// Builds the anonymized FactHub tables for dataset wa-i638-girls-event-placements
// from HeCheated.org's public Washington K-12 results page.
// Usage: node build_placements.js <saved copy of https://hecheated.org/high-school_results_Washington.html> <outDir>
// Athlete names, schools and meet names are dropped; each athlete gets a code (A, B, ...)
// ordered by number of events in the pamphlet-era snapshot (most first), ties by earliest date.
const fs = require('fs'), vm = require('vm'), path = require('path');
const [, , htmlPath, outDir = '.'] = process.argv;
const html = fs.readFileSync(htmlPath, 'utf8');
const s = html.indexOf('const data = [');
const e = html.indexOf('];', s);
// The page embeds its table as a JavaScript array literal; evaluate it in an empty sandbox.
const rows = vm.runInNewContext('(' + html.slice(s + 13, e + 1) + ')', {}, { timeout: 2000 });

const MONTHS = { Jan: '01', Feb: '02', Mar: '03', Apr: '04', May: '05', Jun: '06', Jul: '07', Aug: '08', Sept: '09', Sep: '09', Oct: '10', Nov: '11', Dec: '12' };
const isoDate = d => { const [y, m, dd] = d.split('-'); return `${y}-${MONTHS[m]}-${dd.padStart(2, '0')}`; };
const isFirst = x => /^1st/.test(x.sort);
const isTop3 = x => /^(1st|2nd|3rd|Semi)/.test(x.sort);
const counted = x => x.sort !== 'DNS';
// Rows added after the pamphlet text was written: the three September 2026 rows.
const inSnapshot = x => !(x.year === '2026' && /^2026-Sept/.test(x.date));
const relay = x => /4x|DMR|SMR/.test(x.eventl);
const meetGroup = x => {
  if (/State\/Provincial/.test(x.category)) return 'State championship';
  if (/Regional|District|Sectional|Conference\/League Championship|Conference\/League Standings/.test(x.category)) return 'District, regional or league championship';
  if (/National/.test(x.category)) return 'National invitational (out of state)';
  if (/Dual|Tri-|Quad|Invitational|Conference Meet/.test(x.category)) return 'Regular-season meet';
  return 'Other or unclassified';
};
const finishGroup = x => {
  if (x.sort === 'DNS') return 'Did not start (not counted)';
  if (isFirst(x)) return '1st';
  if (isTop3(x)) return '2nd-3rd';
  const m = String(x.place).match(/^(\d+)/);
  if (!m) return 'DQ or no place listed';
  const n = +m[1];
  if (n <= 10) return '4th-10th';
  return '11th or lower';
};

// Athlete codes
const stats = {};
rows.forEach(x => {
  const k = x.name; stats[k] = stats[k] || { n: 0, first: isoDate(x.date) };
  if (inSnapshot(x) && counted(x)) stats[k].n++;
  if (isoDate(x.date) < stats[k].first) stats[k].first = isoDate(x.date);
});
const order = Object.keys(stats).sort((a, b) => stats[b].n - stats[a].n || stats[a].first.localeCompare(stats[b].first));
const code = {}; order.forEach((k, i) => code[k] = String.fromCharCode(65 + i));

const out = rows
  .map(x => ({ x, d: isoDate(x.date) }))
  .sort((a, b) => a.d.localeCompare(b.d) || a.x.eventl.localeCompare(b.x.eventl))
  .map(({ x, d }, i) => ({
    row_id: i + 1,
    athlete_code: code[x.name],
    date: d,
    year: +x.year,
    level: x.level,
    sport: x.sport,
    meet_category: x.category,
    meet_group: meetGroup(x),
    wiaa_state_meet: /State\/Provincial/.test(x.category) && /WIAA/.test(x.competition),
    event: x.eventl || '(team result)',
    relay: relay(x),
    team_sport: x.sport === 'Ultimate Frisbee',
    place_as_listed: x.place,
    finish_group: finishGroup(x),
    counted_event: counted(x),
    first_place: counted(x) && isFirst(x),
    top_three: counted(x) && isTop3(x),
    meet_location_state: x.location,
    in_pamphlet_count: inSnapshot(x),
  }));

const cols = Object.keys(out[0]);
const esc = v => { const t = String(v); return /[",\n]/.test(t) ? '"' + t.replace(/"/g, '""') + '"' : t; };
fs.writeFileSync(path.join(outDir, 'wa_girls_event_entries.csv'), [cols.join(','), ...out.map(r => cols.map(c => esc(r[c])).join(','))].join('\n') + '\n');
fs.writeFileSync(path.join(outDir, 'entries.json'), JSON.stringify(out, null, 1));
console.log('rows', out.length);