#!/usr/bin/env python3
"""
Affiche les détails des faux positifs à partir des résultats d'évaluation.
"""

import json
from pathlib import Path
from collections import defaultdict, Counter

# Charger l'évaluation
eval_file = Path("tests/ground_truth/quality_evaluation/baseline_quality_evaluation.json")
with open(eval_file, 'r', encoding='utf-8') as f:
    eval_data = json.load(f)

# Analyser les types problématiques
problematic_types = {
    'EPISODE': 106,
    'VILLE': 20,
    'CODE_POSTAL': 10,
    'ADRESSE': 10,
    'TEL': 8
}

print("=" * 80)
print("ANALYSE DES FAUX POSITIFS PAR TYPE")
print("=" * 80)

# Collecter tous les exemples de détections
all_detections = defaultdict(list)

for doc in eval_data['per_document']:
    pdf_name = doc['pdf']
    audit_file = Path(f"tests/ground_truth/pdfs/baseline_anonymized/{pdf_name}.audit.jsonl")
    
    if not audit_file.exists():
        continue
    
    with open(audit_file, 'r', encoding='utf-8') as f:
        for line in f:
            det = json.loads(line)
            kind = det.get('kind', 'UNKNOWN')
            original = det.get('original', '')
            page = det.get('page', -1)
            
            all_detections[kind].append({
                'text': original,
                'page': page,
                'file': pdf_name
            })

# Afficher les statistiques pour chaque type problématique
for pii_type, expected_fp in problematic_types.items():
    detections = all_detections.get(pii_type, [])
    
    print(f"\n{'=' * 80}")
    print(f"Type: {pii_type}")
    print(f"Faux positifs attendus: {expected_fp}")
    print(f"Détections totales: {len(detections)}")
    print(f"{'=' * 80}")
    
    # Compter les occurrences
    text_counter = Counter(d['text'] for d in detections)
    
    print(f"\nTextes les plus fréquents:")
    for text, count in text_counter.most_common(30):
        print(f"  {count:3d}x '{text}'")
    
    # Afficher quelques exemples avec contexte
    print(f"\nExemples avec fichier:")
    seen = set()
    for d in detections[:20]:
        key = (d['text'], d['file'])
        if key not in seen:
            seen.add(key)
            print(f"  '{d['text']}' (page {d['page']}) - {d['file']}")

print("\n" + "=" * 80)