Update app.py

app.py

@@ -1,24 +1,28 @@
 import os
 import re
 import random
+import subprocess
 import requests
 import streamlit as st
+import spacy  # for additional NLP processing
 from langchain_huggingface import HuggingFaceEndpoint
 from langchain_core.prompts import PromptTemplate
 from langchain_core.output_parsers import StrOutputParser
 from transformers import pipeline
 
-# Use environment variables for keys
-HF_TOKEN = os.getenv("HF_TOKEN")
-if HF_TOKEN is None:
-    raise ValueError("HF_TOKEN environment variable not set. Please set it in your Hugging Face Space settings.")
+# Must be the first Streamlit command!
+st.set_page_config(page_title="HAL - NASA ChatBot", page_icon="🚀")
 
-NASA_API_KEY = os.getenv("NASA_API_KEY")
-if NASA_API_KEY is None:
-    raise ValueError("NASA_API_KEY environment variable not set. Please set it in your Hugging Face Space settings.")
+# --- Helper to load spaCy model with fallback ---
+def load_spacy_model():
+    try:
+        return spacy.load("en_core_web_sm")
+    except OSError:
+        st.warning("Downloading spaCy model en_core_web_sm... This may take a moment.")
+        subprocess.run(["python", "-m", "spacy", "download", "en_core_web_sm"], check=True)
+        return spacy.load("en_core_web_sm")
 
-# Set up Streamlit UI
-st.set_page_config(page_title="HAL - NASA ChatBot", page_icon="🚀")
+nlp_spacy = load_spacy_model()
 
 # --- Initialize Session State Variables ---
 if "chat_history" not in st.session_state:
@@ -28,6 +32,38 @@ if "response_ready" not in st.session_state:
 if "follow_up" not in st.session_state:
     st.session_state.follow_up = ""
 
+# --- Appearance CSS ---
+st.markdown("""
+    <style>
+    .user-msg {
+        background-color: #696969;
+        color: white;
+        padding: 10px;
+        border-radius: 10px;
+        margin-bottom: 5px;
+        width: fit-content;
+        max-width: 80%;
+    }
+    .assistant-msg {
+        background-color: #333333;
+        color: white;
+        padding: 10px;
+        border-radius: 10px;
+        margin-bottom: 5px;
+        width: fit-content;
+        max-width: 80%;
+    }
+    .container {
+        display: flex;
+        flex-direction: column;
+        align-items: flex-start;
+    }
+    @media (max-width: 600px) {
+        .user-msg, .assistant-msg { font-size: 16px; max-width: 100%; }
+    }
+    </style>
+""", unsafe_allow_html=True)
+
 # --- Set Up Model & API Functions ---
 model_id = "mistralai/Mistral-7B-Instruct-v0.3"
 sentiment_analyzer = pipeline(
@@ -41,12 +77,12 @@ def get_llm_hf_inference(model_id=model_id, max_new_tokens=128, temperature=0.7)
         repo_id=model_id,
         max_new_tokens=max_new_tokens,
         temperature=temperature,
-        token=HF_TOKEN,
+        token=os.getenv("HF_TOKEN"),
         task="text-generation"
     )
 
 def get_nasa_apod():
-    url = f"https://api.nasa.gov/planetary/apod?api_key={NASA_API_KEY}"
+    url = f"https://api.nasa.gov/planetary/apod?api_key={os.getenv('NASA_API_KEY')}"
     response = requests.get(url)
     if response.status_code == 200:
         data = response.json()
@@ -59,19 +95,24 @@ def analyze_sentiment(user_text):
     return result['label']
 
 def predict_action(user_text):
-    if "NASA" in user_text or "space" in user_text:
+    if "nasa" in user_text.lower() or "space" in user_text.lower():
         return "nasa_info"
     return "general_query"
 
+def extract_context(text):
+    """Extract key entities using spaCy."""
+    doc = nlp_spacy(text)
+    entities = [ent.text for ent in doc.ents]
+    return ", ".join(entities) if entities else ""
+
 def generate_follow_up(user_text):
     """
     Generates two variant follow-up questions and randomly selects one.
-    It also cleans up any unwanted quotation marks or extra meta commentary.
     """
     prompt_text = (
-        f"Based on the user's question: '{user_text}', generate two concise, friendly follow-up questions "
-        "that invite further discussion. For example, one might be 'Would you like to know more about the six types of quarks?' "
-        "and another might be 'Would you like to explore another aspect of quantum physics?' Do not include extra commentary."
+        f"Based on the user's question: '{user_text}', generate two concise, friendly follow-up questions that invite further discussion. "
+        "For example, one might be 'Would you like to know more about the six types of quarks?' and another 'Would you like to explore another aspect of quantum physics?'. "
+        "Answer exclusively in English, and do not include extra commentary."
     )
     hf = get_llm_hf_inference(max_new_tokens=80, temperature=0.9)
     output = hf.invoke(input=prompt_text).strip()
@@ -83,14 +124,17 @@ def generate_follow_up(user_text):
 
 def get_response(system_message, chat_history, user_text, max_new_tokens=1024):
     """
-    Generates HAL's answer with depth and a follow-up question.
-    The prompt instructs the model to provide a detailed explanation and then generate a follow-up.
-    If the answer comes back empty, a fallback answer is used.
+    Generates HAL's detailed, in-depth response and a follow-up question.
+    It incorporates sentiment analysis, additional NLP context, and style instructions.
     """
     sentiment = analyze_sentiment(user_text)
     action = predict_action(user_text)
     
-    # Extract style instruction if present
+    # Extract extra context (e.g., named entities)
+    context_info = extract_context(user_text)
+    context_clause = f" The key topics here are: {context_info}." if context_info else ""
+    
+    # Extract style instruction if provided.
     style_instruction = ""
     lower_text = user_text.lower()
     if "in the voice of" in lower_text or "speaking as" in lower_text:
@@ -99,6 +143,9 @@ def get_response(system_message, chat_history, user_text, max_new_tokens=1024):
             style_instruction = match.group(2).strip().capitalize()
             style_instruction = f" Please respond in the voice of {style_instruction}."
     
+    # Force output in English.
+    language_clause = " Answer exclusively in English."
+    
     if action == "nasa_info":
         nasa_url, nasa_title, nasa_explanation = get_nasa_apod()
         response = f"**{nasa_title}**\n\n{nasa_explanation}"
@@ -117,23 +164,25 @@ def get_response(system_message, chat_history, user_text, max_new_tokens=1024):
     
     style_clause = style_instruction if style_instruction else ""
     
-    # Instruct the model to generate a detailed, in-depth answer.
     prompt = PromptTemplate.from_template(
         (
             "[INST] {system_message}\n\nCurrent Conversation:\n{chat_history}\n\n"
             "User: {user_text}.\n [/INST]\n"
-            "AI: Please provide a detailed explanation in depth. "
-            "Ensure your response covers the topic thoroughly and is written in a friendly, conversational style, "
-            "starting with a phrase like 'Certainly!', 'Of course!', or 'Great question!'." + style_clause +
+            "AI: Please provide a detailed, in-depth answer in a friendly, conversational tone that thoroughly covers the topic."
+            + style_clause + context_clause + language_clause +
             "\nHAL:"
         )
     )
     
+    st.write("DEBUG: Prompt sent to model:")
+    st.write(prompt.format(system_message=system_message, chat_history=filtered_history, user_text=user_text))
+    
     chat = prompt | hf.bind(skip_prompt=True) | StrOutputParser(output_key='content')
-    response = chat.invoke(input=dict(system_message=system_message, user_text=user_text, chat_history=filtered_history))
-    # Remove any extra markers if present.
+    raw_output = chat.invoke(input=dict(system_message=system_message, user_text=user_text, chat_history=filtered_history))
+    st.write("DEBUG: Raw model output:")
+    st.write(raw_output)
     
-    # Fallback in case the generated answer is empty
+    response = raw_output  # Using the full raw output without splitting
     if not response:
         response = "Certainly, here is an in-depth explanation: [Fallback explanation]."
     
@@ -147,6 +196,8 @@ def get_response(system_message, chat_history, user_text, max_new_tokens=1024):
     follow_up = generate_follow_up(user_text)
     chat_history.append({'role': 'assistant', 'content': follow_up})
     
+    st.write("DEBUG: Generated follow-up question:", follow_up)
+    
     return response, follow_up, chat_history, None
 
 # --- Chat UI ---
@@ -159,36 +210,13 @@ if st.sidebar.button("Reset Chat"):
     st.session_state.follow_up = ""
     st.experimental_rerun()
 
-st.markdown("""
-    <style>
-    .user-msg {
-        background-color: #696969;
-        color: white;
-        padding: 10px;
-        border-radius: 10px;
-        margin-bottom: 5px;
-        width: fit-content;
-        max-width: 80%;
-    }
-    .assistant-msg {
-        background-color: #333333;
-        color: white;
-        padding: 10px;
-        border-radius: 10px;
-        margin-bottom: 5px;
-        width: fit-content;
-        max-width: 80%;
-    }
-    .container {
-        display: flex;
-        flex-direction: column;
-        align-items: flex-start;
-    }
-    @media (max-width: 600px) {
-        .user-msg, .assistant-msg { font-size: 16px; max-width: 100%; }
-    }
-    </style>
-""", unsafe_allow_html=True)
+st.markdown("<div class='container'>", unsafe_allow_html=True)
+for message in st.session_state.chat_history:
+    if message["role"] == "user":
+        st.markdown(f"<div class='user-msg'><strong>You:</strong> {message['content']}</div>", unsafe_allow_html=True)
+    else:
+        st.markdown(f"<div class='assistant-msg'><strong>HAL:</strong> {message['content']}</div>", unsafe_allow_html=True)
+st.markdown("</div>", unsafe_allow_html=True)
 
 user_input = st.chat_input("Type your message here...")
 
@@ -202,11 +230,3 @@ if user_input:
         st.image(image_url, caption="NASA Image of the Day")
     st.session_state.follow_up = follow_up
     st.session_state.response_ready = True
-
-st.markdown("<div class='container'>", unsafe_allow_html=True)
-for message in st.session_state.chat_history:
-    if message["role"] == "user":
-        st.markdown(f"<div class='user-msg'><strong>You:</strong> {message['content']}</div>", unsafe_allow_html=True)
-    else:
-        st.markdown(f"<div class='assistant-msg'><strong>HAL:</strong> {message['content']}</div>", unsafe_allow_html=True)
-st.markdown("</div>", unsafe_allow_html=True)