fix PixivFavoriteExtractor regex

[beesdotjson]

Output below is the matches of each URL in a list [uid, kind, self.tag, query], based on

gallery-dl/gallery_dl/extractor/pixiv.py

Line 300 in 91c2e15

uid, kind, self.tag, query = match.groups()

Before:

https://www.pixiv.net/en/users/173530/bookmarks/artworks ['173530', 'bookmarks/artworks', None, None]
https://www.pixiv.net/bookmark.php?id=173530 [None, None, None, 'id=173530']
No match: https://www.pixiv.net/en/users/3137110
https://www.pixiv.net/bookmark.php?id=3137110&tag=%E3%81%AF%E3%82%93%E3%82%82%E3%82%93&p=1 [None, None, None,
'id=3137110&tag=%E3%81%AF%E3%82%93%E3%82%82%E3%82%93&p=1']
https://www.pixiv.net/bookmark.php [None, None, None, None]
https://www.pixiv.net/bookmark.php?tag=foobar [None, None, None, 'tag=foobar']
https://www.pixiv.net/en/users/173530/following ['173530', 'following', None, None]
https://www.pixiv.net/bookmark.php?id=173530&type=user [None, None, None, 'id=173530&type=user']
https://touch.pixiv.net/bookmark.php?id=173530 [None, None, None, 'id=173530']
https://touch.pixiv.net/bookmark.php [None, None, None, None]
https://www.pixiv.net/en/users/173530/bookmarks/artworks?rest=hide ['173530', 'bookmarks/artworks', None,
None]
https://www.pixiv.net/en/users/173530/bookmarks/artworks/未分類?rest=hide ['173530', 'bookmarks/artworks/>
未分類', '未分類', None]

Of note:

• https://www.pixiv.net/en/users/173530/bookmarks/artworks?rest=hide ['173530', 'bookmarks/artworks', None, None] -> does not group "hide" option (private bookmarks)
• https://www.pixiv.net/en/users/173530/bookmarks/artworks/未分類?rest=hide ['173530', 'bookmarks/artworks/> 未分類', '未分類', None] -> tag "未分類" included in "kind" field (i think this is undesired behavior)

After:

https://www.pixiv.net/en/users/173530/bookmarks/artworks ['173530', 'bookmarks/artworks', None, None]
https://www.pixiv.net/bookmark.php?id=173530 [None, None, None, 'id=173530']
No match: https://www.pixiv.net/en/users/3137110
https://www.pixiv.net/bookmark.php?id=3137110&tag=%E3%81%AF%E3%82%93%E3%82%82%E3%82%93&p=1 [None, None, None,
'id=3137110&tag=%E3%81%AF%E3%82%93%E3%82%82%E3%82%93&p=1']
https://www.pixiv.net/bookmark.php [None, None, None, None]
https://www.pixiv.net/bookmark.php?tag=foobar [None, None, None, 'tag=foobar']
https://www.pixiv.net/en/users/173530/following ['173530', 'following', None, None]
https://www.pixiv.net/bookmark.php?id=173530&type=user [None, None, None, 'id=173530&type=user']
https://touch.pixiv.net/bookmark.php?id=173530 [None, None, None, 'id=173530']
https://touch.pixiv.net/bookmark.php [None, None, None, None]
https://www.pixiv.net/en/users/173530/bookmarks/artworks?rest=hide ['173530', 'bookmarks/artworks', None,
'rest=hide']
https://www.pixiv.net/en/users/173530/bookmarks/artworks/未分類?rest=hide ['173530', 'bookmarks/artworks',
 '未分類', 'rest=hide']

[mikf]

tag "未分類" included in "kind" field (i think this is undesired behavior)

That doesn't really matter, since kind is only used to check whether to fetch followed users or not:

gallery-dl/gallery_dl/extractor/pixiv.py

Line 312 in 62cfee4

if kind == "following":

Of course it'd be better if this didn't happen, but I'd take the current behavior over a regex lookbehind.

does not group "hide" option (private bookmarks)

So essentially the issue is that new-style URLs (/users/173530/bookmarks/artworks) don't include query parameters, isn't it? Then how about:

    pattern = (r"(?:https?://)?(?:www\.|touch\.)?pixiv\.net/(?:(?:en/)?"
               r"users/(\d+)/(bookmarks/artworks|following)(?:/([^/?#]+))?"
               r"|bookmark\.php)(?:\?([^#]*))?")

Fetches query parameters for old- and new-style URLs, doesn't include tags after /artworks, and doesn't use a lookbehind.

https://www.pixiv.net/en/users/173530/bookmarks/artworks ('173530', 'bookmarks/artworks', None, None)
https://www.pixiv.net/bookmark.php?id=173530 (None, None, None, 'id=173530')
https://www.pixiv.net/en/users/3137110/bookmarks/artworks/%E3%81%AF%E3%82%93%E3%82%82%E3%82%93 ('3137110', 'bookmarks/artworks', '%E3%81%AF%E3%82%93%E3%82%82%E3%82%93', None)
https://www.pixiv.net/bookmark.php?id=3137110&tag=%E3%81%AF%E3%82%93%E3%82%82%E3%82%93&p=1 (None, None, None, 'id=3137110&tag=%E3%81%AF%E3%82%93%E3%82%82%E3%82%93&p=1')
https://www.pixiv.net/bookmark.php (None, None, None, None)
https://www.pixiv.net/bookmark.php?tag=foobar (None, None, None, 'tag=foobar')
https://www.pixiv.net/en/users/173530/following ('173530', 'following', None, None)
https://www.pixiv.net/bookmark.php?id=173530&type=user (None, None, None, 'id=173530&type=user')
https://touch.pixiv.net/bookmark.php?id=173530 (None, None, None, 'id=173530')
https://touch.pixiv.net/bookmark.php (None, None, None, None)
https://www.pixiv.net/en/users/173530/bookmarks/artworks?rest=hide ('173530', 'bookmarks/artworks', None, 'rest=hide')
https://www.pixiv.net/en/users/173530/bookmarks/artworks/未分類?rest=hide ('173530', 'bookmarks/artworks', '未分類', 'rest=hide')

[beesdotjson]

Avoiding the look behind is a good idea. I should have tinkered with my regex a little more before settling upon it.

[mikf]

Thank you