Update src/streamlit_app.py

src/streamlit_app.py

@@ -1,4 +1,3 @@
-
 import streamlit as st
 import feedparser
 import requests
@@ -7,94 +6,135 @@ import os
 from bs4 import BeautifulSoup
 from transformers import MarianMTModel, MarianTokenizer
 import whisper
-from streamlit_audio_recorder import st_audiorec
+from streamlit_audio_recorder import st_audiorec # Make sure this library is installed
 
+# Configure the Streamlit page
 st.set_page_config(page_title="📍 BharatPulse - Local News & Sentiment", layout="wide")
 
-# ------------------------------
-# User Auth Functions
-# ------------------------------
+# --- User Authentication Functions ---
+# IMPORTANT: For production, this 'users.json' file will not persist across deployments
+# on cloud platforms like Streamlit Community Cloud. Consider using a database (e.g., Firebase, PostgreSQL)
+# for persistent user data.
 USERS_FILE = "users.json"
 
 def load_users():
+    """Loads user data from the JSON file."""
     if not os.path.exists(USERS_FILE):
         return {}
-    with open(USERS_FILE, "r") as f:
-        return json.load(f)
+    try:
+        with open(USERS_FILE, "r") as f:
+            return json.load(f)
+    except json.JSONDecodeError:
+        # Handle case where file might be empty or corrupted
+        st.error("Error loading user data. Initializing empty user database.")
+        return {}
 
 def save_users(users):
+    """Saves user data to the JSON file."""
     with open(USERS_FILE, "w") as f:
         json.dump(users, f, indent=2)
 
 def register_user(username, email, password, location):
+    """Registers a new user."""
     users = load_users()
     if email in users:
-        return False, "User already exists."
+        return False, "User already exists with this email."
     users[email] = {
         "username": username,
         "email": email,
-        "password": password,
+        "password": password, # In a real app, hash passwords!
         "location": location
     }
     save_users(users)
-    return True, "Registration successful."
+    return True, "Registration successful. You can now log in."
 
 def login_user(email, password):
+    """Logs in an existing user."""
     users = load_users()
     user = users.get(email)
-    if user and user["password"] == password:
+    if user and user["password"] == password: # In a real app, compare hashed passwords!
         return True, user
-    return False, "Invalid credentials."
+    return False, "Invalid email or password."
 
-# ------------------------------
-# Whisper Model
-# ------------------------------
-@st.cache_resource
+# --- Whisper Model for Audio Transcription ---
+@st.cache_resource # Caches the model to avoid reloading on rerun
 def load_whisper_model():
-    return whisper.load_model("base")
+    """Loads the Whisper base model."""
+    st.info("Loading Whisper model (this may take a moment)...")
+    model = whisper.load_model("base")
+    st.success("Whisper model loaded!")
+    return model
 
 def transcribe_audio(audio_bytes):
+    """Transcribes audio bytes to text using Whisper."""
+    # Save audio to a temporary file
     path = "/tmp/voice.wav"
-    with open(path, "wb") as f:
-        f.write(audio_bytes)
-    model = load_whisper_model()
-    result = model.transcribe(path, language="te")
-    return result["text"]
-
-# ------------------------------
-# Translation Model
-# ------------------------------
-@st.cache_resource
+    try:
+        with open(path, "wb") as f:
+            f.write(audio_bytes)
+        model = load_whisper_model()
+        # Specify language for better accuracy if known
+        result = model.transcribe(path, language="te", fp16=False) # fp16=False for CPU compatibility
+        return result["text"]
+    except Exception as e:
+        st.error(f"Error during audio transcription: {e}")
+        return ""
+    finally:
+        # Clean up the temporary file
+        if os.path.exists(path):
+            os.remove(path)
+
+# --- MarianMT Model for Translation ---
+@st.cache_resource # Caches the model to avoid reloading on rerun
 def load_translation_model():
+    """Loads the MarianMT English-Telugu translation model."""
+    st.info("Loading translation model (this may take a moment)...")
     model_name = "Helsinki-NLP/opus-mt-en-te"
     tokenizer = MarianTokenizer.from_pretrained(model_name)
     model = MarianMTModel.from_pretrained(model_name)
+    st.success("Translation model loaded!")
     return tokenizer, model
 
 def translate_to_telugu(text):
-    tokenizer, model = load_translation_model()
-    tokens = tokenizer.prepare_seq2seq_batch([text], return_tensors="pt", truncation=True, max_length=512)
-    translated = model.generate(**tokens)
-    return tokenizer.decode(translated[0], skip_special_tokens=True)
-
-# ------------------------------
-# Extract Article Summary
-# ------------------------------
+    """Translates English text to Telugu."""
+    try:
+        tokenizer, model = load_translation_model()
+        # Prepare input for the model, ensuring truncation for long texts
+        tokens = tokenizer.prepare_seq2seq_batch([text], return_tensors="pt", truncation=True, max_length=512)
+        translated = model.generate(**tokens)
+        return tokenizer.decode(translated[0], skip_special_tokens=True)
+    except Exception as e:
+        st.error(f"Error during translation: {e}")
+        return "Translation failed."
+
+# --- Article Content Extraction ---
 def extract_article_content(url):
+    """Extracts the first few paragraphs from an article URL."""
     try:
-        response = requests.get(url, timeout=5)
+        # Add headers to mimic a browser, as some sites block direct requests
+        headers = {
+            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
+        }
+        response = requests.get(url, timeout=10, headers=headers) # Increased timeout
+        response.raise_for_status() # Raise an HTTPError for bad responses (4xx or 5xx)
         soup = BeautifulSoup(response.text, "html.parser")
         paragraphs = soup.find_all('p')
+        # Join the text of the first 5 paragraphs, or fewer if not available
         content = ' '.join([p.get_text() for p in paragraphs[:5]])
-        return content
+        if not content: # If no paragraphs found, try to get text from body
+            content = soup.body.get_text(separator=' ', strip=True)[:1000] # Get first 1000 chars
+        return content if content else "Could not extract article content."
+    except requests.exceptions.RequestException as e:
+        return f"Network or HTTP error: {e}"
     except Exception as e:
-        return str(e)
+        return f"Error extracting content: {e}"
 
-# ------------------------------
-# Auth UI
-# ------------------------------
+# --- Authentication UI ---
+# Initialize session state for login status
 if "logged_in" not in st.session_state:
     st.session_state.logged_in = False
+if "user" not in st.session_state:
+    st.session_state.user = None
 
 if not st.session_state.logged_in:
     st.title("🔐 Login to BharatPulse")
@@ -102,100 +142,155 @@ if not st.session_state.logged_in:
     tab1, tab2 = st.tabs(["🔑 Login", "📝 Register"])
 
     with tab1:
-        email = st.text_input("Email")
-        password = st.text_input("Password", type="password")
-        if st.button("Login"):
+        st.subheader("Existing User Login")
+        email = st.text_input("Email", key="login_email")
+        password = st.text_input("Password", type="password", key="login_password")
+        if st.button("Login", key="login_button"):
             success, user_or_msg = login_user(email, password)
             if success:
                 st.session_state.logged_in = True
                 st.session_state.user = user_or_msg
-                st.rerun()
+                st.success(f"Welcome, {st.session_state.user['username']}!")
+                st.rerun() # Rerun to switch to the main app view
             else:
                 st.error(user_or_msg)
 
     with tab2:
-        new_username = st.text_input("Username")
-        new_email = st.text_input("New Email")
-        new_password = st.text_input("New Password", type="password")
-        new_location = st.text_input("Location")
-        if st.button("Register"):
-            success, message = register_user(new_username, new_email, new_password, new_location)
-            if success:
-                st.success(message)
+        st.subheader("New User Registration")
+        new_username = st.text_input("Username", key="register_username")
+        new_email = st.text_input("New Email", key="register_email")
+        new_password = st.text_input("New Password", type="password", key="register_password")
+        new_location = st.text_input("Location (e.g., Hyderabad)", key="register_location")
+        if st.button("Register", key="register_button"):
+            if new_username and new_email and new_password and new_location:
+                success, message = register_user(new_username, new_email, new_password, new_location)
+                if success:
+                    st.success(message)
+                else:
+                    st.error(message)
             else:
-                st.error(message)
-    st.stop()
+                st.warning("Please fill in all registration fields.")
+    st.stop() # Stop execution if not logged in
 
-# ------------------------------
-# Main News UI
-# ------------------------------
+# --- Main News UI (only runs if logged in) ---
+st.sidebar.header("User Info")
 st.sidebar.success(f"Logged in as: {st.session_state.user['username']}")
+st.sidebar.info(f"Your registered location: {st.session_state.user['location']}")
+
 if st.sidebar.button("🚪 Logout"):
     st.session_state.logged_in = False
     st.session_state.user = None
-    st.rerun()
+    st.rerun() # Rerun to go back to login screen
 
-st.title("📍 BharatPulse")
-tabs = st.tabs(["📰 Headlines", "🔍 Search by City", "🎤 Voice Search"])
+st.title("📍 BharatPulse - Local News & Sentiment")
 
+# Define RSS feeds
+# You can expand this dictionary with more RSS feeds for different regions/newspapers
 feeds = {
     "Sakshi - Medak": "https://www.sakshi.com/rss/district-news/medak.xml",
-    "Eenadu - AP": "https://www.eenadu.net/rss/andhra-pradesh.xml"
+    "Eenadu - AP": "https://www.eenadu.net/rss/andhra-pradesh.xml",
+    # Add more feeds here, e.g., "Eenadu - Telangana": "https://www.eenadu.net/rss/telangana.xml"
 }
 
-# ------------------------------
-# 📰 Headlines Tab
-# ------------------------------
+tabs = st.tabs(["📰 Headlines", "🔍 Search by City", "🎤 Voice Search"])
+
+# --- 📰 Headlines Tab ---
 with tabs[0]:
-    st.header("📰 Local Telugu News")
+    st.header("📰 Latest Local Telugu News Headlines")
     for name, url in feeds.items():
         st.subheader(name)
-        feed = feedparser.parse(url)
-        for entry in feed.entries[:3]:
-            st.markdown(f"### {entry.title}")
-            st.write(entry.published)
-            st.markdown(f"[Read Full Article]({entry.link})")
-            summary = extract_article_content(entry.link)
-            st.markdown("**English Summary:** " + summary)
-            telugu_summary = translate_to_telugu(summary)
-            st.markdown("**తెలుగు అనువాదం:** " + telugu_summary)
-            st.markdown("---")
-
-# ------------------------------
-# 🔍 Search by City
-# ------------------------------
+        try:
+            feed = feedparser.parse(url)
+            if not feed.entries:
+                st.info(f"No entries found for {name}.")
+                continue
+
+            for entry in feed.entries[:3]: # Display top 3 articles per feed
+                st.markdown(f"### [{entry.title}]({entry.link})") # Make title a clickable link
+                st.write(f"Published: {entry.published}")
+                
+                # Extract and summarize article content
+                summary = extract_article_content(entry.link)
+                st.markdown(f"**English Summary:** {summary}")
+                
+                # Translate summary to Telugu
+                telugu_summary = translate_to_telugu(summary)
+                st.markdown(f"**తెలుగు అనువాదం:** {telugu_summary}")
+                st.markdown("---")
+        except Exception as e:
+            st.error(f"Could not fetch news from {name}: {e}")
+
+# --- 🔍 Search by City Tab ---
 with tabs[1]:
-    st.header("🔎 Search by City / District")
-    city = st.text_input("Enter your city/district name in Telugu or English")
+    st.header("🔎 Search News by City / District")
+    # Pre-fill with user's registered location if available
+    default_city = st.session_state.user['location'] if st.session_state.user else ""
+    city_search_input = st.text_input("Enter your city/district name (e.g., Medak, Hyderabad)", value=default_city)
 
-    if city:
-        st.success(f"Showing news for: {city}")
+    if city_search_input:
+        st.success(f"Searching news for: **{city_search_input}**")
+        found_news = False
         for name, url in feeds.items():
-            if city.lower() in name.lower():
-                st.subheader(name)
-                feed = feedparser.parse(url)
-                for entry in feed.entries[:2]:
-                    st.markdown(f"**🗞️ {entry.title}**")
-                    summary = extract_article_content(entry.link)
-                    telugu_summary = translate_to_telugu(summary)
-                    st.markdown("**తెలుగు అనువాదం:** " + telugu_summary)
-
-# ------------------------------
-# 🎤 Voice Search Tab
-# ------------------------------
+            # Basic matching: checks if the city name is in the feed's name (case-insensitive)
+            if city_search_input.lower() in name.lower():
+                st.subheader(f"News from {name}")
+                try:
+                    feed = feedparser.parse(url)
+                    if not feed.entries:
+                        st.info(f"No entries found for {name} matching '{city_search_input}'.")
+                        continue
+
+                    for entry in feed.entries[:2]: # Display top 2 articles per matching feed
+                        st.markdown(f"**🗞️ [{entry.title}]({entry.link})**")
+                        summary = extract_article_content(entry.link)
+                        telugu_summary = translate_to_telugu(summary)
+                        st.markdown(f"**తెలుగు అనువాదం:** {telugu_summary}")
+                        st.markdown("---")
+                        found_news = True
+                except Exception as e:
+                    st.error(f"Could not fetch news from {name}: {e}")
+        if not found_news:
+            st.info(f"No news found for '{city_search_input}' in the available feeds. Try a different city or check our 'Headlines' tab.")
+
+# --- 🎤 Voice Search Tab ---
 with tabs[2]:
     st.header("🎤 Search by Telugu Voice")
-    audio = st_audiorec()
-    if audio:
-        transcribed = transcribe_audio(audio)
-        st.success(f"Voice Input Recognized: {transcribed}")
-        city = transcribed
-        for name, url in feeds.items():
-            if city.lower() in name.lower():
-                st.subheader(name)
-                feed = feedparser.parse(url)
-                for entry in feed.entries[:2]:
-                    st.markdown(f"**🗞️ {entry.title}**")
-                    summary = extract_article_content(entry.link)
-                    telugu_summary = translate_to_telugu(summary)
-                    st.markdown("**తెలుగు అనువాదం:** " + telugu_summary)
+    st.info("Click 'Record' to start, 'Stop' when done. Speak the city/district name in Telugu.")
+    
+    # Use the audio recorder component
+    audio_bytes = st_audiorec()
+
+    if audio_bytes:
+        with st.spinner("Transcribing audio..."):
+            transcribed_text = transcribe_audio(audio_bytes)
+        
+        if transcribed_text:
+            st.success(f"Voice Input Recognized: **{transcribed_text}**")
+            city_from_voice = transcribed_text.strip() # Use the transcribed text as the city
+            
+            st.subheader(f"Searching news for: **{city_from_voice}**")
+            found_news_voice = False
+            for name, url in feeds.items():
+                # Basic matching: checks if the transcribed city name is in the feed's name
+                if city_from_voice.lower() in name.lower():
+                    st.subheader(f"News from {name}")
+                    try:
+                        feed = feedparser.parse(url)
+                        if not feed.entries:
+                            st.info(f"No entries found for {name} matching '{city_from_voice}'.")
+                            continue
+
+                        for entry in feed.entries[:2]: # Display top 2 articles per matching feed
+                            st.markdown(f"**🗞️ [{entry.title}]({entry.link})**")
+                            summary = extract_article_content(entry.link)
+                            telugu_summary = translate_to_telugu(summary)
+                            st.markdown(f"**తెలుగు అనువాదం:** {telugu_summary}")
+                            st.markdown("---")
+                            found_news_voice = True
+                    except Exception as e:
+                        st.error(f"Could not fetch news from {name}: {e}")
+            if not found_news_voice:
+                st.info(f"No news found for '{city_from_voice}' in the available feeds. Try speaking clearly or check our 'Headlines' tab.")
+        else:
+            st.warning("Could not transcribe audio. Please try again.")
+