feat: AI news summarization - top 10 articles with 500-word Chinese briefs

2026-06-14 13:21:03 +08:00
parent b3a3034b2f
commit 4cd8089480
1 changed files with 118 additions and 0 deletions
--- a/scripts/summarize_news.py
+++ b/scripts/summarize_news.py
@ -0,0 +1,118 @@
+"""Use DeepSeek to summarize top news into 500-word Chinese briefs."""
+import json, os, requests
+from datetime import datetime, timezone, timedelta
+
+API_KEY = "sk-bbca4a0380d549389f0d27cdea0b5228"
+API_URL = "https://api.deepseek.com/v1/chat/completions"
+MODEL = "deepseek-v4-flash"
+
+NEWS_FILE = "/var/www/nav/data/news.json"
+OUTPUT = "/var/www/nav/data/news_summary.json"
+tz = timezone(timedelta(hours=8))
+
+
+def load_raw_news():
+    if not os.path.exists(NEWS_FILE):
+        return []
+    with open(NEWS_FILE) as f:
+        data = json.load(f)
+    return data.get("items", [])
+
+
+def summarize(items):
+    """Send raw news to DeepSeek and get summarized articles."""
+    # Build a condensed input for the AI (titles + summaries)
+    news_text = ""
+    for i, item in enumerate(items[:20], 1):
+        title = item.get("title", "")
+        summary = item.get("summary", "")[:150].replace("<", "").replace(">", "")
+        news_text += f"{i}. [{item['source']}] {title}\n   {summary}\n\n"
+
+    prompt = f"""你是一个新闻整理助手。以下是来自BBC世界新闻、BBC中文、Hacker News的最新资讯。
+
+请从以上内容中精选出最重要的10条新闻，每条用中文写成一段300-500字的精炼摘要。
+要求：
+1. 选择真正重要、有信息量的新闻（优先世界大事、科技突破、财经动向）
+2. 每条摘要独立、完整，包含核心事实和背景
+3. 语言简洁有力，适合快速阅读
+4. 按重要性排序
+5. 每条附上来源标签
+
+输出格式为JSON数组：
+[
+  {{
+    "rank": 1,
+    "title": "精炼标题（20字以内）",
+    "source": "BBC World",
+    "summary": "300-500字中文摘要..."
+  }},
+  ...
+]
+
+只输出JSON，不要其他内容。
+
+今日要闻原始数据：
+{news_text}"""
+
+    resp = requests.post(
+        API_URL,
+        headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
+        json={
+            "model": MODEL,
+            "messages": [{"role": "user", "content": prompt}],
+            "temperature": 0.5,
+            "max_tokens": 6000,
+        },
+        timeout=120,
+    )
+
+    if resp.status_code != 200:
+        print(f"API error: {resp.status_code} {resp.text[:200]}")
+        return None
+
+    result = resp.json()
+    content = result["choices"][0]["message"]["content"]
+
+    # Parse JSON from response
+    content = content.strip()
+    if content.startswith("```"):
+        content = content.split("\n", 1)[1]
+        content = content.rsplit("\n", 1)[0]
+        if content.endswith("```"):
+            content = content[:-3]
+    content = content.strip()
+
+    try:
+        articles = json.loads(content)
+    except:
+        print(f"JSON parse failed, content: {content[:300]}")
+        return None
+
+    return articles
+
+
+if __name__ == "__main__":
+    now = datetime.now(tz).strftime("%Y-%m-%d %H:%M")
+    items = load_raw_news()
+    if not items:
+        print("No news data")
+        exit(0)
+
+    print(f"Loaded {len(items)} raw news items. Summarizing...")
+    articles = summarize(items)
+
+    if not articles:
+        print("Summarization failed")
+        exit(1)
+
+    result = {
+        "updated_at": now,
+        "total": len(articles),
+        "articles": articles,
+    }
+
+    os.makedirs(os.path.dirname(OUTPUT), exist_ok=True)
+    with open(OUTPUT, "w", encoding="utf-8") as f:
+        json.dump(result, f, ensure_ascii=False, indent=2)
+
+    print(f"Written {len(articles)} summarized articles to {OUTPUT}")