dvc.lock

schema: '2.0'
stages:
  bingsearch:
    cmd: python stages/01_bingquery.py
    outs:
    - path: download/search_results.parquet
      hash: md5
      md5: cbb02378c80cdefe3f2a9d8754aac6a9
      size: 434703
  download_pdfs:
    cmd: python stages/02_download_pdf.py
    deps:
    - path: download/search_results.parquet
      hash: md5
      md5: cbb02378c80cdefe3f2a9d8754aac6a9
      size: 434703
    outs:
    - path: brick/riskder.parquet
      hash: md5
      md5: 47ca58ef9790ace14d67662fa3d6ba76
      size: 498859
    - path: brick/riskder.pdf
      hash: md5
      md5: 555f872ce98d136793bd801edd7afd7f.dir
      size: 2050153290
      nfiles: 1809
  extract_data:
    cmd: python stages/03_data_extractor.py
    deps:
    - path: brick/riskder.pdf
      hash: md5
      md5: 555f872ce98d136793bd801edd7afd7f.dir
      size: 2050153290
      nfiles: 1809
    outs:
    - path: brick/extraction.parquet
      hash: md5
      md5: 18a22e58f4ee7675398964d8573fe19a
      size: 11356