Function report

src-tauri/src-python/scripts/test_activeness_ocr.py:308–346

Match a flat list of OCR names against guild members and print a report.

(label: str, all_names: list[str])

def report(label: str, all_names: list[str]) -> None:
    """Match a flat list of OCR names against guild members and print a report."""
    # Deduplicate keeping last occurrence (mirrors "best activeness" logic)
    seen: dict[str, int] = {}
    for name in all_names:
        seen[name] = seen.get(name, 0) + 1

    matched: dict[str, tuple[str, int]] = {}
    unmatched: list[tuple[str, int, str, float]] = []

    for ocr_name, count in seen.items():
        m, r = best_match(ocr_name)
        if r >= THRESHOLD:
            if m not in matched or count > matched[m][1]:
                matched[m] = (ocr_name, count)
        else:
            unmatched.append((ocr_name, count, m, r))

    missing = [m for m in GUILD_MEMBERS if m not in matched]

    print(f"\n{'=' * 60}")
    print(f"  {label}")
    print(f"  Matched {len(matched)}/{len(GUILD_MEMBERS)} guild members")
    print("=" * 60)

    if missing:
        print(f"\n  Missing ({len(missing)}):")
        for m in missing:
            print(f"    {m!r}")

    if unmatched:
        print(f"\n  Unmatched OCR reads (ratio < {THRESHOLD}):")
        for ocr, cnt, m, r in sorted(unmatched, key=lambda x: -x[1]):
            print(f"    {ocr!r:32s} x{cnt}  best={m!r} ({r:.2f})")

    print("\n  All matched:")
    for gm, (ocr, cnt) in sorted(matched.items(), key=lambda x: x[0].lower()):
        flag = " *" if ocr != gm else ""
        print(f"    {gm!r:30s} <- {ocr!r:30s} x{cnt}{flag}")


# ── Activeness parsing ────────────────────────────────────────────────────────