run_extract.py

import sys
import json
from pathlib import Path

from sdser import extract_from_pdf


def dump(path: str):
    cas, ec, cas_name_map, ec_name_map, pairs = extract_from_pdf(path)
    out = {
        'CAS': cas,
        'EC': ec,
        'cas_name_map': cas_name_map,
        'ec_name_map': ec_name_map,
        'pairs': pairs,
    }
    # Ensure pairs are JSON-serializable lists
    out['pairs'] = [[c, e] for (c, e) in out['pairs']]
    print(json.dumps(out, indent=2, ensure_ascii=False))


def main():
    if len(sys.argv) < 2:
        print("Usage: python run_extract.py <path-to-pdf> [out.json]")
        sys.exit(1)
    pdf = sys.argv[1]
    out_path = None
    if len(sys.argv) >= 3:
        out_path = Path(sys.argv[2])

    s = None
    try:
        from sdser import extract_from_pdf
        cas, ec, cas_name_map, ec_name_map, pairs = extract_from_pdf(pdf)
        out = {
            'CAS': cas,
            'EC': ec,
            'cas_name_map': cas_name_map,
            'ec_name_map': ec_name_map,
            'pairs': [[c, e] for (c, e) in pairs],
        }
        j = json.dumps(out, indent=2, ensure_ascii=False)
        if out_path:
            out_path.write_text(j, encoding='utf-8')
            print(f"Wrote expected JSON to {out_path}")
        else:
            print(j)
    except Exception as exc:
        print("Error extracting:", exc)
        raise


if __name__ == '__main__':
    main()