navidrome/core/external_metadata.go
Deluan Quintão c795bcfcf7
feat(bfr): Big Refactor: new scanner, lots of new fields and tags, improvements and DB schema changes (#2709)
* fix(server): more race conditions when updating artist/album from external sources

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(scanner): add .gitignore syntax to .ndignore. Resolves #1394

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(ui): null

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(scanner): pass configfile option to child process

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(scanner): resume interrupted fullScans

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(scanner): remove old scanner code

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(scanner): rename old metadata package

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(scanner): move old metadata package

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: tests

Signed-off-by: Deluan <deluan@navidrome.org>

* chore(deps): update Go to 1.23.4

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: logs

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(test):

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: log level

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: remove log message

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: add config for scanner watcher

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: children playlists

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: replace `interface{}` with `any`

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: smart playlists with genres

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: allow any tags in smart playlists

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: artist names in playlists

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: smart playlist's sort by tags

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add moods to child

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add moods to AlbumID3

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor(subsonic): use generic JSONArray for OS arrays

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor(subsonic): use https in test

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add releaseTypes to AlbumID3

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add recordLabels to AlbumID3

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor(subsonic): rename JSONArray to Array

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add artists to AlbumID3

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add artists to Child

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(scanner): do not pre-populate smart playlists

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): implement a simplified version of ArtistID3.

See https://github.com/opensubsonic/open-subsonic-api/discussions/120

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add artists to album child

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add contributors to mediafile Child

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add albumArtists to mediafile Child

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add displayArtist and displayAlbumArtist

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add displayComposer to Child

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add roles to ArtistID3

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(subsonic): use " • " separator for displayComposer

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor:

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(subsonic):

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(subsonic): respect `PreferSortTags` config option

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor(subsonic):

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: optimize purging non-unused tags

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: don't run 'refresh artist stats' concurrently with other transactions

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor:

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: log message

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: add Scanner.ScanOnStartup config option, default true

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: better json parsing error msg when importing NSPs

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: don't update album's imported_time when updating external_metadata

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: handle interrupted scans and full scans after migrations

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: run `analyze` when migration requires a full rescan

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: run `PRAGMA optimize` at the end of the scan

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: don't update artist's updated_at when updating external_metadata

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: handle multiple artists and roles in smart playlists

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(ui): dim missing tracks

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: album missing logic

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: error encoding in gob

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: separate warnings from errors

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: mark albums as missing if they were contained in a deleted folder

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: add participant names to media_file and album tables

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: use participations in criteria, instead of m2m relationship

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: rename participations to participants

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add moods to album child

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: albumartist role case

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(scanner): run scanner as an external process by default

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(ui): show albumArtist names

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(ui): dim out missing albums

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: flaky test

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(server): scrobble buffer mapping. fix #3583

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: more participations renaming

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: listenbrainz scrobbling

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: send release_group_mbid to listenbrainz

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): implement OpenSubsonic explicitStatus field (#3597)

* feat: implement OpenSubsonic explicitStatus field

* fix(subsonic): fix failing snapshot tests

* refactor: create helper for setting explicitStatus

* fix: store smaller values for explicit-status on database

* test: ToAlbum explicitStatus

* refactor: rename explicitStatus helper function

---------

Co-authored-by: Deluan Quintão <deluan@navidrome.org>

* fix: handle album and track tags in the DB based on the mappings.yaml file

Signed-off-by: Deluan <deluan@navidrome.org>

* save similar artists as JSONB

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: getAlbumList byGenre

Signed-off-by: Deluan <deluan@navidrome.org>

* detect changes in PID configuration

Signed-off-by: Deluan <deluan@navidrome.org>

* set default album PID to legacy_pid

Signed-off-by: Deluan <deluan@navidrome.org>

* fix tests

Signed-off-by: Deluan <deluan@navidrome.org>

* fix SIGSEGV

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: don't lose album stars/ratings when migrating

Signed-off-by: Deluan <deluan@navidrome.org>

* store full PID conf in properties

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: keep album annotations when changing PID.Album config

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: reassign album annotations

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: use (display) albumArtist and add links to each artist

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: not showing albums by albumartist

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: error msgs

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: hide PID from Native API

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: album cover art resolution

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: trim participant names

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: reduce watcher log spam

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: panic when initializing the watcher

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: various artists

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: don't store empty lyrics in the DB

Signed-off-by: Deluan <deluan@navidrome.org>

* remove unused methods

Signed-off-by: Deluan <deluan@navidrome.org>

* drop full_text indexes, as they are not being used by SQLite

Signed-off-by: Deluan <deluan@navidrome.org>

* keep album created_at when upgrading

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(ui): null pointer

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: album artwork cache

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: don't expose missing files in Subsonic API

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: searchable interface

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: filter out missing items from subsonic search

* fix: filter out missing items from playlists

* fix: filter out missing items from shares

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(ui): add filter by artist role

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): only return albumartists in getIndexes and getArtists endpoints

Signed-off-by: Deluan <deluan@navidrome.org>

* sort roles alphabetically

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: artist playcounts

Signed-off-by: Deluan <deluan@navidrome.org>

* change default Album PID conf

Signed-off-by: Deluan <deluan@navidrome.org>

* fix albumartist link when it does not match any albumartists values

Signed-off-by: Deluan <deluan@navidrome.org>

* fix `Ignoring filter not whitelisted` (role) message

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: trim any names/titles being imported

Signed-off-by: Deluan <deluan@navidrome.org>

* remove unused genre code

Signed-off-by: Deluan <deluan@navidrome.org>

* serialize calls to Last.fm's getArtist

Signed-off-by: Deluan <deluan@navidrome.org>

xxx

Signed-off-by: Deluan <deluan@navidrome.org>

* add counters to genres

Signed-off-by: Deluan <deluan@navidrome.org>

* nit: fix migration `notice` message

Signed-off-by: Deluan <deluan@navidrome.org>

* optimize similar artists query

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: last.fm.getInfo when mbid does not exist

Signed-off-by: Deluan <deluan@navidrome.org>

* ui only show missing items for admins

Signed-off-by: Deluan <deluan@navidrome.org>

* don't allow interaction with missing items

Signed-off-by: Deluan <deluan@navidrome.org>

* Add Missing Files view (WIP)

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: merged tag_counts into tag table

Signed-off-by: Deluan <deluan@navidrome.org>

* add option to completely disable automatic scanner

Signed-off-by: Deluan <deluan@navidrome.org>

* add delete missing files functionality

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: playlists not showing for regular users

Signed-off-by: Deluan <deluan@navidrome.org>

* reduce updateLastAccess frequency to once every minute

Signed-off-by: Deluan <deluan@navidrome.org>

* reduce update player frequency to once every minute

Signed-off-by: Deluan <deluan@navidrome.org>

* add timeout when updating player

Signed-off-by: Deluan <deluan@navidrome.org>

* remove dead code

Signed-off-by: Deluan <deluan@navidrome.org>

* fix duplicated roles in stats

Signed-off-by: Deluan <deluan@navidrome.org>

* add `; ` to artist splitters

Signed-off-by: Deluan <deluan@navidrome.org>

* fix stats query

Signed-off-by: Deluan <deluan@navidrome.org>

* more logs

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: support legacy clients (DSub) by removing OpenSubsonic extra fields - WIP

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: support legacy clients (DSub) by removing OpenSubsonic extra fields - WIP

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: support legacy clients (DSub) by removing OpenSubsonic extra fields - WIP

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: support legacy clients (DSub) by removing OpenSubsonic extra fields - WIP

Signed-off-by: Deluan <deluan@navidrome.org>

* add record label filter

Signed-off-by: Deluan <deluan@navidrome.org>

* add release type filter

Signed-off-by: Deluan <deluan@navidrome.org>

* fix purgeUnused tags

Signed-off-by: Deluan <deluan@navidrome.org>

* add grouping filter to albums

Signed-off-by: Deluan <deluan@navidrome.org>

* allow any album tags to be used in as filters in the API

Signed-off-by: Deluan <deluan@navidrome.org>

* remove empty tags from album info

Signed-off-by: Deluan <deluan@navidrome.org>

* comments in the migration

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: Cannot read properties of undefined

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: listenbrainz scrobbling (#3640)

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: remove duplicated tag values

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: don't ignore the taglib folder!

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: show track subtitle tag

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: show artists stats based on selected role

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: inspect

Signed-off-by: Deluan <deluan@navidrome.org>

* add media type to album info/filters

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: change format of subtitle in the UI

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: subtitle in Subsonic API and search

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: subtitle in UI's player

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: split strings should be case-insensitive

Signed-off-by: Deluan <deluan@navidrome.org>

* disable ScanSchedule

Signed-off-by: Deluan <deluan@navidrome.org>

* increase default sessiontimeout

Signed-off-by: Deluan <deluan@navidrome.org>

* add sqlite command line tool to docker image

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: resources override

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: album PID conf

Signed-off-by: Deluan <deluan@navidrome.org>

* change migration to mark current artists as albumArtists

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(ui): Allow filtering on multiple genres (#3679)

* feat(ui): Allow filtering on multiple genres

Signed-off-by: Henrik Nordvik <henrikno@gmail.com>
Signed-off-by: Deluan <deluan@navidrome.org>

* add multi-genre filter in Album list

Signed-off-by: Deluan <deluan@navidrome.org>

---------

Signed-off-by: Henrik Nordvik <henrikno@gmail.com>
Signed-off-by: Deluan <deluan@navidrome.org>
Co-authored-by: Henrik Nordvik <henrikno@gmail.com>

* add more multi-valued tag filters to Album and Song views

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(ui): unselect missing files after removing

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(ui): song filter

Signed-off-by: Deluan <deluan@navidrome.org>

* fix sharing tracks. fix #3687

Signed-off-by: Deluan <deluan@navidrome.org>

* use rowids when using search for sync (ex: Symfonium)

Signed-off-by: Deluan <deluan@navidrome.org>

* fix "Report Real Paths" option for subsonic clients

Signed-off-by: Deluan <deluan@navidrome.org>

* fix "Report Real Paths" option for subsonic clients for search

Signed-off-by: Deluan <deluan@navidrome.org>

* add libraryPath to Native API /songs endpoint

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(subsonic): add album version

Signed-off-by: Deluan <deluan@navidrome.org>

* made all tags lowercase as they are case-insensitive anyways.

Signed-off-by: Deluan <deluan@navidrome.org>

* feat(ui): Show full paths, extended properties for album/song (#3691)

* feat(ui): Show full paths, extended properties for album/song

- uses library path + os separator + path
- show participants (album/song) and tags (song)
- make album/participant clickable in show info

* add source to path

* fix pathSeparator in UI

Signed-off-by: Deluan <deluan@navidrome.org>

* fix local artist artwork (#3695)

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: parse vorbis performers

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: clean function into smaller functions

Signed-off-by: Deluan <deluan@navidrome.org>

* fix translations for en and pt

Signed-off-by: Deluan <deluan@navidrome.org>

* add trace log to show annotations reassignment

Signed-off-by: Deluan <deluan@navidrome.org>

* add trace log to show annotations reassignment

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: allow performers without instrument/subrole

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: metadata clean function again

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: optimize split function

Signed-off-by: Deluan <deluan@navidrome.org>

* refactor: split function is now a method of TagConf

Signed-off-by: Deluan <deluan@navidrome.org>

* fix: humanize Artist total size

Signed-off-by: Deluan <deluan@navidrome.org>

* add album version to album details

Signed-off-by: Deluan <deluan@navidrome.org>

* don't display album-level tags in SongInfo

Signed-off-by: Deluan <deluan@navidrome.org>

* fix genre clicking in Album Page

Signed-off-by: Deluan <deluan@navidrome.org>

* don't use mbids in Last.fm api calls.

From 1337574018:

With MBID:
```
GET https://ws.audioscrobbler.com/2.0/?api_key=XXXX&artist=Van+Morrison&format=json&lang=en&mbid=a41ac10f-0a56-4672-9161-b83f9b223559&method=artist.getInfo

{
artist: {
name: "Bee Gees",
mbid: "bf0f7e29-dfe1-416c-b5c6-f9ebc19ea810",
url: "https://www.last.fm/music/Bee+Gees",
}
```

Without MBID:
```
GET https://ws.audioscrobbler.com/2.0/?api_key=XXXX&artist=Van+Morrison&format=json&lang=en&method=artist.getInfo

{
artist: {
name: "Van Morrison",
mbid: "a41ac10f-0a56-4672-9161-b83f9b223559",
url: "https://www.last.fm/music/Van+Morrison",
}
```

Signed-off-by: Deluan <deluan@navidrome.org>

* better logging for when the artist folder is not found

Signed-off-by: Deluan <deluan@navidrome.org>

* fix various issues with artist image resolution

Signed-off-by: Deluan <deluan@navidrome.org>

* hide "Additional Tags" header if there are none.

Signed-off-by: Deluan <deluan@navidrome.org>

* simplify tag rendering

Signed-off-by: Deluan <deluan@navidrome.org>

* enhance logging for artist folder detection

Signed-off-by: Deluan <deluan@navidrome.org>

* make folderID consistent for relative and absolute folderPaths

Signed-off-by: Deluan <deluan@navidrome.org>

* handle more folder paths scenarios

Signed-off-by: Deluan <deluan@navidrome.org>

* filter out other roles when SubsonicArtistParticipations = true

Signed-off-by: Deluan <deluan@navidrome.org>

* fix "Cannot read properties of undefined"

Signed-off-by: Deluan <deluan@navidrome.org>

* fix lyrics and comments being truncated (#3701)

* fix lyrics and comments being truncated

* specifically test for lyrics and comment length

* reorder assertions

Signed-off-by: Deluan <deluan@navidrome.org>

---------

Signed-off-by: Deluan <deluan@navidrome.org>
Co-authored-by: Deluan <deluan@navidrome.org>

* fix(server): Expose library_path for playlist (#3705)

Allows showing absolute path for UI, and makes "report real path" work for playlists (Subsonic)

* fix BFR on Windows (#3704)

* fix potential reflected cross-site scripting vulnerability

Signed-off-by: Deluan <deluan@navidrome.org>

* hack to make it work on Windows

* ignore windows executables

* try fixing the pipeline

Signed-off-by: Deluan <deluan@navidrome.org>

* allow MusicFolder in other drives

* move windows local drive logic to local storage implementation

---------

Signed-off-by: Deluan <deluan@navidrome.org>

* increase pagination sizes for missing files

Signed-off-by: Deluan <deluan@navidrome.org>

* reduce level of "already scanning" watcher log message

Signed-off-by: Deluan <deluan@navidrome.org>

* only count folders with audio files in it

See https://github.com/navidrome/navidrome/discussions/3676#discussioncomment-11990930

Signed-off-by: Deluan <deluan@navidrome.org>

* add album version and catalog number to search

Signed-off-by: Deluan <deluan@navidrome.org>

* add `organization` alias for `recordlabel`

Signed-off-by: Deluan <deluan@navidrome.org>

* remove mbid from Last.fm agent

Signed-off-by: Deluan <deluan@navidrome.org>

* feat: support inspect in ui (#3726)

* inspect in ui

* address round 1

* add catalogNum to AlbumInfo

Signed-off-by: Deluan <deluan@navidrome.org>

* remove dependency on metadata_old (deprecated) package

Signed-off-by: Deluan <deluan@navidrome.org>

* add `RawTags` to model

Signed-off-by: Deluan <deluan@navidrome.org>

* support parsing MBIDs for roles (from the https://github.com/kgarner7/picard-all-mbids plugin) (#3698)


* parse standard roles, vorbis/m4a work for now

* fix djmixer

* working roles, use DJ-mix

* add performers to file

* map mbids

* add a few more tests

* add test

Signed-off-by: Deluan <deluan@navidrome.org>

* try to simplify the performers logic

Signed-off-by: Deluan <deluan@navidrome.org>

* stylistic changes

---------

Signed-off-by: Deluan <deluan@navidrome.org>
Co-authored-by: Deluan <deluan@navidrome.org>

* remove param mutation

Signed-off-by: Deluan <deluan@navidrome.org>

* run automated SQLite optimizations

Signed-off-by: Deluan <deluan@navidrome.org>

* fix playlists import/export on Windows

* fix import playlists

* fix export playlists

* better handling of Windows volumes

Signed-off-by: Deluan <deluan@navidrome.org>

* handle more album ID reassignments

Signed-off-by: Deluan <deluan@navidrome.org>

* allow adding/overriding tags in the config file

Signed-off-by: Deluan <deluan@navidrome.org>

* fix(ui): Fix playlist track id, handle missing tracks better (#3734)

- Use `mediaFileId` instead of `id` for playlist tracks
- Only fetch if the file is not missing
- If extractor fails to get the file, also error (rather than panic)

* optimize DB after each scan.

Signed-off-by: Deluan <deluan@navidrome.org>

* remove sortable from AlbumSongs columns

Signed-off-by: Deluan <deluan@navidrome.org>

* simplify query to get missing tracks

Signed-off-by: Deluan <deluan@navidrome.org>

* mark Scanner.Extractor as deprecated

Signed-off-by: Deluan <deluan@navidrome.org>

---------

Signed-off-by: Deluan <deluan@navidrome.org>
Signed-off-by: Henrik Nordvik <henrikno@gmail.com>
Co-authored-by: Caio Cotts <caio@cotts.com.br>
Co-authored-by: Henrik Nordvik <henrikno@gmail.com>
Co-authored-by: Kendall Garner <17521368+kgarner7@users.noreply.github.com>
2025-02-19 20:35:17 -05:00

607 lines
17 KiB
Go

package core
import (
"context"
"errors"
"net/url"
"sort"
"strings"
"time"
"github.com/Masterminds/squirrel"
"github.com/navidrome/navidrome/conf"
"github.com/navidrome/navidrome/core/agents"
_ "github.com/navidrome/navidrome/core/agents/lastfm"
_ "github.com/navidrome/navidrome/core/agents/listenbrainz"
_ "github.com/navidrome/navidrome/core/agents/spotify"
"github.com/navidrome/navidrome/log"
"github.com/navidrome/navidrome/model"
"github.com/navidrome/navidrome/utils"
. "github.com/navidrome/navidrome/utils/gg"
"github.com/navidrome/navidrome/utils/random"
"github.com/navidrome/navidrome/utils/slice"
"github.com/navidrome/navidrome/utils/str"
"golang.org/x/sync/errgroup"
)
const (
maxSimilarArtists = 100
refreshDelay = 5 * time.Second
refreshTimeout = 15 * time.Second
refreshQueueLength = 2000
)
type ExternalMetadata interface {
UpdateAlbumInfo(ctx context.Context, id string) (*model.Album, error)
UpdateArtistInfo(ctx context.Context, id string, count int, includeNotPresent bool) (*model.Artist, error)
SimilarSongs(ctx context.Context, id string, count int) (model.MediaFiles, error)
TopSongs(ctx context.Context, artist string, count int) (model.MediaFiles, error)
ArtistImage(ctx context.Context, id string) (*url.URL, error)
AlbumImage(ctx context.Context, id string) (*url.URL, error)
}
type externalMetadata struct {
ds model.DataStore
ag *agents.Agents
artistQueue refreshQueue[auxArtist]
albumQueue refreshQueue[auxAlbum]
}
type auxAlbum struct {
model.Album
Name string
}
type auxArtist struct {
model.Artist
Name string
}
func NewExternalMetadata(ds model.DataStore, agents *agents.Agents) ExternalMetadata {
e := &externalMetadata{ds: ds, ag: agents}
e.artistQueue = newRefreshQueue(context.TODO(), e.populateArtistInfo)
e.albumQueue = newRefreshQueue(context.TODO(), e.populateAlbumInfo)
return e
}
func (e *externalMetadata) getAlbum(ctx context.Context, id string) (auxAlbum, error) {
var entity interface{}
entity, err := model.GetEntityByID(ctx, e.ds, id)
if err != nil {
return auxAlbum{}, err
}
var album auxAlbum
switch v := entity.(type) {
case *model.Album:
album.Album = *v
album.Name = str.Clear(v.Name)
case *model.MediaFile:
return e.getAlbum(ctx, v.AlbumID)
default:
return auxAlbum{}, model.ErrNotFound
}
return album, nil
}
func (e *externalMetadata) UpdateAlbumInfo(ctx context.Context, id string) (*model.Album, error) {
album, err := e.getAlbum(ctx, id)
if err != nil {
log.Info(ctx, "Not found", "id", id)
return nil, err
}
updatedAt := V(album.ExternalInfoUpdatedAt)
if updatedAt.IsZero() {
log.Debug(ctx, "AlbumInfo not cached. Retrieving it now", "updatedAt", updatedAt, "id", id, "name", album.Name)
album, err = e.populateAlbumInfo(ctx, album)
if err != nil {
return nil, err
}
}
// If info is expired, trigger a populateAlbumInfo in the background
if time.Since(updatedAt) > conf.Server.DevAlbumInfoTimeToLive {
log.Debug("Found expired cached AlbumInfo, refreshing in the background", "updatedAt", album.ExternalInfoUpdatedAt, "name", album.Name)
e.albumQueue.enqueue(&album)
}
return &album.Album, nil
}
func (e *externalMetadata) populateAlbumInfo(ctx context.Context, album auxAlbum) (auxAlbum, error) {
start := time.Now()
info, err := e.ag.GetAlbumInfo(ctx, album.Name, album.AlbumArtist, album.MbzAlbumID)
if errors.Is(err, agents.ErrNotFound) {
return album, nil
}
if err != nil {
log.Error("Error refreshing AlbumInfo", "id", album.ID, "name", album.Name, "artist", album.AlbumArtist,
"elapsed", time.Since(start), err)
return album, err
}
album.ExternalInfoUpdatedAt = P(time.Now())
album.ExternalUrl = info.URL
if info.Description != "" {
album.Description = info.Description
}
if len(info.Images) > 0 {
sort.Slice(info.Images, func(i, j int) bool {
return info.Images[i].Size > info.Images[j].Size
})
album.LargeImageUrl = info.Images[0].URL
if len(info.Images) >= 2 {
album.MediumImageUrl = info.Images[1].URL
}
if len(info.Images) >= 3 {
album.SmallImageUrl = info.Images[2].URL
}
}
err = e.ds.Album(ctx).UpdateExternalInfo(&album.Album)
if err != nil {
log.Error(ctx, "Error trying to update album external information", "id", album.ID, "name", album.Name,
"elapsed", time.Since(start), err)
} else {
log.Trace(ctx, "AlbumInfo collected", "album", album, "elapsed", time.Since(start))
}
return album, nil
}
func (e *externalMetadata) getArtist(ctx context.Context, id string) (auxArtist, error) {
var entity interface{}
entity, err := model.GetEntityByID(ctx, e.ds, id)
if err != nil {
return auxArtist{}, err
}
var artist auxArtist
switch v := entity.(type) {
case *model.Artist:
artist.Artist = *v
artist.Name = str.Clear(v.Name)
case *model.MediaFile:
return e.getArtist(ctx, v.ArtistID)
case *model.Album:
return e.getArtist(ctx, v.AlbumArtistID)
default:
return auxArtist{}, model.ErrNotFound
}
return artist, nil
}
func (e *externalMetadata) UpdateArtistInfo(ctx context.Context, id string, similarCount int, includeNotPresent bool) (*model.Artist, error) {
artist, err := e.refreshArtistInfo(ctx, id)
if err != nil {
return nil, err
}
err = e.loadSimilar(ctx, &artist, similarCount, includeNotPresent)
return &artist.Artist, err
}
func (e *externalMetadata) refreshArtistInfo(ctx context.Context, id string) (auxArtist, error) {
artist, err := e.getArtist(ctx, id)
if err != nil {
return auxArtist{}, err
}
// If we don't have any info, retrieves it now
updatedAt := V(artist.ExternalInfoUpdatedAt)
if updatedAt.IsZero() {
log.Debug(ctx, "ArtistInfo not cached. Retrieving it now", "updatedAt", updatedAt, "id", id, "name", artist.Name)
artist, err = e.populateArtistInfo(ctx, artist)
if err != nil {
return auxArtist{}, err
}
}
// If info is expired, trigger a populateArtistInfo in the background
if time.Since(updatedAt) > conf.Server.DevArtistInfoTimeToLive {
log.Debug("Found expired cached ArtistInfo, refreshing in the background", "updatedAt", updatedAt, "name", artist.Name)
e.artistQueue.enqueue(&artist)
}
return artist, nil
}
func (e *externalMetadata) populateArtistInfo(ctx context.Context, artist auxArtist) (auxArtist, error) {
start := time.Now()
// Get MBID first, if it is not yet available
if artist.MbzArtistID == "" {
mbid, err := e.ag.GetArtistMBID(ctx, artist.ID, artist.Name)
if mbid != "" && err == nil {
artist.MbzArtistID = mbid
}
}
// Call all registered agents and collect information
g := errgroup.Group{}
g.SetLimit(2)
g.Go(func() error { e.callGetImage(ctx, e.ag, &artist); return nil })
g.Go(func() error { e.callGetBiography(ctx, e.ag, &artist); return nil })
g.Go(func() error { e.callGetURL(ctx, e.ag, &artist); return nil })
g.Go(func() error { e.callGetSimilar(ctx, e.ag, &artist, maxSimilarArtists, true); return nil })
_ = g.Wait()
if utils.IsCtxDone(ctx) {
log.Warn(ctx, "ArtistInfo update canceled", "elapsed", "id", artist.ID, "name", artist.Name, time.Since(start), ctx.Err())
return artist, ctx.Err()
}
artist.ExternalInfoUpdatedAt = P(time.Now())
err := e.ds.Artist(ctx).UpdateExternalInfo(&artist.Artist)
if err != nil {
log.Error(ctx, "Error trying to update artist external information", "id", artist.ID, "name", artist.Name,
"elapsed", time.Since(start), err)
} else {
log.Trace(ctx, "ArtistInfo collected", "artist", artist, "elapsed", time.Since(start))
}
return artist, nil
}
func (e *externalMetadata) SimilarSongs(ctx context.Context, id string, count int) (model.MediaFiles, error) {
artist, err := e.getArtist(ctx, id)
if err != nil {
return nil, err
}
e.callGetSimilar(ctx, e.ag, &artist, 15, false)
if utils.IsCtxDone(ctx) {
log.Warn(ctx, "SimilarSongs call canceled", ctx.Err())
return nil, ctx.Err()
}
weightedSongs := random.NewWeightedChooser[model.MediaFile]()
addArtist := func(a model.Artist, weightedSongs *random.WeightedChooser[model.MediaFile], count, artistWeight int) error {
if utils.IsCtxDone(ctx) {
log.Warn(ctx, "SimilarSongs call canceled", ctx.Err())
return ctx.Err()
}
topCount := max(count, 20)
topSongs, err := e.getMatchingTopSongs(ctx, e.ag, &auxArtist{Name: a.Name, Artist: a}, topCount)
if err != nil {
log.Warn(ctx, "Error getting artist's top songs", "artist", a.Name, err)
return nil
}
weight := topCount * (4 + artistWeight)
for _, mf := range topSongs {
weightedSongs.Add(mf, weight)
weight -= 4
}
return nil
}
err = addArtist(artist.Artist, weightedSongs, count, 10)
if err != nil {
return nil, err
}
for _, a := range artist.SimilarArtists {
err := addArtist(a, weightedSongs, count, 0)
if err != nil {
return nil, err
}
}
var similarSongs model.MediaFiles
for len(similarSongs) < count && weightedSongs.Size() > 0 {
s, err := weightedSongs.Pick()
if err != nil {
log.Warn(ctx, "Error getting weighted song", err)
continue
}
similarSongs = append(similarSongs, s)
}
return similarSongs, nil
}
func (e *externalMetadata) ArtistImage(ctx context.Context, id string) (*url.URL, error) {
artist, err := e.getArtist(ctx, id)
if err != nil {
return nil, err
}
e.callGetImage(ctx, e.ag, &artist)
if utils.IsCtxDone(ctx) {
log.Warn(ctx, "ArtistImage call canceled", ctx.Err())
return nil, ctx.Err()
}
imageUrl := artist.ArtistImageUrl()
if imageUrl == "" {
return nil, agents.ErrNotFound
}
return url.Parse(imageUrl)
}
func (e *externalMetadata) AlbumImage(ctx context.Context, id string) (*url.URL, error) {
album, err := e.getAlbum(ctx, id)
if err != nil {
return nil, err
}
info, err := e.ag.GetAlbumInfo(ctx, album.Name, album.AlbumArtist, album.MbzAlbumID)
if errors.Is(err, agents.ErrNotFound) {
return nil, err
}
if utils.IsCtxDone(ctx) {
log.Warn(ctx, "AlbumImage call canceled", ctx.Err())
return nil, ctx.Err()
}
// Return the biggest image
var img agents.ExternalImage
for _, i := range info.Images {
if img.Size <= i.Size {
img = i
}
}
if img.URL == "" {
return nil, agents.ErrNotFound
}
return url.Parse(img.URL)
}
func (e *externalMetadata) TopSongs(ctx context.Context, artistName string, count int) (model.MediaFiles, error) {
artist, err := e.findArtistByName(ctx, artistName)
if err != nil {
log.Error(ctx, "Artist not found", "name", artistName, err)
return nil, nil
}
return e.getMatchingTopSongs(ctx, e.ag, artist, count)
}
func (e *externalMetadata) getMatchingTopSongs(ctx context.Context, agent agents.ArtistTopSongsRetriever, artist *auxArtist, count int) (model.MediaFiles, error) {
songs, err := agent.GetArtistTopSongs(ctx, artist.ID, artist.Name, artist.MbzArtistID, count)
if errors.Is(err, agents.ErrNotFound) {
return nil, nil
}
if err != nil {
return nil, err
}
var mfs model.MediaFiles
for _, t := range songs {
mf, err := e.findMatchingTrack(ctx, t.MBID, artist.ID, t.Name)
if err != nil {
continue
}
mfs = append(mfs, *mf)
if len(mfs) == count {
break
}
}
if len(mfs) == 0 {
log.Debug(ctx, "No matching top songs found", "name", artist.Name)
} else {
log.Debug(ctx, "Found matching top songs", "name", artist.Name, "numSongs", len(mfs))
}
return mfs, nil
}
func (e *externalMetadata) findMatchingTrack(ctx context.Context, mbid string, artistID, title string) (*model.MediaFile, error) {
if mbid != "" {
mfs, err := e.ds.MediaFile(ctx).GetAll(model.QueryOptions{
Filters: squirrel.And{
squirrel.Eq{"mbz_recording_id": mbid},
squirrel.Eq{"missing": false},
},
})
if err == nil && len(mfs) > 0 {
return &mfs[0], nil
}
return e.findMatchingTrack(ctx, "", artistID, title)
}
mfs, err := e.ds.MediaFile(ctx).GetAll(model.QueryOptions{
Filters: squirrel.And{
squirrel.Or{
squirrel.Eq{"artist_id": artistID},
squirrel.Eq{"album_artist_id": artistID},
},
squirrel.Like{"order_title": str.SanitizeFieldForSorting(title)},
squirrel.Eq{"missing": false},
},
Sort: "starred desc, rating desc, year asc, compilation asc ",
Max: 1,
})
if err != nil || len(mfs) == 0 {
return nil, model.ErrNotFound
}
return &mfs[0], nil
}
func (e *externalMetadata) callGetURL(ctx context.Context, agent agents.ArtistURLRetriever, artist *auxArtist) {
artisURL, err := agent.GetArtistURL(ctx, artist.ID, artist.Name, artist.MbzArtistID)
if err != nil {
return
}
artist.ExternalUrl = artisURL
}
func (e *externalMetadata) callGetBiography(ctx context.Context, agent agents.ArtistBiographyRetriever, artist *auxArtist) {
bio, err := agent.GetArtistBiography(ctx, artist.ID, str.Clear(artist.Name), artist.MbzArtistID)
if err != nil {
return
}
bio = str.SanitizeText(bio)
bio = strings.ReplaceAll(bio, "\n", " ")
artist.Biography = strings.ReplaceAll(bio, "<a ", "<a target='_blank' ")
}
func (e *externalMetadata) callGetImage(ctx context.Context, agent agents.ArtistImageRetriever, artist *auxArtist) {
images, err := agent.GetArtistImages(ctx, artist.ID, artist.Name, artist.MbzArtistID)
if err != nil {
return
}
sort.Slice(images, func(i, j int) bool { return images[i].Size > images[j].Size })
if len(images) >= 1 {
artist.LargeImageUrl = images[0].URL
}
if len(images) >= 2 {
artist.MediumImageUrl = images[1].URL
}
if len(images) >= 3 {
artist.SmallImageUrl = images[2].URL
}
}
func (e *externalMetadata) callGetSimilar(ctx context.Context, agent agents.ArtistSimilarRetriever, artist *auxArtist,
limit int, includeNotPresent bool) {
similar, err := agent.GetSimilarArtists(ctx, artist.ID, artist.Name, artist.MbzArtistID, limit)
if len(similar) == 0 || err != nil {
return
}
start := time.Now()
sa, err := e.mapSimilarArtists(ctx, similar, includeNotPresent)
log.Debug(ctx, "Mapped Similar Artists", "artist", artist.Name, "numSimilar", len(sa), "elapsed", time.Since(start))
if err != nil {
return
}
artist.SimilarArtists = sa
}
func (e *externalMetadata) mapSimilarArtists(ctx context.Context, similar []agents.Artist, includeNotPresent bool) (model.Artists, error) {
var result model.Artists
var notPresent []string
artistNames := slice.Map(similar, func(artist agents.Artist) string { return artist.Name })
// Query all artists at once
clauses := slice.Map(artistNames, func(name string) squirrel.Sqlizer {
return squirrel.Like{"artist.name": name}
})
artists, err := e.ds.Artist(ctx).GetAll(model.QueryOptions{
Filters: squirrel.Or(clauses),
})
if err != nil {
return nil, err
}
// Create a map for quick lookup
artistMap := make(map[string]model.Artist)
for _, artist := range artists {
artistMap[artist.Name] = artist
}
// Process the similar artists
for _, s := range similar {
if artist, found := artistMap[s.Name]; found {
result = append(result, artist)
} else {
notPresent = append(notPresent, s.Name)
}
}
// Then fill up with non-present artists
if includeNotPresent {
for _, s := range notPresent {
// Let the ID empty to indicate that the artist is not present in the DB
sa := model.Artist{Name: s}
result = append(result, sa)
}
}
return result, nil
}
func (e *externalMetadata) findArtistByName(ctx context.Context, artistName string) (*auxArtist, error) {
artists, err := e.ds.Artist(ctx).GetAll(model.QueryOptions{
Filters: squirrel.Like{"artist.name": artistName},
Max: 1,
})
if err != nil {
return nil, err
}
if len(artists) == 0 {
return nil, model.ErrNotFound
}
artist := &auxArtist{
Artist: artists[0],
Name: str.Clear(artists[0].Name),
}
return artist, nil
}
func (e *externalMetadata) loadSimilar(ctx context.Context, artist *auxArtist, count int, includeNotPresent bool) error {
var ids []string
for _, sa := range artist.SimilarArtists {
if sa.ID == "" {
continue
}
ids = append(ids, sa.ID)
}
similar, err := e.ds.Artist(ctx).GetAll(model.QueryOptions{
Filters: squirrel.Eq{"artist.id": ids},
})
if err != nil {
log.Error("Error loading similar artists", "id", artist.ID, "name", artist.Name, err)
return err
}
// Use a map and iterate through original array, to keep the same order
artistMap := make(map[string]model.Artist)
for _, sa := range similar {
artistMap[sa.ID] = sa
}
var loaded model.Artists
for _, sa := range artist.SimilarArtists {
if len(loaded) >= count {
break
}
la, ok := artistMap[sa.ID]
if !ok {
if !includeNotPresent {
continue
}
la = sa
la.ID = ""
}
loaded = append(loaded, la)
}
artist.SimilarArtists = loaded
return nil
}
type refreshQueue[T any] chan<- *T
func newRefreshQueue[T any](ctx context.Context, processFn func(context.Context, T) (T, error)) refreshQueue[T] {
queue := make(chan *T, refreshQueueLength)
go func() {
for {
select {
case <-ctx.Done():
return
case <-time.After(refreshDelay):
ctx, cancel := context.WithTimeout(ctx, refreshTimeout)
select {
case item := <-queue:
_, _ = processFn(ctx, *item)
cancel()
case <-ctx.Done():
cancel()
}
}
}
}()
return queue
}
func (q *refreshQueue[T]) enqueue(item *T) {
select {
case *q <- item:
default: // It is ok to miss a refresh request
}
}