-- ============================================================
-- LegiScan Legislative Analysis Queries
-- Database: data_centers  Schema: public
-- ============================================================
--
-- SETUP
--   Populate the database first:
--     python ingest_legiscan.py --all
--   This downloads ~646 sessions (2016-2026, all US states + federal),
--   loads ~1.3M bills, and tags ~60K as relevant.
--
-- To refresh (weekly dataset updates from LegiScan):
--     python ingest_legiscan.py --fetch --load
--   Already-imported sessions with unchanged dataset_hash are skipped.
--
-- To retag after editing keyword lists in ingest_legiscan.py:
--     python ingest_legiscan.py --tag
--
-- RELEVANCE TAGS (stored in legiscan_bills.relevance_tags[]):
--   data_center         - Bills naming data centers, hyperscale, colocation, AI campuses
--   large_load          - Crypto mining, large industrial loads, extraordinary load
--   ratepayer_protection- Cost shifting, cross-subsidy, rate design, affordability
--   grid_impact         - Grid reliability, transmission, interconnection queue
--   tax_incentive       - Tax exemptions/abatements/credits for facilities
--   energy_policy       - Renewable PPAs, green tariffs, clean electricity
--   water_use           - Cooling water, evaporative cooling, water footprint
--   siting_permitting   - Zoning, conditional use permits, local control
--
-- STATUS CODES (legiscan_bills.status):
--   1=Introduced  2=Engrossed  3=Enrolled  4=Passed  5=Vetoed
--   6=Failed      7=Override   8=Chaptered 9=Referred 12=Draft
-- ============================================================

-- ── Quick overview ──────────────────────────────────────────

SELECT
    COUNT(*)                                           AS total_bills,
    COUNT(*) FILTER (WHERE is_relevant)                AS relevant_bills,
    COUNT(DISTINCT state)                              AS states,
    MIN(ls.year_start)                                 AS year_from,
    MAX(ls.year_end)                                   AS year_to
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id);

-- ── Bills per relevance tag ─────────────────────────────────

SELECT
    tag,
    COUNT(*)                                           AS bill_count,
    COUNT(*) FILTER (WHERE lb.status = 4)             AS passed,
    COUNT(*) FILTER (WHERE lb.status IN (4,8))        AS enacted
FROM legiscan_bills lb, unnest(relevance_tags) AS tag
GROUP BY tag
ORDER BY bill_count DESC;

-- ── Top states for relevant legislation ────────────────────

SELECT
    state,
    COUNT(*)                                           AS relevant_bills,
    COUNT(*) FILTER (WHERE 'data_center'       = ANY(relevance_tags)) AS data_center,
    COUNT(*) FILTER (WHERE 'large_load'        = ANY(relevance_tags)) AS large_load,
    COUNT(*) FILTER (WHERE 'ratepayer_protection' = ANY(relevance_tags)) AS ratepayer,
    COUNT(*) FILTER (WHERE 'tax_incentive'     = ANY(relevance_tags)) AS tax_incentive,
    COUNT(*) FILTER (WHERE 'grid_impact'       = ANY(relevance_tags)) AS grid_impact
FROM legiscan_bills
WHERE is_relevant
GROUP BY state
ORDER BY relevant_bills DESC
LIMIT 20;

-- ── Trend by year ───────────────────────────────────────────

SELECT
    ls.year_start                                      AS year,
    COUNT(lb.bill_id)                                  AS total_bills,
    COUNT(lb.bill_id) FILTER (WHERE lb.is_relevant)   AS relevant_bills,
    COUNT(lb.bill_id) FILTER (WHERE lb.is_relevant AND lb.status IN (4,8)) AS enacted,
    ROUND(100.0 * COUNT(lb.bill_id) FILTER (WHERE lb.is_relevant)
          / NULLIF(COUNT(lb.bill_id), 0), 1)           AS pct_relevant
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id)
GROUP BY ls.year_start
ORDER BY ls.year_start;

-- ── Data center bills specifically ─────────────────────────

SELECT
    lb.state,
    lb.bill_number,
    ls.year_start                                      AS year,
    lb.status,
    lb.title,
    lb.relevance_tags,
    lb.url
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id)
WHERE 'data_center' = ANY(lb.relevance_tags)
ORDER BY
    CASE lb.status WHEN 4 THEN 0 WHEN 8 THEN 1 WHEN 3 THEN 2 ELSE 3 END,
    ls.year_start DESC,
    lb.state;

-- ── Ratepayer protection bills ──────────────────────────────

SELECT
    lb.state,
    lb.bill_number,
    ls.year_start                                      AS year,
    lb.status,
    lb.title,
    lb.relevance_tags,
    lb.url
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id)
WHERE 'ratepayer_protection' = ANY(lb.relevance_tags)
ORDER BY
    CASE lb.status WHEN 4 THEN 0 WHEN 8 THEN 1 WHEN 3 THEN 2 ELSE 3 END,
    ls.year_start DESC,
    lb.state;

-- ── Bills at intersection of data center + ratepayer ───────

SELECT
    lb.state,
    lb.bill_number,
    ls.year_start                                      AS year,
    lb.status,
    lb.title,
    lb.relevance_tags,
    lb.url
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id)
WHERE 'data_center'           = ANY(lb.relevance_tags)
  AND 'ratepayer_protection'  = ANY(lb.relevance_tags)
ORDER BY ls.year_start DESC, lb.state;

-- ── Large load + grid impact ────────────────────────────────

SELECT
    lb.state,
    lb.bill_number,
    ls.year_start                                      AS year,
    lb.status,
    lb.title,
    lb.relevance_tags,
    lb.url
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id)
WHERE 'large_load'  = ANY(lb.relevance_tags)
  AND 'grid_impact' = ANY(lb.relevance_tags)
ORDER BY ls.year_start DESC, lb.state;

-- ── Tax incentive bills passed/enacted ─────────────────────

SELECT
    lb.state,
    lb.bill_number,
    ls.year_start                                      AS year,
    lb.status,
    lb.title,
    lb.url
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id)
WHERE 'tax_incentive' = ANY(lb.relevance_tags)
  AND lb.status IN (4, 8)   -- Passed or Chaptered
ORDER BY ls.year_start DESC, lb.state;

-- ── Join to data centers: states with both DCs and active legislation ──

SELECT
    dc.state,
    COUNT(DISTINCT dc.id)                              AS data_centers,
    COUNT(DISTINCT lb.bill_id)                         AS relevant_bills,
    COUNT(DISTINCT lb.bill_id)
        FILTER (WHERE 'ratepayer_protection' = ANY(lb.relevance_tags)) AS ratepayer_bills,
    COUNT(DISTINCT lb.bill_id)
        FILTER (WHERE 'data_center' = ANY(lb.relevance_tags))          AS dc_specific_bills,
    COUNT(DISTINCT lb.bill_id)
        FILTER (WHERE lb.status IN (4,8))                              AS enacted_bills
FROM master_data_centers dc
LEFT JOIN legiscan_bills lb ON dc.state = lb.state AND lb.is_relevant
GROUP BY dc.state
ORDER BY relevant_bills DESC;

-- ── Full-text search: find bills mentioning specific terms ──
-- Replace 'hyperscale' with any keyword of interest

SELECT
    lb.state,
    lb.bill_number,
    ls.year_start                                      AS year,
    lb.status,
    lb.title,
    lb.description,
    lb.url
FROM legiscan_bills lb
JOIN legiscan_sessions ls USING (session_id)
WHERE to_tsvector('english', COALESCE(lb.title,'') || ' ' || COALESCE(lb.description,''))
      @@ to_tsquery('english', 'hyperscale | colocation | "data center"')
ORDER BY ts_rank(
    to_tsvector('english', COALESCE(lb.title,'') || ' ' || COALESCE(lb.description,'')),
    to_tsquery('english', 'hyperscale | colocation | "data center"')
) DESC
LIMIT 50;

-- ── Session coverage check ──────────────────────────────────

SELECT
    state_abbr,
    COUNT(*)                                           AS sessions_loaded,
    SUM(bill_count)                                    AS total_bills,
    MIN(year_start)                                    AS earliest,
    MAX(year_end)                                      AS latest
FROM legiscan_sessions
GROUP BY state_abbr
ORDER BY state_abbr;