Detect Content Theft with SERP API Monitoring

import requests, os, re
from urllib.parse import urlparse

API_KEY = os.environ["SCAVIO_API_KEY"]
HEADERS = {"x-api-key": API_KEY, "Content-Type": "application/json"}
SEARCH_URL = "https://api.scavio.dev/api/v1/search"

YOUR_DOMAINS = {"yourdomain.com", "www.yourdomain.com"}

def extract_fingerprints(text, count=5):
    """Pick unique mid-article sentences as fingerprints."""
    sentences = [s.strip() for s in re.split(r'[.!?]', text)
                 if len(s.strip()) > 60]
    # Skip first and last -- intros/outros are generic
    middle = sentences[2:-2] if len(sentences) > 6 else sentences
    return [f'"{s}"' for s in middle[:count]]

def search_for_copies(phrase):
    """Search Google for an exact phrase and return non-owned results."""
    resp = requests.post(SEARCH_URL, headers=HEADERS, json={
        "platform": "google",
        "query": phrase,
        "country": "us"
    })
    results = resp.json().get("data", {}).get("organic", [])
    copies = []
    for r in results:
        domain = urlparse(r["link"]).netloc.replace("www.", "")
        if domain not in YOUR_DOMAINS:
            copies.append({
                "title": r["title"],
                "url": r["link"],
                "domain": domain,
                "snippet": r.get("snippet", "")
            })
    return copies

def detect_theft(article_text, article_url):
    """Run full theft detection on one article."""
    fingerprints = extract_fingerprints(article_text)
    all_copies = {}

    for phrase in fingerprints:
        copies = search_for_copies(phrase)
        for copy in copies:
            domain = copy["domain"]
            if domain not in all_copies:
                all_copies[domain] = {
                    "url": copy["url"],
                    "title": copy["title"],
                    "phrase_matches": 0
                }
            all_copies[domain]["phrase_matches"] += 1

    # Sort by confidence (more phrase matches = more likely a copy)
    ranked = sorted(all_copies.values(),
                    key=lambda x: x["phrase_matches"], reverse=True)

    print(f"Checked: {article_url}")
    print(f"Fingerprints searched: {len(fingerprints)}")
    print(f"Suspect domains found: {len(ranked)}")
    for r in ranked:
        confidence = r["phrase_matches"] / len(fingerprints) * 100
        print(f"  [{confidence:.0f}%] {r['url']}")

    return ranked

# Cost: 5 fingerprints x 1 credit = 5 credits ($0.025) per article
detect_theft(open("my-article.txt").read(), "https://yourdomain.com/my-article")

import json
from datetime import datetime

def weekly_scan(articles):
    """Scan all articles and save results."""
    report = {"date": datetime.now().isoformat(), "findings": []}
    for article in articles:
        copies = detect_theft(article["text"], article["url"])
        if copies:
            report["findings"].append({
                "article": article["url"],
                "copies": copies
            })

    with open(f"theft-report-{datetime.now().strftime('%Y-%m-%d')}.json", "w") as f:
        json.dump(report, f, indent=2)

    high_confidence = [f for f in report["findings"]
                       if any(c["phrase_matches"] >= 3 for c in f["copies"])]
    print(f"High-confidence theft detected on {len(high_confidence)} articles")
    return report