Update tools/wikipedia_search.py

tools/wikipedia_search.py

@@ -1,33 +1,21 @@
 import requests
 
 def wikipedia_search(query: str) -> str:
-    """
-    Search Wikipedia and return the full page content (up to 200,000 chars) of the best matching article.
-    If the page contains a 'Discography' section, it will try to keep it intact.
-    """
     try:
         base_url = "https://en.wikipedia.org/w/api.php"
         headers = {"User-Agent": "langgraph-gaia-agent/1.0"}
         
-        # 1. 先尝试作为精确页面标题访问
+        # 1. 尝试直接作为页面标题访问
         direct_title = query.replace(" ", "_")
-        check_params = {
-            "action": "query",
-            "titles": direct_title,
-            "format": "json"
-        }
+        check_params = {"action": "query", "titles": direct_title, "format": "json"}
         r = requests.get(base_url, params=check_params, headers=headers, timeout=10)
         data = r.json()
         pages = data.get("query", {}).get("pages", {})
         if any(pid != "-1" for pid in pages):
             # 精确页面存在
             extract_params = {
-                "action": "query",
-                "prop": "extracts",
-                "explaintext": True,
-                "redirects": 1,
-                "titles": direct_title,
-                "format": "json",
+                "action": "query", "prop": "extracts", "explaintext": True,
+                "redirects": 1, "titles": direct_title, "format": "json"
             }
             r = requests.get(base_url, params=extract_params, headers=headers, timeout=10)
             data = r.json()
@@ -35,28 +23,12 @@ def wikipedia_search(query: str) -> str:
                 extract = page.get("extract", "")
                 title = page.get("title", direct_title)
                 url = "https://en.wikipedia.org/wiki/" + title.replace(" ", "_")
-                # 提高截断阈值至 200,000 字符（约 50k tokens，仍可能溢出但概率极低）
-                max_len = 200000
-                if len(extract) > max_len:
-                    # 尝试保留 Discography 章节
-                    discog_idx = extract.lower().find("discography")
-                    if discog_idx != -1:
-                        # 保留从 Discography 前 5000 到文末（或后 150000）
-                        start = max(0, discog_idx - 5000)
-                        end = min(len(extract), discog_idx + 150000)
-                        extract = extract[start:end] + "\n...[truncated from Wikipedia full page, but Discography section preserved]"
-                    else:
-                        extract = extract[:max_len] + "\n...[truncated]"
+                if len(extract) > 8000:
+                    extract = extract[:8000] + "\n...[truncated]"
                 return f"# {title}\nURL: {url}\n\n{extract}"
         
-        # 2. 精确页面不存在，执行搜索
-        search_params = {
-            "action": "query",
-            "list": "search",
-            "srsearch": query,
-            "srlimit": 1,
-            "format": "json",
-        }
+        # 2. 搜索
+        search_params = {"action": "query", "list": "search", "srsearch": query, "srlimit": 1, "format": "json"}
         r = requests.get(base_url, params=search_params, headers=headers, timeout=10)
         data = r.json()
         hits = data.get("query", {}).get("search", [])
@@ -64,29 +36,17 @@ def wikipedia_search(query: str) -> str:
             return f"No Wikipedia article found for '{query}'."
         title = hits[0]["title"]
         
-        # 3. 获取完整页面内容
         extract_params = {
-            "action": "query",
-            "prop": "extracts",
-            "explaintext": True,
-            "redirects": 1,
-            "titles": title,
-            "format": "json",
+            "action": "query", "prop": "extracts", "explaintext": True,
+            "redirects": 1, "titles": title, "format": "json"
         }
         r = requests.get(base_url, params=extract_params, headers=headers, timeout=10)
         data = r.json()
         for page in data.get("query", {}).get("pages", {}).values():
             extract = page.get("extract", "")
             url = "https://en.wikipedia.org/wiki/" + title.replace(" ", "_")
-            max_len = 200000
-            if len(extract) > max_len:
-                discog_idx = extract.lower().find("discography")
-                if discog_idx != -1:
-                    start = max(0, discog_idx - 5000)
-                    end = min(len(extract), discog_idx + 150000)
-                    extract = extract[start:end] + "\n...[truncated]"
-                else:
-                    extract = extract[:max_len] + "\n...[truncated]"
+            if len(extract) > 8000:
+                extract = extract[:8000] + "\n...[truncated]"
             return f"# {title}\nURL: {url}\n\n{extract}"
         
         return f"Could not retrieve content for '{query}'."