Update app.py

app.py

@@ -4,57 +4,48 @@ import gradio as gr
 from langchain_community.document_loaders import PyPDFLoader
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_google_genai import GoogleGenerativeAIEmbeddings, ChatGoogleGenerativeAI
-from langchain_community.vectorstores import Pinecone as LangChainPinecone
+from langchain_community.vectorstores import Pinecone
 from langchain.chains import RetrievalQA
 
-from pinecone import Pinecone, ServerlessSpec
+import pinecone
 
 # Constants
 INDEX_NAME = "rag-demo-index"
-DIMENSION = 768  # Use 768 for Gemini embeddings
 
 def process_rag(pdf_file, user_question):
-    # 🔐 Load from Hugging Face Secrets
+    # Load secrets from Hugging Face (or env manually)
     pinecone_api_key = os.environ.get("PINECONE_API_KEY")
-    pinecone_env = os.environ.get("PINECONE_ENVIRONMENT")
+    pinecone_env = os.environ.get("PINECONE_ENVIRONMENT")  # Example: "gcp-starter"
     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."
+        return "❌ Missing API key(s). Please check Pinecone & Google Gemini keys."
 
     if not pdf_file:
-        return "❌ Please upload a PDF file first."
+        return "❌ Please upload a PDF file."
 
     try:
-        # Step 1: Load PDF and chunk it
+        # 1. Load and split PDF
         loader = PyPDFLoader(pdf_file.name)
         documents = loader.load()
 
         text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
         docs = text_splitter.split_documents(documents)
 
-        # Step 2: Embeddings via Gemini
+        # 2. Gemini Embeddings
         embeddings = GoogleGenerativeAIEmbeddings(
             model="models/embedding-001",
             google_api_key=google_api_key
         )
 
-        # Step 3: Connect to Pinecone v3
-        pc = Pinecone(api_key=pinecone_api_key)
-
-        if INDEX_NAME not in pc.list_indexes().names():
-            pc.create_index(
-                name=INDEX_NAME,
-                dimension=DIMENSION,
-                metric="cosine",
-                spec=ServerlessSpec(
-                    cloud="aws",
-                    region=pinecone_env
-                )
-            )
-
-        # Step 4: Store docs in Pinecone
-        vectorstore = LangChainPinecone.from_documents(
+        # 3. Init Pinecone (old client)
+        pinecone.init(api_key=pinecone_api_key, environment=pinecone_env)
+
+        if INDEX_NAME not in pinecone.list_indexes():
+            pinecone.create_index(name=INDEX_NAME, dimension=768, metric="cosine")
+
+        # 4. Create LangChain-compatible Vector DB
+        vectorstore = Pinecone.from_documents(
             documents=docs,
             embedding=embeddings,
             index_name=INDEX_NAME
@@ -62,20 +53,21 @@ def process_rag(pdf_file, user_question):
 
         retriever = vectorstore.as_retriever()
 
-        # Step 5: Gemini chat model
+        # 5. Gemini LLM
         llm = ChatGoogleGenerativeAI(
             model="gemini-pro",
             google_api_key=google_api_key,
             temperature=0
         )
 
-        # Step 6: RAG chain
+        # 6. Retrieval QA chain
         qa_chain = RetrievalQA.from_chain_type(
             llm=llm,
             retriever=retriever,
             return_source_documents=False
         )
 
+        # 7. Ask question
         result = qa_chain({"query": user_question})
         return result["result"]
 
@@ -84,7 +76,7 @@ def process_rag(pdf_file, user_question):
 
 # Gradio UI
 with gr.Blocks() as demo:
-    gr.Markdown("## 🔍 Ask Questions from PDF using Gemini + Pinecone RAG")
+    gr.Markdown("## 🔍 Ask Questions from PDF using Gemini + Pinecone (LangChain RAG)")
 
     with gr.Row():
         pdf_input = gr.File(label="📄 Upload PDF", file_types=[".pdf"])