tools/find_logement_sample.py

#!/usr/bin/env python3
"""Find a test sample with logement fields."""
import json
from pathlib import Path

# Find a test sample with logement fields
data = json.loads(Path('data_combined/combined_test_v2.json').read_text(encoding='utf-8'))
samples = [r for r in data if r.get('box_labels') and any('log' in b.lower() for b in r.get('box_labels',[]))]

if samples:
    s = samples[0]
    print(f"Test sample: {s['image_file']}")
    print(f"Doc class: {s['doc_class']}")
    print(f"Logement fields in sample:")
    for lbl, lid, bbox in zip(s.get('box_labels',[]), s.get('box_label_ids',[]), s.get('boxes',[])):
        if 'log' in lbl.lower():
            print(f"  {lbl}: {bbox}")
else:
    print("No test samples with logement fields found")