/hydrus/ - Hydrus Network

Archive for bug reports, feature requests, and other discussion for the hydrus network.

Index Catalog Archive Bottom Refresh
Mode: Reply
Name
Options
Subject
Message

Max message length: 8000

Files

Max file size: 32.00 MB

Max files: 5

Supported file types: GIF, JPG, PNG, WebM, OGG, and more

E-mail
Password

(used to delete files and postings)

Misc

Remember to follow the rules

The backup domain is located at 8chan.se. .cc is a third fallback. TOR access can be found here, or you can access the TOR portal from the clearnet at Redchannit 2.0.

Visit >>>/h/ for all your lewd needs!

8chan.moe is a hobby project with no affiliation whatsoever to the administration of any other "8chan" site, past or present.

(4.11 KB 300x100 simplebanner.png)
Hydrus Network General #1 Anonymous Board volunteer 01/20/2021 (Wed) 04:21:14 No. 15143
This is a thread for releases, bug reports, and other discussion for the hydrus network software. The hydrus network client is an application written for Anon and other internet-fluent media nerds who have large image/swf/webm collections. It browses with tags instead of folders, a little like a booru on your desktop. Advanced users can share tags and files anonymously through custom servers that any user may run. Everything is free, privacy is the first concern, and the source code is included with the release. Releases are available for Windows, Linux, and macOS. I am continually working on the software and try to put out a new release every Wednesday by 8pm EST. If you would like to learn more, please check out the extensive help and getting started guide here: https://hydrusnetwork.github.io/hydrus/
I am the developer of Hydrus. After procrastinating for a long time, I finally shut down the old /hydrus/ on 8kun. I have never been a great or comfortable BO, and Codexx kindly offered to host me here on /t/. I hope we can bundle the old Release, Q&A, and Bug Report threads into a General here and enjoy a much smoother and more reliable site. The old /hydrus/ is archived at >>>/hydrus/ . The plan is that whenever this General fills up, we will migrate it there to continue the archive. If you are new to hydrus, there are several new concepts, and it can get quite complicated. I strongly advise checking out the getting started guide in the help linked above. I am just a normal Anon, and I work on it alone. I am not a very good or professional programmer, and I am always overwhelmed with items to add or fix or just bad code to clean up on my todo list, but hydrus can do some fun stuff anyway. I am happy to answer any questions anyone has. I am sometimes slow to get back to my messages, but I try to get to everything.
I like to make a post on Tuesday nights talking about my week and the release coming the day after. I had a great week. I fixed up some issues with new systems; one where tags were sometimes being forgotten by the new tag search cache, and another where late-created network sessions were sometimes causing harmless but annoying errors on client shutdown. There are also fixes for the gelbooru downloaders, improved handling for when you try to import a downloader your client cannot yet support, and some help and code cleanup. The release should be as normal tomorrow.
>>15143 Glad to know.
https://www.youtube.com/watch?v=SpXsBNBtqP0 windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v426/Hydrus.Network.426.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v426/Hydrus.Network.426.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v426/Hydrus.Network.426.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v426/Hydrus.Network.426.-.Linux.-.Executable.tar.gz I had a great week. I mostly fixed and cleaned things. 8kun board deleted The 8kun /hydrus/ board is now deleted. An archive has been migrated to >>>/hydrus/, and all the links across the program and help are updated to reflect the changes. I appreciate the help from Codexx to get this going. downloaders from the future It seems gelbooru changed their markup recently, and our default downloader stopped pulling tags, at least for some. A user helpfully created an update--which is rolled into today's release if you missed it--but unfortunately users who were running older clients ran into an unusual versioning bug. The updated downloader would only work correctly on above v422 or so, leading older clients to get a variety of annoying errors about 'md5', 'hex' or 'additional_info' when either booting or trying to inspect the broken object. I have fixed the problem here. Anyone who recently imported a 'downloader from the future' should have it fixed today. Furthermore, the various ways you import downloaders now try to check against versions your current client can support, and if they are too advanced, you now get a nice non-spammy popup message and the too-complex objects are discarded from the import. If the advanced object is buried deeper inside the downloader, it may not be able to discard neatly yet, and you will just get a 'soft' popup message about it, but I hope to tighten this up in future. This version checking is also applied more softly across the whole program. It is less likely that one of your internal objects will be from the 'future', but the client will now make a popup if this is so. Please let me know if you get a whole bunch of these. With luck, this is ultimately a rare problem and I can toughen up this error and actually stop clients from running if their objects ever turn invalid. I also fixed up error recovery in the downloader system UI and multi-column lists across the program. When some borked object enters them, they now try to show an appropriate 'cannot render' style text, rather than raising an error. the rest I also updated the gelbooru pools downloader. Some users get different pool 'gallery' markup from the site, we are not sure why, but it should be fixed now. Last week's network object breakup went well overall. There was one unfortunate but harmless error popup that could appear on client shutdown, typically when certain websites had downloads start in the last five minutes of the program, which I have fixed. Also fixed is the new fast tag search sometimes dropping subtag results. If you noticed that 'samus aran' was not appearing in results, even though 'character:samus aran' existed and could appear in a 'character:anything' query, this is now fixed. Thank you to the users who reported on this and helped me figure it out. A new maintenance routine will run on update to fill in any gaps that may have appeared for you. It shouldn't take more than a minute, even if you sync with the PTR. I added a new 'profile' mode to the help->debug menu, 'callto profile mode'. This one tracks mostly UI-level jobs that work in the background. If I have been talking to you recently about heavy downloaders or similar making the UI juddery, please give it a go and send me the profile logs as usual. Also, profiles in general should be less spammy with the popups. The 'bandwidth used this session' section of the main window's status bar is now more accurate. In certain cases, after a delay, it could report bandwidth used in a session in the hours or days before boot, but now it is exclusive to this session. The difference here is usually not noticeable, but if you restarted a client after pausing all network traffic and then left it open for three hours, it could be confusing!
full list - misc: - thanks to help from Codexx at 8chan.moe, the old 8kun board is completely migrated and archived at 8chan.moe /hydrus/. going forward I will be maintaining a Hydrus Network General there on /t/ for merged release posts, Q&A, and Bug Reports. the plan is that whenever it fills up, it will be moved to the /hydrus/ archive. the links across the program and help are updated, please let me know if I missed any. Endchan /hydrus/ remains as a bunker - fixed a bug where subtag entries in the new tag fast search cache were being deleted for all namespaces when a single namespaced version was went to count 0. it meant some autocomplete results were not appearing, often after some sibling changes. a new 'repopulate' job has been added to the database regenerate menu to fix this efficiently if something like it happens again. this routine will be run on update to fix all users, it shouldn't take long (issue #785) - fixed a bug where the new network objects would throw an error on save when a 'dirty' object was quickly deleted. I think this was typically sessions that only have ephemeral session cookies being created in the final five minutes of the program and then being cleared during program exit - when an archive/delete filter finishes, it now fires off all its changes in one go. previously they would go in ~64-file chunks over the next few hundred milliseconds. this will add a small amount of 'refresh lag', delaying page refreshes etc..., on bigger filter jobs for some clients, but it will guarantee that if you hit F5 real quick after finishing filtering on a processing page with non-random sort, you won't see the same files again at the top, only for them to be swiftly archived/deleted as you watch. trash file performance is much better these days, let me know how this goes for you if you do megafilters - the tag import options whitelist now checks subtags of parsed tags. if you add 'samus aran' to the whitelist, but the site delivers 'character:samus aran', this now passes the whitelist - thanks to a user's submission, the gelbooru 0.2.5 post parser is updated and should get tags again, for those users who stopped getting them last week--however, I never experienced this myself, so please let me know if you still have trouble. there could be something more complicated going on here - updated the gelbooru 0.2.x gallery parser to handle an alternate form of gelbooru pools--we did not figure out why different users are being given different markup, it wasn't as simple as being logged in or not, but there is a difference for some. this parser is folded in on update, so the gelb pool downloader should be fixed for users who had trouble with it - also updated the gelbooru pool gallery url class to infer next page url, as in the alternate form the next page is difficult to parse - the 'clear all closed pages' command under the 'undo' menu now asks for yes/no confirmation - added a 'callto' profile mode, which will be very useful in diagnosing GPU lag in future. the 'callto' jobs are little off-main-thread things like image rendering and async panel preparation. should help us figure out where big download pages etc... are eating up CPU - the different profile modes in the debug menu now all show popup messages, but only when their job exceeds the particular profile's interesting time, usually 3-20ms. this should reduce spam - the 'this session' bandwidth tracker on the status bar is now a special tracker that only includes data from boot. previously, it was using the 'global' tracker, which after certain time intervals (four minutes, three hours, three days), will compress bandwidth history into larger time windows to save space. if one of these windows covered time before the client started, it could spookily report a little bandwidth used on a client started with network traffic paused - bandwidth data usage in times shorter than the last ten seconds (which are smoothed out to avoid bumps) now also get the 'don't get bandwidth from the future on motherboards that had a briefly crazy system clock' fix from last week - fixed some disengaged database tuning that was leading to worse cancel times on certain jobs - updated a whole bunch of the help so section headings are links with nice #fragment/anchor ids, making it easy to link other users to a particular section. I will continue this work, and future help will follow this new format - fixed some bad character encodings in the changelog document, siblings help, and tagging schema help. these should now be utf-8 valid - . - object load improvements: - the client now detects serialisable (saveable) objects that were generated in a future version format your client does not yet support. this mostly affects downloader objects like parsers, where you might import an object a user in a much newer version of the client made. for instance, this week some users imported a fixed gelbooru parser in an older client, which was then saved and double-updated later on, and that caused other problems down the line. downloader imports deal with this situation cleanly, but otherwise it mostly makes a popup notifying you of the problem and asking to contact me. there are about 170 places in the program where objects are deserialised and I am not ready to make this a fullblown error until I know more about people's IRL situations. let's hope this is not widespread. if you run into this, please let me know! - if you were running an older client and manually imported the updated gelbooru parser that was going around, and then you got errors about 'md5', hex' or 'additional_info' something, it _should_ be automatically fixed on update. you should be able to update from previous to ~422, see it in network->downloader components->manage parsers, and it should just work. many users will have the entry overwritten anyway in the above gelb update I am rolling in. if any of this does still give you trouble, please delete and re-import the affected object(s) - importing one of these future-versioned serialised objects using the import/export buttons on a multi-column list, either clipboard, json, or png, will cleanly discard future objects with a non-spammy notification - the Lain drag-and-drop easy downloader import does the same - the parser 'show what this can parse in nice text' routine now fails gracefully - multi-column lists now handle a situation where either the display or sort data for a row cannot be generated. a single error popup per list will be generated so as not to spam, bad sorts will be put at the top, and 'unable to render' will occupy all display cells - . - network server stuff - fixed being able to delete an account type in the server admin menu - the way accounts are checked for permissions serverside now works how the client api does it, unified into a neater system that checks before the job starts - did some misc server code cleanup, and clientside, prepped for restoring account modification and future improvements next week I started on the network updates this week. I will be cleaning more server code and reworking my ancient unit tests, and getting some older admin commands that were 'temporarily' turned off working again. I'll also continue reformatting the help to make sure all the headers have #fragment links, to help external linking to specific sections.
>>15147 >It seems gelbooru changed their markup recently, and our default downloader stopped pulling tags just use gallery-dl and youtube-dl https://github.com/mikf/gallery-dl
>>15144 This is a newfag question that might seem nitpicky, but is there any formal way to have multiple separate databases/clients/boorus? If you wanted to lets say fully separate your SFW with your NSFW, I'm aware you could just make a tag for both of them, maybe a parent tag, but is there a way to also just have different databases so you don't accidentally open porn while looking at classical artwork?
(14.46 KB 363x527 image.png)
>>15150 Use launch arguments via Windows shortcuts. Take a look at this example I've provided, notice the -d argument, which specifies where on the disk that the Hydrus DB should be located, and if it doesn't exist; create it. You'll realise the same version of Hydrus (i.e. a single install) can be used to manage muiltiple databases. Steps; Unzip Hydrus Client to somewhere on your PC. Right click client.exe (the one with the psi icon); then send to desktop (create shortcut). Right click the newly created icon on the desktop, then hit properties, followed up by the shortcut tab as shown in my screenshot.
>>15150 Yeah, this >>15151 is the way to do different ones. This process works for several users. More here: https://hydrusnetwork.github.io/hydrus/help/database_migration.html I have never liked it though, as having two clients means two sets of options and storage and backup to sort out, it means files can't exist in both domains without being duplicated, and if you end up 'syncing with the PTR', which is a hydrus way of getting the approx 1 billion tag mappings hydrus users have publicly shared, and results in storing about 42GB of database files locally, then that sync only works on one client at a time, so if both sfw/nsfw had booru files and would benefit from the PTR, you would have to have two separate syncs, which is pretty ridiculous. So, I am hoping in Q2 of this year to finally get to allowing the client to store multiple 'local file services', which is basically internal file partitions. At the moment, everything goes in a domain called 'my files', but I'll make it so you can add more, like 'art reference', 'sfw', 'nsfw', 'family photos', whatever you like, and then choose which to include in any tag lookup or search page. Files will be moveable, in multiple domains at once without needing to be duplicated on the hard drive, and you'll only need one sync of the PTR. This will also hopefully fairly decently solve the 'my sister is looking over my shoulder' problem of trying to hide nsfw results when looking up something sfw. You just change a search page to 'sfw' domain, and I will never give you accidental lewd tag autocomplete results when you type in a search, and so on, because of the hard 'wall' between the partitions. This will be a lot more effective than ideas like 'soft' filters that hide all files with tag x unless they are explicitly searched for. When I add multiple local file services, I will also add ways to merge separate clients together neatly. Please feel free to start on separate clients with '-d', and I will have a better solution I hope later in the year.
>>15149 Thanks. When I first started on hydrus, I didn't know what the hell I was doing, and I ended up reinventing a bunch of wheels. The downloader system is one of these, and it ended up getting iterated on as we found more complicated sites to parse, and now hydrus has its own system of user-editable objects to define gallery searches and url types and data to parse that is all fed into the different sorts of download the client supports. Were I to start again, I would likely use more tech from gallery-dl or imgbrd or others, but my system is somewhat more tightly coupled now, and hydrus likes some odd information that many common downloaders do not provide, like source URLs to associate with files stored in the client, the 'types' of tags (which many booru APIs also do not provide) so a hydrus tag 'namespace' can be applied, and post time to calculate the current 'file velocity' of a download source to throttle future check times on recurring downloaders. That said, some sites like twitter have always been tricky, and we can't keep up with changes as well as larger projects like youtube-dl. I do plan in my next expansion of the downloader system the ability to patch it into arbitrary external executables, so you can make simple rules like 'if you hit a twitter tweet url and it looks like video, run youtube-dl with these commands and import the file it makes back'. I am sure similar help could be found from gallery-dl and others. In the meantime, there are still some sites that are a real pain to pull from. There are a variety of user-made downloaders here: https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders But when a site is only supported by another program, I know several hydrus users who have that program drop its downloaded files into a folder they have hydrus monitor for automatic import. Updating downloaders in hydrus as sites change has been a long time pain. I hope in the next iteration of the network engine that I can add versioning of download objects and automatic updates as authors publish new versions, a bit like how a vidya Mod Manager works.
After talking about multiple local file services above, I should mention for anyone new to hydrus that when I finish a big project and have no emergencies to get to, I like to put up a big poll of what to work on next. As well as giving me my next work, it serves as a useful way to see what people are generally interested in. I ran one just last month, over Christmas, and this was the result, where multiple local file services won: https://www.survey-maker.com/results3310902xA574481e-102#tab-2 Users can vote on multiple items in these. It is always a gigantic list. In Q1 of this year, my 'big job' will be working on some behind the scenes network code cleanup and new tools for server admins. Assuming I can get that a good whack of work done by the end of March, Q2 will be the file services.
>>15153 >drop its downloaded files into a folder they have hydrus monitor for automatic import. the issue with that is there is no url history (no way to regex add urls through importer), no txt file tag sideloading through auto import mode, no hash collision checking (no md5 lookup in client api)
The drag and drop thread watcher does not work for 420chan.org at this time. The watcher reports the thread is dead and the parser could not find anything. While fixing the URL class, I discovered this bug. The example string is converted correctly into the API string in the string converter dialog, but upon applying the changes, the API URL box lists the normalized string instead. I have applied changes and reopened the client several times now and it remains like this.
(5.00 KB 512x118 420chan_thread_bug.png)
Attached is the exported URL class.
Does anyone have a kemono.party downloader? I heard there was one but I couldn't find it in the downloader repo.
>>15158 I'd love this, and one for The Permanent Booru would be great as well. Haven't even looked into Hydrus yet personally, it seems a bit confusing and all over the place. But if it can grab from kemono or TPB I'd dedicate my life to it. Yiff's death made me finally get in the mood to start archiving things on my own.
I've been getting "file not found" errors when I import my own pictures; for whatever reason it seems to be specific to png files.
Are the file lookup scripts accessible from the tag manager being deprecated? I've always used those to populate tags when downloading from supported sites but recently (even on 426) the gelbooru one stopped working
I used to just keep all my import tabs open, just for a reference of what I've ever imported. Only if they were full artist galleries, not 1-off images, or anything. But today I took note of them in a text document, and closed the tabs, just trying to temper Hydrus's ram usage, if at all possible. I think it may have helped a little: at the time I decided to do this, hydrus was using 12 gigs of my 32 gigs of ram, and when it finished doing whatever it was doing, it lowered to 9.5 gigs, which was about normal for me. After closing all my import tabs, it was still using 9.5 gigs of ram but my "session weight" had decreased from 1.5 mil to 1.3 mil. I restarted the program, and it's still freezing and shit from just having booted, which is normal for me and will go away in an hour or something, but hydrus is currently using 6 gigs or ram, and my session weight is 1.2 mil. The literal only tab I have open is my "gallery" tab, which has 2.8k queries, 1.2 million files. I have a url import tab as well (1.9k images), but otherwise nothing. Is there anything I can do here? I don't want to lose the history of the gallery tab. If I'm now at a consistent 6 gigs ram usage, maybe I won't feel hydrus's ram usage anymore. But this still seems like a waste.
Thanks so much for making hydrus
Bug: Tabs on Mac OS High Sierra Tabs on the newest version act weird for me. Switching between tabs is rarely possible. Instead the names tabs which are clicked on become just grey, as they normally do. Does not seem to affect the performance of the program. Happens every session. Sometimes only one tab works, sometimes two. Bug: Sort on Mac OS High Sierra When clicked on, sort by becomes dark grey as normal, but does not show the menu and stays dark grey. Sometimes seems to cause program freeze. Does not happen every session. Log messages: no errors apart from OpenSSL being outdated (probably not related).
So, I set Hydrus's SOCKS proxy to use Tor. I tested it, and it seems to be working. But now I'm getting hit by cuckflare captchas on most sites in the downloader even after I import cookies from a working log-in via Hydrus Companion. What do? I've tried changing my Tor IP already.
>>15165 You can try making your user agent the same as the browser you sent the cookies from. That will sometimes work temporarily but you probably just shouldn't use TOR.
>>15155 Yeah, it is only a patch for now. I want a wider import pipeline, more ways of getting different external metadata into the client, and ways for external programs to push too. I'll keep working on unifying the import pipeline and expanding the client api. >>15156 >>15157 Thank you for this report. I am not sure what is going on here. I'll check it out and fix that bad API converter as needed and roll an update into 427! >>15158 >>15159 I am not totally up to date with the kemono situation, but I believe some lads have been looking at making a downloader for it. If I remember right, kemono switched up their layout as they launched, so perhaps the hydrus side was delayed until things were settled down. When a downloader is completed, it should appear here https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders , and since kemono is popular, like yiff was, I'll roll it in the default downloaders.
>>15160 Thank you for this report. When you get this error, is this appearing as a popup in the bottom-right corner of the main window, or as an 'import failed' result? Or is it once they are imported, they then fail to load as thumbnails/files? Have you changed any of the locations under database->migrate database, or are you running a fresh database with default locations? Can you talk about the problem a bit more generally? If it is a popup, can you click the 'copy traceback' button and paste it here, or pastebin it, or email it to me? If it is in a hard drive import page, can you click the little 'table' icon on the left panel, scroll to the 'failed' results, and right-click->copy notes? You can scrub any private path info, I am just interested in the code. >>15161 Not so much deprecated, but updated significantly. I like what they do, so while I will replace them with something that uses newer and saner parsing tech, I hope to have a system in place to update all the existing objects to the new system. We'll see how truly feasible that is. >>15162 I know a couple of other users who have very large sessions. Unfortunately I just can't support it well yet. I know I can improve the CPU load significantly, and I have a plan to reduce the session save time (which would be a lump of lag every 15 minutes or whatever you have session save time set to), but the best solution to having a laggy session is to reduce the amount of stuff in active memory. I am sorry to say, but I generally recommend having a 'session weight' of less than 80k at the moment. If your gallery tab is finished doing its work, so you don't actually need it in active memory to do more downloading and you just want a record, one neat solution could be to save that 'page of pages' tab as its own session, call it 'gallery log' or whatever, and then only load it in when you need to check on it. You should be able to do session load/save by right-clicking 'page of pages' page tabs now. All that said, I introduced this new 'callto' profile mode this week. If you could run a bit of that (there is an explainer link under help->debug->profile modes if you are new to it), and send them in to me, that would be really helpful to see where a bigass session is spending its CPU.
is there any idea about when we will get comparability with MacOS Big Sur?
>>15169 I hope soon. Some guys are working on it, I think they have a prototype going. Once that is running ok, I'd like to fold it into my normal releases. If you are comfortable with discord and would like to help test it, please check in with Suika there.
>>15163 Thank m8, I am glad you like it. >>15164 Thank you for this report. I am sorry for the trouble. I do not think I changed anything in Qt or tab behaviour, so I wonder if High Sierra got some update. I'll check this out on my own client. We may need to do some back and forth here to figure out what is going on, so please let me know how this works for you in future versions. BTW, when we move to this new Big Sur compatible build, that OpenSSL shit should go away and cloudscraper will start working on macOS, as the build will be made with newer libraries on github rather than my old laptop. >>15165 >>15166 Yeah, I think TOR is probably the problem here. By definition, TOR will connect to sites with different IPs, right? And CF tie your identity to IP, so even if you populate hydrus with your (even TOR) browser's cookies, CF will still see you as a new lad soon enough and will want to re-authenticate. CF may also recognise TOR and just have super strict TOR rules, but I don't know enough to be confident.
>>15171 cf does recognize tor, iirc they run a fuckton of exit nodes as well
>>15168 It seemed to be an issue related to FFMpeg, I replaced the version that came with it with my own and it works fine now.
>>15168 Ok, I just sent you an email of the callto profile mode log + all older logs that were in the same folder.
is there a way to exclude certain namespaces from suggested tags? it's a really useful feature but the list is always cluttered by creator: tags for me.
(908.83 KB 1280x720 moondancer.png)
This software offers a neat way to import and export tags using a .TXT sidecar files. Does exist a way to do the same with notes?
>>15171 Thanks for your reply. No High Sierra didn't get any updates recently and won't get any ever since the last update in November (though Apple would never make an official statement saying so, of course). If you aren't familiar with the issue then it's probably just my old potato. Just wish I could give you some error logs.
(410.00 KB 1000x1750 rarjpg.jpg)
Replying to https://8kun.top/hydrus/res/11542.html#15137. Please point to the right thread in case I'm doing it wrong. > Thank you for this report. I am sorry for the continuing trouble. If searches are slow when you are searching a page of results, but fast when you search an empty page (or one with 'searching immediately' off, which then searches the database, rather than the tags for the files in front of you), then the routine that is generating search results from thumbnails is slow. I doubt it has something to do with the thumbnail related routines as I have the same slow searching problem in the tag manager (F3→public tag repository→search for keyword) where there is no thumbnails. > Some more feedback would be very helpful. Please run 'db profile mode' and do some slow autocomplete searches with files in front of you, and then pastebin or email me that profile log. It is the 'media_predicates' routine that is running slow for you. > > The >>15000 query on client.caches.db would be helpful. Let's see what SQLite wants to be doing here. I put the results in rarjpg attached to my post. Download the picture, change the extension from jpg to rar, and you should be able to extract it. Please, read readme.txt first.
(493.26 KB client-logs.tgz)
>>15178 Dammit, the board engine turned rarjpg to plain jpg. I'll try to attach the archive to this post then.
Has anyone tried the "minimize to system tray" thing on Linux? It says that it may be buggy so I don't really want to risk trying it on my own.
(35.90 KB 813x612 Screenshot_20210124_213032.png)
>>15180 In KDE Plasma works fine. --Desktop: KDE Plasma 5.14.5 tk: Qt 5.11.3 wm: kwin_x11 dm: SDDM --Distro: MX-19.3_KDE_x64 patito feo November 11 2020 --base: Debian GNU/Linux 10 (buster)
(60.80 KB 691x759 h.png)
So this just happened after Windows Defender updated
>>15182 Got that one too
>>15147 >I also updated the gelbooru pools downloader. Some users get different pool 'gallery' markup from the site, we are not sure why, but it should be fixed now. Thanks for doing that.
>>15153 Giving Hydrus pluggable support for other downloaders as an alternative to its downloader engine would eventually give a lot of flexibility, so as long as there was some option for said external downloaders to communicate additional metadata (source URLs, tags, etc.)
Maybe its because im retarded, but I want to test hosting a local booru and I keep getting the upnpc error. I already installed miniUPnP (Manjaro) but there's no executable in the library that I can move. What can I do?
Let me know if this is the wrong place for this. Having Linux Trouble: I'm pretty new to Linux and I've been trying to get Hydrus working to no avail. Running Debian 9 (I know I gotta update, I'm just a lazy shit scared to fuck things up) I downloaded the Linux executable. Extracted (tar -xzvf [filename].tar.gz). After some fucking around, got it to launch (from client.py, is that correct?). Worked on my database for an hour before the program crashed(?) and now it won't launch again. When I run it from the command line (python3 client.py) I get the error:"ImportError: No module named 'send2trash'. I'm hopelessly fucking lost here. Don't know if I'm going about this totally the wrong way or how I got it to work at all. Also, not sure if this is the right place for this, if not lmk.
>>15188 Btw, the version that I got working that one time was 376
>>15182 >>15183 >>15184 Thank you for these reports, and the definition scan. These unfortunately happen from time to time. I used to bundle a copy of upnpc with the program to help some optional UPNP stuff when you run servers, but it triggered virus scan problems evelry few months, even though the byte content of the exe I bundled hadn't changed in years, that I just took it out. Now client.exe is the problem. ;_; This is most likely a false positive. It always has been in the past. I am always interested in these reports though, and of course, if there is ever one that really seems legit, I would like to know as soon as possible. It would be nice if they said on these reports exactly what triggered the problem, but I guess keeping all that secret is part of the game. If you would like, you can roll back to 425 and wait a bit or try a future release. New definitions tend to roll out in a few weeks and these problems go away, and they often go away with the next release as well. I presume in these cases the detection is an accidental collision of bytes or a statistical spike that stands out by accident. Or you can tell Windows Defender to ignore the hydrus install folder. 427 will be out tomorrow, we'll see how that does with Defender. I just ran a test prebuild against Virus Total, it gave this: https://www.virustotal.com/gui/file/6e4a3c0db5af923feb37f5f170d080a63fce2f4e5979d335ab8d9868e13c360d/details Same basic thing. Seems like SecureAge Apex are not too chuffed for the moment, but all else looks clear. Another option here is to run from source. This is much more complicated to set up, but then you only have to trust my code. https://hydrusnetwork.github.io/hydrus/help/running_from_source.html
>>15173 Ah, thank you. This is on Windows, right? Can you say roughly which version you swapped to, was it a newer one? I scan pngs with ffmpeg on import to check if they are really apngs. If my current bundled ffmpeg has a problem with pathnames with accents in or whatever the problem was, I'll be interested to fix it in my actual release. >>15174 Thank you! >>15175 Not yet. I want to bulk out this whole system in future, particularly in prep for alternate and better 'suggest' sources. Attaching a tag filter is a great idea.
>>15176 Not yet. I am planning a rework of all that sidecar import stuff. I want to generalise it to take and parse json and xml as well, be more flexible with the name of the sidecar, and plug it into the whole metadata import pipeline so it is all running on the same basic code as any downloader and can add any info that can take like source URLs. Then I will have to add 'note import options' to that main import pipeline. I never got around to this when I last expanded notes, and had trouble fitting it in since, so it was on the last 'big jobs poll'. It will come, I'll likely just fit a simple version in one week and iterate on it, along with Client API support. Please hang in there for now. Keep some sort of record of which notes info goes with which file or file hash, and I am confident we can figure out a nice import in future. >>15178 >>15179 Thank you. It is still that one query that is running slow for you, but that EXPLAIN QUERY PLAN looks fine. I am not sure why. I can't have this done for tomorrow, but am going to investigate this more. >>15187 Ah, damn, you are hitting the problem related to my upnpc virus scanning issues in >>15190 . If it can't find an executable in install_dir/bin, it is supposed to try 'minupnpc' on your PATH. If you just open a shell, can you run 'miniupnpc', or does it have a different name for you? If you just want a cheat solution here, grab this older version https://github.com/hydrusnetwork/hydrus/releases/download/v400/Hydrus.Network.400.-.Linux.-.Executable.tar.gz and just pull the upnpc_linux exe out of the bin dir to your current install. It should just work™.
>>15188 if you downloaded the executable you should launch the "client" file, simply doubleclick or open a terminal window and type ./client and press enter also iirc if running from source you should use client.pyw, not client.py, but not certain about that
>>15188 >>15189 No worries. If you download the Linux Executable build, there should be a simple 'client' executable file in the base directory you extract, right beside the client.py file. I build on Ubuntu, so if your Linux is anywhere similar to that, just try double-clicking that file, and that's all you have to do. No need to mess around with the py and all the pain in the ass pip stuff. If you can't double click, just run it with ./client from a terminal. If the executable just doesn't work for you, you'll likely have to run from source. I have some help for that here: https://hydrusnetwork.github.io/hydrus/help/running_from_source.html Although if you got it to run from client.py, perhaps you already did some pip/venv work? Maybe while you were trying to get the .py to go you accidentally ran the executable from terminal, and that's what worked? Let me know how you get on. I'm happy to help. The program is more stable in Linux now, but I know it crashes still for some users with very large databases or slightly unusual window managers. Running from source can help with that, if you still have issues.
>>15179 I just remembered something--can you open help->about and tell me what version it has for 'sqlite'?
>>15193 >>15194 Thanks. That seems to have done it. >Maybe while you were trying to get the .py to go you accidentally ran the executable from terminal, and that's what worked? Yeah, that was probably it
I had an ok week. I mostly fixed and cleaned things. Another instance of 'ghost' pending tags is fixed, with bad counts and tags corrected nice and quickly with a new maintenance routine. Also, the various undocumented shortcuts in the tag autocomplete search box are now exposed and customisable. The release should be as normal tomorrow.
Sankaku gallery tag searches that include the 6+girls tag are stopping at the end of the first page of results for me. I assume the + sign is messing it up somehow. Is there some way I can tweak the query so that it will run properly?
>>15198 Fugg, yeah tags like that need special handling. I'll look into why this is failing for sank. It might be a downloader object or my url handling code that is failing. If you want to brute-force it for now, you should be able to drag and drop individual page URLs from your browser onto hydrus and they'll be queued up just on their own in a new 'urls' downloader page. You can also likely generate them with a script and then copy/paste as well.
https://www.youtube.com/watch?v=E97ftpTjeC4 windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v427/Hydrus.Network.427.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v427/Hydrus.Network.427.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v427/Hydrus.Network.427.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v427/Hydrus.Network.427.-.Linux.-.Executable.tar.gz I had an ok week. I mostly fixed and cleaned things. more ghosts There was another 'ghost' pending tag issue. Sometimes, on committing pending tags with siblings up to the PTR, the pending record and its count was not clearing properly. I reproduced the problem and fixed it, and a new test ensures it will not happen again. Thank you to the reports about this, they helped me to figure it out. Since we have had some trouble with pending tags in particular, I have written a new maintenance routine to recalculate pending tags and their counts efficiently without having to wait for siblings/parents re-sync. This will run on update, so with luck, your tags and autocompletes should be fixed up (until the next problem!). the rest There is a new shortcut set under file->shortcuts to handle tag autocomplete shortcuts. Any 'switch searching immediately' that was previously on 'main gui' will be migrated over to this on update, and I have also moved several previously hardcoded and undocumented shortcuts to the new shortcut system, so you can now edit them! The new actions (with their defaults) are: forcing tag autocomplete lookup now (ctrl+space), flipping an IME-friendly input mode on/off (insert), navigating autocomplete results tabs left/right when input is empty (left/right arrow), navigating the service pages in manage tags dialog left/right when input is empty (up/down arrow), and navigating a parent media viewer previous/next when input is empty (page up/down). I fixed a bad auto-setup link in the gelbooru downloader last week that messed up some users. Also rolled in is a new URL definition that should fix 420chan thread watching for their newer url format. full list - ghost pending tags: - fixed another ghost pending tags bug. this may have been new or there since the display cache started, I am not sure, but it shouldn't happen again. it was occuring when a pending tag was being committed to 'current' and another tag in its sibling group already existed as a current tag for that file. the pending tag and its count would not clear for non-'all known files' domains, causing ghosts to appear in search pages but not typically manage tags. you may have noticed ghost tags hanging around after a pending commit--this was it. the true problem here was in the 'rescind pending' action that occurs just before an add/commit. a new unit test tests for this situation both for two non-ideal tags being pend-merged, and a non-ideal tag being pend-merged into the existing ideal - wrote a routine to regenerate _pending_ tag storage and autocomplete counts from scratch for the combined and specific display tag caches. this job is special in that it regens tags instantly without having to reset sibling/parent sync. you can run the job from the database->regen menu. this is the start of 'fix just this tag' maintenance ability - the pending regen routines will occur on update. it shouldn't take long at all, unless you have five million tags pending, where it could be a couple minutes - . - autocomplete shortcuts: - there is a new shortcut set under _file->shortcuts_ just for tag autocomplete shortcuts. any 'switch searching immediately' shortcut previously on 'main gui' will be migrated over - the tag autocomplete input text box is now plugged into the new shortcut system and uses this set - migrated previously hardcoded autocomplete shortcuts to the shortcut system (defaults): - - force search now, for when you have automatic searching turned off (ctrl+space) - - enable IME-friendly mode (insert) - - if input empty, move left/right a tab (left/right arrow) - - if input empty, move left/right a service page (up/down arrow) - - if input empty and on media viewer, move to previous/next media (page up/down) - misc improvements to my shortcut handler - misc shortcut code cleanup - . - the rest: - I fixed a bad example url in the new gelbooru file page parser that was sometimes leading to a link to the gallery url class. this was an artifact of an old experiment with md5-search parsing, now fixed with newer redirection tech. the updated parser is folded into update, and if you ended up with the incorrect link, it should be detected, dissolved, and re-linked with the file page parser - thanks to a user report, wrote a new url class for 420chan's newer thread url format - sorting a gallery downloader or thread watcher multi-column list by 'status' should now group 'done' and 'paused' items separately - fixed a bug in the /add_tags/add_tags Client API call when checking some petitioned tags data types. cleaned all that code, it was ugly (issue #788) - added unit tests for /add_tags/add_tags to test the service_names_to_actions_to_tags parameter better and repository actions, including petitioning with and without specified reason - . - code refactoring: - finally addressing the near-1MB ClientDB file, I have started a framework to break the db into separate modules with their own creation/repair/work responsibilities. this will make the file easier to work on, maintain, update, and test. this week starts off simple, with the master definitions being peeled off into hashes, tags, urls, and texts submodules - cleaned some misc code around here, including a bunch of related decoupling - ClientDB.py is now in its own 'db' module as well. the db will further fracture and this module will gain more files in future - the boot code in the launch scripts is now migrated to the 'hydrus' directory, with the actual launch scripts now doing nicer main checks to not launch the program if you want to play around with importing hydrus. more work to come here - finished the help's header linking job--all headers across the help are now #fragment links - misc help cleanup next week While I am pleased with the cleanup and fixes, I had trouble getting in all the work time I wanted this week. Next week is scheduled to be a 'medium size' job week. I will try to improve my 'taglist' widget to better display implied parents and siblings in the manage tags dialog. I never got this done properly in the tag display cache work, and workflow suffers for it, so I will reserve some time and hammer it out. There's a small chance it will take two weeks, but we'll see.
Small bug report: From 427, "working..." and "already added" statuses are last on the list when sorting by status; maybe related to the "sorting a gallery downloader or thread watcher multi-column list by 'status' should now group 'done' and 'paused' items separately" change? I'll also report that sorting by status reversed (last first) also sorts by subject reversed (z first), not sure how intended this is. Using the opportunity to thank you for all the hard work, Hydrus is amazing.
Maybe I'm just blind, but I can't find where to initiate a slideshow. I see it mentioned in changelogs and stuff, but no mention of it in the help that I can see.
>>15202 Open an image in the media viewer and right click on it; you should see a menu with an option to start a slideshow.
>subscription help page >Make a 'page of pages' called 'subs' and put all your subscription landing pages in there Sorry for the retard question, but how do I do this? Do I need to edit something in the manage subs dialog, or is dragging the pages onto the page-of-pages enough?
>>15204 In the subscription options there is an option to set it to publish to a page, make sure that is checked. After the subs run and make their pages, you can just drag those pages into the pages-of-pages and your subs will continue to publish to those pages,
I think I remember reading somewhere that the blacklist can be more powerful than a simple "don't download a file if it has a tag on this list". I'm interested in setting up exceptions in some cases. Say I blacklist the tag 'loli', but want to allow it if it also has the tag 'rating:safe'. Can you do stuff like that?
Found a bug in File Import "Add tags before the import": deleting a quick namespace does not refresh the tags displayed in the file list to reflect the deletion (until you do something else to trigger a refresh)
>>15206 The tag blacklist has some advanced functionality but exceptions only really work for broad tags, e.g. blacklist rating tags except for rating:safe. So in your scenario you could query loli and it would only get rating:safe ones but you could just query loli + rating:safe to achieve the same thing.
>>15195 Here it is. BTW is there a way to change the temp directory?
Starting today I have been getting Trojan:Win32/CryptInject!ml Trojan:Win32/Fuerboos.D!cl threat results on the client.exe from Windows Security. I assume they're false positives, but I wanted to let you know
>>15201 Yeah, I didn't think this through enough, it is annoying me on my IRL client as well. I like that things are grouped better, but the actual order needs some more work. I'll put some more time in. I'll also see if I can get 'working' showing up more permanently--it sometimes blits out atm, and I'd love to have a 'pending' status for downloaders waiting to get in, atm they just sit blank, silent, which is often confusing. >>15204 >>15205 Yeah, this works by page name, so wherever your (leftmost) 'gelbooru artists' page or whatever is, the files will end up there once published. I highly recommend overriding the publish name/label in the subscription options to merge several subs of similar type together. Many users end up with 50+ pages for all different queries and 100,000 files and a laggy session. Better to merge and clear it out asap. >>15206 >>15208 Hmm, yeah, I don't think it'll do an exception like that yet. The tag filter exception stuff is just for within namespaces really, so you can say 'no "character:" tags except "character:samus aran" and "character:lara croft"'. Best answer for right now is probably to have a separate downloader/subscription query that looks for "content rating:safe", and you'll save everyone time by doing the filtering serverside. I only recommend the blacklist atm for filtering small amounts of stuff you hate, like guro, diaper, vore, kind of stuff, from broader character/creator/series searches. It is inefficient when the stuff you want to get through is less than, say, 80% of the content in the search.
>>15207 Thank you, I will fix this! >>15209 Hmm, I was wondering if you had an ancient SQLite somehow and the query planner was not clever enough yet, but that seems fine. My Windows here is 3.33.0. I'll keep working and try 3.22 out just in case. You can change the temp directory with a launch parameter, should work for all temp stuff across the program: https://hydrusnetwork.github.io/hydrus/help/launch_arguments.html >>15210 JESAS, they are just going wild with this shit recently. Thank you for letting me know. Someone was telling me this recent rash is due to Defender detecting shit in .NET runtime for some reason. And apparently I can buy into their program for the low low price of $200-600 a year to get certified code, lmao.
(208.87 KB 768x941 0yInqnf.png)
>>15212 >And apparently I can buy into their program for the low low price of $200-600 a year to get certified code, lmao. >pay us or we'll tell everybody that your code is dangerous It's a racket.
>>15212 >And apparently I can buy into their program for the low low price of $200-600 a year to get certified code, lmao. >paying shekels to get a blue checkmark in their "ok software" database conspiring hands intensifies
Ok. This entire post, In formatting it, I think I gained better understanding of the terminology and what's possible. So I just deleted it all and remade. So, I have an artist mega I'm trying to import. Deumosden's, linked from their newgrounds links. It has a ton of folders and subfolders, but hydrus can import those folders under a namespace. It's nice so far. But, to have a "collection" sort for this, I have to manually set it up, despite importing already observing the linearity of each folder. Not only that, but any collection I manually make is now put among all the other generic ones that see far more use. So I mean at least it's still possible to do what I want. But a bit more automation/ sorting options would be better, if possible. Because even without trying to toss the artist-specific collections into a bin of some sort, I have collections based on website, downloader, and even among websites, there are galleries, boorus, imageboards, etc. While we're on the topic, I wish the menuing didn't punish you depending on how many submenus deep you are. There is a clear incentive to keep the ones you use most often under the least amount of subfolders. I wish you could just visualize all your options on the same screen, somehow, while still being sorted.
(164.68 KB 1366x768 Screenshot_20210131_044454.png)
(62.44 KB 514x581 Screenshot_20210131_045001.png)
Bug found. Video won't play and after clicking the seek bar to try to force it to play, a system error message popped up. ---------- >SystemError >('Error running mpv command', -12, (<MpvHandle object at 0x7f37e6e71730>, <mpv.c_char_p_Array_5 object at 0x7f380472d510>)) > File "hydrus/hydrus/client/gui/ClientGUICanvasMedia.py", line 792, in mousePressEvent > File "hydrus/hydrus/client/gui/ClientGUICanvasMedia.py", line 739, in _ScanToCurrentMousePos > File "hydrus/hydrus/client/gui/ClientGUIMPV.py", line 323, in Seek > File "hydrus/venv/lib/python3.6/site-packages/mpv.py", line 732, in seek > File "hydrus/venv/lib/python3.6/site-packages/mpv.py", line 719, in command > File "hydrus/venv/lib/python3.6/site-packages/mpv.py", line 126, in raise_for_ec
I was wondering if it's possible to remove the PTR after syncing with it and make my DBs small again. Not a huge deal if it isn't, I'm close to finishing the sync and was just wondering if I could remove it if I don't get much use out of it.
Is there any way to update a pixiv id as if it were a subscription? Often times I visit a pixiv user page, and it turns out hydrus companion shows me that I've ripped everything the user has posted, save for some recent ones. But I currently don't know how to add those new images to my hydrus besides manually sending each artwork link to my hydrus via a right click of each one, adding the pixiv user as a subscription, or adding the entire pixiv user to a gallery download tab.
Is there a way to automatically apply a duplicate relationship in all pixel by pixel duplicates?
>>15218 What's wrong with the last two? I set up a subscription for pixiv, it just werks. If you add the whole gallery to a downloader, it should only download the new ones. >>15219 I don't think that a way exists right now, but I'm pretty sure dev said he wants to work on a duplicates overhaul sometime Soon™, which would include automatic pixel-by-pixel duplicate filtering.
>>15220 I feel like it doesn't only download the new ones. Maybe I'm wrong, but I thought it was running up my limit. Like, if I'm currently able to download, because I have bandwidth available for pixiv, I should theoretically be able to add a pixiv id to a gallery downloader tab, and if I've already downloaded everything, it shouldn't download anything, right? But I feel that is not the case. I really feel subscriptions do it in a much more efficient way than adding the same pixiv id to a gallery downloader tab. But I don't want to subscribe to every random pixiv user page I wanted to rip at the time. If there's a way to subscribe only until it finishes downloading everything, in which case the subscription is purged, that's still not as streamlined as somehow being able to send all new posts to hydrus using hydrus companion.
>>15221 If you run a gallery search that contains files you already have, Hydrus skips downloading the existing files again (unless you tell it to under file import options or something). It might use a tiny bit of bandwidth checking what the checksum of each file is depending on the site, but probably not much.
>>15212 Thank you! One more issue I just found (v427 on Windows): on a files page, if I right-click a tag in the tag list and "select files with", pressing the "manage tags" keyboard shortcut (f3) afterwards doesn't work (nothing happens). Selecting "manage tags" from the right-click menu still works, though.
I'm in the process of attempting to recover my client.mappings.db and I'm unclear about the third step in the manual import/export. It says to create a clean database from the modified SQL file using the command >>sqlite3.exe client.mappings_recovered.db < client.mappings.db.sql but I feel like I'm missing something with how to run this.
Is there a way to automatically/permanently hide certain images from view? I have an image that I want to keep but it is too large in size or resolution for my PC to display. When I open it my PC crashes so I'd like to hide it to prevent any accidents.
>>15225 Holy shit! Do you just have a bad computer or what are the dimensions and shit of that image? When I was new to hydrus, hydrus straight didn't upload my entire old folders because it decided some of them were "compression bombs". Unfortunately I didn't know what that meant at the time, so I deleted the folders without enabling importing them.
I had an ok week. I was thinking of taking two weeks so I could do the taglist improvements nicely, and since that would give me a little extra time, I decided to pull the trigger on finally upgrading my main dev computer as well. I spent a bit of this week sorting that out, replacing my six year old office computer with a nice mini pc with an SSD. Beyond that, I only have some little bug fixes and cleanup done, so I will spend tomorrow as a normal work day and put off the release. There will be no release tomorrow. 428 should be on the 10th!
>>15213 First build on my new dev machine. :^))))))))))))))))))))
>>15215 Yes, I am sorry, I have never really liked 'collect by' in hydrus. It has worked well for collecting really well tagged manga chapters, but when sorting miscellaneous content by artist or something, it becomes frustrating to deal with the reality of messy tagging. The collect-by rules are strict intersections of namespace sets. My hope is that with the file alternates expansion coming almost certainly later in the year, we will gain some tools to have more containers for groups of files that work like a virtual cbr file, or other semi-sticky relationships, including sorting for small 4-page comics, so you can set up and restore nicer groupings of related files. When we get to that work, I would appreciate your feedback on what works well and badly for you. I also like the idea of populating collections automatically by the namespaces you have on your system. Menus are a constant fight. Different users use different functions more or less, and I am grappling with several sections of legacy code. My current strategy is to keep rewriting hardcoded blocks into more dynamic sections as I go, with the final aim of having everything pop in dynamically so I can hang an options structure around it and let you rearrange things yourself. In the meantime, though, I will have to focus on little areas. Which would you say is the most annoyingly buried menu item for you?
>>15216 Damn, I saw something like this recently. I was working with a guy who was running from source on Linux, getting libmpv1 and python-mpv, and we discovered the latest versions were suddenly failing to talk to each other. It looks like you are running from source. Did you have to install libmpv1 with apt or similar? If you are running from source, can you open a terminal in your hydrus venv environment and go: pip3 show python-mpv If you are using the Arch package, I believe that runs from source but sets it up for you, so you may not know how to do this. Let me know if this is so. That should show your version. I discovered 0.5.4 suddenly caused issues. Rolling back to 0.5.2 may help: pip3 install python-mpv==0.5.2 But I also had problems on my Windows using the latest libmpv1, so it sounds like there are multiple problems here. My current working Windows is 0.5.2 reporting API version 1.109. Please let me know if you discover some nice version numbers here. I don't know how you roll back with apt, but I guess it is a similar version parameter?
>>15217 You can shrink the db a lot, but not all the way. I don't have a maintenance routine to 'recycle' the definitions in client.master.db yet, but it will come. To remove the PTR, just remove the service under services->manage services. It may take a few minutes to finish, just let it work. Once it is done, run database->maintenance->vacuum->force, which will truncate your db files. A full sync with the PTR should shrink from 42GB down to 9GB or so, almost all space lost from client.mappings.db. >>15218 >>15220 >>15221 >>15222 Just a side thing for Pixiv. Because they are weird and allow multiple files for art 'post' URLs, hydrus has to do a little more network traffic when it works URLs it has seen before. You are correct that it isn't much--it just has to grab the HTML for each page and check the list of actual file links. This only happens once per subscription, on initial setup. Normal boorus can skip this work much quicker, without pulling any data from the site. If you add a pixiv id as a subscription, it should work like any other subscription. They have a weird way of delivering 'pages' of results, but hydrus compensates. If you aren't seeing new results, it could be the artist does not upload much and the sub isn't planning to check for a week or so. A fix here would be to tell it 'check now' in the edit subs dialog. If the sub has checked, and still didn't find anything, the problem could just be Pixiv being Pixiv. They mess around with logins and cookies and nsfw permissions all the time. Hydrus Companion is usually great against this by just copying cookies, but I have learned not to expect typical behaviour from the site. Give the subs time, and they should catch anything they miss in subsequent checks.
>>15219 >>15220 Yep, there is no cache for pixel dupes yet, so I can't search for it. I calculate it on the fly as images load. One of these 'medium size' job weeks, which I try to do every four weeks, I'll sort out a pixel dupe cache with search code and a maintenance routine to retroactively populate it. Before I do that, I do want to fix some ICC colour profile stuff, as I know some pixel dupes checks are giving false-positive and false-negative matches. >>15223 Thanks. I'll bet that is a keyboard focus issue. The shortcut handler that launches manage tags is on the thumbnail grid, and when you click the taglist, focus moves to that. I will make all those 'select files' actions off the taglist move focus to the thumbgrid immediately, since that is what it is affecting. >>15224 Forgive me, I do not think I wrote that section--it came from another user--so I am not totally expert on the method. That line should be executed in a command line terminal. If you are on Windows, go shift+right-click on the db folder, and you should get 'open command prompt/powershell here' option. If it launches the powershell, type 'cmd' and hit enter to get an old style terminal. Then paste that line and hit enter. I think the 'in' arrow in windows terminal means 'once the program is loaded, read that file as stdin', basically typing it all as if you did. ALTERNATELY, I think you can do it from the sqlite3 program. Do this I think: .open client.mappings.db .read client.mappings.db.sql .exit Let me know if you still have trouble.
>>15225 >>15226 Decompression bombs are a pain. They used to be actual things designed by hackers having fun making clever png compression tables that would blow up to gigabytes of ram when loaded, but now it is mostly Patreon artists making 25,000 x 40,000 pngs that accidentally temporarily need 2.8GB to render. I consider this the '34MB gif conversion of a 10 second 2.4MB 720p webm' filetype sin of our modern age. Hydrus no longer blocks these files as there aren't good ways to detect them without false positives and many users still want them anyway. Unfortunately, there is no way to hide just one image in hydrus atm. Given it crashes your program, I recommend you move it out of hydrus until you have more ram or the guys who make image decoders figure out a solution to this class of problem. Clicking on it is obviously dangerous, but I think if you hide your preview window, you'll be able to click it and drag and drop it to your desktop and then delete it from hydrus ok. To hide your preview window, double-click the little horizontal sash that separates the bottom-left preview with the top-left search/tag lists. Hit up the pages menu to restore sashes.
(59.30 KB 496x640 20162285_6_496_640.jpg)
>>15159 It has it's quirks, but given enough time it'll pretty easy to work with. In the extended downloader repository, there is a partially finished parser for kemono https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders/Kemono.party Other than that, you can fetch low quality versions from E621 and other boards, and later replace them via duplicate processing with higher quality versions pulled from another source.
>>15210 Linucs maeter raaace
(125.59 KB 1366x738 Screenshot_20210202_230511.png)
(102.99 KB 1366x738 Screenshot_20210202_230557.png)
(151.35 KB 1366x738 Screenshot_20210202_232133.png)
(118.31 KB 900x900 unnamed.jpg)
>>15230 >It looks like you are running from source. I'm not. The package used is Hydrus.Network.426.-.Linux.-.Executable.tar.gz libmpv1 was installed with apt-get python-mpv is not available in my package manager. As shown in the screenshots, I downgraded libmpv1 from version 0.29.1-2~mx19+3 to 0.29.1-1 (stable), but the problem is still there. By the way, I have inside Hydrus perhaps 100 videos, only 2 won't play throwing this error message. Anyway, this is a minor occurrence and I encourage you for the time being to dismiss it, as its importance is secondary.
(242.04 KB 1366x768 Screenshot_20210203_014055.png)
>>15230 SOLVED! I downloaded the same video but with a different resolution and file extension. Now it's working great.
>>15237 I didn't think anyone could make MX KDE look that fucking ugly. What the fuck have you done to that panel?
>>15229 >In the meantime, though, I will have to focus on little areas. Which would you say is the most annoyingly buried menu item for you? My usage of the program is very superficial, but even in my limited usage, the one that sticks out to me is after enabling help > 'advanced mode, the reverse image search option is under [right click] > manage > file relationships > find similar-looking files > exact match/ very similar/ similar/ speculative. To me this is completely bogus. I don't mind the threshold options, even though I have only ever chosen the lowest threshold. But for comparison the reverse image search addon I've installed in my browser is only [right click] > Image Search Options > [any site on this submenu]. For that matter, why is reverse image search functionality locked behind advanced mode. To me that's pretty fucked up. Unless I'm mistaken, because I just toggled advanced mode off to check and still had reverse search, but I thought it used to be under advanced mode. Maybe it still is, but I would need to restart to see that.
(9.76 KB 538x133 ClipboardImage.png)
>Used hydrus for years no problem, >today I loaded it up and found pic related on a shit ton of my files. I split my database into 2 drive directories, 25/75 but now hydrus only sees the 2nd hdd use for thumbnails instead of pulling the actual files which are the mass bulk of the db. I told it to use the 2nd hdd db which is over tb but it only registers 17gb of the files in the program. what can I do to fix this?
>search "maniacbox" on Sankaku from within Hydrus while logged in >18 results >search "maniacbox" on the Sankaku website while logged in >104 results It appears that images tagged with loli are blocked in Hydrus somehow, even when logged in. Has this happened for anyone else?
>>15226 25 MB, 9000x9000 >>15233 thank you, will do that
>>15242 Are you cookies set up properly? It works on my machine.
>>15232 Wanted to let you know that I managed to run the command successfully, it's just in the extremely slow process of recovery. >>Error: near line 142: table current_mappings_5 has 2 columns but 1 values were supplied This error cropped up so far, but I hope it can be ignored.
I fucked up on tag parenting and made the intended parent a child and vice versa. Is there a way to invert the parent/child tag relationship? If not how do I remove the status entirely so I can redo it? Looking through the tutorial it looks like there's a delete option on an old version but I'm on 416 and don't see it.
>>15246 I don't necessarily know what I'm talking about, but a few versions ago (probably for months now) Hydrus has had "virtual" tag parents, which for hydrus means tag parents don't exactly behave like actual tags anymore, but rather are calculated. So before if you had a cleavage/boobies tag parent, all images tagged "cleavage" would have an actual "boobies" tag applied. With virtual tag parents, that "boobies" tag is now just calculated as being applied, and upon its parent being removed, it's calculated away. I don't know if version 416 was before or after this change, but still.
>>15231 I made a backup and tried removing the PTR (fully synced) to see what happened. Removing it from the services list took several hours. I don't know how many, I ended up letting it run overnight. After removal, I did a full vacuum and my Hydrus install went from 61GB to just under 28GB (12GB of media according to Mr. Bones). But now my install seems borked. I get a "Service id error in database" when I try to search. I've attached the full error. I tried checking the DB integrity, and it found no errors. I'll keep trying the stuff under database that have to do with fixing it, and post if I can get it working again.
>>15248 A restart fixed it.
(145.17 KB 448x427 hollow.png)
Is there some hidden option I am missing in the gallery downloader that would let it skip errored links/push them back to the end? Any query with even a single link that errors(with a state of "error",the "ignored" ones don't have any problems) is stuck in timeout for ages and even manually going in and setting that url to skip and/or "retry and continue search" on another entry does not save that query from the eternal Requiem it fell into.I basically end up having to start them paused and check the domains for validty before letting it run. I found the option to change the retry timer but that still doesn't help it if the url is completely inaccessable. TL;DR:Hitting the retry timer leaves the gallery downloader query unresponsive.Is there any way to tell it to skip/postpone them instead of retrying?
>>15244 I imported cookies from both chan.sankaku.com and sankaku.com after logging in. Also added my login to the logins manager and it is 'valid'. Searching for 'loli' with both the default downloader and the beta API one return 0 URLs for me. I tried putting a sankaku URL tagged with loli into a URL downloader and got a 500 error.
Is there a currently working solution for twitter downloads?
>>15252 Use the Nitter downloader instead.
>>15247 so you're saying I'll have to do it all by hand?
>>15236 >>15237 Ah, sorry for the confusion! Those paths are I guess actually my laptop's venv that is being replicated virtually in the build. The problem I saw before was for all videos. If this is only two videos, perhaps it is a bug in mpv for now. Your different video resolution probably had a different (maybe non-broken) header, maybe without borked unicode characters or something. I will make sure that error state is handled more gracefully. Thank you for the update. >>15240 Thanks. I agree, I will move that up. I guess I tucked it into the 'file relationships' submenu in some rewrite, and for a while that search code was real slow and bad, so it was probably hidden from non-advanced. I'll correct both, thank you. >>15241 I am very sorry for this trouble. If you look at "C:\Hydrus Network\db\client_files", does that 'f2d' subdirectory actually exist, and does the 2d41... jpg exist inside it? In your other client_files folder on your other drive, do you have an f2d subdirectory there? If you have both, is the one on C empty? I am not sure why the client would get confused about these locations. When you split the database, did you do it manually, outside the program, and then boot and 'repair' the missing locations, or was it done in the database->migrate database dialog? If you look at that dialog now, does it think everything is in its 'ideal' location, or does it think files need to be moved?
>>15242 >>15251 I don't know enough about sank's login system to comment too cleverly here, but it could be there are surplus cookies in hydrus confusing it. If you want to wipe all your hydrus cookies and resync using Hydrus Companion from scratch, check out network->data->review session cookies. A possible further problem can be your User-Agent. I don't know about Sank, but some Cloudflare-similar services match that to your cookies as well. You can try copying your browser's to hydrus under network->data->manage http headers. >>15245 Yes, I think that can be ignored. It is odd though. Maybe the original routine that created that sql file could only pull one value from that row. If the import continued despite the error, you are most likely fine. >>15246 >>15247 >>15254 I am sorry, there is no quick 'invert these pairs' command for parents yet. The good news as >>15247 says is that the tags that seem to have been added were not--they are virtual, and completely undoable--so when you delete the pairs and remake them the correct way around, you won't have a bunch of spare child tags hanging around. To delete a row in manage tag parents dialog, I think if you click 'show all' and then select some rows and hit the delete key, they will be queued for removal. Or you can double-click rows, if that is faster. Also, if you just try to enter the rows as you actually want, it may/should say 'yo, there's a conflict here, do you want to remove all the existing conflicting rows?', and it'll remove them for you. If we are talking thousands of rows here, it would be easiest to export them to txt and then use a script to reverse them, and then reimport. I have always hated the UI for sibling and parent pairs. I don't have time right now to make my dream graph UI, but as you do this job, if you can think up some simple things that would have made your life easier, please let me know. I think better labeling on what the left and right do is probably a good start.
>>15248 >>15249 Thank you for this report. I am sorry for the trouble. Despite being annoying, this was harmless, no need to worry, I think it was a search domain button that was still stuck on 'PTR' tag domain after the service was deleted, and that error is not being handled well. Or it was a service cache issue, with PTR hanging around somewhere despite the deletion. I will fix up the handling here.
>>15250 Sorry, the downloader has some hardcoded patches to catch broken domains and stop spam/retry-loops and the handling is shit atm. I will be cleaning that code up, writing a proper manager for it all, and exposing current domain status with new UI in future. I also plan better 'retry later' tech for individual jobs in a queue. At the moment, each one has to be worked in order until it gets a status other than 'blank', but in future they'll be able to set a 'I had 2 connection errors, I can try again in four hours' kind of thing so complex queues can work what they can and not hold everything else up.
>>15253 Doesn't work. I'm getting constant "out of bandwidth" errors from nitter. It's been that way for days now and I've concluded it's not going to work anymore.
>>15256 >I have always hated the UI for sibling and parent pairs. I don't have time right now to make my dream graph UI, but as you do this job, if you can think up some simple things that would have made your life easier, please let me know. I think better labeling on what the left and right do is probably a good start. To me, my current main complaint actually comes from the introduction of virtual tag parents, and it's not specific to the tag parents window. For a real example, I recently (post-virtual tag parents) started using PixivUtil2 to download from fanbox. In order to sort it (besides importing the PixivUtil2 filename under a namespace), I give the entire batch the namespace "fanbox:[username]". Then I give that namespace the parent "source:fanbox.cc". However, "source:fanbox.cc" only exists as a virtual tag, so I cannot search it and have it appear under suggestions. Even when it's typed-in verbatim, it doesn't show any results at all. Only when you select it does it show you the parent/child pairs you've already made that use it. And, again, it's not specific to this window. For another real example, I use to archive MMD videos in my hydrus (before deciding not to bother anymore, since I never tag them). I use to use the generic tag "mmd", before of course realizing that something that generic will be used by other people. So I manually changed all instances of my personally-tagged "mmd" videos to "mmd (mikumikudance)". But the problem is that since I made them parent tags under the mmd creator's name, there is, again, no suggestion in search for the tag. The only reason I'm capable of searching for that tag is because I've applied it as a normal tag to videos without a creator name.
>>15258 Thanks anyways,I at least have an idea of which domains cause problems so I am fine for now as long as I remember to start the queries with search paused. >>15259 Go to network->downloader components->manage url classes and switch these out for another nitter mirror.You can also duplicate them and change those(as you can see I did). github .com /zedeus/nitter/wiki/Instances Speaking of; Devman,is there any syntax I am missing to add entirely seperate domains into a single url class or is that another one for the to-do list?
>>15261 Thanks. This nitter situation is a pain. I just looked into it properly yesterday and came across that mirror list. Hydrus isn't set up to do manual mirroring very well, but just switching out the url classes for a different domain is a simple solution. I am thinking of just rolling in an alternate downloader for one of those mirrors. Is there one you would recommend? URL Classes can have only one domain. I was just having a conversation two minutes ago with someone regarding FLAANation about how it would be useful to have multiple domains for URL Classes. Then this nitter problem could be solved just with multiple GUGs. I am not sure how difficult this would be to do, but I am thinking about it.
>>15260 Ah, I am sorry, I think there has been some confusion. Virtual tags are supposed to actually be better at appearing in search results. They work basically the same on the front end in how they display, but behind the scenes of all this is now much more neat and logically consistent. The only proviso is that it now needs a bit more CPU time to do that calculation. If you are not seeing parents appearing on your files, and thus not appearing in tag autocomplete results, is there any chance your tag service is not 'syncing' in the background? If you check the tags->sibling/parent sync menu, is your client set to sync in 'normal' time? In the review maintenance window under that menu, are the siblings and parents synced for the tag service you want? Also, once change with sibling and parent virtualisation is that siblings and parents now have more options where they can apply. In the old system, you could only set them to apply either on their own services or to all services, but now, under tags->manage where tag siblings and parents apply, you can set the siblings on 'my tags' to apply to the PTR and similar. If your parents are on 'my tags', but the 'source:fanbox.cc' tags are on PTR, you may need to set up those rules to apply that way. Once your services' siblings and parents are synced, if you are set to sync in 'normal' time, then any changes you make in manage tag parents are supposed to apply within seconds. All parents should appear quickly and comprehensively. If you are completely synced and you think your sibling and parent rules are applied correct, please let me know, and let me know which tag autocompletes are not showing results--I think they should appear in search pages, but not in manage tags--as this could well be a bug. Parents and siblings are supposed to appear in search page search results. EDIT: Shit, I just reread your post. When you say you gave the 'fanbox' namespace the parent 'source:fb', did you add something like 'fanbox:*' as your child? I am very sorry, but namespace-wide parents and siblings are not supported yet. I want to add these, but this whole system proved complicated and I ran out of time.
(54.76 KB 972x829 Untitled.png)
>>15263 I understand the non-bold text was from a misunderstanding of my complaint, but I think I am still focused. Pic related is illustrating no suggestions, despite being typed-in verbatim, and when you select it anyway, it shows where you have used it.
>>15264 >focused confused*
>>15242 Your post finally motivated me to fix my setup for SC, so thanks! Here's what I did, and now I can get all the results: - Add https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders/Sankaku - Follow instructions from https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders/Sankaku#capi-v2-auth - Try your search again, making sure that you're using "Sankaku chan beta tag search" and not the default Sankaku chan tag search"; on the "main" website, some tags are obscured behind a paywall, but not on the API (which is what beta uses)
>>15264 >>15265 Ah, I think the thing here is that now in the virtual system, I now have two ways of looking at tags: 1) 'Display'. This is what you see when you are in a 'search/read' place like a search page. You get nice parents and siblings here, with correct counts. 2) 'Storage'. This is what you see when you are in a 'edit/write' place like manage tags or manage parents/siblings. Here you are looking at tags as they actually are stored, without the 'virtual' tags calculated. I see that you do not have a nice result there in your autocomplete list, because that tag doesn't exist in storage. I think I need to rework how some of this displays. I told you I hated the UI, ha ha. In any case, as soon as you enter a tag into the left or the right of 'manage tag parents/siblings', I display all the current pairs that include any of those tags to help you see what you are editing. This was easier than always having 'show all pairs' on all the time, which was laggy. Your parents there look great. All the 'fanbox:' are set up to get 'source:fanbox'. If you open up a normal search page and type 'fanbox', you should get a 'source:fanbox' result for it, if you enter that tag, all those files with a 'fanbox:blah' tag should come back. I am sorry for the confusion here. This is messy and not well documented. I will queue up a job to look through all the help here and see if I can add some tooltips and so on. I hope next week to have an improvement to 'manage tags' to show parents and siblings better, even though they don't actually 'exist' in storage.
(30.58 KB 300x1279 Untitled.png)
>>15267 >Your parents there look great. All the 'fanbox:' are set up to get 'source:fanbox'. Thanks, lol. I feel really behind the curb, because the only site I've ripped through hydrus is pixiv. Everything else I use a different tool and import. I think the filenames generated are useful, but I worry that the namespaces I've occupied doing this would overlap with the ones hydrus would use, if I ever started using hydrus to rip instead. Also, you can see what I did with the "exhentai page0" namespace that I have a weird complaint I'm not sure there could ever be anything done about. For exhentai specifically, it's irresponsible to import the pages under a generic "page" namespace, because the images could come from literally any source. That was sort of my reasoning for only using hydrus to rip from pixiv, because if any other site uses a "page" namespace, it could potentially overlap. I wish I could somehow use a "page" namespace specific to any import, without worry. For exhentai galleries where I have to use "exhentai page" and "exhentai page0" namespaces, it's only to avoid any images appearing in both sullying the page order of either. But it would be convenient to on import choose "this namespace is specific to this import." So regardless of the namespace being generic, it manages to be isolated anyway. This was really hard to explain, and I'm sure I could've gotten the point across more tersely, but still.
>>15266 Hi, i'm new with hydrus and don't really understand the instructions you posted. I'm mainly trying to use hydrus as an offline sankaku, basically organizing my files by tag. I don't quite get the https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders/Sankaku#capi-v2-auth part of the instruction. What's the beta page? What should i do? It assumes i know what i'm doing but i really don't.
>>15266 >CAPI-V2 Auth This must be new. After following those steps, I seem to be able to get files with tags like loli again. It seems to work with the regular SC downloader too. Thanks mate. >>15269 Here's what I did (firefox-based browser): >go to beta.sankakucomplex.com, navigate to log in screen >hit F12 and click the "network" tab >log in while the dev tools are still up >after logging in, look for a request from capi-v2.sankakucomplex.com (I had 4) >click on one, and look for "Authorization" on the right hand panel. Check a different one if it isn't there. >Copy the text after Authorization and put it in Hydrus>network>data>http headers Hopefully this helps.
>>15270 can i use hydrus to tag all the videos i already have stored? i don't quite understand.
One of the major issues with the PTR, besides massive tagging errors slipping through (at one point "rating:18+" got "guro" as a parent, seriously wtf) and outright vandalism, is the enormous amount of borked automated tag imports in the form of "some tag" becoming "some_tag". I was thinking about writing a scripts that helps quickly fixing such errors, mostly by detecting whether for any "some_tag" containing underscores there is already the corresponding "some tag": would that be a good idea or is there some issue I didn't notice?
Is there any chance of a future option to compress files on import and decompress them on opening and export? Maybe with some level options for compression level/performance tradeoff? It's also possible this is already a thing in Hydrus and I just don't know about it.
>>15271 Using the method described in >>15266 and >>15270 allows you to download more stuff (these are specifically aimed at Sakaku Complex); importing existing files from your hard drive to tag them can be done in the file menu, with "import files".
>>15267 Hey this is stupid, because it seems like I'm making an excuse for no reason, since I missed nothing. But I had this thread open in another tab, and it stopped updating replies after my earlier reply. I would've made this post a day earlier had I realized sooner. I didn't understand what you were saying at first, but I just tried what you said here: >If you open up a normal search page and type 'fanbox', you should get a 'source:fanbox' result for it, if you enter that tag, all those files with a 'fanbox:blah' tag should come back. And it's true. The only thing is, I have never been able to experience this, because, well firstly I never knew it would work that way on that page, but more than that, I have only ever searched from a tagging window since I started using Hydrus. You said you hope to amend the way this works anyway, which was also kinda why I didn't try too hard to understand what you were explaining at first. But I keep hydrus open 24/7, and the active tab is always highlighting an image. So I always just hit f3 to open the tag window, then I search any tag, and right click > "open a new search page for x". The behaviour of the tag window was what I thought the only behaviour for searching ever was. I also just have way too many tabs open at any given time, so ctrl + T > up > up is just always more keypresses than just hitting F3 on an image that is already highlighted on my current page. I assume if there is space on the tab bar that you can merely double click to open a new tab. But then it will probably still just open the same "new page" window to specify what tab you're trying to open. I think everyone's workflow is probably the same as mine. In hindsight, I have searched from a normal tab, in order to disable "searching immediately", to search multiple tags at once without loading a huge tag during the downtime I'm typing another. Also when my client just came back from lagging, or something, so search results weren't showing up yet, I'll search it anyway using a wildcard search. But I never managed to notice that virtual tags were counted there.
(119.24 KB 502x497 hydrus.png)
Is there a way for hydrus to find files that have been imported to delete them from their source location? I imported my files without checking that one checkbox and now want to remove them, can I only do it by hand?
>shadow (1) (will display as character:shadow the hedgehog) Who the fuck thought this was a good idea?
(16.60 KB 225x499 Untitled.png)
>>15276 Just reimport them,it won't make duplicates.
>>15276 No, plus if you reimport files that are already in the DB it won't delete them even if you set it to delete on import. The best way to go about it would probably be reimporting, just to make sure all the files are already in hydrus, them then deleting the folder.
I had a good couple of weeks. As well as fixing a bunch of bugs and improving quality of life, I was able to finish an important backend update to the taglists used across the program. Implied parents and siblings are now shown nicely in the manage tags dialog. There is a whole bunch to test, so the release may be late tomorrow.
>>15273 >Is there any chance of a future option to compress files on import and decompress them on opening and export? JPEG XL support was discussed before and IIRC Hydrus dev said he would enable support for the format and possibly look at how to deal with collections recompressed to JPEG XL once the relevant python libs support it. Almost anything else won't do much. Even if you enable lz4 or something else that is fast enough for this use (which you possibly can anyhow do in your filesystem, or at the block layer or something without needing hydrus), it just won't save much space on current media formats.
>>15281 Always appreciated.
A few feature requests: A shortcut to temporarily show/hide the progress bar More customizable image scaling settings, e.g. enable nearest nearest neighbor scaling at >=500% zoom for tiny pixel art or faster zooming When using mpv to view animations or videos, allow changing scaling settings (I know it's possible in standalone mpv to change them on the fly) Blend videos and animations with the background color or a custom color where possible, whether using the built-in viewer or mpv I know mpv has support for blending with the set background color and rendering to a framebuffer with an alpha channel, whichever's simpler I ordered them by my very rough estimate of least to most complicated. Not sure if there's a better place for anonymous feature requests, but I hope you can add these to your to-do list.
>>15284 >More customizable image scaling settings I would like an option to only scale down, and not up. Scaling up usually looks bad, especially pixel art.
>>15285 >Scaling up usually looks bad, especially pixel art. Not him, but "especially" pixel art? This is complete nonsense. https://pawoo.net/@ohgaki/1855549 Pic related was upscaled 4x using https://lospec.com/pixel-art-scaler/, because waifu2x kept trying to reduce noise, even when I tried to disable the setting. Even in my image viewer JPEGView, I can hit "F3" to scale without noise reduction, getting the same effect. I think you mean the standard noise reduction on scaling looks bad for pixel art, in which case you should've said you wished for the option to disable it.
>>15285 There is an option for that, in options -> media -> filetype: image (in the table) -> if the media is smaller than the media viewer canvas: show at 100%
I know I read about this on the old board (maybe in the suggestions thread?), but I'm wondering if there is an option to use something like the archive/delete filter, but for a tag? Something like "is this tag applicable to this picture?" kind of thing, where the would-be "archive" button is "yes" and the would-be "trash" button is "no"? I download and add a lot of media in bulk, and I'd love to be able to quickly tag some of the pictures in there with a tag, but with the picture in full view (where grid view can be "too small" and inconclusive).
>>15288 You can configure the grid view to be larger. Or you can just filter by whatever tag you want + recent time and then go through the enlarged/full screen images as usual.
I wanted to ask: We currently have private and PTR public tags in the UI and probably in some database tables. But we could use external tools like hydrus-dd. Is there a way to make it use its own tag table so it does not affect either the private or the PTR public tags?
Sorry lads, the end of this week was a crunch and I ran out of time, I will post the release now and catch up on this thread properly on Saturday.
https://www.youtube.com/watch?v=RNPnBVHSeZc windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v428/Hydrus.Network.428.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v428/Hydrus.Network.428.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v428/Hydrus.Network.428.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v428/Hydrus.Network.428.-.Linux.-.Executable.tar.gz I had a good couple weeks working on the taglist code and some other jobs. If you are on Windows and use the 'extract' release, you may want to do a 'clean' install this week. Extra notes below. taglists So, I took some time to make taglists work a lot cleaner behind the scenes and support more types of data. A heap of code is cleaner, and various small logical problems related to menus are fixed. The tag right-click menu is also more compact, quicker to see and do what you want. The main benefits though are in the manage tags dialog. Now, the '(will display as xxx)' sibling suffix colours in the correct namespace for the sibling, and parents 'hang' underneath all tags in all the lists. It is now much easier to see why a parent or sibling is appearing for a file. This is a first attempt. I really like how these basically work, but it can get a bit busy with many tags. With the cleaner code, it will be much easier to expand in future. I expect to add 'expand/collapse parents' settings and more sorts, and maybe shade parents a bit transparent, in the coming weeks. Please let me know how it works for you IRL and I'll keep working. the rest The main nitter site seems to be overloaded. They have a bunch of mirrors listed here: https://github.com/zedeus/nitter/wiki/Instances I picked two roughly at random and added new downloaders for them. If you have Nitter subs, please move their 'sources' over, and they should start working again (they might need to do a bit of 'resync' and will complain about file limits being hit since the URLs are different, but give them time). If you would rather use another mirror, feel free to duplicate your own downloaders as well. Thanks to a user who helped here with some fixed-up parsers. I gave the recently borked grouped 'status' sort in thread watchers and downloader pages another go, and I improved the reporting there overall. The 'working' status shouldn't flicker on and off as much, there is a new 'pending' status for downloaders waiting for a work slot, and the 'file status' icon column in galleries now shows the 'stop' symbol when files are all done. The menu entry to 'open similar-looking files' is now further up on thumbnails' 'open' submenus. The duplicate filter has its navigation buttons on the right-hand hover window rearranged a bit. It is silly to have both 'previous' and 'next' when there are only two files, so I merged them. You can also set 'view next' as a separate shortcut for the duplicate filter, if you want to map 'flip file' to something else just for the filter. windows clean install If you use the Windows installer, do not worry, these issues are fixed automatically for you from now on. I updated to a new dev machine this week. Some libraries were updated, and there is now a dll conflict, where a dll from an older version is interfering with a new one. As it happens, the library that fails to load is one I made optional this week, so it doesn't seem to actually stop you from booting the client, but it will stop you from running the Client API in https if you never did it before (the library does ssl certificate generation). It is good to be clean, so if you extract the Windows release, you may want to follow this guide this week: https://hydrusnetwork.github.io/hydrus/help/getting_started_installing.html#clean_installs Remember: Don't delete your db folder. full list - interesting taglist changes: - taglists work way better behind the scenes - when siblings display with the '(will display as xxx)' suffix, this text is now coloured by the correct namespace! - parents now show in 'manage tags dialog' taglists! they show up just like in a write/edit tag autocomplete results list - the tag right-click menu has had a pass. 'copy' is now at the top, the 'siblings and parents' menu is split into 'siblings' and 'parents' with counts on the top menu label and the submenus for each merged, and the 'open in new page' commands are tucked into an 'open' submenu. the menu is typically much tighter than before - when you hit 'select files with these tags' from a taglist, the thumbgrid now takes keyboard focus if you want to hit F7 or whatever - custom tag presentation (_options->tag presentation_, when you set to always hide namespaces or use custom namespace separator in read/search views) is more reliable across the program. it isn't perfect yet, but I'll keep working - a heap of taglist code has been cleaned up. some weird logical issues should be better - now the code is nicer to work with, I am interested in feedback on how to further improve display and workflows here
- the rest: - added two mirrors for nitter, whose main site is failing due to load. I added them randomly from the page here: https://github.com/zedeus/nitter/wiki/Instances . if you have nitter subs, please move their download source to one of the mirrors or set up your own url classes to other mirror addresses. thanks to a user for providing other parser fixes here - gallery download pages now show the 'stop' character in the small file column when the files are done - gallery download pages now report their 'working' status without flicker, and they report 'pending' when waiting for a download slot (this situation is a legacy hardcoded bottleneck that has been confusing) - thread watchers also now have the concept of 'pending', and also report when they are next checking - improved the new grouped status sort on gallery downloader and watcher pages. the ascending order is now DONE, working, pending, checking later (for watchers), paused - the network request delay after a system resume is now editable under the new options->system panel. default is 15 seconds - the 'wait on files too' option is moved from 'files and trash' to this panel - when the 'just woke' status is active, you now get a little popup with a cancel button to override it - 'open similar-looking files' thumbnail menu entry is moved up from file relationships to the 'open' menu - the duplicate filter right-hand hover window no longer has both 'previous' and 'next' buttons, since they both act as 'flip', and the merged button is moved down, made bigger, and has a new icon - added 'view next' to the duplicate filter shortcut set, so you can set a custom 'flip between pair' mapping just for that filter - thanks to a user helping me out, I was able to figure out a set of lookups in the sibling/parent system that were performing unacceptably slow for some users. this was due to common older versions of sqlite that could not optimise a join with a multi-index OR expression. these queries are now simpler and should perform well for all clients. if your autocomplete results from a search page with thumbs were achingly slow, let me know how they work now! - the hydrus url normalisation code now treats '+' more carefully. search queries like 6+girls should now work correctly on their own on sites where '+' is used as a tag separator. they no longer have to be mixed with other tags to work - . - small/specific stuff: - the similar files maintenance search on shutdown now reports file progress every 10 files and initialises on 0. it also has faster startup time in all cases - when a service is deleted, all currently open file pages will check their current file and tag domains and update to nicer defaults if they were pointed at the now-missing services - improved missing service error handling for file searches in general--this can still hit an export folder pointed at a missing service - improved missing service error handling for tag autocomplete searches, just in case there are still some holes here - fixed a couple small things in the running from source help and added a bit about Visual Studio Build Tools on Windows - PyOpenSSL is now optional. it is only needed to generate the crt/key files for https hosting. if you try to boot the server or run the client api in https without the files and without the module available to generate new ones, you now get a nice error. the availability of this library is now in the client's about window - the mpv player will no longer throw ugly errors when you try to seek on a file that its API interface cannot support - loading a file in the media viewer no longer waits on the file system lock on the main thread (it was, very briefly), so the UI won't hang if you click a thumb just after waking up or while a big file job is going on - the 'just woke' code is a little cleaner all around - the user-made downloader repository link is now more obvious on Lain's import dialog - an old hardcoded url class sorting preference that meant gallery urls would be matched against urls before post, and post before file, is now eliminated. url classes are now just preferenced by number of path components, then how many parameters, then by example url length, with higher numbers matching first (the aim is that the more 'specific' and complicated a url class, the earlier it should attempt to match) - updated some of the labelling in manage tag siblings and parents - when you search autocomplete tags with short inputs, they do not currently give all 'collapsed' matching results, so an input of 'a' or '/a/' does not give the '/a/' tag. this is an artifact of the new search cache. after looking at the new code, there is no way I can currently provide these results efficiently. I tested the best I could figure out, but it would have added 20-200ms lag on all PTR searches, so instead I have made a plan to resurrect an old cache in a more efficient way. please bear with me on this problem - tag searches that only include unusual characters like ? or & are now supported without having to lead the query with an asterisk. they will be slower than normal text search - fixed a bug in the 'add tags before import' dialog for local imports where deleting a 'quick namespace' was not updating the tag list above - . - windows clean install: - I moved to a new windows dev machine this week and a bunch of libraries were updated. I do not believe the update on Windows _needs_ a clean install this week, as a new dll conflict actually hits the coincidentally now-optional PyOpenSSL, but it is worth doing if you want to start using the Client API soon, and it has been a while, so let's be nice and clean. if you extract the release on Windows, please check out this guide: https://hydrusnetwork.github.io/hydrus/help/getting_started_installing.html#clean_installs - the Windows installer has been updated to remove many old files. it should now do clever clean installs every week, you have nothing to worry about!™
- boring db breakup: - the local tags cache, which caches tags for your commonly-accessed hard drive files, is now spun off to its own module - on invalid tag repair, the new master tags module and local tags cache are now better about forgetting broken tags - the main service store is spun off to its own module. several instances of service creation, deletion, update and basic fetching are merged and cleaned here. should improve a couple of logical edge cases with update and reset - . - boring taglist changes: - taglists no longer manage text and predicates, but a generalised item class that now handles all text/tag/predicate generation - taglist items can occupy more than one row. all position index calculations are now separate from logical index calculations in selection, sizing, sorting, display, and navigation - all taglist items can present multiple colours per row, like OR predicates - items are responsible for sibling and parent presentation, decoupling a heap of list responsibility mess - tag filter and tag colour lists are now a separate type handled by their own item types - subordinate parent predicates (as previously shown just in write/edit autocomplete result lists) are now part of multi-row items. previously they were 'quiet' rows with special rules that hung beneath the real result. some related selection/publish logic is a bit cleaner now - string tag items are now aware of their parents and so can present them just like autocomplete results in write/edit contexts - the main taglist content update routines have significantly reduced overhead. the various expansions this week add some, so we'll see how this all shakes out - the asynchronous sibling/parent update routine that populates sibling and parent data for certain lists is smarter and saves more work when data is cached - old borked out selection/hitting-skipping code that jumped over labels and parents is now removed - 'show siblings and parents' behaviour is more unified now. basically they don't show in read/search, but do in write/edit - a heap of bad old taglist code has been deleted or cleaned up next week This was a big couple of weeks. Setting up the new dev machine--I replaced my six year old HP office computer with a nice mini-pc with an SSD--worked out great, but there were some headaches as always. The taglist work was a lot too. I'll take next week a little easier, just working misc small jobs.
(207.89 KB 764x540 Check'em - (01).jpg)
>>15285 Digits confirm.
>>15290 Just create another tag service and configure hydrus-dd to use it. Or what I like to do is run hydrus-dd as a file lookup server and manually add tags, since it can occasionally be wildly inaccurate.
>>15286 If you can manage to scale up in way such that each 'pixel' is a perfect square, it looks good. Anything else looks bad. >>15287 Shit, guess I should look in the options first.
>>15290 >hydrus-dd This seems cool but I could never get it working. It looks like tensorflow (or something it depends on) is fucked up and can't read its own config files.
I'd like to preserve the current folder structures but I'd like to have an easy way to tag my local images and videos. Can hydrus work like this, without duplicating the files?
>>15299 Oh, the FAQ says no.
Is the in-client backup gone, or am I just blind? The documentation says database->set up a database backup location, but no such thing exists. This is on 427.
>>15301 I haven't updated to the latest version yet (I'm on 427) but I have that option there. It's the second one in the list for me. My database is only across one hard drive though, maybe that's why?
I'm trying to run the Tumblr user lookup and it isn't grabbing images even from posts that have them. Is the parser outdated or something? I'm mostly trying to grab reblogs (It's a blog for an event where multiple artists contributed) but I think there were a few originally posted images it didn't grab either.
Does anyone know what exactly Sankaku's request limits are? Trying to tune my bandwidth usage to download at a decent speed while also having spare bandwidth for manual browsing of the site.
>>15304 At least for the beta sankaku parser on the cuddlebear repo, the bandwidth is 1 request per 5 seconds and 80 requests per 7 minutes. They don't seem to care how fast or how much you download as long as you space out the requests. I haven't been blocked by them for downloading huge videos back to back, but I have been blocked once for forcing too many requests.
Love Hydrus. Recently started putting in videos in it too. Is there a way for me to put video series in the correct order? If there is I definitely will move all my videos to hydrus.
Alright I have just imported the filename for sorting videos. Creates a lot of tag bloat but it just werks.
(8.63 KB 233x208 Untitled.png)
>>15302 I do have my thumbnails on an ssd and my files on a different drive, so maybe that's the case. I'll use the manual backup I suppose, thanks for the sanity check!
>>15268 Thanks. Yeah, I don't really like how 'page:' tag has worked out. My next push on this is going to be when I do the 'file relationships' and 'cbr/cbz support' expansion. I am going to try to associate files with each other, in groups that support ordering, rather than trying to store sort in tags. As you say, page tags can overlap in different contexts, and they are hell to edit if one is wrong. If instead I can store 'this file goes after this file', and present manga chapters and mini 4-page comic-things to you as a virtual 'book' like a cbr, I hope that'll be a lot easier to use. It is tricky though, so I'll appreciate your feedback as we continue here. >>15272 Yeah. I am supposed to be fitting in some nicer janitor tools at the moment to help improve sibling and parent decision-making, although finding the time is tricky. I have definitely solved the 'make adding shit to the PTR easy' problem, and now we have a billion mappings, it is definitely time for better tag management. If you can figure out some nice scripts (Client API will help a lot here) to improve tags, please do. All feedback would be useful. At some point I'll be doing another push on tag siblings. As well as broad 'rename all artist: to creator:' namespace siblings, I'd like to just add a checkbox for 'replace all underscores to be spaces m8'. I had considered making this an official hydrus rule, just to remove all underscores from valid tags, but it turns out some users like them, so I'll have to make it optional. >>15273 >>15282 Yes, I am considering this. It will always be an optional thing. Compression and other sorts of file conversions are tricky for several reasons, including it messes up our hash-based mappings. But there are good reasons for it, so it will come. As >>15282 says, most media files are well compressed by design. Most tech is lucky to save 10% lossless, but JPEG XL seems like a winner in future. I know a guy with a nice machine who turned on NTFS compression for his database files lmao. It actually seems to work well, and saves about 40% space or so.
>>15275 Ah, cool, thank you for explaining. It sounds like I could do with improving my 'getting started' help to emphasize how page searching works. If you skipped the current help, or you found it confusing, can you explain how I could change it so you would have noticed this? I've had trouble figuring out a balance between making the help comprehensive but also simple and easy to get started. I have a crippling addiction to wordy sentences and effort-posting. If it helps, there's a simpler guide made by another user here: https://github.com/Zweibach/text/blob/master/Hydrus/Hydrus%20Help%20Docs/00_tableOfContents.md Also, have you noticed the 'system' tags that hang below an empty page search? They are really powerful and let you search things other than tags like filesize or your inbox/archive. BTW: If you middle-click a tag, it should also do that 'open in a new page' thing.
>>15280 I have had this report about 'on "already in db", the "delete from source" function does not work' a couple times, but I am afraid I am having trouble reproducing the problem. It seems to work ok for me. It doesn't delete if the file has status of "previously deleted". Can you say any more about the situation where this has not worked for you? I'd like to expand hard-drive file import options to work more like import folders, with different actions for each ending import status, but I also want to explore what is actually going on here. >>15276 >>15279 Yes, I believe just re-importing with the checkbox hit will delete if it was 'already in db', but as above, it seems I need to look at this. Hydrus does not remember where files come from at the moment.
>>15277 >>15278 This is some old bullshit, my fault from when I was running the PTR. I accidentally approved that bad petition. I guess the user didn't know or didn't think about how siblings actually would work, and I must have misclicked. It should be corrected now, so if you aren't synced, please wait a bit, the fix will get processed soon. The good news is siblings are undoable now, so future problems like this won't cause so much harm. Back in the day though, ha ha, I accidentally spammed shadow the hedgehog all over the place. If you are fully synced and still see it, let me know.
>>15284 Thank you for your feedback. When you say a shortcut to hide the progress bar, do you mean the popup windows that come up like a toaster on the bottom-right of the main window? A shortcut to flip the collapse/expand button there is a great idea. I agree about better image scaling options. All that code is a shitshow atm unfortunately, I am pending a big rewrite of the whole rendering pipeline. Please bear with me, I will make big zooms non-laggy and clean all the code, and then we can explore nicer and more reliable zoom quality tech. I wish there were a nice way to say 'zoom pixel art like this, and shaded images like this', but I am still vacillating on nice ways to do it. Probably just a flippable zoom mode in the viewer is the least headache for everyone. MPV is another one I want to revisit. It has all this customisation, but my ability to access that stuff is often a bit arcane, and I am not an experienced MPV user, so I only set some basic 'good' settings as default. If you know what you want, you can override your MPV settings permanently by editing the mpv.conf that should be in your install_dir/db folder. Transparency in videos is something I will get into once the main renderer pipeline is cleaned up. I want it though. This place is great for feature requests, thank you. I'm happy to talk any place.
>>15288 >>15289 Custom filters where you have the nice 'go forward/go back/complete filter' are something I am planning. I'll be generalising the archive/delete filter code and then ultimately let you hang whatever media commands you want on the final commit. For now, your best option is the shortcuts system. You can either emulate a filter by just also using previous/next shortcuts, or you can fold it into your normal archive/delete filter. You can set shortcuts for tags to the 'media viewer' shortcut set, and if you are feeling advanced, you can turn on help->advanced mode and then get access to 'custom' shortcut sets under file->shortcuts. These custom sets you can turn on and off in the media viewer with the little 'keyboard' icon on the top hover. I know some a guy who has a custom shortcut set for about ten or fifteen tags, and he does a clever archive/delete filter + tags while sitting on his couch using a tablet doing VNC-something with a touch button overlay plugged into autohotkeys. It is all jank as shit, but if you play around with it, you can set up some common tags to the number keys or similar.
>>15290 >>15296 Yeah, hit services->manage services and add a new 'local tag service'. That'll work exactly like 'my tags', but be its own separate partition, and its own separate tab or panel on all tag edit/import UI. Pipe your hydrus-dd into there, and it'll be its own thing. >>15299 >>15300 Yep, sorry, Hydrus makes a strict rule about getting away from old file and folder systems. If you like, give hydrus a go for a couple of weeks without deleting from your old structure to see how it works and see how often you miss accessing things the old way. I know several people who were skeptical but found they liked it once they tried it. If you don't like it, no worries, you can delete hydrus with nothing lost. >>15301 >>15302 >>15308 Yes, sorry, once you migrate the database, backup gets more complicated, so I disable my simple routine. I can't do a better job here than the third-party software out there, so please set up your own, as here: https://hydrusnetwork.github.io/hydrus/help/getting_started_installing.html#backing_up I have a job to clarify this across the UI and help, sorry for the confusion.
Is there something instead of favorites where I can keep track of my organizational tags? There are certain tags I only use once or twice to label a specific file to have an unique combination of tags but there are tags I use all the time and are very important but they're already too numerous for me to keep track of them and get buried under the thousands of one time tags. It's about 100 important tags so I also can't put them under favorites as that would defeat the point of having favorites, it's just too much.
>>15303 I am afraid that parser has never been great. It can't grab images from any 'custom' post, so if the user adds the image as a text post or uses third-party posting software that makes that sort of 'html' post that happens to include images somewhere, the hydrus parser can't figure out what it should pull. Only simple 'image' posts are supported. >>15306 >>15307 File relationships and multi-file grouping just isn't very good yet, I am afraid. I hope to do a big push on it later this year. For now, you can play around with 'page:'-style numbered tags and namespace sorting, but they have never been very smooth to work with. It is best if your files actually have a common number in their filename, like 'artist - series - 01 - title.mp4', and then if you get some experience with regex, you can pull that page:01, episode:01, whatever:01 from the filename right when you import, for hundreds of files at once.
>>15316 I have long thought of having some sort of larger 'tag cloud' or bigger list of manual or auto-populated 'best/favourite/most popular' tags, and I have it on the 'big jobs' poll, but it has unfortunately never proved very popular. I know a guy who does a ton of personal tag organisation, including rating by tag. He does it through clever namespaces, and then sets his autocomplete options under tags->manage tag display and search so his local tag service that stores all those tags can search namespaces (giving 'character:samus aran' even though you only typed in 'char'). Then he sets his search page to that local tag domain (instead of 'all known tags'), and he can search his relatively small personal shit quick and easy, typing namespaces to narrow down, without it getting lost in the PTR's gubbins.
Hydrus dev I just updated from 427 to 428, and on boot I'm getting tons of errors: TypeError 'NoneType' object cannot be interpreted as an integer TypeError 'NoneType' object cannot be interpreted as an integer File "hydrus\client\gui\lists\ClientGUIListBoxes.py", line 1937, in paintEvent self._parent._Redraw( painter ) File "hydrus\client\gui\lists\ClientGUIListBoxes.py", line 1562, in _Redraw for logical_index in range( first_visible_logical_index, last_visible_logical_index + 1 ): It's made 33 popups of this so far in like 5 minutes and it keeps going. I'm on Windows 7.
Hydrus is amazing, nothing compares, thanks for making it. Feature request: ability to increment/decrement ratings, instead of setting them to a given number.
>>15310 I've been using hydrus for a few years, so I haven't looked at the help at all since then. I don't even remember taking anything away from the help tbh. Not to out of context discredit the entire thing. >Also, have you noticed the 'system' tags that hang below an empty page search? They are really powerful and let you search things other than tags like filesize or your inbox/archive. >BTW: If you middle-click a tag, it should also do that 'open in a new page' thing. I never knew about the middle click thing, but I'm on a laptop, so I can't middle click anyway. And I have seen the system tags before, but I haven't really used them or checked what they do except the one that lets you filter by filetype, which I've used to filter video. Which by the way, good call on changing "mime" to "filetype". I had never seen the word "mime" before, so I just never clicked it for the first few months I used hydrus. >If you skipped the current help, or you found it confusing, can you explain how I could change it so you would have noticed this? Maybe a line about it in the "manage tag parents" window, because otherwise, again, I would only ever search from a tagging window, so I would never notice on my own. It's kind of brutal still that the tagging window doesn't have the same suggestions as an empty search page. It will always be more convenient (to me) to search from a tagging window, so in a way, it doesn't even matter that much. >>15309 >I am going to try to associate files with each other, in groups that support ordering, rather than trying to store sort in tags. As you say, page tags can overlap in different contexts, and they are hell to edit if one is wrong. If instead I can store 'this file goes after this file', and present manga chapters and mini 4-page comic-things to you as a virtual 'book' like a cbr, I hope that'll be a lot easier to use. This would be great. As it is if you just tag 10 images page:1, page:2 etc, there is no way to just freely insert another image somewhere between. It's also just a pain in the ass, anyway, to give tags for order rather than just visually sort. This was always one of my main gripes with hydrus. If you do add this, though, I hope you can make it work with infinite scope, because I would probably never tag again- I would only sort an image relative to another image. >>15315 >Yep, sorry, Hydrus makes a strict rule about getting away from old file and folder systems. I mean, this is kind of discouraging. Not that guy, but I thought you basically just said in different wording that you would add infrastructure to support this. It obviously won't (and shouldn't) be displayed nor behave the same way as a folder, but I imagined it would largely be the same, save for filenames being discarded for a pure sort independent of tags.
>>15319 Thank you for this report, I am very sorry for the trouble. I was just talking to another user about it, apparently it could be related to 'select files with this tag'? Did the popups appear out of the blue, or after you did an action? Does restarting the client stop them, or do they come back?
>>15322 I only booted the program and didn't do anything, I hadn't even clicked an image yet. When I sit there doing nothing, I watch it produce another error. And it keeps producing more. But I am able to make it produce the error every time I open/ close the "manage tags" dialogue. The program takes a long time to shut down/ boot for me, and it's really, really unresponsive for the first half hour or so, so I can restart if you need the context, but if there's anything else I can do before restarting, I'll do that first, if I can.
>>15322 >>15323 Sorry, I want to amend my previous post. >But I am able to make it produce the error every time I open/ close the "manage tags" dialogue. Rather, after taking a better look at it, what happens is when I open the "manage tags" dialogue, it produces another error. Then a moment later, the "recent" tags list displays the parent relationship of each tag where it has one, which produces another error.
>>15320 I am glad you like it! You should be able to set a shortcut command for that under the 'media actions' set under file->shortcuts. Make a new shortcut, and change the 'simple command' bit to 'numerical rating increment/decrement' and you should be set. Let me know if you have any trouble with it.
>>15323 >>15324 I am sorry. I found the code where this is happening, but I am not yet sure why it is. It seems like it is trying to draw tags on a list that doesn't exist yet or similar. It is possible, if your client is running very slow (do you have a slightly older computer, or is the slow boot something that just started this week?), that some 'draw' events are occurring in an odd order, either on initialisation or as those parent tags pop in. I will wrap my tag rendering code in additional safety checks for 429. Since this is affecting you so bad, I think I will recommend you roll back to 427 for a week. This is best done by rolling back to a backup, if you made one just before updating. If you don't have a backup, this week had no huge changes to the database, so I believe it is possible to just run 427 on a 428 database. Just install 427 like you normally would any other release. If you are on Windows, you will likely have to do a 'clean' install as my 428 release post talks about. If you don't want to mess around with rolling back, please wait for 429. I am confident I can fix this problem. It is caused by those taglists, so if you discover that an empty taglist, or one that is exactly the length of the list widget, or one that exceeds it is the problem, please hobble along with whatever you figure out can work until then. If you have a ton of open pages, it may be you'll get one error for every open page and then it will stop.
>>15321 >Yep, sorry, Hydrus makes a strict rule about getting away from old file and folder systems. >I mean, this is kind of discouraging. Not that guy, but I thought you basically just said in different wording that you would add infrastructure to support this. It obviously won't (and shouldn't) be displayed nor behave the same way as a folder, but I imagined it would largely be the same, save for filenames being discarded for a pure sort independent of tags. You can now store hydrus files in multiple different locations, if that is what you mean. I just can't support tracking of files with "/my_manga/berserk/chapter 600/page226.jpg" user locations. My FAQ that he was talking about is here. I explain there my feelings on filenames in general and human-readable locations when you get to 100,000+ files: https://hydrusnetwork.github.io/hydrus/help/faq.html If you are interested in spreading out your hydrus media over several hard drives, check out: https://hydrusnetwork.github.io/hydrus/help/database_migration.html
>>15325 Oh it already exists, that's great. Nevermind then.
>>15326 >It is possible, if your client is running very slow (do you have a slightly older computer, or is the slow boot something that just started this week?), that some 'draw' events are occurring in an odd order, either on initialisation or as those parent tags pop in. My computer is not bad. It's a laptop with an i7 and 32gb ram. I'm currently at 10% CPU usage, and 76% ram usage. My HDD is only 5400RPM though. >It is caused by those taglists, so if you discover that an empty taglist, or one that is exactly the length of the list widget, or one that exceeds it is the problem, please hobble along with whatever you figure out can work until then. If you have a ton of open pages, it may be you'll get one error for every open page and then it will stop. I currently have 155 pages open, but I just hit F3 enough times to make the error count exceed 160, so I don't understand. My list widget length is actually shorter than the max that can be displayed, because I upgraded to a larger screen res at some point, but never updated that list to make use of the extra space. Even after it updates to display the parent relationships, it doesn't fill out the whole widget. Because I still don't understand what you mean about the error potentially only happening for every open page, I guess I'll keep the program open for a while, but I will probably roll back, I guess. About backups, I didn't do a manual one, so unless there's an automatic one, I have none. If these errors don't hinder usage in any way, I might just minimize it like I always do and keep using the program until the next release anyway.
>>15327 Oh, sorry, I didn't read the full context of the conversation. When he said "I'd like to preserve the current folder structures", I thought he meant he was fine with the files being in hydrus, but simply wanted the folder/subfolder sort to be preserved within hydrus. Which I was trying to say, was behaviour I wanted without hydrus- a sorting without tags. He did explicitly say in the next line "can hydrus work like this, without duplicating the files", so that was my bad not realizing you were responding to that.
>>15330 behaviour I wanted within* hydrus
>>15326 >>15329 Sorry, I forgot to add that, no, the slow boot has been happening for me forever. It's probably my 1.2m session weight from my gallery tab.
>>15329 >>15332 Sorry, by 'one per page', I meant that if you are getting the popups when you just sit there and don't do anything, it could be that each taglist in the program is going to make that error once, and since there is one per page, it may be each list in each page will make it once and then it'll all shut up until you open the manage tags dialog. I am not sure how troublesome the errors actually are. On the surface I think they are harmless, but if they build up enough, they may crash the program since they are happening inside a 'paint' event, which is a delicate time. The 1.2m session size is definitely the lag. If you are interested in having a faster client all around, give all those files a processing tag like 'my sub files' on your 'my tags' domain, and then kill those pages. You can always summon that bigass group of files later in a search. Reducing the overall page count would also help. You do have a nice machine, so I'd say 150,000 file session is probably a good estimate limit before things start to get CPU laggy.
>>15333 >The 1.2m session size is definitely the lag. If you are interested in having a faster client all around, give all those files a processing tag like 'my sub files' on your 'my tags' domain, and then kill those pages. Sorry, I really have no idea what this means. "processing tag"? "My tags domain"? I'm the same guy earlier who asked the 1.2m session weight question and send you the email of my "callto" profile mode log, and your advice then seemed completely different. But this is a convenient segue anyway to consolidate the other conversation I was having with you into this one. To bring up the quote of yours I responded to again: >I am going to try to associate files with each other, in groups that support ordering, rather than trying to store sort in tags. As you say, page tags can overlap in different contexts, and they are hell to edit if one is wrong. If instead I can store 'this file goes after this file', and present manga chapters and mini 4-page comic-things to you as a virtual 'book' like a cbr, I hope that'll be a lot easier to use. Hydrus currently (as far as I can tell) does not value history very much. Which goes completely contrary to this potential prospective tagless sort you might add down the line. Hydrus (as far as I can tell) doesn't preserve the date you added a subscription, it doesn't preserve image viewing history, it doesn't preserve search history, it doesn't preserve tag history, it doesn't preserve my gallery tab if I were to ever lose it. All of this would just be supplementary to tagless sorting. But still. All I ever wanted was to sort images relative to other images. The philosophy behind hydrus appears (to me) to be standardized tagging, which isn't in itself relative at all. I mean you said you would like to add it, but I just wanted to say that, yeah, this is kinda completely missing the point of hydrus, in a way. What I want is ultimately a completely different behavior than hydrus has ever been. In hindsight I didn't really have to word all this this way. I would be sorta fine with getting rid of my gallery tab, but preserving the contents of it. It's not that important to see the biggest number keep increasing over time, I think.
(36.54 KB 1278x252 Untitled.png)
Hydrus dev, in 428 it appears you put "[open] a new search page [for x tag]" under a submenu, when previously it was just straight there on the right click menu, and it's annoying the shit out of me. I only see the one option under the "open" button anyway. It's only when you select more than one tag does it change to 3: "a new search page for 2 selected" "a new OR page for 2 selected" "new search pages for each in selection"
Just thinking out loud. Hydrus has most features that boorus do. The only things it is missing that I know of are parent/child file relationships, which I think will be addressed sometime Soon™, and translation bubbles. Hydrus can also be run as a server. I'm no expert on the server, but if we can get to the point of perfectly replicating entire boorus, what's to stop some madman with a great Internet connection and tons of storage from downloading fucking everything from fucking everywhere and letting other Hydrus users connect to his server and use it as the ultimate booru? That sounds like it would be a lot better than everyone having to juggle downloaders and logins and cookies and API keys that expire every 48 hours.
I am trying to write a downloader for nicovideo seiga. The problem is that you cannot get the tags with HTML alone - the page justs sends you some javascript to run. I found an API-endpoint that gives me the tags, but it does not give me the file URLs or anything similar - at least not in one page. So I can call the endpoint, and the client just says that it did not find anything to download on that page and throws the tags away. Is there a way that I can execute multiple "network jobs" for the same post? Currently, I have 3 parsers; - The first parses the gallery page and gets me links to the individual post "download pages" *and* API pages (using IDs) - both using "additional url to follow" - Second parser gets the raw file URL from that download page - The third is the tag API, which I also need to call and parse seperately (additional URL from the first parser) and I would like to have it like this: - post page -> API call for the tags -> raw image URL (client understands that API call and raw image URL belong together) Am I doing this wrong?
(79.11 KB 850x862 Amai-Gondola.jpg)
>>15336 >what's to stop some madman with a great Internet connection and tons of storage from ... finding some madlad with a great internet connection and tons of storage
>>15336 >downloading fucking everything How much storage would that take? You can get raw storage for as cheap as $13 per terabyte if you get lucky, so the problem would be the internet connection and the ability to host 24/7 indefinitely. Maybe there could be some sort or extension to or alternative version of the PTR that lets you find hashes by tag and download their files over IPFS or via a direct web link to them. That may be easier to start, and probably better in the long term since there is no central server needed for it to work.
>>15339 >How much storage would that take? I run a small booru made mostly of static images and a few animated gifs; with over 80k pictures the storage needed just for them is around 30GB +/- some depending on image quality and duplicates. I would assume a more generic booru with over a million files (including longplay videos) would need at least some terabytes of storage.
>>15340 Apparently danbooru is 3.4TB https://archive.rebeccablacktech.com/g/thread/79755679/#79814780 Same guy who made that claim linked this in a later post, when asked if it's possible to download danbooru, to which he answered yes: https://www.gwern.net/Danbooru2020
(155.00 KB 487x740 1412222153649.jpg)
>>15293 >- thanks to a user helping me out, I was able to figure out a set of lookups in the sibling/parent system that were performing unacceptably slow for some users. this was due to common older versions of sqlite that could not optimise a join with a multi-index OR expression. these queries are now simpler and should perform well for all clients. if your autocomplete results from a search page with thumbs were achingly slow, let me know how they work now! Dude, it works like a charm again, I love you! Also the new tag lists displaying their siblings are super cool. Thanks a lot.
>>15319 >>15322 It actually did stop producing errors at some point. I have 315 messages minimized right now (of which some are subscriptions), and opening the "manage tags" window doesn't produce any more. I never restarted my client since updating and booting it for the first time post-update. Maybe it just hit a max amount of error messages possible considering it's probably the same same error text but I wouldn't know.
(203.59 KB 1366x737 Screenshot_20210215_012844.png)
(62.61 KB 542x615 Screenshot_20210215_014944.png)
Bug Report. It is about keyboard input while typing tags. It happens once every perhaps 200 entries since I begun using Hydrus with v420/1 ------- Description: Type the tag ---> hit Enter ---> nothing happens Type again that tag ---> hit Enter ---> the tag successfully is shown in the "Manage Tags" dialog. ------- I'm reporting it now because this time an error message popped up. >TypeError >'NoneType' object cannot be interpreted as an integer >File "hydrus/hydrus/client/gui/lists/ClientGUIListBoxes.py", line 1937, in paintEvent >File "hydrus/hydrus/client/gui/lists/ClientGUIListBoxes.py", line 1562, in _Redraw
>>15344 >Type the tag ---> hit Enter ---> nothing happens When I say "nothing happens" I mean that the typed tag disappears from the "input field" but it won't show in the tag list field.
(71.67 KB 654x723 Screenshot_20210215_024001.png)
>>15313 I actually meant the progress bar below the video, but a shortcut to hide/collapse the popups would be even nicer. Consider my feature request about both.
Anyone know a way to extract pending petitions/siblings/parents along with the given reason? I had to restore my db from about two weeks ago, but I lost about 500 pending siblings that I spent quite a bit of time on. I know a little sql, and I'm fine with just extracting them to a human-readable format so I can manually re-pend the siblings.
(72.48 KB 595x755 ClipboardImage.png)
(72.08 KB 629x768 ClipboardImage.png)
>>15255 >does that 'f2d' subdirectory actually exist, and does the 2d41... jpg exist inside it? It does not exist on my primary drive where I had stored 25%. but I can find it on the 2nd drive. >When you split the database, did you do it manually, outside the program, and then boot and 'repair' the missing locations, or was it done in the database->migrate database dialog? Initially I split the database through Hydrus, not manually. This was probably something I caused, not Hydrus when rejigging some of my drives to load a dual boot, some naming structure got shuffled by the creation of recovery images. It isnt just Hydrus that got jacked up, also my scratch disks in photoshop >If you look at that dialog now, does it think everything is in its 'ideal' location, or does it think files need to be moved? I've looked at that a few times and for now it says everything is in the right space, but no there's way for me to tell until i try to open a file if its actually working, Every so often when I reopen hydrus it says it removed some files that couldnt be loaded but idealy id rather fix the main problem instead of redownloading. The other missing information is a ton of tags but I am subscribed to the ptr, so I'm not sure what to do about that.
>>15181 What about GNOME/Budgie, or XFCE4?
(75.70 KB 1920x336 ClipboardImage.png)
I finally started working on my inbox instead of just downloading more shit due to space limitations and I found out that hydrus does funny shit once you get too many files on screen. Surprisingly there wasn't a memleak, it just got to 5GB and stopped.
>>15348 Mostly figured it out with an abomination (probably) of sql, but the sibling petitions seem to have no distinguishing features other than the reason id and the good and bad tags. Do the PTR jannies see multiple separate petitions with the same reasons or are they aggregated?
>>15348 >>15352 Oh I forgot to mention everything was intact except for client.mappings.db but I wasn't willing to wait around for ptr processing again so I just restored from the backup
TypeError 'NoneType' object cannot be interpreted as an integer File "hydrus\client\gui\lists\ClientGUIListBoxes.py", line 1937, in paintEvent self._parent._Redraw( painter ) File "hydrus\client\gui\lists\ClientGUIListBoxes.py", line 1562, in _Redraw for logical_index in range( first_visible_logical_index, last_visible_logical_index + 1 ): Windows 10, 428, was trying to scroll the left sidebar while the window was quite small (so that I could drag and drop to post something onto an imageboard)
I had a good week fixing and cleaning little issues. The annoying popup errors some people got with the new taglists are fixed, thread watchers and multi-column lists are nicer to work with, .clip files are supported, and many file database lookups are faster. I have some IRL going on, shuffling my day about, so I expect the release to be out several hours early. >>15354 Should be fixed tomorrow. Sorry for the trouble!
>>15143 OP should join a pleroma instance instead of trash twitter.
(579.54 KB 1054x1080 43t.gif)
>>15355 Ping to the developer. I'm thinking that reporting minor bugs might be a monkey wrench in the scheduled release. Should those non critical reports stop, so you can focus on changes to the core code?
>>15357 Not him but I don't understand. Are you implying he choose between being told about bugs, or not being told? I don't mean to be rude, but you make it difficult to say anything nice.
>>15358 >Are you implying he choose between being told about bugs, or not being told? Not at all. My concern is originated in the way OP works, with self imposed death lines. I said that by reporting small bugs (I'm the author of two of them), the developer might be derailed in the release schedule. Chasing bugs isn't pretty and many times is frustrating. Also, why to fix something that perhaps will be re-written in the future with upgraded code. So, for me a simple Yes or No will suffice.
>>15359 I really don't understand why being educated on conceivable things to do means necessarily forcing those things in before a deadline. Again, not trying to be rude. But I do not think someone of sound mind can say "don't report bugs to me, because [muh schedule coincidentally means I can't do it before the following deadline. Or maybe it doesn't. Who knows. But for this week you shouldn't report bugs]." ???
>>15357 >>15358 >>15359 >>15360 Please send me all bug reports. I appreciate them. The smaller the problem, the easier I can tuck a fix into the next release. Bigger stuff gets put on the pile with everything else and I'll get to it when I have time or may magically get fixed by accident. I'm always overwhelmed with stuff to do, please don't worry about it.
https://www.youtube.com/watch?v=ezC4ybLJsPM windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v429/Hydrus.Network.429.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v429/Hydrus.Network.429.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v429/Hydrus.Network.429.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v429/Hydrus.Network.429.-.Linux.-.Executable.tar.gz I had a good simple week just working on smaller jobs and cleanup. If you have a large client, it may take a minute for you to update this week. all misc this week The new taglists seem to have gone generally well. There was one display/update error that hit some users, which I have fixed, along with some little menu logic cleanup. I will work on more display and sort options for them in future. Since I added 'pending' status for thread watchers last week, you may have seen some apparently stuck on that status. This was a harmless legacy issue, just old check jobs that were never closed out correctly, which I have fixed. I also cut out the 'checking in' prefix from the waiting-to-check status text, let's see if that fits better. Dialogs with multi-column lists should be less of a pain this week. I think I fixed the issue where the last column could sometimes shrink to a few pixels on the next dialog launch, and I think I also fixed the issue where columns would sometimes get a few pixels thinner every launch. Just set them once more to the size you want, and they should stick now! '.clip' files, those for Clip Studio Paint, are now recognised and importable. Too complicated for me to support thumbs, but they have their own icon. I cleaned up the behaviour of the increasingly tall filetype selection widget, so I am happy to start adding simple new filetypes for a bit. If you would like a new media filetype added to hydrus, please send an example of that file to me with a bit of info about it. Nothing too obscure, please, and I can't promise I can add everything quick, but many are actually fairly simple. I'd like to see about .sai and .wav, and maybe start on an xml scanner so I can at recognise things like .svg. I took a bit of time to check out animated webp this week. Unfortunately, gathering duration and frame info about webps seems to be much more difficult than actually making webps, so I am afraid we will have to wait for an update to FFMPEG or one of the other libraries we rely on to get animated webps going. A new cache speeds up a bunch of file database lookups this week. If you have a large client, it may take a minute to update as this is populated.
full list - misc: - fixed a bug in the new taglist backend that would sometimes error out in a paint event(!) on display initialisation or data changes for some clients - improved the taglist 'tag' vs 'copyable string' copy/select/action menu logic. e.g. 'namespace:*' is copyable, but it is not a tag - thread watchers now skip/clean up unactioned check log entries (this usually happens when a check is due during network traffic paused, queueing the job, and then the client shuts down). if you noticed some odd perpetually 'pending' checkers in last week's status overhaul, this was the issue, and they should clean up. this was always harmless, just revealed with new status code - thread watchers now record serious network error detail in the check log - thread watchers are quicker about notifying UI on checker log changes - thread watchers now report 'time delta' as their simple status when waiting to check, rather that 'checking in (time delta)'. let's see if that fits better in the columns - fixed an issue where several dialogs with multi-column lists would reset their 'last column' size to the minimum three characters on the next load if they did not receive certain size events while they were open. you should just have to fix any broken dialogs once and you'll be good again - I believe I also improved/fixed the issue of dialogs with multi-column lists sometimes shrinking by a few pixels every open/close - the 'we just woke from sleep' detection is now more aggressive. it should now detect a wake after sleeps as short as 60 seconds (down from 5 mins). let's see if we get any false positives during maintenance or other busy periods - if you have a complicated database (one stored across multiple locations), the 'database' menu now has a label in place of the simple database's 'backup/restore' commands - improved the 'directory is writeable-to' check used in the program. on windows, due to some python tempfile weirdness, this was actually hanging on Program Files. - improved the related 'is db dir writeable-to' test in the boot script. if you try to run the program on a custom non-writeable db directory, the crash error should now be nicer, and running a straight client.exe installed to 'Program Files' should now auto-place your db in your user folder, no complaints, like the macOS App - corrected 'writable' typo to 'writeable' across the program lmao - fixed the new header links in the FAQ file, which I accidentally messed up - started work on updating neighbouring .txt tag sidecar export. it isn't ready yet, but it will add tag filters and tag display type to sidecar export with easier expansions in future, and fold it nicely into Export Folders - improved some log-off detection + clean shutdown code, but I do not yet have nice multiplatform support - . - filetypes: - the stacked expand/collapse checkbox widget that lets you select filetypes now always starts collapsed. also, some 'partially clicked' logic is improved when you click through filetype group - application/clip (clip studio paint) files are now supported! thanks to a user for helping out here - just a side note: I looked into animated webp support this week, but it turns out decoding support is rarer than encoding. my normal and fairly new FFMPEG can't reliably render subsequent frames or figure out duration, nor can PIL or OpenCV. I think we will simply have to wait for an update on one of their ends - . - boring db cleanup: - wrote a local hashes cache to store hashes for all the files on your disk, much like the tag one. this should speed up all normal searches and other common file lookups in the db - the raw storage mapping tables are spun off to their own module - basic file info and inbox is spun off to its own module - improved and sped up some inboxing file count calculations - cleaned up some more misc file metadata and inbox code - improved logic in local tags cache next week Some IRL stuff has been eating my time and energy recently. That should be easing up, so I'll be back at network improvements and hopefully some autocomplete improvements for short inputs.
(168.47 KB 820x1060 147_commander-rainbow-dash.jpeg)
>>15361 >Please send me all bug reports Understood.
Will 8chan downloads work soon?
(44.77 KB 694x532 Untitled.png)
Another menu thing I wanted to complain about was pic related: right click > copy > [options for child tag] right click > [2 parents, 0 children] > [options for the 2 parent tags] I had no idea the bottom path copied to your clipboard too. Not trying to say that putting the bottom path under the "copy" menu would be more convenient than how it is right now, but it currently doesn't communicate that it also copies to your clipboard.
(43.21 KB 489x481 screenshot.jpg)
This just came up, care to explain?
>>15367 >care to explain? You sound pretty passive-aggressive m8. This is wording you use against children to abuse your power dynamic with them. This shit doesn't fly when talking to adults. Just remember that you are surrounded by people who have been in hydrus balls-deep for years and have no easy way out, even if it were one day compromised. This is also not the first mention of "trojan" in the thread.
Here's an overly autistic idea, gif like compression for configurably close alternates. There's often image sets with very little differences or translated versions that only change the text. You could set a base image and, much like gifs, only save the difference between the two and apply them on the fly to save disk space(which would vary between a little for recolors to huge saves for high res sets with <10% differences), with an option to rebuild the image and export. It'd be great for all formats except jaypegs because of how fucky modifying jpgs is.
(18.17 KB 203x652 parent tag.png)
Hey, just updated for the first time in a while and i was wondering if this change to the way parent tags display in the manage tags window is permanent or if it can be reduced back to just displaying the tags separately again instead of hanging under the character: tag or at least only make a tag appear only once on the list. Because as it is now its taking up a lot of space (depending on the series and character) and makes it a bit hard to see if i've missed adding a character or other tag.
>>15334 Sorry, let me re-explain: When there are lots of things in the client, it slows down several update routines. When I want to say 'a new tag got added to file x, so let's go through all the files in the current session and make sure any file "x" gets updated', it is obviously faster to scan 50,000 thumbnails than a million. There are similar slowdowns in the download system if you have lots of URLs in downloaders. I am always working to make that code faster, which is why those profiles are useful. But there are limits, and even when I make things faster, I can only raise the limit a bit. A couple of years ago, I estimated most decent computers were 'ok' with a session weight of about 50,000. I have done work on this, and I now estimate about 150,000. The best thing you can do right now to improve UI performance is to reduce the size of your current session. The easiest things to remove from your session are files just waiting for you to 'go through them some day' and URLs and gallery queries that are 'DONE'. Since you may want to find those files again when you want to process them, before you delete the queries or close the pages they are in, you might like to attach a little marker so you can load them up in a normal search page. Let's take an example: If you have a page with 3,000 files in it that you grabbed from several 'pretty women' threads, if you select them all with ctrl+a, then hit f3 to bring up 'manage tags', you can then give them a little 'processing tag', something like 'pretty women to filter'. Then, if you want to see those files again, in any search page you can do something like: pretty women to filter system:inbox And you'll find them again. You can safely close that page and recover memory and CPU time. You might like to use one big tag, like 'stuff I downloaded', or many small ones, or if you quite like processing your inbox in random order, you may find not adding any tags is ok. If you add tags when you download, many 'artist' queries may not need any other processing tags, since you might like to re-summon those files just with the creator:blah tag. If you sync with the PTR, then 'manage tags' dialog will have two tabs--'my tags' and 'PTR'. But if you have not changed your tag services, you will just see 'my tags'. --- Yes, I agree about hydrus and history. I have never added much logging. I think it is in part because I have been overwhelmed just trying to keep on top of the complexities of regular metadata before even adding a time dimension, and also my own aversion to logging in general, and a lack of interest personally and more widely from the community in the features it would bring. Hydrus is at the moment focused on single files in a stateless snapshot, a bit like imageboard posts. I expect to do a real push on file relationships later in the year to add some more ways of collecting things and with the UI to create and edit and view those relationships, and we'll see how it goes.
>>15335 Thanks. I wrestle with menu placement constantly. I am never comfortable with the balance. The nice thing about offloading to this sort of 'verb->actions' submenu thing is the literal labels on the top-level menu are thin (not many characters), making it small and meaning you don't have to move your mouse much to get to any submenus. Bundling more things into submenus generally makes the top-level short as well, meaning you can get to the verb you want with smaller movement. But then shit is in submenus. Different users use different actions more and less frequently, so whether I put 'refresh' or 'add siblings' or 'delete' on the top or buried deeper is always going to be inconvenient. I suppose the whole point of UI design is to try to optimise for the best and least offensive case, but I have trouble finding it. I'm also fighting a bunch of legacy code that I can't break apart until I clean it to newer and more dynamic formats. I notice that 'add xxxxxxxxxxx to favourites' is fucking up my nice compact menu, fugg. An end goal here is to have completely dynamic menus that you can customise and reorder, but until I can get close to that, I appreciate all thoughts on better layouts. If it helps your situation, since opening the selection in a new page is common, you can just middle-click your selection of tags, and it'll do that. Bringing the taglist into the shortcuts system would be another good mitigation against menus the user doesn't like, then they can map middle-click or whatever to whatever action they do most.
>>15372 I don't know if I can do the "processing tag" suggestion, because virtually all my session weight comes from a gallery tab. Earlier when I closed every tab except the gallery tab, I took note of the search queries in a text document. So it's just the gallery tab that's left. And about hydrus history, I think this is just, again, discrepant philosophy. To me it seems so natural to value history, since it's in itself a relative sort independent of tags. Would go hand in hand with actual tagless sorting infrastructure. Since hydrus' tools offer standardization that isn't relative at all, the audience for hydrus and their expectations reflect that. Again, tagless anything is really anti-hydrus. Were it introduced, I would basically exclusively use that, and not reach tagging at all. I also did want an ETA on this ever being introduced, and "later in the year" is ok, considering, again, this is pretty anti-hydrus. Actually not anyone has ever installed hydrus for at all. >>15373 >If it helps your situation, since opening the selection in a new page is common, you can just middle-click your selection of tags, and it'll do that. I still can't do that on my laptop. In my browser, I can alt + click to open in a new tab. Not in hydrus, though.
>>15336 >>15338 >>15339 >>15340 >>15341 I know a couple of madlads who have copied whole boorus. It yeah obviously takes TB of personal storage, and you end up with a client with extra millions of files. The ones I have spoken to sync with the PTR, which currently has ~1.1 billion tag mappings (17 million unique tags over 59 million files), which as a mass is bigger than any booru. I have tuned for 'add things fast', so the quality is obviously less, but there are some 'mostly' complete copies of several boorus' tags in the PTR. The main problem I think is files. Bandwidth and storage is still a killer. Some projects like IPFS attempt to solve this, but they aren't quite ready for prime time imo. There are also numerous legal issues with openly sharing actual file content. I built the file repository to potentially be a bigass file storage server like the PTR, but it never proved too popular and hasn't had the dev time as a result. It could store millions of files though, with work, and I could improve the clientside navigation and downloading. The other problem is that most humans want have names and DM each other and comment and have favourite lists and other social stuff. I have never cared for that, but it is a block when you want to talk about user support for things. All that takes CPU and coding time. But for file spergs like us who mostly care just about file content and metadata, I'd like to help make things easier to find and share. I expect a URL repository to happen in the next couple of years, where users can share file_hash->URL relationships. That could ultimately be expanded to file_hash->Location, for things like IPFS hashes. At that point, you no longer need to do a gallery search, as you can cross reference tags to URLs in a hydrus db and go "hey, get all known files that have 'samus aran' and 'contrapposto'". If we can slowly slowly divorce ourselves from specific sources and just push and grab from an amorphous and anonymous cloud, that'd be neat.
>>15337 Unfortunately I don't have support for this yet, I don't think. If you can generate the url where you get the file url from the tag api page, then maybe you can pipe one to the other. Something like: hit tag api page get tags use 'context variable' parser to pull the tag api page url, then regex it to file url, queue this up as url to pursue hydrus should then add that url and attach the tags you parsed then it'll get the file page you'll pull raw file url from that it downloads file, adds tags on import, everything works™ But that is so jank, and you probably want to reserve the file url as a 'primary' url that can start new downloads by itself, not something that is pointed to by something else. One advanced option, though it is not convenient, is to grab this data yourself externally and then push it through the Client API. Otherwise, complex situations like this might have to wait for the next overhaul of the whole network and parser engine. It is on my mind.
>>15344 >>15345 >>15346 Thank you for this report. I am sorry for the trouble. It should have been the same bug as here >>15354 . Should be fixed in 429, please let me know if you still have problems.
>>15347 Ah I see, thanks. That's a good idea. I hate that when you go fullscreen you still have the bar at the bottom--a lot of the UI is still prototype-tier, from first dev. I have a long-time queued up job to plug the mpv player properly into the shortcuts system, I will add this to that. >>15348 >>15352 >>15353 Sounds like you are basically ok with SQLite. Should be 'tag_parent_petitions' and 'tag_sibling_petitions' tables. You'll want to use the service_id from the 'services' table, and then reason_id lines up with the text_id in the 'texts' table in client.master.db. The tag_ids line up with the 'tags' table, which itself links to 'namespaces' and 'subtags' in client.master.db. If you have a python terminal in front of you, I recommend doing it in several jumps. If you can only do it in one line of SQL, then you'll have some ugly complex JOIN shit like you say. I can help you with this if you have trouble. In the UI, jannies generally see the same reason grouped together as a bunch of rows with checkboxes, like: "I added these underscore fixes" samus_aran -> character:samus aran character:princess_peach -> character:princess peach lara_croft -> character:lara croft And then they can tick all, none, or some for approval.
>>15356 I don't use social media outside of hydrus work. I don't keep up with what is actually good. I expect to move off the big sites in future, or at least have backups on smaller sites. I'll likely be migrating my email to Protonmail. I don't know anything about mastodon or pleroma or any of the other alternatives. Can you say why you like Pleroma? >>15365 I think it is some issue with referral URLs on spicier boards. Hydrus may have the tech to get around that, but I am afraid I haven't looked into the issue yet, so I can't talk confidently. >>15366 Yeah, I agree. I don't think I can do a nice and reliable tooltip on a menu, but it'd be nice to have 'click to copy' without it taking up too much space. >>15367 Thank you for this report. They just keep coming. A guy on Endchan got Wacatac.B!ml. As always, I believe this is a false positive. Something changed in some virus scanners recently, or my code, that is causing more false positives all of a sudden. I hope it calms down, or we pin down which bit of my code they don't like so I can change it. I appreciate the notice. These have always been false positives so far, but if one is ever legit, I'd want to know asap. The warnings usually go away after a couple of weeks once new anti-virus definitions roll out, if you would prefer to wait. Or you can just whitelist the exe/directory or whatever. If you know python, the nuclear solution for this is to run from source, but if you aren't familiar with python, it would be a whole bunch of work. https://hydrusnetwork.github.io/hydrus/help/running_from_source.html
>>15370 Ha ha, that's an interesting idea. Unfortunately I'd never have time to implement something this clever by myself, and the space savings wouldn't justify the work either on my end with the additional CPU/storage complexities, but if you are into programming, it could be a neat project to do. Our best bet so far for image compression is likely JpegXL once it matures, but the boring simple answer I have come to is that the cheapest way to liberate disk space is to just buy a new disk. They are growing so much faster than any software is right now. >>15371 I am going to add some quick settings to how this displays, and some new sorts too, and hopefully some workflow improvements (e.g. 'removing this tag also removes this implied parent (since nothing else remaining implies it)'). I agree it is too busy atm as soon as you get to more than a handful of tags. >>15374 For my gallery queries, I tend to do something like: >I think I like artist 'blah' >I put 'blah' into a gallery downloader and wait to get everything in it (with tags) >If the files that were downloaded need an extra processing tag for special filtering, I add it to them all >I copy the query text >I add the query text to a subscription for the site, so I get blah's new stuff in future >I delete the query from the downloader page Are any of your gallery downloads finished, or are they all very very large and incomplete? I added a routine for a user with a huge single unfinished query some time ago, I think if you right-click on the 'file import status' button, you can 'clear out completed items'. That reduces your queue size a lot by deleting stuff that was 'successful' or 'already in db' etc... Adding various sorts of logging can add significant database complexity. Can you describe which sort of logging you would find most useful of all? And please forgive me for not understanding, but can you clarify what you mean by 'anti-hydrus'? Do you mean 'files without tags' or 'untraceable files' or similar? To look at it another way, what would 'pro-hydrus' mean to you? I'll see about adding shortcuts to the taglist so you can customise alt+click to do 'open in new page' and so on.
>>15380 By "anti-hydrus" I meant, I think tags postulate intrinsic value, whereas sorting taglessly forfeits intrinsic value to instead depend on relative value. Say you just have the color blue. Hydrus would give it the tag "blue". Without tags, the only way the color "blue" could be distinguished were if it were relative to a spectrum of color otherwise. Something I'd consider "pro-hydrus" would be tagging standards. You standardize the term "blue" to mean such and such. That's fine. But at best it reaches relative content. It's not contingent on it. I find using tags as a basis to be needlessly constricting and I would rather it be forfeit entirely, which hydrus simply has never given me the tools to do. To me it's a matter of starting from x [relative], to conceivably reach y [standardization]. But hydrus has thus far seemed comfortable only being standardization. I dunno, it feels like straight up dissent to me. Maybe I'm forcing a false dichotomy, but to me it feels like these were really the only two options. >Adding various sorts of logging can add significant database complexity. Can you describe which sort of logging you would find most useful of all? This was really hard to describe. I already wrote then deleted tons of paragraphs trying to format a reply that gets to a cohesive point. I think the reason for history (for me) ultimately boils down to two things: disproportionate attention, and being a makeshift note. Sometimes you just like something a lot, so you try to "favorite" it in some way. If you remember the favorite, that's fine. But sometimes "favorites" get saturated, and you end up having to constantly elevate priority, to find the crème de la crème of your favorites. It's a whole thing. So history at all, or the sheer density of your having gone to that thing would give it a higher weight in this regard than other things. Mind you, some "history" logs don't even work this way- I just tested it in Firefox, and if you visit a site, you lose the timestamp of the last time you visted it. At least for that day. And the other reason for wanting history is just treating it as a makeshift note. Sometimes when browsing the internet, I find myself having extremely little interest in something, but I will open it in a new tab anyway, just to add it to my history, to be able to find it again. I wish I could go further and add notes about each item if I so chose, but I think that's the main point of history, so I think it would be best to add that. Instead of adding a history log, can you instead add a way to take note of things, where each note is timestamped to the current date? If this is asking too much, I'll just find a 3rd-party program to do it, I guess. But using a hydrus tag to tell myself "this artist draws x fetish I like" feels extremely punishing. Every time I view the tags of any image by that artist, I see my shitty note, not properly compensated for at all. If there was instead some sort of dropdown that said "view notes that cite this tag", and I can expand it to view my shitty note and the timestamp I made it, I would feel less bad about it. Again, this is all wildly off the basis of just sorting images relatively. At the base level, I just wish to have the color blue, and put a lighter shade of blue next to it, without tags. Anything else is just extra.
>>15380 Also all my gallery downloads are finished, it's just that having the tab open costs 1.2m session weight at this point, and I wanted the history of it.
This gif breaks the file viewer for some reason.
>>15378 Yeah I got it done in one line of SQL with the exact JOIN method you mention since I'm not familiar with python's sqlite binding yet. I recovered everything I needed to from the corrupt client, but in future I might play with Hydrus' DB again to get familiar with SQL and its bindings. Knowing how the jannies see petitions helps a bunch, thanks. This will help shape how I do petitions in the future, maybe it deserves a place in the docs if it doesn't have one? Thanks for all the help btw.
>>15383 yes yes i want sex pls
>>15376 That did it! Tags getting added to the downloaded images! So now, I just have to get namespaces, metadata (dates added), maybe some checksums and source URLs to work properly. Depending on the quality of the downloader, I will share it, of course. Thank you very much!
>>15380 Alright, looking forward to the parent/sibling tag settings. Guess i'll just be scrolling a lot when it comes to pictures with a bunch of characters for now. Another question though, i think i've set up a subscription page correctly but is there a way to clear it out when im done sorting everything or do i just make a new page, name it the same and delete the old one?
Is there a list of keyboard shortcuts? I'm specifically looking for the "copy file path" keyboard shortcut. But I've also needed other hotkeys in the past and don't know where to look them up.
>>15387 >i think i've set up a subscription page correctly but is there a way to clear it out when im done sorting everything or do i just make a new page, name it the same and delete the old one? You mean the page the subscription fills with files? If you close the page the subscription will automatically create a new one when it downloads new files.
>>15387 >>15389 Actually it will only do that if you set the subscription to publish the files to a page.
(1.04 KB 164x47 subs.png)
>>15389 >>15390 I've set it to publish files to a page yeah, so if i delete the page below it will automatically create a new one there if i remove it?
>>15391 >there It'll automatically create a new page at the same level you're at.
>>15392 So it does, a lot less complicated than i thought it was. Thanks.
I'm the same 1.2m session weight guy from before, and I just wanna say that my ram usage for hydrus is currently at 8.1 gigs, and my total ram usage is 72% (32gb ram). But yesterday hydrus spiked to 11.5 gigs, and my total ran usage was 96% at the worst. My computer basically grinded to a halt at around 90%. I've already been told how to get rid of my gallery tab responsible for my session weight, so saying this doesn't really add anything. But I tend to run out of ram a lot, and sometimes drivers crash. My audio drivers crashed when this happened yesterday. One time in the past, it caused a BSOD. Not right away, but somehow, after restarting the drivers, I randomly BSOD'd. It gets really bad. I dunno if I was looking for anything in making this post.
(245.35 KB 462x502 ONE MORE GOD REJECTED.png)
>>15394 If you were on linux you could have just set hydrus to a really high niceness value so that it swaps out the memory for the more critical stuff,but assuming you are taking the Micro$haft™ up your ass the only advice I can give is to mess with your virtual memory settings,maybe increase the size of your page file,however the fuck it is that you that on windows. Also whatever is spending 20 gigs of your memory might be a bit more of a priority problem. .t person that only had 8 gigs on deck for the last two years
>>15395 It's just my browser and hydrus. Also I don't think I understood much of what you said. I just googled "page file" thoguh, and I think I've seen that being used. When I look at task manager when I run out of ram, all my prograns using several gigs/ hundreds of megs get reduced to nothing. And the entire time my computer is basically frozen for hours. I don't think increasing the page file will do much, since it seems nowhere near comparable in speed to ram. It's literally just a last resort. I can't really do much to lower ram usage in my browser. But I guess I really need to get rid of my gallery tab in hydrus. I wish easy access to your history didn't cost so much, and was so unsustainable.
>>15395 Actually maybe I do need to increase my page file. I remember when upgrading my HDD, I cloned it and had to manually allocate the unused space. I probably didn't properly compensate for what the page file should've been. But still though. The speeds are so slow, I don't think it's used as anything but a last resort.
>>15397 Actually, no. Responding to myself to say, I think I remember my storage space decreasing a lot when I upgraded my ram. Sometimes windows just doesn't properly calculate my actual free space available until restart. But I thought that time was different. Windows page files might be proportional to your ram, is the thing. Either way, shit is still slow, and seems to be used to just drain everything from ram. I don't think it matters.
>>15394 You can save pages in sessions instead of keeping them open all the time. >my gallery tab responsible for my session weight How the fuck did you manage?
Is there a convenient way to pass an image to an image search service like saucenao or iqdb? If not, is this a possible feature that has been discussed?
>>15400 You can use reverse image search under f3->file lookup scripts, and get tags, but you can only do it on one pic at a time.
>>15399 Can you tell me what my workflow would look like were I to actually do that. Not that I've ever really gone back to my gallery tab to browse. But the fact that I wouldn't be able to anymore would suck. Also I don't see a way to save only a single page, unless I'm missing something.
>>15400 Pretty sure you can drag and drop an image from hydrus onto saucenao and it'll work
>>15402 >Also I don't see a way to save only a single page, unless I'm missing something. Just send it down a page of pages and save it as a session. >Can you tell me what my workflow would look like were I to actually do that You save it as a session, close it for browsing, and then bring it back for whatever reason you'd have to bring it back and overwrite the session as needed. Why do you even use it for?
>>15404 Sorry, can you spell it out for me. I don't know what "send it down a page of pages" mean. But my other pages are worth 200k session weight at this point, so if that sounds like what I think it is, I'd be not only saving all of that, but opening all of that in a single tab, on top of them already having their own tab. >and then bring it back for whatever reason you'd have to bring it back and overwrite the session as needed. This is an enormous pain in the arse. As I implied in my last paragraph, I have countless tabs (pages) open at all times. And tbh I don't even know how to preview page contents besides opening them directly, so I largely forget about any page that isn't the immediately preceding one. But that's not the point. The point is overwriting my current session to view my big ass gallery tab is, an enormous pain in the arse. I don't know why it's such a struggle, to just want to sort taglessly. That is all this boils down to. I'm sure if I synced to the PTR I'd have countless tags I don't give a shit about, cause it's no better than browsing a booru, to me. I just want to have the gallery I cherrypicked, and to sort it relatively as I choose. Without tags. This is why I feel hydrus is at odds with all I ever wanted. So I'm glad hydrus is planned to get tagless sorting later in the year. But everything else, it's not extrapolated from tagless sorting. So something like this- I guess it's just a list of tags, in the end. The big numbers going up to represent a total don't matter. It's just the galleries I've ripped, and the timestamp I ripped it. Why can't this just cost less than it does to have open.
To format my unending bitching over myself as a feature request: I wish you could export your current session/ a gallery tab as a legible text document. Maybe there are different page types I haven't taken advantage of, but for me, I only have imports, searches, and a gallery downloader tab. Imports will be extremely spammy to export, since all you have are the filenames of each individual file. But otherwise this would be convenient, for me.
(537.85 KB 1920x1080 out.mp4)
>>15405 >Sorry, can you spell it out for me.
>>15407 This is super cool. I wish the entirety of browsing hydrus was based on this type of thing. Instead I've just been using normal tabs and it sucks. Trying this right now. I hope re-opening it won't crash my computer and set it on fire, or something.
>>15407 I hate restarting my client, because it's so slow for me. But my session weight is still 1.4m, and my ram usage is still 8 gigs. So hopefully in an hour when it finishes booting and all that, my session weight will be only the 200k from my irresponsible other tabs I have open.
>>15407 I'm currently at 150k session weight, and hydrus is at 1.4m ram usage, for the first time since I installed hydrus. That's cool. I mean it helps, a lot. I can't imagine myself running out of ram anymore.
Is there a keyboard shortcut for switching pages outside a "page of pages"? Once you switch to a "page of pages" tab it loops.
How do you select multiple pages at once? Trying to consolidate them by theme.
Also random thought, but many months ago hydrus changed the default under "file" > options > gui pages > "When there are too many tabs to fit, '...'elide their names so they fit" to being checked, so the page names are useless. Since I'm trying to use page of pages to sort pages now, I had to uncheck that to make them not useless. The greater density of tabs visible might be useful to some, but when trying to sort, it's pure nonsense. I didn't know it was an option to change that.
Considering how much I bitched about wanting tagless sorting, this "page of pages" thing seems super finicky and sucks pretty badly. If there's a way to highlight and drag multiple pages at once, I can't do it with ctrl + click, which is how I would do it in my browser. When dragging a page into a "pages of pages", I have to drag to this extremely thin white bar at the end of the latest page, which fails more often than not for me. Let alone that I'm constantly accidentally shifting focus away from that white bar. And, again, I still don't know how to actually tab away from a "pages of pages" page- once I enter one, shift+tab or ctrl+shift+tab loops in the "page of pages".
Hydrus also should have put this "page of pages" thing in your face. To me, something like this should be the main drag of the program. Instead, I've been using it for years and never even knew it existed. But it remained kinda bad those years anyway. And no one bitched about it, cause it's not what anyone installed hydrus for. To me it's all a shame.
>>15415 main draw* of the program
I think I was giving far too much credit to "page of pages". I just dumped all my pages to the right in a "page of pages", and it's exactly the same as if I hadn't done that at all, except now it's under a single page. I thought it would be a different ui to put everything you have open at your fingertips.
My entire hydrus just crashed. I don't even feel I was doing anything, except tagging a gif, which I guess doesn't happen often. I hope this doesn't have to do with me trying to use "page of pages" for the first time, cause I seriously can't look at any hydrus user the same way again after seeing the current state of "page in pages". I think it's worse than the precedent not being set at all. I feel I shouldn't have said anything at all, and should have taken my expectations elsewhere.
>>15418 Anon, what the fuck are you doing?
>>15379 >Can you say why you like Pleroma? it's full of normal people and not far leftists who censor you for not being a communist
I have a censored doujin tagged for which I recently discovered a decensored edit. What is the easiest way to move all the tags from one to the other for each page?
>>15401 I can't see anything like this on the f3 screen, though I might just be retarded. Can you elaborate?
>>15421 1. Select censored image→f3→copy tags 2. Select corresponding uncensored image→f3→paste tags
>>15407 Noice. This procedure needs to be added to Hydrus' help.
(189.46 KB 1220x956 09.png)
>>15420 >Pleroma >it's full of normal people Normies.
Has support for text files ever been proposed? I have a bunch of porn stories that I'd love to import if possible
I had an ok week. I fixed the tumblr downloader, which accidentally broke recently, and some lagging text display in the download UI, added show/hide options for the new tag parents in the manage tags dialog (along with some other quality of life), and made it so short tag autocomplete inputs can match odd characters, so 'a' will give a result of '/a/'. The release should be as normal tomorrow.
(1.30 MB 1920x1050 a.mp4)
>>15428 ah, thanks. I'm using an older version, so that feature isn't there
Is there a Zerochan downloader, and if not, what would be a way to set it up?
https://www.youtube.com/watch?v=Giy5iQ9gGvc windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v430/Hydrus.Network.430.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v430/Hydrus.Network.430.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v430/Hydrus.Network.430.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v430/Hydrus.Network.430.-.Linux.-.Executable.tar.gz I had an ok week. I mostly fixed and improved things. If you sync with the PTR, it will take a minute or two to update. all misc this week When I did the autocomplete lookup speedup recently, we unfortunately lost the ability to match unusual characters with short tag inputs. Entering 'a' would not return a match of '/a/', and in fact there was no way to look that tag up without some slow cleverness like wildcards. I have fixed this this week. If you sync with the PTR, it will take a minute or two to update while a small new cache is constructed. You can now hide/show the 'hanging' parents in taglists (in manage tags dialogs etc...) from the list right-click menu. You can also set in options->tags whether they should start shown/hidden for normal taglists and autocomplete dropdown results. Taglists that show '(displays as xxx)' sibling info now sort by that sibling for lexicographic purposes. I had thought to make this optional, but when I tried it out, it felt great to me, so I left it as default. Let me know what you think. I fixed an issue that meant some texts across the program were staying blank for a few seconds when they should have been updating. This mostly affected the 'highlighted downloader' panel in gallery download pages. This was personally driving me nuts IRL. I fixed the tumblr downloader. Sorry for the problems here, I misunderstood initial reports. This was due to a logical change in hydrus that just happened to hit tumblr's URL format. If you have tumblr subscriptions, please go into them and see if any recent URLs have the status 'Found 0 new URLs.'. Right click them and say to 'try again' and they should fix up. full list - misc: - fixed 'unusual character' collapse logic for short text inputs in tag autocomplete lookups. in human, this means typing 'a' now correctly gives you the tag '/a/' and _vice versa_ (issue #799) - to make this work, an old database subtag map cache is revived this week in a more efficient form. if you sync with the PTR, it will take a couple minutes to update. the regen routine is also added to the database->regen menu, in case it ever desynchronises in future - absent an override referral url, api-linked url fetches now use the original url as referrer. previously they were sending no referrer. this fixes watching spicy boards on 8chan.moe - updated a 'get all this stuff' database routine to report more info, and a handful of supermassive jobs (mostly db maintenance regen) now report x/y progress with y, rather than just a nebulous increasing x - fixed an odd bug in a common UI text-clearing call that was causing real text not to show up for a while after the clear. this was most apparent in the downloader highlight panels, where status text on file/gallery/network status could sometimes stay blank until a change - the manage tags dialog's "there are several things you can do" button box when you enter tags in complicated situations is now clearer. there are several sorts of intro text on the dialog, the button labels are clearer, and button tooltips have more action information - fixed the tumblr downloader! sorry for the trouble here, I hadn't realised the situation from some reports. if you have tumblr subs, please go into them and set to 'try again' any recent urls that say 'Found 0 new URLs.'
- taglists: - you can now right-click any edit/write taglist (like those across the manage tags dialog) and choose to hide/show the implied parents that now hang underneath tags - you can set whether this defaults to hide or show, separately for the regular taglists and the autocomplete results dropdown, under options->tags - the taglist now sorts lexicographically using sibling tag data where available. I had expected to make options here to use storage or ideal tag, but once I tried it out, using the ideal all the time felt proper to me, so let's see how it goes - fixed the routine that removes mutually exclusive predicates (e.g. system:inbox/archive) when adding to the active search predicates taglist. this fixes the 'exclude xxx from search' menu action and other add/swap actions (issue #815) - gave the taglist right-click menu another quick pass. since there are all sorts of actions that may or not appear, and menu items can get pretty wide with tag text, I am trying out an intentionally short and thin top-level menu of 'verbs' that is quick to navigate with your mouse, and then tuck longer and taller stuff in secondary menus - . - boring code cleanup: - cleaned and unified a bunch of the new taglist sibling and parents display logic and other legacy variables. it now basically all derives from one storage/display state, so behaviour across the program should be more unified. this may cause confusion in some more advanced dialogs, so let me know anywhere it looks weird - the 'favourites' autocomplete tab in 'edit/write' a/c dropdowns now show siblings and parents for the current display service - the tag suggestions favourites dropdowns and taglists in the options now show siblings/parents according to the current service - the 'url class precedence' routine, which tests more 'specific' url classes first when trying to match an url, has a subtle logic change--now, url classes are first considered more 'specific' according to number of path components and parameters that have no default. this stops an url class with multiple optional parameters overriding another with a single fixed parameter (this is what affected the tumblr downloader above). the specific (descending) sort key is now (required components, total components, required parameters, total parameters, len normalised example url) - refactored client object serialisation access routines to a new db module - refactored database transaction code and status tracking to a separate object - refactored some more tag definition routines to the master tag module next week I am having trouble fitting in the network improvements. This coming week is supposed to be a 'medium size' job week, where I would usually push hard on something heavy, but I'll save it for the network stuff. I'll look at my big network improvements job, triage what is reasonable to do, and try to just grind some proper work out.
Minor suggestion: if a downloader uses a domain that ends in .onion, automatically run the requests with the local SOCKS5H proxy for Tor (localhost:9050). I could set that as my proxy in the options, but most sites shit themselves when a 'bot' using Tor comes knocking. Onion sites wouldn't, though.
>>15431 I mistyped, I meant a way to click-and-drag a Zerochan gallery (in this case, my favorites list) and have Hydrus pull it all, the downloaders in the link work for individual pictures but not for this, nor for searching Zerochan images through Hydrus.
(10.66 KB 219x137 Untitled.png)
Grammatical error (missing word) on the latest release. >some [have] that tag
Is there a way to paste multiple tags when searching for files?
I clicked on a gif/video again while tagging, and my entire hydrus crashed again. Instantly gone. This is the second time in the past few days, and also the second time ever. Maybe it has to do with my audio drivers crashing earlier when I maxed out my ram. I guess I'll restart my PC. Though I hate doing that.
>>15438 To elaborate on what I mean by "maybe it has to do with my audio drivers crashing earlier when I maxed out my ram", I mean in the past when my audio drivers crashed after maxing out my ram, I was unable to open mpv (returned an error). I stupidly tried clicking a video in hydrus in this state, which crashed the program. My audio drivers have been restarted since the crash, but as I said before, in the past when my audio drivers crashed, even after restarting them, I at some point BSOD'd. So something might just permanently not be right after the driver crash. So I'll restart my PC.
Is there a downloader for tiktok? I want to auto download all of echo0210's posts
>>15429 Not that guy, but I was poking around in the settings trying to undo the new "recent" order default from the new release, and found the option you want under file > options > tag suggestions > file lookup scripts > show file lookup scripts on single-file manage tags windows
>>15436 Not only this, but I think the buttons on this window are smaller than they were on the previous release. Not to say that all buttons need to be distinctly big to be convenient. but why make them smaller.
>>15426 Don't think this was answered, so I'm replying to bring attention to it
Can I have a QRD on the IPFS support? Is this feature now mature enough to effectively have decentralized boorus that laymen could use?
Was the "find files similar to this one" under the right click menu removed in a recent release?
>>15445 moved to open > similar-looking files
>>15444 Unless I missed something IPFS itself currently still lacks the infrastructure necessary for distributed content discovery,though if you know what you are looking for (and are reasonably sure someone is serving it out there) it seems to work quite well.
>>15446 That's pretty good actually.
>>15448 Yeah it was specifically requested earlier >>15240 >>15255
>>15432 > Taglists that show '(displays as xxx)' sibling info now sort by that sibling for lexicographic purposes. I had thought to make this optional, but when I tried it out, it felt great to me, so I left it as default. Let me know what you think. f3→recent tags sorting now works kinda weird. It was sorted in the order of usage before, and now the tags jump up and down in pretty unclear (at least for me) way.
version: hydrus 430-1 1. Some areas of gui are white, when they should be whatever color is in the qt theme. Same thing with some dialog buttons, where "cancel" is in red font and the other option is in green. Pcmanfm-qt, the qt version of pcmanfm, doesn't have this problem. 2. When on gallery page with search results highlighted, can't multiselect to open in external viewer (sxiv). Only one image is opened, when it should just load all of them for sxiv to view for me to be able to scroll through them. 3. If pictures from the same gallery search weren't tossed all into different directories (f00, f0a, f0b, ...), there would've been a workaround for 2. What's up with that? 4. Consider removing the "help for this panel -->" (in blue text xD) clutter and instead make a "F1" hotkey that will open help. Also it would be awfully nice if that help was not a link to a page on the internet, since for some hydrus is literally an _offline_ booru. 5. Is there a reason the "migrate database" window is not a simple spreadsheet? What the fuck is this "thumbnail location override" bullshit, I was actually looking for just that but couldn't find it for far too long (read: I have autism). 6. Couldn't find built-in viewer options in "options->gui" and zooming to fit screen (which is default or even only option with this viewer) fucks up oekaki art. 7. Couldn't find the "open in external viewer by default" option in "options->gui". ps. captcha as an alert sucks cocks pss. the point of this post is to say that lower-case letters for everything look fucking good
(1.66 KB 104x25 thumbnailtext.png)
What are people using for viewing images? I feel like I'm doing something wrong, because the internal viewer makes every image look like trash, no matter if it's larger or smaller than the viewport, and I've tried all rescaling methods. External viewers only seem to be able to get one image at a time, so that's not really practical for now. I'm using linux, kde, xorg, and on a 2k monitor. Is one of those causing problems? I've noticed the text above thumbnails is also very poorly rendered, pic related.
(900.22 KB 1778x1050 Untitled.png)
>>15452 works on my machine (winblows 7)
>>15453 That looks much nicer than what I'm looking at. The thumbnails also look pixelated with my setup (v430). Text elsewhere in the program looks as intended. My only guess is that it might have something to do with QT_SCALE_FACTOR.
>>15454 So using "QT_ENABLE_HIGHDPI_SCALING=1 QT_SCALE_FACTOR=0.75 hydrus-client" makes it not blurry, though now it's small as fuk (apparently you need both environment variables?). Seeing if I can make it bigger without being blurry.
(2.27 KB 112x26 thumbnailtext.png)
>>15455 "QT_ENABLE_HIGHDPI_SCALING=1 QT_SCALE_FACTOR=0.8 QT_AUTO_SCREEN_SCALE_FACTOR=1 hydrus-client" does the trick, pic related. Not sure why this needs to be set up manually? I searched the codebase and didn't see these being disabled or anything.
I had a challenging week, but I did get some nice work done. I mostly fixed bugs and updated a couple of downloaders. For advanced users, there is also a new sorting system for parsers. The release should be as normal tomorrow. I am way behind on this thread. I apologise, I will triage and try to catch up on Saturday.
Hi, I'm wondering if Hydrus has support for something like a Gallery. Some way to collect things like themed artist packs, variant sets, and other grouped or sequential images. The ability to make the galleries ordered would be appreciated, too. Example: Say I have a simple variant set of a character in various states of undress. Going from fully clothed with a jacket on top all the way to being nude and having just finished some sort of sex act. What I imagine is an ability to select a handful of images and being able to click a button and make them into a gallery with a random name. This can probably be done with a tag namespace. The reason I want it to have an automatically generated name is because I don't want to name all the sequences I have manually. A simple mask like " gallery:1592146707 " would do. I think the numbers generated are based on something like unix time, or how many seconds have passed since midnight on October 1971 or something. I also want to be able to see a front page of the gallery. Being able to give it a manual title and about a paragraph or so of description in it would also be good features. I did a scan of the user documentation and a ctrl+f search of this thread for the word "gallery", and didn't find my answer. As far as the program itself, I imported about 800 images in one of my "sorting" folders to try it out. I... I don't think Hydrus was made for that kind of workflow. Tagging and creating the tag relationships is taking forever. I don't have my own standardized tagging system, so what images to tag with "boob focus", "oral", "deepthroat", and "legs" seems extremely haphazard. There's got to be a better way to do this. Exploring the Hydrus downloader options is interesting, but I've got a bunch more reading to do before I can hack that. I'm also running Grabber to take images from boorus, but I'm making it resize any images wider than 1920 pixels, so that I don't end up with too many 36 megabyte png files of posed SFM models.
>>15458 Hydrus dev sorta has a feature like that in mind, ETA later this year: >>15309 >Yeah, I don't really like how 'page:' tag has worked out. My next push on this is going to be when I do the 'file relationships' and 'cbr/cbz support' expansion. I am going to try to associate files with each other, in groups that support ordering, rather than trying to store sort in tags. As you say, page tags can overlap in different contexts, and they are hell to edit if one is wrong. If instead I can store 'this file goes after this file', and present manga chapters and mini 4-page comic-things to you as a virtual 'book' like a cbr, I hope that'll be a lot easier to use. >>15317 >File relationships and multi-file grouping just isn't very good yet, I am afraid. I hope to do a big push on it later this year. I made a lot of posts ITT bitching a moaning about wishing this existed, too. But I think the typical hydrus workflow doesn't involve manual tagging. I think what you do is sync to the ptr (public tag repository), as well as using DeepDanbooru, which estimates tags for you, somehow. I dunno how this works since I've never done it myself.
https://www.youtube.com/watch?v=CU5NKzKiAGU windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v431/Hydrus.Network.431.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v431/Hydrus.Network.431.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v431/Hydrus.Network.431.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v431/Hydrus.Network.431.-.Linux.-.Executable.tar.gz I had a challenging week, but I got some decent fixes in. all misc this week Pretty much everything was little fixes this week. Nothing hugely significant, but check the changelog if you have been waiting on something. The e621 parser now pulls 'lore' tags, which seems to be some new 'canonical' tag. I have put them in 'lore' namespace, so you can exclude them easily if you want. One user suggested they are useful for certain blacklists. For advanced users, I added a 'sorter' to string processing. It has a neat thing where you can sort by a regex'd substring of the main string. I was happy about this, but then I realised this morning that it is only half of the job I wanted to do. I will add a String Slicer next week so you can just sample the first item or the last three items or whatever you like, once you have your sort figured out. full list - misc: - when parents are hidden in the edit/write taglists (e.g. in manage tags), there is now a '(n parents)' suffix - thread watchers that are DEAD or 404 but still have files downloading now report 'working' status until that is done - search terms with ';' like 'steins;gate' should now work in downloaders. sorry for the trouble! - fixed an issue where un-ideal tags were sometimes becoming non-searchable when they were entirely replaced in manage tags with their 'ideal' siblings (i.e. their autocomplete count went to 0). this was due to overzealous deletion in the new tag definitions cache not filtering out sibling/parent chain members. a small routine will run on update to resynchronise affected tags - fixed an issue when loading up files in the main 'import files' dialog where a critical error (as opposed to a nice 'couldn't figure it out, sorry') in mime detection would cause the whole job to hang - that main 'import files' dialog now counts 'missing' files separately in the error count - fixed tags not updating on the filename tagging dialog when double-clicking to remove from the simple taglists - fixed the sort on the manage tags dialog's suggestion taglists--they now preserve their original sort, rather than alphabetising once sibling/parent data is populated - the manage parser test panel now catches all network errors. error data back from the server is presented better, and the traceback is now viewable in a special new button on the network job control - the edit shortcut set dialog now gives a veto text popup if you try to ok with a shortcut set twice (previously, it would ok and merge down to one command randomly). support for multiple commands per shortcut will come in future - entering alt+number in the shortcut entry dialog on windows will no longer spam some errors about 'null character' - fixed a 'this object is too huge' check in the database, which mostly affects gui sessions with millions of objects, to check against 1 billion bytes max size rather than 1GB, as here https://sqlite.org/limits.html (issue #816) - fixed the 8chan.moe parser, which was pulling hash incorrectly. it should now save more bandwidth - updated the e621 parser to pull their (new?) 'lore' tags, which all end in _(lore) and refer to canonical gender and some spice. they come with the 'lore' namespace for now, we'll see how it works out. a user reports these are useful for blacklists - . - new string processing sort step (for advanced users): - string processing objects have a new processing step: String Sorter - this sorter can sort the whole list of strings, either strict lexicographic or 'human sort' that does numbers properly, asc/desc - it can also take a regex for the sort 'key', so you can sample just the number or name you want for sort purposes - content parsers no longer have the 'sort formulae results' controls. any content parser with existing sort has been converted so its string processing object has a String Sort step appended - the string processing UI is still built around single string processing, so the test UI here is essentially non-functional, but you can see the sort happen in the formula test parse panel - I will add a String Slicer in future to sample the list of strings, so you'll be able to grab the top item etc... - . - boring code cleanup: - refactored the refresh call in filename tagging dialog to nicer Qt signals - the add/remove taglists on the simple panel are also moved to Qt signals - and so are 'filter' taglists - fixed some typos in new help text - removed a 'needs restart' string in 'gui pages' options that no longer does - . - here is a hieroglyph falcon: - 𓅃 next week A thing I wanted to move forward this week, the network improvements, just got lost in the mix, fugg. I also fell way behind on some messages. I think I may simply be ill. Not the coof, and I am not dying or anything, but I'm a little low productivity at the moment. No worries, I'll make sure to rest enough and get back to it. Otherwise, I want to get this String Slicer sorted. Since network improvements are proving such a problem to break up and fold into 'normal' work, I may end up scheduling several weeks over March/April to do it in one go so I can move on multiple local file services, we'll see.
This seems like a pretty common use case so I'm not sure why I'm struggling to find a solution: I have a list of about 200 artists I follow on a booru already (through a userscript), and I'm trying to transfer that list into hydrus via subscriptions. The problem is that I already keep up with all these artists and have all of the art that I want from them. When trying to import them, it tries to download their latest 100 posts, which is both bandwidth-intensive and redundant. Is there a way to say "I only want to be notified about new posts from this point in time on", or maybe to dry run the downloads without actually getting the images?
Just upgraded to 431, getting the following error on startup: A multi-column list was unable to generate text or sort data for one or more rows! Please send hydrus dev the traceback! So there you go: TypeError '>' not supported between instances of 'NoneType' and 'int' Traceback (most recent call last): File "hydrus/hydrus/client/gui/lists/ClientGUIListCtrl.py", line 269, in _GetDisplayAndSortTuples File "hydrus/hydrus/client/gui/ClientGUIManagement.py", line 2668, in _ConvertDataToListCtrlTuples File "hydrus/hydrus/client/importing/ClientImportWatchers.py", line 420, in GetWatcherSimpleStatus File "hydrus/hydrus/client/importing/ClientImportWatchers.py", line 1233, in GetSimpleStatus TypeError: '>' not supported between instances of 'NoneType' and 'int' I'm using Python 3.6.9, if it matters. I'm unable to find anything obviously broken and as far as I can see, I can use Hydrus normally, so I guess that's a low priority one. Thanks for reviewing and let me know if I can provide anything more!
>>15461 Maybe you're looking for the "on first check, get at most this many files" option, available under "synchronisation" when you look at the subscription's options (after clicking "add")?
>>15463 yeah I tried that but the next time it syncs it still grabs the normal number, so it's largely pointless.
>>15462 Damn, thank you for this report. Just looking at it now, it looks to be an edge case situation perhaps for a thread watcher that has not done any checks yet and has not booted fully (maybe a thread watcher that was queued but could never check before last shutdown, or a very slow processing queue). This is code I haven't touched in a while, so I do not think it is new, and it may just be you hit this small edge case once and then it fixed itself when the thread started work. If you do not get a whole bunch of them, you can ignore it for now as basically harmless. I will investigate it more and fix it for next week in any case. Please let me know if you get the error more.
>>15461 >>15463 >>15464 Thank you for this report. In downloading using your user-script, was that outside of hydrus, so hydrus did not do the downloading itself? In this case, it would not 'remember' the URLs, so some time cannot be saved if it encounters them again. That 'first check' number is definitely the one you want to shrink, but I can see on subsequent checks it may then look beyond that in a slow way, expecially if hydrus hasn't ever downloaded from those URLs before. I'll see if I can improve this logic. I think your best bet is to try to set the 'first check' number to the page size of the booru, often 42. If the first page of results is all in the sub, it won't want to look for page 2, 3 and so on. If you are downloading from pixiv, they use a slightly odd format that messes with subscription url calculations at the extremes, and your best bet is to set both 'first check' and 'normal check' to 20 for now and expand 'normal check' later on busy subs. Or wait for me to improve the logic.
>>15466 Yeah this was outside of hydrus. I think the best solution would be to just do dry downloads where it gets the URL and metadata but not the actual image. I don't see any other way where we can know what has already been checked. Subscriptions aren't necessarily ordered by time either as far as I was reading (though they're recommended to be). Will give the suggestions a try in the meantime
Since updating to the latest version (431) my "gallery downloader" tab no longer displays what it's doing in the page name. It just permanently displays "gallery (0)"
(403.26 KB 1123x689 Peek 2021-03-05 20-03.gif)
(66.96 KB 542x615 Screenshot_20210305_201706.png)
>Undesirable dialog behavior when typing tags. I'm not sure if it is a feature or a bug. The issue is that if I input the same tag twice by mistake, such tag will be discarded from the "list field". This is troublesome as input tasks require extra fine attention to not to make a double entry.
>>15469 Since this was brought up, I just wanna say I have never willingly removed a tag by tagging an image with a tag it already had. Because of this behaviour I exclusively tag from the main hydrus window, rather than the image viewer. Because the image viewer has no confirm dialog box, so any changes are instantly made there, requiring me to remember what tag(s) I just undid. The only time I've purposely done this was when I entered the wrong tag. So I would enter the wrong tag again in quick succession. But for any other instance (i.e. if the file was already tagged as such before I opened the tagging window), I have never wanted a tag removed in this manner.
I've been using hydrus all these years and never tried actually browsing my "gallery" tab before. But just now I tried, and I think it ignores my collection sort. When you double click an entry in your gallery tab, it seems to sort by time imported, by latest to oldest, no matter what. You only get your collection sort if you choose it after you've already selected an entry from your gallery tab.
>>15471 I see this can be fixed by hitting "F5", so it's an annoyance but otherwise does work as I expected.
I love Hydrus but I do have two major issues with it that I'd like to speak about: >Hydrus is extremely bloated I think it would be amazing to see it come with the bare minimum, then allow users to add on whatever features they want. This would allow for better security, better performance, less bloat, and I would imagine a bit easier to maintain and focus on stuff individually during development. I do understand this may be a lot of work upfront, however in the long-term I think it would be extremely beneficial. You're not limiting features, or removing anything, simply making them modular with a one click install/uninstall via a plugin. >Spaces vs Underscores I've seen many people have issue with the fact that Hydrus uses spaces, so I simply don't understand why not just make the client and beyond both. For example, the PTR can use whatever you please, spaces or underscores. But just one of those two. Then the users can select if they want to have tags search and show with spaces or underscores. If for example, the PTR uses spaces, and the user chooses underscores, the client simply interprets and shows all spaces as underscores. This would go beyond the client, and push to the Client API as well, so any external viewers would be able to search with what the Hydrus client specifies. This also solves the issue of having two tags in the PTR such as arms_up and arms up, where the user can submit either arms_up or arms up, but it would act as the same EXACT tag, so there'd be zero issues. Also if the user chooses to use underscores in Hydrus, but accidentally tags something with a space, it would just swap the space to an underscore automatically. I really think this would be nice. Currently it is a mess, and can be confusing seeing most boorus use underscores. It would give the user the option, and give the PTR their own option, it would also help reduce redundancy automatically in both the client's local tags, as well as any tags being submitted into the PTR. One more related thing, I think it would be really nice to have the PTR over IPFS, or some sort of decentralized/distributed system. Users could choose to use that to download the PTR faster without hitting rate limits. And also would add a layer of archival redundancy because to my knowledge right now, the PTR is centralized on the server it resides on, with I assume a few backups. Besides that you have what are on users computers from what they download. But it's not really a great system long term imo. Other minor things: >Capital Letters in UI The all lowercase everywhere in Hydrus Client UI looks kinda dumb. No other programs I'm aware of does this. Idk if it's to "be hip" or just lazyness. But I think at least all "titles" and "tab titles" should have capital first letters like every other program. Or at least give that as an option. >Improved Docs I also think the docs need a major revision and update, and a search box. Honestly, I don't know why we're hosting on Github, but can we not just setup a website? Anyways, my main issues are on Spaces vs Underscores, and the bloat issue. And one question: Is it possible to move the PTR DB to one drive, but keep hydrus and the PTR db on another? I see no options to select PTR DB location. It's like 50gb, if I could put that on the SSD for faster searches, then my files onto the HDD that would be nice. Thanks!
>>15473 I disagree about lowercase looking "dumb". But hydrus dev did say a few words about the docs here: >>15310 >I've had trouble figuring out a balance between making the help comprehensive but also simple and easy to get started. I have a crippling addiction to wordy sentences and effort-posting. If it helps, there's a simpler guide made by another user here: https://github.com/Zweibach/text/blob/master/Hydrus/Hydrus%20Help%20Docs/00_tableOfContents.md Can you elaborate on how you would amend the docs?
>>15474 >I disagree about lowercase looking "dumb". Sure, I completely get it's personal preference and it is a niche thing. But at least make it optional, seeing most other programs use capitals. To be clear, I have zero issues with tag capitalization (I know lowercase tags are standard), it's the client texts, tabs, etc. that I have issues with. Again my biggest issues are the first two (Spaces/Underscores with tags, and bloat)
>>15474 With the docs, this was part of my issue. All the docs aren't on the Hydrus website. Going through the Hydrus github and external links, I constantly find other areas with tons of information, or more condensed such as the one you linked. I think all this information should be directly on the site, or at least be linked on the site. A wiki might be a step in a good direction for the docs? Also I think "help" should just be changed to documentation on the website. And the separation between the documents makes it harder to search for stuff, again I suggested a search box, but you can't even ctrl+f the page to find stuff because it's on multiple different pages. documentation is very important, having clear and easily understandable documentation is extremely important, as well as for users to easily be able to find the information they're looking for. No reason to have a bunch of different documentation spread across different links. Unless it's redundant, I'd say it should go directly on the site, and if it's redundant (a shorter version like you posted) at least provide a link to it on the website. Stuff like this is what I mean: https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts https://github.com/CuddleBear92/Hydrus-guides https://github.com/Zweibach/text/blob/master/Hydrus/Hydrus%20Help%20Docs/00_tableOfContents.md
Hey, I am going to catch up best I can after a long delay. Sorry if you have been waiting on something specific. >>15467 Yeah, I'll do something where if total_num_urls in the sub is less than the normal check file limit, it won't look beyond that number of files when it walks the gallery. That'll fix this particular issue and let people start 'quick' subs easily in future. Just for interest: At the moment, subs' basic 'I should stop now, and shouldn't get the next gallery page' logic is: - Have we hit 'normal' limit of new files yet? - Were the last five urls of the current page all stuff we have seen before? There's more complicated exceptions that kick in in unusual cases, but that's basically it. So, if you set the first fetch to get at least one whole page of results, then the next check will likely see five already seen before in the final urls of that first page. I'll fix the logic in any case. >>15468 Thank you for this report. I will check it out, I have been changing some status stuff recently, maybe some update routine got borked. Can you just double-check that your options were not somehow changed? Should be options->gui pages, there should be 'page file count' for some/all pages and 'show import page x/y progress after its name' set.
>>15469 >>15470 Thank you for this report. If you hit that little 'cog' icon, I believe there is an option there to say 'don't remove stuff when I type it in'. Basically add-only mode. Let me know if it doesn't work exactly how you want. Now I've done this recent taglist overhaul, some of my tools are better. I'll keep working on neater workflow and logic and customisation here, some of the behind the scenes is still complete garbage. >>15471 >>15472 Yeah, the sort from a highlighted gallery or watcher (or hard drive import page) is the order of files in the actual importer as they were worked on. If you click the little 'data table' icon button, called "file import status", you'll see the underlying rows of import objects the importer works on. For a booru, this is typically newest-posted first, for a thread watcher, it is oldest posted first, in the same order you see a thread in your browser. As files are imported and the new thumbnails fade in on the end, that's the sort that is being shown. This matches up with 'time imported' as long as the files are all new. When there are files 'already in db', then they get inserted into their download queue position. I'm mixed on how to handle this best. Some users before you have asked for their default sort, maybe I should just have an option to auto-override on any highlight, if that's what you like best. I'll also be adding the 'collect-by' dropdown to import pages, which is currently absent for not dissimilar reasons.
>>15473 Thank you for your feedback. I agree hydrus is huge. I have a very long term plan to make some things more modular, but for reducing size it won't help the biggest offenders, which are bigass dll/so files like mpv and opencv, which most users would consider essential. I am also not an expert in PyInstaller, so perhaps there are ways of trimming files or similar. If you want the real slim version, clone the github repo and run from source, as per here: https://hydrusnetwork.github.io/hydrus/help/running_from_source.html As for making things modular to make it nicer to code, I am very slowly pivoting in that direction. My dream is to increasingly platformise hydrus and spin off features into modules and allow user-made add-ons, and then I can focus more on core and less on specifics. I'll keep working. I agree about underscores as well. When I next put time into the siblings system, I am going to go for some broader sibling 'rules', first for namespace siblings so you can rename 'artist:' tags to 'creator:', and second a specific hardcoded checkbox that just says 'hey, make all underscores spaces' or vice versa. That may ultimately be too expensive to put on the actual siblings system, so I may just apply it as a render rule. But there are too many users who prefer underscores or not to make a single happy standard, so I just need an option to swing it one way or the other efficiently. There was a thought some time ago to have distributed PTR updates. The tech is all basically there--you only need to get a little manifest file from the server itself, and that could be converted to a signed file to even make that distributable--but ultimately my own bandwidth limit problems were relieved when another user took on the hosting for me with a more powerful server. He says feel free to remove the limits I have in place now (I think 512MB/day by default?). I'll be removing that default at some point when I figure out some maintenance stuff that benefits from the rule for newly synced client. The redundancy benefits of IPFS are also neat. Update files don't change after they are created, so they fit into distributed storage very well. The IPFS bridge I have right now is still barely more than prototype. My own experiences with IPFS have somewhat shown it not 'ready for prime time', but some people have telling me the new versions are doing some cool stuff, so I'll check in again some time and see how reasonable it would be to extend our access. I expect to make a 'Public URL Repository' in the next year or two, which would probably really be a 'Public Location Repository'. I could throw up sha256/ipfs_hash pairs up on that, and before you know it, we will have half the job done. I don't really know why I have lower case titles in the program UI any more. I think when I started, I idiosyncratically thought it looked neat, I guess it was my old history in web 2.0 or something. I had a 'capitalise UI stuff' option for a while, but I think it is gone now--I never kept it up to date. I don't have strong feelings about the lowercase titles now, but I do know they are a good normalfriend filter, so I put them to the back of my queue for now of things to think about. >>15476 As for the docs, I think I saw on the discord today that some users are figuring out a wiki of some sort. If that is the sort of thing you like to participate in, please feel free to get involved. Unfortunately, hydrus is always a bit of a mess and a mishmash of different unprofessional things, and my help is the same. I'll keep working on improving little things, and I have tried to write a decent getting started guide, but it doesn't work for everyone. I do not feel hydrus is 'finished' anywhere, and things are always being updated and then going out of date, so my overall philosophy is generally 'get started with this, and then have a play and learn what you like to do by doing'. I have a general distaste for anything too professional, and I can't work in teams, so I doubt I would be able to spearhead the work of a nice unified website, but if some users want to figure something out, I'd be delighted to link it from the program and my help.
>>15477 about the subs issue, I ended up just batching all the artists into 5 separate queries with "OR" logic, then going through the much smaller amount that it grabbed manually. I know this isn't recommended because hydrus tries to be smart about delaying sub queries based on activity but I don't see how that could be more efficient than batching 50 artists together. Adding and removing subs like this will be a pain in the ass but I feel better running 5 queries every day than running 200 queries every few days
>>15480 (maybe hydrus could intelligently combine queries behind the scenes?)
Catching up with older posts now >>15381 Thanks. You'll have to forgive me, but I think you are a little too clever for me. I'll try to explain myself as best I can. In general, most users--and myself--have found that the number of incoming files far exceeds our ability to tag or sort or even just run the archive/delete filter. Most of my efforts so far in setting metadata to files have been making it easier to deal with the firehose of content, rather than zooming in to minute timestamped detail. I am not opposed to additional optional logging such as 'you viewed this file at (timestamp)', but in terms of prioritisation, it tends to have less return on investment to paint with broader strokes at higher levels. That said, for your idea of notes, I believe you can do something here. Hydrus lets you set multiple notes to files--you can set a shortcut for the 'manage notes' dialog, too, under the 'media actions' set under file->shortcuts--so perhaps you can attach notes to your files as you go through them, and give them a title of YYYY-MM-DD or similar? I hope to be making a push on notes this year, which as well as including parsing notes from websites (stuff like artist descriptions, perhaps the beginnings of translation boxes), will also have optional note display in the background of the media viewer. I understand better your idea of attaching blue, and then a lighter blue next to it. If you would like to play around with a metadata that has several relative states, you might like to try a multi-star 'rating' service. Hit services->manage services and then add a new local 'numerical rating service'. Give it say five stars, and then in the media viewer you'll be able to one-click set a star rating. You can sort by that rating in any page. I'll eventually add more custom drawing options so you could, say, colour the thumbnail border different based on the star rating of a file.
>>15480 >>15481 Hey, that is pretty clever! Yeah, this should work, for the most part. The sub will adapt its speed to whatever it is getting, so as long as the OR query is, say, producing 10 items a day and you set it to try to get 5 items every go, it'll hit it every 12 hours. If you are hitting up artist queries, they tend to only need to check once every few weeks, so the bandwidth of many subscription queries tends to fade into the background. The sub only checks the first page, which tends to be easy for the server to generate and likely is already cached from some other user recently hitting it. Bigass OR queries would be more computationally complex to generate serverside, and would not be cached, but then you are doing fewer requests in general, so it is probably a mixed bag of whether you are saving time one way or another. Whichever works for you. When I fix up that logic, I'd recommend you move to individual queries since it is so much easier to manage. Subs are already doing a lot, so I'll try not to be too clever behind the scenes, ha ha. It does save a ton of time if you have a query for, say 'bodysuit' followed by one for 'samus aran'. There are obviously some shared URLs there, and on any normal booru, the sub doesn't need to hit up those URLs twice, it skips, 'already in db', with no network job needed.
>>15383 WEW, I get the same. I may be wrong, but that seems to be a bug in mpv. It renders fine in my native viewer and in firefox. Basically, when I make an mpv window, I set the resolution ratio correct and just tell it to load the file. The black bars on the sides are due to me giving it space, but then mpv is only using the thin area in the middle. I don't have a lot of levers to pull here--it isn't my code--or at least I don't know enough about mpv to know if there is a command I can send to the API to try to fix this. Perhaps there is some metadata in that file for pixel ratio or something that mpv is trying to pull and getting confused about. EDIT: I actually get the same thin size when I load the gif in MPC-HC. It reports 500x333, but then an AR of 140:333 and SAR or DAR whatever the hell it is called of 125:296 (0.422?). Best guess is mpv is actually rendering this correct, and the guy or server script who made that gif published it with a borked setting somewhere. I don't know the best solution here. I didn't even know gifs could have SAR, maybe ffmpeg could tell me?
(84.98 KB 500x333 sar.mp4)
>>15383 >>15484 Behold, through the power of ffmpeg SAR bullshit I threw together, here's the same lad fixed and smaller as an mp4, if you are interested. It looks similar enough to be ok, I think.
>>15479 >But there are too many users who prefer underscores or not to make a single happy standard, so I just need an option to swing it one way or the other efficiently. Would my suggestion not work? Again to be clear, the PTR could in theory use whatever works best for you/the PTR maintainers. Then all Hydrus clients would have the OPTION of selecting between everything being underscores, or everything being spaces. If the PTR chooses to go with spaces, and a user decides they want to select underscores on their client, it would simply replace all the spaces with underscores client side. Then if a user submits a tag that has underscores, to the PTR which uses spaces, it would simply submit as spaces to the PTR because that's what the PTR chooses. Again, this would also solve issues of double tags, like arms_up and arms up both being submitted to the PTR, which should help reduce massive redundancy on the PTR automatically. Basically the client has the choice between either, as does the PTR. Whatever the PTR goes with, when it reaches client side, the user can then choose to keep whatever the PTR uses or swap to the other. It'd be fully interpretable. Basically a substitute for spaces or underscores. Hope I'm clear on that, not sure if it's possible but it sounds like it would be and probably the best most universal option. >I don't have strong feelings about the lowercase titles now, but I do know they are a good normalfriend filter, so I put them to the back of my queue for now of things to think about. Again, I'm not shitting on them, I know some people prefer them being lower case. Just at least provide the option is what I ask. >As for the docs, I think I saw on the discord today that some users are figuring out a wiki of some sort. If that is the sort of thing you like to participate in, please feel free to get involved. I do not use Discord, however if the wiki goes up on the website and I see it, or see it posted somewhere, I will definitely help make edits and work on stuff directly through the wiki. I have mentioned this before, but I encourage you to create a a seperate chatting area, that's more privacy friendly. Matrix, XMPP, even IRC. Discord sucks, I know you know that. >I do not feel hydrus is 'finished' anywhere I love hydrus genuinely, but I feel this way simply by using it. It feels incomplete. I get it, it's software, it's never really complete and I appreciate all the work that goes into it. I just really think Hydrus simply needs a more consistent development flow, and focus on stability and consistency. With the amount of people hopping onto Hydrus, I think something needs to be worked out soon. I agree with your distaste for professionalism/teams, however I think making adjustments to how you and the other community developers work on things would be a big start. Maybe take a look at how Jellyfin works on their stuff to get some ideas. I'm not really a developer, otherwise I'd be more than happy to try to help work some things out. I'm just looking at it from a overall face value. I think polishing things up, and maybe adding a bit more of a focused development workflow would help. The plugin system would be huge for this, but I feel you'd run into many issues attempting it at the stage hydrus is currently in to be honest. Please do keep up your amazing work, I love Hydrus and hope to see it
(84.98 KB 500x333 7.mp4)
(297.11 KB 748x494 dab_peko.webm)
>>15485 >Doesn't render in a browser sasuga hydev, attempt 2
(89.65 KB 500x333 output.webm)
>>15487 SHIT it just made the same file with different settings. Ok, fuck you firefox, here it is in vp9 and av1. EDIT: AV1 not allowed in upload form, ha ha ha. >>15486 Yeah, I agree with you mostly. I'll let the user go with underscores or not. The issue of redundancies on the PTR is something I am not ready to act on yet. There are other cleanup jobs to do there as well, and maybe some sibling stuff, and tag filtering, and I want to get some backend 'recycling' tech going so I can clear out masses of tags cleverly. I am still thinking about it. Mostly just legacy shit to think through. Wiki seems to be here, I know nothing about it, but it looks like the PTR jannies are mostly working on it atm. https://wiki.hydrus.network/ I'll check out Jellyfin, thanks!
>>15384 Great. Feel free to ask me any questions if you want to know why something is weird. I have a decent amount of experience in tags in SQLite now. I'm supposed to be reworking the janny UI as part of these 'network updates' so they can work through petitions quicker. I'll make a note to take some screenshots so users know what it looks like on the other end. Basically it is like: REASON: I accidentally scraped bad tags, please remove. [x] skirt [x] character:samus aran [x] shirt [select all] - [select none] [approve] - [deny]
>>15386 Great, I am glad, thanks for letting me know. >>15387 >>15389 >>15390 >>15391 >>15392 >>15393 Good trick is to hit ctrl+a, ctrl+r, to remove the files once you are done with the sub page, leaving it empty for the next set of files. Also telling your subs to publish to merged overridden page names, so like 'characters' 'artists' and 'sfw' or similar, so you are filtering categories rather than individual queries. Also you can set archive/delete filter to remove processed files once done (options->files and trash->remove files from view when they are filtered), you can sort of merge everything into a 'go away once I am done with you' pattern. If and when your subs overwhelm you, another trick is to give them an 'additional tag' for 'my tags' in the sub's tag import options of something like 'blah sub, process later' or again a broader merged 'nsfw filter' tag and then set them not to publish anywhere. Then you can load them up in a favourited tag search whenever you like. >>15400 >>15403 Yeah, drag and drop is the solution. Hanging an action off the thumbnail menu that ends up with a browser window open on the results page is actually way more complicated than it sounds. >>15412 I don't think you can yet. The page tabs are still working on some older code that is actually emulating the old wx system. I'll slowly bring them more to Qt standard, and then I think you can do neat stuff like shift+click selects.
(165.66 KB 1366x738 Screenshot_20210306_221112.png)
(83.98 KB 770x770 clapping.gif)
>>15478 >If you hit that little 'cog' icon, I believe there is an option there to say 'don't remove stuff when I type it in' Thank you! It works like a charm!!!
>>15482 I'm not understanding what I'm supposed to see under "the 'media actions' set under file->shortcuts", and I don't know how to add notes otherwise, can you hold my hand through this a bit harder. >>15478 >If you hit that little 'cog' icon, I believe there is an option there to say 'don't remove stuff when I type it in'. Basically add-only mode. Let me know if it doesn't work exactly how you want. Wish I knew about this years ago. >>15479 >As for the docs, I think I saw on the discord today that some users are figuring out a wiki of some sort. If that is the sort of thing you like to participate in, please feel free to get involved. I doubt I would have much to contribute to this, but this would be nice, yeah. Like the "cog" icon thing I am just finding out about years into using hydrus. Something like this could be buffered by a user saying "if you're a new user, you may want to do this."
>>15426 >>15443 Yes, I'd like to add support for text files, html, that sort of thing. It'll take a bit of extra effort, since atm hydrus assumes it can figure out filetype from file content, which is easy for a jpeg and often tricky for a document. This job is currently on the big jobs poll, it'll get some proper work eventually. xml formats (some doc formats work this way) may come sooner, since you often can figure out what they are. If you have some example mobi files or whatever format you are working in, please email me some or point me to some examples, and I'll add them to my 'new formats to add' directory and give them a look. >>15434 Ah, thanks, I will write this down. In future, I'll have optional proxy settings for each domain, but this is worth hardcoding in. >>15436 Thanks, fixed! >>15442 If the buttons got smaller, I didn't choose it! God knows why Qt chooses to wrap and size text the way it does, I am still learning the intricacies of layout. >>15437 If you turn on help->advanced mode, you'll get a new 'OR' button on search pages that lets you enter complex input. If you do 'skirt and glasses', I think it'll do roughly what you want. Not as easy as pasting a multiline list, but I think it works. If the tags are in hydrus, try selecting the ones you want all at once with ctrl+ or shift+click. Then, the right-click menu should have something like 'open a new page for all these' and alternately 'require all these in the current search'.
>>15492 If you right-click a single thumbnail, you should be able to go manage->notes to launch on edit notes dialog. This is the file notes feature. There is no shortcut for this action mapped by default, so if you edit file->shortcuts->media actions and add a new shortcut, say F6, and set it to 'manage file notes', you'll have quick access in the thumbnail or media views. Hydrus can do a lot, and different users prefer different specific things. Please don't be afraid to hover over or click a button if you aren't sure what it does, it may solve the same issue someone else had before you.
>>15494 Oh. sorry, I understand now. Despite you saying there is no shortcut for it, I was looking for the shoutcuts already set out the box, and trying to find a mention of "notes" among them. I think the notes feature is a start, but without it being relative at all, it seems like it would be the same complaint I have against tags. At least it exists at all, I guess. But it's not of the behavior I wanted. >Hydrus can do a lot, and different users prefer different specific things. Please don't be afraid to hover over or click a button if you aren't sure what it does, it may solve the same issue someone else had before you. Yeah. But at a certain point, it's more a fault of human nature than not wanting things to get better. I don't remember the whole context, but I remember I used irfanview for years, and I was so used to it producing an error I just closed-out, that when it actually produced the error in a context where I should read it, I just closed it. I ended up posting a screenshot of the error online asking what the fuck, to which someone told me to read the fucking error. This logic is what I heard is compensated for in game testers and such, since the ones with experience gloss-over fundamental quality-of-life features missing, whereas the new guys always question what's possible.
>>15440 Almost all user-made downloaders are here: https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders I am afraid I don't see a tiktok downloader there. I'm afraid I don't know much about the service myself, so I can't talk too cleverly about it. Maybe one of the downloader makers on the hydrus discord would be interested, if you want to get in touch? https://discord.gg/wPHPCUZ This is crass, but maybe the reddit downloader with something like https://www.reddit.com/r/tiktokthots/ would help for related work? Some searches of that username popped up with that. >>15444 >>15447 I don't know enough about IPFS to know what is good and ready to go yet. I'd like to do a push later in the year. I'll just paste what I said up in >>15479 : The IPFS bridge I have right now is still barely more than prototype. My own experiences with IPFS have somewhat shown it not 'ready for prime time', but some people have telling me the new versions are doing some cool stuff, so I'll check in again some time and see how reasonable it would be to extend our access. I expect to make a 'Public URL Repository' in the next year or two, which would probably really be a 'Public Location Repository'. I could throw up sha256/ipfs_hash pairs up on that, and we could have legit filesharing working out the box. >>15450 I may have fucked this by accident in 430. I fixed it back to usage in 431, please let me know if this is working better now.
>>15497 Can you keep this zero-effort non-contribution to the conversation on 4chan?
>>15498 Recoiling at any mention of discord is a contribution in and of itself.
>>15499 I really don't think so. Try using your words you illiterate emotional cunt.
>>15451 Thank you for this feedback. Forgive me if I am brief in responses: 1: Yeah, I am still coming off a wx->Qt conversion that happened a year and a bit ago. Stuff like styling is still all over the place, I will slowly grind it all into one system that works correct. 2: Yes, at the moment, you can only open one file at once in an external program. I hope to add some sort of 'playlist' support, but as you saw, there are some additional technical problems to get around as well. 3: Several users have millions of files. OS file systems often get upset at directories that large, for several operations, so I break the file storage into smaller pieces. You can also see this exposed in some booru CDN URLs, when you get a URL ending in like /cf/cf0123456789abcedf...jpg. This also allows me to do simple database migration by moving subdirectories as discreet chunks. It is 256 pieces at the moment, and in future will move to a dynamic number. 4: If you hit help->help, it should take you to the help html stored on your hard drive. I think that is mostly true across the rest of the program, but let me know if I messed up somewhere. I'd like everything to be available portably offline. A lot of the help is hacky and hardcoded, I am bad at unifying presentation. I hope I will figure out a good balance in future. 5: It is an ugly dialog reflecting an awkward technical situation behind the scenes. Believe it or not, it once looked worse! I'd like to give it a full pass and unify the file and thumbnail storage rows. (note: whenever you wonder if there is a reason why I did something, it is always because the other way was easier to do :^))))) 6: Try options->media. You can set different zoom rules according to different media types. It has ugly presentation again, sorry! 7: options->media again.
>>15500 I don't really care what you think. Discord is bad, having to use discord to communicate with people is bad. Deferring people to discord to get problems solved is bad. Normalization of discord as the only way to get access to software and give feedback on software is bad. Therefore recoiling at being pointed towards discord as a solution to an issue is a productive action. Sowing dissent against the normalization of proprietary spyware as the one and only means of direct communication between people online is good. Feel free to "use your words" and point out where you disagree, nigger.
>>15502 Nice of you to retroactively jerk your illiterate emotional cunt greentext+reaction image combo as being "sowing dissent". You offer nothing with your shitty reply beyond saying that you're emotional in a negative way about thing. Now you add the information (after 5 sentences of baseless out of context assertions) that discord is "proprietary spyware". Why couldn't you have said that the first time you fucking retard. By this being your criticism now, you forfeit your shitty """""""criticism""""""" that I was giving you shit for, which was comprised of nothing but greentext+reaction image. You're so fucking stupid. Look at how I had to beseech your bitch ass to render your useless post non-useless by appending context to it. Fuck you.
(582.18 KB 724x720 ahhhhh.mp4)
>>15452 >>15453 >>15454 >>15455 >>15456 QT_SCALE_FACTOR and its friends are a massive pain. iirc, one of them is 'nice' and is UI scale (which is secretly just font scale, and all other elements follow), one of them is bad, and is 'legacy pixel scale' (which is draw it small, then scale it up with nearest neighbour garbage on screen), and something in there enables different modes. After trying to get this right for different window systems, I believe I cast it from my hands, and now I just inherit whatever the launch environment has. 'Default' Qt generally does the 'right thing', which is modern font-based UI scaling, but perhaps your environment already has one of those set, for another program or system Qt? If you are using my built installer, there is a double problem here as my PyInstaller Qt can't talk much to your system Qt, so maybe some setting is being inherited but another isn't. I am afraid I don't know enough to talk confidently. I am very sorry for the trouble. vid related is my previous experience with Qt environment variables wrapped in a pyinstaller virtual environment across multiple platforms and window managers. >>15458 >>15459 Yeah, I'd love to have good features like this. Until now, though, hydrus has mostly been focused on imageboard-tier images, where things are mostly just on their own and ephemeral. Later in the year, I will be massively extending the 'file alternates' system, which at the moment is a prototype stub for file association, into something that can group files, with UI tuned for smaller numbers, typically 2-10, into 'these are costume changes' or 'these are messy/clean versions' or 'this is WIP of this' or 'these are four pages of a mini manga, they go in this order'. Thumbnails ideally will be groupable and sortable and collapsible by this info. Then there will be substantial updates to the media viewer so it can present that data, a bit like how danbooru does it, where you are looking at a file but it says on the same page 'hey, this file has a master'. A different expansion that will have related and shared tech will be cbr/cbz (comic archives) support. I am not sure what it will ultimately be called, but 'gallery' for us tends to be an abstract name for 'site you download from' like a booru or hentai foundry or something. 'pool' is a booru word for generally larger collections of files. We'll see. I say more in the help somewhere about going nuts tagging lots of things, I have two rules about tags: 1) Don't try to be perfect. 2) Try to only tag what you actually search. If you browse boorus mostly with creator tags, focus on those. Or if you mostly want to look up 'boob focus', try tagging that more than anything else. Tags are for searching, not describing. No one gains from 'brown shirt' or the infamous 'cum on knee'.
(139.93 KB 614x360 Nah.webm)
>>15503 >you aren't allowed to express distaste for anything unless you write a dissertation on it and going out of your way to completely derail the thread
>>15505 >context = dissertation ok retard. thanks for your greentext+reaction image combo, and nothing else
>>15459 Oh BTW, deepdanbooru is ML, legit machine learning neural network bullshit like you read about in the news. It is some research project that looks at your image and compares it to everything it looked at across danbooru and can recognise asuka and ribbons and a heap of other stuff with decent accuracy, and a hydrus user make a shell for it so hydrus can talk to it. Really neat tech, and just starting to work out with good results. I deeply hope we'll have more of this in hydrus in future. I know the big corps aren't going to allow hard nsfw or imageboard banter in their ML tech, so if we end up using our large private datasets to train the world's most advanced bepis recognisers, I'll be pretty chuffed.
(642.21 KB 656x532 Mia.png)
>>15506 Hey, take it easy
(607.01 KB 600x762 This.png)
>>15499 This! Discord is a ZOG's honeypot.
>>15509 Not going to continue taking part in this "conversation" where you upvote an emotional reaction that appeals to your biases. But in case you really lack the self-awareness. You notice how his emotional reaction and that unapologetic reply you quoted, neither of those posts communicate any reason to hate discord. At that point in the conversation, he has offered nothing beyond "this program/service hurts my fefes." Now you come in, providing context (for the first time) as to why one may rationally be averse to discord. You can criticize things beyond the non-criticism of an emotional reaction image. Like, are you serious? But not responding anymore. Continue with the greentext+reaction image and nothing else. It's very nice.
>>15510 Back to /leftypol/ commie.
>>15510 >taking imageboard banter this seriously
>>15510 discord is spyware and non-free. i think that is reason enough to exile it to the void. If you need to use something like discord you could use: telegram, matrix, signal, tox. some are debatable but all even still are ten times better compared to discord in terms of user privacy and integrity defending discord and their awful shit stain policies only hurts people more as it can stir people into loosing private information which they wouldn't otherwise. Discord deserves to be shunned at all times, and is perfectly valid criticism in any situation when discussing possible platforms for discussion. Fuck off with your cuckchan/reddit-tier proprietary loving shit
>>15513 All I said was an emotional reaction image+greentext is non-criticism. If I were to actually make a contextual statement on the coincidental subject matter, what you said here piqued my interest: >defending discord and their awful shit stain policies only hurts people more as it can stir people into loosing private information which they wouldn't otherwise. I am a huge proponent of archival. Obviously. Because I'm a user of hydrus, and am in its general. I don't know what discord is about. I've only used it 3 times so far, and two times I was kicked from the server, and lost my messages there. The 3rd time, I had DM'd people, and they had all conspired to block me, so they all appeared offline, despite obviously being online. It was some shit. But at least I had access to my messages sent to them. For as much as 4chan jerks themselves off over IRC, the only experience I have with it is trying to appeal my bans to 4chan mods, which leads to a mod kicking me from the server with an epic passive-aggressive ban message. I lose access to my messages sent in the server. This somehow manages to be worse than discord, since discord doesn't appear to give you the tools to force yourself to get the last word in. If anything, I am the one who would benefit most from context, any context at all being given to why I shouldn't use discord. But instead I receive greentext and an emotional reaction image, and very unapologetically so. If you purpose is to edify on how I can do well for myself, you can do so by providing context, instead of dribbling some emotional bullshit from your infant mouth down your infant chin. Then doubling-down when someone asks in different words why they should conform.
>>15514 I'm not the anon you're replying to but I don't think anyone mentioned IRC. IRC messages are easily archived because you have a running log of them through your client. Getting kicked means nothing to the log on your disk. I also don't understand why you give "losing access to messages" as the same reason that you don't like IRC and discord, then at the end you seem to be saying that discord is better? The anon mentioned these alternatives: telegram, matrix, signal, tox. Of those the most common choice for a project like this is matrix, and matrix is a super easy drop-in replacement to discord. Many bigger open source software projects have switched to matrix (mozilla, kde, opensuse(?), etc). I'm not going to go on a crusade and say this project needs to be moved from discord to matrix, but I can understand why some people are fiercely against having anything to do with discord. Discord is a massive tracking behemoth, and the idea of letting it track and sell your talk about huge obscure porn collections is, not surprisingly, easily unnerving to people in these sorts of circles who are already predisposed to being paranoid, anti-cloud, pro-privacy, pro-encryption etc. (For the record, I personally don't like telegram because it associates your phone number with the account, and its based on flaky home-made encryption that's never even turned on in 99% of cases. There are many articles about the problems with telegram. I don't think telegram means harm but there are better alternatives and telegram has very few selling points compared to those alternatives (basically only stickers uwu))
>>15515 If you're saying I can export IRC messages, then that's fine. I still don't respect the ability to append a message to kicking someone from a server. It wasn't that anyone brought up IRC; I was only mentioning it for reference. The only comparison I made between IRC and discord was that IRC gave mods the ability to force themselves to get the last word in. Which to me, is only a downside. I can't say much about tracking. But I don't see why the precedent of my being tracked needs to be set. For that reason alone discord couldn't be my first choice, but I assume hydrus dev uses it simply because it's popular. Another downside about telegram is the longest you can go without activity is a year, beyond which it deletes your account. IRC is obviously worse in this regard, since your account only lasts 3 months. But either way, I'm not that educated on this topic, or anything. My original criticism didn't consider the context of discord at all, I was just saying how could there possibly be any defense for out of context greentext+reaction image. Then anons proceeded to upvote the post, not because it enlightened anyone of shit, but because the out of context sentiment appeals to their biases. It's all so stupid.
An annoying bug: When importing files from my drive, tags I add (in the add tags before import screen) don't auto-sort anymore, instead staying in the order I added them. They also don't show siblings/parents.
Is there any way to scrape threads from boards like /b/?
>>15518 Yeah, that's what the watcher does; open a new watcher page and paste the /b/ URL in the "watcher url" box (or click on the paste icon). You may also want to consider https://gitgud.io/prkc/hydrus-companion which is clever enough to ask Hydrus and let you know when a URL is supported by your downloaders and send it to be watched.
>>15493 Wrt to text files, I mostly mean in terms of short stories from Reddit and Literotica. The latter especially can be difficult to save as stories are often spread over many pages and chapters. I'm not sure how it might work to import a "gallery" of stories from an artist's url or something, but I guess it's something to think about.
I had a great week. I caught up on some small fixes and improvements, finished off a new parsing tool for advanced users, improved behind the scenes server code, and reworked the tag sorting dropdown to be easier to use. The release should be as normal tomorrow.
>>15520 Yeah. I think it is all doable, it will just take some cleverer re-jiggering of some storage and import routines. Hydrus isn't great at grouping multiple files together yet, but I'd like some more tools in the next year and beyond. BTW: Check out FanFicFare. They are roughly the youtube-dl of multi-page stories from fanfic sites and similar. I don't know about reddit, but they support literotica and 100-odd other sites as here: https://github.com/JimmXinu/FanFicFare/wiki/SupportedSites There are a bunch of options in the python CLI, like '-f html' to compile a really nice chapter-linked simple flat offline html file. I believe it is even clever enough to replace a smaller/older file with a newer one with the latest chapters, but I haven't played with it enough to speak too confidently.
>>15518 >>15519 You can also drag and drop the URL from your browser address bar straight on to the hydrus client, just like you would drop a file to import. It should recognise it and put it in a watcher page for you. On my IRL client, I have network->downloaders->watch clipboard->watcher urls on, and then any thread I want to grab, I just copy the URL (which for me is mapped to a mouse button), works really well for me.
>>15465 I have not been able to reproduce this issue ever since, but I see from the 431 changelog that you've fixed this already. Thanks a lot!
https://www.youtube.com/watch?v=_ZDUHbIVaz8 ^ After years of seeing all sorts of smaller edits using this footage, I finally discovered this source vid this week. If you like fine art, please watch it in full, it is eight minutes of unbroken kino. It gets a bit heavy at the end, not appropriate for kids. If you don't like old art, no worries, I was just chuffed this week to have finally found this thing. windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v432/Hydrus.Network.432.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v432/Hydrus.Network.432.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v432/Hydrus.Network.432.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v432/Hydrus.Network.432.-.Linux.-.Executable.tar.gz I had a great week. There are a bunch of different little fixes and improvements this week, and the tag sort dropdown is now easier to use. tag sort Rather than a big list, the tag sort dropdown is now a dynamic control that splits the different variables (sort type, asc/desc, group by) into different mini-dropdowns that show and hide when available. Confusing labels like 'lexicographic' and 'ignore namespace' are replaced with clean 'sort by tag/subtag/count'. I hope it is a bit easier to work with now. Give it a go and let me know what you think. If you don't like your default tag sort, you can set it under options->tags. Your existing default should be converted to the new format on update. I also added the option to sort by sibling text or the actual tag text when you have the manage tags dialog open. There is more work to do. One thing I don't like is if, say, on the asc/desc dropdown, you have 'a-z' selected, you can hover your mouse over and scroll down to get 'z-a', but you can't scroll up to get it. I prefer to have scroll work like 'flip' when there are only two values, where either scroll direction will make a change, so I'll figure that out, and in the file sort/collect controls, so there is even less to think about when you edit. I'll also see about saving the tag sort in the session. I also updated the back-end of tag sorting here, so it'll be easier to update in future. Some users have asked for custom namespace sorts (e.g. creator, character, series, then the rest (a-z), then unnamespaced), so I'd like to start thinking about that properly. misc As some users have not realised they were getting a big session that was lagging them, the client now makes a once-per-boot gentle warning popup if your session exceeds 500k weight. If you cannot conveniently reduce your session size, you can turn this warning off under options->gui pages. Subscriptions now work more efficiently with a very small 'first run' file limit and a larger 'normal' file limit. Previously, the second run was often charging right through the handful of files caught on first sync and getting more than it should, but now it should better recognise it has 'caught up' and stop at the right position. advanced users I added a String Selector/Slicer object to the parsing system. You can select the nth item or the mth to nth. m can be 'start', n can be 'end', and either can be a negative index. Try mixing it with the new String Sorter! The String Processing UI has also had an update pass to better support multiple strings--it now shows everything its parent formula had, so you can better test sorting or how processing is changing the third string in the list or similar. The Server/Client API now support single 'range' requests for files (basically anything not html/json). Apparently Safari was having trouble getting videos from the Client API, let's see if this fixes it. If not, I can try supporting the (more complicated) multiple range requests. The server's error handling is also improved--on certain 'emergency' errors, it was sometimes hanging without closing the connection properly.
full list - tag sorting: - the tag sort dropdown has been replaced with a dynamic control. rather than one big list with all possible permutations, you now work on each variable (sort type, asc/desc, group by) separately. what you are actually sorting is easier to understand and select - my stupid "lexicographic/incidence" labelling is replaced with the simpler and neater 'tag', 'subtag', and 'count' - when in the manage tags dialog and sorting by tag or subtag, you can now turn off the 'use sibling' sort. - I'd like to further neaten the workflow here, making the individual dropdowns flip back and forth with a mouse scroll in either directior rather than being just up/down allowed. let me know overall how you find this new control - the 'tag sort' object is updated behind the scenes as well. your old value should be converted automatically - fixed an issue with count tag sorting where deleted tag counts were being counted even when not displayed - if you try to search tags on a page of thumbnails that holds an invalid tag, this is now caught gracefully and you get a little popup saying 'please run the repair invalid tags routine' - . - misc: - the client now gives a once-per-boot warning popup if your session size exceeds 500k. for those who cannot reduce session size conveniently, this popup can be turned off under _options->gui pages_ - when the file import options prohibit a file due to filesize or resolution etc.., it should now always record that as an 'ignored' result rather than an 'error' - fixed an unusual error popup in thread watcher display that could occur during session load. this problem seems to have been around for a long time, but it required a watcher in a previously saved and still valid 'wait a bit' error state and was only vulnerable for a few milliseconds, so it hadn't come up before. in any case, it is fixed - subscriptions with small 'first run' file limits now work better: if you create a subscription with a fairly small 'first run' file limit (this typically matters when the number is smaller than one of the site's gallery page's worth of results), subsequent normal checks with larger file limits will be more aggressive about noticing that they 'caught up' to that small initial sync (previously, they would sometimes incorrectly think the site just got some files tagged out of order and bump right past that initial 'already in db' batch and keep going until they hit their own file limit) - . - advanced string processing: - added a String Selector/Slicer object to the parsing system. this object allows you to select the nth item in a list of parsed strings or the mth to nth items. m can be 'start' and n can be 'end', and negative indices are allowed for both. pair it with the new Sorter for some neat new tricks! - the string processing edit UI is now _more_ multi-string-aware. the test panel has had a code cleanup pass and now has a list of all the starting strings in the test data (e.g. all the urls parsed by the formula that launched the UI) rather than just the first, and a list of all results from that list. selecting any of the starting strings populates the 'single string' area, so you can now zoom in on one particular string to see what is happening to it - the String Sorter edit UI now gets all the strings at that stage of processing, so you can review the sort properly - the new String Slicer edit UI similarly gets all the strings at that stage of processing - future updates will expand multi-string presentation and testing. I'd like to show the whole list at each stage - . - server/client api core improvements: - I had a go at supporting the Range header for file (basically this means anything non-html/json) requests. I added tests and it seems to work. as I understand this mostly applies to browsers pulling video from the Client API. to start, I am supporting single range requests. if it is needed, I'll try to get Multi Range requests right, but for now they'll 416 - the client now understands 416 ("can't do that requested range m8") errors - I reworked the serverside error handling chain. this has been borked for a long time due to my own lack of understanding of twisted's deferred system, and certain late-stage errors were just not being handled right. the server should no longer hang on these and now should print error info correctly, including a rough 500 in true late emergencies, and terminate the connection correctly - . - boring: - fixed up a handful of typo-borked unit tests - fixed my ordinal (xst, xnd, xrd, xth) text generator to deal with 11, 12, and 13 correctly lmao - started some db maintenance routines and logistics to recover definitions and remove orphans in future, I feel great about it so far, but it'll have to wait for more of my db 'modules' refactoring to be more useful - updated the mpv dll on the Windows release to 2021-02-28, it may improve some video support/performance - updated sqlite dll on the Windows release to 3.34.1 next week I think I am going to do a 'clean up' week and try to polish and finish off little things, no big changes, so that I can take the rest of March and perhaps the start of April to finally focus on these network improvements in one go.
>>15468 >Since updating to the latest version (431) my "gallery downloader" tab no longer displays what it's doing in the page name. It just permanently displays "gallery (0)" >>15477 >Thank you for this report. I will check it out, I have been changing some status stuff recently, maybe some update routine got borked. Can you just double-check that your options were not somehow changed? Should be options->gui pages, there should be 'page file count' for some/all pages and 'show import page x/y progress after its name' set. After updating to latest (432) and noticing this hasn't been patched, I can confirm those two settings are set to work, but they still don't work for me.
>>15527 Never mind, it does? My bad, sorry. Maybe I should just delete this post, but I switched to the tab and saw something was "working" in it. But the tab display showed "gallery (0)". I feel on boot it wasn't displaying anything, but when I opened settings it started displaying progress, just to troll me. But now I'm not sure. But I swear I saw it not displaying progress, which was why I made this post.
>>15528 I think it may be delayed because I just booted the program, or something. It is currently stuck displaying "139/214" despite having finished. But it will probably update at some point.
>>15529 Well, it's been another 30 minutes and it's still stuck displaying "139/214". So I don't know what- Ok. So this time i saw what happened. It was stuck at "139/214" like I just said. But literally when I opened the options window and closed it, it updated to "(0)" because everything had completed, So somehow it only updates when you open the options window... maybe other windows.
(41.88 KB 319x673 1.png)
(109.26 KB 557x679 2.png)
(116.53 KB 573x702 3.png)
(174.05 KB 1366x737 4.png)
When searching for tags, the results displays the the wanted tags but some are shown without the "namespace". As shown in the pics. - Pic 1 Searching for 001974 shows a phantom tag without its namespace, for the record, it doesn't exist any tag 001974 without a namespace - Pic 2 Double clicking such a phantom tag shows the thumbnail of the file containing the tags 001974 with namespaces - Pic 3 Clicking the thumbnail shows the "tag list". As it can be seen, the tag doesn't exist without a namespace. - Pic 4 As explained in v432 "What's New" GitHub page, I executed the command: > Database > Check and Repair > Fix Invalid Tags ; but it didn't work.
>>15531 I detected this bug in v430 a is also present in v432. The packages are Linux executable. BTW, check my trips.
>>15471 >>15472 >>15478 I actually just noticed that it's not only that pressing "F5" on a "gallery downloader" tab changes the order to the collection order you've set, but also that you can only press "F5" once. Not only that, but pressing "F5" in this context doesn't behave like pressing it on normal pages; it just changes the order, rather then reloading all thumbnails.
>>15533 Sorry, want to amend this to clarify that the "you can only press "F5" once" thing doesn't mean anything. In hindsight I'm just commenting on the fact that it just changes the order without reloading thumbnails. Because "F5" on other pages reloads thumbnails, since that can't be observed in this context, it appears to do nothing past the first time.
So i got a question, what does the "how boned am i?" button actually mean. Does it show when i need to do a fresh install or my inbox/untagged to tagged ratio or what.
>>15535 It's just a statistic button that >implies Hydrus is taking over your life. Don't worry about it.
>>15537 Not him but I don't think that's a great answer. On my database with 1.7m files the window took several seconds to open, and it told me I appear to be "unboned". The three "duplicate" stats at the bottom show "0", and my inbox is also "0". Besides the "media view"/ "media preview view" stats, I think if you actually used this as intended, it would tell you how behind you are on whatever standard this is meant to measure.
I'm relatively new to Hydrus so I'm still figuring things out. One thing I'm still having trouble with is the gallery downloader. What I'd like to do is to open a dozen or so -chan threads, download all the images from them, and then archive/delete them. When I try this, it'll stop when it reaches some daily bandwidth limit. Is their a way of increasing or disabling the daily download limit so I don't have to wait 24hrs once the default limit is reached?
>>15539 network>data>review bandwidth usage then edit the cap for whichever sites you want to download more from. And also the global cap if necessary.
>>15215 >>15229 I just wanted to bring up again that it is extremely awful importing mega folders to hydrus. Even if you can automate folder names/ isolate the filenames to the mega. Making it browsable in hydrus is annoying and I haven't done it yet for any of them because I can't be arsed. I actually stopped importing megas and just keep the .zip in a folder outside of hydrus. But today I am updating my import of a mega, and holy shit, it sucks.
>>15527 >>15528 >>15529 >>15530 Thank you for these reports. I think part of what might be happening here is I recently changed some session loading stuff to load the 'initial files' of a page in the background, so if you load a big session or one on a slightly slower computer, you see the pages when they are empty reporting 0 files, and then the correct number pops in later. It seems like your client is for some reason lagging way out or skipping the update routines for these numbers. I will check the code.
>>15542 My session weight is only 195k and I updated (last restarted) 2-3 days ago, and some o my pages still say "initialising" and they display nothing, and the bottom of the window says "no search done yet". Also the word "initialising" is underlined red in my browser, so I guess it's misspelled in the client.
>>15531 >>15532 Thank you, this is sort of a bug, not really a bug. More something that I once thought would be helpful, but isn't always, and isn't explained well. When you have multiple instances of a tag in a namespace, like 'samus aran' and 'character:samus aran', the autocomplete provides the unnamespaced version as an option with combined counts, in your case either 1 or 2 files. An unnamespaced search in hydrus matches all namespaced instances, so '0019...' will match 'filename:0019...', 'title:0019...', whatever. This is mostly helpful when you have mixes of 'samus aran' and 'character:samus aran' and you want to get both, not so much when all instances are namespaced. It may also be superceded these days by OR searching and the siblings system. Btw, the invalid tags thing is for tags that break the rules of hydrus, stuff like broken unicode characters or whitespace in the wrong place like 'character: samus aran '. This is just me 'helping' with an additional and mostly undocumented search type. What do you think, should I remove this? Maybe just in some cases? If I keep it, how could I better explain that unnamespaced tags also search namespaced?
(221.43 KB 376x323 yo.png)
>>15535 >>15537 >>15538 It gives you more if you have more than I think 5% of your files in the inbox. Just a nice reminder that you will never catch up with everything. I fucked myself by resetting everything back to inbox for a second pass years ago, when I had like 400k all archived. Never been inbox zero again since. I'll ultimately add some graphs to this thing to show how much you imported over time and some other fun stuff.
>>15541 Yeah, if your files are connected, I recommend you keep them out of hydrus for now. It is best for 'ephemeral' files that work on their own, like most imageboard posts. We'll see how the 'file alternates' expansion later in the year goes. >>15536 Let me know if you encounter stuff like this elsewhere. I imagine there isn't, though, not at that quality. He really hit lightning in a bottle with that collection, it is still bouncing around seven years later. That new neural net ML tech is being used to make photos of politicians sing stupid songs, but maybe in five or ten years we'll have old art come to life and have Wagnerian operas acted out by renaissance characters, and I can finally ascend to the ultra comfy zone. Fuck all modern art. >>15543 Thanks.
>>15543 Oh btw I am British born, so all the spelling in the program is 'colour' and similar bullshit, I apologise.
>>15547 It's fine. I guess because my browser realizes I'm in Canada somehow, when I spell "color", it tells me I spelled it wrong. But I vehemently disagree with adding letters where it's unnecessary, even if I was forced to conform in school, so I ignore it. I didn't know "initialising" was of the same ilk. Also I was gonna say something about >>15545 by bringing up what you said here: >>15482 >In general, most users--and myself--have found that the number of incoming files far exceeds our ability to tag or sort or even just run the archive/delete filter. Most of my efforts so far in setting metadata to files have been making it easier to deal with the firehose of content, rather than zooming in to minute timestamped detail. I am not opposed to additional optional logging such as 'you viewed this file at (timestamp)', but in terms of prioritisation, it tends to have less return on investment to paint with broader strokes at higher levels. But I couldn't really add any new information, so I discarded the post. But basically I was just gonna say, timestamps would only be a means to facilitate the "file alternates" thing you hope to add later in the year. I don't really care about the timestamps themselves as much as the actual "file alternates" thing itself. I wish something like this were actually useful in practice, to aid the "file alternates" thing coming later, rather than just being a number on a window you take a screenshot of and post online for shits and giggles.
>>15544 >What do you think, should I remove this? Not at all. I was afraid this behavior could mess the data base up in some way, but given your explanation about its purpose and usefulness, I beg you to keep it.
>>15546 >Let me know if you encounter stuff like this elsewhere. In BEAUTY quality and style, I have just one satanist painting from 1797 lodged in a British museum. I attached it here. Its information is in the file name. Regarding what you are looking for, I use Yandex for straight and non entertete kunst results, I believe it is very close to your desires: https://yandex.com/images/search?text=high%20definition%20classical%20art&isize=large https://yandex.com/images/search?text=high%20resolution%20classical%20art&isize=large
(36.36 KB 835x464 qimgv.png)
Has there been thought about improving the image viewer lately? I'm coming from using qimgv, which has a much nicer detail for upscaling (especially when zooming). Going through my collection with hydrus, I was surprised how many images had jpeg artifacts; they're basically just not visible with qimgv. Pic related are my settings in qimgv.
(5.82 KB 489x120 Screenshot_20210314_162537.png)
Hi, i just updated my ancient hydrus 384 install which has been inactive ever since i discovered hydrus became unfeasible on hdd. I'm hyped about how well it runs from a pcie 4.0 nvme now. However there is a pretty big problem: It refuses to import PNG files now. (Pic related) I'm running from source on kubuntu 20.04, and installed all dependencies except the proxy stuff in my venv. I updated skipping versions from 384 to 390, 395, 400... 425, 430, and am now on 432. I had each run for a while and checked it was working. 390 imported the huge backlog that accumulated in my import folder, but skipped the PNGs. All later version up to the current one skipped them too. And the downloaders that added thousands of images from pixiv and boorus today also didn't import any PNGs. I'm sure there should have been a few. Is there any plausible reason for this? The only thing wrong I can see is that it's complaining about my ffmpeg build having an unknown version number. Does it use that for PNG, and will it actually refuse to use it if the version number is weird?
>>15553 Nuking my custom ffmpeg build and replacing it with the default distro build fixed it. Why does it only seem to need it for png files? It even imported mp4 files fine without it.
asked this in endchan but it looks like this board is more active: Is there a way to automatically tag images with the page: namespace? I want to import a 100+ page comic but I'm definitely not trying to tag each page manually
>>15555 When importing files, there should be a button in the lower right corner of the pop-up window that reads "add tags based on filename" that brings up a file tagging window. You can then use regex to extract the page number from the file name. It might take some playing around with, but the advanced tab has regex shortcuts and guides you can mess with. There's more you can do with some regex knowledge as well, like auto-tagging series names and stuff like that.
So I recently moved a big folder of sfw material into my db, but I'm now coming to regret that, as I'd like to keep those files separate from my nsfw stuff. This brings up multiple questions: 1. Is there a way of migrating only part of a db based on tags? Would this also keep the tag relationships? 2. I'm currently running Hydrus from the Linux executable using a shell script. How do I launch a separate instance that only looks at the new db? I'm on Hydrus 335 btw
>>15557 I don't know about 1., but check out -d from https://hydrusnetwork.github.io/hydrus/help/launch_arguments.html for 2.
somebody help! regular sankaku login script stopped working with message 'Missing cookie "pass_hash"!' so followed https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders/Sankaku#capi-v2-auth and >>15270 's instruction At first I got a 401 error when sankaku beta downloader is parsing gallery pages But after reenter the value to HTTP header column gallery parsing working fine but when the sankaku beta downloading all I download is pic related the bandwidth limit is preset and I use beta sankaku downloader
>>15559 If your image also mentions "Expired link - please reload the site", you're in luck, I actually hit that issue! So of course the first step is to get a new value for your header (which you did already). Next, you'll right-click this file and "manage" -> "urls". Delete everything that's in there (this list is the reason why your Hydrus client thinks it already downloaded the file), then you can try downloading again. You may also want to remove the tags on this file (else it'll show up on a lot of your searches) and tag it with some "meta:sankaku-crap" or something, to easily find it next time this happens, but that's an optional step. Hopefully it works for you as well.
If you update a database backup, but stop it partway through, does anything in the backup get updated, or does it just revert all the progress made. I have a very large number of files since the last time I backed up. Can that be done in parts by just canceling the backup when I want to get back to Hydrus, or does it need to be done all at once?
Might be a retarded question, but is there any way for the client to automatically look for your images on image boards with tagging and apply creator/character/series tags based on what the board has tagged them as?
>>15560 Thanks! >If your image also mentions "Expired link - please reload the site" actually not but by following your steps annoying logo image downloading stopped! thanks again! but keep failing to download images the strange thing is I can download some image some of them are even behind censor wall so I think authorization is valid search query is "monster_enako" https://chan.sankakucomplex.com/post/show/22030284 https://chan.sankakucomplex.com/post/show/10484997 https://chan.sankakucomplex.com/post/show/10482590 https://chan.sankakucomplex.com/post/show/7384714 https://chan.sankakucomplex.com/post/show/7384711 don't judge me Maybe I fucked up when following the instruction steps or hydrus settings. I am going to see more further
>>15562 To find tags on boorus, if your file is identical to the one on the image board, then you can copy the md5 hashes in hydrus, use a text editor to add "md5:" in front of the hash, and then run each of those searches through the gallery downloader, choosing whichever booru you want to try. If the files are there, then your local images will be updated with tags from the booru. If the files aren't identical by hash, then you can use the iqdb lookup scripts in the tag manager (only for single files though). If you have tons of these images, you can try Hatate to automate the process: https://github.com/nostrenz/hatate-iqdb-tagger I've had a lot of success with it, but be warned about false positives such as occasionally labeling censored images as uncensored or vice-versa.
>>15563 ummmm I can download images by import queue -> open detailed file import status -> right click url -> copy sources then url import tab starts and my hydrus downloads the image but I get the same annoying log again from gallery downloader.
Is there an option that shows short-cuts in the right-click menu? I don't use Hydrus very often and always forget that F12 is archive/delete, F7 is archive, etc.
Hm, the help says the PTR's bandwidth limits are gone, but I still get stuck at the same 512 MB/month limit as in the past. Running on SSD made it possible for the first time to process all the legacy tags, so as far as I can tell, it now connected to the new PTR for the first time. I had to click "refresh account" to get it working again. Do I need to change anything else? Or is that limit still intended?
>>15147 >The 8kun /hydrus/ board is now deleted. It's not, though. https://8kun.top/hydrus/ is all still there...
The bug wherein Hydrus doesn't launch a GUI and instead hangs with 100% of a core occupied still happens as of Version 432. Last word from Hydrus dev was that >Unfortunately the problem seems to be with PyInstaller, not my code. A couple of guys have been working on a different build method here: >https://github.com/ReAnzu/hydrus/actions/runs/452850300 >I hope to catch up with their work when it is finished and working and fold it into my build too. Is that build method ready to be folded in? I just want my images back...
I had a good week fixing bugs and adding a little quality of life. The release should be as normal tomorrow. I expect it to be the last release for a few weeks so I can work on a larger network overhaul.
>>15568 Ah, I am sorry, I forgot to change my wording when it came time to post everywhere about the change. I originally thought there was a 'delete' button on the 8kun admin panel, but when I actually went to do the final deed, I just couldn't find any method to delete. As far as I could tell from their FAQ, the semi-official way to drop a board was to not log in as BO for two or three weeks and then it ends up in claims. I haven't logged in, so I guess it is in claims now. iirc, I delisted it. The site is not exactly a hotbed of activity any more, so I imagine it will eventually be deleted one way or another. >>15569 Last time I talked to one of the guys working on this, he said they were ironing out some issues and looking for more testers. They are on the discord if you would like to contact them directly. https://discord.gg/wPHPCUZ This seems to be the latest test build: https://github.com/ReAnzu/hydrus/actions/runs/590475620 Are you able to run that dmg? Should be: https://github.com/ReAnzu/hydrus/suites/2098586666/artifacts/42623118 If you try it, make sure you have your existing hydrus database, stored in ~/Library/Hydrus by default, backed up.
https://www.youtube.com/watch?v=twOFs6zJq8Q windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v433/Hydrus.Network.433.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v433/Hydrus.Network.433.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v433/Hydrus.Network.433.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v433/Hydrus.Network.433.-.Linux.-.Executable.tar.gz I had a good week doing simple cleanup and fixes. This is intended to be a 'clean' release. There will be no release for about three weeks. all misc this week Thanks to the work of a user (who was also very patient as I slowly added the features he needed), we now have a new native twitter downloader! It gets video now! (Including gif, which on twitter is just a small mp4!) It is called 'twitter syndication' and uses a new access method. Its main limitation seems to be that its search can only reach about 500 tweets into the past. Please check it out, and if it works well for you, please move your subscriptions over from nitter to save them some bandwidth. This is the first version, so it may well have future updates. I have written a new 'scrollable' menu button, which works like the file sort selector--instead of a dropdown, you get a menu, and you can scroll through it in either direction with your mouse. I have converted several old small choice dropdowns to this new button: file sort ascending/descending, collect-by 'collect/leave unmatched', and all the new tag sort dropdowns I added last week. Try putting your mouse over and scrolling them! This is a little thing, but I hope it smooths out the workflow here. I'll likely use it on more things. Some more mouse changes: Middle-clicking on the collect-by dropdown now clears it. Middle-clicking or right-clicking the 'page tab bar' to open the 'new page' chooser or the page right-click menu should now only work on the tab bar's empty greyspace (previously, this was confusingly happening on a click on greyspace in any child panel, like some random downloader page panel). Namespace file sort (the 'series-creator-title...' stuff) now support a-z/z-a ordering. Files that have no match are still counted as 'less than a' in a-z terms, but I can change this stuff easier these days. I don't use this myself much, so if you do, let me know what you'd like in future. Network job controls (the little panel that shows your downloads) now let you edit their active job's bandwidth rules from the cog menu! It will show all the network contexts and which are currently waiting on bandwidth. If the job is using the 'default' rules, it will say so and allow you to set specific ones just for that domain. full list - misc: - thanks to the effort of a user, this week reintroduces a native twitter downloader! it now gets video! (and 'gifs', which on twitter are just mp4s) please experiment with it and move your nitter subs back to this as you find success. it is called 'twitter syndication' and uses a different access method to get tweet info. it should get the highest resolution videos and images. the search has limited lookup distance, perhaps 500 tweets, but should work for most subscription purposes. this is a first version and may have future updates - on the main gui, middle-clicking and left-double-clicking to open the 'new page' dialog, and right-clicking to open the page menu, should now only work on the page tabs or page greyspace. middle-clicking on some random downloader page greyspace should no longer spawn these commands. also, tiny change--middle-clicking now activates here on click _release_ rather than _press_ - on the file sort widget, the asc/desc sort dropdown is now a 'scrollable' menu button. since sorts are just one of two values, you can now scroll either direction to flip it - the 'collect/leave unmatched' in the collect control is also now a scrollable menu button - the new tag sort dropdowns are now all scrollable menu buttons. go ham with them - middle-clicking the collect-by dropdown now clears it - namespace file sort now supports a-z and z-a sorting. files with none of the matching namespaces still count as 'less than a' in a-z terms, but since I am updating all this code, perhaps this could get more attention. I don't use this much, so if you do, let me know what you would like - network job controls now show their jobs' current bandwidth limits on their cog menu, split up by network context. you can edit the bandwidth rules directly from this menu, and if it is using defaults, set a specific ruleset - userpath generation routines used for database location fallback and default export directory determination now recover from failure in the case of undefined user directory. the client will now not boot if the userpath is needed but undefined - the client api no longer prints empty lists for any tag statuses on file metadata calls, nor service entries that have no tags at all - fixed the new subscription 'caught up to small initial sync' calculation, which last week was only firing properly after one page of results - subscriptions are now better about saving interesting status notes on their gallery logs, rather than overwriting with boring 'no new urls found' messages - fixed a typo bug in the new range header implementation for unended ranges - . - boring cleanup: - refactored ClientGUICommon and ClientGUIControls to a new 'widgets' module - refactored the menu buttons from ClientGUICommon to a new ClientGUIMenuButtons - wrote a mixin for the basic menu button behaviour and simplified the classes - wrote a new 'choice' menu button that has a dropdown checkbox menu and allows mouse scrolling to navigate the list, with wrapping at boundaries - refactored the updated network job control to its own file ClientGUINetworkJobControl - misc db code cleanup and refactoring - misc tag logic cleanup next week I have had trouble fitting in the 'network updates' I have been meaning to do in Q1. These are mostly admin-side updates to server control, brushing up ugly janitor UI and fixing permissions and account management that hasn't been touched in years. I am going to take a few weeks to focus on just this work so I can move on to multiple local file services for Q2. I am expecting to take three weeks. 434 should therefore be on April 7th. I'll make progress updates on tuesdays as normal in case the schedule changes.
Hi again, I'm currently adding some artists on pixiv again, but hydrus tells me the subscription is dead for a few of them. E.g. one of them is this one: https://www.pixiv.net/en/users/9984008 and if I check the gallery import log it shows "The parser found nothing in the document!" for the url: https://www.pixiv.net/ajax/user/9984008/profile/all If I open that url manually, it looks like the ajax url of another subscription that works correctly. Any idea what could be wrong? What's the best practice for pixiv subscriptions anyway? Initially I added a few artists as new subscriptions, but now I just use one pixiv subscription with each artist as a separate query. Are there any pros and cons to this?
(51.02 KB 977x821 Capture.PNG)
I didn't see any parsers for aryion.com on the hydrus presets github, so I tried making one myself. The test parse seems to work, but when I try importing it says it couldn't find a file. It's one of those types of links that just immediately downloads, instead of being an image viewable in your browser that you can right-click and click "save image". So I guess it's some kind of redirect, and not a direct link to the file? Is there any way to find out the direct link to the file? Or maybe some way to tell hydrus that it can get a file there? >>15573 I think this is because all of that user's works are R-18 and so you need to give hydrus your pixiv login to be able to see them.
>>15574 >I think this is because all of that user's works are R-18 and so you need to give hydrus your pixiv login to be able to see them. Actually I added my old account's login ages ago, which still appears to work fine when downloading single R-18 links. E.g. throwing https://www.pixiv.net/en/artworks/10783738 at the url downloader successfully downloads the image. I also tried logging into that old account in my browser, and it still works fine for R-18 there too.
(21.00 KB 1668x158 123.PNG)
>>15575 Well the issue does have to do with logins because https://www.pixiv.net/ajax/user/9984008/profile/all is completely empty if you visit it when not signed in. I don't think the parser itself is broken, you're just not being signed in.
>>15573 >>15575 When giving hydrus your cookies, make sure to use the hydrus companion browser extension to do so. I added pixiv cookies to hydrus with just a random 3rd-party extension, and it randomly stopped downloading r-18 works, and I didn't understand why. I re-added cookies, changed my pixiv password, everything. Then I used hydrus companion instead, and it started working. That downtime made me miss content I was specifically trying to import in time, so it still hurts.
>>15576 >>15577 Sending the cookies using the hydrus companion extension fixed it.
>>15566 Bumping this and adding another question. Is there a full list of hardcoded/default shortcut, including cases where the same key functions differently in different contexts?
>>15571 >This seems to be the latest test build: https://github.com/ReAnzu/hydrus/actions/runs/590475620 >Are you able to run that dmg? Should be: https://github.com/ReAnzu/hydrus/suites/2098586666/artifacts/42623118 The test build page doesn't show a link for the .dmg artifact, and the direct link (https://github.com/ReAnzu/hydrus/suites/2098586666/artifacts/42623118) throws 404.
>>15542 It's been 5 days since you reply (long since the page itself has been active without restarting my client), and it just displays "system:similar to 1 files using max hamming of 8". At the bottom of the window it says "no search done yet". Is there simply no hope at this point of ever knowing what the contents of this page ever was?
(36.63 KB 360x360 hhmph.mp4)
>Went from an unboned fate to 30% inbox in a single query. Mr Bones Wild Ride is real.
A feature request I have is the ability to search for subscriptions by typing/ pasting a name. I can hit a key on my keyboard to be brought to the first in the list that starts with that character. But for entries that are just the japanese artist name, I can't do that. I try to use the "check queries now" button on that window instead of downloading otherwise, because I always thought it was better that way, since I thought that makes it check for new files sooner, since it found new files in a tighter time period.
(34.60 KB 838x248 Screenshot_20210319_164153.png)
(31.31 KB 838x248 Screenshot_20210319_164218.png)
(54.67 KB 576x690 Screenshot_20210319_164129.png)
(51.46 KB 576x690 Screenshot_20210319_164108.png)
Somehow I can't get regexr working. I have a pretty big folder from the pxdownloader extension, that I'm trying to extract post number, artist id, artist name, post title, and page number from. There are single image posts and multi image posts in the folder: /PxDownloader/Bボタン(576198)/ゼオラがアニメEDの水着でうっふんあっはん(14853452)/01.jpg (multi image post with numbered files) /PxDownloader/きょうでゅ(202253)/アリカ体操服(13889606).jpg (single image post) On regexr.com I managed to get the page number and post number using these: page number: [1-9]\d*(?=\.(jpg|png)) post number: [1-9]\d*(?=\)(\/[0-9]\d*|)\.(jpg|png)) but once I tried it in hydrus it returned the lookahead as part of the result for some reason. Is lookahead not supported?
I'm on Hydrus 335 Linux and I keep on getting crashes. The last line in the CLI log is always "Fatal IO error 11 (Resource temporarily unavailable) on X server :0." This can happen when Hydrus is unattended or during use, but the former is less frequent. In active use, Hydrus typically lasts between 5 and 30 minutes, though I've had it crash immediately on startup
When I try to do a gallery download (gelbooru & danbooru) all the images get "ignored." Why is it doing this and how can I prevent this?
I just updated to the latest (433) a few hours ago and the client has done nothing since then. I don't believe it's ever behaved this way before. Right as it booted I added some things to my "gallery downloader" page, and even imported some stuff. It even managed to update a subscription by one image. But literally nothing has happened since then. I can't search (rather, I can't see search results), I can't even see the ones of files that managed to import since boot. Double-clicking an entry in my "gallery downloader" page doesn't make anything show up, either. The weirdest part is the first page in my hydrus is just a search that returns a single file, one of my favorite images. It managed to load on boot. I had clicked it once, so it showed up in the preview window. But then I tabbed away and back, and it stopped displaying in the preview window. Clicking off of it then back on doesn't make it show up again. I still have access to my "my tags" list in the "F3" window. But as I type this, the entire client froze a few minutes ago, and has remained frozen. I don't recall this happening for me since saving my 1.2m session weight "gallery downloader" page under a "page of pages" and closing it. Maybe this is the new behavior of hydrus on boot, since on boot I experience some lag/ freezing and it's unusable for a while. But I don't think it's ever lasted several hours, or ever behaved this way. Client still frozen for me btw.
>>15587 Well it's still frozen 20 minutes later. I just wanted to say that I didn't know hydrus companion still showed you which files are already in your hydrus when hydrus is frozen. I actually thought it specifically didn't. Maybe this freeze is different somehow. But yeah, I'm going afk for now, so I might not immediately be able to tell when it unfroze, if ever it will.
>>15587 >>15588 I dunno how much time has past since I made these posts, cause the time zone is different than where I am. But it's literally still frozen and it's probably been hours.
>>15587 >>15588 >>15589 Holy fuck man. I just noticed that there won't be a release for several weeks according to hydrus dev. This is absolutely awful. I will have to revert to the previous version unless it magically fixes itself from this literal-did-nothing-for-several-hours state, then froze-for-several-hours-and-remains-frozen state.
>>15584 Try these bro These worked your file name example page number:\d*(?=\.jpg|\.png) post number:(?<=\()\d*(?=\).jpg|\).png|\)\/\d) >>15585 my hydrus 433 is working fine two days straight. I'm running from source and my environment is OS is Arch Linux but kernel is a bit old - 5.4.97-lts python is 3.9.2
>>15591 >These worked your file name example Thanks!
>>15587 Client still frozen btw. It's really not ever going to unfreeze at this rate, so I'll stop replying to myself about it. But this is hell.
(298.35 KB 325x720 confused bow.gif)
>>15593 I am curious enough to ask why you haven't attempted to kill the process and reopen it by this point.
>>15594 I was importing stuff. I didn't know how hydrus would treat files that may've only partially imported. I guess I'll do that. I'm about to go to bed, so may as well instead of just another -8 hours.
>>15594 Also I hate that it took me 30 minutes to reply to you. Maybe it's just me, but this site is always failing to update replies, instead it just silently fails until I refresh. It usually says in the tab when there is a new reply. But here I only knew you replied by manually refreshing the page in case exactly this happened again.
>>15594 Maybe the restart fixed it. I closed all the import tabs and re-imported them (wasn't sure if they were going to continue where they left off or what), and they actually display in the preview window, now. Imported 21 hours 49 minutes ago. That's the amount of time my client had been broken/ frozen. If it's really just going to work now, this was the first time this has ever happened to me. I only restart hydrus to update, unless something out of my control happens.
>>15550 >>15551 Thanks. Yeah, I love all old fine art. I am not sure what to call that modern animation. I feel like 'animatic' is the right word, but I think it isn't. I am not sure how good the situation is these days, but I got a bunch of fine art torrents some years ago, probably 50,000 paintings, and a bunch of it tagged with creator at least. A really great trick I discovered is the word you want to search for is 'paintings' not 'art'. Put 'renaissance paintings' into any search, you get a bunch of fine art, but you put anything with 'art' in, you get overwhelmed with 'erotic art' in your results. Same with this latest vid--putting in 'animated art' is a bunch of other stuff, but 'animated paintings' took me straight to some playlists with re-uploads in. Once I knew the guy's name, I could find high quality. A bunch of galleries (like actual museum galleries) are digitising their stuff into publicly accessible sites too, like here: https://www.artic.edu/collection This stuff does often have a json api attached, so at some point I'd love to put a downloader together and parse their stuff. >>15549 Thanks. I'll try and show what it does better for new users. >>15552 Yeah, it is a 'big job' waiting for when I have time. I have a whole bunch of plans to radically improve performance and caching, add more quality options, and dejank some ancient layout code, I just reserve the time to get it done.
>>15553 >>15554 Great. I am sorry for any unhelpful error information here. I have tried to improve it, at least it got you on the correct track, I'll see if I can do more to talk about this. It needs it for png since all pngs are scanned by ffmpeg (like other 'video') to see if it is really an apng (animated png). Perhaps your custom ffmpeg didn't have libapng or whatever it is included? >>15557 >>15558 That help link will help you figure out multiple clients. As it happens after years of wanting this, I will be working on 'multiple local file services' over the next four months, hoping to be done by end of Q2 2021. When that is true, you'll be able to have separate partitions for sfw and nsfw on the same client, with good UI support to help you navigate and process that. You may like to wait a bit or just keep an eye on how I get on with that, I hope to make your situation much more convenient soon. >>15559 >>15560 >>15563 >>15565 Sorry for the trouble. Some of my old login scripts are way out of date. Sankaku is an odd site, so it has always had some funny viewing rules and odd CloudFlare issues. I can see the first of your links, 22030284, but not any of the others. They have 'contentious content' tag, which I think is locked behind the login and whatever content viewing settings are in your account. I am going to add the User-Agent header to the Client API so the guy who writes Hydrus Companion can automatically set this data for you. It is sometimes a real headache to get into some sites.
>>15561 With the internal hydrus backup? You should be good just to go again--it'll resume what progress it made. If you have a big client, you might like to try out some proper third party software like FreeFileSync. They can always do a better job than I can do. >>15566 >>15579 These shortcuts are increasingly no longer hardcoded. Check file->shortcuts to set them yourself. You can see the different contexts there, the 'media' one has the most interesting ones. I have played around with showing the shortcuts in UI (the dupe filter has this on some button tooltips), but I wasn't happy with the result. I'll reintroduce it as the shortcuts system matures. I'd like to show them on the menus themselves like you see in many programs, but my command system backend isn't clean enough yet. >>15567 The guy running the PTR keeps ragging on me for this, ha ha. I haven't removed it yet as my database routines don't like it when a new user suddenly adds 10GB to the empty starting database files. It is a stopgap, and I'll eventually get rid of it by default. You can remove it under network->data->review bandwidth usage and edit rules and then click edit default rules and remove the limit off the hydrus service.
>>15574 Is that data.php?id=673110 URL the download URL? 'associable/source' url types are for the bit you see on booru pages where it says 'Source', it is an URL that gets added to the file in hydrus when it is imported, just as a helpful thing to add twitter links and so on to hydrus. You want 'url to download/pursue' type in the content parser. >>15580 Shit, I see a 'hydrus-dmg' link in the artifacts, and when I click it, it thinks a bit and then gives me a .zip. I am sorry, I don't know the details of how this stuff works, I am waiting on these lads to present me with a 'working script' and then teach me how to fold it into my github build process. I just asked the guy on discord who is doing this what the answer is. I gtg for a bit but I'll be back later and post any reply.
>>15594 I don't think anything changed. I clicked off my first page's one image again, then when I click it from that point onward, it doesn't show up in the preview window. My imports only managed to import ones of images. It'e exactly the same as the first time I booted on this version.
>>15580 >>15601 The macOS build guy says that link, https://github.com/ReAnzu/hydrus/actions/runs/590475620 , should be of a working build. The dmg artifact at the bottom should be it, if you cannot see it, I am not sure what is wrong--maybe some weird github login issue or something? Some people have tested it and said it was ok, and things are mostly lined up for me to integrate now. EDIT: I checked that page out on my not-logged-into-github browser, and I see the artifact as text, not a link. I guess you have to be logged in. I hope to have this integrated soon in any case, I am sorry for the delays. >>15581 I am sorry for the trouble, I have difficulty keeping up with my messages. If that page was showing results for that 'similar to' query, then once your client is working ok, hitting F5 on it will refresh that search and load your files back. First we will have to figure out why your client is having so much trouble starting threaded jobs. If you are the same person I see down below, it sounds like there are more issues. If you are not that person, please let me know. >>15582 💀🚋🚋🚋
>>15583 Ah, thanks, that's interesting. There is a 'select subscriptions' button that lets you type, but it works only for query text--is that helpful, or not? Would you like something similar, but for subscription names? If you would like a sub to grab files quicker and more often, you want the 'checker options/timings' buttons in the edit subscriptions UI. Tell it to try to get '1' new file per check instead of 4, or reduce the 'never check slower than' values. >>15585 Thank you for this report. I am very sorry for the trouble. There are still some Window Managers out there that are touchy about some of the ways I access and lay out UI. I am slowly cleaning up my UI code to reduce the problems here, and I know the basic nature of what they are upset about, but it is difficult to pin down the exact problems because the crash is so abrupt, and crash dumps to figure out what was 'going on last' are not simple in python. Although it is not a super helpful thing to say, some Linux users who had a lot of trouble here found running from source was completely fine, so if you are using my built release, and you have some python experience, you might like to try that: https://hydrusnetwork.github.io/hydrus/help/running_from_source.html >>15586 If you click the little 'file import status' icon button on the download page to bring up the big multi-column list of file imports, and then look at the right-most column, called 'notes', what does it say there is the reason for the 'ignore'? No file found, or a big gif problem, or a tag being blacklisted?
>>15587 >>15588 >>15589 >>15590 >>15593 >>15595 >>15602 I am very sorry that you are still having trouble. I changed the way hydrus loads some pages in 433, giving some background jobs their own space to do their own thing, and freeing up 'front-end' resources to do fast jobs like loading sessions. I do not think this is completely related to what you are experiencing, but it is the thing that I changed last week related to jobs. It sounds like jabs are having trouble scheduling themselves. I think looking at help->debug->data actions->review threads may give some useful information. Are there dozens and dozens of 'CallToThread's? How about 'CallToThreadLongRunning', and how many jobs are the Job Schedulers handling?
(138.84 KB 1920x655 Untitled.png)
>>15605 Here's what it says
(149.47 KB 1920x685 Untitled.png)
>>15605 >>15606 I expanded the "type" column to show its full name in this ss
>>15603 >I am sorry for the trouble, I have difficulty keeping up with my messages. If that page was showing results for that 'similar to' query, then once your client is working ok, hitting F5 on it will refresh that search and load your files back. First we will have to figure out why your client is having so much trouble starting threaded jobs. If you are the same person I see down below, it sounds like there are more issues. If you are not that person, please let me know. I am the same person, yes. Also my comment about it being 5 days was meant to be in reference to the page not displaying its contents, not my at all expecting a reply from you (I didn't, as the conversation had ended on your reply, and I didn't reply again). However that post of mine you're replying to was before the latest update that completely rendered my hydrus unusable. maybe the same problem was somewhat present before, though.
>>15604 >Ah, thanks, that's interesting. There is a 'select subscriptions' button that lets you type, but it works only for query text--is that helpful, or not? Would you like something similar, but for subscription names? Well I would love to try this, but in trying to open the "manage subscriptions" window, nothing happened (same guy as the above 3 posts here).
(36.89 KB 1206x469 python_oBY1CloP4C.png)
>>15606 >>15607 >>15608 Thanks. Attached is from my very lightweight dev machine just after boot. All those Gallery and HDD imports firing off are using and cycling through threads pretty quick, and there are 500-odd additional jobs in your 'job scheduler'. Since you are having UI lock-ups, I suspect this is the issue. The changes I made to thread scheduling won't have affected your situation, so this is something else. It could just be those galleries or hard drive imports are catching up on some older work and having trouble slogging through the backlog and also keeping UI updated. I can work on this and smooth things out. Normally, I tune this stuff to work as fast and hard as possible, but since you have a couple hundred, I expect it is lagging out now it wants to do work. If you can, please now hit up help->debug->data->show scheduled jobs. This will make an absolutely gigantic popup window and print data to your log. I do not need to see everything, but can you confirm for me that it involves a lot of 'REPEATINGWorkOnGallery' and 'REPEATINGWorkOnFiles'? If you would like an immediate relief, my best guess at the moment is to try to pause as many importers as you can. That should turn off these jobs, or at least reduce their overall CPU.
>>15610 I might not be able to open what you say, because of the same reason I can't currently open a subscription window. I did it but nothing happened. The subscription window hasn't opened for 30 minutes now.
>>15610 Well I had 13 import tabs and 18 gallery downloader importers active. I paused the gallery downloader- Actually my subscription window just opened. Maybe the window you asked me to open will open eventually after all. Also I guess I will go ahead and pause my import tabs as well. Maybe it helped.
>>15610 Well. I slept, woke up. That window you asked me to open never opened. My client isn't frozen, but still, for my first page that is always the same search query that returns a single image, when I click it, it doesn't show up in the preview window. Also my client just froze. I didn't do anything except switch to an import page and unpause the import, which imported a single image (without loading the thumbnail), then it didn't import any more after several seconds, so I paused it again. I then switched between pages and noticed that one page had 10 successful imports, and one visible thumbnail. I clicked on what I thought was just blank space with no image, but then it started loading the thumbnail. It was at this point the client froze. It is still frozen, minutes later. This was probably exactly what caused the earlier freeze, because it froze on an import page as well (however, no thumbnails became loaded when it first froze, nor did it say at the bottom of the client that any images were selected. So at this point I have no hope for this ever unfreezing. The last time it froze my client had been frozen for something like 18 hours before I just force terminated it in task manager.
>>15604 >If you click the little 'file import status' icon button on the download page to bring up the big multi-column list of file imports, and then look at the right-most column, called 'notes', what does it say there is the reason for the 'ignore'? No file found, or a big gif problem, or a tag being blacklisted? It says "could not find a file or post URL to download!"
>>15610 Hydrus dev, my client has been literally unusable since the latest update. Even if you think you see parallels in behaviour since before the latest update, at least I could use the client before. Is there anything you would like me to provide, because I really don't like the unexpected random extended downtime caused literally by just updating. I operate based on always having access to my database. For me, this puts my everyday workflow to a halt. And I don't see any end in sight. My client is currently frozen, as it has been for the past several hours, but at no point since updating have I been able to use hydrus for a single thing.
>>15615 Well... I force terminated it for the second time since updating, because it indefinite freezed again, of course... I will boot it and try again to open the window asked of me, but considering it didn't open the entire time I slept, I have little faith.
>>15616 Again I want to mention that right after it boots, thumbnails of the pages I already have open load fine, and when clicking on them, they appear in the preview window. But after hydrus finishes booting, this stops.
(86.05 KB 1567x590 Untitled.png)
>>15617 Ok. So after doing like I said I did (force terminating the frozen client in task manager, and trying to open the window hydrus dev asked of me) nothing happened in the hours between then and now. But what I did differently this time was close my import pages, and NOT re-open them this time. My formerly active "gallery downloader" page entries were still paused from before I force terminated the client. So interestingly, because not a single import had occurred since program boot (somehow not even a subscription), I could still click away from the image already loaded in the pages I had open, and they still displayed in the preview window, unlike what was happening before. My client appears to be behaving completely normally right now (searches work, double-clicking my "gallery downloader" page entries show their results, tagging works, opening "manage subscription" no longer takes 30 minutes, etc). The only prroblem is that I only even booted to open the window hydrus dev asked me to open >>15610, and it didn't work. Also, of course, that trying to import anything breaks the program in the weird way that's been happening since boot. I took another screenshot of the older window hydrus dev asked me to open, that I can manage to open. Again, the newer one he asked of me had not opened for hours, even though I'd booted hydrus for no reason other than to open it. I just tried it again for good measure, and it still did not open. At this point all I can do is revert to a previous version. Unless there's some other feedback I can provide on this version. Again, as far as I can tell, importing breaks my client on this version, rendering it entirely unusable for any purpose.
>>15618 Somehow an import I tried just worked. It was one of the folders of images I was importing on boot. I don't know why now it's working. Maybe all imports work now, so the client may be back to normal? I'll try the rest.
>>15619 To clarify, when I say "I was importing on boot", I meant the first two times I had launched the program, I had imported them on boot. This current time I had closed the old pages entirely, as well as paused active importers in my gallery page, until four hours later. Now (four hours after boot without having imported a single thing, not even a subscription somehow), importing the same folder works.
(609.64 KB 1778x1050 Untitled.png)
>>15620 Hydrus dev, it's gonna seem like I'm trolling, but a large portion of my client has turned into a black bar, where if I click anything there, hydrus loses window focus. Even the parts of the black bar covering the minimize/ maximize buttons of the window behave the same way. I have never seen this before. I didn't even know it was possible for a program to behave this way. This happened after I imported 12 folders at once, all with images from different galleries I'd ripped using gallery-dl. The next time I checked on my client it looked like this. Well not right away, because I had tagged a few images before. But the time after that, it looked like this.
>>15621 Ok, I fucked it. Now I feel like it will definitely seem like I was trolling. When I expanded my window size, the black bar that removes window focus went away. I doubt I will ever be able to reproduce it.
>>15621 >>15622 Actually the black bar happened again as I was tabbing in and out of the client and doing stuff. It might be easily reproducible on this version. For me. I don't know why I end up being the guinea pig for this type of stuff.
>>15623 By the way, I unpaused an entry in my "gallery downloader" page, and it imported successfully, without breaking my client anymore. So at this point, I think the problem was evaded (besides the reproducible black bar that removes window focus). Apparently trying to import while the client is still in its booting stage is a no-no. But beyond that (after I slept and woke up without a single import happening), it seems to work fine. Besides, again, the black bar that removes window focus. But I haven't had a single subscription happen yet. It's been several hours, my sleep and then my wake, and it's not like my bandwidth was being occupied by my gallery "downloader" page. It seems subscriptions probably just won't work for me on this version.
>>15621 lol. the only thing that comes to mind is that this is the area where hydrus shows notifications. have you tried right-clicking the lower right, to close any notifications that might be there?
>>15625 The only notifications I got since booting this time were from having over 165 pages open on boot. I haven't gotten a subscription or anything since. You can compare the scale of my notification thing for my window size >>15453, and it doesn't seem the same, but it was a good idea. If I click the area where the notification would appear, it's as if I'm clicking nothing, so it's not as if it's there in a phantom way, or anything. Also the box that removes window focus disappeared when I changed window size. So It must be different. But it did appear twice for me so far.
>>15626 > >>15453 Holy crap those are some gorgeous legs, looking up egk513 right now, damn! XD No idea what else it could be. The media viewer is a separate window, so it seems unlikely it causes this
>>15627 I know, right. lol. But yeah to me this was completely fucked. It seems like a huge exploit if programs can just intentionally cover the minimize/expand/close buttons of the window with a black box they generate. Just what the fuck.
Made a babepedia downloader. The site doesn't have much, but it does have some rare pics.
I have some feature requests regarding ratings, that I had on my mind for a while: Could we please get the rating names displayed next to the actual ratings in the media viewer? I've been trying to figure out what setting is wrong that causes rating names to be invisible. Now I checked the help and saw that apparently they are intentionally only displayed in tool tips? Alternatively allowing custom icons could be used to differentiate them better. Another thing that would be great is a counter-type rating, that can be pressed up/down repeatedly. Think of counting how many times you came to a pic, or for the meme lords how many times you posted a particular pic. The last thing I've been wanting for a while is a more advanced sorting system. I use a bunch of anatomical fetish ratings that I subjectively use to rate attractiveness and visibility of the features, e.g. that pic from >>15453 would get 5/5 for legs, 0/5 for boobs 0/5 for midriff, etc. In addition I use an overall "liked" rating, and even a "happiness" rating for those days where I don't want to see sad stuff. Currently if I open a search, I can only sort by one of the ratings. The problem is, let's say I sort for the best boobs, I currently have 461 pics with a 5/5 rating for epic boobs, that will all be at the top. Ideally I want to be able to add additional sorting criteria, like the "liked" rating, and image resolution. The criteria could all be added together to give all images a sorting score. Then the different criteria could have a multiplier displayed next to them to set their influence on the score.
>>15630 >Another thing that would be great is a counter-type rating, that can be pressed up/down repeatedly. Think of counting how many times you came to a pic This is easily the hottest thing I've read in months, thanks for that.
I follow 4000 accounts on Pixiv, is there a way to extract all the users IDs automatically so I can add them to a Hydrus Network subscription?
>>15601 >Is that data.php?id=673110 URL the download URL? 'associable/source' url types are for the bit you see on booru pages where it says 'Source', it is an URL that gets added to the file in hydrus when it is imported, just as a helpful thing to add twitter links and so on to hydrus. You want 'url to download/pursue' type in the content parser. Oh, I see. The file post downloader works now. I have a separate problem with the gallery downloader though. There's no way to tell if an aryion URL is a gallery or a post, just based on the URL, unless it has a page parameter. And when you just click on a gallery, the page parameter won't be in the URL until you click "next page". So if you want to download from a gallery, you have to make sure it has a page parameter, and if not, manually add "?p=1" to the end. Otherwise it will match the URL class for post URLs and not find anything. The other option is to just combine the gallery and file URL classes as a file URL class and use content parsers to generate the "next page" URLs instead of using the gallery URL class. But then the gallery page URL will end up getting saved as a URL for the image, which isn't ideal. (I thought setting the "next page" content parser as producer of a "sub-gallery url" would solve this, but it doesn't seem to work at all if I do that) I guess this can't be fixed unless you change how URL classes work. Unless I'm missing something? For now I'll go with the former. It's slightly annoying but it works.
>>15633 Also, it would be nice if you could test what a URL would import without actually importing things. I don't want to download the entire contents of a tag just to make sure my parser works.
>>15633 I guess it's done. contains: >a downloader for tag searching >a downloader for a user's gallery >url class/parser for galleries (grabs all subfolders and pages) >url class/parser for posts (grabs artist username, source time, and tags) Once again, it's important to note that if you want to do a url import on a gallery, make sure that the url contains "?p=1", or else hydrus will think it's a post instead of a gallery and won't find anything. The simplest way to do this is to just click next page and then click previous page. But if the gallery has only one page, you'll have to do manually add that bit yourself.
>>15610 Hydrus dev, I don't need to reply to you again since I'm not adding any new information. But just to summarize: -After booting my client since updating to latest (433), I imported things immediately on boot as I normally do, and after the first few minutes it became incapable of loading thumbnails. For pages where a thumbnail was already loaded, selecting it no longer made it appear in the preview window. Only ones of images (<9) imported per page, with some pages not having successfully imported any. -After you saw my posts complaining about this, you suggested I pause my imports from my "gallery downloader" tab, as well as any others. I did so for my third boot since updating. After going to sleep then waking up without a single import happening, my client appeared to be usable as normal, save for a black bar that removes window focus upon clicking inside of it appearing every so often (>>15621), but the black bar can be removed by changing the resolution of my hydrus window. Also, even though in the first two boots since the update subscriptions had succeeded, on the third boot, I have not had a single subscription notification yet. -the window you asked me to open and give you feedback on in the post of yours I'm quoting has still not succeeded at opening, despite my only booting the client this third time in order to try to open it. I had also tried opening that window after my second boot of the client since updating, though far less time had elapsed. By this point, it has been over 24 hours without the window having opened
>>15632 Use PixivUtil2 or some similar pixiv downloader and write a script for converting json or txt containing tags to the one hydrus can understand. I'm using pixivutil2 and wrote this script. https://github.com/nasu-no-suage/PixivUtil2ToHydrusNetwork This script is rough but works.
sorry if I'm missing something, but is there a way to make my scroll wheel (or any other button) control the volume of the preview window? I've looked through all the shortcuts and shortcut sections but didn't seem to find this option.
>>15631 >This is easily the hottest thing I've read in months, thanks for that. You're welcome! XD
When sorting files, if two files have the same value for the sort criterion, what determines the order?
How do I automatically tag anything with audio, with system:audio or medium:has audio (siblings so it doesn't matter)? Same with video. For example, if I import a video with audio that is completely original (no tags), how do I get Hydrus to automatically apply the video and audio tag to this? Now do that, but for any video/audio content I download/import. Looking at the content I just imported, there are no video or audio tags. I have to add them manually.
>>15641 I'm not sure about audio, but you can sort by mime to find videos. To do that, just double click on system:mime in the tag list and select video
>>15642 >but you can sort by mime to find videos >To do that, just double click on system:mime in the tag list By the tag list, you mean the tag list (search) you use to search for anything else correct? When I search system:mime, I get zero results. Also I wasn't exactly looking for a way to search them, but tag them by default.
>>15643 >By the tag list, you mean the tag list (search) you use to search for anything else correct? Yes. If you have nothing entered in the search box, you should see a list of system tags. Scroll down in that list and you will see "system:mime" >Also I wasn't exactly looking for a way to search them, but tag them by default. I don't think it's possible to use system tags in a parent-child relationship. I'm not sure why you would want a separate tag for them though when the system tag already exists
>>15644 Ah, I think I found it. It's not called system:mime, it's called system:filetype. Also one issue I noticed is these system: tags are not searchable through the Client API (I think?). I use LoliSnatcher Android and they don't work. Although the developer hasn't got Hydrus searching completely functional due to it using spaces instead of commas, however I don't think this is the issue. He also stated that it was not possible to use to search from the Client API from what I can tell. He claimed it was a Hydrus issue (unrelated to the spaces/comma issue). https://github.com/NO-ob/LoliSnatcher_Droid/releases
>>15644 >I'm not sure why you would want a separate tag for them though when the system tag already exists Understandable, however these tags are in heavy use in the PTR. Imo anything recognized by the system with audio, should automatically apply the ideal audio tag so they get synced to the PTR. Same with video.
(3.97 KB 338x145 hydrus_error.png)
I left hydrus on over night to download/process but when I woke up trying to run the client would give me this error. In addition the shortcuts no longer work.
(21.79 KB 838x435 Untitled.png)
For a while I thought I was imagining things. This didn't start right away, it has probably been present in the last version. but just now I actually kept note of where it was before, and have for the first time observed what is happening, which confirms that I'm not just going crazy. Every time you open the "review files to import" window, the "path" bar increases in horizontal length by a few pixels. It just keeps doing it infinitely. When you import enough, the actual buttons you press to import end up partially off your screen, if not entirely off your screen. I had been absentmindedly wondering "what the fuck" and correcting it/ scrolling horizontally to see the buttons again. but this time it was so gratuitous, that, again, I just thought, what the fuck. So I reset it completely and observed it.
I look in my task tray, and I see the Hydrus Client icon, then it just disappears. I look in processes, it's not running (it was), and then I search for it and it's not appearing? I look in the HydrusNetwork directory, and don't see client.exe anymore, I see server.exe (and server shows in search). But it just randomly disappeared?? I'm going to restart my machine, but like tf.
>>15649 Restarted, yeah still gone wtf. Man fuck Windows. I guess I can try to reinstall Hydrus and hope it detects all my previous files as they're still there. I have a backup but that's a pita. Idky it randomly deleted the client exe.
>>15647 Oh maybe this is partially related to the issue I'm having.
>>15647 >>15649 Ok, I reran the installer and after it installed, I say open it, and it says unable to execute file client.exe. Create process failed, code 225. Operation did not complete because the file contains a virus or potentially unwanted software. I take it Windows defender deleted the client.exe file. Not sure why it did it randomly, because I was running v433 for like a few days no issues. I assume (hope) this is a false positive with the python package? We really need to figure that out.
(43.14 KB 1824x853 client_bARN5yNmhI.png)
Hey everyone, I'm new to subscriptions, and I'm setting up my pixiv subscriptions in Hydrus. I'm finding that it's only downloading sfw images (images you don't need an account for), despite it showing that my login is good in hydrus. For accounts that have no sfw images, it states that the subscription is dead. I can drag and drop these pixiv links manually and it works fine. Any suggestions?
>>15655 I had the same problem when I re-activated my old hydrus install. It used to work with just the login, but it seems something changed. I managed to fix it by sending it my browser cookies using the hydrus companion extension. I'm not aware of the official link to that. Maybe someone can link to it.
Also regarding pixiv, was there any progress regarding ugoira in the past year? I see it still displaying "ignored" with reason "ugoira veto", which really sucks. For now all I want is for it to download the zip files, as long as the posts are still there, even if I can't view them internally. Can I disable that stupid "ugoira veto" somewhere?
>>15656 >hydrus companion extension You mean this? https://gitgud.io/prkc/hydrus-companion >>15657 It's not the answer you're looking for but I always manually download them using ffmpeg and use the --ugoira-conv flag to convert it into a .webm, which I then import into hydrus.
Could download support please be added for https://ofans.party/
>>15658 sorry, gallery-dl, not ffmpeg
I'm having trouble pulling images from Sankaku Complex. I'm trying to import images that I've save to my favorites by using the tag "fav:[my account name]", but it's just ignoring every file. Does anyone know what's wrong?
>>15661 I know this barely helps, but this is exactly what I'm doing (subbed on my own fav:<me>) and it works great. Have you read and tried >>15266 already?
>>15661 Huh, restarting the client fixed it. For anyone wondering, I was trying to use the gallery downloader for Sankaku Complex (Channel, not Idol), and I'm on version 433.
>>15662 Thanks for pointing that post out, I hadn't read it. Does this mean I'm missing some files by not using the beta search?
>>15664 You're potentially missing some images (IIRC everything old-ish (a month+ old, maybe?) tagged with contentious content), not because of the beta thing (>>15270 corrected me on this one) but because you're probably not using the logged-in API using the default downloader. Note that if you decide to use the API downloader (the one in the message I linked), you'll need to re-login every day to download more, which may get annoying very fast. If you forget to re-login and try to download more, you'll get issues like >>15559, which can be easily fixed with >>15560.
>>15665 Hmm, it's pulling files with the contentious content tag without me doing anything different. I'll compare the number in my favorites to the number in Hydrus once it finishes. Thanks for the info.
>>15655 Yeah like the other guy said, you need the hyrus companion extension. Another thing to note is that exporting cookies via a 3rd-party browser addon (not hydrus companion) worked for me before, and was how I did it, until it one days stopped working. I had no idea why. I tried everything, resetting password, relogging, nothing worked. I eventually tried the hydrus companion extension and it worked. But by this point I had already missed images I was trying to download. it sucked. But the point is, the hydrus companion extension is what you should use, without a doubt. >>15658 >You mean this? https://gitgud.io/prkc/hydrus-companion Not him but I actually don't know how to get the direct link to the addon by viewing my install of it. Does a link to it not exist in the documentation, such that you don't have to ask random users if you found the correct link?
>>15647 >>15651 >>15652 >>15653 >>15654 >>15649 >>15650 Christ, I am sorry for the trouble. I just answered some guys on twitter with the same deal. We keep getting nailed by these, the past three or four months. I don't know if something changed in my code to start thing, or if the anti-virus guys rolled out some new definition tech that just doesn't like hydrus. Every three or four weeks now, some new randomly-named virus is detected in client.exe, sometimes different names for different users. So far, these have all been false positives. I believe this is likely the same. My best guess for all this is that hydrus does network and file accesses, a little ghetto crypto, and some optional UPnP tech, and some combination of that looks bad to Windows Defender (and regularly sets off some Chinese detectors, if you look at the big list on VirusTotal, check the links in this thread if you'd like some examples). The good(?) news is these flags tend to clear after a week or two, when the next set of definitions roll out. You can try whitelisting your hydrus directory, but obviously that is not an ideal solution. The nuclear option is to run hydrus from source, but this is too technically complicated for most users. I am going to explore signing my executables to see if that can raise my 'professional' profile in any way for these programs.
I had a good week getting started on the network updates. I cleaned out ancient code and improved some admin-side workflow. There is plenty more to do. The release is still expected to be another two weeks, for April 7th.
>>15663 >>15666 I imported a few thousand files, but around 30 that were in my favorites didn't even get detected. Does anyone know if there's a way to find out which files Hydrus couldn't see? I was still able to import of bunch of loli and toddlercon, so I'm confused what isn't showing up.
>>15670 Maybe they are duplicates of files you already have in your db?
>>15670 I didn't read any of your previous replies at all, but I remember when I first installed hydrus, it didn't allow "decompression bombs", but hydrus dev has since diasabled that, so that isn't relevant here. But otherwise, I had dragged a folder into hydrus as a test, then I deleted everything because I wasn't convinced on hydrus yet. I later re-imported everything, but some didn't make it through, because I had previously deleted them. Unless this has changed over the years, you have to enable importing previously deleted images.
>>15671 It lists the files that are already in the database as such, and most of them were already in the DB. It still counts toward the total number. >>15672 "exclude previously deleted files" is already unchecked.
>>15667 >Does a link to it not exist in the documentation If it does exist it's somewhere deep in the middle of a page. I looked through a lot of pages and didn't find anything.
Maybe I'm missing something here, but I think there's a bug in the URL class creator; it's unable to recognize a '#' as a component of a URL, so it won't let me create a class ("Please enter an example url that matches the given rules!"). I'm trying to create a downloader for ofans.party, but that website has a https://ofans.party/#/creator/<creator here> format; removing the '/#' returns a 404, so I can't cheat it like that. What makes me believe this might be a bug is that even fixed characters, the error message is 'Example does not match - "" did not exactly match "#"'; notice the "", maybe Hydrus just does not register the '#' and matches an empty string. Let me know if I can help further or should provide a screenshot. Thank you!
>>15635 I added a parser that grabs the title of the post. Here's the updated downloaders.
>>15496 Replying to this message for a IPFS feature request: a way of storing a CID as hash, directly from the parser's file hash option (next to the existing md5 / sha*).
>>15603 >If that page was showing results for that 'similar to' query, then once your client is working ok, hitting F5 on it will refresh that search and load your files back. I did want to say that I was surprised this worked. I thought if you reverse image searched, that if that page didn't load contents on boot, you could never know what it ever was. But it actually worked.
>put 499,940 urls into hydrus >it crashes wtf bros how could this have happened?
>>15614 I never received a response for this. Does anyone know what the issue is?
Hydrus can recognize if a file has been in its database before. I know there's the "exclude deleted files" option you can uncheck, but is there any way to delete all traces of a file ever being in the database?
Duplicate comparison currently sucks for very similar videos, and also for cropped pics and pics with different sizes. Here's some stuff that could fix it >if you pause a video is stays paused, even if you switch away and back. That allows you to compare individual frames. >if you zoom in on both videos and images is stays zoomed in even if you switch back and forth between images. That allows you to compare individual sections. Currently if the size is different it resets every time you switch. >same but for placement as in some rare cases the image is not centralized by default Here's two videos that have all of those problems (except maybe the second one).
>>15610 >>15636 Actually I want to amend this to say that I actually have been getting subscriptions the entire time. I thought since the popup bottom right telling me about my subscriptions wasn't visible that it meant they haven't been working for me on this release, however they have- I only randomly discovered as much. This validates >>15625's theory that the black-bar-that-removes-window-focus-when-clicked-on is a consequence of the notification button, since they occupy roughly the same space.
(43.06 KB 828x531 Screenshot_20210326_085059.png)
>>15681 >but is there any way to delete all traces of a file ever being in the database? Yes. Go to the Menu "services" and click on "review services", then click on "clear deleted files record".
>>15684 Thanks.
Any plans on having a docker container? There are currently only 2: one is read-only and the other is way behind with some weird patches.
>>15686 What's even the point of having a container for a software like hydrus?
>>15687 Ease of use, really. I have my docker-swarm with traefik for reverse proxy. I have my torrents, audio and video servers there. I want to use Hydrus for all of my images, so when I need them, I can connect to hydrus.my_server.com and browse everything, posting the correct reaction image or forwarding an old memory to my dad. If that’s not the appropriate use for Hydrus, could you please point me to the more appropriate solution?
>>15687 I have a virtual machine dedicated to Hydrus, and I run it as a remote X application.
>>15614 >>15680 Thank you for the extra information. 'Could not find a file or post URL to download' means either: -the parser is unable to find any URLs in the Post page. -you do not have any URL Classes for the URLs given. Since this is happening on the actual file step, I think it must be the former. For some reason, your gelbooru parser is not pulling the info we want from a page. Please hit downloaders->downloader components->manage url class links and see where your 'gelbooru file page'. By default, I think it should be on 'gelbooru 0.2.5 file page parser'. Is it something else for you? You should be able to change it there. 'danbooru file page' should be 'danbooru file page parser'. If it all seems correct, please copy a gelbooru post URL and hit downloaders->downloader components->manage url classes, and then paste the URL in the test area at the top--does it register as the right 'gelbooru file page' url class? If there are conflicts and things are being recognised wrong, you may need to get into the muck and delete some duplicates or whatever seems wrong. If everything seems lined up correct and recognising properly, are you on an old version of the client? Sometimes I update parsers, so if you are 20 versions old, just updating could help. If you are on a new client, we would then be looking at problems with the parser itself or your cookies with gelbooru. Maybe you are weirdly cloudflare blocked or similar. In this case, please hit downloaders->downloader components->manage parsers, zoom into the gelbooru/danbooru post page parsers, and experiment with the 'test' panel and some data fetched from example urls--does it seem correct there, is it getting tags but no URL, or is it getting nothing? If your download objects are all fucked up due to a new downloader import you did that went wrong, just deleting all the url classes and parsers and then using the 'add defaults' button to reset is a decent nuclear solution.
>>15636 >>15683 Thank you, I have read all your posts. It looks like some things are really fucked up with your client, and I am sorry again for the trouble. I am honestly not certain what is going on, especially with the black appearance suddenly (unless help->dark mode was accidentally activated), although that could be an artifact of the UI just failing to keep up to date with some drawing updates. The one thing I am convinced about is your client is not happy with simultaneous action of some sort, either the gallery work loop or imports. Please keep them paused/running slow for now. I will have to write some debug routines to better display what is going on here. I will also work more on thread scheduling options so you can automatically throttle this work behind the scenes. Even though your session is pretty big, I know several users who run much hotter than you, hundreds of queries on half-million sessions, who do not have any of your symptoms. I am less confident this could be a case of 'oh, this computer is a little slow, so it has more lag', and more confident there is an unusual additional overhead occurring in your client. Something like an anti-virus that is spending five seconds scanning every file import or similar. I will just have to improve my debug code so we can see what the hell is dying here. I am up to my waist in these network updates at the moment, so I can't promise I'll have any solid work done on overhauling thread scheduling done for 434, but if I have time, I'll check it.
>>15630 >>15631 >>15639 Ha ha ha, yeah, I want to add a rating that is just a number, no star limit. You can add shortcuts now to do increment/decrement for a rating, but you are stuck in the 'star' system for now. A lot of this code is still functionally a prototype. I hate how it works behind the scenes (just a bunch of floats). I also keep telling myself to add some rating services by default, as many new users don't even notice the system, but then I regret the rough edges and tell myself to improve it first. It is another thing I need to schedule time for. I don't like the way ratings stack in the top-right hover. I agree that more layout options, with labels and so on would be nice. Several users have asked for variants of this. I will write it down properly in my todo. I really like your idea of the clever sort. For a long time, I have been wondering about making a system and UI for computing various scores for media. Basically 'if (property), then add (value)' where property could be all sorts of metadata about the file. This idea was originally for users who wanted to colour thumbnail borders based on different properties, but if I had the system and UI ready, I could plug it into other places, like an arbitrary weighted sort. I'll refine this idea a bit, and maybe one of these medium-size weeks I'll try something out.
>>15633 Ah, man, so the gallery and file URLs are very similar? I am not familiar with the site, but I guess this is not the sort of thing you are grabbing?: https://aryion.com/g4/tags.php?tag=Big_Breasts https://aryion.com/g4/view/680174 In this case, you can check for the 'view' path component in the URL Class to determine which is which. But if you have a site where post urls and gallery urls are very similar, this is tricky. EDIT: I just looked at your downloader and saw: https://aryion.com/g4/view/390220?p=1 Yeah, that sucks. I understand about the optional 'p' parameter, too. Hentai Foundry does a similar thing. I think the best solution for now is to just require the 'p' parameter for Gallery URLs and generate p=1 in a GUG. Make a little explainer in the GUG text like 'enter pool id here', and trust the user will put the right data in. There is a very advanced hacky solution that can let you kind of support two parsers in one (using two subsidiary parsers and 'veto' content parsers in each that turn off one XOR the other parser based on what the parsed page looks like), but this is undocumented and I don't like it. For testing, your best shot at the moment is the test panel of the manage parsers dialog.
>>15638 Not right now. My shortcuts system is incomplete, sorry! I will be integrating all the mpv controls into it in future. >>15640 There's a secondary sort under options->sort/collect. Otherwise, they preserve whatever their previous sort was. For the first time they are sorted, that initial sort choice will be pseudorandom. >>15645 system tags will be added to the Client API in the future. A bunch of the work has already been done by someone else, I just need to find the time to get it all going. >>15646 When I add the tag filter in these network updates I am currently working on, those 'medium' tags may be removed from the PTR. Some users like them, but they are not too useful for sharing. I may add auto-tag systems in future, using the arbitrary metadata scoring tech here (e.g. "if (has audio), then set tags 'has audio'") >>15692
>>15657 A user was telling me the other day about working on a pixiv downloder that uses a third-party site to convert ugoira to webm or something. When it is ready, I expect him to put it up on the github here: https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders/Pixiv >>15659 I don't make downloaders personally any more, but many users do. A bunch hang out on discord if you would like to ask them, if you are comfortable with that. https://discord.gg/wPHPCUZ Otherwise, most user-made downloaders get posted here: https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders >>15674 All the add-ons are now listed at the top here: https://hydrusnetwork.github.io/hydrus/help/client_api.html I will put a link to that page in the help menu.
>>15675 Yes, for now, the fragment # is not parsed. You can keep the fragment in the saved URL using the 'keep fragment?' checkbox. I will be adding this (for Mega URL preservation). Man, having a # in the URL like that is pretty weird though! That is technically an invalid URL I think (anything after # is not sent to the server), unless there is a funny rule about '/' following the '#'. Can you try the url-encoded value instead, %23? https://ofans.party/%23/creator/<creator here> I think that is how it is being handled in your browser address bar. Hydrus may still sperg out at that though. Let me know if %23 doesn't work, I may have to hardcode in an exception. >>15677 Thanks. This makes sense. I don't know the format of CID--is that the new hash they use? What does it look like? >>15678 The hash of the file being 'similar to'-ed is stored in the system search predicate. >>15679 :^) >>15681 >>15684 >>15685 You can also turn on the 'advanced file deletion dialog' under options->files and trash. That dialog lets you delete a file without leaving a deletion record when you do it.
>>15682 Thanks. I would like to add some 'position/zoom locking' tech, similar to the duplicates filter, to the regular viewer. Adding in video view progress memory is a great idea to add to that. >>15686 >>15687 >>15688 >>15689 I am an MS-DOS boomer, and dockershit is unfortunately space voodoo to me. Feel free to package my code however you like, it is all free for any purpose, but I am afraid I cannot offer any official release or technical help with this stuff. I know a guy who runs hydrus using some virtual container gubbins, but he does have some crashes, seemingly due to the odd window manager situation. If you try any of this out and there is any way I can help, like some different PATH initialisation or whatever, let me know.
(18.58 KB 300x344 1616836689613.jpg)
This is actually very cool. Good work, OP!
>>15696 >the format of CID--is that the new hash they use? What does it look like? I was hoping you'd ask, so I researched it in the meantime – it's kind of a mess, honestly, so I don't know if it's a good feature request or a waste of your time anymore. Maybe it's still interesting in the future, when you add more stuff to the IPFS features? It's pretty well documented, but there are two versions of the thing already (and you can transform v0 to v1 and sometimes v1 to v0) and they look nothing alike. https://github.com/multiformats/cid#decoding-algorithm has the general documentation (I'm linking to the decoding phase here because it's the best overview IMO), and there's the Python implementation https://github.com/ipld/py-cid which may help here. I'd also point out the inspector which I found nice – https://cid.ipfs.io/#QmYwAPJzv5CZsnA625s3Xf2nemtYgPpHdWEz79ojWnPbdG . In this example, I'm giving it a v0 CID (they're still the most common; but they're case-sensitive which sucks) and you can see it transforms it into a v1 CID (case-insensitive) at the bottom of the page. >having a # in the URL like that is pretty weird though! It sucks, for sure! It took me by surprise as well; it seems to be running nginx but I'd be curious to see how it all works behind. Unfortunately replacing # with %23 seems to mess with the server's way of handling requests. Don't worry about it though, I'll just skip the URL class for this one; they have an API that has normal-looking paths, which is enough to have galleries / subscriptions working. That's already most of what's needed; I'll clean up my work and reply to >>15659 later with it. Do you know if posting a PNG here is enough to add it to the CuddleBear92 repo? Is CuddleBear92 lurking in the thread or should I join the Discord for this? Thanks for your help as always!
When you're downloading files from a gallery downloader or from subscriptions, is there some way to stop files from being imported if they match a certain search (or at least if they have a certain tag)? For example, could I have a subscription that skips downloading files that match "female only AND duo AND black hair" in the PTR, or something like that? I can't just put those things into the queries that the subscriptions and gallery searches do, because of term limits that the websites have, so having that kinda control client-side would be helpful.
>>15700 NVM I just found it, though it only works on single tags and not entire searches. A suggestion though. I wasn't able to find it originally because I was looking in file import options, but the blacklist setting is in tag import options. Since this affects what files get imported, I think it makes more sense if it was in file import options instead.
>>15700 One more question actually. The blacklist and whitelist settings both talk about the tags blocked and allowed being "on the site" and "from the source". Does this mean that it only checks if these tags are going to be imported from the website, and it doesn't check if these files are tagged with these tags on your client? If that's so, then it wouldn't really be useful for websites that don't really use tags, like imageboard threads, since they're not going to be importing tags at all, so the list settings will never apply.
In the duplicates processing filter, how do you make it show both images at 100% zoom instead of scaling the smaller one up with a shitty nearest neighbour filter to overlap the area the larger takes up? I get the intention, but it makes it really hard to judge quality. What if the smaller image is a png and the larger is a 85% quality jpg? You wouldn't be able to tell that the png is better because it'll look like some blocky piece of shit.
>>15691 For what it's worth, I restarted the client and the notification button started working as usual. No black-bar-that-removes-window-focus-when-clicked-on anymore.
When you download from Pixiv and certain URLs are skipped with the note "Ugoira veto", are the URLs saved to some global list? I'd like to be able to get every such link that was vetoed.
(8.13 KB 512x142 ofans.party-2021.03.28.png)
>>15659 There you go – note that this only works with gallery or subscription mode and if you're using the main site, Hydrus won't be able to recognize the URL. Also note: this downloads from IPFS. What this means is that the content is distributed P2P, but I'm setting a gateway in the parser for compatibility (so you don't have to host your own node), specifically https://ipfs.io/ipfs/. Sometimes, the gateway won't have the content right away and may return a 504 because it did not get your content fast enough; try it again and it'll work eventually. You can also choose another gateway, but you'll have to enter it manually in the parser, so I wouldn't recommend it. If you still decide to look into it, a list is available at https://ipfs.github.io/public-gateway-checker/.
I'm not extremely versed in docker so I have a question, if I download Hydrus in docker, can I store the files elsewhere on my filesystem? For example, I run docker in the container, then say I want all the database and files stored on the system in /home/user/pictures/lewds and then Hydurs monitors that. Also add directories from around my system like /home/user/videos Can I do these things with Hydrus Docker? Instead of everything being stored inside the container. Also on a sidenote, why is the Hydrus Docker image read only? is this normal?
could a web admin be added to hydrus or the ability to connect to hydrus from another computer running hydrus and import files, download files, view files, etc. but all the downloading and stuff isn't on the local machine, it's done on the machine you're connecting to. like if i want to manage my hydrus client on desktop, from my laptop, i could easily open a web admin page on my laptop, or go into hydrus install on my laptop and connect to the ip on the desktop then manage it. without using third party web clients or anything, built in i mean.
Why aren't files that are "previously deleted" counted as part of the trash file repository? I sometimes do gallery downloads, then use the number of archived files versus the number of trashed files to judge how much I like the files in that gallery search, but if a file has been deleted from the trash, it's instead counted as being a remote file (I don't even know what that is). Is there a way to see the number of all trashed files (including previously deleted) and select them together? The way it is, it's more difficult for me to figure out which searches are worth making subscriptions, because I have to remember to check 2 different things and add them together before comparing them with archived files.
>>15695 >A user was telling me the other day about working on a pixiv downloder that uses a third-party site to convert ugoira to webm or something. When it is ready, I expect him to put it up on the github here Hm, personally I want to archive the original files though, not lossy conversions. For now an option to disable that annoying ugoira veto would be enough for me. I can figure out how to view those files later. Maybe a shell script that unpacks them to a temp folder and opens them using djv view. The point is, artists seem to delete their accounts all the time, and I would like to archive this stuff while it's still there.
>>15710 Not sure if it's what you want, but the last time I used PixivUtil2 it downloaded Ugoira's as zip-files of jpg's.
My power went out in the middle of some gallery imports. I turned my computer back on and restarted Hydrus and its imports. Can I expect that all the images would be imported, or is there a possibility that Hydrus thinks some URLs have been visited but the images are not saved to disk?
Is there a way to force a subscription to keep trying until it gets all files? Sankaku, and at rare times other sites, keep closing down after each subscription after I've downloaded enough files. The only way to restart it is to clean the bandwidth usage, open and then close the subscription menu.
>>15713 You want to use the gallery downloader to get all the files first then subscribe to your tags; that's the preferred method anyway.
>>15714 It still needs to check the URLs, and even that process keeps stopping every time I set it to check my subscriptions.
I had another good week grinding on the network code. I fixed and cleaned old admin functions and started on future extensions for these systems. The main tasks remaining are improvements to the janitor petition processing UI and new features for core repository storage like a tag filter and dynamic update times. Unless I really fall into a hole, the release is still expected for next week, for April 7th.
>>15691 >>15604 >Ah, thanks, that's interesting. There is a 'select subscriptions' button that lets you type, but it works only for query text--is that helpful, or not? Would you like something similar, but for subscription names? I just got around to trying this, but I guess because I'm searching the subscription names (which I've formatted to be the artist name + a personal note about why I subscribed to them), it appears to do nothing at all. Yes I would like something similar but for subscription names. But I will admit that even if such a button existed, it appears I would never have tried it, since I didn't even know this button existed.
>>15717 I quoted the wrong two posts; I meant to quote the actual post I was quoting, and my earlier reply to it >>15609
>>15716 I just want to say that I'm having a really positive experience with Hydrus now. I tried using it about a year or two ago, and I found that it was unstable and would crash overnight often when importing or even when navigating files. I left it untouched since then until recently when I updated it and it's a lot more stable now. Hydrus has crossed over the threshold where I would trust it if I moved my pictures from the native filesystem to it.
Heads up for anyone using the kemono parser. Due to the anti-DDOS faggotry the client will fail to download virtually all files. To fix this just go through the anti-DDOS thing with your browser and then copy the cookies over to hydrus.
(312.11 KB 1093x1077 illya thinky.jpg)
Question. What happens if I edit an image that is the hydrus database? Say, hypothetically, I go into the database file structure, open an image in gimp, edit it, and save it. What will happen? Will anything happen? I know that hydrus file structure is based around hashes, and changing the file changes the hash. Will hydrus detect the changed hash? Will this cause any problems with the database?
When deleting a file, is there any difference between selecting a predefined reason (the ones you set in options) and writing in a custom reason, aside from the former being quicker of course? Like, is there any sort of deduplication to save space, or something like that, or is choosing one of the pre-defined reasons effectively equivalent to writing in that reason yourself?
>>15721 When opening the file through hydrus it'll notice the hash is different and say the file is fucked. >>15722 It's mostly just for organization purposes. When you import a deleted file it specifies the reason you chose, which can be useful for some people.
Getting 'zero length tag' error from trying to process the official tag repo. Any easy fixes?
>>15724 Works on my machine. Are you on linux? Have you tried restarting hydrus and trying again?
>>15725 Yup. I actually reinstalled Windows for a different reason, moved the database and it still persists. I guess I'll try it on Linux once I am done setting up my home server.
I just got a warning that I shouldn't have more than 500 pages open, as it leads to program instability. Is this a warning I should actually heed, or is it safe to just ignore it? I thought if I could afford the page weight that it didn't matter, but I guess not. My page weight is only 343k, in contrast to before when I hadn't closed my "gallery downloader" page I'd been using since I first installed hydrus, the "gallery downloader" page alone was worth 1.2m page weight. Unless there is a significant difference between what produces page weight in these scenarios, I think I will just ignore it for now. But I am concerned.
>>15727 The program gets slower but I don't think it actually crashes.
>>15728 I also remembered that on boot it warns me about having 165 pages open, so in hindsight I don't think the number actually matters, it must just be a warning of some sort. I don't know why so great of a discrepancy in the threshold for these warnings, though. It's good to know it won't crash though.
>>15727 >500 pages Sweet shit, what the fuck are you doing nigger? I thought me having about 20 was a lot.
>>15730 I've only ever used hydrus to rip from pixiv, so that's my "gallery downloader" page. Then since I started using "page of pages" to group, I made a "page of pages" for browsing, and one for imports with gallery-dl filenames. The "browsing" "page of pages" is currently 74,100 files, and my "imports" one is currently 153,243 files. But they both feel pretty useless, since I just want to view what the search query for each page is. Cause I re-opened all my gallery-dl artist gallery rips as simple search queries. So it would be nice if this was browseable in any way beyond tabbing between each page one at a time. But right now it's just history until my page weight gets too high, at which point I'll export it using >>15407 again, then the contents really become out of reach. I feel like nothing is really ever at my fingertips in hydrus, but my expectations are really lonely, so...
>>15727 You can rename pages by doubleclicking, also use tags. Hydrus is meant to be used more like a booru than an image browser. Paged searches would be neat.
>>15732 Meant for >>15731
>>15732 >>15733 Well yeah I renamed the parent pages so they're titled "browsing"/"imports", but it would be redundant to name a page after its respective search query, instead of just having the ability to see the search query of all pages in a column or something. Also I would still be tabbing back and forth one by one. >Hydrus is meant to be used more like a booru than an image browser. This sucks.
(786.10 KB 480x270 1392737928043.gif)
I noticed opening bigger images (2+ MB) tends to be very slow (a few seconds) and it makes my hdd do noises. If I open them from the actual folder, I don't get any noises and they open fast. It's entirely possible my hdd is dying, but I do find it curious that everything works fast and as expected if I don't use hydrus.
>>15735 Do you have a lot of pages open? A few days ago it was slow as shit and files would take too long to open, turns out it was slow because of a 50k+ results search that I accidentally did.
>>15735 I mean that's fucked up, but if you're on wangblows you can run HD Tune Pro, which is a terrible program since it gives you a timer until it stops working until you pay, but it was what was recommended to me before, so I ended up torrenting it to use it again. Also maybe defragging your hard drive would help.
>>15737 >HD Tune Pro Tried it out, nothing seemed out of the ordinary, and I have kept an eye on SMART stats for a while now. I'll give defragging a try as I haven't done that in ages, it's 38% fragmented now. >>15736 I have ~10k images open on about 10 tabs. No ongoing dumb searches. If I search for images over 20MB, the first one can take like 20 seconds to load. Ones after it will take less time. If I then jump 100 images ahead, I get the long load time again. Also I should mention the hdd noises aren't the alarming screeches and clicks, just regular noises but waaay too loud.
>>15738 I don't know all HD Tune Pro can do, but the "benchmark" shows you read/write speeds, "error scan" is really slow, but shows you bad sectors, and "health" will notify you if... something, I don't know. But one of my two HDDs showed up yellow for something here, which sucks.
>>15738 I just tried it out and yeah, there's some overhead with hydrus and opening images. I'm guessing python throttles the data as to not freeze the rest of the program. Maybe having that process on its own thread would fix that. If the process already has its own thread shit's fucked
>>15739 I did everything and there was literally nothing wrong as far as I could see. Not an expert. >>15740 Okay, I won't be terribly concerned with the hdd for now then, as there aren't any statistics saying I should be. These long load times are kinda annoying though.
>>15741 I know your hard drive seems to be in good health, but I just remembered "CrystalDiskInfo" as another HDD health-checking software. It seems to only show one screen worth of info, but at least it isn't timewalled-premium bullshit. Also it has a version that makes the background an anime girl. https://crystalmark.info/en/software/crystaldiskinfo/
>>15742 That's what I've used before, yes, and will keep using.
I installed the Fantia parser but it doesn't work with NSFW images. I created an account but II had no luck with the usual copying cookies method. I even tried using the companion but no luck. I think there's some extra security rather than just cookies. Is there any way to actually login through hydrus? github.com/CuddleBear92/Hydrus-Presets-and-Scripts/tree/master/Downloaders/Fantia
My client gets stuck uploading to the PTR with 56 tag petition data left. I tried leaving it overnight and when I woke up it said the PTR was busy or something to that effect.
Please add the ability to export (copy, without deleting them from hydrus) all files from a specific tag search into one folder. For example, person:belledelphine then export all files into their own folder labeled BelleDelphine or something.>>15143
>>15706 This worked for a few posts in subscription mode, then it is erroring saying bandwidth up to 12hours, is this the rate limit you were talking about? Thats fine and all if so, however the issue is if I try to add anyone else to my subscriptions from ofansparty, it marks them as dead and pauses them because it couldn't grab any files (because the other creator capped the limit)
>>15747 Glad it works (ish) for you. Are these bandwidth errors sent by Hydrus (https://hydrusnetwork.github.io/hydrus/help/getting_started_downloading.html#bandwidth) or by the server (you're getting an error back)? If it's a Hydrus error, you can bypass it by going to network -> data -> review bandwidth usage and edit rules. If it's a server error, my bad, sorry, I did not know (I'm using my own IPFS endpoint so I can't hit any limits); you can choose another server at https://ipfs.github.io/public-gateway-checker/ and update it in the parser (it sucks, I know). Let me know if you need help with this or if you have a server you'd like to try and I'll export a new downloader png for you. Note that this would re-download images (as Hydrus does not recognize the unique ID of the IPFS file yet, see >>15677; it would recognize this as a new URL because of the new domain); they will still be recognized as the correct file (no dupes) after they're downloaded, but you'd consume double the bandwidth for the images (and worse, vids) you already have. You could alternatively host your own personal IPFS gateway (this would fix both of the above), but be aware of the P2P implications.
>>15746 You can search, select all the files on the grid view then right click -> share -> export -> files, which gives you the ability to do this; deleting them from Hydrus is an option, disabled by default.
>>15746 >>15749 file->import and export folders->manage export folders Set up your search query from there and hydrus will periodically dump files that match it into a specified folder.
Is there a downloader for https://kemono.party/ ?
(1.56 KB 221x93 sub.png)
>>15748 >Are these bandwidth errors sent by Hydrus (https://hydrusnetwork.github.io/hydrus/help/getting_started_downloading.html#bandwidth) or by the server (you're getting an error back)? That link says under bandwidth the following, >It will not be too long until you see a "bandwidth free in xxxxx..." message. As a long-term storage solution However, my error message when going to my subscriptions, while similar, is a bit different. See pic related. Is this a Hydrus issue? Or a gateway issue? Also if it is a Hydrus limit issue, do I edit the ofans.party web domain bandwidth, the api.ofans.party web domain bandwidth, or the IPFS.io web domain bandwidth?
>>15753 This looks like a gateway issue; the best way to know for sure is to go to the manage subscription menu, double click on the subscription then on the creator (the menu where you can set the query text); you have a button on top of the progress bar that will display the downloaded files / the errors you may have encountered (and you can also alternatively copy try the URL in a browser, just to see if you can reproduce an error there as well). I looked for any notion of bandwidth limits for ipfs.io's IPFS gateway but was unable to find anything, so I don't know how high or low it could be. Have you considered downloading everything in a gallery first, then using the subscription as an update method only? This would make errors easier to spot IMO. Also if you're fine with the wait, you could also try lowering the Hydrus bandwidth settings for the gateway, so Hydrus would wait rather than hit errors. The ofans.party and its API subdomain aren't used much, just to list the files to download; everything else is going to the gateway. Again, let me know if you want me to generate another png with an updated gateway (you'd have to select one first though); note that you can of course change this yourself it in the parser, but it's pretty far in. Hydrus dev, what happens if a parser provides multiple URLs? If I set several gateways with different priorities, would they be tried eventually as fallbacks, or would an error on the first prevent the others from working? I've looked for that in the doc but didn't find the answer to this question.
>>15699 Thanks for this info. Normally when you have a legit # in the URL, they treat it as instructions for clientside javascript. The server never sees fragment text, but js can access it, so it can do some later AJAX calls (or whatever the kids call this stuff), so the initial root you load is just the 'browser' of the content, and the actual content comes as JSON later. You see this in Mega links, and I think gmail does it for some 'last viewed email' kind of stuff. Not sure if that is what happening here. I'll upload this >>15706 to the github now. Thank you for your work. I think some of the discord guys follow this thread in and out, but I don't know if anyone is religious about it. >>15747 >>15748 >>15753 >>15753 For this, this looks like normal hydrus bandwidth rules applying. By default I think subscriptions only try to do 256MB a day, to be conservative and generally smooth things out. Hit network->data->review bandwidth... to see everything. If you sort that dialog by 'blocked', you'll probably see what it is that hit a rule. You can either edit that thing's rules specifically, or hit the 'edit default bandwidth rules' to give you some more space. Once there is more bandwidth available, your sub should restart in a minute or two. Hit network->pause->nudge subscriptions awake to give them a quicker jog.
>>15700 >>15701 >>15702 Yeah, for now, the blacklist is in the parsing pipeline and can act before the client know's the file's identity. An extension to check other known tags for the file would be neat. I think a blacklist that supports multiple tag logic is too complicated for me to support right now. Maybe in the future if I get some objects and UI to handle and test against that stuff easy. If your deletion rules are this specific, I assume they are only a few percent of your total download. If you, so you might like to just save a search for like: system:inbox female only duo black hair And then run that once a month and ctrl+a->delete. Sounds stupid, but it may be a practical solution.
>>15703 I don't have excellent zoom controls yet either in duplicate view or the normal media viewer. I am overdue a complete overhaul of these settings. I want some toolbar buttons or a menu from the top hover or something--some way to easily change this logic with one or two clicks. I am sorry for the trouble. >>15705 The client is not clever enough to remember 'ignored' results. If you make a subscription though, I think they'll be remembered forever? I am not sure if they are eventually cleared out of subs, but I think so. In that case, when we finally get ugoira support, you'll just be able to click 'retry ignored' on your pixiv subs and blitz through them all. If subs forget them over time (like 'successful' results), or we need to search deeper, then at the time we'll instead do a 'reset' on all pixiv subs, and maybe paste their queries into new downloader pages. This will take some time and a bit of bandwidth for some surplus HTML/JSON, but hydrus will be able to skip over all previously downloaded files, so it shouldn't be a huge problem. >>15708 That's my dream with the Client API. I want clients to talk to other clients so you can browse across machines just as if you were searching your own. I just have to keep working adding more commands to the API and then browsing tech across machines. This upcoming 'multiple local file service' work in the next three weeks will be helpful to this, adding more 'virtual location' tools and cleaning out the behind the scenes.
>>15709 I treat the 'trash' service as basically a recycle bin. Files sit in there in case you want to undelete, but then they get permanently deleted. Hit options->files and trash to customise its rules. I agree about a proper 'deleted' domain. There isn't one yet. Hydrus knows which files it has seen before and deleted, but there isn't a way to discriminate in search UI between those files and files it knows about but hasn't actually seen yet. If you have help->advanced mode on, under the subscription UI, on the UI for a particular subscription, select some queries and then hit 'quality info'. This is a totally hacked prototype thing I made for a lad a while ago, but it gives you brief info about archived/deleted ratio and helps you figure out if any of your subs are actually bad. I hope that can help a bit for now. >>15710 >>15711 Yeah, the funny thing about ugoiras is they have custom frame timings, so technically they can come with a separate JSON sidecar file storing the timings that the javascript then reads and uses to switch images. This is one of the roadblocks to ugoira in hydrus. I think it is valid to store the JSON inside the archive, I forget if pixiv store it this way, I think they do, but that's the way hydrus will do it. >>15712 You'll likely have the files on your disk, but hydrus won't know you have them (it will forget a few minutes of session/database work, but Windows will have saved the files to disk, so basically it just rewinds a few minutes on next boot). If the power cut hit your disk at a sensitive time when it was writing, one or more files may be corrupt, your database may have minor corruption. If you got hit by this (and get some scary error message in the coming weeks), it should all be fixable, let me know. When you continued/restarted the downloaders, they likely said "hey, those files were 'already in db'" when they redownloaded and saw them in the right place on disk already. I don't think you'll have missed anything.
>>15713 >>15715 If the issue here is bandwidth, change your bandwidth rules under network->data->review bandwidth.... Either for all domains generally in the defaults, or sank especially. Beware however, sankaku are always short on bandwidth and quite strict about it. I think they'll ban your IP if you exceed the '80 requests per seven minutes' thing too much, and I think there is a '5 requests per 12 seconds' rule as well. Just be careful, and if you can get the content elsewhere first, you'll do them a kindness if you do. >>15719 Thanks for letting me know. I am glad it is better now. Moving from wx to Qt was a huge thing, and I had to learn a lot and clean a ton of code (and there is still a huge pile to do), but the transition has been a huge positive overall for stability and UI overall, particularly for multiplatform. >>15721 >>15723 Yeah, basically my answer is: don't. Hydrus won't actually notice the hash is different when you open it (hash calculation is a little expensive, so I don't do it much), but if a maintenance routine occurs on that file some years from now, you'll get some errors and the file will likely be moved out of the db to a landing zone directory where you can sort out the problem manually. Most importantly, if the file is different, if you then add tags based on the edit to the PTR, other users who received those tags would get bad info, since the hash the tags were shared on would still refer to the old file. It won't crash anything. But if hydrus notices, it will think the file was hit by hard drive corruption.
>>15724 >>15725 >>15726 If you are on an older version, please update. Some weird tags were not being handled correct when I updated the tag cleaning process. 433 should handle it correct. If you are on 433: fugg, and please copy and paste your traceback so I can have a closer look. >>15727 >>15728 >>15729 This was an actual problem in wx. In Qt, I believe it is just memory limit and CPU stuff. Your client will get laggy. Most importantly, if this is a long-term issue of just adding more and more pages, here is my experienced opinion (bear in mind I fail at this too, so I am also being hypocritical): If your current workflow regularly adds pages and fails to remove all those pages, even if you tell yourself you will one day clear them out, then the workflow is not working. You will never get to all those pages, you are in an infinite queue situation. You should reduce input or increase output one way or another. I have this problem with my general file imports, I import thousands a week but only archive/delete a few hundred. But I keep my UI clean and fast by giving my imports processing tags so I can load them up in system:limit=256 chunks later. There is no need to sit on a massive bloated session, it just adds frustration due to lag and five hundred things to look at, and without change, it will never get better.
>>15735 >>15736 >>15737 My image rendering pipeline sucks for some situations. It is another thing way overdue for an overhaul, given how more frequent bigger files are these days. I apologise for the inconvenience. I am not sure what would cause the extra HDD activity, but it may be the neighbour render-caching that is loading the two or three images previous and next to the current one you are viewing in order to display them faster when you browse. Best answer for now, if you can afford it, is to boost your image cache under options->speed and memory. There are several systems I can improve here. >>15744 It could be your 'User-Agent' header? That's often a final trick for more highly protected CDNs. Hit network->data->manage http headers and change your fantia domain or global User-Agent to exactly what your browser has. The Client API will support this in future, so Hydrus Companion will copy it over for you. >>15745 Thank you for this report. Try it again now, and if it doesn't clear, please hit database->regenerate->clear service info cache. Might be a miscount.
hydrus dev did you miss this >>15648 I made this post and since there hasn't been a release since I made it, it's obviously still happening. Even if it doesn't break anything, it's still annoying and I haven't seen anyone else bring it up yet.
Say dev, what's the rough roadmap on "big plans"? In this thread you said first comes behind-the-scenes stuff on network code and admin tools, and that next comes file services. I'm assuming that 434 is (most of) the former, and that you're about to start on the file services. Personally I'm already pretty satisfied with Hydrus, so I'm not going to beg for certain features. I do like the idea of cleaning up the media viewer, and also the option to 'recycle' definitions mentioned in >>15231 though.
>>15755 Thanks for posting the png on Github! Feels good. Reposting my question from >>15754 as Hydrus dev already replied to all of the related posts but this: What happens if a parser provides multiple URLs? If I set several gateways with different priorities, would they be tried eventually as fallbacks, or would an error on the first prevent the others from working? I've looked for that in the doc but didn't find the answer to this question.
>>15754 >Again, let me know if you want me to generate another png with an updated gateway (you'd have to select one first though); note that you can of course change this yourself it in the parser, but it's pretty far in. Haven't gotten a chance to look at the other stuff yet, but just for reference, is it possible to use IPFS Desktop as my "gateway"? I'm not extremely familiar with IPFS still.
>>15765 Absolutely, IPFS Desktop contains an IPFS node, and a node can act as a gateway (should be enabled by default). If you've changed nothing to your settings, try getting http://127.0.0.1:8080/ipfs/QmYwAPJzv5CZsnA625s3Xf2nemtYgPpHdWEz79ojWnPbdG/about in a browser on the host that is running IPFS Desktop (this link is SFW). If it works, you can just replace https://ipfs.io with http://127.0.0.1:8080 everywhere and your links will work the same. I'm getting on the P2P disclaimer again, but be warned that getting files mean that you'll also be sharing them. Also to note: sometimes, "warming up" peer connections can take a while, so try to get a file from ofans.party from your local IPFS node first (click on an image / video link then replace the original domain with the local one); subsequent downloads should then be much faster.
>>15757 Glad to hear that redoing the zoom stuff and downloading ugoira are both definitely on the todo list. >If subs forget them over time (like 'successful' results), or we need to search deeper, then at the time we'll instead do a 'reset' on all pixiv subs, and maybe paste their queries into new downloader pages. This will take some time and a bit of bandwidth for some surplus HTML/JSON, but hydrus will be able to skip over all previously downloaded files, so it shouldn't be a huge problem. Not a problem, thank you. I'll probably just keep the gallery import page up until then. Worst case, I keep a separate document of artists I care about so I can run the queries again with no data loss unless the artist has deleted the ugoira in the meantime. >>15758 >When you continued/restarted the downloaders, they likely said "hey, those files were 'already in db'" when they redownloaded and saw them in the right place on disk already. I don't think you'll have missed anything. Fortunately I did not lose anything.
>>15764 Ah, sorry--at the moment it just pulls the URL with highest priority. If multiple URLs have the same priority, they are each added as separate 'child' import objects in the import list. I think the pixiv 'manga' parser does this, if you want to check out an example. This is good when the URLs point to different files, but not good when they are different resolutions of the same file (user would get all resolutions as separate files) or just the same file in different locations (user wastes a bunch of bandwidth). In the next big rewrite of the system, there will be a way to say 'try this top priority URL, but if it 404/401/403/whatevers, try this lower priority URL', but that'll take some work.
I had a great three weeks working on network functions and serverside admin controls I had neglected. The old account management controls are working again with many improvements, account types can now be set up as user-creatable straight from the client (no need for registration keys!), and much of the code is cleaner and handles errors better. In the rewrites, I also laid the groundwork for custom update periods, serverside tag filtering, and faster petition processing, which I hope to get done in the coming weeks. Regular users who do not admin a server do not have so many changes, although I have fixed bugs and added more controls to repositories, so you can now pause account sync, update downloading, and update processing separately, and I fixed a logical problem with processing that was sticking some users with deleted siblings. The 'network version' goes up tomorrow, and the PTR should be updating, so if you sync with the PTR, you will want to update some time to keep talking to it! There's no rush, but you'll get a little error popup if you try to talk to an updated server on an older client. The release should be as normal tomorrow.
I got a small problem. When I set a tag to be added to the public tag repository with a shortcut, if the tag is already present, it'll ask for a removal reason as if I'm trying to delete the tag, instead of just doing nothing. The shortcuts I have are to set the tag, not toggle them, so it shouldn't try to remove the tags if they're already present. It should just do nothing. This doesn't happen with tags in "my tags". With those, it just does nothing, as expected.
https://www.youtube.com/watch?v=f1nvZ74OY_I windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v434/Hydrus.Network.434.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v434/Hydrus.Network.434.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v434/Hydrus.Network.434.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v434/Hydrus.Network.434.-.Linux.-.Executable.tar.gz I had a great three weeks working on long overdue network updates. Most of the changes this week restore old account management UI for server admins, but there are a couple of fixes and improvements for regular users as well. The network version goes up to 20 today. Clients and servers can only talk to each other if they are on the same version, so if you want to talk to the PTR, you will want to update at some point. No rush, but if you try to talk to the server on an older client, you will get a polite error message. regular user stuff On review services, you can now pause repository account sync, update downloading, and update processing separately. If you don't want to process right now but still want the client to be grabbing update files for later, or you want to pause all service network traffice for a while but still process your backlog, it should now be easy to set up. I fixed an annoying bug where going from 'random' file sort to one with asc/desc sort order by loading a favourite search would always default the asc/desc to the top value. Guess what hydev's default sort is. Server administrators can now set messages to accounts. If you get a message, your client should get the update within a day or two, at which point it will appear in a popup box and be viewable on review services. Server administrators can now set some account types to be user-creatable--no need to mess around with registration keys or access keys yourself, the client can fetch them for you from the server. A new button on the manage services edit panel checks this and can do it for you. I understand the PTR is likely to move to this in future for accounts with sibling and parent petition permissions, since grouping these by account will be useful for janny petition workflow. Having separate accounts for these complicated uploads will also allow better account management (including the above rudimentary messaging) for users who make common errors with sibling or parent submissions. I believe I fixed a bug in repository processing that caused some newer users to get stuck with siblings and other data that was later deleted. This should not happen again, and I am now planning ways to efficiently fix the problem retroactively. (If you were hit by this and want the non-efficient fix, hit 'reset processing->reprocess content' for the repo under review services). Some users recently got some scary errors about an invalid repository update file. This was the result of the client becoming confused about which update files it should have for a repository. I am still investigating why this happened, but to fix the immediate issue I have added a maintenance routine (under the new review services->reset downloading) that triggers a full resync of update metadata and delete surplus records. This function will fire automatically in future if the service encounters borked update files of any kind. If you got these errors, please try firing it and let me know how you get on.
admin improvements After a very long delay, account management is finally back. If you have permission to modify accounts, the 'modify an account' service admin menu item works again and launches a fully revamped window. You can also launch account modification from manage tags for a particular tag selection, the thumbnail right-click for file repository files, or the petition processing page. The new window shows better data about accounts, and lets you zoom in on individual ones to see current account type, expiration, banned status, or message. It also shows if you are editing yourself! The account modifications are: change account type, set/extend/clear account expiration, ban, unban, and the new 'set message'. All modifications now print janny and subject account key to the server log. I removed the old 'superban' (which deleted all that user's content contributions), but it will likely return in future. The service admin menu also has a new 'review all accounts' entry that lets you see a simple summary list of all accounts, which you can then select for modification. The manage account types dialog now lets you set up auto-account creation. It works a bit like subscriptions, with a 'account creation velocity' limit of 'x accounts per y time delta', like '10 per day'. This defaults to off obviously, but if you set some account types to auto-create, users will be able to see that in manage services. future admin improvements As always, there was much more I wanted to do than could get done. I wasn't able to finish improving petition processing workflow or add custom update duration, a serverside tag filter, or a server message, but I was able to get the important network stuff done I needed for these things. Since incrementing the network version is always such an obstacle, I am confident I will now be able to work on these things in normal weekly work. Please let me know how they work out for you as I roll them out. full list (1,700 word changelog this week, please check github or the help) https://github.com/hydrusnetwork/hydrus/releases/tag/v434 next week I will catch up with the regular work I missed over the three weeks. I will also try to get optional update duration done for the server. And lastly, I absolutely want to figure out macOS releases that work on Big Sur, which is another thing I have left for too long. The test github build some users have been working on apparently works, so I just have to integrate their script into my own github build routine and, fingers crossed, we should be good.
>>15691 >>15704 Since updating to latest (434) this whole behavior from last version seems to have stopped, and everything is back to normal as far as I can tell. However, the window you asked me to open >>15610 still doesn't appear to be able to be opened for me. I just tried on this version, but nothing happened. I just noticed though that your exact wording was: >If you can, please now hit up help->debug->data->show scheduled jobs. This will make an absolutely gigantic popup window and print data to your log. I do not need to see everything, but can you confirm for me that it involves a lot of 'REPEATINGWorkOnGallery' and 'REPEATINGWorkOnFiles'? >This will make an absolutely gigantic popup window and print data to your log. So I went to my database path and found the log folder, and in "client - 2021-3.log", "REPEATINGWorkOnGallery" appears 803 times, and "REPEATINGWorkOnFiles" appears 845 times. Neither of these appear a single time in the log for the previous month. Sorry I didn't give feedback on this by checking this way at the time or at all before this date. It was only cause stuff seems to be normal for me now and it still didn't open anything why I re-read what you said.
Non-interpolated upscaling for pixel art when?
I have a huge number of absurdly large pics. I can just convert them to pngs in batch with honeyview, but tagging is the bottleneck in the process. Is there a way to automatically map tags from existing files in to new files? I know I can just export some tags in the filenames and then import with regex, but that limits the tags, doesn't allow for different tag services, inbox/archive, or rating operations.
>>15775 You can actually export a txt file per pic containing all the tags in the right click -> share -> export -> file menu. Right above the "export" button there's a "export tags to .txt files?" box; don't forget to select a service. To re-import them after resize, you can then go to file -> import files, select the resized versions, go to "add tags before the import" and there there's a "try to load tags from neighbouring .txt. files" box, under "misc". I think that as long as your .txt has the same name as the original file, it'll be imported correctly and you'll have no issues re-tagging the resized versions. For inbox/arcbive, I'd add an extra "meta:archive" tag that I'd remove after adding them to the archive. I don't know how to do this the ratings though, maybe adding an extra tag as well would work if you don't have too many ratings.
>>15776 >You can actually export a txt file per pic containing all the tags in the right click -> share -> export -> file menu. Right above the "export" button there's a "export tags to .txt files?" box; don't forget to select a service. The problem is that it just merges all tags together without separating by service. That means I need to export and import the files multiple times to get the different tags.
>>15777 And in addition to that you need to rename all the txt files to match the new format. It works but it's an extremely clunky process, at least in windows. >0 - add in all temporary tags (e.g. archive and ratings) >1a - export files with tags for each tag service >1b - restart the client in admin mode instead to export symlinks, which is quicker >2 - do conversions >3 - rename all txts, which in windows requires you to separate the txt files and run for %i in (*) do ren %i %~ni for %i in (*) do ren %i %~ni for %i in (*) do ren %i %i.[format].txt >4 - put files and txt together and import with the first set of tags >4b - if you are in admin mode you need to either restart the client to be able to drag files in or go all the way through the file import menu like a dork >5 - change the txt files for the next tag service and reimport the files for every single tag service exported >6 - fix the temporary tags if you used them It's just too time consuming to be worth it right now. Stuff like being able to export just tags or to export separated tags for different services, and being able to import tags based on regex would make it a lot easier.
(176.36 KB 430x160 ptrstuck.webm)
>>15761 Clear service info cache didn't fix it. I've attached a webm of what the lower right looks like if that helps.
Can I exclude zip files from being picked up by Hydrus import?
Is it possible to select multiple images (directly in Hydrus), and have it queue them for IQDB tag fetch? Instead of doing it manually one by one? Also does that IQDB downloader also check on the 3D IQDB (https://3d.iqdb.org/)? If not, is there a parser for that? One more thing, is it possible to run a search with multiple parsers? Instead of one at a time. For example, try IQDB then another parser if that one returns nothing? Doing this stuff individually for lots of files is a pita.
could a lightweight (emphasis on lightweight) web gui be introduced for managing hydrus on other machines also the ability to connect to a host hydrus client from other hydrus clients and download directly to the host, import files over to the host from other machines, and manage tags and stuff
>>15782 Is this maybe what you're looking for: https://archive.rebeccablacktech.com/g/thread/80755655/#80758410 https://hydrusnetwork.github.io/hydrus/help/client_api.html https://github.com/floogulinc/hydrus-web https://hydrus.app/ I thoroughly have no idea what this does in any way, but it seems irresponsible to just to it without knowing, but maybe it's what you wanted.
>>15781 >>15428 Is this maybe what you wanted. Note that I have no idea what's going on, but maybe adding that tab would help, I wouldn't know.
>>15783 to my knowledge the hydrus web is only for viewing i'm requesting basically a full fledged (but lightweight) web interface that does everything the hydrus client can. so i can manage tags, use parsers, delete, import files, add urls to watcher, manage subscriptions, etc. from the browser or from the browser on other devices. lots of self-hosted programs offer web gui's to do stuff like this, so it's easier to use the software witjout being on the host machine in this case, it could probably utilize the client api, but it would still need a frontend
>>15784 not sure i can't tell what the video is doing, but i dont think so. basically asking if i can highlight multiple files, then click fetch tags, instead of needing to do it one by one.
>Bear in mind the Client API is still under construction and is http-only for the moment--be careful about transmitting sensitive content outside of localhost. The access key will be unencrypted across any connection, and if it is included as a GET parameter, as simple and convenient as that is, it could be cached in all sorts of places. Is this still true? That's pretty cucked if a user wants to use client api outside of lan. Or is this not an issue for reverse proxy setups through NGINX? Could an NGINX template be added to that page btw? https://hydrusnetwork.github.io/hydrus/help/client_api.html
>>15694 >those 'medium' tags may be removed from the PTR To clarify, do you intend to only remove the obvious medium:* tags describing the file itself? It's an useful namespace to share otherwise. Anyway, can something be done about database writes? I have the database on SSD with full image directories symlinked to HDD, and hydrus writes to SSD a lot. Especially during PTR processing and when using downloaders (especially with tag importing but still a lot without it). Either one can easily write a few dozen gigabytes per day, that's pretty bad for SSD health.
>>15762 Ah, sorry, I did miss it. Thank you. I am aware of this issue and am going to put some more time into it. It affects some other dialogs and panels as well. I have improved the size calculation logic here, but something is a few pixels off in some situations. >>15763 I have never really had any sort of firm roadmap. Although I know it is unhealthy for software development, I don't really have a strongly defined end-state or mission statement other than I'd like to make managing imageboard stuff easier for Anons. When it comes to big stuff like an overhaul of the downloader system or client-to-client comms, I have about thirty or forty of those, and my 'little' todo is so overwhelmed that I stopped counting at about 3,600 items. I think I am mostly just bad at managing my time and organising things. I'll just keep going every week with a mix of small improvements mostly suggested by users and trying to do a 'big job' every quarter or so. In very broad strokes, my main worry right now is that the client database is getting a bit 'big' for many users. I have spent a few years making file and tag storage and search work efficiently, and now I am starting to think hard about the actual difficult and important part of hydrus, which is management. Creating human friendly workflows for discovery, searching, description, browsing is something often on my mind. I am really happy with some of the search term improvements and saved searches recently. I want to make editing tags easy and fun, or at least less of a pain than it currently is. I'd also like universally portable metadata. Pulling notes through the downloader engine, sending ratings to another client through the API, importing URLs from an external repository, all that sort of stuff. But if you get me talking, I'll say 'I'd also like xxx' for a bunch of subjects, ha ha ha. I'll just keep working. The work is never done. My very long term plan, which is half-dream, half-actually-happening-now-oh-shit, is to integrate some ML into hydrus. Machine Learning benefits from large and diverse datasets, and what do you know, we are all collecting these giant collections of personalised file description and preference data. I want the hydrus of five years from now to (optionally!) learn from you and do work for you like simple tagging and archiving and 'seems like you should add a subscription for this artist' suggestions, all trained privately on your own CPU. And in ten or twenty years, maybe hydrus will draw new memes and art for you, who the hell knows where this tech is going. I have been skeptical of this, despite my personal excitement for the tech, but the success of https://gitgud.io/koto/hydrus-dd and some experiments users have been doing on auto-tagging models have convinced me that this is viable for a certain subset of our problems. If there is one end-state of the hydrus network project, it is to create the world's largest neural network models of bepis and bepis-adjacent metadata.
>>15774 There are the options in options->media->image/png, but I am not happy with how static they are. I'm due a zoom rewrite soon, making various zoom logic quickly changeable in the media viewer as you browse, and I'd like the same for current interpolation algorithms. I am also getting concerned about the image rendering pipeline. It is old and busted in many situations, particularly high-res pngs, and needs attention. I should be able to improve zoom quality then. I don't know much about different interpolation algorithms, but OpenCV currently gives me these: https://docs.opencv.org/4.5.1/da/d54/groupimgproctransform.html#ga5bb5a1fea74ea38e1a5445ca803ff121 Are there different clever interpolation algorithms that people into emulation and pixel art commonly use? What are their names? If there is a python library I can search up that can wangle this stuff off of an opencv numpy image array, or something I can easily convert from that, then I can add these as new options. I won't have time to code my own instantiations of the algorithms though, most likely. Having different zooms for images tagged 'pixelart' is something that should be possible in future. I want that sort of thing possible. But for now, the filetype is the only thing that can change zoom type. >>15776 >>15777 >>15778 >>15775 Yeah, I regret how shit .txt tag export/import currently is. It was a prototype experiment that proved useful enough that it needs an overhaul. I want service- and metadata- distinct import from/export to for sidecars in .txt and json and xml. For now, though, your best bet if you can do some scripting is the Client API: https://hydrusnetwork.github.io/hydrus/help/client_api.html >>15779 Christ! Thank you, I will check the code and figure out how this could be happening. It may be a tag state that is invalid due to a bug, something like a tag that is both 'current' and 'pending', and it can't be uploaded because it is being filtered out. You might like to do a search for just pending tags and 'system:has tags' and see if there you get 56 (+) tags in the result, or if you get no files at all, or what is going on here. If you can remove any 'surplus' pending tags manually, that may fix you up. In any case, let me know how you get on. EDIT: Yes, I think the problem here is that the tags are being sent out, but they aren't being cleared due to my new filtering tech on the content update pipeline, so your client thinks there is still stuff to go out, but then it never clears. I will look at this.
>>15780 Not from all imports yet. You can from an 'import folder' (file->import and export folders), which has a filetype selection filter. I'll be adding that filter to all 'file import options' in future so you can filter on any import context. >>15781 >>15784 >>15786 I am hesitant to hack in more than one 'filename lookup' fetch at once since I am afraid of it going wrong (e.g. a user not knowing what they are doing queuing up 100,000 queries to iqdb and wasting a bunch of bandwidth for everyone). I want to write a proper system for deferred tag lookups in the foreground on the tag dialog and the background, with maintenance timers and tag application rules for the background job, but the whole lookup system is also still running on legacy parsing code. It is on my mind, but it will have to wait some time for me to get all the systems up to date and do it as a proper project done right. I agree it will be nice to have, though. I'd like the new system to work on any url class the client understands, so you'll be able to do tag fetches from any known url for the file, as well as md5 or file lookups. >>15782 >>15783 >>15785 Yep, these are the current state of this tech. There is lots of interest in managing a client from mobile devices, browsers, or other clients, but it is still in the early technical stages. Much of the work is waiting on me to expand the Client API and some back-end network tech, but it is coming. I will work on client-to-client comms, other users will work on apps and browser interfaces. I will keep the https://hydrusnetwork.github.io/hydrus/help/client_api.html page updated with links to all current projects. Someone told me today that this https://ififfy.github.io/flipflip/#/ recently added a hydrus plugin, too. >>15787 Thanks, I will update that help. You can do https now, but it is a self-signed certificate unless you have your own domain and cert. I'll be improving the way certs are generated in future--some users gave me some neat ideas--and also improving the client's verification so you can manually choose to trust a self-signed cert (atm they are unverified). I do not know what an NGINX template is. Can you describe it in a couple of sentences, and how it is useful, and maybe post an easy-mode guide for me to follow?
>>15788 I'll say that I don't run the PTR any more. The guy who does and the janny team are on discord, and I know them and everything, but I do not enjoy making rules or running a community, so I try not to make any decisions or speak for them too much. I sometimes use 'medium' tags as an example of a 'bad' tag since you can usually already search for all that stuff using the 'system:' tags. I know some users like them as tags anyway, but in the case of sharing, you are sharing data about files that a recipient already knows about every file they have imported. I'm open to adding local tools like that 'if has audio, then add "has audio" tag to "my metadata local tag service"' tech, since it affects the same outcome and doesn't waste bandwidth or storage space for those who sync with the PTR. Basically, if I reach across the internet and tell you a webm is of 'character:samus aran', that is useful, but if I tell you 'that is a webm', I haven't actually added knowledge. EDIT: Ah shit, I realised I read your post wrong! Sorry! Yes, this is only to remove medium:webm etc... 'medium:monochrome background' or whatever is still useful to share. There is a limit to how much I can control the writes to the database. There is a lot of data in the PTR, and I am subject to SQLite's rules. A huge table ends up needing a lot of writes after 2 million changes. If there are ways to reduce this, I'll be interested to know. I may be able to reduce it a bit by extending my commit period (currently 30s) on the database while under heavy load, which I understand with WAL journalling would let changes to a page that was already altered in the current transation only require one write, but experimentally I have not seen huge gains unless I extend the commit period to five or ten minutes to cover a significant proportion (and hence more shared altered pages) of the current repository processing job, but this then makes for uninterruptable processing (hence why I would probably do it only under heavy load, and probably in idle or shutdown time), and increases overall spare drive space needed for the transaction as the WAL file (and perhaps some temp files in your temp dir) grow with the long transaction. However, I am not super concerned about the SSD health aspect. My general understanding, and please correct me if I am wrong, is that most consumer SSD drives can take a few hundred TB writes until they die. It seems to be an open question, some estimates are up to 1 or 2 PBW, but a decent conservative estimate for a 512 SSD is about 300 TBW. If we assume that a busy hydrus client with many files is writing 20 GB a day, then if that drive does nothing else, it would use all writes after about 15,000 days, or 41 years. And is it reasonable to say you'll get a new SSD every 5-10 years, or so? So in a conservative situation, as long as your hydrus is less than 12.5-25% of the writes you are doing on that drive (and I'll assume it is typically the biggest user of drive resources by far), the actual bottleneck on the drive life is sheer replacement. Also, since SSD drives seem to be growing in size and reliability pretty fast (TBW goes up with total size of drive, too), I think at a faster rate than the PTR is growing, I think the pressure here will relieve over time. Nonetheless, if you are worried, I recommend you pause PTR processing. I hope within this year to have a clientside tag filter that will let you choose a much smaller subset of the PTR to process with (e.g. just character and series tags), which will greatly reduce storage requirements. My biggest concern with the PTR at the moment, rather than resource usage, is that it has a lot of stuff people don't want. Just getting it smaller clientside will make for faster clients with fewer tags people do not care about.
>>15792 Oh, I should clarify, for this SSD calculation example, this only matters if you sync with the PTR. If you don't sync with 1.2 billion tags (and keep up to date day after day), your hard drive use is so much smaller, it isn't significant to think about.
>>15792 ...and that should be 'as long as your hydrus is more than 12.5-25% of your writes'. As long as you aren't running 3-7 other programs as heavy as PTR-syncing hydrus on the same drive, then the drive should be spending writes at a slower rate than your normal schedule of drive replacement, if you replace even 5-10 years. Since most users will be, say, 80%+ of drive writes being hydrus writes, and since practical TBW on modern hardware is typically larger, 700TBW or so, I think we can assume in most situations the drive will last 80 years or something before write failure. I figure it'll be unplugged for other reasons by then. Let me know if my numbers or understanding are wrong though!
>>15792 It may not seem like much but it writes 15-20 gigabytes per day just syncing to the PTR. Trying to download actual files with downloaders writes a lot more to a large database, especially if you import tags along with them. Also, in some partitioning setups, actual writes are amplified significantly. It adds up really fast, to more than projected daily writes. I don't know sqlite very well so I can't speak about it accurately, however: Is it possible to add an option to keep journaling in RAM as long as the database doesn't end up in an incomplete state on crash (in which case, discarding incomplete transactions is fine)? As a backup solution, is it possible to relocate WAL to a different partition? Is it possible to preallocate sparse blocks for mappings so that database fragmentation doesn't cause too many block rewrites? Speaking of fragmentation, I recently did a full vacuum of the database and, anecdotally, it's writing even more now. I'm not sure what else may cause write amplification. Unfortunately, character and series tags aren't nearly enough. Anyway, it would be nice to be able to do something about excess writes. PTR sync is the worst offender, but so are gallery downloads and subscriptions on larger databases. It all adds up to far more than just 20 gigabytes per day and it writes at an alarming rate compared to design TBW.
>>15791 >but it is a self-signed certificate unless you have your own domain and cert. yes I am referring to a domain+cert. However, the fact that the client api still sends the key unencrypted sounds like it might be a bad idea? >I do not know what an NGINX template is https://jellyfin.org/docs/general/networking/nginx.html Basically a configuration file for using Hydrus with Nginx. The user can then just copy the template file, and fill in the information like domain name, etc. The template can be as basic as just saying listen on 80 and 443 like a basic web page, or it could include other lines like compression, security flags, etc. It all kinda depends on the program so it's nice when programs provide a template instead of figuring it out. I'm sure someone more experienced with NGINX and networking can help you easily.
>>15792 >discord please consider providing an alternative to discord btw. it's fucking trash and i know you already know that. maybe a matrix or something. at minimum you could bridge it over to the discord, however please provide an alternative. https://spyware.neocities.org/articles/discord.html https://usemumble.neocities.org/ (the parts speaking about discord issues)
Hydev, you should set up a Monero wallet for donations. You're one of the very few people I actually feel like supporting, and a Monero donation address would let people like me anonymously send you magic internet money without the need for even opening a web browser. >>15797 I agree, but it sounds like the Discord is run by the PTR guys so you would have to talk to them. Bridging the two wouldn't do much, considering you can't use Matrix's encryption with it. I guess you could use it without having to directly touch Discord servers though.
>>15798 >Discord is run by the PTR guys so you would have to talk to them. The issue being needing a Discord account to do this. >Bridging the two wouldn't do much Two seperate channels could be made, a bridged one to Discord to allow for intercommunication, and then one not bridged. Or just a non-bridged one. Just threw in the idea of bridging to help with stubborn people not wanting to leave discord.
>>15797 >pitch black background thanks now im blind
>>15676 Maybe i just missed it in your posts but is there a reason that the artist gallery downloader sometimes gets the whole gallery and sometimes skips entire sub-galleries?
>>15735 >>15738 Here. Well, the drive is starting to fail. SMART stats are now showing warnings. Didn't manage to clone the partition to another drive because of an error, but I'll try again to see if it was just unlucky. How does Hydrus behave when reinstalling Windows, if I have my files split up?
What's the difference between "hash recognized" and "file recognzied" in the file import status page? "url recognized" obviously means that Hydrus has already seen a file that has that url as a known url, but the other 2 sound like the same thing. Also, if Hydrus recognizes a url, does it still check to see if the file is the same or if there are additional files at that url, or does it abandon immediately without checking?
>>15802 I realized it probably doesn't care as long as everything's where I left it when I start it up again..
When I open a normal file page, the left frame (the one that has the search box, selection tags and preview) keeps expanding in width beyond what I want it to be. When I shrink the width of the frame again, it'll stay, but any time I open a new page, that page will have the larger width again. Is there a way to explicitly set the size of that part of the window, so that new pages open with it being that width, instead of the larger width that I don't want? I didn't have this problem before, so I don't know what changed. Thanks!
The LoliSnatcher dev has stated that Hydrus has no way to query tags on it's api. This means he can't get tag suggestions (when searching) to show up in his app. Can this be worked on? https://github.com/NO-ob/LoliSnatcher_Droid
I had a good week. I fixed several bugs and added some quality of life, got a Big Sur build working for macOS users to try out, and for users who run smaller servers on a LAN or between friends, hydrus repositories now support changing their 'update periods'. The release should be as normal tomorrow.
>>15802 >>15804 Well, I managed to avoid a full OS reinstall. Opening large images still takes a while though, so it wasn't just my drive dying that was causing it.
https://www.youtube.com/watch?v=-ZdLbiF9gCo windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v435/Hydrus.Network.435.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v435/Hydrus.Network.435.-.Windows.-.Installer.exe macOS NEW app: https://github.com/hydrusnetwork/hydrus/releases/download/v435/HydrusNetwork-v435.dmg OLD app: https://github.com/hydrusnetwork/hydrus/releases/download/v435/Hydrus.Network.435.-.macOS.old.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v435/Hydrus.Network.435.-.Linux.-.Executable.tar.gz I had a good week. I continued last week's server work and have a macOS Big Sur build available for testing. new macOS build This was delayed far longer than I wanted, but I think I have it figured out now. Thanks to work from Suika and ReAnzu on github figuring out the script, the macOS build is now put together on github itself, on Catalina. This new build thus works on Big Sur, and should generally have less compatibility jank on any newer macOS machine. I have included my old 10.12-built release as well today, just in case I messed something up, but if you are a macOS person, please give the new one a try and let me know how you get on. Assuming this works out ok for the majority of people, I'll be moving to it as the official macOS build. Anyone who is still on an old macOS will have to run from source to stay updated with hydrus. The new release is just out of testing, so I'll clean up the filename and add the readme.rtf and so on so it looks prettier. I will also explore reducing the size (300MB vs 137MB wew), but I am told this may be unavoidable due to some new macOS thing that bundles both old and new versions of libraries in the App. editable repository update periods Just in case, it is probably a good idea to back up your server before you try this! Hydrus repositories now have an editable update period! This was an oft-requested feature for users running a server on a LAN or between just a few friends, so if this is you, please give it a go. There is a new 'edit options' permission for accounts, which any account with 'manage account types' (admins basically) will inherit on update. If you have this, the 'admin services' menu will have a new 'edit update period' entry, which launches a simple time delta widget where you can change it. Changes occur immediately. The server will create any updates that are now due, and your client will do a resync to keep up to date with the new numbers, so if you have 'review services' open when you make the change, you should see everything update in realtime. Other clients that sync with the server will catch up with the new check times when they next do their normal sync, unless they too force a resync with the new 'review services' button I added last week. You can change the update several times, up or down, and it shouldn't be a big deal--the client no longer cares if one set of updates were short or long--so please feel free to play around with this a bit, just don't go crazy with super short or long times. The minimum is 10 minutes, but I recommend you only go down to about 60 minutes if you really want to be that fast, at least as we experiment with this. The system was originally built with 100,000 seconds in mind, so I think super short periods will add overhead bloat over time (not to mention, you only see content after processing, which atm only happens in idle time anyway). the rest I had another go at multi column list 'last column' resizing. Dialogs shouldn't judder around so much when you resize them, and more instances of dialogs growing or shrinking slightly on the every re-open should be mitigated. Tag upload to a repository has a new safety check for a particular error state. I know one user got stuck with some pending mappings that wouldn't clear--in fact it lead to an endless cycle of uploading!--which is now caught, and a new maintenance routine can fix it. The Client API's file search now converts tags with asterisks into wildcard or 'namespace' search terms. Its version is now 16. Also added to the Client API help page is a link to https://ififfy.github.io/flipflip/#/ , an advanced slideshow engine that now supports hydrus as a source.
full list - misc: - a new macOS build that should run on Big Sur is now ready, it should be attached to this release. it is built on github automatically, and is thanks to hard work from Suika and ReAnzu. I am attaching my old release as well, just in case I messed up somewhere on my end. if you are a macOS user, please try the new App! it will not work on very old macOS like 10.12, but if this works out today for the majority of macOS users, I will be moving to just putting this new build out going forward. I'll add some polish like the readme.rtf and harmonise the filename etc.. too. I'd love to cut the filesize down, but this may not be possible (it is apparently some modern macOS thing where it bundles old and new versions of libraries in the same App so you basically get it twice) - the bottom-right corner of the regular media viewer canvas now also shows media zoom - the StringSorter object now has a simple 'reverse' sort type - the infamous multi-column list 'last column' width calculations are improved: first, dialogs with multi-column lists should no longer judder back and forth a single character's width as you expand the parent window. also, the last column saved size (which is used in dialog relaunch width initialisation) is now snapped to rounded 5-character intervals, which should mitigate various 'fuzzy' reasons for some dialogs to remember a larger or smaller size and grow or shrink one or more characters' width on the next launch - the _help->debug->gui actions_ menu has a new entry to reset all multi-column list saved widths back to default - the 'edit OR predicate' panel when you shift+double-click an OR predicate now expands horizontally and vertically with the window - the 'edit search predicates' list in the 'edit favourite search' panel now expands vertically with the window - the client now detects some invalid tag mapping states on tag upload--when a mapping is both current & pending or when it is both deleted and petitioned. these pair-states are mutually exclusive, normally