proton-stream/docs/DEVELOPMENT.md

# Development

## Where things stand

| Layer | State |
|---|---|
| SDK streaming surface (`../proton-sdk-rs` 0.4.0, from crates.io) | **done, live-verified** — `open_revision` / `download_range` / thumbnails / `refresh_session` on the visitor path; node-key cache; chunked `enumerate_nodes`; paged `download_file_to` |
| `pstr-core` — shares, config, catalog, naming, library grouping | **done** — 106 offline tests |
| `pstr-cli` — add / shares / remove / crawl / list / metadata / play / bench | **done** — verified against a real 354-node share: 310 playable files, 22 titles, correct seasons and episodes |
| `pstr-stream` — block cache, read-ahead | **done, live-verified** — 68 offline tests; benchmarked against a real 761 MiB episode (see `docs/TESTING.md`) |
| `pstr-player` — libmpv `stream_cb` + render API | **done, live-verified** — 33 offline tests; plays a real 1080p HEVC episode over a public link, seeks, resumes from the disk cache, and renders into the app's GL context |
| `pstr-app` — egui UI, poster grid, embedded video | **done, offline-tested** — 15 offline tests; library wall, title pages, player page with overlay controls, themes, share management |
| Metadata providers (AniList, then TVmaze / TMDB) | **done** — 32 offline tests; posters, synopses, per-episode titles, hand-pinned matches. Opt-in, off by default |
| Packaging | **done, locally verified** — `scripts/build.sh` (tarball, `.deb`, `.rpm`, `.app`, `.dmg`) and `scripts/build.ps1` (portable `.zip`, WiX `.msi`), plus `packaging/PKGBUILD`. See `packaging/README.md`. The macOS half is written but has not been run on a Mac |

## Build order from here

1. Packaging is in place (`packaging/README.md`); what is left there is running
   the macOS half on a real host, and a release workflow that drives every
   platform.

Done: **`pstr-player` step two** — `mpv_render_context_create` with
`MPV_RENDER_API_TYPE_OPENGL` against eframe's glow context
(`pstr-player/src/render.rs`), painted from a framebuffer this app owns
(`pstr-app/src/video.rs`) so the picture can sit in a rectangle rather than
filling the window. `VideoOutput::Embedded` is what the app asks for; the
standalone-window mode the CLI uses is unchanged, and nothing about it reaches
the streaming path.

Done: `pstr-player` step one — `mpv_stream_cb_add_ro`, mpv in its own window.
`pstr play --file <text>` plays a catalogued episode over a public link.

Done: `pstr-app` step one — `cargo run -p pstr-app` opens a window with the
library wall, a title page per series, a transport bar and share management. It
plays through the same `Player` the CLI uses — at that point still in mpv's own
window; the app's transport drives it from a command channel. Two pieces of that are
worth knowing before touching it:

- **Drawing never mutates.** Pages are handed `&`-state and push `Action`s onto
  a list the app applies after the frame. That is what lets a card click change
  the page it was drawn on.
- **The UI thread never opens a share, reads the catalog or decodes a JPEG.**
  All of it goes through `pstr_app::engine::Engine`, which spawns onto the tokio
  runtime and answers with an `Event` — plus a `ctx.request_repaint()`, without
  which the event sits in the channel until the mouse moves.

Done: player controls beyond play/pause/seek — volume with mute, an audio-track
picker and a subtitle picker, in the transport bar (so on the player page and on
the bar under the library alike), plus `M` and `↑`/`↓` on the player page. All
four preferences persist to `playback.json` (`pstr_core::prefs`), and a track
chosen by hand sets `alang`/`slang` for every file loaded afterwards — the point
being that picking Japanese audio once picks it for the rest of the season.
A dragged volume slider writes to disk on release, not per frame.

Done: **the track choice follows the title, by name as well as language.** A
track picked by hand is stored per title in the catalog (`title_track_prefs`,
schema v8 added `audio_title`/`subtitle_title`) as its language *and* the name
the muxer gave it. On `FileLoaded`, `playback::restore_tracks` re-selects the
track whose name matches (`pstr_player::pick_track`) — which is what tells
"Signs & Songs" from "Full Subtitles" when both are English, and finds an
untagged track at all; `alang`/`slang` could only ever say the language. A name
in the wrong language is not a match, and a title with only a language stored
keeps mpv's own pick. Android stores languages only and keeps a desktop-stored
name while its language is unchanged.

Done: the player page's overlay — a full-width seek bar with the file's chapters
marked on it and the current chapter named beside the clock, a transport cluster
(previous / −10s / play / +30s / next), and volume plus the audio, subtitle and
chapter pickers to the right. Chapters come from mpv's `chapter-list`
(`pstr-player/src/chapters.rs`); a chapter whose name reads as an opening,
ending or preview also puts a **skip** button in the bottom-right corner, which
is drawn whether or not the rest of the controls are up — an opening lasts
ninety seconds and a button you have to wake with the mouse is one nobody uses.
The strip under the library is the same controls at a lower density plus a
**Back to video** button, which is what makes leaving the player page with Esc
reversible.

Done: **next / previous episode and autoplay.** The neighbours come from
`Title::following`/`preceding` in `pstr-core` — display order, so it walks off
the end of a season into the next one — and autoplay fires only on a clean end
of file, never on a failure. `P`/`N` on the player page, and the page stays put
while the next file opens rather than dropping back to the library.

Done: **what a chapter is, and the "up next" countdown.** Chapter names are read
in two passes (`pstr-player/src/chapters.rs`). The first reads the name alone —
`OP`, `ED`, `Next Episode`, and the Japanese forms — and yields a *claim*; the
second (`roles`) settles the claims that a name cannot settle, using the rest of
the file. `Intro` is the opening only when the file has no chapter that says so
outright, the chapter is the length of a theme song and it is near the front:
episode one of Oshi no Ko opens with eleven minutes called `Intro` that are the
story. `Credits` / `Cast` are the ending at about that length or in the last
quarter. Both ambiguities fail *towards content*, because not offering a skip
costs a button and skipping the story costs the story.

`credits_start` then takes the run of chapters that **ends** the file — walking
back from the last one and stopping at the first that is content. That is the
part worth stating: the ending is not the tail if something follows it.
Fullmetal Alchemist Brotherhood episode 46 has a Part C after the ED and the
preview after that, so only the preview is skippable there, while the ordinary
`… ED · Preview` gives up both. Reaching that run starts a ten-second **Up next**
card (`App::tick_up_next`, `ui::player::up_next_card`) with *Play now* and
*Watch till the end*; the latter holds for the rest of the file. It needs
chapters, autoplay on, a next episode, and playback actually running — a paused
episode is one somebody walked away from.

Done: **per-episode metadata.** `Provider::enrich` fetches the episode list —
and, for AniList, the series' fanart — of a title that already matched, and
`Catalog::set_enrichment` stores both and records that it asked (schema **v9**),
so an empty answer is not re-asked every run. TMDB is `/tv/{id}` plus every
season appended onto one more request. AniList walks the match's `SEQUEL`
relations and takes each entry's episodes from ani.zip, which carries TVDB's
season, number and absolute number — the numbering release groups use. The
reconciling between a release's numbering and a provider's lives in one tested
place, `EpisodeGuide::get`. [METADATA.md](METADATA.md) has the audit that drove
this, the order `get` tries things in, and what comes next.

The first version searched AniList for `<title> 2nd Season` per season and took
episode names from `streamingEpisodes`; both were wrong often enough to show —
the search found the base entry again and captioned Attack on Titan's later
seasons with season one's names, and `streamingEpisodes` is whatever a streaming
site published. v9 deletes everything they stored.

Done: **matching a title by hand.** `MATCH_FLOOR` is set where a wrong poster
costs more than none, and the price of that is a few titles it refuses to decide
about — `Fate/stay night [Heaven's Feel]` is the standing example, three films in
one folder against a provider that files each of them as its own entry, so there
is no single answer for the scorer to find. *Change match* on the title page opens
`ui::matcher`: the provider's own answers for whatever is typed, unscored and in
its own order (`MetadataService::search`), and a click pins one. Nothing is
ranked there on purpose — the floor exists to stop the *scorer* guessing, and a
person reading eight entries with their posters is not guessing.

A chosen row is stored `manual` (schema **v5**), which means two things:
`MetadataRecord::is_fresh` never expires it, and `run_match` skips it even when
forced — "match again" must not undo the one title someone fixed by hand. The
way back is *Clear*, which drops the row entirely (`Catalog::forget_metadata`)
rather than storing a miss, so the next run treats the title as one it has never
asked about. Switching provider still clears everything, hand-picked rows
included: a `remote_id` means nothing to the other provider.

Worth knowing before chasing a missing name: ani.zip only lists what TVDB has
placed. A file it cannot answer is usually a special (`S02E00`), an episode TVDB
skipped, or one too new to be listed — Attack on Titan's last two files and two
of Bleach's forty-four are the standing examples. `pstr metadata match` prints
how many episodes each title got.

Note the glyph trap: egui's bundled fonts cover the transport icons
(`⏮ ⏭ ⏸ ▶ 🔊 🔇 ⛶`) and *not* `← 💬 ▣`, which draw as an empty box. Anything new
on a button wants checking on screen before it ships.

**Sidecar subtitle files.** A `.srt`, `.ass`, `.ssa` or `.vtt` beside a video
is paired with it by name at crawl time (`catalog::build_sidecars`): the video's
name followed by a dot or nothing (`S01E01.en.forced.ass`), the same inside a
`Subs` folder beside the videos, the RARBG layout `Subs/<video>/2_English.srt`,
and failing those the only video in the folder. The words past the video's
name give the language and the picker's title. They are stored in `sidecars`
(schema v12), rebuilt by every crawl like `nodes`.

Each client opens them while mpv opens the video and adds them with `sub-add
pstr://<token>` through the same registry once mpv reports the file loaded. One
is turned on only when subtitles are on and the file has none of its own
playing, and only in the preferred language (`catalog::preferred_sidecar`); the
rest wait in the subtitle menu. A downloaded episode played without a session
goes without them, and they are not downloaded with it.

## Design decisions already taken

- **Fat client, no server.** Each viewer's app talks to Proton directly.
- **Public links first**, several, merged into one catalog. A Proton account
  is optional: it adds folders of the viewer's own Drive as shares, and watch
  history synced through a `proton-stream` device in that Drive — one file per
  installation, so no two devices ever write the same file.
- **Embedded libmpv**, not a bundled decoder stack. HEVC 10-bit, HDR and ASS
  subtitles with hardware decode, for free.
- **No transcoding.** mpv demuxes and decodes natively; there is no remux or
  re-encode step and no ffmpeg dependency of our own.
- **`stream_cb`, not a loopback HTTP server.** No port to open, no local attack
  surface, no plaintext leaving the process.
- **Metadata enrichment is opt-in and off by default**, with the privacy cost
  stated in the UI: enabling it sends your library's titles to a third party.
  With it off, posters fall back to Proton thumbnails.
- **WASM is out.** Not a small port and not worth reopening: the SDK pulls tokio
  with `rt-multi-thread` + `fs`, has four `spawn_blocking` sites, resolves two
  `getrandom` versions with no wasm backend, and drags `aws-lc-rs`/`ring` through
  rustls. Separately a browser cannot reach the Proton API at all (CORS) without
  a proxy server, which the fat-client decision rules out.

- **One palette, one type ramp, one shape radius — named, not repeated.**
  `pstr_core::appearance` already settled colour for both clients;
  `pstr_app::theme::Role` settles size the same way. A `.size(17.0)` at a call
  site is not a scale, it is one of forty independent decisions that happen to
  sit near each other, and it is why the two clients read as different apps even
  when their colours agreed. Anything new asks for a role.
- **Depth and motion are the theme's, not the widget's.** Shadows come from
  `theme::tile_shadow` / `theme::bar_shadow`, which halve themselves on a light
  flavour — the same black that separates a card from a dark page is a smudge on
  a near-white one. Transitions go through `Context::animate_*`, which is driven
  by wall-clock time and keeps no state of its own, so a hover that fades needs
  no field anywhere.
- **The grid flexes; `CARD_WIDTH` is a minimum.** `ui::columns` hands back the
  column count *and* the width they should be drawn at. Dividing by the constant
  and flooring leaves up to a full card of dead space at the right of every row.
- **Android copies the ramp; a test says so.** `ui/theme/Type.kt` and
  `ui/theme/Shape.kt` are `Role` and radius 8 restated for Compose, because
  Material's own defaults are a different scale (57sp displays, a five-step
  corner ramp ending in a fully-round button) and there is no bridge to resolve
  type the way `appearance` resolves colour. `TypeRampTest` names all fifteen
  Material styles and fails on a sixteenth, so the copy cannot drift quietly.
  Note that Material's filled buttons read their own shape token rather than
  `MaterialTheme.shapes` — that is what `AccentButton`/`TonalButton`/
  `EdgedButton`/`QuietButton` in `ui/theme/Accent.kt` exist for, and new call
  sites should use them rather than Material's directly.
- **The gradient toggle is not read where the gradient is drawn.**
  `Palette::resolve` sets `accent_alt` equal to `accent` when gradients are off,
  so a two-stop brush collapses to a flat fill on its own. Both clients rely on
  that instead of branching on the flag, which is what keeps them from
  disagreeing about what the stored setting means (see
  [B52](BUGS.md#b52--paint-the-accent-as-a-gradient-did-nothing-on-android)).

## Traps

- **Block sizes are not uniform.** Always `RevisionReader::block_sizes()`; never
  a hardcoded 4 MiB. A padded vector shifts every later block's start and serves
  full-length reads of the wrong bytes, silently.
- **Cache validity is the revision id**, not `(mtime, size)`. It advances if and
  only if a new revision was sealed.
- **Read-ahead must yield to seeks, and must not outrun the ring.** Both were
  measured, both cost more than they bought when got wrong: prefetches left
  running from the position the viewer left took worst-case seek latency from
  572 ms to 2895 ms, and a window deeper than half the ring evicts its own blocks
  before the player reaches them. `pstr-stream::stream` cancels on seek and
  clamps the depth; the numbers are in its module docs.
- **Read-ahead depth is a property of the *consumer*, not of the block layer.**
  mpv brings its own, so under mpv the block layer's read-ahead should be **0**:
  at depth 12 the first frame took 5.5 s instead of 2.5 s and a seek 3.6 s
  instead of 1.5 s. `pstr_player::READAHEAD_BLOCKS` is what a player-driven
  caller passes; `pstr_stream::DEFAULT_READAHEAD_BLOCKS` is for readers that have
  none of their own. Do not "unify" them. See `docs/BUGS.md` B8.
- **mpv's buffer is in seconds, not bytes.** A byte budget means two minutes of
  buffer on a 4 Mbit/s episode and eight seconds on a 4K remux. The byte figure
  survives only as a ceiling, and it must stay under the ring or B7 recurs a
  layer up.
- **The anonymous session expires mid-film.** The SDK renews it in place and
  readers survive; the app should still be able to reopen and resume at the same
  byte offset as a belt-and-braces fallback.
- **A share filled by the FUSE client has no thumbnails at all.** Proton renders
  a thumbnail at upload time and the SDK does not generate one, so every
  `download_thumbnail` on this library answers `Ok(None)` and every tile falls
  back to initials. Nothing is broken — it is why metadata providers matter more
  than they looked like they did, and why an empty `thumbs/` cache directory is
  not evidence the thumbnail path is wrong.
- **Negative caching matters.** Without it, every grid render pays a round-trip
  per unmatched title. `proton-drive-linux`'s photo grid learned this the hard
  way. `ThumbnailCache` remembers both the textures it has *and* the files that
  turned out to have no thumbnail.
- **A dragged seek bar must seek on release, not per frame.** Every seek cancels
  outstanding read-ahead and refetches; a bar that seeks while dragging spends
  the whole drag throwing away blocks it just paid for. `ui::transport` acts on
  `clicked() || drag_stopped()`.
- **egui 0.35 moved the panels.** `eframe::App` is `fn ui(&mut self, ui, frame)`,
  not `update(ctx, frame)`; panels are `egui::Panel::top(id).show(ui, …)` rather
  than `TopBottomPanel::top(id).show(ctx, …)`, and `Context::style_mut` is now
  `all_styles_mut`. Older egui snippets will not compile as written.