Update app.py

app.py

@@ -9,93 +9,91 @@ from langchain.chains import RetrievalQA
 
 from pinecone import Pinecone, ServerlessSpec
 
-# Set a consistent index name
+# Constants
 INDEX_NAME = "rag-demo-index"
+DIMENSION = 768  # Use 768 for Gemini embeddings
 
-def process_rag(api_key_gemini, api_key_pinecone, pinecone_region, pdf_file, user_question):
-    if not api_key_gemini or not api_key_pinecone or not pinecone_region:
-        return "❌ Please provide all required API keys and region."
+def process_rag(pdf_file, user_question):
+    # 🔐 Load from Hugging Face Secrets
+    pinecone_api_key = os.environ.get("PINECONE_API_KEY")
+    pinecone_env = os.environ.get("PINECONE_ENVIRONMENT")
+    google_api_key = os.environ.get("GOOGLE_API_KEY")
+
+    if not all([pinecone_api_key, pinecone_env, google_api_key]):
+        return "❌ Missing required secrets. Check PINECONE_API_KEY, PINECONE_ENVIRONMENT, or GOOGLE_API_KEY."
 
     if not pdf_file:
-        return "❌ Please upload a PDF file."
+        return "❌ Please upload a PDF file first."
 
     try:
-        # Step 1: Load and split PDF
+        # Step 1: Load PDF and chunk it
         loader = PyPDFLoader(pdf_file.name)
         documents = loader.load()
 
-        splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
-        docs = splitter.split_documents(documents)
+        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
+        docs = text_splitter.split_documents(documents)
 
-        # Step 2: Gemini embedding model
+        # Step 2: Embeddings via Gemini
         embeddings = GoogleGenerativeAIEmbeddings(
             model="models/embedding-001",
-            google_api_key=api_key_gemini
+            google_api_key=google_api_key
         )
 
-        # Step 3: Init Pinecone client
-        pc = Pinecone(api_key=api_key_pinecone)
+        # Step 3: Connect to Pinecone v3
+        pc = Pinecone(api_key=pinecone_api_key)
 
-        # Step 4: Create index if it doesn't exist
         if INDEX_NAME not in pc.list_indexes().names():
             pc.create_index(
                 name=INDEX_NAME,
-                dimension=768,  # for Gemini embedding
+                dimension=DIMENSION,
                 metric="cosine",
-                spec=ServerlessSpec(cloud="aws", region=pinecone_region)
+                spec=ServerlessSpec(
+                    cloud="aws",
+                    region=pinecone_env
+                )
             )
 
-        # Step 5: Store vectors using LangChain wrapper (not using pc.Index)
-        vectordb = LangChainPinecone.from_documents(
-            docs,
+        # Step 4: Store docs in Pinecone
+        vectorstore = LangChainPinecone.from_documents(
+            documents=docs,
             embedding=embeddings,
-            index_name=INDEX_NAME,
-            api_key=api_key_pinecone,
-            environment=pinecone_region
+            index_name=INDEX_NAME
         )
 
-        # Step 6: Create retriever
-        retriever = vectordb.as_retriever()
+        retriever = vectorstore.as_retriever()
 
-        # Step 7: Load Gemini model for answering
+        # Step 5: Gemini chat model
         llm = ChatGoogleGenerativeAI(
             model="gemini-pro",
-            google_api_key=api_key_gemini,
+            google_api_key=google_api_key,
             temperature=0
         )
 
+        # Step 6: RAG chain
         qa_chain = RetrievalQA.from_chain_type(
             llm=llm,
             retriever=retriever,
             return_source_documents=False
         )
 
-        # Step 8: Ask the question
         result = qa_chain({"query": user_question})
         return result["result"]
 
     except Exception as e:
         return f"❌ Error: {str(e)}"
 
-# Gradio Interface
-with gr.Blocks() as app:
-    gr.Markdown("## 📄🔍 Ask Questions About Your PDF (Gemini + Pinecone RAG)")
+# Gradio UI
+with gr.Blocks() as demo:
+    gr.Markdown("## 🔍 Ask Questions from PDF using Gemini + Pinecone RAG")
 
     with gr.Row():
-        gemini_key = gr.Textbox(label="🔐 Gemini API Key", type="password", placeholder="Paste your Gemini API key")
-        pinecone_key = gr.Textbox(label="🌲 Pinecone API Key", type="password", placeholder="Paste your Pinecone API key")
-        pinecone_region = gr.Textbox(label="📍 Pinecone Region", placeholder="e.g., us-east-1")
-
-    pdf_file = gr.File(label="📄 Upload your PDF", file_types=[".pdf"])
-    user_question = gr.Textbox(label="❓ Ask a Question")
-    answer_output = gr.Textbox(label="🤖 Gemini Answer", lines=10)
+        pdf_input = gr.File(label="📄 Upload PDF", file_types=[".pdf"])
+        question_input = gr.Textbox(label="❓ Ask your question")
+    
+    answer_output = gr.Textbox(label="🤖 Gemini Answer", lines=8)
 
-    ask_button = gr.Button("🔍 Ask")
+    ask_button = gr.Button("🔎 Run RAG")
 
-    ask_button.click(
-        fn=process_rag,
-        inputs=[gemini_key, pinecone_key, pinecone_region, pdf_file, user_question],
-        outputs=answer_output
-    )
+    ask_button.click(fn=process_rag, inputs=[pdf_input, question_input], outputs=answer_output)
 
-app.launch()
+demo.launch()