music-assistant-server

1"""Several helper/utils to compare objects."""
2
3from __future__ import annotations
4
5import re
6from collections.abc import Sequence
7from difflib import SequenceMatcher
8from enum import Enum
9from functools import lru_cache
10
11from music_assistant_models.enums import ExternalID, MediaType
12from music_assistant_models.helpers import create_safe_string
13from music_assistant_models.media_items import (
14    Album,
15    Artist,
16    Audiobook,
17    ItemMapping,
18    MediaItem,
19    MediaItemMetadata,
20    MediaItemType,
21    Playlist,
22    Podcast,
23    Radio,
24    Track,
25)
26
27from music_assistant.helpers.external_ids import is_valid_isrc, normalize_external_id
28
29IGNORE_VERSIONS = (
30    "explicit",  # explicit is matched separately
31    "music from and inspired by the motion picture",
32    "original soundtrack",
33    "hi-res",  # quality is handled separately
34)
35
36_VERSION_IGNORE_WORDS = {
37    "album",
38    "at",
39    "edition",
40    "variant",
41    "versie",
42    "version",
43    "versione",
44}
45_VERSION_WORD_ALIASES = {
46    "remastered": "remaster",
47}
48_IGNORE_VERSION_KEYS = {create_safe_string(value) for value in IGNORE_VERSIONS}
49
50# version tokens that signal a fundamentally different recording (not just packaging),
51# so they must never be treated as an ambiguous/mergeable edition difference
52_RECORDING_CONFLICT_VERSION_TOKENS = {
53    "acoustic",
54    "cover",
55    "demo",
56    "instrumental",
57    "karaoke",
58    "live",
59    "remix",
60    "session",
61}
62
63# duration tolerances (seconds) for the album-track fingerprint comparison
64_ISRC_DURATION_TOLERANCE = 8
65_FALLBACK_DURATION_TOLERANCE = 2
66
67
68class AlbumMatchEvidence(Enum):
69    """Confidence level for an album identity comparison."""
70
71    MATCH = "match"
72    NO_MATCH = "no_match"
73    INSUFFICIENT = "insufficient"
74
75
76def compare_media_item(
77    base_item: MediaItemType | ItemMapping,
78    compare_item: MediaItemType | ItemMapping,
79    strict: bool = True,
80) -> bool | None:
81    """Compare two media items and return True if they match."""
82    if base_item.media_type == MediaType.ARTIST and compare_item.media_type == MediaType.ARTIST:
83        assert isinstance(base_item, Artist | ItemMapping)  # for type checking
84        assert isinstance(compare_item, Artist | ItemMapping)  # for type checking
85        return compare_artist(base_item, compare_item, strict)
86    if base_item.media_type == MediaType.ALBUM and compare_item.media_type == MediaType.ALBUM:
87        assert isinstance(base_item, Album | ItemMapping)  # for type checking
88        assert isinstance(compare_item, Album | ItemMapping)  # for type checking
89        return compare_album(base_item, compare_item, strict)
90    if base_item.media_type == MediaType.TRACK and compare_item.media_type == MediaType.TRACK:
91        assert isinstance(base_item, Track)  # for type checking
92        assert isinstance(compare_item, Track)  # for type checking
93        return compare_track(base_item, compare_item, strict)
94    if base_item.media_type == MediaType.PLAYLIST and compare_item.media_type == MediaType.PLAYLIST:
95        assert isinstance(base_item, Playlist | ItemMapping)  # for type checking
96        assert isinstance(compare_item, Playlist | ItemMapping)  # for type checking
97        return compare_playlist(base_item, compare_item, strict)
98    if base_item.media_type == MediaType.RADIO and compare_item.media_type == MediaType.RADIO:
99        assert isinstance(base_item, Radio | ItemMapping)  # for type checking
100        assert isinstance(compare_item, Radio | ItemMapping)  # for type checking
101        return compare_radio(base_item, compare_item, strict)
102    if (
103        base_item.media_type == MediaType.AUDIOBOOK
104        and compare_item.media_type == MediaType.AUDIOBOOK
105    ):
106        assert isinstance(base_item, Audiobook | ItemMapping)  # for type checking
107        assert isinstance(compare_item, Audiobook | ItemMapping)  # for type checking
108        return compare_audiobook(base_item, compare_item, strict)
109    if base_item.media_type == MediaType.PODCAST and compare_item.media_type == MediaType.PODCAST:
110        assert isinstance(base_item, Podcast | ItemMapping)  # for type checking
111        assert isinstance(compare_item, Podcast | ItemMapping)  # for type checking
112        return compare_podcast(base_item, compare_item, strict)
113    assert isinstance(base_item, ItemMapping)  # for type checking
114    assert isinstance(compare_item, ItemMapping)  # for type checking
115    return compare_item_mapping(base_item, compare_item, strict)
116
117
118def compare_artist(
119    base_item: Artist | ItemMapping,
120    compare_item: Artist | ItemMapping,
121    strict: bool = True,
122) -> bool | None:
123    """Compare two artist items and return True if they match."""
124    # return early on exact item_id match
125    if compare_item_ids(base_item, compare_item):
126        return True
127    # return early on (un)matched external id
128    for ext_id in (ExternalID.MB_ARTIST, ExternalID.DISCOGS, ExternalID.TADB):
129        external_id_match = compare_external_ids(
130            base_item.external_ids, compare_item.external_ids, ext_id
131        )
132        if external_id_match is not None:
133            return external_id_match
134    # return early if artist_types don't match
135    if (
136        isinstance(base_item, Artist)
137        and isinstance(compare_item, Artist)
138        and base_item.artist_type != compare_item.artist_type
139    ):
140        return False
141    # finally comparing on (exact) name match
142    return compare_strings(base_item.name, compare_item.name, strict=strict)
143
144
145def compare_album(
146    base_item: Album | ItemMapping,
147    compare_item: Album | ItemMapping,
148    strict: bool = True,
149) -> bool | None:
150    """Compare two album items and return True if they match."""
151    return compare_album_evidence(base_item, compare_item, strict) == AlbumMatchEvidence.MATCH
152
153
154def compare_album_evidence(
155    base_item: Album | ItemMapping,
156    compare_item: Album | ItemMapping,
157    strict: bool = True,
158    base_tracks: Sequence[Track] | None = None,
159    compare_tracks: Sequence[Track] | None = None,
160) -> AlbumMatchEvidence:
161    """
162    Return the match evidence for two album items.
163
164    Unlike `compare_album`, this distinguishes a confident non-match from
165    insufficient metadata (e.g. an edition difference that cannot be resolved from
166    the album's own fields), so a caller that can fetch tracklists knows when doing
167    so may still resolve the comparison. If `base_tracks`/`compare_tracks` are
168    supplied, an ordered track fingerprint comparison is used to resolve that
169    remaining ambiguity, and a conflicting fingerprint overrides an otherwise
170    nominally-matching album (e.g. identical title/version/year but a different
171    number of tracks).
172
173    :param base_tracks: Ordered tracklist for base_item, if already available to the caller.
174    :param compare_tracks: Ordered tracklist for compare_item, if already available.
175    """
176    # return early on exact item_id match
177    if compare_item_ids(base_item, compare_item):
178        return AlbumMatchEvidence.MATCH
179
180    # return early on (un)matched authoritative external id
181    for ext_id in (
182        ExternalID.MB_ALBUM,
183        ExternalID.DISCOGS,
184        ExternalID.TADB,
185    ):
186        external_id_match = compare_external_ids(
187            base_item.external_ids, compare_item.external_ids, ext_id
188        )
189        if external_id_match is not None:
190            return AlbumMatchEvidence.MATCH if external_id_match else AlbumMatchEvidence.NO_MATCH
191
192    # barcode/ASIN are shared across pressings and are non-unique corroboration only,
193    # so they are never used on their own, only to corroborate a year mismatch below
194    secondary_external_id_match = any(
195        compare_external_ids(base_item.external_ids, compare_item.external_ids, ext_id) is True
196        for ext_id in (ExternalID.ASIN, ExternalID.BARCODE)
197    )
198
199    # a real edition conflict (e.g. deluxe vs. live) is decisive, an ambiguous
200    # subset/superset wording (e.g. "2022 Remaster" vs "Deluxe 2022 Remaster") is not
201    version_evidence = _compare_album_version(base_item.version, compare_item.version)
202    if version_evidence == AlbumMatchEvidence.NO_MATCH:
203        return AlbumMatchEvidence.NO_MATCH
204    # compare name
205    if not _compare_album_name(base_item.name, compare_item.name):
206        return AlbumMatchEvidence.NO_MATCH
207
208    ambiguous = version_evidence == AlbumMatchEvidence.INSUFFICIENT
209    if not strict and (isinstance(base_item, ItemMapping) or isinstance(compare_item, ItemMapping)):
210        return _finalize_album_evidence(ambiguous, base_tracks, compare_tracks)
211    # for strict matching we REQUIRE both items to be a real album object
212    assert isinstance(base_item, Album)
213    assert isinstance(compare_item, Album)
214    # compare year: without corroboration this is provider drift, not proof either way
215    if (
216        base_item.year
217        and compare_item.year
218        and base_item.year != compare_item.year
219        and not secondary_external_id_match
220    ):
221        ambiguous = True
222    # compare explicitness
223    if compare_explicit(base_item.metadata, compare_item.metadata) is False:
224        return AlbumMatchEvidence.NO_MATCH
225    # compare album artist(s)
226    if not compare_artists(base_item.artists, compare_item.artists, not strict):
227        return AlbumMatchEvidence.NO_MATCH
228    return _finalize_album_evidence(ambiguous, base_tracks, compare_tracks)
229
230
231def compare_album_track_fingerprint(
232    base_tracks: Sequence[Track] | None,
233    compare_tracks: Sequence[Track] | None,
234) -> AlbumMatchEvidence:
235    """
236    Compare two album tracklists position-by-position and return match evidence.
237
238    Requires an identical disc/track shape to consider two tracklists the same
239    edition; a tracklist that never reports a disc number is treated as insufficient
240    (not assumed disc 1) when compared against a genuinely multi-disc tracklist. At
241    each position, a shared (normalized) ISRC with a compatible duration is preferred
242    as identity evidence; conflicting ISRCs indicate a different recording/remaster.
243    Positions without a usable ISRC on either side fall back to a normalized
244    title/version match with a tight duration tolerance.
245
246    :param base_tracks: Ordered tracklist for the base album.
247    :param compare_tracks: Ordered tracklist for the album being compared.
248    """
249    if not base_tracks or not compare_tracks:
250        return AlbumMatchEvidence.INSUFFICIENT
251    base_positions = _track_positions(base_tracks)
252    compare_positions = _track_positions(compare_tracks)
253    if not base_positions or not compare_positions:
254        return AlbumMatchEvidence.INSUFFICIENT
255    base_is_multi_disc = any(disc_number > 1 for disc_number, _ in base_positions)
256    compare_is_multi_disc = any(disc_number > 1 for disc_number, _ in compare_positions)
257    if (base_is_multi_disc and _has_unknown_disc_layout(compare_tracks)) or (
258        compare_is_multi_disc and _has_unknown_disc_layout(base_tracks)
259    ):
260        # one side never reports a disc number while the other is genuinely multi-disc:
261        # assuming disc 1 for the unknown side would produce a false shape conflict
262        return AlbumMatchEvidence.INSUFFICIENT
263    if base_positions.keys() != compare_positions.keys():
264        # different disc/track shape (e.g. a bonus disc or missing tracks): different edition
265        return AlbumMatchEvidence.NO_MATCH
266
267    evidence = AlbumMatchEvidence.MATCH
268    for position, base_track in base_positions.items():
269        position_evidence = _compare_track_fingerprint(base_track, compare_positions[position])
270        if position_evidence == AlbumMatchEvidence.NO_MATCH:
271            return AlbumMatchEvidence.NO_MATCH
272        if position_evidence == AlbumMatchEvidence.INSUFFICIENT:
273            evidence = AlbumMatchEvidence.INSUFFICIENT
274    return evidence
275
276
277def compare_track(
278    base_item: Track,
279    compare_item: Track,
280    strict: bool = True,
281    track_albums: list[Album] | None = None,
282) -> bool:
283    """Compare two track items and return True if they match."""
284    # return early on exact item_id match
285    if compare_item_ids(base_item, compare_item):
286        return True
287    # tracks on the same album but different discs are always distinct,
288    # even if they share external IDs (e.g. same recording on multiple discs)
289    if (
290        base_item.album
291        and compare_item.album
292        and base_item.disc_number
293        and compare_item.disc_number
294        and base_item.disc_number != compare_item.disc_number
295        and compare_album(base_item.album, compare_item.album, False)
296    ):
297        return False
298    # return early on (un)matched primary/unique external id
299    for ext_id in (
300        ExternalID.MB_RECORDING,
301        ExternalID.MB_TRACK,
302        ExternalID.ACOUSTID,
303    ):
304        external_id_match = compare_external_ids(
305            base_item.external_ids, compare_item.external_ids, ext_id
306        )
307        if external_id_match is not None:
308            return external_id_match
309    # check secondary external id matches
310    for ext_id in (
311        ExternalID.DISCOGS,
312        ExternalID.TADB,
313        ExternalID.ISRC,
314        ExternalID.ASIN,
315    ):
316        external_id_match = compare_external_ids(
317            base_item.external_ids, compare_item.external_ids, ext_id
318        )
319        if external_id_match is True:
320            # we got a 'soft-match' on a secondary external id (like ISRC)
321            # but we do a double check on duration
322            if abs(base_item.duration - compare_item.duration) <= 8:
323                return True
324
325    # compare name
326    if not compare_strings(base_item.name, compare_item.name, strict=True):
327        return False
328    # track artist(s) must match
329    if not compare_artists(base_item.artists, compare_item.artists, any_match=not strict):
330        return False
331    # track version must match
332    if strict and not compare_version(base_item.version, compare_item.version):
333        return False
334    # check if both tracks are (not) explicit
335    if base_item.metadata.explicit is None and isinstance(base_item.album, Album):
336        base_item.metadata.explicit = base_item.album.metadata.explicit
337    if compare_item.metadata.explicit is None and isinstance(compare_item.album, Album):
338        compare_item.metadata.explicit = compare_item.album.metadata.explicit
339    if strict and compare_explicit(base_item.metadata, compare_item.metadata) is False:
340        return False
341
342    # exact albumtrack match = 100% match
343    if (
344        base_item.album
345        and compare_item.album
346        and compare_album(base_item.album, compare_item.album, False)
347        and base_item.disc_number
348        and compare_item.disc_number
349        and base_item.track_number
350        and compare_item.track_number
351        and base_item.disc_number == compare_item.disc_number
352        and base_item.track_number == compare_item.track_number
353    ):
354        return True
355
356    # fallback: exact album match and (near-exact) track duration match
357    if (
358        base_item.album is not None
359        and compare_item.album is not None
360        and (base_item.track_number == 0 or compare_item.track_number == 0)
361        and compare_album(base_item.album, compare_item.album, False)
362        and abs(base_item.duration - compare_item.duration) <= 3
363    ):
364        return True
365
366    # fallback: additional compare albums provided for base track
367    if (
368        compare_item.album is not None
369        and track_albums
370        and abs(base_item.duration - compare_item.duration) <= 3
371    ):
372        for track_album in track_albums:
373            if compare_album(track_album, compare_item.album, False):
374                return True
375
376    # fallback edge case: albumless track with same duration
377    if (
378        base_item.album is None
379        and compare_item.album is None
380        and base_item.disc_number == 0
381        and compare_item.disc_number == 0
382        and base_item.track_number == 0
383        and compare_item.track_number == 0
384        and base_item.duration == compare_item.duration
385    ):
386        return True
387
388    if strict:
389        # in strict mode, we require an exact album match so return False here
390        return False
391
392    # Accept last resort (in non strict mode): (near) exact duration,
393    # otherwise fail all other cases.
394    # Note that as this stage, all other info already matches,
395    # such as title, artist etc.
396    return abs(base_item.duration - compare_item.duration) <= 2
397
398
399def compare_playlist(
400    base_item: Playlist | ItemMapping,
401    compare_item: Playlist | ItemMapping,
402    strict: bool = True,
403) -> bool | None:
404    """Compare two Playlist items and return True if they match."""
405    # require (exact) name match
406    if not compare_strings(base_item.name, compare_item.name, strict=strict):
407        return False
408    # require exact owner match (if not ItemMapping)
409    if isinstance(base_item, Playlist) and isinstance(compare_item, Playlist):
410        if not compare_strings(base_item.owner, compare_item.owner):
411            return False
412    # a playlist is always unique - so do a strict compare on item id(s)
413    return compare_item_ids(base_item, compare_item)
414
415
416def compare_radio(
417    base_item: Radio | ItemMapping,
418    compare_item: Radio | ItemMapping,
419    strict: bool = True,
420) -> bool | None:
421    """Compare two Radio items and return True if they match."""
422    # return early on exact item_id match
423    if compare_item_ids(base_item, compare_item):
424        return True
425    # a dynamic station is its provider's own, so a same-named station is a different one
426    if _is_dynamic_radio(base_item) or _is_dynamic_radio(compare_item):
427        return False
428    # compare version
429    if not compare_version(base_item.version, compare_item.version):
430        return False
431    # finally comparing on (exact) name match
432    return compare_strings(base_item.name, compare_item.name, strict=strict)
433
434
435def compare_audiobook(
436    base_item: Audiobook | ItemMapping,
437    compare_item: Audiobook | ItemMapping,
438    strict: bool = True,
439) -> bool | None:
440    """Compare two Audiobook items and return True if they match."""
441    # return early on exact item_id match
442    if compare_item_ids(base_item, compare_item):
443        return True
444
445    # return early on (un)matched external id
446    for ext_id in (
447        ExternalID.ASIN,
448        ExternalID.BARCODE,
449    ):
450        external_id_match = compare_external_ids(
451            base_item.external_ids, compare_item.external_ids, ext_id
452        )
453        if external_id_match is not None:
454            return external_id_match
455
456    # compare version
457    if not compare_version(base_item.version, compare_item.version):
458        return False
459    # compare name
460    if not compare_strings(base_item.name, compare_item.name, strict=True):
461        return False
462    if not strict and (isinstance(base_item, ItemMapping) or isinstance(compare_item, ItemMapping)):
463        return True
464    # for strict matching we REQUIRE both items to be a real Audiobook object
465    assert isinstance(base_item, Audiobook)
466    assert isinstance(compare_item, Audiobook)
467    # compare publisher
468    if (
469        base_item.publisher
470        and compare_item.publisher
471        and not compare_strings(base_item.publisher, compare_item.publisher, strict=True)
472    ):
473        return False
474
475    def _audiobook_artist_name(value: str | Artist | ItemMapping) -> str:
476        return value.name if isinstance(value, Artist | ItemMapping) else value
477
478    # compare narrator(s) — different narrators indicate different recordings and must not be merged
479    if base_item.narrators and compare_item.narrators:
480        base_narrators = {
481            create_safe_string(_audiobook_artist_name(n)) for n in base_item.narrators
482        }
483        compare_narrators = {
484            create_safe_string(_audiobook_artist_name(n)) for n in compare_item.narrators
485        }
486        if base_narrators.isdisjoint(compare_narrators):
487            return False
488    # compare author(s)
489    for author in base_item.authors:
490        author_safe = create_safe_string(_audiobook_artist_name(author))
491        if author_safe in [
492            create_safe_string(_audiobook_artist_name(x)) for x in compare_item.authors
493        ]:
494            return True
495    return False
496
497
498def compare_podcast(
499    base_item: Podcast | ItemMapping,
500    compare_item: Podcast | ItemMapping,
501    strict: bool = True,
502) -> bool | None:
503    """Compare two Podcast items and return True if they match."""
504    # return early on exact item_id match
505    if compare_item_ids(base_item, compare_item):
506        return True
507
508    # return early on (un)matched external id
509    for ext_id in (
510        ExternalID.ASIN,
511        ExternalID.BARCODE,
512    ):
513        external_id_match = compare_external_ids(
514            base_item.external_ids, compare_item.external_ids, ext_id
515        )
516        if external_id_match is not None:
517            return external_id_match
518
519    # compare version
520    if not compare_version(base_item.version, compare_item.version):
521        return False
522    # compare name
523    if not compare_strings(base_item.name, compare_item.name, strict=True):
524        return False
525    if not strict and (isinstance(base_item, ItemMapping) or isinstance(compare_item, ItemMapping)):
526        return True
527    # for strict matching we REQUIRE both items to be a real Podcast object
528    assert isinstance(base_item, Podcast)
529    assert isinstance(compare_item, Podcast)
530    # compare publisher
531    return not (
532        base_item.publisher
533        and compare_item.publisher
534        and not compare_strings(base_item.publisher, compare_item.publisher, strict=True)
535    )
536
537
538def compare_item_mapping(
539    base_item: ItemMapping,
540    compare_item: ItemMapping,
541    strict: bool = True,
542) -> bool | None:
543    """Compare two ItemMapping items and return True if they match."""
544    # return early on exact item_id match
545    if compare_item_ids(base_item, compare_item):
546        return True
547    # return early on (un)matched external id
548    # check all ExternalID, as ItemMapping is a minimized obj for all MediaItems
549    for ext_id in ExternalID:
550        external_id_match = compare_external_ids(
551            base_item.external_ids, compare_item.external_ids, ext_id
552        )
553        if external_id_match is not None:
554            return external_id_match
555    # compare version
556    if not compare_version(base_item.version, compare_item.version):
557        return False
558    # finally comparing on (exact) name match
559    return compare_strings(base_item.name, compare_item.name, strict=strict)
560
561
562def compare_artists(
563    base_items: list[Artist | ItemMapping],
564    compare_items: list[Artist | ItemMapping],
565    any_match: bool = True,
566) -> bool:
567    """Compare two lists of artist and return True if both lists match (exactly)."""
568    if not base_items or not compare_items:
569        return False
570    # match if first artist matches in both lists
571    if compare_artist(base_items[0], compare_items[0]):
572        return True
573    # compare the artist lists
574    matches = 0
575    for base_item in base_items:
576        for compare_item in compare_items:
577            if compare_artist(base_item, compare_item):
578                if any_match:
579                    return True
580                matches += 1
581    return len(base_items) == len(compare_items) == matches
582
583
584def compare_albums(
585    base_items: list[Album | ItemMapping],
586    compare_items: list[Album | ItemMapping],
587    any_match: bool = True,
588) -> bool:
589    """Compare two lists of albums and return True if a match was found."""
590    matches = 0
591    for base_item in base_items:
592        for compare_item in compare_items:
593            if compare_album(base_item, compare_item):
594                if any_match:
595                    return True
596                matches += 1
597    return len(base_items) == matches
598
599
600def compare_item_ids(
601    base_item: MediaItem | ItemMapping, compare_item: MediaItem | ItemMapping
602) -> bool:
603    """Compare item_id(s) of two media items."""
604    if not base_item.provider or not compare_item.provider:
605        return False
606    if not base_item.item_id or not compare_item.item_id:
607        return False
608    if base_item.provider == compare_item.provider and base_item.item_id == compare_item.item_id:
609        return True
610
611    base_prov_ids = getattr(base_item, "provider_mappings", None)
612    compare_prov_ids = getattr(compare_item, "provider_mappings", None)
613
614    if base_prov_ids is not None:
615        assert isinstance(base_item, MediaItem)  # for type checking
616        for prov_l in base_item.provider_mappings:
617            if (
618                prov_l.provider_instance == compare_item.provider
619                and prov_l.item_id == compare_item.item_id
620            ):
621                return True
622
623    if compare_prov_ids is not None:
624        assert isinstance(compare_item, MediaItem)  # for type checking
625        for prov_r in compare_item.provider_mappings:
626            if (
627                prov_r.provider_instance == base_item.provider
628                and prov_r.item_id == base_item.item_id
629            ):
630                return True
631
632    if base_prov_ids is not None and compare_prov_ids is not None:
633        assert isinstance(base_item, MediaItem)  # for type checking
634        assert isinstance(compare_item, MediaItem)  # for type checking
635        for prov_l in base_item.provider_mappings:
636            for prov_r in compare_item.provider_mappings:
637                if prov_l.provider_domain != prov_r.provider_domain:
638                    continue
639                if (
640                    prov_l.is_unique or prov_r.is_unique
641                ) and prov_l.provider_instance != prov_r.provider_instance:
642                    continue
643                if prov_l.item_id == prov_r.item_id:
644                    return True
645    return False
646
647
648def compare_external_ids(
649    external_ids_base: set[tuple[ExternalID, str]],
650    external_ids_compare: set[tuple[ExternalID, str]],
651    external_id_type: ExternalID,
652) -> bool | None:
653    """Compare external ids and return True if a match was found."""
654    base_ids = {
655        normalize_external_id(external_id_type, value)
656        for current_type, value in external_ids_base
657        if current_type == external_id_type
658    }
659    if not base_ids:
660        # return early if the requested external id type is not present in the base set
661        return None
662    compare_ids = {
663        normalize_external_id(external_id_type, value)
664        for current_type, value in external_ids_compare
665        if current_type == external_id_type
666    }
667    if not compare_ids:
668        # return early if the requested external id type is not present in the compare set
669        return None
670    if base_ids.intersection(compare_ids):
671        return True
672    if external_id_type.is_unique:
673        return False
674    return None
675
676
677def loose_compare_strings(base: str, alt: str) -> bool:
678    """Compare strings and return True even on partial match."""
679    # this is used to display 'versions' of the same track/album
680    # where we account for other spelling or some additional wording in the title
681    if len(base) <= 3 or len(alt) <= 3:
682        return compare_strings(base, alt, True)
683    word_count = len(base.strip().split(" "))
684    if word_count == 1 and len(base) < 10:
685        return compare_strings(base, alt, False)
686    base_comp = create_safe_string(base)
687    alt_comp = create_safe_string(alt)
688    if base_comp in alt_comp:
689        return True
690    return base_comp in alt_comp
691
692
693def compare_strings(str1: str, str2: str, strict: bool = True) -> bool:
694    """Compare strings and return True if we have an (almost) perfect match."""
695    if not str1 or not str2:
696        return False
697    str1_lower = str1.lower()
698    str2_lower = str2.lower()
699    if strict:
700        return str1_lower == str2_lower
701    # return early if total length mismatch
702    if abs(len(str1) - len(str2)) > 4:
703        return False
704    # handle '&' vs 'And'
705    if " & " in str1_lower and " and " in str2_lower:
706        str2 = str2_lower.replace(" and ", " & ")
707    elif " and " in str1_lower and " & " in str2:
708        str2 = str2_lower.replace(" & ", " and ")
709    if create_safe_string(str1) == create_safe_string(str2):
710        return True
711    # last resort: use difflib to compare strings
712    required_accuracy = 0.9 if (len(str1) + len(str2)) > 18 else 0.8
713    return SequenceMatcher(a=str1_lower, b=str2_lower).ratio() > required_accuracy
714
715
716def compare_version(base_version: str, compare_version: str) -> bool:
717    """Compare version string."""
718    return _normalize_version_tokens(base_version) == _normalize_version_tokens(compare_version)
719
720
721def compare_explicit(base: MediaItemMetadata, compare: MediaItemMetadata) -> bool | None:
722    """Compare if explicit is same in metadata."""
723    if base.explicit is not None and compare.explicit is not None:
724        # explicitness info is not always present in metadata
725        # only strict compare them if both have the info set
726        return base.explicit == compare.explicit
727    return None
728
729
730@lru_cache(maxsize=1024)
731def _normalize_version_tokens(value: str) -> tuple[str, ...]:
732    """Return meaningful version tokens in stable order."""
733    if not value or create_safe_string(value) in _IGNORE_VERSION_KEYS:
734        return ()
735    tokens = (
736        _VERSION_WORD_ALIASES.get(token, token)
737        for token in re.findall(r"[^\W_]+", value.casefold())
738    )
739    return tuple(sorted(token for token in tokens if token not in _VERSION_IGNORE_WORDS))
740
741
742def _is_dynamic_radio(item: Radio | ItemMapping) -> bool:
743    """Return True if the item is a dynamic radio station."""
744    return isinstance(item, Radio) and item.is_dynamic
745
746
747def _compare_album_version(base_version: str, compare_version: str) -> AlbumMatchEvidence:
748    """Return match evidence for an album version/edition comparison."""
749    base_tokens = set(_normalize_version_tokens(base_version))
750    compare_tokens = set(_normalize_version_tokens(compare_version))
751    if base_tokens == compare_tokens:
752        return AlbumMatchEvidence.MATCH
753    # a recording-changing qualifier (live, karaoke, remix, ...) makes an otherwise
754    # unequal pair of editions unsafe to merge, wherever it appears in either wording,
755    # not only when it is the token that happens to differ between the two, and even
756    # when the other side omits version metadata entirely
757    if (base_tokens | compare_tokens) & _RECORDING_CONFLICT_VERSION_TOKENS:
758        return AlbumMatchEvidence.NO_MATCH
759    if not base_tokens or not compare_tokens:
760        # a provider commonly omits edition metadata entirely (e.g. a remaster tagged
761        # without a version string), so a blank version next to a real one is
762        # undecided rather than a proven conflict: let a tracklist resolve it
763        return AlbumMatchEvidence.INSUFFICIENT
764    if base_tokens < compare_tokens or compare_tokens < base_tokens:
765        # one version's wording is a strict subset of the other's (e.g. "2022 Remaster"
766        # vs. "Deluxe 2022 Remaster"): an ambiguous packaging difference a tracklist can resolve
767        return AlbumMatchEvidence.INSUFFICIENT
768    return AlbumMatchEvidence.NO_MATCH
769
770
771def _compare_album_name(base_name: str, compare_name: str) -> bool:
772    """Return True if two album titles are the same identity, ignoring formatting drift."""
773    base_safe = _normalize_album_name(base_name)
774    compare_safe = _normalize_album_name(compare_name)
775    if base_safe and compare_safe:
776        return base_safe == compare_safe
777    if base_safe or compare_safe:
778        return False
779    # both titles collapse to nothing under normalization (e.g. pure punctuation):
780    # fall back to a whitespace-normalized raw comparison so unrelated titles don't match
781    return " ".join(base_name.split()).casefold() == " ".join(compare_name.split()).casefold()
782
783
784def _normalize_album_name(name: str) -> str:
785    """Return a punctuation/diacritic/whitespace-normalized album title for identity checks."""
786    return " ".join(create_safe_string(name).split())
787
788
789def _track_positions(tracks: Sequence[Track]) -> dict[tuple[int, int], Track]:
790    """Return tracks keyed by their (disc_number, track_number) position."""
791    if len({bool(track.disc_number) for track in tracks}) > 1:
792        # some tracks report a disc number and others don't: the shape can't be trusted
793        return {}
794    positions: dict[tuple[int, int], Track] = {}
795    for track in tracks:
796        if not track.track_number:
797            return {}
798        key = (track.disc_number or 1, track.track_number)
799        if key in positions:
800            # duplicate position: the tracklist shape cannot be trusted
801            return {}
802        positions[key] = track
803    return positions
804
805
806def _has_unknown_disc_layout(tracks: Sequence[Track]) -> bool:
807    """Return True if a tracklist reports no disc number at all (an assumed single disc)."""
808    return all(not track.disc_number for track in tracks)
809
810
811def _compare_track_fingerprint(base_track: Track, compare_track: Track) -> AlbumMatchEvidence:
812    """Return match evidence for a single album-track position."""
813    base_isrcs = _track_isrcs(base_track)
814    compare_isrcs = _track_isrcs(compare_track)
815    if base_isrcs and compare_isrcs:
816        if base_isrcs.isdisjoint(compare_isrcs):
817            # both sides tagged an ISRC and they disagree: a different recording/remaster
818            return AlbumMatchEvidence.NO_MATCH
819        if not base_track.duration or not compare_track.duration:
820            return AlbumMatchEvidence.INSUFFICIENT
821        if _duration_close(base_track.duration, compare_track.duration, _ISRC_DURATION_TOLERANCE):
822            return AlbumMatchEvidence.MATCH
823        return AlbumMatchEvidence.INSUFFICIENT
824
825    # no usable ISRC on (at least) one side: fall back to title/version + duration
826    if not base_track.name or not compare_track.name:
827        return AlbumMatchEvidence.INSUFFICIENT
828    if not compare_strings(base_track.name, compare_track.name, strict=True):
829        return AlbumMatchEvidence.NO_MATCH
830    if not compare_version(base_track.version, compare_track.version):
831        return AlbumMatchEvidence.NO_MATCH
832    if not base_track.duration or not compare_track.duration:
833        return AlbumMatchEvidence.INSUFFICIENT
834    if _duration_close(base_track.duration, compare_track.duration, _FALLBACK_DURATION_TOLERANCE):
835        return AlbumMatchEvidence.MATCH
836    return AlbumMatchEvidence.NO_MATCH
837
838
839def _track_isrcs(track: Track) -> set[str]:
840    """Return the structurally valid, normalized ISRCs tagged on a track."""
841    return {
842        normalize_external_id(ExternalID.ISRC, value)
843        for current_type, value in track.external_ids
844        if current_type == ExternalID.ISRC and is_valid_isrc(value)
845    }
846
847
848def _duration_close(base_duration: int, compare_duration: int, tolerance: int) -> bool:
849    """Return True if two track durations (in seconds) are within tolerance."""
850    return abs(base_duration - compare_duration) <= tolerance
851
852
853def _finalize_album_evidence(
854    ambiguous: bool,
855    base_tracks: Sequence[Track] | None,
856    compare_tracks: Sequence[Track] | None,
857) -> AlbumMatchEvidence:
858    """Combine an album's metadata ambiguity with an optional track fingerprint override."""
859    fingerprint_evidence = compare_album_track_fingerprint(base_tracks, compare_tracks)
860    if fingerprint_evidence == AlbumMatchEvidence.NO_MATCH:
861        # a conflicting tracklist is decisive even if the album's own metadata looked fine
862        return AlbumMatchEvidence.NO_MATCH
863    if not ambiguous:
864        return AlbumMatchEvidence.MATCH
865    return fingerprint_evidence
866