YouTube AI 内容标签大升级：醒目位置 + 自动检测，创作者该怎么应对

"""
批量检查 YouTube 视频是否包含 AI 生成披露关键词。
运行前需要：
1. 在 Google Cloud Console 创建项目，启用 YouTube Data API v3
2. 生成 OAuth 2.0 凭据（客户端 ID + 密钥），下载 client_secrets.json
3. 安装依赖：pip install google-auth-oauthlib google-api-python-client
"""

import os
import re
from google_auth_oauthlib.flow import InstalledAppFlow
from googleapiclient.discovery import build

# OAuth 范围：只读访问
SCOPES = ["https://www.googleapis.com/auth/youtube.read-only"]
CLIENT_SECRETS_FILE = "client_secrets.json"  # 放在脚本同目录下

# AI 披露相关关键词（中英文）
AI_KEYWORDS = [
    "AI-generated", "AI 生成", "artificial intelligence",
    "generated by AI", "AI 辅助", "AI 合成",
    "synthetic", "deepfake", "AI 修改",
]

def get_authenticated_service():
    flow = InstalledAppFlow.from_client_secrets_file(
        CLIENT_SECRETS_FILE, SCOPES
    )
    credentials = flow.run_local_server(port=0)
    return build("youtube", "v3", credentials=credentials)


def fetch_channel_videos(youtube, channel_id, max_results=50):
    """拉取频道最近上传的视频列表"""
    request = youtube.search().list(
        part="snippet",
        channelId=channel_id,
        type="video",
        order="date",
        maxResults=max_results,
    )
    response = request.execute()
    return response.get("items", [])


def check_ai_disclosure(description: str) -> list[str]:
    """检查描述中是否包含 AI 披露关键词，返回匹配到的关键词列表"""
    if not description:
        return []
    found = []
    for kw in AI_KEYWORDS:
        if re.search(kw, description, re.IGNORECASE):
            found.append(kw)
    return found


def main():
    youtube = get_authenticated_service()

    # 替换为你的频道 ID（在 YouTube 频道页 URL 中可以找到）
    CHANNEL_ID = os.environ.get("YT_CHANNEL_ID", "UC_x5XG1OV2P6uZZ5FSM9Ttw")

    videos = fetch_channel_videos(youtube, CHANNEL_ID)

    print(f"共拉取 {len(videos)} 个视频，开始检查 AI 披露状态...\n")

    undisclosed_ai = []
    for v in videos:
        snippet = v["snippet"]
        title = snippet["title"]
        desc = snippet.get("description", "")
        video_id = v["id"]["videoId"]
        matched = check_ai_disclosure(desc)

        status = "✅ 已披露" if matched else "❌ 未发现披露关键词"
        print(f"[{status}] {title}")
        print(f"  链接: https://youtube.com/watch?v={video_id}")
        if matched:
            print(f"  匹配关键词: {', '.join(matched)}")
        print()

        # 如果标题含 AI 但描述没披露，标记为潜在遗漏
        title_has_ai = any(
            re.search(kw, title, re.IGNORECASE) for kw in AI_KEYWORDS
        )
        if title_has_ai and not matched:
            undisclosed_ai.append((title, video_id))

    if undisclosed_ai:
        print("=" * 50)
        print("⚠️ 以下视频标题含 AI 关键词但描述未披露，建议补充：")
        for t, vid in undisclosed_ai:
            print(f"  - {t} → https://youtube.com/watch?v={vid}")
    else:
        print("✅ 所有视频的 AI 披露状态看起来一致。")


if __name__ == "__main__":
    main()

# 设置频道 ID 环境变量
export YT_CHANNEL_ID="你的频道ID"

# 首次运行会弹出浏览器要求 OAuth 授权
python check_ai_disclosure.py