feat: news categorized by topic (5 categories, 5 each), with subtitles

scripts/summarize_news.py

@@ -1,4 +1,4 @@
-"""Use DeepSeek to summarize top news into 500-word Chinese briefs."""
+"""Use DeepSeek to summarize top news into categorized Chinese briefs (5 per category)."""
 import json, os, requests
 from datetime import datetime, timezone, timedelta
 
@@ -20,38 +20,50 @@ def load_raw_news():
 
 
 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):
+    for i, item in enumerate(items[:30], 1):
         title = item.get("title", "")
         summary = item.get("summary", "")[:150].replace("<", "").replace(">", "")
-        news_text += f"{i}. [{item['source']}] {title}\n   {summary}\n\n"
+        source = item.get("source", "")
+        news_text += f"{i}. [{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。
+    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、新浪新闻、涌报、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。
+请从以上内容中精选出最重要的新闻，按以下分类组织（每个分类最多5条）：
-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：
+分类要求：
-[
+- 国际要闻：全球政治、外交、军事
+- 科技前沿：AI、软件、硬件、互联网  
+- 财经动向：市场、经济、商业
+- 中国热点：中国相关新闻
+- 社会生活：健康、环境、文化、民生
+
+每条新闻格式：
+## 小标题（10字以内概括核心）
+
+正文（200-300字中文精炼摘要，包含核心事实和背景）
+
+输出格式为JSON：
 {{
-    "rank": 1,
+  "categories": [
-    "title": "精炼标题（20字以内）",
+    {{
-    "source": "BBC World",
+      "name": "国际要闻",
-    "summary": "300-500字中文摘要..."
+      "items": [
+        {{"rank": 1, "subtitle": "小标题", "title": "原标题", "source": "BBC World", "summary": "正文..."}},
+        ...
+      ]
     }},
     ...
   ]
+}}
 
-只输出JSON，不要其他内容。
+注意：
+- 每个分类最多5条
+- 小标题要精炼抓人
+- 摘要要有信息量，不要空话
+- 只输出JSON，不要其他内容
 
-今日要闻原始数据：
+原始新闻：
 {news_text}"""
 
     resp = requests.post(
@@ -60,35 +72,39 @@ def summarize(items):
         json={
             "model": MODEL,
             "messages": [{"role": "user", "content": prompt}],
-            "temperature": 0.5,
+            "temperature": 0.4,
-            "max_tokens": 6000,
+            "max_tokens": 8000,
         },
-        timeout=120,
+        timeout=180,
     )
 
     if resp.status_code != 200:
         print(f"API error: {resp.status_code} {resp.text[:200]}")
         return None
 
-    result = resp.json()
+    content = resp.json()["choices"][0]["message"]["content"]
-    content = result["choices"][0]["message"]["content"]
-
-    # Parse JSON from response
     content = content.strip()
     if content.startswith("```"):
-        content = content.split("\n", 1)[1]
+        lines = content.split("\n")
-        content = content.rsplit("\n", 1)[0]
+        content = "\n".join(lines[1:-1]) if lines[-1].strip() == "```" else "\n".join(lines[1:])
-        if content.endswith("```"):
-            content = content[:-3]
-    content = content.strip()
 
     try:
-        articles = json.loads(content)
+        result = json.loads(content)
     except:
-        print(f"JSON parse failed, content: {content[:300]}")
+        # Try to fix common JSON issues
+        content = content.strip()
+        # Remove trailing commas before closing brackets
+        import re
+        content = re.sub(r',\s*}', '}', content)
+        content = re.sub(r',\s*]', ']', content)
+        try:
+            result = json.loads(content)
+        except:
+            print(f"JSON parse failed, len={len(content)}")
+            print(f"First 800 chars: {content[:800]}")
             return None
 
-    return articles
+    return result
 
 
 if __name__ == "__main__":
@@ -98,21 +114,23 @@ if __name__ == "__main__":
         print("No news data")
         exit(0)
 
-    print(f"Loaded {len(items)} raw news items. Summarizing...")
+    print(f"Loaded {len(items)} raw news items. AI summarizing by category...")
-    articles = summarize(items)
+    data = summarize(items)
 
-    if not articles:
+    if not data:
         print("Summarization failed")
         exit(1)
 
     result = {
         "updated_at": now,
-        "total": len(articles),
+        "categories": data.get("categories", []),
-        "articles": articles,
     }
 
+    total = sum(len(c["items"]) for c in result["categories"])
     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}")
+    print(f"Written {total} articles across {len(result['categories'])} categories")
+    for c in result["categories"]:
+        print(f"  {c['name']}: {len(c['items'])} items")