hollywood

Local-first entertainment data platform for research corpus building. A Hono REST API ingests entertainment RSS feeds, browser-only directories like WGA, and structured metadata sources (TMDB, Wikidata, IMDb) into a unified SQLite database, with raw-payload archiving and JSONL exports.

Run

just api-setup
just api-dev

The API starts at http://localhost:4000. OpenAPI docs are at /openapi.

Endpoints

• GET /sources — list built-in ingest sources
• POST /ingest — LLM extraction from raw submission text (single doc or batch)
• POST /ingest/source — ingest a single source (variety, deadline, hollywood_reporter, the_wrap, tmdb, wikidata, wga, imdb, spotify)
• POST /ingest/group — ingest all sources in a named group (news, entities, directories, all)
• POST /normalize — re-derive normalized tables from already-archived raw records
• GET /export — export normalized tables as JSONL
• GET /doctor — health checks (data dir, DB, API keys, per-source config)
• GET /candidates, /projects, /submissions, /search, /tags, /users — entity graph CRUD/search
• POST /assistant/chat — Director-routed assistant (calendar/financial/marketing/touring), grounded in the entity graph

Environment

HOLLYWOOD_DATA_DIR=data
HOLLYWOOD_DB_PATH=~/.hominem/hollywood.db
HOLLYWOOD_USER_AGENT=...
HOLLYWOOD_REQUEST_TIMEOUT_SECONDS=30
TMDB_API_KEY=...
SPOTIFY_CLIENT_ID=...
SPOTIFY_CLIENT_SECRET=...
OPENROUTER_API_KEY=...

The default storage layout is:

• data/raw/ for archived payloads
• ~/.hominem/hollywood.db for the unified SQLite database
• data/parquet/ for JSONL exports (despite the directory name — parquet export is not yet implemented)

Development

just api-setup
just api-dev        # dev server with hot reload
just api-typecheck
just api-build
just api-start       # run the compiled build