Improve Filesystem provider (#953)

Author: marcelveldt

music_assistant/common/models/enums.py

@@ -1,45 +1,11 @@
 """All enums used by the Music Assistant models."""
 from __future__ import annotations
 
-from enum import Enum
-from typing import Any, TypeVar
-
-# pylint:disable=ungrouped-imports
-try:
-    from enum import StrEnum
-except (AttributeError, ImportError):
-    # Python 3.10 compatibility for strenum
-    _StrEnumSelfT = TypeVar("_StrEnumSelfT", bound="StrEnum")
-
-    class StrEnum(str, Enum):
-        """Partial backport of Python 3.11's StrEnum for our basic use cases."""
-
-        def __new__(
-            cls: type[_StrEnumSelfT], value: str, *args: Any, **kwargs: Any
-        ) -> _StrEnumSelfT:
-            """Create a new StrEnum instance."""
-            if not isinstance(value, str):
-                raise TypeError(f"{value!r} is not a string")
-            return super().__new__(cls, value, *args, **kwargs)
-
-        def __str__(self) -> str:
-            """Return self."""
-            return str(self)
-
-        @staticmethod
-        def _generate_next_value_(
-            name: str, start: int, count: int, last_values: list[Any]  # noqa
-        ) -> Any:
-            """Make `auto()` explicitly unsupported.
-
-            We may revisit this when it's very clear that Python 3.11's
-            `StrEnum.auto()` behavior will no longer change.
-            """
-            raise TypeError("auto() is not supported by this implementation")
+from enum import StrEnum
 
 
 class MediaType(StrEnum):
-    """StrEnum for MediaType."""
+    """Enum for MediaType."""
 
     ARTIST = "artist"
     ALBUM = "album"
@@ -62,8 +28,24 @@ def ALL(cls) -> tuple[MediaType, ...]:  # noqa: N802
         )
 
 
+class ExternalID(StrEnum):
+    """Enum with External ID types."""
+
+    # musicbrainz:
+    # for tracks this is the RecordingID
+    # for albums this is the ReleaseGroupID (NOT the release ID!)
+    # for artists this is the ArtistID
+    MUSICBRAINZ = "musicbrainz"
+    ISRC = "isrc"  # used to identify unique recordings
+    BARCODE = "barcode"  # EAN-13 barcode for identifying albums
+    ACOUSTID = "acoustid"  # unique fingerprint (id) for a recording
+    ASIN = "asin"  # amazon unique number to identify albums
+    DISCOGS = "discogs"  # id for media item on discogs
+    TADB = "tadb"  # the audio db id
+
+
 class LinkType(StrEnum):
-    """StrEnum with link types."""
+    """Enum with link types."""
 
     WEBSITE = "website"
     FACEBOOK = "facebook"
@@ -79,7 +61,7 @@ class LinkType(StrEnum):
 
 
 class ImageType(StrEnum):
-    """StrEnum with image types."""
+    """Enum with image types."""
 
     THUMB = "thumb"
     LANDSCAPE = "landscape"
@@ -94,7 +76,7 @@ class ImageType(StrEnum):
 
 
 class AlbumType(StrEnum):
-    """StrEnum for Album type."""
+    """Enum for Album type."""
 
     ALBUM = "album"
     SINGLE = "single"
@@ -182,7 +164,7 @@ def from_bit_depth(cls, bit_depth: int, floating_point: bool = False) -> Content
 
 
 class QueueOption(StrEnum):
-    """StrEnum representation of the queue (play) options.
+    """Enum representation of the queue (play) options.
 
     - PLAY -> Insert new item(s) in queue at the current position and start playing.
     - REPLACE -> Replace entire queue contents with the new items and start playing from index 0.
@@ -207,7 +189,7 @@ class RepeatMode(StrEnum):
 
 
 class PlayerState(StrEnum):
-    """StrEnum for the (playback)state of a player."""
+    """Enum for the (playback)state of a player."""
 
     IDLE = "idle"
     PAUSED = "paused"
@@ -323,7 +305,6 @@ class ProviderFeature(StrEnum):
     ARTIST_METADATA = "artist_metadata"
     ALBUM_METADATA = "album_metadata"
     TRACK_METADATA = "track_metadata"
-    GET_ARTIST_MBID = "get_artist_mbid"
 
     #
     # PLUGIN FEATURES

music_assistant/common/models/media_items.py

@@ -3,7 +3,7 @@
 
 from dataclasses import dataclass, field, fields
 from time import time
-from typing import Any
+from typing import Any, Self
 
 from mashumaro import DataClassDictMixin
 
@@ -12,6 +12,7 @@
 from music_assistant.common.models.enums import (
     AlbumType,
     ContentType,
+    ExternalID,
     ImageType,
     LinkType,
     MediaType,
@@ -66,10 +67,6 @@ class ProviderMapping(DataClassDictMixin):
     audio_format: AudioFormat = field(default_factory=AudioFormat)
     # url = link to provider details page if exists
     url: str | None = None
-    # isrc (tracks only) - isrc identifier if known
-    isrc: str | None = None
-    # barcode (albums only) - barcode identifier if known
-    barcode: str | None = None
     # optional details to store provider specific details
     details: str | None = None
 
@@ -154,14 +151,12 @@ class MediaItemMetadata(DataClassDictMixin):
     mood: str | None = None
     style: str | None = None
     copyright: str | None = None
-    lyrics: str | None = None
-    ean: str | None = None
+    lyrics: str | None = None  # tracks only
     label: str | None = None
     links: set[MediaItemLink] | None = None
     chapters: list[MediaItemChapter] | None = None
     performers: set[str] | None = None
     preview: str | None = None
-    replaygain: float | None = None
     popularity: int | None = None
     # last_refresh: timestamp the (full) metadata was last collected
     last_refresh: int | None = None
@@ -204,11 +199,10 @@ class MediaItem(DataClassDictMixin):
     item_id: str
     provider: str  # provider instance id or provider domain
     name: str
-    metadata: MediaItemMetadata
     provider_mappings: set[ProviderMapping]
 
     # optional fields below
-    # provider_mappings: set[ProviderMapping] = field(default_factory=set)
+    external_ids: set[tuple[ExternalID, str]] = field(default_factory=set)
     metadata: MediaItemMetadata = field(default_factory=MediaItemMetadata)
     favorite: bool = False
     media_type: MediaType = MediaType.UNKNOWN
@@ -238,14 +232,55 @@ def image(self) -> MediaItemImage | None:
             return None
         return next((x for x in self.metadata.images if x.type == ImageType.THUMB), None)
 
+    @property
+    def mbid(self) -> str | None:
+        """Return MusicBrainz ID."""
+        return self.get_external_id(ExternalID.MUSICBRAINZ)
+
+    @mbid.setter
+    def mbid(self, value: str) -> None:
+        """Set MusicBrainz External ID."""
+        if not value:
+            return
+        if len(value.split("-")) != 5:
+            raise RuntimeError("Invalid MusicBrainz identifier")
+        if existing := next((x for x in self.external_ids if x[0] == ExternalID.MUSICBRAINZ), None):
+            # Musicbrainz ID is unique so remove existing entry
+            self.external_ids.remove(existing)
+        self.external_ids.add((ExternalID.MUSICBRAINZ, value))
+
+    def get_external_id(self, external_id_type: ExternalID) -> str | None:
+        """Get (the first instance) of given External ID or None if not found."""
+        for ext_id in self.external_ids:
+            if ext_id[0] != external_id_type:
+                continue
+            return ext_id[1]
+        return None
+
     def __hash__(self) -> int:
         """Return custom hash."""
         return hash(self.uri)
 
-    def __eq__(self, other: ItemMapping) -> bool:
+    def __eq__(self, other: MediaItem | ItemMapping) -> bool:
         """Check equality of two items."""
         return self.uri == other.uri
 
+    @classmethod
+    def from_item_mapping(cls: type, item: ItemMapping) -> Self:
+        """Instantiate MediaItem from ItemMapping."""
+        # NOTE: This will not work for albums and tracks!
+        return cls.from_dict(
+            {
+                **item.to_dict(),
+                "provider_mappings": {
+                    "item_id": item.item_id,
+                    "provider_domain": item.provider,
+                    "provider_instance": item.provider,
+                    "available": item.available,
+                },
+            }
+        )
+
 
 @dataclass(kw_only=True)
 class ItemMapping(DataClassDictMixin):
@@ -259,13 +294,12 @@ class ItemMapping(DataClassDictMixin):
     sort_name: str | None = None
     uri: str | None = None
     available: bool = True
+    external_ids: set[tuple[ExternalID, str]] = field(default_factory=set)
 
     @classmethod
     def from_item(cls, item: MediaItem):
         """Create ItemMapping object from regular item."""
-        result = cls.from_dict(item.to_dict())
-        result.available = item.available
-        return result
+        return cls.from_dict(item.to_dict())
 
     def __post_init__(self):
         """Call after init."""
@@ -290,7 +324,6 @@ class Artist(MediaItem):
     """Model for an artist."""
 
     media_type: MediaType = MediaType.ARTIST
-    mbid: str | None = None
 
 
 @dataclass(kw_only=True)
@@ -302,7 +335,6 @@ class Album(MediaItem):
     year: int | None = None
     artists: list[Artist | ItemMapping] = field(default_factory=list)
     album_type: AlbumType = AlbumType.UNKNOWN
-    mbid: str | None = None  # release group id
 
 
 @dataclass(kw_only=True)
@@ -312,7 +344,6 @@ class Track(MediaItem):
     media_type: MediaType = MediaType.TRACK
     duration: int = 0
     version: str = ""
-    mbid: str | None = None  # Recording ID
     artists: list[Artist | ItemMapping] = field(default_factory=list)
     album: Album | ItemMapping | None = None  # optional
 

music_assistant/constants.py

@@ -5,7 +5,7 @@
 
 API_SCHEMA_VERSION: Final[int] = 23
 MIN_SCHEMA_VERSION: Final[int] = 23
-DB_SCHEMA_VERSION: Final[int] = 25
+DB_SCHEMA_VERSION: Final[int] = 26
 
 ROOT_LOGGER_NAME: Final[str] = "music_assistant"
 

music_assistant/server/controllers/media/albums.py

@@ -102,10 +102,26 @@ async def add_item_to_library(
         # grab additional metadata
         if metadata_lookup:
             await self.mass.metadata.get_album_metadata(item)
-        # actually add (or update) the item in the library db
-        # use the lock to prevent a race condition of the same item being added twice
-        async with self._db_add_lock:
-            library_item = await self._add_library_item(item)
+        # check for existing item first
+        library_item = None
+        if cur_item := await self.get_library_item_by_prov_id(item.item_id, item.provider):
+            # existing item match by provider id
+            library_item = await self.update_item_in_library(cur_item.item_id, item)  # noqa: SIM114
+        elif cur_item := await self.get_library_item_by_external_ids(item.external_ids):
+            # existing item match by external id
+            library_item = await self.update_item_in_library(cur_item.item_id, item)
+        else:
+            # search by name
+            async for db_item in self.iter_library_items(search=item.name):
+                if compare_album(db_item, item):
+                    # existing item found: update it
+                    library_item = await self.update_item_in_library(db_item.item_id, item)
+                    break
+        if not library_item:
+            # actually add a new item in the library db
+            # use the lock to prevent a race condition of the same item being added twice
+            async with self._db_add_lock:
+                library_item = await self._add_library_item(item)
         # also fetch the same album on all providers
         if metadata_lookup:
             await self._match(library_item)
@@ -139,6 +155,7 @@ async def update_item_in_library(
         else:
             album_type = cur_item.album_type
         sort_artist = album_artists[0].sort_name
+        cur_item.external_ids.update(update.external_ids)
         await self.mass.music.database.update(
             self.db_table,
             {"item_id": db_id},
@@ -152,7 +169,9 @@ async def update_item_in_library(
                 "artists": serialize_to_json(album_artists),
                 "metadata": serialize_to_json(metadata),
                 "provider_mappings": serialize_to_json(provider_mappings),
-                "mbid": update.mbid or cur_item.mbid,
+                "external_ids": serialize_to_json(
+                    update.external_ids if overwrite else cur_item.external_ids
+                ),
                 "timestamp_modified": int(utc_timestamp()),
             },
         )
@@ -219,28 +238,8 @@ async def versions(
 
     async def _add_library_item(self, item: Album) -> Album:
         """Add a new record to the database."""
-        # safety guard: check for existing item first
-        if cur_item := await self.get_library_item_by_prov_id(item.item_id, item.provider):
-            # existing item found: update it
-            return await self.update_item_in_library(cur_item.item_id, item)
-        if item.mbid:
-            match = {"mbid": item.mbid}
-            if db_row := await self.mass.music.database.get_row(self.db_table, match):
-                cur_item = Album.from_dict(self._parse_db_row(db_row))
-                # existing item found: update it
-                return await self.update_item_in_library(cur_item.item_id, item)
-        # fallback to search and match
-        match = {"sort_name": item.sort_name}
-        for db_row in await self.mass.music.database.get_rows(self.db_table, match):
-            row_album = Album.from_dict(self._parse_db_row(db_row))
-            if compare_album(row_album, item):
-                cur_item = row_album
-                # existing item found: update it
-                return await self.update_item_in_library(cur_item.item_id, item)
-
-        # insert new item
-        album_artists = await self._get_artist_mappings(item, cur_item)
-        sort_artist = album_artists[0].sort_name
+        album_artists = await self._get_artist_mappings(item)
+        sort_artist = album_artists[0].sort_name if album_artists else ""
         new_item = await self.mass.music.database.insert(
             self.db_table,
             {
@@ -250,11 +249,11 @@ async def _add_library_item(self, item: Album) -> Album:
                 "favorite": item.favorite,
                 "album_type": item.album_type,
                 "year": item.year,
-                "mbid": item.mbid,
                 "metadata": serialize_to_json(item.metadata),
                 "provider_mappings": serialize_to_json(item.provider_mappings),
                 "artists": serialize_to_json(album_artists),
                 "sort_artist": sort_artist,
+                "external_ids": serialize_to_json(item.external_ids),
                 "timestamp_added": int(utc_timestamp()),
                 "timestamp_modified": int(utc_timestamp()),
             },