// Builds the king-county-overdose-deaths-by-place dataset: CSV tables + dataset JSON.
// Inputs: research/whs/overdoses-data/*.csv (KCMEO dashboard pulls 2026-11-06), data/hic/hic_panel.csv
const fs = require('fs');
const path = require('path');
const crypto = require('crypto');

const ROOT = path.resolve(__dirname, '..', '..');
const IN = path.join(ROOT, 'research', 'whs', 'overdoses-data');
const OUT = path.join(ROOT, 'facthub', 'out', 'housing-od');
fs.mkdirSync(OUT, { recursive: true });

function readCsv(file) {
  const text = fs.readFileSync(file, 'utf8').trim().split(/\r?\n/);
  const parse = (line) => {
    const out = []; let cur = ''; let q = false;
    for (const ch of line) {
      if (ch === '"') { q = !q; continue; }
      if (ch === ',' && !q) { out.push(cur); cur = ''; continue; }
      cur += ch;
    }
    out.push(cur); return out;
  };
  const head = parse(text[0]);
  return text.slice(1).map((l) => { const v = parse(l); const o = {}; head.forEach((h, i) => (o[h] = v[i])); return o; });
}
const toCsv = (rows, cols) => [cols.join(','), ...rows.map((r) => cols.map((c) => {
  const v = r[c]; if (v === null || v === undefined) return '';
  const s = String(v); return /[",]/.test(s) ? '"' + s.replace(/"/g, '""') + '"' : s;
}).join(','))].join('\n') + '\n';

const SIT = {
  'Housing service site': 'Subsidized and supportive housing (housing service site)',
  'Privately owned or rented residence': 'Private home',
  'Outdoors, homeless shelter, or vehicle': 'Outdoors, shelter or vehicle',
  'Other temporary arrangement': 'Other temporary arrangement',
  'Unknown': 'Unknown',
};
const SITS = Object.keys(SIT);
const round = (x, d) => Math.round(x * 10 ** d) / 10 ** d;

// ---------- Table 1: Jan 1 - Nov 5 by living situation, 2019-2026 ----------
const ytdRaw = readCsv(path.join(IN, 'kc-me-ytd-jan1-nov5-by-year-2016-2026.csv'));
const ytdTotal = {}; const ytdSit = {}; const ytdRegion = {};
for (const r of ytdRaw) {
  const y = +r.year; const d = +r.deaths_jan1_to_nov5;
  if (r.group === 'King County total') ytdTotal[y] = d;
  else if (r.group.startsWith('Region: ')) (ytdRegion[y] ||= {})[r.group.slice(8)] = d;
  else (ytdSit[y] ||= {})[r.group] = d;
}
const YEARS = [2019, 2020, 2021, 2022, 2023, 2024, 2025, 2026];
const t1 = [];
for (const y of YEARS) {
  const named = SITS.filter((s) => s !== 'Unknown').reduce((a, s) => a + (ytdSit[y][s] || 0), 0);
  for (const s of SITS) {
    let v = ytdSit[y][s]; let vt = 'observed';
    if (s === 'Unknown' && v === undefined) { v = ytdTotal[y] - named; vt = 'derived'; }
    t1.push({ year: y, period: 'Jan 1-Nov 5', living_situation: s, deaths: v, value_type: vt, county_total: ytdTotal[y] });
  }
  const sum = SITS.reduce((a, s) => a + t1.find((r) => r.year === y && r.living_situation === s).deaths, 0);
  if (sum !== ytdTotal[y]) throw new Error('ytd sum mismatch ' + y);
}

// ---------- Table 2: full-year by living situation, 2019-2025 ----------
// this file has an unquoted living_situation containing commas: parse year,<label...>,deaths,share
const annRaw = fs.readFileSync(path.join(IN, 'kc-me-living-situation-2019-2025.csv'), 'utf8').trim().split(/\r?\n/).slice(1).map((l) => {
  const p = l.split(','); return { year: p[0], living_situation: p.slice(1, -2).join(',').trim(), deaths: p[p.length - 2] };
});
const annTotRaw = readCsv(path.join(IN, 'kc-me-overdose-deaths-annual-2015-2025.csv'));
const annTot = {}; annTotRaw.forEach((r) => (annTot[+r.year] = +r.kc_me_overdose_deaths));
const annSit = {};
annRaw.forEach((r) => ((annSit[+r.year] ||= {})[r.living_situation] = +r.deaths));
const t2 = [];
for (let y = 2019; y <= 2025; y++) {
  const named = SITS.filter((s) => s !== 'Unknown').reduce((a, s) => a + annSit[y][s], 0);
  for (const s of SITS) {
    let v = annSit[y][s]; let vt = 'observed';
    if (s === 'Unknown' && v === undefined) { v = annTot[y] - named; vt = 'derived'; }
    t2.push({ year: y, living_situation: s, deaths: v, value_type: vt, county_total: annTot[y] });
  }
}

// ---------- Table 3: by region (place of death), YTD and full year ----------
const annReg = readCsv(path.join(IN, 'kc-me-deaths-by-region-2016-2025.csv'));
const t3 = [];
for (let y = 2016; y <= 2026; y++) for (const reg of ['Seattle', 'South', 'East', 'North']) {
  const full = annReg.find((r) => +r.year === y && r.region === reg);
  t3.push({ year: y, region: reg, deaths_jan1_nov5: ytdRegion[y][reg], deaths_full_year: full ? +full.deaths : null });
}

// ---------- Table 4: HUD Housing Inventory Count beds, WA-500 ----------
const hic = readCsv(path.join(ROOT, 'data', 'hic', 'hic_panel.csv')).filter((r) => r.coc_code === 'WA-500' && +r.year >= 2019);
const t4 = hic.map((r) => ({ year: +r.year, psh_beds: +r.psh_beds, oph_beds: +r.oph_beds, psh_plus_oph_beds: +r.psh_beds + +r.oph_beds, es_beds: +r.es_beds }));

// ---------- Table 5: CDC provisional counts, 12 months ending December ----------
const cdc = readCsv(path.join(IN, 'cdc-vsrr-us-wa-12mo-ending-dec-2015-2026.csv')).filter((r) => r.month === 'December');
const cdcK = readCsv(path.join(IN, 'cdc-vsrr-county-12mo-ending-peers.csv')).filter((r) => r.countyname === 'King' && r.month === '12');
const t5 = [
  ...cdc.map((r) => ({ geography: r.state === 'US' ? 'United States' : 'Washington', year: +r.year, deaths: +r.data_value, basis: 'reported, place of residence' })),
  ...cdcK.map((r) => ({ geography: 'King County (residents)', year: +r.year, deaths: +r.provisional_drug_overdose, basis: 'provisional, place of residence' })),
];

// ---------- write CSVs ----------
const files = {
  'ytd-by-living-situation.csv': toCsv(t1, ['year', 'period', 'living_situation', 'deaths', 'value_type', 'county_total']),
  'annual-by-living-situation.csv': toCsv(t2, ['year', 'living_situation', 'deaths', 'value_type', 'county_total']),
  'deaths-by-region.csv': toCsv(t3, ['year', 'region', 'deaths_jan1_nov5', 'deaths_full_year']),
  'hic-beds-king-county.csv': toCsv(t4, ['year', 'psh_beds', 'oph_beds', 'psh_plus_oph_beds', 'es_beds']),
  'cdc-annual-overdose-deaths.csv': toCsv(t5, ['geography', 'year', 'deaths', 'basis']),
};
for (const [f, s] of Object.entries(files)) fs.writeFileSync(path.join(OUT, f), s);
fs.copyFileSync(__filename, path.join(OUT, 'build_housing_od.js'));

// ---------- analyses ----------
const get1 = (y, s) => t1.find((r) => r.year === y && r.living_situation === s).deaths;
const get2 = (y, s) => t2.find((r) => r.year === y && r.living_situation === s).deaths;
const H = 'Housing service site';

const a1rows = YEARS.map((y) => {
  const known = ytdTotal[y] - get1(y, 'Unknown');
  return { year: String(y), share: round(get1(y, H) / known, 4), housing_deaths: get1(y, H), deaths_with_known_place: known, unknown_place: get1(y, 'Unknown'), share_of_all_deaths: round(get1(y, H) / ytdTotal[y], 4), share_if_all_unknown_were_housing: round((get1(y, H) + get1(y, 'Unknown')) / ytdTotal[y], 4) };
});

const labels = {
  [H]: 'Subsidized and supportive housing',
  'Privately owned or rented residence': 'Private homes',
  'Outdoors, homeless shelter, or vehicle': 'Outdoors, shelters, vehicles',
  'Other temporary arrangement': 'Other temporary stays',
};
const a2rows = [...Object.keys(labels).map((s) => ({ place: labels[s], deaths_2025: get1(2025, s), deaths_2026: get1(2026, s), change: round(get1(2026, s) / get1(2025, s) - 1, 4) })),
  { place: 'All overdose deaths', deaths_2025: ytdTotal[2025], deaths_2026: ytdTotal[2026], change: round(ytdTotal[2026] / ytdTotal[2025] - 1, 4) }];

// sensitivity: unknowns in 2026 beyond 2025's count assigned entirely to one group
const extraUnknown = get1(2026, 'Unknown') - get1(2025, 'Unknown');
const a3rows = Object.keys(labels).map((s) => ({ place: labels[s], change_if_all_extra_unknown_assigned_here: round((get1(2026, s) + extraUnknown) / get1(2025, s) - 1, 4), change_as_reported: round(get1(2026, s) / get1(2025, s) - 1, 4) }));

const a4rows = [...Object.keys(labels).map((s) => ({ place: labels[s], deaths_2019: get2(2019, s), deaths_2025: get2(2025, s), multiple: round(get2(2025, s) / get2(2019, s), 2) })),
  { place: 'All overdose deaths', deaths_2019: annTot[2019], deaths_2025: annTot[2025], multiple: round(annTot[2025] / annTot[2019], 2) }];

const beds = (y) => t4.find((r) => r.year === y);
const a5rows = [2019, 2025].map((y) => ({ year: y, housing_deaths: get2(y, H), psh_plus_oph_beds: beds(y).psh_plus_oph_beds, deaths_per_1000_beds: round(get2(y, H) / beds(y).psh_plus_oph_beds * 1000, 1) }));

const a6rows = [
  ...Object.keys(labels).map((s) => ({ group: labels[s], deaths_2023: get1(2023, s), deaths_2026: get1(2026, s), change: round(get1(2026, s) / get1(2023, s) - 1, 4) })),
  ...['Seattle', 'South', 'East', 'North'].map((g) => ({ group: 'Region: ' + g, deaths_2023: ytdRegion[2023][g], deaths_2026: ytdRegion[2026][g], change: round(ytdRegion[2026][g] / ytdRegion[2023][g] - 1, 4) })),
];
const a6b = ['Seattle', 'South', 'East', 'North'].map((g) => ({ region: g, deaths_2025: ytdRegion[2025][g], deaths_2026: ytdRegion[2026][g], change: round(ytdRegion[2026][g] / ytdRegion[2025][g] - 1, 4) }));

const cdcv = (g, y) => t5.find((r) => r.geography === g && r.year === y).deaths;
const a7rows = [
  { geography: 'United States', deaths_2019: cdcv('United States', 2019), deaths_2025: cdcv('United States', 2025), ratio: round(cdcv('United States', 2025) / cdcv('United States', 2019), 2), source: 'CDC, place of residence' },
  { geography: 'Washington', deaths_2019: cdcv('Washington', 2019), deaths_2025: cdcv('Washington', 2025), ratio: round(cdcv('Washington', 2025) / cdcv('Washington', 2019), 2), source: 'CDC, place of residence' },
  { geography: 'King County', deaths_2019: annTot[2019], deaths_2025: annTot[2025], ratio: round(annTot[2025] / annTot[2019], 2), source: 'King County Medical Examiner, place of death' },
];

fs.writeFileSync(path.join(OUT, 'analyses.json'), JSON.stringify({ a1rows, a2rows, a3rows, a4rows, a5rows, a6rows, a6b, a7rows, extraUnknown }, null, 2));
console.log(JSON.stringify({ a1rows, a2rows, a3rows, a4rows, a5rows, a6rows, a6b, a7rows, extraUnknown }, null, 1));
for (const f of Object.keys(files)) {
  const b = fs.readFileSync(path.join(OUT, f));
  console.log(f, b.length, 'sha256:' + crypto.createHash('sha256').update(b).digest('hex'));
}