ghstedpixel/app.py

1import re2import httpx3from config import SAFE_BROWSING_API4 5def scrub_conversational_bloat(query: str) -> str:6    """Strips conversational noise words and store hooks to maximize index matching accuracy"""7    if query.startswith(("http://", "https://")): return query8    noise_pattern = r'\b(buy|cheap|online|sale|authentic|store|deals?|discount|get|best|price|lowest)\b'9    scrubbed = re.sub(noise_pattern, '', query.lower())10    clean_string = " ".join(scrubbed.split())11    if clean_string != query.lower():12        print(f"Core normalization: Cleaned token string layout from '{query}' -> '{clean_string}'")13    return clean_string if clean_string else query14 15async def check_url_safe(url: str, client: httpx.AsyncClient) -> bool:16    """Vets outbound destination URLs against malware and deceptive targets using Google Safe Browsing APIs"""17    if not SAFE_BROWSING_API:18        print("⚠️ WARNING: SAFE_BROWSING_API key missing. Bypassing live link validation safety verification.")19        return True20    api_url = f"https://safebrowsing.googleapis.com/v4/threatMatches:find?key={SAFE_BROWSING_API}"21    payload = {22        "client": {"clientId": "tarz-ai-search-agent", "clientVersion": "1.1.0"},23        "threatInfo": {24            "threatTypes": ["MALWARE", "SOCIAL_ENGINEERING", "UNWANTED_SOFTWARE", "POTENTIALLY_HARMFUL_APPLICATION"],25            "platformTypes": ["ANY_PLATFORM"],26            "threatEntryTypes": ["URL"],27            "threatEntries": [{"url": url}]28        }29    }30    try:31        response = await client.post(api_url, json=payload, timeout=5.0)32        if response.status_code == 200:33            res_data = response.json()34            if "matches" in res_data and len(res_data["matches"]) > 0:35                print(f"❌ [Security Flag] Blocked malicious link threat match vector: {url}")36                return False37            return True38        else:39            print(f"⚠️ [Safe Browsing Error] API returned state {response.status_code}. Defaulting shut for containment.")40            return False41    except Exception as e:42        print(f"❌ [Safe Browsing Exception] Connection failed: {e}. Isolation fallback activated.")43        return False