beets/beetsplug/discogs.py
2025-09-19 20:46:07 -07:00

689 lines
27 KiB
Python

# This file is part of beets.
# Copyright 2016, Adrian Sampson.
#
# Permission is hereby granted, free of charge, to any person obtaining
# a copy of this software and associated documentation files (the
# "Software"), to deal in the Software without restriction, including
# without limitation the rights to use, copy, modify, merge, publish,
# distribute, sublicense, and/or sell copies of the Software, and to
# permit persons to whom the Software is furnished to do so, subject to
# the following conditions:
#
# The above copyright notice and this permission notice shall be
# included in all copies or substantial portions of the Software.
"""Adds Discogs album search support to the autotagger. Requires the
python3-discogs-client library.
"""
from __future__ import annotations
import http.client
import json
import os
import re
import socket
import time
import traceback
from functools import cache
from string import ascii_lowercase
from typing import TYPE_CHECKING, Sequence
import confuse
from discogs_client import Client, Master, Release
from discogs_client.exceptions import DiscogsAPIError
from requests.exceptions import ConnectionError
from typing_extensions import TypedDict
import beets
import beets.ui
from beets import config
from beets.autotag.distance import string_dist
from beets.autotag.hooks import AlbumInfo, TrackInfo
from beets.metadata_plugins import MetadataSourcePlugin
if TYPE_CHECKING:
from collections.abc import Callable, Iterable
from beets.library import Item
USER_AGENT = f"beets/{beets.__version__} +https://beets.io/"
API_KEY = "rAzVUQYRaoFjeBjyWuWZ"
API_SECRET = "plxtUTqoCzwxZpqdPysCwGuBSmZNdZVy"
# Exceptions that discogs_client should really handle but does not.
CONNECTION_ERRORS = (
ConnectionError,
socket.error,
http.client.HTTPException,
ValueError, # JSON decoding raises a ValueError.
DiscogsAPIError,
)
TRACK_INDEX_RE = re.compile(
r"""
(.*?) # medium: everything before medium_index.
(\d*?) # medium_index: a number at the end of
# `position`, except if followed by a subtrack index.
# subtrack_index: can only be matched if medium
# or medium_index have been matched, and can be
(
(?<=\w)\.[\w]+ # a dot followed by a string (A.1, 2.A)
| (?<=\d)[A-Z]+ # a string that follows a number (1A, B2a)
)?
""",
re.VERBOSE,
)
DISAMBIGUATION_RE = re.compile(r" \(\d+\)")
class ReleaseFormat(TypedDict):
name: str
qty: int
descriptions: list[str] | None
class DiscogsPlugin(MetadataSourcePlugin):
def __init__(self):
super().__init__()
self.config.add(
{
"apikey": API_KEY,
"apisecret": API_SECRET,
"tokenfile": "discogs_token.json",
"source_weight": 0.5,
"user_token": "",
"separator": ", ",
"index_tracks": False,
"append_style_genre": False,
"strip_disambiguation": True,
}
)
self.config["apikey"].redact = True
self.config["apisecret"].redact = True
self.config["user_token"].redact = True
self.setup()
def setup(self, session=None):
"""Create the `discogs_client` field. Authenticate if necessary."""
c_key = self.config["apikey"].as_str()
c_secret = self.config["apisecret"].as_str()
# Try using a configured user token (bypassing OAuth login).
user_token = self.config["user_token"].as_str()
if user_token:
# The rate limit for authenticated users goes up to 60
# requests per minute.
self.discogs_client = Client(USER_AGENT, user_token=user_token)
return
# Get the OAuth token from a file or log in.
try:
with open(self._tokenfile()) as f:
tokendata = json.load(f)
except OSError:
# No token yet. Generate one.
token, secret = self.authenticate(c_key, c_secret)
else:
token = tokendata["token"]
secret = tokendata["secret"]
self.discogs_client = Client(USER_AGENT, c_key, c_secret, token, secret)
def reset_auth(self):
"""Delete token file & redo the auth steps."""
os.remove(self._tokenfile())
self.setup()
def _tokenfile(self):
"""Get the path to the JSON file for storing the OAuth token."""
return self.config["tokenfile"].get(confuse.Filename(in_app_dir=True))
def authenticate(self, c_key, c_secret):
# Get the link for the OAuth page.
auth_client = Client(USER_AGENT, c_key, c_secret)
try:
_, _, url = auth_client.get_authorize_url()
except CONNECTION_ERRORS as e:
self._log.debug("connection error: {}", e)
raise beets.ui.UserError("communication with Discogs failed")
beets.ui.print_("To authenticate with Discogs, visit:")
beets.ui.print_(url)
# Ask for the code and validate it.
code = beets.ui.input_("Enter the code:")
try:
token, secret = auth_client.get_access_token(code)
except DiscogsAPIError:
raise beets.ui.UserError("Discogs authorization failed")
except CONNECTION_ERRORS as e:
self._log.debug("connection error: {}", e)
raise beets.ui.UserError("Discogs token request failed")
# Save the token for later use.
self._log.debug("Discogs token {}, secret {}", token, secret)
with open(self._tokenfile(), "w") as f:
json.dump({"token": token, "secret": secret}, f)
return token, secret
def candidates(
self, items: Sequence[Item], artist: str, album: str, va_likely: bool
) -> Iterable[AlbumInfo]:
return self.get_albums(f"{artist} {album}" if va_likely else album)
def get_track_from_album(
self, album_info: AlbumInfo, compare: Callable[[TrackInfo], float]
) -> TrackInfo | None:
"""Return the best matching track of the release."""
scores_and_tracks = [(compare(t), t) for t in album_info.tracks]
score, track_info = min(scores_and_tracks, key=lambda x: x[0])
if score > 0.3:
return None
track_info["artist"] = album_info.artist
track_info["artist_id"] = album_info.artist_id
track_info["album"] = album_info.album
return track_info
def item_candidates(
self, item: Item, artist: str, title: str
) -> Iterable[TrackInfo]:
albums = self.candidates([item], artist, title, False)
def compare_func(track_info: TrackInfo) -> float:
return string_dist(track_info.title, title)
tracks = (self.get_track_from_album(a, compare_func) for a in albums)
return list(filter(None, tracks))
def album_for_id(self, album_id: str) -> AlbumInfo | None:
"""Fetches an album by its Discogs ID and returns an AlbumInfo object
or None if the album is not found.
"""
self._log.debug("Searching for release {}", album_id)
discogs_id = self._extract_id(album_id)
if not discogs_id:
return None
result = Release(self.discogs_client, {"id": discogs_id})
# Try to obtain title to verify that we indeed have a valid Release
try:
getattr(result, "title")
except DiscogsAPIError as e:
if e.status_code != 404:
self._log.debug(
"API Error: {} (query: {})",
e,
result.data["resource_url"],
)
if e.status_code == 401:
self.reset_auth()
return self.album_for_id(album_id)
return None
except CONNECTION_ERRORS:
self._log.debug("Connection error in album lookup", exc_info=True)
return None
return self.get_album_info(result)
def track_for_id(self, track_id: str) -> TrackInfo | None:
if album := self.album_for_id(track_id):
for track in album.tracks:
if track.track_id == track_id:
return track
return None
def get_albums(self, query: str) -> Iterable[AlbumInfo]:
"""Returns a list of AlbumInfo objects for a discogs search query."""
# Strip non-word characters from query. Things like "!" and "-" can
# cause a query to return no results, even if they match the artist or
# album title. Use `re.UNICODE` flag to avoid stripping non-english
# word characters.
query = re.sub(r"(?u)\W+", " ", query)
# Strip medium information from query, Things like "CD1" and "disk 1"
# can also negate an otherwise positive result.
query = re.sub(r"(?i)\b(CD|disc|vinyl)\s*\d+", "", query)
try:
results = self.discogs_client.search(query, type="release")
results.per_page = self.config["search_limit"].get()
releases = results.page(1)
except CONNECTION_ERRORS:
self._log.debug(
"Communication error while searching for {0!r}",
query,
exc_info=True,
)
return []
return filter(None, map(self.get_album_info, releases))
@cache
def get_master_year(self, master_id: str) -> int | None:
"""Fetches a master release given its Discogs ID and returns its year
or None if the master release is not found.
"""
self._log.debug("Getting master release {}", master_id)
result = Master(self.discogs_client, {"id": master_id})
try:
return result.fetch("year")
except DiscogsAPIError as e:
if e.status_code != 404:
self._log.debug(
"API Error: {} (query: {})",
e,
result.data["resource_url"],
)
if e.status_code == 401:
self.reset_auth()
return self.get_master_year(master_id)
return None
except CONNECTION_ERRORS:
self._log.debug(
"Connection error in master release lookup", exc_info=True
)
return None
@staticmethod
def get_media_and_albumtype(
formats: list[ReleaseFormat] | None,
) -> tuple[str | None, str | None]:
media = albumtype = None
if formats and (first_format := formats[0]):
if descriptions := first_format["descriptions"]:
albumtype = ", ".join(descriptions)
media = first_format["name"]
return media, albumtype
def get_album_info(self, result):
"""Returns an AlbumInfo object for a discogs Release object."""
# Explicitly reload the `Release` fields, as they might not be yet
# present if the result is from a `discogs_client.search()`.
if not result.data.get("artists"):
try:
result.refresh()
except CONNECTION_ERRORS:
self._log.debug(
"Connection error in release lookup: {0}",
result,
)
return None
# Sanity check for required fields. The list of required fields is
# defined at Guideline 1.3.1.a, but in practice some releases might be
# lacking some of these fields. This function expects at least:
# `artists` (>0), `title`, `id`, `tracklist` (>0)
# https://www.discogs.com/help/doc/submission-guidelines-general-rules
if not all(
[
result.data.get(k)
for k in ["artists", "title", "id", "tracklist"]
]
):
self._log.warning("Release does not contain the required fields")
return None
artist, artist_id = self.get_artist(
[a.data for a in result.artists], join_key="join"
)
album = re.sub(r" +", " ", result.title)
album_id = result.data["id"]
# Use `.data` to access the tracklist directly instead of the
# convenient `.tracklist` property, which will strip out useful artist
# information and leave us with skeleton `Artist` objects that will
# each make an API call just to get the same data back.
tracks = self.get_tracks(result.data["tracklist"])
# Extract information for the optional AlbumInfo fields, if possible.
va = result.data["artists"][0].get("name", "").lower() == "various"
year = result.data.get("year")
mediums = [t.medium for t in tracks]
country = result.data.get("country")
data_url = result.data.get("uri")
style = self.format(result.data.get("styles"))
base_genre = self.format(result.data.get("genres"))
if self.config["append_style_genre"] and style:
genre = self.config["separator"].as_str().join([base_genre, style])
else:
genre = base_genre
discogs_albumid = self._extract_id(result.data.get("uri"))
# Extract information for the optional AlbumInfo fields that are
# contained on nested discogs fields.
media, albumtype = self.get_media_and_albumtype(
result.data.get("formats")
)
label = catalogno = labelid = None
if result.data.get("labels"):
label = self.strip_disambiguation(
result.data["labels"][0].get("name")
)
catalogno = result.data["labels"][0].get("catno")
labelid = result.data["labels"][0].get("id")
cover_art_url = self.select_cover_art(result)
# Additional cleanups
# (various artists name, catalog number, media, disambiguation).
if va:
artist = config["va_name"].as_str()
else:
artist = self.strip_disambiguation(artist)
if catalogno == "none":
catalogno = None
# Explicitly set the `media` for the tracks, since it is expected by
# `autotag.apply_metadata`, and set `medium_total`.
for track in tracks:
track.media = media
track.medium_total = mediums.count(track.medium)
if not track.artist: # get_track_info often fails to find artist
track.artist = artist
if not track.artist_id:
track.artist_id = artist_id
# Discogs does not have track IDs. Invent our own IDs as proposed
# in #2336.
track.track_id = f"{album_id}-{track.track_alt}"
track.data_url = data_url
track.data_source = "Discogs"
# Retrieve master release id (returns None if there isn't one).
master_id = result.data.get("master_id")
# Assume `original_year` is equal to `year` for releases without
# a master release, otherwise fetch the master release.
original_year = self.get_master_year(master_id) if master_id else year
return AlbumInfo(
album=album,
album_id=album_id,
artist=artist,
artist_id=artist_id,
tracks=tracks,
albumtype=albumtype,
va=va,
year=year,
label=label,
mediums=len(set(mediums)),
releasegroup_id=master_id,
catalognum=catalogno,
country=country,
style=style,
genre=genre,
media=media,
original_year=original_year,
data_source=self.data_source,
data_url=data_url,
discogs_albumid=discogs_albumid,
discogs_labelid=labelid,
discogs_artistid=artist_id,
cover_art_url=cover_art_url,
)
def select_cover_art(self, result):
"""Returns the best candidate image, if any, from a Discogs `Release` object."""
if result.data.get("images") and len(result.data.get("images")) > 0:
# The first image in this list appears to be the one displayed first
# on the release page - even if it is not flagged as `type: "primary"` - and
# so it is the best candidate for the cover art.
return result.data.get("images")[0].get("uri")
return None
def format(self, classification):
if classification:
return (
self.config["separator"].as_str().join(sorted(classification))
)
else:
return None
def get_tracks(self, tracklist):
"""Returns a list of TrackInfo objects for a discogs tracklist."""
try:
clean_tracklist = self.coalesce_tracks(tracklist)
except Exception as exc:
# FIXME: this is an extra precaution for making sure there are no
# side effects after #2222. It should be removed after further
# testing.
self._log.debug("{}", traceback.format_exc())
self._log.error("uncaught exception in coalesce_tracks: {}", exc)
clean_tracklist = tracklist
tracks = []
index_tracks = {}
index = 0
# Distinct works and intra-work divisions, as defined by index tracks.
divisions, next_divisions = [], []
for track in clean_tracklist:
# Only real tracks have `position`. Otherwise, it's an index track.
if track["position"]:
index += 1
if next_divisions:
# End of a block of index tracks: update the current
# divisions.
divisions += next_divisions
del next_divisions[:]
track_info = self.get_track_info(track, index, divisions)
track_info.track_alt = track["position"]
tracks.append(track_info)
else:
next_divisions.append(track["title"])
# We expect new levels of division at the beginning of the
# tracklist (and possibly elsewhere).
try:
divisions.pop()
except IndexError:
pass
index_tracks[index + 1] = track["title"]
# Fix up medium and medium_index for each track. Discogs position is
# unreliable, but tracks are in order.
medium = None
medium_count, index_count, side_count = 0, 0, 0
sides_per_medium = 1
# If a medium has two sides (ie. vinyl or cassette), each pair of
# consecutive sides should belong to the same medium.
if all([track.medium is not None for track in tracks]):
m = sorted({track.medium.lower() for track in tracks})
# If all track.medium are single consecutive letters, assume it is
# a 2-sided medium.
if "".join(m) in ascii_lowercase:
sides_per_medium = 2
for track in tracks:
# Handle special case where a different medium does not indicate a
# new disc, when there is no medium_index and the ordinal of medium
# is not sequential. For example, I, II, III, IV, V. Assume these
# are the track index, not the medium.
# side_count is the number of mediums or medium sides (in the case
# of two-sided mediums) that were seen before.
medium_is_index = (
track.medium
and not track.medium_index
and (
len(track.medium) != 1
or
# Not within standard incremental medium values (A, B, C, ...).
ord(track.medium) - 64 != side_count + 1
)
)
if not medium_is_index and medium != track.medium:
side_count += 1
if sides_per_medium == 2:
if side_count % sides_per_medium:
# Two-sided medium changed. Reset index_count.
index_count = 0
medium_count += 1
else:
# Medium changed. Reset index_count.
medium_count += 1
index_count = 0
medium = track.medium
index_count += 1
medium_count = 1 if medium_count == 0 else medium_count
track.medium, track.medium_index = medium_count, index_count
# Get `disctitle` from Discogs index tracks. Assume that an index track
# before the first track of each medium is a disc title.
for track in tracks:
if track.medium_index == 1:
if track.index in index_tracks:
disctitle = index_tracks[track.index]
else:
disctitle = None
track.disctitle = disctitle
return tracks
def coalesce_tracks(self, raw_tracklist):
"""Pre-process a tracklist, merging subtracks into a single track. The
title for the merged track is the one from the previous index track,
if present; otherwise it is a combination of the subtracks titles.
"""
def add_merged_subtracks(tracklist, subtracks):
"""Modify `tracklist` in place, merging a list of `subtracks` into
a single track into `tracklist`."""
# Calculate position based on first subtrack, without subindex.
idx, medium_idx, sub_idx = self.get_track_index(
subtracks[0]["position"]
)
position = f"{idx or ''}{medium_idx or ''}"
if tracklist and not tracklist[-1]["position"]:
# Assume the previous index track contains the track title.
if sub_idx:
# "Convert" the track title to a real track, discarding the
# subtracks assuming they are logical divisions of a
# physical track (12.2.9 Subtracks).
tracklist[-1]["position"] = position
else:
# Promote the subtracks to real tracks, discarding the
# index track, assuming the subtracks are physical tracks.
index_track = tracklist.pop()
# Fix artists when they are specified on the index track.
if index_track.get("artists"):
for subtrack in subtracks:
if not subtrack.get("artists"):
subtrack["artists"] = index_track["artists"]
# Concatenate index with track title when index_tracks
# option is set
if self.config["index_tracks"]:
for subtrack in subtracks:
subtrack["title"] = (
f"{index_track['title']}: {subtrack['title']}"
)
tracklist.extend(subtracks)
else:
# Merge the subtracks, pick a title, and append the new track.
track = subtracks[0].copy()
track["title"] = " / ".join([t["title"] for t in subtracks])
tracklist.append(track)
# Pre-process the tracklist, trying to identify subtracks.
subtracks = []
tracklist = []
prev_subindex = ""
for track in raw_tracklist:
# Regular subtrack (track with subindex).
if track["position"]:
_, _, subindex = self.get_track_index(track["position"])
if subindex:
if subindex.rjust(len(raw_tracklist)) > prev_subindex:
# Subtrack still part of the current main track.
subtracks.append(track)
else:
# Subtrack part of a new group (..., 1.3, *2.1*, ...).
add_merged_subtracks(tracklist, subtracks)
subtracks = [track]
prev_subindex = subindex.rjust(len(raw_tracklist))
continue
# Index track with nested sub_tracks.
if not track["position"] and "sub_tracks" in track:
# Append the index track, assuming it contains the track title.
tracklist.append(track)
add_merged_subtracks(tracklist, track["sub_tracks"])
continue
# Regular track or index track without nested sub_tracks.
if subtracks:
add_merged_subtracks(tracklist, subtracks)
subtracks = []
prev_subindex = ""
tracklist.append(track)
# Merge and add the remaining subtracks, if any.
if subtracks:
add_merged_subtracks(tracklist, subtracks)
return tracklist
def strip_disambiguation(self, text) -> str:
"""Removes discogs specific disambiguations from a string.
Turns 'Label Name (5)' to 'Label Name' or 'Artist (1) & Another Artist (2)'
to 'Artist & Another Artist'. Does nothing if strip_disambiguation is False."""
if not self.config["strip_disambiguation"]:
return text
return DISAMBIGUATION_RE.sub("", text)
def get_track_info(self, track, index, divisions):
"""Returns a TrackInfo object for a discogs track."""
title = track["title"]
if self.config["index_tracks"]:
prefix = ", ".join(divisions)
if prefix:
title = f"{prefix}: {title}"
track_id = None
medium, medium_index, _ = self.get_track_index(track["position"])
artist, artist_id = self.get_artist(
track.get("artists", []), join_key="join"
)
artist = self.strip_disambiguation(artist)
length = self.get_track_length(track["duration"])
return TrackInfo(
title=title,
track_id=track_id,
artist=artist,
artist_id=artist_id,
length=length,
index=index,
medium=medium,
medium_index=medium_index,
)
@staticmethod
def get_track_index(
position: str,
) -> tuple[str | None, str | None, str | None]:
"""Returns the medium, medium index and subtrack index for a discogs
track position."""
# Match the standard Discogs positions (12.2.9), which can have several
# forms (1, 1-1, A1, A1.1, A1a, ...).
medium = index = subindex = None
if match := TRACK_INDEX_RE.fullmatch(position.upper()):
medium, index, subindex = match.groups()
if subindex and subindex.startswith("."):
subindex = subindex[1:]
return medium or None, index or None, subindex or None
def get_track_length(self, duration):
"""Returns the track length in seconds for a discogs duration."""
try:
length = time.strptime(duration, "%M:%S")
except ValueError:
return None
return length.tm_min * 60 + length.tm_sec