Implement Qobuz filters (#529)

* Implement artist filters

* Add warning message for filtering other sources

streamrip/config.toml

@@ -108,6 +108,7 @@ bit_depth = 24
 lossy_bitrate = 320
 
 # Filter a Qobuz artist's discography. Set to 'true' to turn on a filter.
+# This will also be applied to other sources, but is not guaranteed to work correctly
 [qobuz_filters]
 # Remove Collectors Editions, live recordings, etc.
 extras = false

streamrip/media/album.py

@@ -50,7 +50,6 @@ class PendingAlbum(Pending):
 
     async def resolve(self) -> Album | None:
         resp = await self.client.get_metadata(self.id, "album")
-
         meta = AlbumMetadata.from_album_resp(resp, self.client.source)
         if meta is None:
             logger.error(

streamrip/media/album_list.py

@@ -1,48 +0,0 @@
-import asyncio
-from dataclasses import dataclass
-
-from ..client import Client
-from ..config import Config
-from .album import PendingAlbum
-from .media import Media
-
-
-@dataclass(slots=True)
-class AlbumList(Media):
-    """Represents a list of albums. Used by Artist and Label classes."""
-
-    name: str
-    albums: list[PendingAlbum]
-    client: Client
-    config: Config
-
-    async def preprocess(self):
-        pass
-
-    async def download(self):
-        # Resolve only 3 albums at a time to avoid
-        # initial latency of resolving ALL albums and tracks
-        # before any downloads
-        album_resolve_chunk_size = 10
-
-        async def _resolve_download(item: PendingAlbum):
-            album = await item.resolve()
-            if album is None:
-                return
-            await album.rip()
-
-        batches = self.batch(
-            [_resolve_download(album) for album in self.albums],
-            album_resolve_chunk_size,
-        )
-        for batch in batches:
-            await asyncio.gather(*batch)
-
-    async def postprocess(self):
-        pass
-
-    @staticmethod
-    def batch(iterable, n=1):
-        total = len(iterable)
-        for ndx in range(0, total, n):
-            yield iterable[ndx : min(ndx + n, total)]

streamrip/media/artist.py

@@ -1,16 +1,176 @@
+import asyncio
+import logging
+import re
 from dataclasses import dataclass
 
 from ..client import Client
-from ..config import Config
+from ..config import Config, QobuzDiscographyFilterConfig
+from ..console import console
 from ..db import Database
 from ..metadata import ArtistMetadata
-from .album import PendingAlbum
-from .album_list import AlbumList
-from .media import Pending
+from .album import Album, PendingAlbum
+from .media import Media, Pending
+
+logger = logging.getLogger("streamrip")
+
+# Resolve only N albums at a time to avoid
+# initial latency of resolving ALL albums and tracks
+# before any downloads
+RESOLVE_CHUNK_SIZE = 10
 
 
-class Artist(AlbumList):
-    pass
+@dataclass(slots=True)
+class Artist(Media):
+    """Represents a list of albums. Used by Artist and Label classes."""
+
+    name: str
+    albums: list[PendingAlbum]
+    client: Client
+    config: Config
+
+    async def preprocess(self):
+        pass
+
+    async def download(self):
+        filter_conf = self.config.session.qobuz_filters
+        if filter_conf.repeats:
+            console.log(
+                "Resolving [purple]ALL[/purple] artist albums to detect repeats. This may take a while."
+            )
+            await self._resolve_then_download(filter_conf)
+        else:
+            await self._download_async(filter_conf)
+
+    async def postprocess(self):
+        pass
+
+    async def _resolve_then_download(self, filters: QobuzDiscographyFilterConfig):
+        """Resolve all artist albums, then download.
+
+        This is used if the repeat filter is turned on, since we need the titles
+        of all albums to remove repeated items.
+        """
+        resolved_or_none: list[Album | None] = await asyncio.gather(
+            *[album.resolve() for album in self.albums]
+        )
+        resolved = [a for a in resolved_or_none if a is not None]
+        filtered_albums = self._apply_filters(resolved, filters)
+        batches = self.batch([a.rip() for a in filtered_albums], RESOLVE_CHUNK_SIZE)
+        for batch in batches:
+            await asyncio.gather(*batch)
+
+    async def _download_async(self, filters: QobuzDiscographyFilterConfig):
+        async def _rip(item: PendingAlbum):
+            album = await item.resolve()
+            # Skip if album doesn't pass the filter
+            if (
+                album is None
+                or (filters.extras and not self._extras(album))
+                or (filters.features and not self._features(album))
+                or (filters.non_studio_albums and not self._non_studio_albums(album))
+                or (filters.non_remaster and not self._non_remaster(album))
+            ):
+                return
+            await album.rip()
+
+        batches = self.batch(
+            [_rip(album) for album in self.albums],
+            RESOLVE_CHUNK_SIZE,
+        )
+        for batch in batches:
+            await asyncio.gather(*batch)
+
+    def _apply_filters(
+        self, albums: list[Album], filt: QobuzDiscographyFilterConfig
+    ) -> list[Album]:
+        _albums = albums
+        if filt.repeats:
+            _albums = self._filter_repeats(_albums)
+        if filt.extras:
+            _albums = filter(self._extras, _albums)
+        if filt.features:
+            _albums = filter(self._features, _albums)
+        if filt.non_studio_albums:
+            _albums = filter(self._non_studio_albums, _albums)
+        if filt.non_remaster:
+            _albums = filter(self._non_remaster, _albums)
+        return list(_albums)
+
+    # Will not fail on any nonempty string
+    _essence = re.compile(r"([^\(]+)(?:\s*[\(\[][^\)][\)\]])*")
+
+    def _filter_repeats(self, albums: list[Album]) -> list[Album]:
+        """When there are different versions of an album on the artist,
+        choose the one with the best quality.
+
+        It determines that two albums are identical if they have the same title
+        ignoring contents in brackets or parentheses.
+        """
+        groups: dict[str, list[Album]] = {}
+        for a in albums:
+            match = self._essence.match(a.meta.album)
+            assert match is not None
+            title = match.group(1).strip().lower()
+            items = groups.get(title, [])
+            items.append(a)
+            groups[title] = items
+
+        ret: list[Album] = []
+        for group in groups.values():
+            best = None
+            max_bd, max_sr = 0, 0
+            # assume that highest bd is always with highest sr
+            for album in group:
+                bd = album.meta.info.bit_depth or 0
+                if bd > max_bd:
+                    max_bd = bd
+                    best = album
+
+                sr = album.meta.info.sampling_rate or 0
+                if sr > max_sr:
+                    max_sr = sr
+                    best = album
+
+            assert best is not None  # true because all g != []
+            ret.append(best)
+
+        return ret
+
+    _extra_re = re.compile(
+        r"(?i)(anniversary|deluxe|live|collector|demo|expanded|remix)"
+    )
+
+    # ----- Filter predicates -----
+    def _non_studio_albums(self, a: Album) -> bool:
+        """Filter out non studio albums."""
+        return a.meta.albumartist != "Various Artists" and self._extras(a)
+
+    def _features(self, a: Album) -> bool:
+        """Filter out features."""
+        return a.meta.albumartist == self.name
+
+    def _extras(self, a: Album) -> bool:
+        """Filter out extras.
+
+        See `_extra_re` for criteria.
+        """
+        return self._extra_re.search(a.meta.album) is None
+
+    _remaster_re = re.compile(r"(?i)(re)?master(ed)?")
+
+    def _non_remaster(self, a: Album) -> bool:
+        """Filter out albums that are not remasters."""
+        return self._remaster_re.search(a.meta.album) is not None
+
+    def _non_albums(self, a: Album) -> bool:
+        """Filter out singles."""
+        return len(a.tracks) > 1
+
+    @staticmethod
+    def batch(iterable, n=1):
+        total = len(iterable)
+        for ndx in range(0, total, n):
+            yield iterable[ndx : min(ndx + n, total)]
 
 
 @dataclass(slots=True)

streamrip/media/label.py

@@ -1,3 +1,4 @@
+import asyncio
 from dataclasses import dataclass
 
 from ..client import Client
@@ -5,12 +6,48 @@ from ..config import Config
 from ..db import Database
 from ..metadata import LabelMetadata
 from .album import PendingAlbum
-from .album_list import AlbumList
-from .media import Pending
+from .media import Media, Pending
 
 
-class Label(AlbumList):
-    pass
+@dataclass(slots=True)
+class Label(Media):
+    """Represents a list of albums. Used by Artist and Label classes."""
+
+    name: str
+    albums: list[PendingAlbum]
+    client: Client
+    config: Config
+
+    async def preprocess(self):
+        pass
+
+    async def download(self):
+        # Resolve only 3 albums at a time to avoid
+        # initial latency of resolving ALL albums and tracks
+        # before any downloads
+        album_resolve_chunk_size = 10
+
+        async def _resolve_download(item: PendingAlbum):
+            album = await item.resolve()
+            if album is None:
+                return
+            await album.rip()
+
+        batches = self.batch(
+            [_resolve_download(album) for album in self.albums],
+            album_resolve_chunk_size,
+        )
+        for batch in batches:
+            await asyncio.gather(*batch)
+
+    async def postprocess(self):
+        pass
+
+    @staticmethod
+    def batch(iterable, n=1):
+        total = len(iterable)
+        for ndx in range(0, total, n):
+            yield iterable[ndx : min(ndx + n, total)]
 
 
 @dataclass(slots=True)