-- 009: Entity cache — stores bulk-downloaded business entity records
-- from state Secretary of State open data portals.
-- Used by Verilex Data Professional API via internal bulk export endpoint.

CREATE TABLE IF NOT EXISTS entity_cache (
    id                  SERIAL PRIMARY KEY,
    jurisdiction        TEXT NOT NULL,          -- US_CO, US_NY, etc.
    entity_name         TEXT NOT NULL,
    entity_number       TEXT,                   -- state filing number
    entity_type         TEXT,                   -- LLC, CORPORATION, LP, LLP, etc.
    status              TEXT,                   -- ACTIVE, DISSOLVED, SUSPENDED, DELINQUENT
    formation_date      DATE,
    dissolution_date    DATE,
    registered_agent    TEXT,
    principal_address   TEXT,
    state               TEXT NOT NULL,          -- 2-letter
    source              TEXT DEFAULT 'socrata', -- socrata, sftp, csv, playwright
    raw_data            JSONB,                  -- original source record for debugging
    last_synced         TIMESTAMPTZ DEFAULT NOW(),
    created_at          TIMESTAMPTZ DEFAULT NOW(),
    UNIQUE(jurisdiction, entity_number)
);

CREATE INDEX IF NOT EXISTS idx_ec_state      ON entity_cache (state);
CREATE INDEX IF NOT EXISTS idx_ec_name       ON entity_cache USING gin (entity_name gin_trgm_ops);
CREATE INDEX IF NOT EXISTS idx_ec_number     ON entity_cache (entity_number);
CREATE INDEX IF NOT EXISTS idx_ec_status     ON entity_cache (status);
CREATE INDEX IF NOT EXISTS idx_ec_synced     ON entity_cache (last_synced);

-- Name search cache — stores results of live name availability searches
-- to avoid hammering state portals. Entries expire after 24 hours.
CREATE TABLE IF NOT EXISTS name_search_cache (
    id              SERIAL PRIMARY KEY,
    state_code      CHAR(2) NOT NULL,
    searched_name   TEXT NOT NULL,
    available       BOOLEAN,
    exact_match     BOOLEAN DEFAULT FALSE,
    similar_names   TEXT[],
    raw_response    TEXT,
    searched_at     TIMESTAMPTZ DEFAULT NOW(),
    expires_at      TIMESTAMPTZ DEFAULT NOW() + INTERVAL '24 hours',
    UNIQUE(state_code, searched_name)
);

CREATE INDEX IF NOT EXISTS idx_nsc_lookup ON name_search_cache (state_code, searched_name) WHERE expires_at > NOW();

-- Enable trigram extension if not already enabled
CREATE EXTENSION IF NOT EXISTS pg_trgm;