- Add MtgonlineCard model (local card data mirror in mtgonline DB)
- Create user_deck.py models (UserDeck, UserDeckCard, DeckPrecedent, CardSuggestion)
- Create user_deck_schemas.py with Pydantic schemas
- Update decks.py router to use local MtgonlineCard instead of cross-DB MtgCard
- Add migration 002 (user deck building tables)
- Add migration 003 (mtgonline_cards table)
- All card lookups now use local mirror for fast queries
- No cross-DB joins in deckbuilding endpoints
- Handoff.md: document 16-table user data schema, Alembic migrations,
and all API endpoints (replays, cards, groups, networks, preferences, activity)
- Handoff.md: consolidate state.json reference to project root
- Roadmap.md: restructure phases to reflect completed work
Phase 1: Backend Foundation (complete)
Phase 2: User Data Schema & API (complete) - 16 tables, 7 routers
Phase 3: Frontend Development (complete)
Phase 4: Testing & Deployment (pending)
- Add Alembic migration setup with async configuration
- Create 16 user data models (users, decks, cards, replays, etc.)
- Implement comprehensive API endpoints with JWT auth
- Add replay, card collection, group, network, preferences, and activity log routers
- Include API documentation and migration test plan
- Update Dockerfile to run migrations on startup
- Created MtgCardMirror and DeckCardLink models in mirror_models.py
- Created card_mirror_service.py with sync_mirrors functionality
- Added sync_mirrors() method to mtgjson_manager.py
- Added mirror_get_db() dependency to database.py
- Added DecklistFile.status column (DRAUGHT/FINAL)
- Updated DeckCreate schema with status field
- Added DeckWithCardsResponse schema with card_count
- Updated deck router to query card mirrors and return card counts
- Added plain text deck content support
- Updated root README with current architecture (dual PostgreSQL, Redis, MTGJSON pipeline)
- Created backend/README.md with detailed architecture, database setup, and troubleshooting
- Updated state.json to reflect completed documentation phase
- Changed REQUIRED_FILES to REQUIRED_JSON_FILES and REQUIRED_ZIP_FILES
- AllSetFiles.zip kept as zip (only available compressed on MTGJSON)
- All other files downloaded as .json (no compression)
- Removed gzip import and unpacking logic for .gz files
- Updated EXPECTED_MIN_SIZES to reflect actual .json file sizes
- Removed validation for non-existent files
- Created MTGJSONManager service for complete data lifecycle
- Handles download, unpack (gzip/zip), and PostgreSQL upsert
- ON CONFLICT DO UPDATE preserves existing data
- Startup triggers initial download on first container init
- Health check verifies MTG data exists in database
- Weekly refresh via MTG_REFRESH_INTERVAL_DAYS setting
- Updated docker-compose start_period to 600s for download time
- Fix MtgSet model to match database schema (removed created_at, added image column)
- Create load_mtgjson_data.py script to load AllSetFiles, AllPrintings.psql, and other MTGJSON data
- Create download_mtgjson_data.py script to download MTGJSON API data files
- Add SPEC_synergy-mapping-engine.md documentation
API endpoints are now working (200 OK) but database needs data loading via download_mtgjson_data.py
then load_mtgjson_data.py
- Added MTGJSON data downloader (downloads all MTGJSON API v5 files)
- Added MTGJSON data loader (imports data into PostgreSQL)
- Added MTGJSON data uploader (alternative upsert logic)
- Fixed route ordering in card_router.py (/sets before /{card_name})
- Added load_mtgjson_data.py entry point script
MTGJSON data sources:
- AllPrintings.psql.gz (main cards database)
- AllSetFiles.zip (set and card data)
- AllDeckFiles.zip (deck data)
- AllIdentifiers.json.gz (card identifiers)
- CardTypes.json.gz (card types)
- DeckList.json.gz (deck list metadata)
- Keywords.json.gz (card keywords)
- SetList.json.gz (set list metadata)
Note: MTGJSON set.json does NOT contain image URLs. Only cards have image_uris.
Sets have iconSvgUrl (SVG icons) but no raster image URLs.
- Converts MTGJSON v5 AllPrintings.json to PostgreSQL format
- Upserts data to mtgdata database using psycopg2
- Handles sets and cards with proper foreign key relationships
- Batch processing of 100 cards at a time
- Proper transaction management with commit/rollback
- Verified: 14,866 sets and 14,826 cards loaded successfully
- Added MTG card ORM models (mtg_cards, mtg_sets tables)
- Created card_database service with search, get_by_name, get_by_set
- Added Redis client with caching layer (3600s TTL default)
- Created card router with caching on all endpoints:
- Search cards (5min cache)
- Get card by name (10min cache)
- Get cards by set (15min cache)
- Get card types/rarities (30min cache)
- Get sets (1hr cache)
- Get statistics (1hr cache)
- Updated settings.py:
- Added JWT_SECRET_KEY field
- Added DB_CONFIG and REDIS_CONFIG dictionaries
- Updated security.py to use JWT_SECRET_KEY with fallback
- Updated auth.py to use timezone-aware datetimes
- Updated refresh_mtg.py to use settings instead of os.environ
- Updated mtg_monitor.py to use settings for connections
- Added services package with __init__.py
All 20 tests passing.
- Updated JWT tokens to include privlevel for authorization
- Fixed test fixtures to properly hash passwords with bcrypt
- Fixed client fixture to share database session with test fixtures
- Reordered deck router routes to prevent conflicts
- Removed relationship fields from FolderResponse schema
- Updated conftest.py with proper async session management
akadmin
wall-o