feat: --repair flag to re-tag existing downloads from source metadata

Walks <root>/<artist>/<source>/ (known yt-dlp source folders only; skips Lidarr
album dirs), re-queries each file's source by the [id] in its filename, and fixes
tags (album/year/artist/title) via mutagen. Honors --dry-run for preview. CLI-only
(not the REST API). Fixes downloads that landed with missing album / wrong year.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

tests/test_repair.py

@@ -0,0 +1,118 @@
+import server.mf  # noqa: F401 — loads musicfetch, registers musicfetch_core
+import musicfetch_core as mf
+
+
+# ---- _is_source_dir ----
+def test_is_source_dir():
+    assert mf._is_source_dir("youtube") is True
+    assert mf._is_source_dir("soundcloud") is True
+    assert mf._is_source_dir("downloads") is True
+    assert mf._is_source_dir("Discovery") is False           # Lidarr album folder
+    assert mf._is_source_dir("Random Access Memories") is False
+    assert mf._is_source_dir("") is False
+
+
+# ---- _parse_track_file ----
+def test_parse_track_file():
+    assert mf._parse_track_file("Under My Skin [nGSNF2l44Zc].opus") == ("Under My Skin", "nGSNF2l44Zc")
+    assert mf._parse_track_file("Ignomon [2202690443].m4a") == ("Ignomon", "2202690443")
+    assert mf._parse_track_file("no-id-here.opus") is None
+    assert mf._parse_track_file("cover.jpg") is None
+
+
+# ---- _repair_probe_url ----
+def test_repair_probe_url():
+    assert mf._repair_probe_url("youtube", "vid") == "https://music.youtube.com/watch?v=vid"
+    assert mf._repair_probe_url("soundcloud", "123") == "https://api.soundcloud.com/tracks/123"
+    assert mf._repair_probe_url("bandcamp", "x") is None
+
+
+# ---- _desired_tags ----
+def test_desired_tags_full():
+    meta = {"artist": "Daft Punk", "title": "Aerodynamic", "album": "Discovery", "release_year": 2001}
+    assert mf._desired_tags(meta) == {"artist": "Daft Punk", "title": "Aerodynamic",
+                                      "album": "Discovery", "date": "2001"}
+
+
+def test_desired_tags_year_fallbacks_and_omits_empty():
+    meta = {"title": "T", "uploader": "Chan", "upload_date": "20230102"}  # no album, no release_year
+    out = mf._desired_tags(meta)
+    assert out["date"] == "2023"
+    assert out["title"] == "T"
+    assert out["artist"] == "Chan"
+    assert "album" not in out  # omitted when empty
+
+
+def test_desired_tags_drops_unknown_artist():
+    meta = {"title": "T"}  # get_artist_from_metadata -> "Unknown Artist"
+    assert "artist" not in mf._desired_tags(meta)
+
+
+# ---- repair_file (fake audio + mocked metadata) ----
+class _FakeAudio(dict):
+    def __init__(self, initial):
+        super().__init__(initial)
+        self.saved = False
+
+    def save(self):
+        self.saved = True
+
+
+def test_repair_file_writes_changed_fields(monkeypatch):
+    monkeypatch.setattr(mf, "run_yt_dlp_get_metadata",
+                        lambda url: {"artist": "Daft Punk", "title": "Aerodynamic",
+                                     "album": "Discovery", "release_year": 2001})
+    audio = _FakeAudio({"artist": ["Daft Punk"], "title": ["Aerodynamic"]})  # album/date missing
+    monkeypatch.setattr(mf, "_open_audio", lambda path: (audio, None))
+    changed = mf.repair_file("X/youtube/Aerodynamic [vid].opus", "youtube", dry_run=False)
+    assert set(changed) == {"album=Discovery", "date=2001"}
+    assert audio["album"] == ["Discovery"]
+    assert audio["date"] == ["2001"]
+    assert audio.saved is True
+
+
+def test_repair_file_dry_run_writes_nothing(monkeypatch):
+    monkeypatch.setattr(mf, "run_yt_dlp_get_metadata",
+                        lambda url: {"artist": "A", "title": "T", "album": "Alb", "release_year": 2020})
+    audio = _FakeAudio({})
+    monkeypatch.setattr(mf, "_open_audio", lambda path: (audio, None))
+    changed = mf.repair_file("X/youtube/T [vid].opus", "youtube", dry_run=True)
+    assert changed  # reports would-change
+    assert audio == {}  # nothing written
+    assert audio.saved is False
+
+
+def test_repair_file_skips_unparseable(monkeypatch):
+    called = {"meta": False}
+    monkeypatch.setattr(mf, "run_yt_dlp_get_metadata",
+                        lambda url: called.update(meta=True) or {})
+    assert mf.repair_file("X/youtube/no-id.opus", "youtube", dry_run=False) == []
+    assert called["meta"] is False  # never hit the network
+
+
+def test_repair_file_skips_unqueryable_source(monkeypatch):
+    monkeypatch.setattr(mf, "run_yt_dlp_get_metadata", lambda url: {"title": "x"})
+    assert mf.repair_file("X/bandcamp/T [id].m4a", "bandcamp", dry_run=False) == []
+
+
+# ---- repair_library (real temp tree, repair_file mocked) ----
+def test_repair_library_scans_only_source_dirs(tmp_path, monkeypatch):
+    root = tmp_path
+    (root / "Daft Punk" / "youtube").mkdir(parents=True)
+    (root / "Daft Punk" / "youtube" / "Aerodynamic [vid].opus").write_text("x")
+    (root / "Daft Punk" / "Discovery").mkdir(parents=True)            # Lidarr album -> skip
+    (root / "Daft Punk" / "Discovery" / "Aerodynamic.flac").write_text("x")
+    (root / "Ephixa" / "soundcloud").mkdir(parents=True)
+    (root / "Ephixa" / "soundcloud" / "Ignomon [123].m4a").write_text("x")
+
+    visited = []
+    monkeypatch.setattr(mf, "repair_file",
+                        lambda path, source, dry_run: visited.append((source, path)) or ["album=X"])
+    scanned, changed = mf.repair_library(str(root), dry_run=False)
+    assert scanned == 2 and changed == 2
+    sources = sorted(s for s, _ in visited)
+    assert sources == ["soundcloud", "youtube"]   # Discovery album folder skipped
+
+
+def test_repair_library_missing_root(monkeypatch):
+    assert mf.repair_library("/no/such/dir", dry_run=False) == (0, 0)