soulsync/core/text/title_match.py

"""Guard against char-level title false positives in track matching.

Issue #769: playlist sync matched tracks that aren't in the library to a
DIFFERENT song by the SAME artist, with high confidence — e.g. "Dani
California" -> "Californication" (Red Hot Chili Peppers), "Under The Bridge"
-> "Around the World". The confidence formula is ``0.5*title + 0.5*artist``,
and a same-artist comparison always yields ``artist = 1.0``, so the title score
is the only thing that can tell two of an artist's songs apart. But the title
score is a ``difflib.SequenceMatcher`` character ratio, which over-credits
unrelated titles that happen to share a long substring ("californi…") or only a
stopword ("the"): 0.67 and 0.62 respectively. With the flat 0.5 artist term
that lands at 0.83 / 0.81 — well over the 0.7 sync threshold.

``titles_plausibly_same`` adds a cheap word-level sanity check on top of the
char ratio: accept a pair only when it's near-identical char-wise (so typos and
punctuation/casing variants — "Beleive"/"Believe", "HUMBLE."/"Humble" — still
match) OR the two titles share at least one significant (non-stopword) token.
Two genuinely different songs by the same artist share no content word, so they
get rejected; the real track is then correctly reported missing.
"""

from __future__ import annotations

import re

# Articles / prepositions / conjunctions only. Deliberately NOT pronouns
# ("you", "me", "i") — those carry meaning in song titles and dropping them
# could strip the only shared word from a real match. "the" MUST stay here:
# without it "Under The Bridge" and "Around the World" would falsely share it.
_TITLE_STOPWORDS = frozenset({
    "the", "a", "an", "of", "and", "or", "to", "in", "on",
    "for", "with", "at", "by", "from",
})

_TOKEN_RE = re.compile(r"[a-z0-9]+")

# Char ratio at/above which two titles are treated as the same regardless of
# shared words — covers typos, punctuation, casing, accents. Tuned so single-
# word typos ("Beleive"/"Believe" = 0.857) pass while the #769 false positives
# ("Dani California"/"Californication" = 0.667) do not.
_NEAR_IDENTICAL = 0.85


def _content_tokens(text: str) -> set[str]:
    return {t for t in _TOKEN_RE.findall((text or "").lower()) if t not in _TITLE_STOPWORDS}


def titles_plausibly_same(
    title_a: str,
    title_b: str,
    char_similarity: float,
    *,
    near_identical: float = _NEAR_IDENTICAL,
) -> bool:
    """Whether two titles could be the same track, given their char similarity.

    ``title_a`` / ``title_b`` should already be normalised/cleaned (lowercased,
    brackets stripped) the same way the caller computed ``char_similarity``.

    Returns ``True`` when the pair is near-identical char-wise OR shares at
    least one significant (non-stopword) token. Returns ``False`` for two
    titles that are only moderately char-similar and share no content word —
    i.e. different songs the char ratio over-credited (#769)."""
    if char_similarity >= near_identical:
        return True
    ta = _content_tokens(title_a)
    tb = _content_tokens(title_b)
    # Word-overlap is only a reliable "different song" signal when at least one
    # side has 2+ content words — that's the #769 case where the char ratio
    # over-credits a shared substring ("Dani California"/"Californication") or
    # a stopword ("Under The Bridge"/"Around the World"). For single-word
    # titles there's no other word to share, so applying it would wrongly fail
    # legitimate stylized spellings ("Grey"/"Gray", "Tonite"/"Tonight",
    # "Thru"/"Through") that the char ratio rightly accepts. In that case defer
    # to the caller's existing char-similarity floor instead of force-failing.
    if max(len(ta), len(tb)) < 2 or not ta or not tb:
        return True
    return not ta.isdisjoint(tb)


_QUALIFIER_RE = re.compile(r"[\(\[]([^\)\]]*)[\)\]]")


def strip_redundant_context_qualifiers(title: str, *context_texts: str) -> str:
    """Remove parenthetical/bracket qualifiers that merely restate known context.

    A qualifier whose text appears (word-bounded) in one of ``context_texts``
    — typically the release's album title, or the other side of a comparison —
    is album context, not a version difference. #808: the wishlist held
    'Champagne Supernova (OurVinyl Sessions)' while the library track was the
    bare 'Champagne Supernova' on the album '… (OurVinyl Sessions)'; the
    qualifier restated the album, but the length-ratio penalty treated the
    pair as different songs and the cleanup never recognised the owned
    edition. Version markers that do NOT appear in any context ('(Live)',
    '(Remix)' on a studio album) are kept, so their mismatch penalty stands.
    """
    if not title:
        return title

    contexts = [c.casefold() for c in context_texts if c]
    if not contexts:
        return title

    def _drop(match: re.Match) -> str:
        inner = match.group(1).strip().casefold()
        if not inner:
            return " "
        pattern = r"\b" + re.escape(inner) + r"\b"
        for ctx in contexts:
            if re.search(pattern, ctx):
                return " "
        return match.group(0)

    out = _QUALIFIER_RE.sub(_drop, title)
    return re.sub(r"\s+", " ", out).strip()


# Qualifier tokens that mark a genuinely DIFFERENT recording/cut — these must
# keep blocking a match. Union of the matching-engine keyword lists plus the
# Spanish markers seen in real libraries (#825: 'En Directo…', 'Versión 1988',
# 'Dueto 2007'). Titles reaching the matcher are unidecode-normalized, so the
# ASCII forms ('version') cover the accented ones ('versión').
_VERSION_MARKER_TOKENS = frozenset({
    # English
    "remix", "mix", "rmx", "live", "acoustic", "unplugged", "instrumental",
    "karaoke", "demo", "demos", "edit", "version", "versions", "remaster",
    "remastered", "slowed", "reverb", "sped", "spedup", "speedup", "extended",
    "club", "mashup", "bootleg", "cover", "covers", "reprise", "session",
    "sessions", "mono", "stereo", "duet", "rework", "dub", "vip", "single",
    "radio", "alt", "alternate", "alternative", "take", "edition", "orchestral",
    "symphonic", "piano", "acapella", "cappella", "nightcore",
    # Distinct-track qualifiers — '(Interlude)' etc. are SEPARATE short tracks
    # that share the base name with the full song; never treat as subtitles.
    "interlude", "intro", "outro", "skit", "freestyle", "medley", "snippet",
    # Part/volume markers whose number can be non-numeric ('Pt. II') — the
    # digit guard below only catches actual digits.
    "pt", "part", "vol", "ii", "iii", "iv", "vi", "vii", "viii",
    # Spanish (unidecode-normalized; 'versión' → 'version' is covered above)
    "directo", "vivo", "dueto",
})


def strip_subtitle_qualifiers(title: str, other_title: str) -> str:
    """Remove bracketed qualifiers that are SUBTITLES, not version markers.

    #825 (carlosjfcasero): the wishlist held 'Llamando a la tierra (Serenade
    From the Stars)' — the song's official subtitle — while the library track
    was the bare 'Llamando a la tierra'. The qualifier appears in no album or
    counterpart title, so :func:`strip_redundant_context_qualifiers` keeps it,
    and the length-ratio penalty then crushes an obviously-same song to ~0.14.
    The sync matcher reported it missing on every run (re-adding it to the
    wishlist) and the cleanup — same matcher — could never remove it.

    A qualifier is stripped only when ALL of:
      * its text does not appear in ``other_title`` (if it does, the direct
        comparison already handles it);
      * it contains no version-marker token ('(Live)', '(Versión 1988)',
        '(Dueto 2007)' keep blocking — they are different recordings);
      * it introduces no digit token absent from ``other_title`` ('(Pt. 2)',
        '(2007)' are different releases, never subtitles).

    Inputs should be normalized the same way the caller compares them
    (lowercased / unidecode'd), like strip_redundant_context_qualifiers.
    """
    if not title:
        return title

    other = (other_title or "").casefold()
    other_tokens = set(_TOKEN_RE.findall(other))

    def _drop(match: re.Match) -> str:
        inner = match.group(1).strip().casefold()
        if not inner:
            return " "
        # Restated in the counterpart title — leave for the direct comparison.
        if re.search(r"\b" + re.escape(inner) + r"\b", other):
            return match.group(0)
        tokens = _TOKEN_RE.findall(inner)
        if any(t in _VERSION_MARKER_TOKENS for t in tokens):
            return match.group(0)
        if any(any(c.isdigit() for c in t) and t not in other_tokens for t in tokens):
            return match.group(0)
        return " "

    out = _QUALIFIER_RE.sub(_drop, title)
    return re.sub(r"\s+", " ", out).strip()


def numeric_tokens_differ(title_a: str, title_b: str) -> bool:
    """True when the digit-bearing tokens of two titles differ — 'Vol.4' vs
    'Vol.4.5', 'Album' vs 'Album 2'. A numeric difference is a different
    release (volume / part / sequel), never a '(Deluxe)'-style suffix:
    string similarity ('Vol.4' vs 'Vol.4.5' = 0.97) and token-subset checks
    both wave these through, which hung volume 4.5's cover art on volume 4
    (Sokhi). Shared digits on both sides ('1989' vs '1989 (Deluxe)') are
    fine."""
    def _digit_tokens(text: str) -> frozenset:
        tokens = re.sub(r"[^a-z0-9]+", " ", (text or "").casefold()).split()
        return frozenset(t for t in tokens if any(c.isdigit() for c in t))

    return _digit_tokens(title_a) != _digit_tokens(title_b)


__all__ = [
    "titles_plausibly_same",
    "strip_redundant_context_qualifiers",
    "strip_subtitle_qualifiers",
    "numeric_tokens_differ",
]