import re, json

with open(r'D:\Evolução categorias\dashboard_categorias.html', encoding='utf-8') as f:
    html = f.read()

m = re.search(r'const DB\s*=\s*(\{[\s\S]*?\});\s*\nconst allLabels', html)
if m:
    db = json.loads(m.group(1))
    labels = db['labels']
    print('Total labels:', len(labels))
    print('First 5:', labels[:5])
    print('Last 5:', labels[-5:])
    # check Jan/2025
    if 'Jan/2025' in labels:
        print('Jan/2025 found at index:', labels.index('Jan/2025'))
    else:
        print('Jan/2025 NOT in labels!')
        # find any Jan
        jan_labels = [l for l in labels if 'Jan' in l or 'jan' in l]
        print('Jan labels found:', jan_labels)
    cats = db['categories']
    print('Categories:', cats)
    # check data for first category
    c0 = cats[0]
    vals = db['data'][c0]
    print(f'{c0} - non-zero values: {sum(1 for v in vals if v > 0)}/{len(vals)}')