[music] Add better track fetching

vrobbler/apps/music/musicbrainz.py

@@ -1,4 +1,5 @@
 import logging
+import re
 from datetime import datetime
 
 import musicbrainzngs
@@ -6,6 +7,149 @@ from dateutil.parser import parse
 
 logger = logging.getLogger(__name__)
 
+TRACK_SUFFIXES = [
+    r" - Bonus Track",
+    r" - Bonus",
+    r" - Remix",
+    r" - Live",
+    r" - Radio Edit",
+    r" - Extended Mix",
+    r" - Instrumental",
+    r" - Acoustic",
+    r" - Mix",
+    r" - Edit",
+    r" \[Explicit\]",
+]
+
+FEATURED_PATTERNS = [
+    (r"\(feat\. (.+?)\)", re.IGNORECASE),
+    (r"\(ft\. (.+?)\)", re.IGNORECASE),
+    (r"featuring (.+)", re.IGNORECASE),
+    (r"feat\. (.+)", re.IGNORECASE),
+    (r"ft\. (.+)", re.IGNORECASE),
+]
+
+
+def strip_track_suffixes(title: str) -> str:
+    for suffix in TRACK_SUFFIXES:
+        title = re.sub(suffix, "", title, flags=re.IGNORECASE)
+    return title.strip()
+
+
+def extract_featured_artists(title: str) -> tuple[str, list[str]]:
+    featured = []
+    cleaned = title
+    for pattern, flags in FEATURED_PATTERNS:
+        m = re.search(pattern, cleaned, flags)
+        if m:
+            featured.append(m.group(1).strip())
+            cleaned = re.sub(pattern, "", cleaned, flags=flags).strip()
+    return cleaned, featured
+
+
+def search_recordings(title: str, artist: str = "") -> list[dict]:
+    kwargs: dict = {"recording": title, "limit": 10}
+    if artist:
+        kwargs["artist"] = artist
+    try:
+        return musicbrainzngs.search_recordings(**kwargs).get("recording-list", [])
+    except (musicbrainzngs.WebServiceError, musicbrainzngs.ResponseError) as e:
+        logger.warning(f"MusicBrainz search error: {e}")
+        return []
+
+
+def pick_best_recording(
+    recordings: list[dict], title: str, artist: str = ""
+) -> tuple[dict | None, str]:
+    query_title = title.strip().casefold()
+    query_artist = artist.strip().casefold() if artist else ""
+    best_score: tuple[dict | None, str] = (None, "")
+
+    for rec in recordings:
+        rec_title = rec.get("title", "").strip()
+        score = int(rec.get("ext:score", 0))
+
+        if rec_title.casefold() != query_title:
+            if score >= 85 and not best_score[0]:
+                best_score = (rec, "score")
+            continue
+
+        if query_artist:
+            artist_credit = rec.get("artist-credit", [])
+            rec_artist = (
+                artist_credit[0]["artist"]["name"].strip()
+                if artist_credit
+                else ""
+            )
+            if rec_artist.casefold() != query_artist:
+                if score >= 85 and not best_score[0]:
+                    best_score = (rec, "score")
+                continue
+
+        return (rec, "exact")
+
+    return best_score
+
+
+def resolve_track(
+    title: str, artist: str = "", album: str = ""
+) -> tuple[dict | None, str]:
+    if not title:
+        return (None, "")
+
+    titles_to_try: list[tuple[str, str]] = [(title, "")]
+    stripped = strip_track_suffixes(title)
+    if stripped != title:
+        titles_to_try.append((stripped, "stripped-"))
+
+    for attempt_title, prefix in titles_to_try:
+        if album and not prefix:
+            try:
+                mbid, length = get_recording_mbid_exact(
+                    attempt_title, artist, album
+                )
+                return (
+                    {
+                        "recording_mbid": mbid,
+                        "length_ms": length,
+                    },
+                    "exact",
+                )
+            except Exception:
+                pass
+
+        if artist:
+            result = get_track_metadata_with_artist(attempt_title, artist)
+            if result and result.get("recording_mbid"):
+                return (result, f"{prefix}exact")
+
+            recordings = search_recordings(attempt_title, artist)
+            result, tag = pick_best_recording(recordings, attempt_title, artist)
+            if result:
+                length_ms = result.get("length")
+                return (
+                    {
+                        "recording_mbid": result["id"],
+                        "length_ms": int(length_ms) if length_ms else None,
+                    },
+                    f"{prefix}{tag}",
+                )
+
+        recordings = search_recordings(attempt_title)
+        result, tag = pick_best_recording(recordings, attempt_title)
+        if result:
+            tag = "title-only" if not prefix else f"{prefix}title-only"
+            length_ms = result.get("length")
+            return (
+                {
+                    "recording_mbid": result["id"],
+                    "length_ms": int(length_ms) if length_ms else None,
+                },
+                tag,
+            )
+
+    return (None, "")
+
 musicbrainzngs.set_useragent("Vrobbler", "1.0", "help@unbl.ink")