Comment surveiller les mentions de marque dans les aperçus IA

import os, requests, json, sqlite3
from datetime import datetime

API_KEY = os.environ['SCAVIO_API_KEY']
SH = {'x-api-key': API_KEY, 'Content-Type': 'application/json'}

CONFIG = {
    'brand_terms': ['scavio', 'scavio.dev'],
    'keywords': ['best search api for developers', 'serp api comparison 2026',
                 'google search api python', 'web scraping api alternative',
                 'ai agent search tool', 'mcp search server',
                 'reddit api for agents', 'tiktok data api']
}

db = sqlite3.connect('ao_monitor.db')
db.execute('''CREATE TABLE IF NOT EXISTS ao_checks (
    keyword TEXT, checked_at TEXT, has_ao INTEGER,
    brand_cited INTEGER, ao_snippet TEXT
)''')
db.commit()
print(f'Monitoring {len(CONFIG["keywords"])} keywords for AI Overview brand citations')

def check_ao(keyword, brand_terms):
    data = requests.post('https://api.scavio.dev/api/v1/search',
        headers=SH, json={'query': keyword, 'country_code': 'us',
                          'include_ai_overview': True}).json()
    ao = data.get('ai_overview', {})
    ao_text = json.dumps(ao).lower() if ao else ''
    cited = any(b.lower() in ao_text for b in brand_terms)
    snippet = ao_text[:300] if ao else ''
    now = datetime.now().isoformat()
    db.execute('INSERT INTO ao_checks VALUES (?,?,?,?,?)',
        (keyword, now, 1 if ao else 0, 1 if cited else 0, snippet))
    db.commit()
    return {'keyword': keyword, 'has_ao': bool(ao), 'cited': cited}

results = []
for kw in CONFIG['keywords']:
    r = check_ao(kw, CONFIG['brand_terms'])
    results.append(r)
    icon = 'CITED' if r['cited'] else 'AO' if r['has_ao'] else '---'
    print(f'  [{icon:6}] {kw}')
print(f'\nCost: ${len(CONFIG["keywords"]) * 0.005:.3f}')

def detect_ao_changes(keyword):
    rows = db.execute(
        'SELECT has_ao, brand_cited, checked_at FROM ao_checks WHERE keyword=? ORDER BY checked_at DESC LIMIT 2',
        (keyword,)).fetchall()
    if len(rows) < 2:
        return None
    curr_ao, curr_cited = rows[0][0], rows[0][1]
    prev_ao, prev_cited = rows[1][0], rows[1][1]
    changes = []
    if curr_ao and not prev_ao: changes.append('AO_APPEARED')
    if not curr_ao and prev_ao: changes.append('AO_DISAPPEARED')
    if curr_cited and not prev_cited: changes.append('BRAND_GAINED')
    if not curr_cited and prev_cited: changes.append('BRAND_LOST')
    return changes if changes else None

print('\nChanges detected:')
alerts = []
for kw in CONFIG['keywords']:
    changes = detect_ao_changes(kw)
    if changes:
        alerts.append({'keyword': kw, 'changes': changes})
        for c in changes:
            priority = 'HIGH' if 'LOST' in c or 'DISAPPEARED' in c else 'INFO'
            print(f'  [{priority}] {kw}: {c}')
if not alerts:
    print('  No changes from previous check.')

def daily_summary():
    today = datetime.now().strftime('%Y-%m-%d')
    rows = db.execute(
        'SELECT keyword, has_ao, brand_cited FROM ao_checks WHERE checked_at LIKE ?',
        (f'{today}%',)).fetchall()
    total = len(rows)
    ao_count = sum(1 for r in rows if r[1])
    cited_count = sum(1 for r in rows if r[2])
    print(f'\n=== AI Overview Monitor - {today} ===')
    print(f'  Keywords checked: {total}')
    print(f'  AI Overview present: {ao_count}/{total} ({ao_count/total*100:.0f}%)')
    print(f'  Brand cited: {cited_count}/{total} ({cited_count/total*100:.0f}%)')
    print(f'  Citation rate: {cited_count/ao_count*100:.0f}% of AOs' if ao_count else '  No AOs found')
    print(f'  Daily cost: ${total * 0.005:.3f}')
    print(f'  Monthly estimate: ${total * 0.005 * 30:.2f}')
    # Keywords where brand is cited
    cited_kws = [r[0] for r in rows if r[2]]
    if cited_kws:
        print(f'\n  Brand cited in AO for:')
        for kw in cited_kws:
            print(f'    - {kw}')

daily_summary()