Update app.py

app.py

@@ -1,67 +1,101 @@
-import os
 import streamlit as st
-from langchain.vectorstores import FAISS
+from langchain_community.vectorstores import FAISS
 from langchain.embeddings import HuggingFaceEmbeddings
+from langchain_community.document_loaders import PyMuPDFLoader, Docx2txtLoader, UnstructuredFileLoader
+from langchain.text_splitter import CharacterTextSplitter
+from langchain_community.llms import ChatGroq
 from langchain.chains import RetrievalQA
-from langchain_groq import ChatGroq
 from gtts import gTTS
+from langdetect import detect
+import os
+import shutil
+import pickle
 import tempfile
 
-# Set your Streamlit page config
-st.set_page_config(page_title="Learning with Fun", page_icon="📚")
+# Title
+st.set_page_config(page_title="Learning with Fun 👦📚", layout="centered")
+st.title("🎓 Learning with Fun – Grade 5 & 6 📘")
+st.markdown("Ask your questions in Urdu or English. Get simple and storytelling-style answers!")
+
+# Temp directory for file upload
+temp_dir = tempfile.mkdtemp()
+
+# File upload
+uploaded_file = st.file_uploader("📄 Upload a textbook file (PDF, DOCX, or TXT)", type=["pdf", "docx", "txt"])
+
+# Load documents
+def load_document(file_path):
+    if file_path.endswith(".pdf"):
+        loader = PyMuPDFLoader(file_path)
+    elif file_path.endswith(".docx"):
+        loader = Docx2txtLoader(file_path)
+    else:
+        loader = UnstructuredFileLoader(file_path)
+    return loader.load()
 
-# Title and description
-st.title("📚 Learning with Fun")
-st.markdown("**بچوں کے لیے سوال و جواب اردو اور انگلش میں**")
+# Vector DB functions
+def create_vectorstore(docs):
+    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=100)
+    texts = text_splitter.split_documents(docs)
+    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+    db = FAISS.from_documents(texts, embeddings)
+    return db
 
-# Load GROQ API key from environment
-GROQ_API_KEY = os.getenv("GROQ_API_KEY")
-if not GROQ_API_KEY:
-    st.error("🚨 GROQ_API_KEY is missing! Please set it in Hugging Face secrets.")
-    st.stop()
+# Load or create vectorstore
+def get_vectorstore(file):
+    file_path = os.path.join(temp_dir, file.name)
+    with open(file_path, "wb") as f:
+        f.write(file.getbuffer())
 
-# Load FAISS index
-try:
-    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
-    vectorstore = FAISS.load_local("faiss_index", embeddings, allow_dangerous_deserialization=True)
-except Exception as e:
-    st.error(f"❌ Failed to load FAISS index: {e}")
-    st.stop()
+    pickle_path = os.path.join(temp_dir, f"{file.name}.pkl")
 
-# Initialize LLaMA3 model via GROQ
-llm = ChatGroq(
-    groq_api_key=GROQ_API_KEY,
-    model_name="llama3-8b-8192"
-)
+    if os.path.exists(pickle_path):
+        with open(pickle_path, "rb") as f:
+            db = pickle.load(f)
+    else:
+        docs = load_document(file_path)
+        db = create_vectorstore(docs)
+        with open(pickle_path, "wb") as f:
+            pickle.dump(db, f)
+    return db
 
-# Create retrieval chain
-qa_chain = RetrievalQA.from_chain_type(
-    llm=llm,
-    retriever=vectorstore.as_retriever()
-)
+# Generate text to speech
+def generate_tts(text, lang):
+    tts = gTTS(text, lang=lang)
+    output_path = os.path.join(temp_dir, "output.mp3")
+    tts.save(output_path)
+    return output_path
 
-# Input box
-query = st.text_input("🧠 اپنا سوال درج کریں (اردو یا انگریزی میں):")
+# Run QA chain
+def run_chain(db, query, response_type):
+    llm = ChatGroq(temperature=0.3, model_name="llama3-8b-8192")
+    qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=db.as_retriever())
+    
+    if response_type == "Explain Simply":
+        prompt = f"Explain this to a Grade 5 kid in simple Urdu:\n{query}"
+    else:
+        prompt = f"Tell a short fun story in Urdu to explain:\n{query}"
+    
+    result = qa_chain.run(prompt)
+    return result
 
-if query:
-    with st.spinner("جواب تیار ہو رہا ہے..."):
-        try:
-            # Get simple answer
-            answer = qa_chain.run(query)
-            
-            # Display basic explanation
-            st.markdown("### 📖 آسان جواب:")
-            st.write(answer)
+# UI Controls
+if uploaded_file:
+    query = st.text_input("❓ Ask a question:")
+    response_type = st.radio("Select answer format:", ["Explain Simply", "Storytelling"])
+    
+    if query:
+        with st.spinner("🤖 Thinking..."):
+            db = get_vectorstore(uploaded_file)
+            answer = run_chain(db, query, response_type)
+            st.markdown("### 🧠 Answer:")
+            st.success(answer)
 
-            # Storytelling version
-            story = f"ایک دن ایک بچے نے پوچھا: '{query}'۔ استاد نے مسکرا کر کہا: {answer}"
-            st.markdown("### 🧚 کہانی کی صورت میں:")
-            st.write(story)
+            # Detect language and speak
+            lang = "ur" if detect(answer) == "ur" else "en"
+            audio_path = generate_tts(answer, lang)
+            st.audio(audio_path, format="audio/mp3")
 
-            # Convert story to Urdu voice
-            tts = gTTS(text=story, lang='ur')
-            with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmpfile:
-                tts.save(tmpfile.name)
-                st.audio(tmpfile.name, format="audio/mp3")
-        except Exception as e:
-            st.error(f"❌ Error generating answer: {e}")
+# Cleanup
+st.markdown("---")
+st.caption("Made with 💙 using Streamlit + LLaMA 3 + Hugging Face")