cb7b1b03a3
The Spotify Web API integration stopped working, leaving /spotify unable to suggest songs at all. Replace the artist/album/track search flow with a single box where the user pastes a link to the track on a music service. Links are checked against a hardcoded allow-list of legal streaming and store domains so links to ripped audio cannot be submitted. The check lives in the service layer, not just the page: any authenticated user can POST to /api/ganjoor/song directly and bypass the UI. Hosts are matched against the full host or a dot-prefixed suffix, so look-alikes such as open.spotify.com.evil.com are rejected; https is required, and userinfo and non-default ports are refused. Accepted URLs are canonicalised - https, lowercased host, tracking parameters stripped - so the same track always yields the same stored URL and duplicate detection actually works. Links are stored under one new type, PoemMusicTrackType.MusicUrl, with the platform derived from the host at render time. That needs no migration or backfill, and supporting another service later needs no new enum value. The duplicate check no longer keys on TrackType, which closes a gap where the same URL could be resubmitted as a different type. Several links per poem remain allowed; only an identical URL for the same poem is refused. The Spotify search page and its OAuth plumbing are kept and simply redirect to /musiclink while the existing SpotifyWorking flag is false, so the old flow can be restored if that API ever works again. Track URLs are no longer written through Html.Raw into href attributes. They previously came from the Spotify API; now that they are user supplied, encoding them prevents stored XSS. Also fixes two latent bugs in SuggestSong that this flow would have hit: a null dereference when TrackUrl is empty, and a singer lookup that matched any singer with an empty Url. |
||
|---|---|---|
| docs/getting-started | ||
| GanjooRazor | ||
| RMuseum | ||
| TajikGanjoor | ||
| .gitattributes | ||
| .gitignore | ||
| GanjoorService.sln | ||
| LICENSE | ||
| README.md | ||
| RUNNING_LOCALLY.md | ||
| SEMANTIC_SEARCH_SETUP.md | ||
GanjoorService
Ganjoor museum and ganjoor.net own backend (ASP.NET Core Web API) and frontend (Razor Pages) code
این کد وب سرویس گنجینهٔ گنجور و گنجور و همچنین کد سایت گنجور است.
Running it locally
New to this codebase and want to run your own copy? See RUNNING_LOCALLY.md for a full step-by-step guide — cloning, database setup, and Visual Studio configuration.
The one thing that guide covers in more depth but is worth knowing up front: the production database is never published, since it contains private/user-linked data. What is published is a git repository of the poetry content itself (poets, categories, poems — allowlisted, no user data): github.com/ganjoor/ganjoor-data. A fresh local install can pull real content from there via Admin → مالی و سایت → درونریزی دادهٔ عمومی (also reachable automatically the first time you run the site against an empty database) instead of starting from nothing.
Semantic search
The "find a poem about..." feature has its own setup guide, separate from the main one above — it needs a downloaded ONNX model and published embeddings data on top of the usual database setup: SEMANTIC_SEARCH_SETUP.md. Also has a real troubleshooting section covering the actual problems hit building this feature, not a generic checklist — worth reading before assuming something new is broken.