how do I pause a search query in the gallery downloader? I right click some queries or just one and I click play/pause search. However it doesn't stop, it still shows "working" and search status is still a blue square. I want to pause them because I'm filling up my storage too fast lmfao

>>15814 I mean that's how I've paused queries before. But since I've only ever paused to be able to restart the client without interrupting anything, sometimes I have to wait for it to actually pause, it appears. It's like, I ran out of bandwidth for the site, so it needs bandwidth available to actually pause, for some reason, I think.

>>15811 >I have unavoidable IRL next week (jury duty)

>>15816 https://www.youtube.com/watch?v=uqH_Y1TupoQ >The Law You Won't Be Told >On a jury you know your options: guilty, or not. But there's another choice that neither the judge nor the lawyers will tell you, often because they're not allowed to, and also because it might be better if you don't know. >This video will tell you that third choice, but be warned, simply watching may prevent you from ever serving on a jury. So this is your last chance to hit the pause button before you learn about... [redacted]

>>15817 Respond to myself here, but to clarify, I'm not saying I agree with the belief that jurors aren't supposed to know their options; I was just parroting the quote pointing that out since it was the only other time I've heard of this.

>>15818 can anyone share permanent booru downloaders? http://owmvhpxyisu6fgd7r2fcswgavs7jly4znldaey33utadwmgbbp4pysad.onion api doc mentions hydrus, so you got to have this, right?

>>15819 I think they meant that one particular option their API has would be useful in Hydrus. I'll try to make a basic parser this weekend.

Is it possible for Hydrus to grab text posts?

>>15821 If you mean text files, then no. This is a somewhat highly requested feature though. It's certainly one of the ones I want the most!

So I've had this bug pop up where Hydrus says that there things that need to be uploaded to the PTR, but why I try to, it just says "upload done" without doing anything. Trying to forget the changes does nothing at all. If there are other things that actually need to be uploaded, the number will increase accordingly and those will upload when I click to upload them, but then it goes back down to 994 and just stays there. Is my DB in trouble, or can this be fixed easily?

>>15819 >>15820 I tried. Here's one that lets you either search via tag or download from URL. It gets metadata like tags and post URL too. This is my first try at a downloader so it's probably shit. If anyone else knows more about creating downloaders and wants to try and improve this, please do.

A downloader for Poipiku; pretty bare-bones as far as metadata goes, but its not like Poipiku shares any tags, plus it hides the download in a JSON that only has the download link. Still, should work well enough for subscribing to specific artists. BTW I barely tested this, who knows how long this will work.

>>15824 Thanks, it's good enough for me.

Hey, while it looks like the macOS build works well, it is having trouble finding an existing database location. I will fix this for 436. If you tried the new app and it started empty with the 'this is the first time you ran the client' popup, please either: - Roll back to the old version for a week - If you know how, run the client with: open -n -a "Hydrus Network.app" --args -d="/path/to/db" The path should be ~/Library/Hydrus by default, which (you should check) should be something like: /Users/(YOU)/Library/Hydrus

Hi. Is there a way to pass the Cloudflare capcha check during login to https://e621.net account?

>>15795 Thanks. Yeah, I am not sure how much I can do here with the current setup. I think there are probably other ways of arranging the database to reduce writes--maybe splitting up the giant mappings tables into smaller pieces in some clever way--but in general, the way that hydrus processes PTR data requires a lot of writes. I will keep investigating this issue. If you are interested in other journalling modes, I now offer everything here: https://sqlite.org/pragma.html#pragma_journal_mode using the --db_journal_mode here https://hydrusnetwork.github.io/hydrus/help/launch_arguments.html There is a MEMORY journal, although that pragma article suggests it is too dangerous to use on a db with permanent data like hydrus. Hydrus default is WAL, which generally needs one write for every change and then a merged write when it checkpoints (for hydrus, every 30 seconds). I am very interested to hear and investigate more about the writes you are seeing with file imports. I do not understand this, and when I tested it on my IRL client (2 million files, syncs with PTR) a bit, I did not see a huge spike in writes. I may have not been looking in the right way though. I let my subs run for a bit while watching task manager, and I generally saw some write spikes of a few MB as files were actually imported and written to my HDD, but I did not see a huge amount of SSD db activity. It didn't break it down into a very detailed form, though. Can you say more about the data you are seeing written to the db on file import? Normally I would expect a few dozen tags every second to only be a handful of kilobytes here and there--it is precisely because the PTR is processing a million or so at a time that it adds up to gigabytes. Forgive me if this sounds stupid, but could the writes you are seeing actually be the media files themselves to your temp dir, and/or to your file storage dir? When hydrus imports files to from the internet, it first stores them in your temp directory, which is usually on your system disk. If you would like to change that location, you can with another launch argument as in the help file above. I think I will make it so you can change the transaction commit period in a launch argument. If you would like to play with the 'commit every 30s' value, I would be interested in your results.

>>15828 Best solution for CF atm is to get Hydrus Companion https://gitgud.io/prkc/hydrus-companion and sync your browser cookies over to hydrus and also copy your browser User-Agent header to your hydrus (network->data->manage http headers). Then your hydrus basically looks like your browser to CF, so you can 'log in' to the site with your browser and then your hydrus should be passed on its behalf. I include 'cloudscraper' in the network engine, but I imagine your are gettings its CF-specific error messages in your UI because CF is giving you a harder time than cloudscraper can handle for whatever reason. If you are on a VPN, I understand IP-hopping to another country is also a sometimes solution. The 'I'm under attack mode's that CF offers sites are apparently sometimes region based.

>>15796 Thanks again, took out that outdated help here https://hydrusnetwork.github.io/hydrus/help/client_api.html If you turn on https for the Client API, then everything in the GET or POST is encrypted with that, and any observer or intermediary who can't see through that https won't see the key no matter where it is in the request. That old worry about the key optionally being in a GET arg is less of a bother now https is available, and as this system matures, I expect I will ultimately phase out http as an option at all. Btw, if there is any confusion about the naming here, the 'access key' in this context is equivalent to the 'password' to the Client API, it isn't anything to do with cipher stuff. That nginx template looks neat. I do not have much experience setting up that sort of server, so that stuff is new to me. Hydrus is a bit idiosyncratic and unhelpful for some of those systems, so it can take a bit of extra work to set up. Same on both counts with docker setup scripts. But if you end up using the Client API or the Server and figure out a neat easy setup script, or you otherwise encounter one, please let me know and I'll fold it into the help! >>15797 >>15798 >>15799 I am pretty terrible at keeping up with social media already, and I don't use it at all outside of my hydrus work. I hate doing it myself, and I am bad at it too. Most of my history getting involved in the running of social platforms either ends in drama bombs or has me not running things with enough attention. I don't run the hydrus discord myself for this same reason--I just visit it. So, if you would like an alternative, I'll have to ask that it be user run. I'll happily put up a link in my help and talk about it in a release post. If it gets a decent population, then I will check it for bug reports and so on. Sorry I can't be more pro-active here, but this is just not my wheelhouse. I keep putting off crypto coin wallet thoughts too, I am sorry. I don't have any real experience with it, and I always have so many other things going on that I can't summon the energy to read up on it, learn what is bullshit and what is real, and figure out what is best to do. I appreciate your kindness, and I am glad you like my program.

>>15802 >>15804 >>15808 I am glad you figured out your drive situation. I lost a drive about ten years ago, somewhere between 50k and 75k files. I always keep a backup (and now a second backup) that I sync regularly. While it is a bit of money to set up, it is hugely worth it--I never really have to worry about a drive blip any more. If it helps, I have more here: https://hydrusnetwork.github.io/hydrus/help/getting_started_installing.html#backing_up You are correct that as long as hydrus's db stays intact, you can move it anywhere. Even when you break it into bits and then need to move them around during a recovery, there are some repair routines on boot to help you stitch it back together. As long as you have your files, we can figure out a recovery. The slow image rendering when it comes to big files is mostly down to my bad code not being able to deal with large files well, as here >>15761 . If you get slow thumbnails, I strongly recommend you move them to an SSD (along with the db itself), as here: https://hydrusnetwork.github.io/hydrus/help/database_migration.html I am going to be blocked up with multiple local file service work for a bit, but I have a 'medium size' job week coming up in two weeks for 438. Perhaps that would be a good time to dive into the image renderer and image neighbour cache and see if I can clean things up.

>>15803 The basic routine when hydrus downloads a booru post page like: https://safebooru.donmai.us/posts/4442487 is: 1) Look at the URL 2) Download the HTML/JSON in the page 3) Download the file If hydrus recognises the URL and can do enough 'I already have that' logic, then it stops at step 1, nothing downloaded. If hydrus can pull (usually md5) hash data from the HTML/JSON and it can do logic on that, it stops at step 2, just the lightweight page content downloaded. If hydrus downloads the file and generates a SH256 hash from it and goes, 'oh, I know that file already', then that resolves step 3, everything downloaded but no file import actually done. In the case of step 2 and 3, the URL is still associated with the file, so only step 1 needs to be done in future. Normally, hydrus tries to save you time and bandwidth. If the logic is good--basically, that the URL only refers to one file in its record, and the file only has one URL at that domain, although there are caveats--then it won't check that URL again. You can change this behaviour under 'file import options' and 'tag import options', which I recommend only for special one-time jobs. Some situations are special. Pixiv is declared in hydrus as a place that can have multiple files per post, so hydrus always proceeds to step 2. Then, if the artist added files since some search you did last year, it'll catch them. And when hydrus gets bad URL relationships (usually from incorrect 'source' info from other boorus) so a file appears to be in different places on the same booru, then hydrus will not trust its URLs for comparisons and may end up going to step 2 for that particular file hash or URL.

>>15805 Yeah, check, pages->management and preview panels. You can save the current page's 'sash' width as the default, force all to the default now, and control whether the current page's width is saved on exit. >>15806 I think the GET /get_files/search_files call https://hydrusnetwork.github.io/hydrus/help/client_api.html#get_files_search_files should do that, but it doesn't work like a normal booru--there is no 'pagination', it will give all results at once. I just added wildcard support to the tags here last week, and I hope to add more system predicates and better sorting in future. EDIT: Ah, shit, I realise this means the autocomplete tag search results, like 'sam*' -> [ 'samus aran (30)' ]. Yes, I would like to add this. If you are interested, here is the masterlist of things I need to spend time on here: https://github.com/hydrusnetwork/hydrus/issues/656

>>15814 >>15815 It'll usually get the last page it was working on. If you double-click the downloader and look at the more detailed panel below, it should say something like 'pausing -- downloading gallery page'. My system isn't clever enough to pause mid-job yet, so pause only triggers between jobs. If the thing seems to be lagging ages before it pauses, it could be it is competing with several other downloaders on a busy domain, so it could take a couple of minutes on a really busy downloader, to finish that job. Pause more and resources will free up. If you want to force it to finish that job right now, highlight the downloader, look at the job it is working on where it says 'waiting xxx seconds for yyy', and click the 'cog' icon. You can force override its 'slot' and 'bandwidth' requirements to fire it off immediately. If the search status is really a square, so searching has stopped, but you see an 'x/y' file progress like '90/120', then the downloader is still downloading files from the search it did. Like you hit 'pause/play search', do 'pause/play files', and it'll stop after the next file is done, and the 'working' status should clear. >>3550 Hey, I think I was just helping you somewhere else. Let me know if that is not true. In any case, I will be cleaning up that error message.

>>15823 Thank you for this report, I am sorry for the trouble! Hey, I have had reports from several people of pending counts that don't exist! This is due to the new 'efficient' tag counting system I added last week, it has revealed some legacy miscounting in the autocomplete system, likely from last year as the siblings system was coming together. These pending counts are fake, please bear with them for now, I will make a maintenance routine that will efficiently correct these old miscounts and get us back to a proper and fast 0!

>>15836 Okay, that's fine for now. My main concern was that this issue was actually a symptom of a deeper corruption of my DB. I just wanted to make sure that there wasn't something seriously wrong with my database that would get written into my backup, and accidentally doom the backup too. Looking through the thread, it looks like I'm not the only one who posted here recently with an issue related to pending tags. That must be what you're talking about. Good to know that there's a fix on the way!

>>15831 >I don't have any real experience with it, and I always have so many other things going on that I can't summon the energy to read up on it, learn what is bullshit and what is real, and figure out what is best to do. That's understandable. At least for Monero, the general consensus is that the official wallet is best. There's a GUI and a CLI version, if you ever get the time and feel like looking in to it.

>>15833 This is interesting info. I think this should be put on the website somewhere, maybe on the help page that explains downloaders. Also, is Pixiv a special case that's hardcoded into Hydrus, or is that special treatment about the possibility of new files on a post something that's part of the downloader. Could that treatment be added to user-made downloaders for other website that can have more files added to a post like Pixiv can?

I'm a former Windows user that has migrated to Manjaro Linux. I have migrated my Hydrus DB to Manjaro and there appears to be no issues, but how do I update to a new version of Hydrus? I am used to using the Windows installer.

>>15840 How did you install Hydrus? I'm on an Arch-based distro as well and I just use the AUR package.

>>15841 The .tar.gz download from GitHub.

If I double click twice to close two pages it closes three (two on the second click) instead. I'm pretty sure it's something with hydrus since it doesn't happen anywhere else. Is there a way to turn this off?

The thread is past 700 replies, and the post limit is 750, it's almost time to start a new one. Afterwards, I will move this one to >>>/hydrus/.

>>15844 Thank you--how about I make one with my release post tomorrow? ----- I had a great few days mostly fixing bugs and cleaning things. The new macOS build is more polished, and I fixed an issue where it could have trouble finding your default database location. I believe I also fixed the issue some users had with 'sticky' pending menu counts, then added some UI quality of life, and wrote an experimental tag display mode that replaces underscores with spaces. The release should be normal time tomorrow. I have unavoidable IRL on Wed, so it is moved up a day.

can I add custom namespaces to collect by?

>>15842 I usually do everything using bash, so I have this script I named hydrupdate: #!/usr/bin/env bash test -z "$1" && exit INSTALLDIR='/your/install/dir/here/' test ! -d "$INSTALLDIR" && echo 'The provided install dir does not exist or is not a directory' && exit tar xf "$1" -C "$INSTALLDIR" --strip-components=1 I can then call the script with hydrupdate /path/to/Hydrus.Network.<release>.-.Linux.-.Executable.tar.gz.

New thread for v436 >>3626 !

>>15839 Thank you, I will! These questions come up enough, and I sometimes forget the details myself, that there should be an official write-up somewhere. >>15840 The other guys are also correct here, but if you are using one of the 'extract' releases, basically you just drag and drop the contents of the archive onto your install directory. Longer explanation here: https://hydrusnetwork.github.io/hydrus/help/getting_started_installing.html#updating >>15843 Thanks, I will have a look. I guess the third click is getting sent by Qt as a second doubleclick event. I'll see if I should be terminating the click-chain or whatever it is called when I eat the initial double-click. >>15846 Yep, check options->sort-collect. The UI is debug-tier bullshit, I apologise, but if you add some new sort types, any namespaces in those sorts will be added in your collect dropdown.