"""
validate.py  —  Compare extracted CSV/JSON against ground truth.
                 Prints per-field accuracy and a detailed diff.
"""
import json, sys, os
from difflib import SequenceMatcher

def similarity(a: str, b: str) -> float:
    return SequenceMatcher(None, a.lower().strip(), b.lower().strip()).ratio()

def main(gt_path: str, extracted_path: str):
    with open(gt_path) as f:
        gt = json.load(f)
    with open(extracted_path) as f:
        ext = json.load(f)["rows"]

    FIELDS = ["begin_time", "end_time", "patient_name", "dob"]
    total = {k: 0 for k in FIELDS}
    exact = {k: 0 for k in FIELDS}
    fuzzy_sum = {k: 0.0 for k in FIELDS}

    matched = 0
    print(f"Ground truth rows: {len(gt)}   Extracted rows: {len(ext)}\n")
    print(f"{'PG':>3} {'FIELD':<15} {'EXPECTED':<28} {'GOT':<28} {'SIM':>5} {'OK':>3}")
    print("─" * 90)

    for g in gt:
        # find matching row by page + approximate begin_time
        candidates = [e for e in ext if e["page"] == g["page"]]
        best, best_sim = None, 0
        for c in candidates:
            s = similarity(c["begin_time"], g["begin_time"])
            s += similarity(c["patient_name"], g["patient_name"])
            if s > best_sim:
                best_sim = s
                best = c
        if best is None:
            print(f"  p{g['page']} — NO MATCH for {g['begin_time']} {g['patient_name']}")
            for k in FIELDS:
                total[k] += 1
            continue

        matched += 1
        for k in FIELDS:
            total[k] += 1
            sim = similarity(str(best.get(k, "")), str(g[k]))
            fuzzy_sum[k] += sim
            ok = sim >= 0.85
            if ok:
                exact[k] += 1
            flag = "✓" if ok else "✗"
            if not ok:  # only print mismatches
                print(f"{g['page']:>3} {k:<15} {str(g[k]):<28} {str(best.get(k,'')):28} {sim:5.2f} {flag:>3}")

    print("\n" + "=" * 60)
    print(f"{'FIELD':<15} {'EXACT':>6} {'TOTAL':>6} {'ACC':>7}  {'FUZZY_AVG':>9}")
    print("─" * 60)
    for k in FIELDS:
        acc = exact[k] / total[k] * 100 if total[k] else 0
        favg = fuzzy_sum[k] / total[k] if total[k] else 0
        print(f"{k:<15} {exact[k]:>6} {total[k]:>6} {acc:>6.1f}%  {favg:>9.3f}")

    overall = sum(exact.values()) / sum(total.values()) * 100 if sum(total.values()) else 0
    print(f"\nOverall field accuracy : {overall:.1f}%")
    print(f"Row match rate         : {matched}/{len(gt)}")

if __name__ == "__main__":
    gt  = sys.argv[1] if len(sys.argv) > 1 else "./test_pdfs/ground_truth.json"
    ext = sys.argv[2] if len(sys.argv) > 2 else "./output/schedule_extracted.json"
    main(gt, ext)