Spaces:
Configuration error
Configuration error
| """Report OCR recall and snapping quality per page, from the built index. | |
| python tools/ocr_report.py | |
| """ | |
| from __future__ import annotations | |
| import sys | |
| from pathlib import Path | |
| sys.path.insert(0, str(Path(__file__).resolve().parent.parent)) | |
| from spareparts import config, index_db # noqa: E402 | |
| def main() -> int: | |
| conn = index_db.connect(config.DB_PATH) | |
| try: | |
| rows = conn.execute( | |
| """ | |
| SELECT p.pdf_page, | |
| (SELECT COUNT(DISTINCT part_no) FROM occurrences o | |
| WHERE o.pdf_page = p.pdf_page) AS table_parts, | |
| (SELECT COUNT(*) FROM labels l WHERE l.pdf_page = p.pdf_page) AS callouts, | |
| (SELECT COUNT(DISTINCT l.part_no) FROM labels l | |
| JOIN occurrences o2 ON o2.part_no = l.part_no | |
| AND o2.pdf_page = l.pdf_page | |
| WHERE l.pdf_page = p.pdf_page) AS matched, | |
| (SELECT COUNT(*) FROM labels l WHERE l.pdf_page = p.pdf_page | |
| AND l.match_kind = 'fuzzy1') AS fuzzy, | |
| (SELECT COUNT(*) FROM labels l WHERE l.pdf_page = p.pdf_page | |
| AND l.in_page_table = 0) AS off_table | |
| FROM pages p ORDER BY p.pdf_page | |
| """ | |
| ).fetchall() | |
| print(f"{'page':>5} {'table':>6} {'found':>6} {'recall':>7} {'callouts':>9}" | |
| f" {'fuzzy':>6} {'not-in-table':>13}") | |
| total_table = total_found = total_callouts = total_fuzzy = total_off = 0 | |
| for r in rows: | |
| recall = 100 * r["matched"] / r["table_parts"] if r["table_parts"] else 0 | |
| print(f"{r['pdf_page']:>5} {r['table_parts']:>6} {r['matched']:>6}" | |
| f" {recall:>6.1f}% {r['callouts']:>9} {r['fuzzy']:>6} {r['off_table']:>13}") | |
| total_table += r["table_parts"]; total_found += r["matched"] | |
| total_callouts += r["callouts"]; total_fuzzy += r["fuzzy"] | |
| total_off += r["off_table"] | |
| print(f"{'ALL':>5} {total_table:>6} {total_found:>6}" | |
| f" {100 * total_found / total_table:>6.1f}% {total_callouts:>9}" | |
| f" {total_fuzzy:>6} {total_off:>13}") | |
| print(f"\nDistinct parts with at least one callout: " | |
| f"{conn.execute('SELECT COUNT(DISTINCT part_no) FROM labels').fetchone()[0]}" | |
| f" of {conn.execute('SELECT COUNT(*) FROM parts').fetchone()[0]}") | |
| print("Exact vs fuzzy snaps: " + ", ".join( | |
| f"{r[0]}={r[1]}" for r in conn.execute( | |
| "SELECT match_kind, COUNT(*) FROM labels GROUP BY match_kind"))) | |
| finally: | |
| conn.close() | |
| return 0 | |
| if __name__ == "__main__": | |
| sys.exit(main()) | |