The union-find algorithm caused transitivity chains: A≈B and B≈C merged
A, B, C into one group even if A and C were completely different photos,
producing the 5000+ "duplicate" groups that were wrong.
- getGroups now returns individual pairs (2 photos each) — what was
actually flagged by the pHash comparison, no chaining
- scan() clears all existing pairs on each run so the result exactly
matches the current threshold (no stale loose pairs from prior scans)
- Default dedupe_threshold lowered from 10 → 5 (fewer false positives;
distance ≤5 means the images are genuinely near-identical)
Run "Scan library" once to replace the old results with clean pairs.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Replace BigInt hammingDistance with 32-bit popcount — safer (no throw
on malformed hashes) and ~10× faster
- Make scan() async with setImmediate chunking so the event loop stays
responsive on large libraries
- Batch getGroups() photo lookup in slices of 900 to stay under SQLite's
999-variable limit
- Add POST /api/duplicates/keep-best endpoint: scores by resolution then
filesize, soft-deletes the losers, dismisses the pair
- DuplicateGroup: green "Keep" badge on the best photo, green "Keep best"
button in footer
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Scans MUSIC_DIR (mounted at /music in Docker) with music-metadata to
read embedded BPM/genre tags and classify tracks as upbeat/cinematic/
dreamy. Falls back to untagged general pool when mood bucket is thin.
Slideshow plays real audio via <audio> element with range-request
streaming; Web Audio synth is kept as fallback when MUSIC_DIR is unset.
Shows now-playing title/artist in the slideshow bottom bar.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Albums tab with manual album CRUD (create/delete, cover photo grid)
and Smart Albums auto-generated by year, city, and camera model using
SQLite window functions — no extra storage, computed from existing data.
Lightbox gets a download-original button (streams raw file with
Content-Disposition: attachment).
DetailPanel Camera section expands to show ISO, aperture, shutter speed,
focal length, and lens model extracted at ingest time; five new DB columns
added via safe ALTER TABLE migration.
Bulk selection: hover a photo to reveal a circle checkbox; selecting
any photo activates a floating action bar for star/unstar and add-to-album
across the whole selection. PhotoCard wrapped in React.memo to limit
re-renders to only affected cards.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The pick option creates an allowlist for the merged EXIF output, which
drops latitude/longitude even when gps:true is set. Fixed by calling
exifr.gps() as a separate step (bypasses the pick filter entirely).
Also adds a startup backfill that re-extracts GPS for already-indexed
photos where gps_lat IS NULL, including reverse geocoding for place names.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Places:
- worker/geocode.js: Nominatim reverse geocoder, 1 req/sec rate limit,
in-memory cache keyed to 3-decimal-place coordinates
- Ingest calls reverseGeocode() for any photo with GPS data; stores
place_city and place_country on the photo row
- DB migrations add place_city / place_country columns + index on place_city
- Stats API returns cities grouped by count
- buildFilter() supports ?city= parameter
- FilterSidebar shows a Places section with city list + country sublabel
Folder browser:
- GET /api/fs/browse?path=… lists subdirectories (path-normalized to prevent traversal)
- Settings panel: 📁 button opens inline directory browser below the add
input — navigate with up arrow / directory clicks, Select fills the input
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The CREATE INDEX on photos(starred) ran inside the main db.exec() block,
which fires before the ALTER TABLE migration that adds the starred column.
On databases created before the starred feature, this crashed the process
on startup with "no such column: starred" — causing the worker container
to exit silently with no logs.
Move the index creation to after the migration so the column always exists.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Settings panel (⚙ in header):
- Watch Folders: add/remove paths, toggle enabled; worker polls every 30s
and calls watcher.add/unwatch without restart
- Ingestion: thumb/preview size and quality, applied per-file at ingest time
- Duplicate Detection: configurable Hamming distance threshold
- Danger Zone: clear all dismissed duplicate pairs
Multi-folder scanning:
- watch_folders table replaces single PHOTO_DIR env var
- PHOTO_DIR env var seeds the first entry on startup
- settings table with defaults for all tunable values
- Worker reads settings from DB on each drain cycle so changes are live
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Load all known filepaths into an in-memory Set on startup (one bulk query).
Chokidar's initial 'add' burst now hits an O(1) Set lookup instead of going
through the serial drain queue with a per-file DB round-trip — a 100k-photo
library restarts in seconds rather than re-processing every file.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Left filter sidebar: All Photos / Favorites / Date by Year / Tags / Collections
with counts, See more/less toggles, and blue active-row highlight
- SSE ingest badge in header: pulsing dot + filename + indexed count during scan;
auto-refreshes the library when a scan finishes
- Starred/Favorites: starred column on photos, star button in lightbox toolbar,
Favorites row in sidebar, ?starred=1 filter on /api/photos
- GET /api/library/stats endpoint (totals, years, tags, collections)
- GET /api/ingest/stream SSE endpoint (polls ingest_status table every 1.5s)
- Worker writes ingest_status (active, current file, scanned count) to SQLite
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
