Added duduping on mb_trackid to the duplicates plugin

beetsplug/duplicates.py

@@ -17,9 +17,10 @@
 import os
 import shlex
 
+from beets.dbcore.query import MatchQuery
 from beets.library import Album, Item
 from beets.plugins import BeetsPlugin
-from beets.ui import Subcommand, UserError, print_
+from beets.ui import Subcommand, UserError, colorize, print_
 from beets.util import (
     MoveOperation,
     bytestring_path,
@@ -141,6 +142,49 @@ def __init__(self):
         )
         self._command.parser.add_all_common_options()
 
+        self.register_listener("import_task_created", self.import_task_created)
+
+    def import_task_created(self, task, session):
+        if "dedupe_mb_trackid_on_import" in self.config and self.config[
+            "dedupe_mb_trackid_on_import"
+        ].get(bool):
+            return self._dedupe_task_on_mb_trackid(task, session)
+
+    def _dedupe_task_on_mb_trackid(self, task, session):
+        # Find all items that already have the same track imported
+        dupes = []
+        for item in task.items:
+            if not item.mb_trackid:
+                continue
+
+            # Query the library for any matching tracks
+            resp = session.lib.items(
+                query=MatchQuery("mb_trackid", item.mb_trackid)
+            )
+            if len(resp.rows) > 0:
+                log_prefix = f"{item.artist} - {item.album} - {item.title}"
+                print_(
+                    f"{colorize('text_warning', log_prefix)}:"
+                    "Item already imported, skipping..."
+                )
+                dupes.append(item)
+
+        # Remove the dupes
+        album = ""
+        for dup in dupes:
+            album = f"{dup.artist} - {dup.album}"
+            task.items.remove(dup)
+
+        # Get rid of the task if all items were removed
+        if len(task.items) == 0:
+            print_(
+                f"{colorize('text_warning', album)}:" if album != "" else "",
+                "All items removed due to duplicates, removing task",
+            )
+            return []
+
+        return [task]
+
     def commands(self):
         def _dup(lib, opts, args):
             self.config.set_args(opts)

beetsplug/tidal/__init__.py

@@ -355,7 +355,6 @@ def _get_album_info(
         track_by_id: dict[str, TidalTrack],
         artist_by_id: dict[str, TidalArtist],
     ) -> AlbumInfo:
-
         track_infos: list[TrackInfo] = []
         for i, track_rel in enumerate(
             album["relationships"]["items"]["data"], start=1

docs/changelog.rst

@@ -39,6 +39,8 @@ New features
   See :doc:`plugins/tidal` for more information.
 
 - Add support for adding or modifying a subtitle (ID3 tag ``TIT3``) field
+- :doc:`plugins/duplicates`: ``dedupe_mb_trackid_on_import`` option added to
+  deduplicate during imports based on already-imported ``mb_trackid``.
 
 Bug fixes
 ~~~~~~~~~

docs/plugins/duplicates.rst

@@ -36,6 +36,11 @@ duplicates or just the duplicates themselves via command-line switches
     -t TAG, --tag=TAG     tag matched items with 'k=v' attribute
     -r, --remove          remove items from library
 
+Generally, beets operates on Albums or Track singletons. As a way to bridge the
+gap, the ``duplicates`` plugin can be configured to duduplicate tracks on import
+by checking for already-imported items with the same ``mb_trackid``. See the
+``dedupe_mb_trackid_on_import`` option for details.
+
 Configuration
 -------------
 
@@ -54,6 +59,8 @@ file. The available options mirror the command-line options:
 - **count**: Print a count of duplicate tracks or albums in the format
   ``$albumartist - $album - $title: $count`` (for tracks) or ``$albumartist -
   $album: $count`` (for albums). Default: ``no``.
+- **dedupe_mb_trackid_on_import**: Deduplicate album tracks when an
+  already-imported track has the same mb_trackid.
 - **delete**: Remove matched items from the library and from the disk. Default:
   ``no``
 - **format**: A specific format with which to print every track or album. This