Update app.py

app.py

@@ -121,28 +121,20 @@ def get_news_headlines(topic: str, count: int = 5) -> str:
         # Define search strategies with different parameters
         search_strategies = [
             {
-                'query': topic,
-                'days_back': 1,
+                'query': f'"{topic}"',  # Exact phrase match
                 'relevance': 'high'
             },
             {
-                'query': f'"{topic}"',  # Exact match
-                'days_back': 7,
+                'query': topic,  # Normal search
                 'relevance': 'high'
             },
             {
-                'query': f"{topic} latest news",
-                'days_back': 30,
-                'relevance': 'medium'
-            },
-            {
-                'query': f"{topic} announcement",
-                'days_back': 30,
+                'query': f"{topic} latest",  # Latest news
                 'relevance': 'medium'
             }
         ]
         
-        all_articles = []
+        relevant_articles = []  # Store only relevant articles
         seen_titles = set()
         required_keywords = set(topic.lower().split())
         
@@ -160,57 +152,69 @@ def get_news_headlines(topic: str, count: int = 5) -> str:
             # Calculate relevance score
             total_score = (title_matches * 3) + (desc_matches * 2) + content_matches
             
+            # For exact phrase matching
             if relevance_level == 'high':
+                # Check if the exact topic phrase appears
+                if topic.lower() in title or topic.lower() in description:
+                    return True
                 return total_score >= len(required_words) * 2
-            elif relevance_level == 'medium':
-                return total_score >= len(required_words)
             else:
-                return total_score > 0
+                return total_score >= len(required_words)
 
         for strategy in search_strategies:
-            if len(all_articles) >= count:
+            if len(relevant_articles) >= count:
                 break
                 
-            # Calculate date range
-            from_date = (datetime.datetime.now() - datetime.timedelta(days=strategy['days_back'])).strftime('%Y-%m-%d')
+            # Calculate how many more articles we need
+            remaining_count = count - len(relevant_articles)
             
-            news = newsapi.get_everything(
-                q=strategy['query'],
-                language='en',
-                sort_by='publishedAt',
-                from_param=from_date,
-                page_size=30  # Get more articles to filter through
-            )
-            
-            if news['articles']:
-                for article in news['articles']:
-                    # Skip if we've seen this title or have enough articles
-                    if article['title'] in seen_titles or len(all_articles) >= count:
-                        continue
-                    
-                    # Check if article is relevant enough
-                    if is_relevant(article, required_keywords, strategy['relevance']):
-                        seen_titles.add(article['title'])
-                        pub_date = datetime.datetime.strptime(article['publishedAt'], '%Y-%m-%dT%H:%M:%SZ')
-                        all_articles.append({
-                            'title': article['title'],
-                            'source': article['source']['name'],
-                            'date': pub_date,
-                            'url': article['url'],
-                            'relevance': strategy['relevance']
-                        })
+            try:
+                news = newsapi.get_everything(
+                    q=strategy['query'],
+                    language='en',
+                    sort_by='relevancy',  # Changed to relevancy sort
+                    page_size=min(50, remaining_count * 5)  # Request more articles to filter through
+                )
+                
+                if news['articles']:
+                    for article in news['articles']:
+                        # Skip if we've seen this title or have enough articles
+                        if article['title'] in seen_titles:
+                            continue
+                        
+                        # Check if article is relevant enough
+                        if is_relevant(article, required_keywords, strategy['relevance']):
+                            seen_titles.add(article['title'])
+                            pub_date = datetime.datetime.strptime(article['publishedAt'], '%Y-%m-%dT%H:%M:%SZ')
+                            relevant_articles.append({
+                                'title': article['title'],
+                                'source': article['source']['name'],
+                                'date': pub_date,
+                                'url': article['url'],
+                                'relevance': strategy['relevance']
+                            })
+                            
+                            # Break if we have enough relevant articles
+                            if len(relevant_articles) >= count:
+                                break
+            except Exception as e:
+                continue  # If one strategy fails, try the next one
         
-        # Sort by date (newest first) and take the top 'count' articles
-        all_articles.sort(key=lambda x: x['date'], reverse=True)
-        all_articles = all_articles[:count]
+        # Sort by date (newest first)
+        relevant_articles.sort(key=lambda x: x['date'], reverse=True)
         
-        if all_articles:
+        if relevant_articles:
             headlines = []
-            for idx, article in enumerate(all_articles, 1):
+            for idx, article in enumerate(relevant_articles, 1):
                 date_str = article['date'].strftime('%Y-%m-%d %H:%M UTC')
                 relevance_indicator = "🎯" if article['relevance'] == 'high' else "✓"
                 headlines.append(f"{idx}. {relevance_indicator} [{date_str}] {article['title']} ({article['source']})")
-            return "\n".join(headlines)
+            
+            # Add a summary of how many relevant articles were found
+            found_count = len(relevant_articles)
+            summary = f"Found {found_count} relevant {'article' if found_count == 1 else 'articles'} out of {count} requested.\n\n"
+            return summary + "\n".join(headlines)
+            
         return f"No relevant news found for topic: {topic}"
     except Exception as e:
         return f"Error fetching news: {str(e)}"