soxoj/socid-extractor

The vBulletinEngine scheme exists, but it extracts no identifier at all — status, country and image, and nothing that identifies the account. Two of those three fields are also dead: sprite_flags and online-status appear on none of the forums still running the engine, measured across all 131 vBulletin records in maigret's database on 2026-09-08. The numeric id is there, in two different places depending on the version: vBulletin 4 links it directly — member.php?u=<id>. Verified: AllTheLyrics gives 144363. vBulletin 3 has no such link, but carries it in the "find all posts by this user" URL — search.php?do=finduser&u=<id>. Verified: http://forum.igromania.ru/member.php?username=blue gives 413781 and http://www.lada-vesta.net/member.php?username=alex gives 106. What to do: add uid and username to the scheme, covering both versions. Removing the two dead fields is welcome. Field names follow FIELDS.md. Where to start: the scheme is in socid_extractor/schemes.py, search for vBullet

Flarum forums answer a JSON:API endpoint, and a scheme for it is a short one — a good place to start if you have not touched this repository before. The endpoint is {forum}/api/users?filter[q]={name}. Verified on 2026-09-11: https://forum.android.com.pl/api/users?filter[q]=rammiro returns data[0] with id 3 and an attributes object holding the username, display name, join date and avatar URL. Twenty of the sites in maigret's database run Flarum. What to extract: uid and username at minimum, plus what maps onto FIELDS.md — fullname, image, created_at. Where to start: schemes live in socid_extractor/schemes.py. Discourse API is the closest model. Note that this endpoint is a search, so data can hold several users or none — take the exact match, not the first element. Watch out for: some instances refuse anonymous API calls. https://forum.exbo.net/api/users?filter[q]=anyone answers 403. The scheme should come back empty there rather than break. Done when: an offline test in tests/tes

Lemmy is the fediverse's link aggregator, and its instances answer a public JSON API that socid_extractor does not use yet. One scheme covers all of them. The endpoint is {instance}/api/v3/user?username={name}. Verified on 2026-09-11: https://lemmy.nz/api/v3/user?username=Dave returns person_view.person with id 2, the display name, avatar, bio and publication date. What to extract: uid and username at minimum, plus what maps onto FIELDS.md — fullname, bio, image, created_at. Where to start: schemes live in socid_extractor/schemes.py. Discourse API is the closest model — a JSON endpoint reached with url_mutations from the profile URL. Watch out for: instances run different API versions, and the path is versioned. https://feddit.org/api/v3/user?username=anyone answers 404 although the site is alive. Don't treat that as your bug, and don't assume v3 everywhere. Done when: an offline test in tests/test_extended_schemes.py passes against a saved JSON response — no live request, no coo

NodeBB forums answer a clean JSON API that socid_extractor does not use yet, and one scheme covers every instance of it. The endpoint is {forum}/api/user/username/{name}. Verified on 2026-09-11: https://bbs.aw-ol.com/api/user/username/whycan returns uid 44 along with username, userslug, joindate, lastonline, reputation and postcount. What to extract: uid, username, and whatever of the rest maps onto FIELDS.md — created_at from joindate, posts_count from postcount. Those two are unix milliseconds, not seconds. Where to start: schemes live in socid_extractor/schemes.py. Copy the shape of Discourse API: same idea, a forum engine with a JSON endpoint reached through url_mutations from the human-facing profile URL. Watch out for: some instances close the API entirely — https://forum.qt.io/api/user/username/anyone answers 401. The scheme has to survive that rather than assume the path always works. Done when: an offline test in tests/test_extended_schemes.py passes against a saved JSON r

phpBB is the largest forum engine socid_extractor has no scheme for — 155 of the sites in maigret's database run it, and one scheme covers all of them at once. There is no API. The numeric user id sits in the profile links of the member list: memberlist.php?mode=viewprofile&u=<id>. Verified on 2026-09-11: https://unixforum.org/search.php?author=alex gives id 119 and http://jeepspb.ru/forum/search.php?author=red gives id 2965. Eight of the thirteen forums that answered exposed it this way. What to extract: uid and username at minimum; registration date and post count are welcome on top. Field names follow FIELDS.md — invent a new one only when nothing there fits. Where to start: schemes live in socid_extractor/schemes.py. The closest model to copy is XenForo — the same situation, a forum engine with no API where the id has to be read out of the markup. Read its comments first: the traps it documents apply here too, in particular that a member search finding nobody falls back to a