extract_pdf_cjs.js 728 B

1234567891011121314151617181920
  1. const fs = require('fs');
  2. const pdf = require('pdf-parse');
  3. const pdfPath = 'd:/Users/chenjun/kyj-yanglao-web-new/src/views/elderly/apply/check-in/颐年集团养老服务合同-2026年6月终版docx.pdf';
  4. const outPath = 'd:/Users/chenjun/kyj-yanglao-web-new/contract_text.txt';
  5. const dataBuffer = fs.readFileSync(pdfPath);
  6. pdf(dataBuffer, { max: 0 }).then(function(data) {
  7. fs.writeFileSync(outPath, data.text, 'utf8');
  8. console.log('Total chars:', data.text.length);
  9. console.log('Total pages:', data.numpages);
  10. console.log('--- START TEXT ---');
  11. console.log(data.text);
  12. console.log('--- END TEXT ---');
  13. console.log('Saved to', outPath);
  14. }).catch(err => {
  15. console.error('ERROR:', err);
  16. process.exit(1);
  17. });