const fs=require('fs');
const specs=[
 ['2019-21','2021 c 334 (ESSB 5092)','sec. 1802','txt/2021-22_5092-S.SL.txt',2],
 ['2021-23','2023 c 475 (ESSB 5187)','sec. 1804','txt/2023-24_5187-S.SL.txt',2],
 ['2023-25','2025 c 424 (ESSB 5167)','sec. 1804','txt/2025-26_5167-S.SL.txt',2],
 ['2025-27','2026 c 268 (ESSB 5998)','sec. 802','txt/2025-26_5998-S.SL.txt',1],
 // intermediates for comparison
 ['2021-23i','2021 c 334 (ESSB 5092)','sec. 805','txt/2021-22_5092-S.SL.txt',1],
 ['2021-23i2','2022 c 297 (ESSB 5693)','sec. 804','txt/2021-22_5693-S.SL.txt',1],
 ['2023-25i','2023 c 475 (ESSB 5187)','sec. 805','txt/2023-24_5187-S.SL.txt',1],
 ['2023-25i2','2024 c 376 (ESSB 5950)','sec. 804','txt/2023-24_5950-S.SL.txt',1],
 ['2025-27i','2025 c 424 (ESSB 5167)','sec. 805','txt/2025-26_5167-S.SL.txt',1],
];
const out=[];
for(const [bi,law,sec,file,occ] of specs){
  let t=fs.readFileSync(file,'utf8').replace(/<!--[^>]*-->/g,'');
  let idx=-1; for(let i=0;i<occ;i++) idx=t.indexOf('FOR THE STATE TREASURER—TRANSFERS',idx+1);
  let rest=t.slice(idx+35);
  let end=rest.search(/\n\*?(NEW SECTION\.|Sec\. \d+\.)/); rest=rest.slice(0,end);
  // split entries by total "$X" lines following ". . . ."
  const blocks=rest.split(/\.\s\.\s\.\s\.[\s\S]*?\n\s*(\$[\d,]+|\(\(\)\)\s*\n?\s*\$[\d,]+)\s*\n/);
  // simpler: regex for entries
  const re=/([\s\S]*?)\.\s\.\s\.\s\.\s*(?:\(\(\)\)\s*)?\$([\d,]+)/g; let m; let tail='';
  let lastIdx=0;
  while((m=re.exec(rest))){ let body=m[1].replace(/\s+/g,' ').trim(); body=body.replace(/\(\(\)\)/g,'').replace(/\s+/g,' ');
    out.push({bi,law,sec,body,total:+m[2].replace(/,/g,'')}); lastIdx=re.lastIndex;}
  out.push({bi,law,sec,body:'__TAIL__ '+rest.slice(lastIdx).replace(/\s+/g,' ').slice(0,3000),total:0});
}
fs.writeFileSync('parsed.json',JSON.stringify(out,null,1));
for(const o of out) console.log(o.bi,'|',o.total,'|',o.body.slice(0,400));