Update app.py

app.py

@@ -118,4 +118,103 @@ def fetch_from_websites(topic, current_time, time_24h_ago):
                     pub_date = (current_time - timedelta(seconds=time_diff)).strftime("%Y-%m-%dT%H:%M:%SZ")
                     
                     desc_elem = item.find_next('p') or item.find_next('div', class_=['summary', 'content', 'desc'])
-                    raw_desc = desc_elem.get_text(strip=True) if desc_elem else f"Latest {topic} update from {url
+                    raw_desc = desc_elem.get_text(strip=True) if desc_elem else f"Latest {topic} update from {url.split('/')[2]}."
+                    
+                    articles.append({
+                        "title": title,
+                        "publishedAt": pub_date,
+                        "raw_description": raw_desc
+                    })
+        except Exception as e:
+            print(f"Failed to fetch from {url}: {e}")
+        if len(articles) >= 10:
+            break
+    
+    while len(articles) < 10:
+        action = random.choice(["Surges", "Expands", "Grows", "Hits"])
+        metric = random.choice(["$500M", "20%", "$1B", "15%"])
+        headline = f"{topic} {action} {metric} in 2025"
+        if headline not in used_headlines:
+            used_headlines.add(headline)
+            time_diff = random.randint(0, 24 * 60 * 60)
+            pub_date = (current_time - timedelta(seconds=time_diff)).strftime("%Y-%m-%dT%H:%M:%SZ")
+            articles.append({
+                "title": headline,
+                "publishedAt": pub_date,
+                "raw_description": f"Simulated {topic} news due to insufficient web data."
+            })
+    
+    return articles
+
+def generate_bedrock_response(model, title, raw_desc, topic, current_date_str):
+    prompt = (
+        f"Generate a 60-80 word news article description for '{topic}' based on: "
+        f"Headline: {title}. Raw data: {raw_desc}. Ensure it’s unique, concise, "
+        f"reflects trends as of {current_date_str}, and focuses on {topic} relevance "
+        f"and economic implications."
+    )
+    body_dict = model["body_template"].copy()
+    body_dict["prompt"] = body_dict["prompt"].format(prompt=prompt)
+    
+    try:
+        response = bedrock.invoke_model(
+            modelId=model["modelId"],
+            contentType=model["contentType"],
+            accept=model["accept"],
+            body=json.dumps(body_dict)
+        )
+        response_body = json.loads(response['body'].read().decode('utf-8'))
+        return (response_body.get('outputs', [{}])[0].get('text', raw_desc) if "mistral" in model["modelId"]
+                else response_body.get('generation', raw_desc)).strip()
+    except Exception as e:
+        print(f"Bedrock model {model['modelId']} failed: {e}")
+        return raw_desc[:400]
+
+def fetch_news(topics, total_articles=10, model_index=0):
+    global latest_articles
+    current_time = datetime.utcnow()
+    time_24h_ago = current_time - timedelta(hours=24)
+    current_date_str = current_time.strftime("%Y-%m-%d")
+    
+    all_articles = []
+    model = MODELS[model_index % len(MODELS)]
+    
+    for topic in topics:
+        raw_articles = fetch_from_websites(topic, current_time, time_24h_ago)
+        for article in raw_articles[:total_articles]:
+            article["description"] = generate_bedrock_response(
+                model, article["title"], article["raw_description"], topic, current_date_str
+            )
+            del article["raw_description"]
+            all_articles.append(article)
+    
+    latest_articles = all_articles[:total_articles * len(topics)]
+    with open('latest_india_news.json', 'w', encoding='utf-8') as f:
+        json.dump(latest_articles, f, indent=4)
+    return latest_articles
+
+def news_fetcher_thread():
+    cycle_count = 0
+    selected_topics = get_user_topics()  # Get topics once at start
+    while True:
+        start_time = time.time()
+        fetch_news(selected_topics, total_articles=10, model_index=cycle_count % len(MODELS))
+        elapsed_time = time.time() - start_time
+        sleep_time = max(300 - elapsed_time, 0)  # 5 minutes = 300 seconds
+        time.sleep(sleep_time)
+        cycle_count += 1
+
+@app.route('/')
+def index():
+    return render_template('index.html', topics=TOPICS)
+
+@app.route('/news')
+def get_news():
+    return jsonify(latest_articles)
+
+if __name__ == "__main__":
+    # Start news fetching in a separate thread
+    fetch_thread = threading.Thread(target=news_fetcher_thread, daemon=True)
+    fetch_thread.start()
+    # Run Flask app
+    app.run(host='0.0.0.0', port=5000, debug=False)