bb.edn

{:min-bb-version "1.3.0"
 
 ;; ╔═══════════════════════════════════════════════════════════════════════╗
 ;; ║  Places Database - Babashka Tasks                                      ║
 ;; ║  Google Maps → SQLite → MCP Server → Claude                           ║
 ;; ║                                                                        ║
 ;; ║  Quick start:                                                          ║
 ;; ║    1. bb setup                                                         ║
 ;; ║    2. cp your_places*.csv data/                                        ║
 ;; ║    3. bb places:import                                                 ║
 ;; ║    4. bb index:fts                                                     ║
 ;; ║    5. bb geo:enrich (optional, adds missing coords)                    ║
 ;; ║    6. bb mcp:config (copy to Claude Desktop config)                    ║
 ;; ║                                                                        ║
 ;; ║  See CLAUDE.md for full docs                                           ║
 ;; ╚═══════════════════════════════════════════════════════════════════════╝
 
 :tasks
 {;; ═══════════════════════════════════════════════════════════════════════
  ;; Import & Update
  ;; ═══════════════════════════════════════════════════════════════════════
  
  places:import
  {:doc "Initial import of CSV files to SQLite
         Usage: bb places:import [file.csv]
         Default: imports all CSVs from data/*.csv
         Creates SQLite database with deduplication and proper indexing"
   :task (shell "python3 scripts/import_places.py" (or (System/getenv "CSV_PATH") "data/*.csv"))}

  places:update
  {:doc "Merge new CSV exports (preserves manual enrichments)
         Usage: bb places:update [file.csv]
         Smart update: only adds new places or fills missing fields
         Never overwrites manual notes or corrections"
   :task (shell "python3 scripts/import_places.py --update" (or (System/getenv "CSV_PATH") "data/*.csv"))}

  places:stats
  {:doc "Show database statistics
         Displays: total places, geocoding coverage, top lists/countries, ratings"
   :task (shell "python3 scripts/db_stats.py")}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; Geocoding (Free via Nominatim/OSM)
  ;; ═══════════════════════════════════════════════════════════════════════

  geo:enrich
  {:doc "Geocode missing coordinates (free via Nominatim)
         Rate limit: 1 req/sec → ~1 hour for 3600 places
         Uses formatted_address + city + country for accuracy
         Usage: bb geo:enrich [--dry-run]"
   :task (shell "python3 scripts/geocode_places.py")}

  geo:status
  {:doc "Show geocoding coverage statistics
         Displays: total/geocoded/missing, sources breakdown"
   :task (shell "python3 scripts/geocode_places.py --status")}

  geo:validate
  {:doc "Validate coordinate accuracy
         Checks for invalid lat/lon ranges and obvious errors"
   :task (shell "python3 scripts/geocode_places.py --validate")}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; Indexing (FTS5, Spatial, Optional Embeddings)
  ;; ═══════════════════════════════════════════════════════════════════════

  index:fts
  {:doc "Build full-text search index (FTS5) - REQUIRED for fast search
         Indexes: name, categories, address, city, region, country, notes
         Run after import or when adding many new places"
   :task (shell "python3 scripts/build_indexes.py --fts")}

  index:embeddings
  {:doc "Generate embeddings for semantic search (optional, ~$0.50)
         Enables queries like: 'find places similar to X'
         Uses voyage-3.1 API (requires ANTHROPIC_API_KEY)
         Estimate cost first: bb cost:embeddings"
   :task (shell "python3 scripts/build_indexes.py --embeddings")}

  index:spatial
  {:doc "Build spatial index (SpatiaLite if available, else Haversine)
         Auto-detects SpatiaLite extension for faster geo queries
         Fallback: Uses manual lat/lon index with Haversine distance"
   :task (shell "python3 scripts/build_indexes.py --spatial")}

  index:all
  {:doc "Build all indexes (FTS5 + spatial)
         Run after initial import for full search capability"
   :depends [index:fts index:spatial]}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; MCP Server (Claude Integration)
  ;; ═══════════════════════════════════════════════════════════════════════

  mcp:serve
  {:doc "Start MCP server for Claude Desktop/Code
         Usually runs automatically via Claude config
         Manual start for debugging: bb mcp:serve"
   :task (shell "python3 scripts/places_mcp_server.py")}

  mcp:config
  {:doc "Print Claude Desktop config snippet
         Copy output to: ~/Library/Application Support/Claude/claude_desktop_config.json
         Then restart Claude Desktop"
   :task (println
          "\n╔════════════════════════════════════════════════════════════╗"
          "\n║ Add to Claude Desktop config.json:                         ║"
          "\n╚════════════════════════════════════════════════════════════╝\n"
          "{\n"
          "  \"mcpServers\": {\n"
          "    \"places\": {\n"
          "      \"command\": \"python3\",\n"
          "      \"args\": [\"" (str (System/getProperty "user.dir") "/scripts/places_mcp_server.py") "\"]\n"
          "    }\n"
          "  }\n"
          "}\n\n"
          "Config location:\n"
          "  ~/Library/Application Support/Claude/claude_desktop_config.json\n\n"
          "After adding, restart Claude Desktop.\n")}

  mcp:test
  {:doc "Test MCP server locally (smoke tests)
         Verifies: database connectivity, FTS index, geocoding coverage"
   :task (shell "python3 scripts/test_mcp.py")}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; Database Maintenance
  ;; ═══════════════════════════════════════════════════════════════════════

  db:backup
  {:doc "Create timestamped database backup
         Saves to: backups/places-YYYYMMDD-HHMMSS.db
         Run before major updates or experiments"
   :task (let [timestamp (.format (java.time.LocalDateTime/now)
                                   (java.time.format.DateTimeFormatter/ofPattern "yyyyMMdd-HHmmss"))
               backup-path (str "backups/places-" timestamp ".db")]
           (shell "mkdir -p backups")
           (shell "cp data/places.db" backup-path)
           (println "✓ Backed up to" backup-path))}

  db:vacuum
  {:doc "Optimize database size (reclaim deleted space)
         Run after large deletions or updates
         Can reduce file size by 30-50%"
   :task (shell "sqlite3 data/places.db 'VACUUM;'")}

  db:shell
  {:doc "Open interactive SQLite shell
         Direct SQL access to places.db
         Exit with: .quit or Ctrl+D"
   :task (shell "sqlite3 data/places.db")}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; Quick Queries (Command Line)
  ;; ═══════════════════════════════════════════════════════════════════════

  query:lists
  {:doc "Show all custom lists with place counts
         Sorted by count (descending)"
   :task (shell "sqlite3 data/places.db" 
                "\"SELECT list, COUNT(*) as count FROM places GROUP BY list ORDER BY count DESC;\"")}

  query:recent
  {:doc "Show 20 most recently added places
         Requires 'added_at' field from Google export"
   :task (shell "sqlite3 data/places.db"
                "\"SELECT name, list, formatted_address FROM places WHERE added_at IS NOT NULL ORDER BY added_at DESC LIMIT 20;\"")}

  query:ungeocoded
  {:doc "Count places still missing coordinates
         Run bb geo:enrich to fill these in"
   :task (shell "sqlite3 data/places.db"
                "\"SELECT COUNT(*) FROM places WHERE lat IS NULL OR lon IS NULL;\"")}

  query:near
  {:doc "Find places near coordinates (requires args: lat lon radius_km)
         Example: bb query:near 59.3293 18.0686 5
         Uses Haversine distance calculation"
   :task (let [[lat lon radius] *command-line-args*]
           (if (and lat lon radius)
             (shell "python3 scripts/spatial_query.py --near" lat lon radius)
             (println "Usage: bb query:near <lat> <lon> <radius_km>")))}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; Setup & Installation
  ;; ═══════════════════════════════════════════════════════════════════════

  setup
  {:doc "Initial setup: create directories, install dependencies
         Run once before first import
         Installs: anthropic, requests, tqdm, python-dotenv"
   :task (do
           (println "🗺️  Setting up Places database...")
           (shell "mkdir -p data backups scripts")
           (shell "pip3 install --break-system-packages -q anthropic requests tqdm python-dotenv 2>/dev/null || true")
           (println "\n✓ Setup complete!")
           (println "\nNext steps:")
           (println "  1. Place CSV files in data/ directory")
           (println "  2. Run: bb places:import")
           (println "  3. Run: bb index:fts")
           (println "  4. Optional: bb geo:enrich")
           (println "  5. Connect to Claude: bb mcp:config\n"))}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; Cost Estimation
  ;; ═══════════════════════════════════════════════════════════════════════

  cost:embeddings
  {:doc "Estimate embedding generation cost
         Uses voyage-3.1 pricing: $0.10 per 1M tokens
         ~500 tokens per place → ~$0.50 for 10k places
         Shows estimate without generating"
   :task (shell "python3 scripts/build_indexes.py --embeddings --estimate")}

  ;; ═══════════════════════════════════════════════════════════════════════
  ;; Help
  ;; ═══════════════════════════════════════════════════════════════════════

  help
  {:doc "Show this help (alias for 'tasks')"
   :task (shell "bb tasks")}

  }	
}