File management for the lost.

Content-addressable storage with a SQLite database to track everything. Files are deduplicated automatically and can live in multiple locations (local + cloud).

Local:

storage/
├── a3/
│   ├── b5/
│   │   ├── a3b5c7d9e1f2a4b6...           # actual file
│   │   └── a3b5c7d9e1f2a4b6....meta.json # optional sidecar backup

• Hash-based sharding (2 levels: 65,536 possible directories)
• ~15 files per directory on average with 1M files
• Same structure mirrored in cloud (S3/Backblaze B2)

Cloud: Same hash-based structure for consistency

CREATE TABLE files (
    hash TEXT PRIMARY KEY,           -- SHA-256 of file content
    size INTEGER,
    mime_type TEXT,
    original_filename TEXT,
    file_extension TEXT,
    created_at TIMESTAMP,
    modified_at TIMESTAMP,
    imported_at TIMESTAMP,
    
    -- Storage locations
    local_path TEXT,                 -- storage/a3/b5/a3b5c7d9...
    s3_url TEXT,                     -- s3://bucket/a3/b5/a3b5c7d9...
    
    -- Origin tracking
    original_paths TEXT,              -- Projects/MyWebsite/images/logo.png
    original_source TEXT,            -- "Dropbox", "iCloud", "OldMacDrive"
    
    -- Flexible data
    tags TEXT,                       -- ['Projects', 'MyWebsite', 'vacation']
    metadata TEXT                    -- EXIF, dimensions, AI analysis, etc.
);

Automatic deduplication - same file stored only once (by hash)
Multiple locations - track local, cloud, and legacy (Dropbox/iCloud) locations
Preserves context - original paths saved, auto-extracted as tags
Flexible metadata - JSON columns for file-specific data
Simple deployment - just SQLite + Python, no servers

Core operations:

1. Hash file → SHA-256
2. Check if hash exists in DB (dedupe)
3. Store in hash-sharded directory
4. Extract tags from original path
5. Record in database
6. Optionally sync to cloud

• One person system - SQLite is perfect, no complexity
• Gradual migration - can index legacy locations first, move files later
• Extensible - JSON lets you add features (AI analysis, etc.) without schema changes
• Resilient - easy DB backups
• Scales to your needs - handles 1TB/1M files easily

Simple, clean, and you're in full control.