const fs=require('fs');
const all=JSON.parse(fs.readFileSync('parsed.json','utf8'));
const finals={'2019-21':[2020,2021],'2021-23':[2022,2023],'2023-25':[2024,2025],'2025-27':[2026,2027]};
const rows=[];
for(const o of all){ if(!finals[o.bi]||o.body.startsWith('__TAIL__')) continue;
  let b=o.body.replace(/^\(\([ .]*\)\)\s*/,'').replace(/^\(\([ .]*\)\)\s*/,'');
  const ci=b.indexOf(':'); const src=b.slice(0,ci).trim(); const rest=b.slice(ci+1);
  const toGF=/for transfer to the (state )?general fund/i.test(rest);
  const fromGF=/^General Fund$/i.test(src);
  if(!toGF&&!fromGF) continue;
  const [y1,y2]=finals[o.bi]; let f1=0,f2=0;
  const re=/\$([\d,]+)\s+for\s+fiscal\s+year\s+(\d{4})|fiscal\s+year\s+(\d{4}),?\s*\$([\d,]+)/g; let m,found=false;
  while((m=re.exec(rest))){found=true; const amt=+(m[1]||m[4]).replace(/,/g,''); const y=+(m[2]||m[3]); if(y===y1)f1+=amt; else if(y===y2)f2+=amt;}
  if(!found||f1+f2!==o.total){ const ys=[...rest.matchAll(/fiscal year (\d{4})/g)].map(x=>+x[1]); if(!found){ if(ys.includes(y1)&&!ys.includes(y2)) f1=o.total; else if(ys.includes(y2)&&!ys.includes(y1)) f2=o.total; } }
  let cat='dedicated', note='';
  if(/Cannabis|Marijuana|Tobacco Settlement/i.test(src)) cat='routine';
  if(/Budget Stabilization/i.test(src)) cat='bsa';
  if(fromGF) cat='';
  if(/repayment/i.test(rest)) note='loan repayment to GF';
  if(/Rescue Plan Transition/i.test(src)) note='GF-funded reserve account (money GF put in during 2021-23)';
  if(/balance in the account/i.test(rest)) note='sweep of remaining balance';
  if(/continue this policy/i.test(rest)) note='recurring sweep (legislative intent to continue)';
  if(/^Air Operating Permit/.test(src)&&o.bi==='2025-27') note='VETOED by Governor (2026 partial veto of sec. 802); not law';
  if(f1+f2!==o.total) note=(note?note+'; ':'')+'FY split unclear in text';
  rows.push({bi:o.bi,law:o.law,sec:o.sec,src,f1:f1/1000,f2:f2/1000,tot:o.total/1000,dir:toGF?'toGF':'fromGF',cat,quote:(src+': '+rest.trim()).slice(0,400),note,dest: fromGF?(rest.match(/for transfer to the ([^,$]*?)(,| \$| for| pursuant| under| solely|$)/i)||[])[1]:'state general fund'});
}
const q=s=>'"'+String(s).replace(/"/g,'""')+'"';
let csv='biennium,sessionLaw,section,sourceAccount,fy1Thousands,fy2Thousands,totalThousands,direction,category,quote,note\n';
for(const r of rows) csv+=[r.bi,q(r.law),q(r.sec),q(r.src),r.f1,r.f2,r.tot,r.dir,r.cat,q(r.quote),q(r.note)].join(',')+'\n';
fs.writeFileSync('C:/dev/FactHub-Tester/millionaires-tax/data/wa_fund_transfers_to_gf.csv',csv);
fs.writeFileSync('rows.json',JSON.stringify(rows,null,1));
// counts
const groups={}; const vetoed=r=>/VETOED/.test(r.note);
for(const r of rows){ if(vetoed(r)) continue; const k=r.bi; groups[k]=groups[k]||{to:[],from:[]}; (r.dir==='toGF'?groups[k].to:groups[k].from).push(r);}
const sum=a=>Math.round(a.reduce((s,r)=>s+r.tot,0)); const uniq=a=>new Set(a.map(r=>r.src.replace(/Marijuana/,'Cannabis').replace(/Rental and Lease/,'Rent and Lease'))).size;
let T={to:[],from:[]};
for(const k of Object.keys(groups)){const g=groups[k]; T.to.push(...g.to); T.from.push(...g.from);
 const d=g.to.filter(r=>r.cat==='dedicated'); const d2=d.filter(r=>!r.note.match(/repayment|reserve/));
 console.log(k,'toGF items',g.to.length,'$K',sum(g.to),'| distinct src',uniq(g.to),'| excl routine+bsa',d.length,'$K',sum(d),'distinct',uniq(d),'| also excl loan repay+WRPTA',d2.length,sum(d2),'| fromGF',g.from.length,sum(g.from));
 if(rows.some(r=>r.bi===k&&r.note.includes('unclear'))) console.log('  unclear:',rows.filter(r=>r.bi===k&&r.note.includes('unclear')).map(r=>r.src+' '+r.tot+' '+r.f1+'/'+r.f2));
}
for(const [lab,keys] of [['2021-23..2025-27',['2021-23','2023-25','2025-27']],['all incl 2019-21',['2019-21','2021-23','2023-25','2025-27']]]){
 const to=T.to.filter(r=>keys.includes(r.bi)); const d=to.filter(r=>r.cat==='dedicated'); const d2=d.filter(r=>!r.note.match(/repayment|reserve/));
 console.log(lab,'toGF',to.length,sum(to),'distinct',uniq(to),'| excl',d.length,sum(d),'distinct',uniq(d),'| excl2',d2.length,sum(d2),uniq(d2),'| fromGF',T.from.filter(r=>keys.includes(r.bi)).length,sum(T.from.filter(r=>keys.includes(r.bi))));}