Update app.py

app.py

@@ -7,8 +7,8 @@ import os
 import pypdf
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 
-# Set Groq API key
-GROQ_API_KEY = os.getenv("GROQ_API_KEY", "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")
+# Set Groq API key (replace with your actual key or use HF secrets)
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")  # Use HF secrets tab in Space settings
 
 # Load embedding model
 embedding_model = SentenceTransformer("all-MiniLM-L6-v2")
@@ -16,86 +16,65 @@ embedding_model = SentenceTransformer("all-MiniLM-L6-v2")
 # Set up Groq client
 client = Groq(api_key=GROQ_API_KEY)
 
-# Function to extract text from PDF
+# Extract text from PDF
 def extract_text_from_pdf(uploaded_file):
     reader = pypdf.PdfReader(uploaded_file)
     text = "\n".join([page.extract_text() for page in reader.pages if page.extract_text()])
     return text
 
-# Function to create text chunks
+# Split text into chunks
 def create_chunks(text, chunk_size=500, chunk_overlap=100):
     text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
     return text_splitter.split_text(text)
 
-# Function to create and save FAISS index
+# Create FAISS index
 def create_faiss_index(chunks):
     embeddings = embedding_model.encode(chunks, convert_to_numpy=True)
-
-    # Create FAISS index
     dimension = embeddings.shape[1]
     index = faiss.IndexFlatL2(dimension)
     index.add(embeddings)
-
-    # Save FAISS index and chunks
-    faiss.write_index(index, "faiss_index.bin")
-    np.save("chunks.npy", chunks)
-
     return index, chunks
 
-# Function to search FAISS
+# Search FAISS index
 def search_faiss(query, index, chunks, top_k=2):
     if index is None:
         return ["No database found. Please upload a PDF first."]
-
     query_embedding = embedding_model.encode([query], convert_to_numpy=True)
     distances, indices = index.search(query_embedding, top_k)
     return [chunks[i] for i in indices[0]]
 
-# Function to query Groq
+# Query Groq
 def query_groq(query):
     chat_completion = client.chat.completions.create(
         messages=[{"role": "user", "content": query}],
-        model="llama-3.3-70b-versatile",
+        model="llama-3-70b-8192",  # or llama-3-8b-8192, depending on your access
     )
     return chat_completion.choices[0].message.content
 
 # Streamlit UI
 st.set_page_config(page_title="RAG Chatbot", page_icon="🤖")
-st.title("📄 RAG-Based Chatbot with FAISS & Groq")
+st.title("📄 RAG Chatbot with FAISS & Groq")
 
-# Upload PDF
-uploaded_file = st.file_uploader("📤 Upload a PDF file", type="pdf")
+uploaded_file = st.file_uploader("📤 Upload a PDF", type="pdf")
 
 if uploaded_file:
     with st.spinner("🔄 Processing PDF..."):
         text = extract_text_from_pdf(uploaded_file)
         chunks = create_chunks(text)
-
-        # Create FAISS index
         index, chunks = create_faiss_index(chunks)
-
-        # Store in session state
         st.session_state["faiss_index"] = index
         st.session_state["chunks"] = chunks
+        st.success("✅ PDF processed!")
 
-        st.success("✅ PDF processed successfully!")
+index = st.session_state.get("faiss_index", None)
+chunks = st.session_state.get("chunks", [])
 
-# Load FAISS index if available
-if "faiss_index" in st.session_state and "chunks" in st.session_state:
-    index = st.session_state["faiss_index"]
-    chunks = st.session_state["chunks"]
-else:
-    index = None
-    chunks = []
-
-# User query input
-user_query = st.text_input("💬 Ask me something about the document:")
+user_query = st.text_input("💬 Ask something about the document:")
 
 if st.button("Search") and user_query:
-    with st.spinner("🔎 Retrieving response..."):
-        retrieved_text = search_faiss(user_query, index, chunks)
-        response = query_groq("\n".join(retrieved_text))
-
+    with st.spinner("🔎 Generating response..."):
+        retrieved = search_faiss(user_query, index, chunks)
+        context = "\n".join(retrieved)
+        response = query_groq(context + "\n\nAnswer the question: " + user_query)
         st.write("### 🤖 AI Response:")
         st.write(response)
-