[sankaku] many "404 Not Found" images with URL https://v.sankakucomplex.com/data/... are actually https://s.sankakucomplex.com/data/...

[mo-han]

it's actually an old issue, which i have encountered a lot of times, but not intensive.
so i thought they were just some "expired resources", until now...

https://v.sankakucomplex.com/data/...
https://s.sankakucomplex.com/data/...
most time the galledy-dl extracts good URLs
never pay attention to whether they are v or s
but when it goes bad, the URLs seem all be v, which is 404
paste the wrong v URL into browser and also get 404
then replace it with s, image reloaded successfully
these 404 images seem to be concealed by sankaku, they don't appeal in web search results and have a different URL domain

example:
https://chan.sankakucomplex.com/post/show/24774189

i didn't dig into the source code so have no idea if the v is hard coded or returned by sankaku server (which is wrong)
anyway please fix this asap
i'm really frustrated to realize that i've downloaded a lot of incomplete galleries (there were some 404 during downloading) in the recent months, but what happened is happened, i guess i'll just let it go

[kattjevfel]

Your example works just fine here, does it only happen sometimes?

[Twi-Hard]

I noticed the same problem a few hours before this issue came up. I just ran the same search again and now some of the images that 404'd downloaded successfully but not all of them. Changing the v into an s on them makes the links work. Something interesting about this is all of the images that 404'd were next to each other. Everything before and after these images that 404'd didn't 404. I didn't get far into the download before I stopped (~1500 images) so it's likely this happens later on too.
Edit: out of the 20 images that 404'd, 9 downloaded when I reran it just now

[mikf]

v.sankakucomplex.com was the sankaku file server for quite some time, judging by the expected test URLs from before commit 32c75d1:

gallery-dl/gallery_dl/extractor/sankaku.py

Line 82 in d5ded11

"pattern": r"https://v\.sankakucomplex\.com/data/[^/]{2}/[^/]{2}"

Sankaku probably tried to move everything (or at least all new files) to the new s. domain, but made some mistakes along the way.