From adbdda7b0eeecaaabce5f5b2fa52c026ff84400a Mon Sep 17 00:00:00 2001
From: BoulderBadgeDad <batmanesignal@gmail.com>
Date: Thu, 4 Jun 2026 10:21:30 -0700
Subject: [PATCH] Library Re-tag: add light/full depth setting, default source
 to active, fix dropdown CSS

- depth setting (light = core tags + matched source ids; full = same
  multi-source enrichment cascade a fresh download gets, run additively
  via embed_source_ids). Threaded through scan/finding/auto-apply and the
  repair_worker fix handler.
- source now defaults to 'auto' (= your source priority / active source)
  instead of blank.
- give native <option> popups a solid dark background (were white-on-white).
- tests for full-depth full_meta payload + enrich invocation + light no-op.
---
 core/repair_jobs/library_retag.py | 86 ++++++++++++++++++++++++++++---
 core/repair_worker.py             |  8 ++-
 tests/test_library_retag_job.py   | 68 ++++++++++++++++++++++++
 webui/static/style.css            |  7 +++
 4 files changed, 159 insertions(+), 10 deletions(-)

diff --git a/core/repair_jobs/library_retag.py b/core/repair_jobs/library_retag.py
index 6219e6c0..9bd0b463 100644
--- a/core/repair_jobs/library_retag.py
+++ b/core/repair_jobs/library_retag.py
@@ -45,7 +45,33 @@ def _read_current_tags(file_path):
         return {}
 
 
-def apply_track_plans(track_plans, cover_action=None, cover_url=None) -> dict:
+def _run_full_enrich(file_path, full_meta) -> bool:
+    """'full' depth: run the same multi-source enrichment a fresh download gets
+    (MusicBrainz/Deezer/AudioDB/Tidal/… via embed_source_ids), ADDITIVELY — it
+    adds rich frames without clearing existing tags. Slow + API-heavy per track.
+    """
+    if not full_meta:
+        return False
+    try:
+        from core.metadata.common import get_mutagen_symbols
+        from core.metadata.source import embed_source_ids
+        symbols = get_mutagen_symbols()
+        if not symbols:
+            return False
+        audio = symbols.File(file_path)
+        if audio is None:
+            return False
+        if getattr(audio, 'tags', None) is None and hasattr(audio, 'add_tags'):
+            audio.add_tags()
+        embed_source_ids(audio, full_meta, context=None, runtime=None)
+        audio.save()
+        return True
+    except Exception as e:
+        logger.warning("full enrich failed for %s: %s", file_path, e)
+        return False
+
+
+def apply_track_plans(track_plans, cover_action=None, cover_url=None, full=False) -> dict:
     """Write each plan's tags in place (+ optionally embed/refresh cover art),
     reusing tag_writer.write_tags_to_file. ``file_path`` on each plan must be a
     real, reachable path (caller resolves Docker paths). Shared by the dry-run=
@@ -76,6 +102,8 @@ def apply_track_plans(track_plans, cover_action=None, cover_url=None) -> dict:
             if res.get('success'):
                 result['written'] += 1
                 last_dir = _os.path.dirname(fp)
+                if full and tp.get('full_meta'):
+                    _run_full_enrich(fp, tp['full_meta'])
             else:
                 result['failed'] += 1
         except Exception as e:
@@ -114,6 +142,34 @@ def _add_source_ids(db_data, source, album_source_id, source_track):
             db_data[track_key] = tid
 
 
+_FULL_META_ID_KEYS = (
+    'spotify_album_id', 'spotify_track_id',
+    'itunes_album_id', 'itunes_track_id',
+    'musicbrainz_release_id', 'musicbrainz_recording_id',
+    'deezer_id',
+)
+
+
+def _build_full_meta(db_data, src, album_title, artist_name, lib_title):
+    """Metadata dict for the 'full' depth enrichment cascade. Carries the matched
+    source's ids so embed_source_ids resolves the right entity instead of guessing
+    by name."""
+    src_title = None
+    for k in ('name', 'title', 'track_name'):
+        v = src.get(k) if isinstance(src, dict) else getattr(src, k, None)
+        if v:
+            src_title = v
+            break
+    meta = {
+        'title': src_title or lib_title,
+        'album': album_title,
+        'album_artist': artist_name,
+        'artist': artist_name,
+    }
+    meta.update({k: db_data[k] for k in _FULL_META_ID_KEYS if db_data.get(k)})
+    return meta
+
+
 def _track_list(result):
     """Normalize a get_album_tracks result into a plain list of track items."""
     if result is None:
@@ -144,6 +200,10 @@ class LibraryRetagJob(RepairJob):
         'comes from. Each finding lists every tag that would change (old -> new) per '
         'track so you can review before applying — nothing is written until you do.\n\n'
         'Settings:\n'
+        '- Depth: "light" writes the core tags + the matched source\'s ids (fast, '
+        'additive). "full" also runs the same multi-source enrichment a fresh '
+        'download gets (MusicBrainz / Deezer / AudioDB / Tidal / etc. — BPM, ISRC, '
+        'lyrics, moods, …); much richer but slower and API-heavy on a big library.\n'
         '- Dry run (default ON): only create findings to review; nothing is written. '
         'Turn it off to auto-apply on scan.\n'
         '- Mode: "overwrite" rewrites every field the source provides; "fill_missing" '
@@ -156,14 +216,16 @@ class LibraryRetagJob(RepairJob):
     default_interval_hours = 168
     default_settings = {
         'dry_run': True,
+        'depth': 'light',
         'mode': MODE_OVERWRITE,
         'cover_art': 'replace',
-        'source': '',
+        'source': 'auto',
     }
     setting_options = {
+        'depth': ['light', 'full'],
         'mode': [MODE_OVERWRITE, MODE_FILL_MISSING],
         'cover_art': ['replace', 'fill_missing', 'skip'],
-        'source': ['', 'spotify', 'itunes', 'deezer', 'musicbrainz'],
+        'source': ['auto', 'spotify', 'itunes', 'deezer', 'musicbrainz'],
     }
     auto_fix = True
 
@@ -186,6 +248,7 @@ class LibraryRetagJob(RepairJob):
         mode = settings.get('mode', MODE_OVERWRITE)
         cover_mode = settings.get('cover_art', 'replace')
         dry_run = settings.get('dry_run', True)
+        depth = settings.get('depth', 'light')
         source_order = self._source_order(settings)
         if not source_order:
             logger.warning("Library re-tag: no usable metadata sources configured")
@@ -233,7 +296,7 @@ class LibraryRetagJob(RepairJob):
 
             try:
                 self._scan_album(context, result, album_id, album_title, artist_name,
-                                 source, album_source_id, mode, cover_mode, dry_run)
+                                 source, album_source_id, mode, cover_mode, dry_run, depth)
             except Exception as e:
                 logger.debug("Library re-tag: album %s failed: %s", album_id, e)
                 result.errors += 1
@@ -245,7 +308,7 @@ class LibraryRetagJob(RepairJob):
         return result
 
     def _scan_album(self, context, result, album_id, album_title, artist_name,
-                    source, album_source_id, mode, cover_mode, dry_run=True):
+                    source, album_source_id, mode, cover_mode, dry_run=True, depth='light'):
         # Local tracks for this album.
         with context.db._get_connection() as conn:
             cur = conn.cursor()
@@ -297,13 +360,17 @@ class LibraryRetagJob(RepairJob):
             if plan['changes'] or cover_action:
                 db_data = plan['db_data']
                 _add_source_ids(db_data, source, album_source_id, src)
-                track_plans.append({
+                tp = {
                     'file_path': lib['file_path'],
                     'track_id': lib['id'],
                     'title': lib['title'],
                     'changes': plan['changes'],
                     'db_data': db_data,
-                })
+                }
+                if depth == 'full':
+                    tp['full_meta'] = _build_full_meta(
+                        db_data, src, album_title, artist_name, lib['title'])
+                track_plans.append(tp)
 
         tag_change_tracks = sum(1 for tp in track_plans if tp['changes'])
         if not tag_change_tracks and not cover_action:
@@ -313,7 +380,7 @@ class LibraryRetagJob(RepairJob):
         # Not dry-run: apply the tags in place now (the track paths were already
         # isfile-checked above) and count it as an auto-fix — no finding.
         if not dry_run:
-            res = apply_track_plans(track_plans, cover_action, cover_url)
+            res = apply_track_plans(track_plans, cover_action, cover_url, full=(depth == 'full'))
             if res['written'] or res['cover_written']:
                 result.auto_fixed += 1
             else:
@@ -326,6 +393,8 @@ class LibraryRetagJob(RepairJob):
             summary_bits.append(f"{tag_change_tracks} track(s), {total_changes} tag change(s)")
         if cover_action:
             summary_bits.append(f"cover art ({cover_action})")
+        if depth == 'full':
+            summary_bits.append("full multi-source enrichment")
         desc = (f'Album "{album_title}" by {artist_name or "Unknown"} would be re-tagged from '
                 f'{source} ({", ".join(summary_bits)}).')
         if unmatched:
@@ -347,6 +416,7 @@ class LibraryRetagJob(RepairJob):
                     'artist': artist_name,
                     'source': source,
                     'album_source_id': album_source_id,
+                    'depth': depth,
                     'mode': mode,
                     'cover_mode': cover_mode,
                     'cover_url': cover_url,
diff --git a/core/repair_worker.py b/core/repair_worker.py
index 2ddca93d..74d3a634 100644
--- a/core/repair_worker.py
+++ b/core/repair_worker.py
@@ -1375,10 +1375,14 @@ class RepairWorker:
                 continue
             rp = _resolve_file_path(raw, self.transfer_folder, download_folder,
                                     config_manager=self._config_manager) or raw
-            resolved_plans.append({'file_path': rp, 'db_data': t.get('db_data') or {}})
+            plan = {'file_path': rp, 'db_data': t.get('db_data') or {}}
+            if t.get('full_meta'):
+                plan['full_meta'] = t['full_meta']
+            resolved_plans.append(plan)
 
         from core.repair_jobs.library_retag import apply_track_plans
-        res = apply_track_plans(resolved_plans, details.get('cover_action'), details.get('cover_url'))
+        res = apply_track_plans(resolved_plans, details.get('cover_action'), details.get('cover_url'),
+                                full=(details.get('depth') == 'full'))
 
         if res['written'] == 0 and not res['cover_written']:
             return {'success': False,
diff --git a/tests/test_library_retag_job.py b/tests/test_library_retag_job.py
index 58a88960..46f8456e 100644
--- a/tests/test_library_retag_job.py
+++ b/tests/test_library_retag_job.py
@@ -111,6 +111,74 @@ def test_scan_dry_run_off_auto_applies_no_finding(tmp_path, monkeypatch):
     assert writes and writes[0]['title'] == 'Real Title'   # actually wrote
 
 
+def test_scan_full_depth_attaches_full_meta_to_finding(tmp_path, monkeypatch):
+    """depth=full: each track plan carries a full_meta dict (title/album/artist +
+    source ids) for the enrichment cascade, and details record the depth."""
+    track = tmp_path / 'track.flac'; track.write_bytes(b'')
+    conn = _db_with_album(str(tmp_path / 'm.db'), str(track), current_title='Old Title')
+    ctx = _context(conn, {'mode': 'overwrite', 'cover_art': 'skip', 'source': 'spotify', 'depth': 'full'})
+    _patch_source(monkeypatch, {
+        'title': 'Old Title', 'album_artist': 'Real Artist', 'album': 'Real Album',
+        'year': '2021', 'genre': 'Rock', 'track_number': 1, 'disc_number': 1,
+    })
+
+    result = lr.LibraryRetagJob().scan(ctx)
+
+    assert result.findings_created == 1
+    d = ctx.findings[0]['details']
+    assert d['depth'] == 'full'
+    fm = d['tracks'][0]['full_meta']
+    assert fm['title'] == 'Real Title'
+    assert fm['album'] == 'Real Album'
+    assert fm['album_artist'] == 'Real Artist'
+    assert fm['spotify_album_id'] == 'sp_alb'
+    assert fm['spotify_track_id'] == 'sp_trk'
+
+
+def test_scan_full_depth_auto_apply_runs_enrich(tmp_path, monkeypatch):
+    """depth=full + dry_run off: after the light write, the full enrichment
+    cascade runs once per written track."""
+    track = tmp_path / 'track.flac'; track.write_bytes(b'')
+    conn = _db_with_album(str(tmp_path / 'm.db'), str(track), current_title='Old Title')
+    ctx = _context(conn, {'mode': 'overwrite', 'cover_art': 'skip', 'source': 'spotify',
+                          'depth': 'full', 'dry_run': False})
+    _patch_source(monkeypatch, {
+        'title': 'Old Title', 'album_artist': 'Real Artist', 'album': 'Real Album',
+        'year': '2021', 'genre': 'Rock', 'track_number': 1, 'disc_number': 1,
+    })
+    monkeypatch.setattr('core.tag_writer.write_tags_to_file',
+                        lambda fp, db_data, **k: {'success': True})
+    enriched = []
+    monkeypatch.setattr(lr, '_run_full_enrich',
+                        lambda fp, meta: enriched.append((fp, meta)) or True)
+
+    result = lr.LibraryRetagJob().scan(ctx)
+
+    assert result.auto_fixed == 1
+    assert len(enriched) == 1
+    assert enriched[0][1]['spotify_track_id'] == 'sp_trk'
+
+
+def test_scan_light_depth_does_not_run_enrich(tmp_path, monkeypatch):
+    """depth=light (default): no full_meta, enrichment cascade never invoked."""
+    track = tmp_path / 'track.flac'; track.write_bytes(b'')
+    conn = _db_with_album(str(tmp_path / 'm.db'), str(track), current_title='Old Title')
+    ctx = _context(conn, {'mode': 'overwrite', 'cover_art': 'skip', 'source': 'spotify',
+                          'dry_run': False})  # depth defaults to light
+    _patch_source(monkeypatch, {
+        'title': 'Old Title', 'album_artist': 'Real Artist', 'album': 'Real Album',
+        'year': '2021', 'genre': 'Rock', 'track_number': 1, 'disc_number': 1,
+    })
+    monkeypatch.setattr('core.tag_writer.write_tags_to_file',
+                        lambda fp, db_data, **k: {'success': True})
+    enriched = []
+    monkeypatch.setattr(lr, '_run_full_enrich',
+                        lambda fp, meta: enriched.append(fp) or True)
+
+    lr.LibraryRetagJob().scan(ctx)
+    assert enriched == []
+
+
 def test_scan_skips_album_already_correct(tmp_path, monkeypatch):
     track = tmp_path / 'track.flac'; track.write_bytes(b'')
     conn = _db_with_album(str(tmp_path / 'm.db'), str(track), current_title='Real Title')
diff --git a/webui/static/style.css b/webui/static/style.css
index 8c521124..36d9919f 100644
--- a/webui/static/style.css
+++ b/webui/static/style.css
@@ -52587,6 +52587,13 @@ tr.tag-diff-same {
     accent-color: var(--accent-color, #6366f1);
 }
 
+/* Native <option> popups don't inherit the translucent select background, so
+   they rendered as white-on-white. Give the options a solid dark background. */
+.repair-setting-input option {
+    background: #1e1e24;
+    color: #fff;
+}
+
 .repair-save-settings-btn {
     margin-top: 10px;
     background: var(--accent-color, #6366f1);