yt-dlp pulls from many sites, so a fixed "YouTube" label was wrong for
bandcamp, soundcloud, etc. Derive the provider name from the source page's
host instead (www.youtube.com -> YouTube, *.bandcamp.com -> Bandcamp),
falling back to the bare host for unmapped sites. Other origins keep their
fixed PROVIDER_NAMES labels.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The web player decides when a track counts as listened (caught near its
start and heard to ~90%, capped at 4 min) and triggers POST /scrobble.
The token stays server-side (RADIEO_LISTENBRAINZ_TOKEN), submitting the
listen with the canonical MusicBrainz MBID when available. Each airing is
deduplicated so multiple tabs submit it once.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bandcamp label/artist pages list releases newest-first, so give the first
YTDLP_RECENT_COUNT entries a YTDLP_RECENT_BOOST multiplier when picking,
leaning the radio towards fresh music. Flat extraction carries no dates, so
list position is used as the recency proxy rather than a real date lookup.
The boost is gated on the source URL being a bandcamp discography listing:
single /album/ and /track/ pages (whether a URL-file line or one the pick
recursed into) list tracks in track order, not by recency, and non-bandcamp
sources have unverified ordering, so all of those keep a uniform pick.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
When every candidate is within the anti-repeat window, the fallback replayed
at random, ignoring how long ago each was heard. With a small feed (or a
window larger than a source's pool) this is the *normal* path, and random
picking clusters the same tracks together.
Play the least-recently-heard candidate instead, so tracks rotate at the
widest spacing the pool allows.
- db: add last_played_at(keys) -> key -> most-recent play timestamp.
- providers/listenbrainz: sort the exhausted pool oldest-first.
- scheduler: on exhaustion, return the oldest-played of the drawn candidates
rather than the last one drawn.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
RADIEO_SUBSONIC_PLAYLIST now accepts a comma-separated list of playlist
names or ids, each optionally weighted with a '=<number>' suffix (e.g.
'Chill=3, Focus, Party=2'); default weight is 1 and 0 disables an entry.
The provider caches each playlist independently and walks them in
weighted-random order on every pick, falling through past empty, renamed
or unreachable playlists so the source never stalls. Picking the
playlist first (by weight) then a song uniformly gives each its
configured share regardless of length.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Source lines may now start with a numeric weight (e.g. 3:https://…) to
bias how often each is drawn. The prefix is unambiguous since a URL
scheme can never be numeric; unprefixed lines default to weight 1 and a
weight of 0 disables the line. Selection uses Efraimidis-Spirakis
weighted sampling without replacement, preserving the fall-through to
the next source when one fails.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Extract the HTTP surface out of radio.liq into two included files: web.liq
(static assets, PWA, local player API) and ingest_proxy.liq (relays to the
ingest daemon). radio.liq keeps only the streaming pipeline and ends with the
%include directives, evaluated after the pipeline so the handlers see radio,
now_playing, history, etc.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Wire up a web manifest, service worker and icon routes so the player can
be installed on mobile. The static manifest stays generic; the page
regenerates it at runtime from STATION_NAME so an instance keeps its name.
The service worker only caches the app shell, never the live stream or the
playback APIs. Icons (192/512, maskable and apple-touch) are rasterized
from the favicon.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add an input in the queue tab to enqueue a yt-dlp URL: a single track, or
a whole playlist/album. Requests are a priority lane in the ingest queue —
pop_next serves them before the auto radio, so the next /next plays the
request without cutting the current track. They download lazily (a few
ahead), so a large playlist queues instantly and bypasses anti-repeat.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The player's "source" link only worked for direct yt-dlp URLs. Two other
cases had no linkable page: ListenBrainz picks resolved via ytsearch1: (the
locator is a search query) and Subsonic library tracks (an opaque song id).
Centralise the rule in Track.page_url and cover both: the yt-dlp fetcher now
records the concrete video URL it resolved into source_url, and a Subsonic
track links to the stream's new /share endpoint, which asks ingest to mint a
public share (createShare) on demand and redirects to it — so a share is only
created when a listener actually clicks, never per played track.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Wire the Media Session previous-track control to a new POST /restart-track
route that requeues the current song from the start for every listener, and
keep next-track as skip. Exposing both handlers also makes Android (Chrome)
show the skip button, which it hides when only nexttrack is set.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Wire the web player into system media controls (MPRIS, Control Center,
lock screen) and keyboard media keys: push track metadata and handle
play/pause/next, with seek neutralized on the live stream.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The provider only uses standard OpenSubsonic endpoints (getPlaylists,
getPlaylist, search3, stream), so it works with any compatible server
(Navidrome, Gonic, Airsonic…), not just Navidrome.
BREAKING: environment variables are renamed
RADIEO_NAVIDROME_URL/USER/PASSWORD/PLAYLIST -> RADIEO_SUBSONIC_*
RADIEO_WEIGHT_NAVIDROME -> RADIEO_WEIGHT_SUBSONIC
Update your .env accordingly.
Internally the source key and provider are renamed navidrome -> subsonic,
aligning with the existing 'subsonic' backend and fetcher.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A yt-dlp source pointing at a label/artist page flat-extracts to a mix of
/track/ and /album/ URLs. The provider used each verbatim as a locator, so an
/album/ URL was handed to the fetcher as if it were a track: yt-dlp then
(mis)downloaded the whole album into one file and tagged it from the
playlist-level info, which carries the album title and no artist — surfacing as
"Unknown artist" on the stream.
Drill picked container entries down to a single track before emitting a
locator, bounded by a small depth so nested containers (label -> album ->
track) resolve while a real track (which flat-extracts to just itself) is the
base case. Locators are now always downloadable tracks, so the existing
tag/fetch path and anti-repeat keying work as intended.
Also make guess_metadata trust the explicit artist tag over the "Artist -
Title" title split: some label uploads double the artist into the title
("Artist - Artist - Title"), which the blind last-" - " split mis-parsed. When
that artist prefixes the title we peel it off (repeatedly), falling back to the
split only when there is no artist tag.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds special jingle folders (midi, gouter, bisous) that take priority
over the default jingle rotation once per day, briefly after 11h00,
15h00, 16h30 and 21h00.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
For yt-dlp tracks the locator is the original web page, so pass it as a
url annotation, carry it through the stream metadata and history, and
turn the track title into a link back to that page (both live and in the
history). Subsonic ids are opaque and stay plain text.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Default it to "Nemu FM" and derive the tab title, logo, and dynamic
document title from one STATION_NAME constant for easy renaming.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
If the stream drops (server restart, network loss), the player now retries
goLive() on error/ended/stalled with a delay that doubles each failure, capped
at 30s, and resets once audio flows again.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
yt-dlp's flat extraction mis-attributes bandcamp uploads: on a label
account the artist becomes the label, and many files ship untagged. After
a download, guess (artist, title, album) from the richer download metadata
and the "Artist - Title" filename shape, confirm against MusicBrainz, and
write ID3 tags into the file. Existing tags are respected; MusicBrainz is
primary, filename parsing only a fallback.
The file's tags are the single source of truth: the provider reads them
back on the next pick, so the corrected identity (incl. MBID) is in place
before the scheduler keys and anti-repeat-checks the track, and cache
reuse inherits it too.
Fetchers now return (path, Track) so the corrected metadata flows back.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
/download now accepts ?file=<name> to fetch any cached file, not just the
current track. History entries carry that filename token (via /history), so
the web UI renders each aired track as a download link. A shared
serve_attachment helper validates the request (basename-only, real audio file,
no hidden/.part files) before streaming it; LRU-evicted tracks return 404.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
On the pinned savonet/liquidsoap:v2.4.5 image, `source.on_track` no longer
exists as a module function — on_track is a method on the source value. The
old `music = source.on_track(music, …)` form fails type-checking, so the whole
radio.liq is rejected and the stream container never starts. Register the
handler in place via the method form instead.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a read-only field with the bare /radio.mp3 URL and a copy button
(clipboard API, with a select+execCommand fallback), so the stream can
easily be opened in VLC or another external player.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Firefox kept resuming the <audio> element from a stale buffer after a
pause, drifting behind the live point. Load the stream with an anti-cache
query parameter, and on resume-from-pause reconnect to live rather than
replaying the buffered audio.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The fallback played the whole /cache directory, which at cold start holds
only the 2-3 tracks being pre-fetched — so it looped them until the
request.dynamic buffer filled. Restrict the fallback to tracks already
aired: the ingest daemon exposes them at GET /fallback.m3u (played_at set,
still on disk), and the stream fetches that into a local /tmp/fallback.m3u
that playlist watches. Cold start is now silent (assumed) instead of a tight
loop, and a mid-stream drain degrades across the whole listening history.
A local file (not a remote playlist URL) is used to avoid Liquidsoap's http
resolver mis-sniffing the response as text/html; mime_type is forced so an
empty header-only m3u still parses.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add Drone pipelines building both container images per architecture and
merging them into multi-arch manifests under registry.nemunai.re/radieo.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Give the outgoing HTTP clients (Navidrome, MusicBrainz, ListenBrainz) a
transport-level retry budget (RADIEO_HTTP_RETRIES, default 2) so a brief
connection blip on a remote endpoint no longer drops a fetch or a lookup.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a Docker healthcheck probing ingest's /healthz (via python, since the slim
image has neither curl nor wget) and make the stream start only once ingest is
service_healthy, so it never briefly falls back on an empty cache at boot.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Handle SIGTERM (and SIGINT) to stop the HTTP server cleanly instead of being
killed after Docker's grace period, so `docker compose down` returns in ~1s and
the queue, HTTP clients and database are closed properly. shutdown() runs on a
helper thread since it must not run on the serving thread.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Serve a small web page at http://<host>:8000/ (stream/index.html) from the
Liquidsoap harbor, alongside the /radio.mp3 stream. It shows the track
currently on air and refreshes it from a /nowplaying JSON endpoint, fed by the
broadcast source's live metadata — accurate even though the ingest daemon runs
a track ahead (prefetch).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The fallback playlist probed every file in /cache, including .gitkeep and
in-progress .part downloads, logging ffmpeg "Invalid data" warnings at startup.
Filter candidates with check_next to keep only real audio files and skip hidden
ones, removing the noise.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a third playback source: a ListenBrainz recommendations Atom feed. Each
suggestion already carries a MusicBrainz recording MBID, title and artist, so
it is keyed directly by MBID (source-agnostic identity, no extra lookup) and
resolved to a concrete file — Navidrome search3 first, then a yt-dlp
ytsearch1: fallback.
- providers/listenbrainz.py: parse the Atom/HTML feed, anti-repeat on the MBID
key, resolve Navidrome-then-yt-dlp. Feed may be an http(s) URL or a local
path (for testing).
- subsonic.py: add search_songs (search3) for resolution.
- canonicalizer.py: short-circuit when a Track already has an MBID, so
feed-provided MBIDs are trusted and MusicBrainz is not hit.
- __main__.py: wire the provider in; register the yt-dlp fetcher as a
resolution backend even when the yt-dlp source is off; close providers on
shutdown.
- config/compose/.env.example: RADIEO_LISTENBRAINZ_URL + weight.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Give tracks a source-agnostic identity so the same song from different
sources no longer replays in a loop.
- Canonicalizer resolves (artist, title) to a MusicBrainz recording MBID
(no API key; ~1 req/s, descriptive User-Agent, best-effort). Hits and
confirmed misses are cached in SQLite; transient errors are not.
- Track.key becomes mbid:<id> when resolved, else a normalized
name:<artist>|<title> fallback — still source-agnostic.
- Scheduler now owns the authoritative anti-repeat on the canonical key,
canonicalizing the drawn track with a bounded retry; providers keep a
cheap recent-locator filter to limit retries.
- db: canonical_cache table, history.locator column with migration for
existing databases, recent_locators().
- Canonicalization can be turned off via RADIEO_CANONICAL_ENABLED=0.
Verified: MBID hit/cache/miss, cross-source key collapse, scheduler
dodging a recent play, schema migration, and full stack (Navidrome +
yt-dlp) with zero Python tracebacks and a valid 192 kbps MP3 stream.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
