slaj/docs/_parse.js

const fs = require('fs');
const zlib = require('zlib');
const buf = fs.readFileSync('docs/工作考核对象.docx');
let pos = 0; const sig = 0x04034b50; let xml = null;
while (pos < buf.length - 4) {
  if (buf.readUInt32LE(pos) !== sig) { pos++; continue; }
  const method = buf.readUInt16LE(pos + 8);
  const compSize = buf.readUInt32LE(pos + 18);
  const nameLen = buf.readUInt16LE(pos + 26);
  const extraLen = buf.readUInt16LE(pos + 28);
  const name = buf.toString('utf8', pos + 30, pos + 30 + nameLen);
  const dataStart = pos + 30 + nameLen + extraLen;
  const dataEnd = dataStart + compSize;
  if (name === 'word/document.xml') { let d = buf.slice(dataStart, dataEnd); if (method===8) d=zlib.inflateRawSync(d); xml = d.toString('utf8'); }
  pos = dataEnd;
}
function txt(s){let r='';const re=/<w:t\b[^>]*>([\s\S]*?)<\/w:t>/g;let m;while((m=re.exec(s))!==null)r+=m[1];return r;}
function findPara(s, from){let idx=s.indexOf('<w:p',from);while(idx!==-1){const c=s[idx+4];if(c===' '||c==='>'||c==='\t'||c==='\n')return idx;idx=s.indexOf('<w:p',idx+5);}return -1;}
const blocks=[]; let i=0; const body=xml;
while(i<body.length){
  const tIdx=body.indexOf('<w:tbl>',i);
  const pIdx=findPara(body,i);
  if(tIdx===-1&&pIdx===-1)break;
  if(pIdx!==-1&&(tIdx===-1||pIdx<tIdx)){const end=body.indexOf('</w:p>',pIdx);blocks.push({type:'p',text:txt(body.substring(pIdx,end+6))});i=end+6;}
  else {const end=body.indexOf('</w:tbl>',tIdx);blocks.push({type:'tbl',raw:body.substring(tIdx,end+8)});i=end+8;}
}
const out=[];let ti=0;
for(let b=0;b<blocks.length;b++){
  if(blocks[b].type!=='tbl')continue;ti++;
  const pre=[];for(let k=b-1;k>=0;k--){if(blocks[k].type==='tbl')break;if(blocks[k].text.trim())pre.unshift(blocks[k].text);}
  out.push('\n=== TABLE '+ti+' ===');
  out.push('PRE: '+JSON.stringify(pre));
  const rowRe=/<w:tr[ >]([\s\S]*?)<\/w:tr>/g;let rm;
  while((rm=rowRe.exec(blocks[b].raw))!==null){
    const cellRe=/<w:tc[ >]([\s\S]*?)<\/w:tc>/g;let cm;const cells=[];
    while((cm=cellRe.exec(rm[1]))!==null)cells.push(txt(cm[1]).trim());
    out.push(cells.join(' | '));
  }
}
fs.writeFileSync('docs/_parsed.txt', out.join('\n'), 'utf8');
console.log('tables:',ti);