Update app.py

app.py

@@ -30,7 +30,7 @@ def search_news_headlines(source: str) -> str:
         Formatted news results or error message
     """
     domain_map = {
-        "BBC News": "bbc.com",
+        "BBC News": "bbc.com/news",
         "CNN": "cnn.com",
         "Reuters": "reuters.com",
         "AP": "apnews.com",
@@ -45,20 +45,29 @@ def search_news_headlines(source: str) -> str:
         if not domain:
             return f"Error: {source} is not a supported news source"
         
-        # Create search query with site filter and time filter
-        search_query = f"site:{domain} latest news after:2023-01-01"
+        # Modified search query to better target actual news articles
+        search_query = f"site:{domain} -inurl:video -inurl:about -inurl:contact -inurl:search news article when:7d"
         
         # Use existing DuckDuckGoSearchTool implementation
         ddg_tool = DuckDuckGoSearchTool(max_results=5)
         raw_results = ddg_tool.forward(search_query)
         
         # Parse the raw results to extract titles and links
-        # DuckDuckGo results come in markdown format: [title](link)
         import re
         headlines = re.findall(r'\[(.*?)\]\((.*?)\)', raw_results)
         
+        # Filter out navigation/section pages
+        filtered_headlines = []
+        for title, link in headlines:
+            if not any(x in title.lower() for x in ['breaking news updates', 'latest news headlines', 'news video', 'video clips', 'news article', 'breaking news', 'news']):
+                filtered_headlines.append((title, link))
+        
         # Format the results
-        formatted_headlines = [f"{i+1}. {title} {link}" for i, (title, link) in enumerate(headlines[:5])]
+        formatted_headlines = [f"{i+1}. {title} {link}" for i, (title, link) in enumerate(filtered_headlines[:5])]
+        
+        if not formatted_headlines:
+            return f"No recent news articles found from {source}. Please try again."
+            
         return f"Here are the latest news from {source}:\n\n" + "\n".join(formatted_headlines)
     
     except Exception as e:
@@ -82,7 +91,7 @@ with open("prompts.yaml", 'r') as stream:
 agent = CodeAgent(
     model=model,
     tools=[final_answer, search_news_headlines], ## add your tools here (don't remove final answer)
-    max_steps=6,
+    max_steps=10,
     verbosity_level=1,
     grammar=None,
     planning_interval=None,