/hydrus/ - Hydrus Network

>I am also fairly IRL busy and a bit stressed, although I expect that to clear up in a couple of weeks.

Take a break as you need it my dude. You work on Hydrus practically every week so it's not like you don't deserve a vacation. Besides, I doubt an extra week of waiting to update will kill anyone.

>downloader engine overhaul

does this include fixes to parsing script issues?

Thank you so much for doing

>>5645

>I will be adding a way to manually edit dupe relationships from the thumbnail menu. Everything you can do in the dupe filter will be possible through it.

I am very grateful!

Please make sure you take some rest as needed. Better to take a short break when you can still go on than to take a long break because you can no longer go on.

>>6036

Yes.

>>6037

Imagine forcing him to wear a diaper and feeding him grapes. Wouldn't it be funny? Just a joke haha

This fixed a whole bunch of my failed webms. Cheers, Hydrus_dev, for fixing an issue that probably wasn't affecting anyone else but me.

feature request:

please add a "trash both" option to duplicate sorting. oftentimes, i come across a pair of images where neither of them are images i want to keep. being able to discard both with one click would make things way, way nicer.

Awesome job! I only have one question about the downloader. I am trying to download off of Rule34Hentai and I'm wondering if there is a way to add an account login like you can with pixiv? Some of the images don't show up unless you're logged in, and they aren't getting downloaded. Is it somehow possible to do this?

>>6063

Not to get in the way of your dreams, but making this an icon is risky. People will click the wrong icon and get unexpected total information loss results.

Alt-del or something would probably be safer already.

>>6063

You can already click 'delete selected'? No need to make two buttons for the exact same action I'd say.

>>6066

That only deletes one and immediately moves on to the next pair so you don't get the chance to delete the other one

>>6065

Could be hold shift while clicking the button

>>6063

>>6065

>>6069

>>6070

You can already do that with the custom option button, by selecting any option from it and then "delete both files".

Slideshow functionality seems to be broken on Windows 10.

So the update one maybe 2 weeks ago had a comment that said they've noticed an increased load speed when they opened up the browser.

Normally for me it takes may be 4 to 5 minutes for the program to load up, I use it in a very stupid way, and to top it off I am very lazy with going through the images so this leaves may be 80 to 90 tabs open, possibly more, about 10 of which are over 3000 files big and one is for over 50,000. However after the person mentioned that and the program crashed twice, I noticed that took under a minute to load up. Now going to one of the download tabs in the program at the time and trying to sort the errors would crash it 9 times out of 10 if not every time I stopped after it started crash constantly so I had to restart the program multiple times and I watched it one of the times, the program loaded up the images that may be 5-600 images a second at some points. I'm not sure what changed to make that happen, I don't know if it's a setting that I could make happen all the time, but having the program crash while going through the settings menu just now and loading it back up and it reverting back to being really slow makes me really miss how fast that was.

Something is in there that makes loading the program potentially fast, but I don't know what it is. I don't know how useful this would be to everyone else, but I'd look into the reason why it's faster and see if you can make it happen every time.

If somebody knows how to make this happen every time I'm all years and would be glad to learn.

>>6084

loading off a 5200 rpm archive drive till nvme 2tb gets under 300$ and when the program normally loads, it will go 120 images, freeze, then 250 images 5 or so seconds later.

500 may be low considering how fast it loads some times, but even 500 is significantly faster compared to what I am use to.

Ok going through the back catalogue of shit to download/go through made me think.

Thread watcher only works for 4chan/8chan at the moment, and effectively it does the same job as the download page of images, I don't know what raw url and gallery do at the moment so I cant comment on them.

However, would there be a way to contextually filter what you want the program to do through dragging and dropping urls?

An example, drop a 4chan or 8chan link to the program pops up thread watcher with all you needing to do is press the start button, the 4chan archive hosts would default to page of image downloads along with a 4chan page if its seen to have 404'ed, honestly dont use 8chan enough to know how bumped off threads work here to know if that's an option or possible.

It also seems like sites like hentaifoundry could be smartly scrapped, I'm not 100% sure on that as help is a bit lacking/ui for it is a bit lacking, however lets assume that it works perfectly, something like

https://www.hentai-foundry.com/search/index?query=anal

could bring up a partially filled out section to either subscribe to it, or scrape it, one or the other, possibly a scrape with a subscribe at the same time.

Or from the same site

dropping a link like

https://www.hentai-foundry.com/user/Rampage0118/profile

would tell it to scrape + subscribe

Or, like in the help

https://gelbooru.com/index.php?page=post&s=list&tags=anal+pokemon+hetero

could see that it want gel, that it wants anal, pokemon, and hetero, will scrape it then subscribe.

With 4chan like sites with thread watcher function, possibly board specific rules, while /3/ will be fine if you checked it just every other week, a thread on /b/ needs to be checked every 5 minutes or it could 404. Per board settings with a global setting for how often it checks would be very helpful, along with a till 404 setting instead of finite check numbers. For some reason I got déjà vu writing this.

have another thing for you to look at when it comes to downloading links.

https://archived.moe/h/thread/4343362/

This is from the /h/ archive, apparently every I didnt let this one go all the way as its 200+ links but the other page was fully failed because of this.

loading the page up in a different browser without scripts, I can clearly see the issue,

for example, this image here

https://archiveofsins.com/h/full_image/1464049607696.png

shows up as

https://archived.moe/h/redirect/1464049607696.png

With the script.

A potential solution would be keeping up with the archives and making workarounds for said archives as they change up their method of displaying images, or, allow the program to go down the rabbit hole by a user definable level.

lets say you have a website that has 1 link on it, that would be the surface, and what it currently does is go down one hole to find it, but what if that hole leads to a redirect page to the image, letting the user tell it to make 2 jumps would solve the problem,

for the sake of keeping up to date with archives and how they work, a user definable/importable script would be for the best long term, but for the immediate letting the program go further then one jump would solve the issue, Thankfully for me i only have 3 pages of links and I can go through them one at a time and not care too much, everything else is on desu archives and that's been pretty friendly so long as the images aren't lost.

One thing I'm really missing from the system is a way to force hydrus to forget a file's dupe status. Say if I reimport after months a file that I previously marked as dupe or whatever, or even a batch of them. There's no way to force hydrus to make me go through them again.

An option in the right click menu to tell hydrus to "reset this file's potential dupes" or something would save me.

So i'm going through the boards on 4chan like normal for me, 3am ish looking for images and threads the pique my interest, and I realise something, if I say imported each thread that I wanted the images posted in it to hydrus, I would have well over 50 thread watching tabs

If there is a way to watch threads all in one thread watcher tab, like there is for download page of images that would be very helpful

something like this

[site] [board] [thread] [check time] [check amount] [time till check]

in a list it would be like this

[4] (as there are only two sites currently working thats all thats needed) [b]

[2183904712] [7m] [5m45s] [td] (till dead)

[8] [pone] [289896] [4h] [1h34m52s] [td]

[8] [monster] [191634] [4h] 3h21m42s] [td]

so on so forth, with text being black thread isn't done, red for it's done, and bold red for something is happening and program doesn't know. the bold one would be something like the website is down, or something unexpected happened and it doesn't know what to make of it.

if the watched thread is selected/highlighted, you should be able to edit parameters, such as the re check time, amount of times, a global pause, a stunted pause, and a priority (priority would be something like /b/ where the thread could 404 if not checked right away, and priority could be handled through check times) and when right clicked you can get the option to visit the thread it came from. pair this with a smart drag and drop function and it could be very powerful/useful.

>>6093

CTRL+C on a thumbnail will already copy the file path to the clipboard.

Discord's upload issue is not hydrus' fault, it's discord's.

>>6094

Figured it as much, however one program not working with it means others will too,

Did not know about the copy paste was an option to get the path, with that said, with that said, looking into the context menu a bit more, most of the options beside show in folder are under the share menu, a menu that I have never touched in other programs due to how useless it is.

I can understand the reason they are under share in this one, but it did take till you told me the function is in the program to even look.

>>6103

> however one program not working with it means others will too,

Discord is a special case due to their security/privilege protocols. In any case the dev can do nothing about it.