This commit is contained in:
Simon
2026-07-06 11:51:52 +00:00
parent 4dc1072d86
commit 1f9977062c
3 changed files with 678 additions and 0 deletions

View File

@@ -49,6 +49,7 @@ This is the current implementation inventory as of this snapshot of the repo. Us
| `pornmz` | `mainstream-tube` | no | no | Mainstream archive. |
| `pornzog` | `mainstream-tube` | no | no | Basic list/detail scraper. |
| `porntrex` | `mainstream-tube` | no | no | KVS-style HTML archive with direct MP4 formats and tag-aware search shortcuts. |
| `redgifs` | `amateur-homemade` | yes | no | Direct integration against the public RedGifs v2 API (same backend `xxxtik` proxies through, but consumed natively here). Auth: anonymous bearer token from `GET api.redgifs.com/v2/auth/temporary`, cached in an `Arc<RwLock<Option<String>>>` with double-checked locking, refreshed once on a 401 via a centralized `authed_get` helper. Default feed: `GET /v2/feeds/trending/{popular,established}?page=&count=` (honors `count` exactly). Free-text query and `tag:`/`category:`/`cat:` shortcuts hit `GET /v2/gifs/search?query=|tags=&order=&page=&count=`; `user:`/`uploader:`/`creator:` shortcuts hit `GET /v2/users/{username}/search?page=&count=`. Uploader profile (`GET /v1/users/{username}`) backs `/api/uploaders`, with `profileContent:true` pulling the creator's own gif listing for `videoCount`/`totalViews`. Listing JSON already carries duration/poster/tags/uploader per item, so no per-item enrichment call is needed (`build_video_item` is a plain sync function, unlike `xxxtik`'s `buffer_unordered` resolve step). `video.url` is the yt-dlp-native `redgifs.com/watch/{id}` page; `formats` intentionally left unset. Thumbnails (`urls.poster`/`urls.thumbnail`) are directly hotlinkable, no proxy needed. 20 curated tags exposed via `categories`. **Known upstream quirk:** `/v2/gifs/search` (query/tag targets only — trending and creator listings are unaffected) returns noticeably fewer items than the requested `count` on some pages (e.g. `count=20` → ~14); not compensated for by over-fetching, since inflating `count` while keeping the same `page` would desync the API's own offset-based pagination cursor (`page × count`) and risk skipped/duplicated items across pages. |
| `redtube` | `mainstream-tube` | no | no | Mainstream archive. |
| `rule34gen` | `ai` | no | no | AI group example. |
| `rule34video` | `hentai-animation` | no | no | Hentai group example. |