TikTok Brand Safety Audit via API

import os, requests, json

TOKEN = os.environ["SCAVIO_API_KEY"]
AUTH = {"Authorization": f"Bearer {TOKEN}"}
BASE = "https://api.scavio.dev/api/v1/tiktok"

RISK_KEYWORDS = [
    "scam", "fake", "controversy", "cancelled", "exposed",
    "problematic", "lawsuit", "banned", "offensive", "hate"
]

def brand_safety_audit(username: str) -> dict:
    """Run a brand safety audit on a TikTok creator."""
    audit = {"username": username, "risks": [], "score": 100}

    # Check 1: Recent video content analysis
    videos_resp = requests.post(f"{BASE}/user/videos", headers=AUTH,
        json={"username": username, "count": 20})
    videos = videos_resp.json().get("videos", [])

    risky_content = []
    for v in videos:
        desc = (v.get("desc", "") or "").lower()
        for kw in RISK_KEYWORDS:
            if kw in desc:
                risky_content.append({
                    "video": v.get("id"),
                    "keyword": kw,
                    "desc": v.get("desc", "")[:80],
                })

    if risky_content:
        audit["risks"].append({
            "type": "content_risk",
            "details": risky_content,
            "severity": "high" if len(risky_content) > 3 else "medium",
        })
        audit["score"] -= len(risky_content) * 10

    # Check 2: Engagement authenticity
    if videos:
        plays = [v.get("stats", {}).get("playCount", 0) for v in videos]
        likes = [v.get("stats", {}).get("diggCount", 0) for v in videos]
        avg_plays = sum(plays) / len(plays)
        avg_likes = sum(likes) / len(likes)

        # Suspicious: very high likes-to-plays ratio (possible bought engagement)
        if avg_plays > 0 and avg_likes / avg_plays > 0.3:
            audit["risks"].append({
                "type": "engagement_anomaly",
                "details": f"Like/play ratio {avg_likes/avg_plays:.2%} unusually high",
                "severity": "medium",
            })
            audit["score"] -= 15

    # Check 3: Web search for creator controversies
    search_resp = requests.post(
        "https://api.scavio.dev/api/v1/search",
        headers={"x-api-key": TOKEN},
        json={"query": f"@{username} tiktok controversy OR cancelled OR scam",
              "num_results": 5},
    )
    web_results = search_resp.json().get("organic_results", [])
    controversy_hits = [r for r in web_results
                       if any(kw in r.get("snippet", "").lower()
                              for kw in RISK_KEYWORDS)]

    if controversy_hits:
        audit["risks"].append({
            "type": "web_reputation",
            "details": [r["snippet"][:100] for r in controversy_hits],
            "severity": "high",
        })
        audit["score"] -= len(controversy_hits) * 20

    audit["score"] = max(audit["score"], 0)
    audit["recommendation"] = (
        "approve" if audit["score"] >= 70
        else "review" if audit["score"] >= 40
        else "reject"
    )
    return audit

# Audit costs: 1 credit (videos) + 1 credit (web search) = $0.01
# For deeper audit, add profile lookup: +1 credit = $0.015 total

candidates = ["creator_a", "creator_b", "creator_c"]
results = []

for username in candidates:
    audit = brand_safety_audit(username)
    results.append(audit)
    print(f"@{username}: Score {audit['score']}/100 -> {audit['recommendation']}")
    for risk in audit["risks"]:
        print(f"  [{risk['severity']}] {risk['type']}")

# Output:
# @creator_a: Score 85/100 -> approve
# @creator_b: Score 35/100 -> reject
#   [high] content_risk
#   [high] web_reputation
# @creator_c: Score 70/100 -> approve

def monitor_brand_mentions(brand_name: str) -> dict:
    """Check TikTok for brand mentions and sentiment."""
    resp = requests.post(f"{BASE}/search", headers=AUTH,
        json={"query": brand_name, "count": 20})
    videos = resp.json().get("videos", [])

    positive_signals = ["love", "best", "recommend", "favorite", "amazing"]
    negative_signals = ["worst", "terrible", "scam", "avoid", "waste"]

    mentions = {"positive": 0, "negative": 0, "neutral": 0, "total": len(videos)}
    for v in videos:
        desc = (v.get("desc", "") or "").lower()
        if any(s in desc for s in positive_signals):
            mentions["positive"] += 1
        elif any(s in desc for s in negative_signals):
            mentions["negative"] += 1
        else:
            mentions["neutral"] += 1

    mentions["sentiment_score"] = (
        (mentions["positive"] - mentions["negative"])
        / max(mentions["total"], 1) * 100
    )
    return mentions

brand = monitor_brand_mentions("Your Brand Name")
print(f"Sentiment: {brand['sentiment_score']:.0f} "
      f"({brand['positive']} pos / {brand['negative']} neg)")