mirror of
https://github.com/JustinZeus/Wholphin.git
synced 2026-11-05 18:30:59 +01:00
<!-- By submitting this pull request, you acknowledge that you have read the [contributing guide](https://github.com/damontecres/Wholphin/blob/main/CONTRIBUTING.md, including the AI/LLM policy, and [developer's guide](https://github.com/damontecres/Wholphin/blob/main/DEVELOPMENT.md) --> ## Description Person pages that get a filmography from Seerr had three problems, all in the same place. The release date was taken from `firstAirDate` alone — the field TMDB fills for series — so movie cards showed no year at all. The library person page capped the row at 25 cast + 25 crew while `DiscoverPersonPage` had no cap, because each screen carried its own copy of the same logic. And `cast + crew` left the row in no order a viewer can read, with a title appearing twice when a person is both cast and crew on it. This PR: * maps credit dates as `releaseDate ?: firstAirDate` — the expression the search result mapping already uses; * moves the credits logic into a single `SeerrService.personCredits()` used by both screens, so they cannot drift apart again; * drops the 25 + 25 cap, de-duplicates by TMDB id and sorts by release date descending, with undated entries last. Rows on movie and series pages are untouched: only the `BaseItemKind.PERSON` branch of `similar()` changed. Two things worth your opinion: * Without the cap, the row for a person with a long filmography is long — 135 cards for Cameron Diaz, for example. If you would rather keep a cap, the row could keep the newest 25 and gain a `ViewMoreCard` opening the full grid — `Destination.DiscoveredItem` with a `PERSON` item already routes to `DiscoverPersonPage`. I am happy to add that in a second commit. * The sort is fixed to release date descending. It now lives in one function, so a different key or a user-facing choice would be cheap to add later. ### Related issues Fixes #1873 ### Testing CI in my fork ran the repository's own PR workflow on this branch: `pre-commit` (ktlint) and `assembleDefaultDebug` with `testDefaultDebugUnitTest` are both green. I then installed a debug build of the same code on an NVIDIA Shield TV Pro 2019 (Android 11), against Jellyfin 10.11.11 with Jellyseerr, and opened Cameron Diaz from the cast row of *Charlie's Angels* before and after the change: * movie cards show a year now; before, the only two cards in the row with a year were the two TV entries; * the row is ordered by release date descending, movies and series interleaved, instead of the API order of all cast followed by all crew; * the row is no longer cut at 25 + 25 — it now runs back to her 1952 credit, and the entries TMDB has no date for sit at the very end of the row. Not tested on any other device; nothing in the change is device specific. ## Screenshots Person page of Cameron Diaz, reached from the cast row of *Charlie's Angels*. The library rows above the Discover row are the same in every shot and are useful as a reference: they always had their years. Before, start of the Discover row — no year under any card: <img width="1920" height="1080" alt="pr1873-1-before-start" src="https://github.com/user-attachments/assets/585d0670-ae39-4214-9004-0c92262991da" /> Before, end of the Discover row — the row stops here, and the only two years in it belong to the two TV entries: <img width="1920" height="1080" alt="pr1873-2-before-end" src="https://github.com/user-attachments/assets/85c58575-2ad0-4369-a103-60f04c8d2350" /> After, start of the row — years on the movie cards, ordered by date descending: <img width="1920" height="1080" alt="pr1873-3-after-start" src="https://github.com/user-attachments/assets/804745c5-e1a6-4fe4-93bd-dc6c97a7a876" /> After, end of the row — the list now reaches 1952, with the undated entries last: <img width="1920" height="1080" alt="pr1873-4-after-end" src="https://github.com/user-attachments/assets/2c32389c-ec0b-4df5-b9d1-f564ef10d8b3" /> ## AI or LLM usage The code was written with Claude Opus 5 (Claude Code): the two mapping lines, the extracted `personCredits()` function and the two call sites. I reviewed every line and can explain it. This description was written by me in Russian and translated into English with the same assistant. --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> Co-authored-by: Damontecres <damontecres@gmail.com> |
||
|---|---|---|
| .. | ||
| config | ||
| schemas/com.github.damontecres.wholphin.data.AppDatabase | ||
| src | ||
| .gitignore | ||
| build.gradle.kts | ||
| proguard-rules.pro | ||