Update app.py

app.py

@@ -118,39 +118,87 @@ def get_news_headlines(topic: str, count: int = 5) -> str:
     newsapi = NewsApiClient(api_key=API_KEY)
     
     try:
-        # Try different search queries to get more relevant results
-        queries = [
-            topic,  # Original topic
-            f"{topic} latest",  # Latest news
-            f"{topic} important",  # Important news
+        # Define search strategies with different parameters
+        search_strategies = [
+            {
+                'query': topic,
+                'days_back': 1,
+                'relevance': 'high'
+            },
+            {
+                'query': f'"{topic}"',  # Exact match
+                'days_back': 7,
+                'relevance': 'high'
+            },
+            {
+                'query': f"{topic} latest news",
+                'days_back': 30,
+                'relevance': 'medium'
+            },
+            {
+                'query': f"{topic} announcement",
+                'days_back': 30,
+                'relevance': 'medium'
+            }
         ]
         
         all_articles = []
-        seen_titles = set()  # To avoid duplicates
+        seen_titles = set()
+        required_keywords = set(topic.lower().split())
         
-        for query in queries:
+        # Function to check article relevance
+        def is_relevant(article, required_words, relevance_level):
+            title = article['title'].lower()
+            description = (article.get('description') or '').lower()
+            content = (article.get('content') or '').lower()
+            
+            # Count how many required words appear in the article
+            title_matches = sum(1 for word in required_words if word in title)
+            desc_matches = sum(1 for word in required_words if word in description)
+            content_matches = sum(1 for word in required_words if word in content)
+            
+            # Calculate relevance score
+            total_score = (title_matches * 3) + (desc_matches * 2) + content_matches
+            
+            if relevance_level == 'high':
+                return total_score >= len(required_words) * 2
+            elif relevance_level == 'medium':
+                return total_score >= len(required_words)
+            else:
+                return total_score > 0
+
+        for strategy in search_strategies:
+            if len(all_articles) >= count:
+                break
+                
+            # Calculate date range
+            from_date = (datetime.datetime.now() - datetime.timedelta(days=strategy['days_back'])).strftime('%Y-%m-%d')
+            
             news = newsapi.get_everything(
-                q=query,
+                q=strategy['query'],
                 language='en',
                 sort_by='publishedAt',
-                page_size=count
+                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 before
-                    if article['title'] in seen_titles:
+                    # Skip if we've seen this title or have enough articles
+                    if article['title'] in seen_titles or len(all_articles) >= count:
                         continue
                     
-                    seen_titles.add(article['title'])
-                    # Convert UTC timestamp to datetime
-                    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']
-                    })
+                    # 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']
+                        })
         
         # Sort by date (newest first) and take the top 'count' articles
         all_articles.sort(key=lambda x: x['date'], reverse=True)
@@ -160,9 +208,10 @@ def get_news_headlines(topic: str, count: int = 5) -> str:
             headlines = []
             for idx, article in enumerate(all_articles, 1):
                 date_str = article['date'].strftime('%Y-%m-%d %H:%M UTC')
-                headlines.append(f"{idx}. [{date_str}] {article['title']} ({article['source']})")
+                relevance_indicator = "🎯" if article['relevance'] == 'high' else "✓"
+                headlines.append(f"{idx}. {relevance_indicator} [{date_str}] {article['title']} ({article['source']})")
             return "\n".join(headlines)
-        return f"No news found for topic: {topic}"
+        return f"No relevant news found for topic: {topic}"
     except Exception as e:
         return f"Error fetching news: {str(e)}"