[nijie] Extractor not working for galleries with multiple images

[cane357]

Broken since the start of this month. For galleries with multiple images, gallery-dl now only downloads tiny thumbnails past the first image.

D:\gallery-dl>gallery-dl -v --config-ignore --cookies-from-browser firefox https://nijie.info/view.php?id=465211
[gallery-dl][debug] Version 1.26.3 - Executable
[gallery-dl][debug] Python 3.8.10 - Windows-10-10.0.19045
[gallery-dl][debug] requests 2.31.0 - urllib3 2.1.0
[gallery-dl][debug] Configuration Files []
[gallery-dl][debug] Starting DownloadJob for 'https://nijie.info/view.php?id=465211'
[nijie][debug] Using NijieImageExtractor for 'https://nijie.info/view.php?id=465211'
[cookies][debug] Extracting cookies from C:\Users\john\AppData\Roaming\Mozilla\Firefox\Profiles\a7zsuovt.default-release\cookies.sqlite
[cookies][info] Extracted 551 cookies from Firefox
[urllib3.connectionpool][debug] Starting new HTTPS connection (1): nijie.info:443
[urllib3.connectionpool][debug] https://nijie.info:443 "GET /view.php?id=465211 HTTP/1.1" 200 None
[urllib3.connectionpool][debug] Starting new HTTPS connection (1): pic.nijie.net:443
[urllib3.connectionpool][debug] https://pic.nijie.net:443 "GET /08/nijie/21/39/581639/illust/0_0_3a5a4225477d44f3_80cc87.png HTTP/1.1" 200 751482
* .\gallery-dl\nijie\581639\465211_p0.png
[urllib3.connectionpool][debug] https://pic.nijie.net:443 "GET /05/__s_rs_l120x120/9ae9d6be25e55b3096cfeaa2a1e7e22d6332f78b5fa01043950bcf93104c4e1d47f35bb1aff89596717387b6a663a4de68860eee0bee008f231a.png HTTP/1.1" 200 5136
* .\gallery-dl\nijie\581639\465211_p1.jpg
[urllib3.connectionpool][debug] https://pic.nijie.net:443 "GET /06/__s_rs_l120x120/9ae9d6be25e55b3096cfeaa2a1e7e22d6332f78b5fa01043950bcf93104c4e1d47f25be5a2a49dcd247687b9f535f4843e8155ee0bb1058a741c.png HTTP/1.1" 200 5424
* .\gallery-dl\nijie\581639\465211_p2.jpg

[KennethSamael]

I see the problem.
https://github.com/mikf/gallery-dl/blob/master/gallery_dl/extractor/nijie.py#L114C67-L114C67
The extractor looks for thumbnail URLs and replaces the /__rs_l120x120/ part to get the source URLs. But now thumbnail URLs seem to contain /__s_rs_l120x120/ instead.

Perhaps a more flexible and thus future-proof regex replace is order? Something like re.sub(r"/\w+?l120x120/", "/", "https:" + src)

[mikf]

I already played around with the new thumbnail format in this regard and transforming them into fullres versions does not seem to work anymore. They now appear to have a completely different format than their fullres counterpart.

It appears to be necessary to also query the /view_popup.php?id= page in addition to just /view.php?id= for posts with more than 1 image.