Release v2.3.1

Bug fixes:

• Path Formats: Fixed a regression where path legalization incorrectly removed parts of user-configured path formats that followed a dot (.). 🐛 (#5771)

For packagers:

• Force poetry version below 2 to avoid it mangling file modification times in sdist package. 🐛 (#5770)

Release v2.3.0

Beets now requires Python 3.9 or later since support for EOL Python 3.8 has been dropped.

New features

• LastGenre Plugin: The new configuration option, keep_existing, provides more fine-grained control over how pre-populated genre tags are handled. The force option now behaves in a more conventional manner. 🐛 (#4982)
• Lyrics Plugin: Add new configuration option dist_thresh to control the maximum allowed distance between the lyrics search result and the tagged item's artist and title. This is useful for preventing false positives when fetching lyrics.
• Lyrics Plugin: Rewrite lyrics translation functionality to use Azure AI Translator API and add relevant instructions to the documentation.
• MBSync Plugin: Add support for all metadata sorces.
• Missing Plugin: Add support for all metadata sources.

Bug fixes

• Fix ambiguous column name sqlite3.OperationalError that occured in album queries that filtered album track titles, for example beet list -a keyword title:foo.
• ImageMagick 7.1.1-44 is now supported.
• Synchronise files included in the source distribution with what we used to have before the introduction of Poetry. 🐛 (#5531) 🐛 (#5526)
• FetchArt Plugin: Fix fetchart bug where a tempfile could not be deleted due to never being properly closed. 🐛 (#5521)
• LastGenre Plugin: Fix track-level genre handling. Now when an album-level genre is set already, single tracks don't fall back to the album's genre and request their own last.fm genre. Also log messages regarding what's been tagged are now more polished. 🐛 (#5582)
• ListenBrainz Plugin: Fix UnboundLocalError in cases where 'mbid' is not defined.
• ListenBrainz Plugin: Fix rST formatting for URLs of Listenbrainz API Key documentation and config.yaml.
• Lyrics Plugin: Do not attempt to search for lyrics if either the artist or title is missing and ignore artist_sort value if it is empty. 🐛 (#2635)
• Lyrics Plugin: Fix fetching lyrics from lrclib source. If we cannot find lyrics for a specific album, artist, title combination, the plugin now tries to search for the artist and title and picks the most relevant result. Update the default sources configuration to prioritize lrclib over other sources since it returns reliable results quicker than others. 🐛 (#5102)
• Lyrics Plugin: Fix plugin crash when genius backend returns empty lyrics. 🐛 (#5583)
• Lyrics Plugin: Fix the issue with genius backend not being able to match lyrics when there is a slight variation in the artist name. 🐛 (#4791)
• Lyrics Plugin: LRCLib will fallback to plain lyrics if synced lyrics are not found and synced flag is set to yes.
• Lyrics Plugin: Rewrite lyrics tests using pytest to provide isolated configuration for each test case. This fixes the issue where some tests failed because they read developers' local lyrics configuration. 🐛 (#5133)
• ParentWork Plugin: Only output parentwork changes when running in verbose mode.
• Sort Order: Fix a bug that would raise an exception when sorting on a non-string field that is not populated in all items. 🐛 (#5512)
• Thumbnails Plugin: Fix API call to GIO on big endian architectures (like s390x) in thumbnails plugin. 🐛 (#5708)
• import: Fix MemoryError and improve performance tagging large albums by replacing munkres library with lap.lapjv. 🐛 (#5207)
• write: Fix the issue where for certain files differences in mb_artistid, mb_albumartistid and albumtype fields are shown on every attempt to write tags. Note: your music needs to be reimported with beet import -LI or synchronised with beet mbsync in order to fix this! 🐛 (#5265) 🐛 (#5371) 🐛 (#4715)

For packagers

• External plugin developers: beetsplug/__init__.py file can be removed from your plugin as beets now uses native/implicit namespace package setup.
• The minimum supported Python version is now 3.9.

Other changes

• Database models are now serializable with pickle.
• Release workflow: fix the issue where the new release tag is created for the wrong (outdated) commit. Now the tag is created in the same workflow step right after committing the version update. 🐛 (#5539)
• FtInTitle Plugin: Optimize the plugin by avoiding unnecessary writes to the database.
• Smart Playlist Plugin: URL-encode additional item fields within generated EXTM3U playlists instead of JSON-encoding them.
• typehints: ./beets/importer.py file now has improved typehints.
• typehints: ./beets/plugins.py file now includes typehints.

Release v2.2.0

New features

• /plugins/substitute: Allow the replacement string to use capture groups from the match. It is thus possible to create more general rules, applying to many different artists at once.

Bug fixes

• Bring back test files and the manual to the source distribution tarball. 🐛 (#5513)
• Fix bug where matcher doesn't consider medium number when importing. This makes it difficult to import hybrid SACDs and other releases with duplicate tracks. 🐛 (#5148)
• Check if running python from the Microsoft Store and provide feedback to install from python.org. 🐛 (#5467)

Other changes

• Changed bitesize label to good first issue. Our contribute page is now automatically populated with these issues. 🐛 (#4855)

Release v2.1.0

New features

• Ability to query albums with track db fields and vice-versa, for example beet list -a title:something or beet list artpath:cover. Consequently album queries involving path field have been sped up, like beet list -a path:/path/.
• Beets now uses platformdirs to determine the default music directory. This location varies between systems -- for example, users can configure it on Unix systems via user-dirs.dirs(5).
• New template function added: %capitalize. Converts the first letter of the text to uppercase and the rest to lowercase.
• Plugin autobpm: Add new configuration option beat_track_kwargs which enables adjusting keyword arguments supplied to librosa's beat_track function call.
• Plugin ftintitle: New keep_in_artist option for the plugin, which allows keeping the "feat." part in the artist metadata while still changing the title.

Bug fixes

• Album flexible fields are now correctly saved. For instance MusicBrainz external links such as bandcamp_album_id will be available on albums in addition to tracks. For albums already in your library, a re-import is required for the fields to be added. Such a re-import can be done with, in this case, beet import -L data_source:=MusicBrainz.
• Fix lyrics plugin only getting part of the lyrics from Genius.com 🐛 (#4815)
• Fix the TypeError when set_fields is provided non-string values. 🐛 (#4840)
• Fix the auto value for the reflink config option.
• Improve naming of temporary files by separating the random part with the file extension.
• Plugin autobpm: Fix the TypeError where tempo was being returned as a numpy array. Update librosa dependency constraint to prevent similar issues in the future. 🐛 (#5289)
• Plugin convert: Fixed the convert plugin no_convert option so that it no longer treats "and" and "or" queries the same. To maintain previous behaviour add commas between your query keywords. For help see combiningqueries.
• Plugin discogs: Fix the TypeError when there is no description.
• Plugin ftintitle: The detection of a "feat. X" part in a song title does not produce any false positives caused by words like "and" or "with" anymore. 🐛 (#5441)
• Plugin ftintitle: The detection of a "feat. X" part now also matches such parts if they are in parentheses or brackets. 🐛 (#5436)
• Plugin lyrics: Update tekstowo backend to fetch lyrics directly since recent updates to their website made it unsearchable. 🐛 (#5456)
• Use single quotes in all SQL queries 🐛 (#4709)

For packagers

• The beet script has been removed from the repository.
• The typing_extensions is required for Python 3.10 and below.
• The minimum supported Python version is now 3.8.

Other changes

• Added caching for dependency installation in all CI jobs which speeds them up a bit, especially the tests.
• GitHub workflows have been reorganised for clarity: style, linting, type and docs checks now live in separate jobs and are named accordingly.
• Installation instructions have been made consistent across plugins documentation. Users should simply install beets with an extra of the corresponding plugin name in order to install extra dependencies for that plugin.
• Plugin autobpm: Add plugin dependencies to pyproject.toml under the autobpm extra and update the plugin installation instructions in the docs. Since importing the bpm calculation functionality from librosa takes around 4 seconds, update the plugin to only do so when it actually needs to calculate the bpm. Previously this import was being done immediately, so every beet invocation was being delayed by a couple of seconds. 🐛 (#5185)
• The linting workflow has been made to run only when Python files or documentation is changed, and they only check the changed files. When dependencies are updated (poetry.lock), then the entire code base is checked.
• The long-deprecated beets.util.confit module has been removed. This may cause extremely outdated external plugins to fail to load.
• contributing: Since poetry now manages local virtual environments, tox has been replaced by a task runner poethepoet. This change affects beets developers and contributors. Please see updates in the development-tools section for more details. Type poe while in the project directory to see the available commands.
• contributing: The project now uses poetry for packaging and dependency management. This change affects project management and mostly affects beets developers. Please see updates in getting-the-source and testing for more information.

Release v2.0.0

With this release, beets now requires Python 3.7 or later (it removes support for Python 3.6).

Major new features

• The beets importer UI received a major overhaul. Several new configuration options are available for customizing layout and colors: ui_options. 🐛 (#3721) 🐛 (#5028)

New features

• edit: Prefer editor from VISUAL environment variable over EDITOR.
• config: Prefer editor from VISUAL environment variable over EDITOR.
• listenbrainz: Add initial support for importing history and playlists from ListenBrainz 🐛 (#1719)
• mbsubmit: add new prompt choices helping further to submit unmatched tracks to MusicBrainz faster.
• spotify: We now fetch track's ISRC, EAN, and UPC identifiers from Spotify when using the spotifysync command. 🐛 (#4992)
• discogs: supply a value for the cover_art_url attribute, for use by fetchart. 🐛 (#429)
• update: added `-e` flag for excluding fields from being updated.
• deezer: Import rank and other attributes from Deezer during import and add a function to update the rank of existing items. 🐛 (#4841)
• resolve transl-tracklisting relations for pseudo releases and merge data with the actual release 🐛 (#654)
• Fetchart: Use the right field (spotify_album_id) to obtain the Spotify album id 🐛 (#4803)
• Prevent reimporting album if it is permanently removed from Spotify 🐛 (#4800)
• Added option to use cover_art_url as an album art source in the fetchart plugin. 🐛 (#4707)
• fetchart: The plugin can now get album art from spotify.
• Added option to specify a URL in the embedart plugin. 🐛 (#83)
• list singleton:true queries have been made faster
• list singleton:1 and singleton:0 can now alternatively be used in queries, same as comp
• --from-logfile now parses log files using a UTF-8 encoding in beets/beets/ui/commands.py. 🐛 (#4693)
• bareasc lookups have been made faster
• list lookups using the pattern operator :: have been made faster
• Added additional error handling for spotify plugin. 🐛 (#4686)
• We now import the remixer field from Musicbrainz into the library. 🐛 (#4428)
• mbsubmit: Added a new mbsubmit command to print track information to be submitted to MusicBrainz after initial import. 🐛 (#4455)
• Added spotify_updated field to track when the information was last updated.
• We now import and tag the album information when importing singletons using Spotify source. 🐛 (#4398)
• spotify: The plugin now provides an additional command spotifysync that allows getting track popularity and audio features information from Spotify. 🐛 (#4094)
• spotify: The plugin now records Spotify-specific IDs in the spotify_album_id, spotify_artist_id, and spotify_track_id fields. 🐛 (#4348)
• Create the parental directories for database if they do not exist. 🐛 (#3808) 🐛 (#4327)
• musicbrainz-config: a new musicbrainz.enabled option allows disabling the MusicBrainz metadata source during the autotagging process
• kodiupdate: Now supports multiple kodi instances 🐛 (#4101)
• Add the item fields bitrate_mode, encoder_info and encoder_settings.
• Add query prefixes = and ~.
• A new configuration option, duplicate_keys, lets you change which fields the beets importer uses to identify duplicates. 🐛 (#1133) 🐛 (#4199)
• Add exact match <exact-match> queries, using the prefixes = and =~. 🐛 (#4251)
• discogs: Permit appending style to genre.
• discogs: Implement item_candidates for matching singletons.
• discogs: Check for compliant discogs_client module.
• convert: Add a new auto_keep option that automatically converts files but keeps the originals in the library. 🐛 (#1840) 🐛 (#4302)
• Added a -P (or --disable-plugins) flag to specify one/multiple plugin(s) to be disabled at startup.
• import-options: Add support for re-running the importer on paths in log files that were created with the -l (or --logfile) argument. 🐛 (#4379) 🐛 (#4387)
• Preserve mtimes from archives 🐛 (#4392)
• Add %sunique{} <sunique> template to disambiguate between singletons. 🐛 (#4438)
• Add a new import.ignored_alias_types config option to allow for specific alias types to be skipped over when importing items/albums.
• smartplaylist: A new --pretend option lets the user see what a new or changed smart playlist saved in the config is actually returning. 🐛 (#4573)
• fromfilename: Add debug log messages that inform when the plugin replaced bad (missing) artist, title or tracknumber metadata. 🐛 (#4561) 🐛 (#4600)
• musicbrainz-config: MusicBrainz release pages often link to related metadata sources like Discogs, Bandcamp, Spotify, Deezer and Beatport. When enabled via the musicbrainz.external_ids options, release ID's will be extracted from those URL's and imported to the library. 🐛 (#4220)
• convert: Add support for generating m3u8 playlists together with converted media files. 🐛 (#4373)
• Fetch the release_group_title field from MusicBrainz. 🐛 4809
• discogs: Add support for applying album information on singleton imports. 🐛 4716
• smartplaylist: During explicit runs of the splupdate command, the log message "Creating playlist ..."" is now displayed instead of hidden in the debug log, which states some form of progress through the UI. 🐛 (#4861)
• subsonicupdate: Updates are now triggered whenever either the beets database is changed or a smart playlist is created/updated. 🐛 4862
• importfeeds: Add a new output format allowing to save a playlist once per import session. 🐛 4863
• Make ArtResizer work with PIL/pillow 10.0.0 removals. 🐛 (#4869)
• A new configuration option, duplicate_verbose_prompt, allows changing how duplicates are presented during import. 🐛 4866
• embyupdate: Add handling for private users by adding userid config option. 🐛 (#4402)
• substitute: Add the new plugin substitute as an alternative to the rewrite plugin. The main difference between them being that rewrite modifies files' metadata and substitute does not. 🐛 (#2786)
• Add support for artists and albumartists multi-valued tags. 🐛 (#505)
• autobpm: Add the autobpm plugin which uses Librosa to calculate the BPM of the audio. 🐛 (#3856)
• fetchart: Fix the error with CoverArtArchive where the maxwidth option would not be used to download a pre-sized thumbnail for release groups, as is already done with releases.
• fetchart: Fix the error with CoverArtArchive where no cover would be found when the maxwidth option matches a pre-sized thumbnail size, but no thumbnail is provided by CAA. We now fallback to the raw image.
• advancedrewrite: Add an advanced version of the rewrite plugin which allows to replace fields based on a given library query.
• lyrics: Add LRCLIB as a new lyrics provider and a new synced option to prefer synced lyrics over plain lyrics.
• import: Expose import.quiet_fallback as CLI option.
• import: Expose import.incremental_skip_later as CLI option.
• smartplaylist: Expose config options as CLI options.
• smartplaylist: Add new option smartplaylist.output.
• smartplaylist: Add new option smartplaylist.uri_format.
• Sorted the default configuration file into categories. 🐛 (#4987)
• convert: Don't treat WAVE (.wav) files as lossy anymore when using the never_convert_lossy_files option. They will get transcoded like the other lossless formats.
• Add support for barcode field. 🐛 (#3172)
• smartplaylist: Add new config option smartplaylist.fields.

Bug fixes

• lastimport: Improve error handling in the process_tracks function and enable it to be used with other plugins.
• spotify: Improve handling of ConnectionError.
• deezer: Improve Deezer plugin error handling and set requests timeout to 10 seconds. 🐛 (#4983)
• spotify: Add bad gateway (502) error handling.
• spotify: Add a limit of 3 retries, instead of retrying endlessly when the API is not available.
• Fix a crash when the Spotify API timeouts or does not return a Retry-After interval. 🐛 (#4942)
• scrub: Fixed the import behavior where scrubbed database tags were restored to newly imported tracks with config settings scrub.auto: yes and import.write: no. 🐛 (#4326)
• deezer: Fixed the error where Deezer plugin would crash if non-Deezer id is passed during import.
• fetchart: Fix fetching from Cover Art Archive when the maxwidth option is set to one of the supported Cover Art Archive widths.
• discogs: Fix "Discogs plugin replacing Feat. or Ft. with a comma" by fixing an oversight that removed a functionality from the code base when the MetadataSourcePlugin abstract class was introduced in PR's #3335 and #3371. 🐛 (#4401)
• convert: Set default max_bitrate value to None to avoid transcoding when this parameter is not set. 🐛 (#4472)
• replaygain: Avoid a crash when errors occur in the analysis backend. 🐛 (#4506)
• We now use Python's defaults for command-line argument encoding, which should reduce the chance for errors and "file not found" failures when invoking other command-line tools, especially on Windows. 🐛 (#4507)
• We now respect the Spotify API's rate limiting, which avoids crashing when the API reports code 429 (too many requests). 🐛 (#4370)
• Fix implicit paths OR queries (e.g. beet list /path/ , /other-path/) which have previously been returning the entire library. 🐛 (#1865)
• The Discogs release ID is now populated correctly to the discogs_albumid field again (it was no longer working after Discogs changed their release URL format). 🐛 (#4225)
• The ...

beets 1.6.0

This release is our first experiment with time-based releases! We are aiming to publish a new release of beets every 3 months. We therefore have a healthy but not dizzyingly long list of new features and fixes.

With this release, beets now requires Python 3.6 or later (it removes support for Python 2.7, 3.4, and 3.5). There are also a few other dependency changes---if you're a maintainer of a beets package for a package manager, thank you for your ongoing efforts, and please see the list of notes below.

Major new features:

• When fetching genres from MusicBrainz, we now include genres from the release group (in addition to the release). We also prioritize genres based on the number of votes. Thanks to @aereaux.
• Primary and secondary release types from MusicBrainz are now stored in a new albumtypes field. Thanks to edgars-supe{.interpreted-text role="user"}. #2200
• An accompanying new albumtypes includes some options for formatting this new albumtypes field. Thanks to edgars-supe{.interpreted-text role="user"}.

Other new things:

• permissions: The plugin now sets cover art permissions to match the audio file permissions.
• unimported: A new configuration option supports excluding specific subdirectories in library.
• info: Add support for an --album flag.
• export: Similarly add support for an --album flag.
• beet move now highlights path differences in color (when enabled).
• When moving files and a direct rename of a file is not possible (for example, when crossing filesystems), beets now copies to a temporary file in the target folder first and then moves to the destination instead of directly copying the target path. This gets us closer to always updating files atomically. Thanks to @catap. #4060
• fetchart: Add a new option to store cover art as non-progressive image. This is useful for DAPs that do not support progressive images. Set deinterlace: yes in your configuration to enable this conversion.
• fetchart: Add a new option to change the file format of cover art images. This may also be useful for DAPs that only support some image formats.
• Support flexible attributes in %aunique. #2678 #3553
• Make %aunique faster, especially when using inline fields. #4145

Bug fixes:

• lyrics: Fix a crash when Beautiful Soup is not installed. #4027
• discogs: Support a new Discogs URL format for IDs. #4080
• discogs: Remove built-in rate-limiting because the Discogs Python library we use now has its own rate-limiting. 🐛 4108
• export: Fix some duplicated output.
• aura: Fix a potential security hole when serving image files. #4160

For plugin developers:

• :pybeets.library.Item.destination{.interpreted-text role="meth"} now accepts a replacements argument to be used in favor of the default.
• The pluginload event is now sent after plugin types and queries are available, not before.
• A new plugin event, album_removed, is called when an album is removed from the library (even when its file is not deleted from disk).

Here are some notes for packagers:

• As noted above, the minimum Python version is now 3.6.
• We fixed a flaky test, named test_album_art in the test_zero.py file, that some distributions had disabled. Disabling this test should no longer be necessary. #4037 #4038
• This version of beets no longer depends on the six library. #4030
• The gmusic plugin was removed since Google Play Music has been shut down. Thus, the optional dependency on gmusicapi does not exist anymore. #4089

beets 1.5.0

This long overdue release of beets includes far too many exciting and useful features than could ever be satisfactorily enumerated. As a technical detail, it also introduces two new external libraries: MediaFile and Confuse used to be part of beets but are now reusable dependencies---packagers, please take note. Finally, this is the last version of beets where we intend to support Python 2.x and 3.5; future releases will soon require Python 3.6.

One non-technical change is that we moved our official #beets home on IRC from freenode to Libera.Chat.

Major new features:

• Fields in queries now fall back to an item's album and check its fields too. Notably, this allows querying items by an album's attribute: in other words, beet list foo:bar will not only find tracks with the foo attribute; it will also find tracks on albums that have the foo attribute. This may be particularly useful in the path-format-config`, which matches individual items to decide which path to use. Thanks to @FichteFoll. #2797 #2988
• A new reflink config option instructs the importer to create fast, copy-on-write file clones on filesystems that support them. Thanks to @rubdos.
• A new unimported lets you find untracked files in your library directory.
• The aura has arrived! Try out the future of remote music library access today.
• We now fetch information about works from MusicBrainz. MusicBrainz matches provide the fields work (the title), mb_workid (the MBID), and work_disambig (the disambiguation string). Thanks to @dosoe. #2580 #3272
• A new parentwork gets information about the original work, which is useful for classical music. Thanks to @dosoe. #2580 #3279
• bpd: BPD now supports most of the features of version 0.16 of the MPD protocol. This is enough to get it talking to more complicated clients like ncmpcpp, but there are still some incompatibilities, largely due to MPD commands we don't support yet. (Let us know if you find an MPD client that doesn't get along with BPD!) #3214 #800
• A new deezer can autotag tracks and albums using the Deezer database. Thanks to @rhlahuja. #3355
• A new bareasc provides a new query type: "bare ASCII" queries that ignore accented characters, treating them as though they were plain ASCII characters. Use the # prefix with list or other commands. #3882
• fetchart: The plugin can now get album art from last.fm. #3530
• web: The API now supports the HTTP DELETE and PATCH methods for modifying items. They are disabled by default; set readonly: no in your configuration file to enable modification via the API. #3870

Other new things:

• beet remove now also allows interactive selection of items from the query, similar to beet modify.
• Enable HTTPS for MusicBrainz by default and add configuration option https for custom servers. See musicbrainz-config for more details.
• mpdstats: Add a new strip_path option to help build the right local path from MPD information.
• convert: Conversion can now parallelize conversion jobs on Python 3.
• lastgenre: Add a new title_case config option to make title-case formatting optional.
• There's a new message when running beet config when there's no available configuration file. #3779
• When importing a duplicate album, the prompt now says "keep all" instead of "keep both" to reflect that there may be more than two albums involved. #3569
• chroma: The plugin now updates file metadata after generating fingerprints through the submit command.
• lastgenre: Added more heavy metal genres to the built-in genre filter lists.
• A new subsonicplaylist can import playlists from a Subsonic server.
• subsonicupdate: The plugin now automatically chooses between token- and password-based authentication based on the server version.
• A new extra_tags configuration option lets you use more metadata in MusicBrainz queries to further narrow the search.
• A new fish adds Fish shell tab autocompletion to beets.
• plugins/fetchart and plugins/embedart: Added a new quality option that controls the quality of the image output when the image is resized.
• plugins/keyfinder: Added support for keyfinder-cli. Thanks to @BrainDamage.
• plugins/fetchart: Added a new high_resolution config option to allow downloading of higher resolution iTunes artwork (at the expense of file size). #3391
• plugins/discogs: The plugin applies two new fields: discogs_labelid and discogs_artistid. #3413
• export: Added a new -f (--format) flag, which can export your data as JSON, JSON lines, CSV, or XML. Thanks to @austinmm. #3402
• convert: Added a new -l (--link) flag and link option as well as the -H (--hardlink) flag and hardlink option, which symlink or hardlink files that do not need to be converted (instead of copying them). #2324
• replaygain: The plugin now supports a per_disc option that enables calculation of album ReplayGain on disc level instead of album level. Thanks to @samuelnilsson. #293
• replaygain: The new ffmpeg ReplayGain backend supports R128_ tags. #3056
• plugins/replaygain: A new r128_targetlevel configuration option defines the reference volume for files using R128_ tags. targetlevel only configures the reference volume for REPLAYGAIN_ files. #3065
• discogs: The plugin now collects the "style" field. Thanks to @thedevilisinthedetails. #2579 #3251
• absubmit: By default, the plugin now avoids re-analyzing files that already have AcousticBrainz data. There are new force and pretend options to help control this new behavior. Thanks to @SusannaMaria. #3318
• discogs: The plugin now also gets genre information and a new discogs_albumid field from the Discogs API. Thanks to @thedevilisinthedetails. #465 #3322
• acousticbrainz: The plugin now fetches two more additional fields: moods_mirex and timbre. Thanks to @malcops. #2860
• playlist and smartplaylist: A new forward_slash config option facilitates compatibility with MPD on Windows. Thanks to @MartyLake. #3331 #3334
• The data_source field, which indicates which metadata source was used during an autotagging import, is now also applied as an album-level flexible attribute. #3350 #1693
• beatport: The plugin now gets the musical key, BPM, and genre for each track. #2080
• A new bpsync can synchronize metadata changes from the Beatport database (like the existing mbsync for MusicBrainz).
• hook: The plugin now treats non-zero exit codes as errors. #3409
• subsonicupdate: A new url configuration replaces the older (and now deprecated) separate host, port, and contextpath config options. As a consequence, the plugin can now talk to Subsonic over HTTPS. Thanks to @jef. #3449
• discogs: The new index_tracks option enables incorporation of work names and intra-work divisions into imported track titles. Thanks to cole-miller. #3459
• web: The query API now interprets backslashes as path separators to support path queries. Thanks to @nmeum. #3567
• beet import now handles tar archives with bzip2 or gzip compression. #3606
• beet import also now handles 7z archives, via the py7zr library. Thanks to @arogl. #3906
• plexupdate: Added an option to use a secure connection to Plex server, and to ignore certificate validation errors if necessary. #2871
• convert: A new delete_originals configuration option can delete the source files after conversion during import. Thanks to logan-arens. #2947
• There is a new --plugins (or -p) CLI flag to specify a list of plugins to load.
• A new genres option fetches genre information from MusicBrainz. This functionality depends on functionality that is currently unreleased in the python-musicbrainzngs library: see PR #266 _. Thanks to :user:aereaux`.
• replaygain: Analysis now happens in parallel using the command and ffmpeg backends. #3478
• plugins/replaygain: The bs1770gain backend is removed. Thanks to @SamuelCook.
• Added trackdisambig which stores the recording disambiguation from MusicBrainz for each track. #1904
• plugins/fetchart: The new max_filesize configuration sets a maximum target image file size.
• badfiles: Checkers can now run during import with the check_on_import config option.
• export: The plugin is now much faster when using the --include-keys option is used. Thanks to @ssssam.
• The importer's set_fields` option now saves all updated fields to on-disk metadata. #3925 #3927
• We now fetch ISRC identifiers from MusicBrainz. Thanks to @aereaux.
• metasync: The plugin now also fetches the "Date Added" field from iTunes databases and stores it in the itunes_dateadded field. Thanks to @sandersantema.
• lyrics: Added a new Tekstowo.pl lyrics provider. Thanks to various people for the implementation and for reporting issues with the initial version. #3344 #3904 #3905 #3994
• beet update will now confirm that the user still wants to update if their library folder cannot be found, preventing the user from accidentally wiping out their beets database. Thanks to user: logan-arens. #1934

Fixes:

• Adapt to breaking changes in Python's ast module in Python 3.8.
• beatport: Fix the assignment of the genre field, and rename musical_key to initial_key. #3387
• lyrics: Fixed the Musixmatch backend for lyrics pages when lyrics are divided into multiple elements on the webpage, and when the lyrics are missing.
• web: Allow use of the backslash character in regex queries. #3867
• web: Fixed a small bug that caused the album art path to be redacted e...

beets 1.4.9

This small update is part of our attempt to release new versions more often! There are a few important fixes, and we're clearing the deck for a change to beets' dependencies in the next version.

The new feature is:

• You can use the NO_COLOR environment variable to disable terminal colors. #3273

There are some fixes in this release:

• Fix a regression in the last release that made the image resizer fail to detect older versions of ImageMagick. #3269
• gmusic: The oauth_file config option now supports more flexible path values, including ~ for the home directory. #3270
• gmusic: Fix a crash when using version 12.0.0 or later of the gmusicapi module. #3270
• Fix an incompatibility with Python 3.8's AST changes. #3278

Here's a note for packagers:

• pathlib is now an optional test dependency on Python 3.4+, removing the need for a Debian patch. #3275

beets 1.4.8

This release is far too long in coming, but it's a good one. There is the usual torrent of new features and a ridiculously long line of fixes, but there are also some crucial maintenance changes. We officially support Python 3.7 and 3.8, and some performance optimizations can (anecdotally) make listing your library more than three times faster than in the previous version.

The new core features are:

• A new config-aunique configuration option allows setting default options for the aunique template function.
• The albumdisambig field no longer includes the MusicBrainz release group disambiguation comment. A new releasegroupdisambig field has been added. #3024
• The modify command now allows resetting fixed attributes. For example, beet modify -a artist:beatles artpath! resets artpath attribute from matching albums back to the default value. #2497
• A new importer option, ignore_data_tracks, lets you skip audio tracks contained in data files. #3021

There are some new plugins:

• The playlist can query the beets library using M3U playlists. Thanks to @Holzhaus and @Xenopathic. #123 #3145
• The loadext allows loading of SQLite extensions, primarily for use with the ICU SQLite extension for internationalization. #3160 #3226
• The subsonicupdate can automatically update your Subsonic library. Thanks to @maffo999. #3001

And many improvements to existing plugins:

• lastgenre: Added option -A to match individual tracks and singletons. #3220 #3219
• play: The plugin can now emit a UTF-8 BOM, fixing some issues with foobar2000 and Winamp. Thanks to @mz2212. #2944
• gmusic:
  • Add a new option to automatically upload to Google Play Music library on track import. Thanks to @shuaiscott.
  • Add new options for Google Play Music authentication. Thanks to @thetarkus. #3002
• replaygain: albumpeak on large collections is calculated as the average, not the maximum. #3008 #3009
• chroma:
  • Now optionally has a bias toward looking up more relevant releases according to the preferred configuration options. Thanks to @Archer4499. #3017
  • Fingerprint values are now properly stored as strings, which prevents strange repeated output when running beet write. Thanks to @Holzhaus. #3097 #2942
• convert: The plugin now has an id3v23 option that allows you to override the global id3v23 option. Thanks to @Holzhaus. #3104
• spotify:
  • The plugin now uses OAuth for authentication to the Spotify API. Thanks to @rhlahuja. #2694 #3123
  • The plugin now works as an import metadata provider: you can match tracks and albums using the Spotify database. Thanks to @rhlahuja. #3123
• ipfs: The plugin now supports a nocopy option which passes that flag to ipfs. Thanks to @wildthyme.
• discogs: The plugin now has rate limiting for the Discogs API. #3081
• mpdstats, mpdupdate: These plugins now use the MPD_PORT environment variable if no port is specified in the configuration file. #3223
• bpd:
  • MPD protocol commands consume and single are now supported along with updated semantics for repeat and previous and new fields for status. The bpd server now understands and ignores some additional commands. #3200 #800
  • MPD protocol command idle is now supported, allowing the MPD version to be bumped to 0.14. #3205 #800
  • MPD protocol command decoders is now supported. #3222
  • The plugin now uses the main beets logging system. The special-purpose --debug flag has been removed. Thanks to @arcresu. #3196
• mbsync: The plugin no longer queries MusicBrainz when either the mb_albumid or mb_trackid field is invalid. See also the discussion on Google Groups Thanks to @arogl.
• export: The plugin now also exports path field if the user explicitly specifies it with -i parameter. This only works when exporting library fields. #3084
• acousticbrainz: The plugin now declares types for all its fields, which enables easier querying and avoids a problem where very small numbers would be stored as strings. Thanks to @rain0r. #2790 #3238

Some improvements have been focused on improving beets' performance:

• Querying the library is now faster:
  • We only convert fields that need to be displayed. Thanks to @pprkut. #3089
  • We now compile templates once and reuse them instead of recompiling them to print out each matching object. Thanks to @simonpersson. #3258
  • Querying the library for items is now faster, for all queries that do not need to access album level properties. This was implemented by lazily fetching the album only when needed. Thanks to @simonpersson. #3260
• absubmit, badfiles: Analysis now works in parallel (on Python 3 only). Thanks to @bemeurer. #2442 #3003
• mpdstats: Use the currentsong MPD command instead of playlist to get the current song, improving performance when the playlist is long. Thanks to @ray66. #3207 #2752

Several improvements are related to usability:

• The disambiguation string for identifying albums in the importer now shows the catalog number. Thanks to @8h2a. #2951
• Added whitespace padding to missing tracks dialog to improve readability. Thanks to @jams2. #2962
• The move command now lists the number of items already in-place. Thanks to @RollingStar. #3117
• Modify selection can now be applied early without selecting every item. #3083
• Beets now emits more useful messages during startup if SQLite returns an error. The SQLite error message is now attached to the beets message. #3005
• Fixed a confusing typo when the convert plugin copies the art covers. #3063

Many fixes have been focused on issues where beets would previously crash:

• Avoid a crash when archive extraction fails during import. #3041
• Missing album art file during an update no longer causes a fatal exception (instead, an error is logged and the missing file path is removed from the library). #3030
• When updating the database, beets no longer tries to move album art twice. #3189
• Fix an unhandled exception when pruning empty directories. #1996 #3209
• fetchart: Added network connection error handling to backends so that beets won't crash if a request fails. Thanks to @Holzhaus. #1579
• badfiles: Avoid a crash when the underlying tool emits undecodable output. #3165
• beatport: Avoid a crash when the server produces an error. #3184
• bpd: Fix crashes in the bpd server during exception handling. #3200
• bpd: Fix a crash triggered when certain clients tried to list the albums belonging to a particular artist. #3007 #3215
• replaygain: Avoid a crash when the bs1770gain tool emits malformed XML. #2983 #3247

There are many fixes related to compatibility with our dependencies including addressing changes interfaces:

• On Python 2, pin the jellyfish requirement to version 0.6.0 for compatibility.
• Fix compatibility with Python 3.7 and its change to a name in the re module. #2978
• Fix several uses of deprecated standard-library features on Python 3.7. Thanks to @arcresu. #3197
• Fix compatibility with pre-release versions of Python 3.8. #3201 #3202
• web: Fix an error when using more recent versions of Flask with CORS enabled. Thanks to @rveachkc. #2979: #2980
• Avoid some deprecation warnings with certain versions of the MusicBrainz library. Thanks to @zhelezov. #2826 #3092
• Restore iTunes Store album art source, and remove the dependency on python-itunes, which had gone unmaintained and was not Python-3-compatible. Thanks to @ocelma for creating python-itunes in the first place. Thanks to @nathdwek. #2371 #2551 #2718
• lastgenre, edit: Avoid a deprecation warnings from the PyYAML library by switching to the safe loader. Thanks to @translit and @sbraz. #3192 #3225
• Fix a problem when resizing images with PIL/pillow on Python 3. Thanks to @architek. #2504 #3029

And there are many other fixes:

• R128 normalization tags are now properly deleted from files when the values are missing. Thanks to @autrimpo. #2757
• Display the artist credit when matching albums if the artist_credit configuration option is set. #2953
• With the from_scratch configuration option set, only writable fields are cleared. Beets now no longer ignores the format your music is saved in. #2972
• The %aunique template function now works correctly with the -f/--format option. #3043
• Fixed the ordering of items when manually selecting changes while updating tags Thanks to @TaizoSimpson. #3501
• The %title template function now works correctly with apostrophes. Thanks to @GuilhermeHideki. #3033
• lastgenre: It's now possible to set the prefer_specific option without also setting canonical. #2973
• fetchart: The plugin now respects the ignore and ignore_hidden settings. #1632
• hook: Fix byte string interpolation in hook commands. #2967 #3167
• the: Log a message when something has changed, not when it hasn't. Thanks to @arcresu. #3195
• lastgenre: The force config option now actually works. #2704 #3054
• Resizing image files with ImageMagick now avoids problems on systems where there is a convert command that is not ImageMagick's by using the magick executable when it is available. Thanks to @ababyduck. #2093 #3236

There is one new thing for plugin developers to know about:

• In addition to prefix-based field queries, plugins can now define named queries that are not associated with any specific field. For example, the new playlist supports queries like playlist:name although there is no field named playlist. See extend-query for details.

And some messages for packagers:

• Note the changes to the dependencies on jellyfish and munkres.
• The optional python-itunes dependency has been removed.
• Python ...

beets 1.4.7

This new release includes lots of new features in the importer and the metadata source backends that it uses. We've changed how the beets importer handles non-audio tracks listed in metadata sources like MusicBrainz:

• The importer now ignores non-audio tracks (namely, data and video tracks) listed in MusicBrainz. Also, a new option, ignore_video_tracks, lets you return to the old behavior and include these video tracks. #1210
• A new importer option, ignored_media, can let you skip certain media formats. #2688

There are other subtle improvements to metadata handling in the importer:

• In the MusicBrainz backend, beets now imports the musicbrainz_releasetrackid field. This is a first step toward #406. Thanks to @Rawrmonkeys.
• A new importer configuration option, artist_credit, will tell beets to prefer the artist credit over the artist when autotagging. #1249

And there are even more new features:

• replaygain: The beet replaygain command now has --force, --write and --nowrite options. #2778
• A new importer configuration option, incremental_skip_later, lets you avoid recording skipped directories to the list of "processed" directories in incremental mode. This way, you can revisit them later with another import. Thanks to @sekjun9878. #2773
• fetchart: The configuration options now support finer-grained control via the sources option. You can now specify the search order for different matching strategies within different backends.
• web: A new cors_supports_credentials configuration option lets in-browser clients communicate with the server even when it is protected by an authorization mechanism (a proxy with HTTP authentication enabled, for example).
• A new sonosupdate plugin automatically notifies Sonos controllers to update the music library when the beets library changes. Thanks to @cgtobi.
• discogs: The plugin now stores master release IDs into mb_releasegroupid. It also "simulates" track IDs using the release ID and the track list position. Thanks to @dbogdanov. #2336
• discogs: Fetch the original year from master releases. #1122

There are lots and lots of fixes:

• replaygain: Fix a corner-case with the bs1770gain backend where ReplayGain values were assigned to the wrong files. The plugin now requires version 0.4.6 or later of the bs1770gain tool. #2777
• lyrics: The plugin no longer crashes in the Genius source when BeautifulSoup is not found. Instead, it just logs a message and disables the source. #2911
• lyrics: Handle network and API errors when communicating with Genius. #2771
• lyrics: The lyrics command previously wrote ReST files by default, even when you didn't ask for them. This default has been fixed.
• lyrics: When writing ReST files, the lyrics command now groups lyrics by the albumartist field, rather than artist. #2924
• Plugins can now see updated import task state, such as when rejecting the initial candidates and finding new ones via a manual search. Notably, this means that the importer prompt options that the edit provides show up more reliably after doing a secondary import search. #2441 #2731
• importadded: Fix a crash on non-autotagged imports. Thanks to @m42i. #2601 #1918
• plexupdate: The Plex token is now redacted in configuration output. Thanks to @Kovrinic. #2804
• Avoid a crash when importing a non-ASCII filename when using an ASCII locale on Unix under Python 3. #2793 #2803
• Fix a problem caused by time zone misalignment that could make date queries fail to match certain dates that are near the edges of a range. For example, querying for dates within a certain month would fail to match dates within hours of the end of that month. #2652
• convert: The plugin now runs before other plugin-provided import stages, which addresses an issue with generating ReplayGain data incompatible between the source and target file formats. Thanks to @autrimpo. #2814
• ftintitle: The drop config option had no effect; it now does what it says it should do. #2817
• Importing a release with multiple release events now selects the event based on the order of your preferred countries rather than the order of release events in MusicBrainz. #2816
• web: The time display in the web interface would incorrectly jump at the 30-second mark of every minute. Now, it correctly changes over at zero seconds. #2822
• web: Fetching album art now works (instead of throwing an exception) under Python 3. Additionally, the server will now return a 404 response when the album ID is unknown (instead of throwing an exception and producing a 500 response). #2823
• web: Fix an exception on Python 3 for filenames with non-Latin1 characters. (These characters are now converted to their ASCII equivalents.) #2815
• Partially fix bash completion for subcommand names that contain hyphens. Thanks to @jhermann. #2836 #2837
• replaygain: Really fix album gain calculation using the GStreamer backend. #2846
• Avoid an error when doing a "no-op" move on non-existent files (i.e., moving a file onto itself). #2863
• discogs: Fix the medium and medium_index values, which were occasionally incorrect for releases with two-sided mediums such as vinyl. Also fix the medium_total value, which now contains total number of tracks on the medium to which a track belongs, not the total number of different mediums present on the release. Thanks to @dbogdanov. #2887
• The importer now supports audio files contained in data tracks when they are listed in MusicBrainz: the corresponding audio tracks are now merged into the main track list. Thanks to @jdetrey. #1638
• keyfinder: Avoid a crash when trying to process unmatched tracks. #2537
• mbsync: Support MusicBrainz recording ID changes, relying on release track IDs instead. Thanks to @jdetrey. #1234
• mbsync: We can now successfully update albums even when the first track has a missing MusicBrainz recording ID. #2920

There are a couple of changes for developers:

• Plugins can now run their import stages early, before other plugins. Use the early_import_stages list instead of plain import_stages to request this behavior. #2814
• We again properly send albuminfo_received and trackinfo_received in all cases, most notably when using the mbsync plugin. This was a regression since version 1.4.1. #2921