import os
import json
from barcode_scanner import scan_all_barcodes
from ocr import read_chassis, postprocess_with_hint

BARCODE_DIR = "images/barcode"
CHASSIS_DIR = "images/chassis"
RESULTS_DIR = "results"


def get_pairs():
    barcode_files = {os.path.splitext(f)[0]: f
                     for f in os.listdir(BARCODE_DIR)
                     if f.lower().endswith(('.jpg', '.jpeg', '.png'))}
    chassis_files = {os.path.splitext(f)[0]: f
                     for f in os.listdir(CHASSIS_DIR)
                     if f.lower().endswith(('.jpg', '.jpeg', '.png'))}

    common = sorted(set(barcode_files.keys()) & set(chassis_files.keys()))
    pairs = []
    for key in common:
        pairs.append({
            "key": key,
            "barcode_path": os.path.join(BARCODE_DIR, barcode_files[key]),
            "chassis_path": os.path.join(CHASSIS_DIR, chassis_files[key]),
        })
    return pairs


def evaluate():
    os.makedirs(RESULTS_DIR, exist_ok=True)

    print("=" * 60)
    print("CHASSIS OCR EVALUATION")
    print("=" * 60)

    print("\n[1/3] Scanning barcodes for ground truth...")
    barcode_results = scan_all_barcodes(BARCODE_DIR)

    pairs = get_pairs()
    print(f"\n[2/3] Found {len(pairs)} matching image pairs")

    print(f"\n[3/3] Running OCR pipeline on chassis images...\n")

    results = []
    exact_match   = 0
    corrected_match = 0
    failed        = 0

    for pair in pairs:
        key          = pair["key"]
        expected     = barcode_results.get(key)
        chassis_path = pair["chassis_path"]

        if not expected:
            print(f"  [WARN] {key} - barcode not decoded, skipping")
            continue

        ocr_text, conf = read_chassis(chassis_path, save_comparison=True)

        corrected, is_match = postprocess_with_hint(ocr_text, expected)

        if ocr_text == expected:
            status = "[EXACT]"
            exact_match += 1
        elif is_match:
            status = "[CORRECTED]"
            corrected_match += 1
        else:
            status = "[FAILED]"
            failed += 1

        print(f"  {status} | {key}")
        print(f"           Expected : {expected}")
        print(f"           Got      : {ocr_text}  (conf: {conf:.0%})")
        if is_match and ocr_text != expected:
            print(f"           Fixed to : {corrected}")
        print()

        results.append({
            "key": key,
            "expected": expected,
            "ocr_raw": ocr_text,
            "corrected": corrected,
            "confidence": round(conf, 3),
            "match": is_match,
            "exact": ocr_text == expected,
        })

    total = len(results)
    total_correct = exact_match + corrected_match
    print("=" * 60)
    print("RESULTS SUMMARY")
    print("=" * 60)
    print(f"Total pairs evaluated : {total}")
    print(f"Exact matches         : {exact_match}/{total} ({exact_match/total*100:.1f}%)")
    print(f"Corrected matches     : {corrected_match}/{total} ({corrected_match/total*100:.1f}%)")
    print(f"Total correct         : {total_correct}/{total} ({total_correct/total*100:.1f}%)")
    print(f"Failed                : {failed}/{total} ({failed/total*100:.1f}%)")
    print("=" * 60)

    if failed > 0:
        print("\nFailed images (focus preprocessing tuning here):")
        for r in results:
            if not r["match"]:
                print(f"  - {r['key']}: expected '{r['expected']}', got '{r['ocr_raw']}'")

    report_path = os.path.join(RESULTS_DIR, "report.json")
    with open(report_path, "w") as f:
        json.dump(results, f, indent=2)
    print(f"\nFull report saved -> {report_path}")
    print(f"Comparison images -> {RESULTS_DIR}/")


if __name__ == "__main__":
    evaluate()