Fix mb search term formatting (#6354)

Fixes #6347

- Fixed MusicBrainz Lucene query formatting in
`MusicBrainzAPI.format_search_term()` (lowercase + trim + escape Lucene
special chars).
- Fixed `plugins.musicbrainz:extra_tags` support by mapping `alias` and
`tracks` into MusicBrainz search fields.
- Adjusted logging to make MusicBrainz API logs visible under the shared
`beets` logger (and removed an unused per-module logger in
`beetsplug.bpd`).

Author: snejus

.github/CODEOWNERS

@@ -3,5 +3,11 @@
 
 # Specific ownerships:
 /beets/metadata_plugins.py @semohr
+
 /beetsplug/titlecase.py @henry-oberholtzer
+
 /beetsplug/mbpseudo.py @asardaes
+
+/beetsplug/_utils/requests.py @snejus
+/beetsplug/_utils/musicbrainz.py @snejus
+/beetsplug/musicbrainz.py @snejus

beetsplug/_utils/musicbrainz.py

@@ -11,9 +11,10 @@
 from __future__ import annotations
 
 import operator
+import re
 from dataclasses import dataclass, field
 from functools import cached_property, singledispatchmethod, wraps
-from itertools import groupby
+from itertools import groupby, starmap
 from typing import TYPE_CHECKING, Any, Literal, ParamSpec, TypedDict, TypeVar
 
 from requests_ratelimiter import LimiterMixin
@@ -30,7 +31,10 @@
 
     from .._typing import JSONDict
 
-log = logging.getLogger(__name__)
+log = logging.getLogger("beets")
+
+
+LUCENE_SPECIAL_CHAR_PAT = re.compile(r'([-+&|!(){}[\]^"~*?:\\/])')
 
 
 class LimiterTimeoutSession(LimiterMixin, TimeoutAndRetrySession):
@@ -181,6 +185,21 @@ def _lookup(
     def _browse(self, entity: Entity, **kwargs) -> list[JSONDict]:
         return self._get_resource(entity, **kwargs).get(f"{entity}s", [])
 
+    @staticmethod
+    def format_search_term(field: str, term: str) -> str:
+        """Format a search term for the MusicBrainz API.
+
+        See https://lucene.apache.org/core/4_3_0/queryparser/org/apache/lucene/queryparser/classic/package-summary.html
+        """
+        if not (term := term.lower().strip()):
+            return ""
+
+        term = LUCENE_SPECIAL_CHAR_PAT.sub(r"\\\1", term)
+        if field:
+            term = f"{field}:({term})"
+
+        return term
+
     def search(
         self,
         entity: Entity,
@@ -195,10 +214,8 @@ def search(
           - 'value' is empty, in which case the filter is ignored
         * Values are lowercased and stripped of whitespace.
         """
-        query = " AND ".join(
-            ":".join(filter(None, (k, f'"{_v}"')))
-            for k, v in filters.items()
-            if (_v := v.lower().strip())
+        query = " ".join(
+            filter(None, starmap(self.format_search_term, filters.items()))
         )
         log.debug("Searching for MusicBrainz {}s with: {!r}", entity, query)
         kwargs["query"] = query

beetsplug/bpd/__init__.py

@@ -30,7 +30,7 @@
 
 import beets
 import beets.ui
-from beets import dbcore, logging
+from beets import dbcore
 from beets.library import Item
 from beets.plugins import BeetsPlugin
 from beets.util import as_string, bluelet
@@ -39,8 +39,6 @@
 if TYPE_CHECKING:
     from beets.dbcore.query import Query
 
-log = logging.getLogger(__name__)
-
 
 try:
     from . import gstplayer

beetsplug/musicbrainz.py

@@ -56,6 +56,8 @@
     "label": "label",
     "media": "format",
     "year": "date",
+    "tracks": "tracks",
+    "alias": "alias",
 }
 
 

docs/changelog.rst

@@ -11,6 +11,10 @@ New features:
 
 Bug fixes:
 
+- :doc:`plugins/musicbrainz`: Fix search terms escaping. :bug:`6347`
+- :doc:`plugins/musicbrainz`: Fix support for ``alias`` and ``tracks``
+  :conf:`plugins.musicbrainz:extra_tags`.
+
 For packagers:
 
 Other changes:

docs/plugins/index.rst

@@ -100,15 +100,15 @@ databases. They share the following configuration options:
     listenbrainz
     loadext
     lyrics
-    mbcollection
-    mbpseudo
-    mbsubmit
     mbsync
     metasync
     missing
     mpdstats
     mpdupdate
     musicbrainz
+    mbcollection
+    mbpseudo
+    mbsubmit
     parentwork
     permissions
     play

docs/plugins/musicbrainz.rst

@@ -93,15 +93,23 @@ Default
     This setting should improve the autotagger results if the metadata with the
     given tags match the metadata returned by MusicBrainz.
 
-    Note that the only tags supported by this setting are: ``barcode``,
-    ``catalognum``, ``country``, ``label``, ``media``, and ``year``.
+    Tags supported by this setting:
+
+    * ``alias`` (also search for release aliases matching the query)
+    * ``barcode``
+    * ``catalognum``
+    * ``country``
+    * ``label``
+    * ``media``
+    * ``tracks`` (number of tracks on the release)
+    * ``year``
 
     Example:
 
     .. code-block:: yaml
 
         musicbrainz:
-            extra_tags: [barcode, catalognum, country, label, media, year]
+            extra_tags: [alias, barcode, catalognum, country, label, media, tracks, year]
 
 .. conf:: genres
     :default: no

test/plugins/utils/test_musicbrainz.py

@@ -1,3 +1,5 @@
+import pytest
+
 from beetsplug._utils.musicbrainz import MusicBrainzAPI
 
 
@@ -80,3 +82,15 @@ def test_group_relations():
             },
         ],
     }
+
+
+@pytest.mark.parametrize(
+    "field, term, expected",
+    [
+        ("artist", '  AC/DC + "[Live]"  ', r"artist:(ac\/dc \+ \"\[live\]\")"),
+        ("", "Foo:Bar", r"foo\:bar"),
+        ("artist", "   ", ""),
+    ],
+)
+def test_format_search_term(field, term, expected):
+    assert MusicBrainzAPI.format_search_term(field, term) == expected