马斯克诉 OpenAI 败诉：诉讼时效掐断 1350 亿美元索赔，但核心争议远未结束

"""
openai_policy_tracker.py
定期抓取 OpenAI 公开政策/使用条款页面的关键段落，
将每次抓取结果存入本地 JSON 文件，便于对比变更历史。

依赖：pip install requests beautifulsoup4
运行：python openai_policy_tracker.py
"""

import json
import hashlib
import datetime
import os
from pathlib import Path

import requests
from bs4 import BeautifulSoup

# 要监控的页面及其关键段落选择器
PAGES = {
    "usage_policies": {
        "url": "https://openai.com/policies/usage-policies/",
        "selectors": ["h2", "h3"],  # 抓取所有二级和三级标题段落
    },
    "terms_of_use": {
        "url": "https://openai.com/policies/terms-of-use/",
        "selectors": ["h2", "h3"],
    },
}

DATA_DIR = Path("policy_snapshots")
DATA_DIR.mkdir(exist_ok=True)


def fetch_page(url: str) -> str:
    resp = requests.get(url, timeout=30, headers={"User-Agent": "PolicyTracker/1.0"})
    resp.raise_for_status()
    return resp.text


def extract_sections(html: str, selectors: list[str]) -> dict[str, str]:
    soup = BeautifulSoup(html, "html.parser")
    sections = {}
    for sel in selectors:
        for tag in soup.select(sel):
            # 取标题及其后续段落文本
            title = tag.get_text(strip=True)
            content_parts = []
            for sibling in tag.next_siblings:
                if sibling.name in selectors:
                    break
                if sibling.name in ("p", "li", "ul", "ol"):
                    content_parts.append(sibling.get_text(strip=True))
            sections[title] = " ".join(content_parts)
    return sections


def compute_hash(sections: dict[str, str]) -> str:
    raw = json.dumps(sections, sort_keys=True, ensure_ascii=False)
    return hashlib.sha256(raw.encode()).hexdigest()[:16]


def save_snapshot(page_name: str, sections: dict[str, str]):
    now = datetime.datetime.now().isoformat(timespec="seconds")
    content_hash = compute_hash(sections)
    filepath = DATA_DIR / f"{page_name}.json"

    history = []
    if filepath.exists():
        history = json.loads(filepath.read_text())

    # 只在内容变化时追加新记录
    if history and history[-1]["hash"] == content_hash:
        print(f"[{page_name}] 内容无变化，跳过记录")
        return

    record = {
        "timestamp": now,
        "hash": content_hash,
        "sections": sections,
    }
    history.append(record)
    filepath.write_text(json.dumps(history, ensure_ascii=False, indent=2))
    print(f"[{page_name}] 新快照已保存，hash={content_hash}")


def main():
    for page_name, config in PAGES.items():
        try:
            html = fetch_page(config["url"])
            sections = extract_sections(html, config["selectors"])
            save_snapshot(page_name, sections)
        except Exception as e:
            print(f"[{page_name}] 抓取失败: {e}")


if __name__ == "__main__":
    main()