当 AI 垃圾淹没开源：Archestra 的白名单自救实验

#!/usr/bin/env python3
"""ai_spam_detector.py — 检测 PR 中的常见 AI 垃圾模式"""

import re
import sys
import json
from pathlib import Path

# 白名单贡献者（GitHub 用户名）
WHITELIST = [
    "your-trusted-contributor-1",
    "your-trusted-contributor-2",
    # 从项目 CODEOWNERS 或长期贡献者列表中提取
]

# AI 垃圾的典型信号
PATTERNS = {
    "hollow_pr_title": r"^(Update|Improve|Refactor|Fix|Enhance|Optimize)\s+(README|docs|comments|typo)",
    "trivial_style_change": r"^\s*(if\s+\w+\s*==\s*True|for\s+\w+\s+in\s+range\(len\(",
    "ai_generated_disclaimer": r"(This (PR|change) was (generated|created|assisted) by (AI|Copilot|ChatGPT|GPT))",
    "generic_issue_template": r"(I noticed|It seems|This appears to be)\s+(a|an)\s+(issue|bug|problem)",
}

def load_pr_data(pr_json_path: str) -> dict:
    """从 GitHub API 输出或事件 payload 加载 PR 数据"""
    with open(pr_json_path) as f:
        return json.load(f)

def detect_spam_signals(pr: dict, changed_files: list[str]) -> list[str]:
    """返回检测到的垃圾信号列表"""
    signals = []

    # 白名单贡献者直接放行
    author = pr.get("user", {}).get("login", "")
    if author in WHITELIST:
        return signals  # 空列表 = 无垃圾信号

    # 检查 PR 标题
    title = pr.get("title", "")
    for name, pattern in PATTERNS.items():
        if re.search(pattern, title, re.IGNORECASE):
            signals.append(f"标题命中模式: {name}")

    # 检查 PR 正文
    body = pr.get("body", "") or ""
    for name, pattern in PATTERNS.items():
        if re.search(pattern, body, re.IGNORECASE):
            signals.append(f"正文命中模式: {name}")

    # 检查改动文件——全是文档/注释类改动是典型信号
    code_files = [f for f in changed_files
                  if not f.endswith((".md", ".txt", ".rst", ".po"))]
    if len(changed_files) > 3 and len(code_files) == 0:
        signals.append("所有改动文件均为文档类型，无代码变更")

    # 检查改动行数——大量微小改动是批量脚本特征
    additions = pr.get("additions", 0)
    deletions = pr.get("deletions", 0)
    if additions > 50 and deletions > 50 and abs(additions - deletions) < 5:
        signals.append("增删行数几乎对称，疑似批量替换式改动")

    return signals

def main():
    pr_path = sys.argv[1] if len(sys.argv) > 1 else "pr_data.json"
    files_path = sys.argv[2] if len(sys.argv) > 2 else "changed_files.json"

    pr = load_pr_data(pr_path)
    with open(files_path) as f:
        changed_files = json.load(f)

    signals = detect_spam_signals(pr, changed_files)

    if signals:
        print("⚠️  检测到 AI 垃圾信号:")
        for s in signals:
            print(f"  - {s}")
        print("\n建议: 标记 PR 为 needs-human-review，通知维护者人工复核。")
        sys.exit(1)  # 非0退出码，CI 可据此判断
    else:
        print("✅ 未检测到明显垃圾信号，进入正常审查流程。")
        sys.exit(0)

if __name__ == "__main__":
    main()

# 先用 GitHub CLI 获取 PR 数据
gh pr view 42 --json title,body,user,additions,deletions > pr_data.json
# 获取改动文件列表
gh pr diff 42 --name-only > changed_files.json
# 运行检测
python3 ai_spam_detector.py pr_data.json changed_files.json

# .github/workflows/ai-spam-guard.yml
name: AI Spam Guard

on:
  pull_request_target:
    types: [opened, edited, synchronize]

jobs:
  detect:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Fetch PR metadata
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          gh pr view ${{ github.event.pull_request.number }} \
            --repo ${{ github.repository }} \
            --json title,body,user,additions,deletions \
            > pr_data.json

          gh pr diff ${{ github.event.pull_request.number }} \
            --repo ${{ github.repository }} \
            --name-only \
            | python3 -c "import sys,json; print(json.dumps(sys.stdin.read().strip().split('\n')))" \
            > changed_files.json

      - name: Run spam detection
        run: python3 ai_spam_detector.py pr_data.json changed_files.json

      - name: Label suspicious PRs
        if: failure()
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          gh pr edit ${{ github.event.pull_request.number }} \
            --repo ${{ github.repository }} \
            --add-label "needs-human-review"

          gh pr comment ${{ github.event.pull_request.number }} \
            --repo ${{ github.repository }} \
            --body "⚠️ 此 PR 被自动检测标记为疑似 AI 垃圾贡献，需维护者人工复核后再决定是否合并。"