Add MusicWorkshop application

backend/tests/test_metadata_normalization.py

@@ -0,0 +1,95 @@
+import os
+import tempfile
+import unittest
+from pathlib import Path
+
+os.environ['MUSIC_WORKSHOP_DB_PATH'] = str(
+  Path(tempfile.gettempdir()) / f'music_workshop_metadata_normalization_{next(tempfile._get_candidate_names())}.db'
+)
+
+from backend.app.metadata_normalization import MetadataNormalizationService, can_ingest_metadata, parse_artist_string
+from backend.app.task_store import TaskStore
+
+
+class MetadataNormalizationTests(unittest.TestCase):
+  def setUp(self):
+    self.db_path = Path(os.environ['MUSIC_WORKSHOP_DB_PATH'])
+    if self.db_path.exists():
+      self.db_path.unlink()
+    self.task_store = TaskStore(self.db_path)
+    self.task = self.task_store.create_task_if_idle(
+      {
+        'input': '/tmp/input',
+        'output': '/tmp/output',
+        'trash': '/tmp/trash'
+      }
+    )
+    self.service = MetadataNormalizationService(self.task_store)
+
+  def test_parse_artist_string_supports_common_delimiters(self):
+    self.assertEqual(parse_artist_string('A / B')['tokens'], ['A', 'B'])
+    self.assertEqual(parse_artist_string('A; B')['tokens'], ['A', 'B'])
+    self.assertEqual(parse_artist_string('A & B')['tokens'], ['A', 'B'])
+    self.assertEqual(parse_artist_string('A feat. B')['tokens'], ['A', 'B'])
+    self.assertEqual(parse_artist_string('A、B')['tokens'], ['A', 'B'])
+
+  def test_single_artist_album_derives_album_artist(self):
+    item = self._insert_item('track-01.flac', {'title': 'Song 1', 'artist': 'Artist A', 'album': 'Album X'})
+    self._insert_item('track-02.flac', {'title': 'Song 2', 'artist': 'Artist A', 'album': 'Album X'})
+
+    normalized = self.service.normalize_item(item)
+    self.assertEqual(normalized['album_artist'], 'Artist A')
+    self.assertEqual(normalized['normalization_strategy'], 'single_artist')
+    self.assertTrue(can_ingest_metadata({**normalized, 'title': 'Song 1'}))
+
+  def test_feat_album_uses_dominant_primary_artist(self):
+    item = self._insert_item('track-01.flac', {'title': 'Song 1', 'artist': 'Artist A feat. Guest', 'album': 'Album X'})
+    self._insert_item('track-02.flac', {'title': 'Song 2', 'artist': 'Artist A', 'album': 'Album X'})
+    self._insert_item('track-03.flac', {'title': 'Song 3', 'artist': 'Artist A & Another', 'album': 'Album X'})
+
+    normalized = self.service.normalize_item(item)
+    self.assertEqual(normalized['album_artist'], 'Artist A')
+    self.assertEqual(normalized['normalization_strategy'], 'main_artist_feat')
+
+  def test_compilation_album_sets_various_artists(self):
+    item = self._insert_item('track-01.flac', {'title': 'Song 1', 'artist': 'Artist A', 'album': 'Top Hits 2025'})
+    self._insert_item('track-02.flac', {'title': 'Song 2', 'artist': 'Artist B', 'album': 'Top Hits 2025'})
+    self._insert_item('track-03.flac', {'title': 'Song 3', 'artist': 'Artist C', 'album': 'Top Hits 2025'})
+
+    normalized = self.service.normalize_item(item)
+    self.assertEqual(normalized['album_artist'], 'Various Artists')
+    self.assertEqual(normalized['compilation'], 1)
+    self.assertEqual(normalized['normalization_strategy'], 'compilation')
+
+  def test_existing_album_artist_is_preserved(self):
+    item = self._insert_item(
+      'track-01.flac',
+      {'title': 'Song 1', 'artist': '阿信', 'album': 'Solo Album', 'album_artist': '五月天'}
+    )
+
+    normalized = self.service.normalize_item(item)
+    self.assertEqual(normalized['album_artist'], '五月天')
+    self.assertEqual(normalized['normalization_strategy'], 'source_preserved')
+
+  def _insert_item(self, filename: str, matched_metadata_json: dict):
+    return self.task_store.insert_task_item(
+      self.task['task_id'],
+      original_path=f'/tmp/input/{filename}',
+      current_file_path=f'/tmp/input/{filename}',
+      relative_path=f'Artist/Album/{filename}',
+      filename=filename,
+      extension='.flac',
+      size_bytes=123456,
+      modified_at='2024-01-01T00:00:00Z',
+      local_cover=None,
+      local_lyric=None,
+      scan_status='queued',
+      scan_reason=None,
+      scan_message=None,
+      match_status='matched_fallback',
+      matched_metadata_json=matched_metadata_json
+    )
+
+
+if __name__ == '__main__':
+  unittest.main()