Update app.py

app.py

@@ -1,5 +1,6 @@
 import os
 import gradio as gr
+import json
 from llama_index.core import SimpleDirectoryReader, VectorStoreIndex
 from langchain_community.embeddings import HuggingFaceEmbeddings
 from llama_index.llms.llama_cpp import LlamaCPP
@@ -10,17 +11,11 @@ from llama_index.llms.llama_cpp.llama_utils import (
 
 model_url = 'https://huggingface.co/bartowski/Llama-3.2-3B-Instruct-GGUF/resolve/main/Llama-3.2-3B-Instruct-Q4_K_M.gguf'
 llm = LlamaCPP(
-    # You can pass in the URL to a GGML model to download it automatically
     model_url=model_url,
     temperature=0.1,
     max_new_tokens=256,
     context_window=2048,
-    # kwargs to pass to __call__()
-    generate_kwargs={},
-    # kwargs to pass to __init__()
-    # set to at least 1 to use GPU
     model_kwargs={"n_gpu_layers": 1},
-    # transform inputs into Llama2 format
     messages_to_prompt=messages_to_prompt,
     completion_to_prompt=completion_to_prompt,
     verbose=True,
@@ -29,13 +24,20 @@ llm = LlamaCPP(
 embeddings = HuggingFaceEmbeddings(model_name="BAAI/bge-small-en-v1.5")
 
 def initialize_index():
-    """Initialize the vector store index from PDF files in the data directory"""
-    # Load documents from the data directory
-    loader = SimpleDirectoryReader(
-        input_dir="data",
-        required_exts=[".pdf"]
-    )
-    documents = loader.load_data()
+    """Initialize the vector store index from JSON data."""
+    # Load JSON data
+    with open("dummy_users_with_tasks.json", "r") as file:
+        json_data = json.load(file)
+    
+    # Convert JSON data to plain text for embedding
+    documents = []
+    for user in json_data:
+        tasks_summary = "\n".join(
+            [f"Task: {task['title']} | Start: {task['start_date']} | Due: {task['due_date']}" 
+             for task in user["tasks"]]
+        )
+        doc_text = f"User: {user['name']} | Email: {user['email']}\nTasks:\n{tasks_summary}"
+        documents.append(doc_text)
     
     # Create index
     index = VectorStoreIndex.from_documents(
@@ -53,25 +55,22 @@ def process_query(
     message: str,
     history: list[tuple[str, str]],
 ) -> str:
-    """Process a query using the RAG system"""
+    """Process a query using the RAG system."""
     try:
         # Get response from the query engine
         response = query_engine.query(
             message,
-            #streaming=True
         )
-        return str(response)
+        return response
     except Exception as e:
         return f"Error processing query: {str(e)}"
 
-# Create the Gradio interface
-demo = gr.ChatInterface(
-    process_query,
-    title="PDF Question Answering with RAG + Llama",
-    description="Ask questions about the content of the loaded PDF documents using Llama model",
-    #undo_btn="Delete Previous",
-    #clear_btn="Clear",
+# Gradio interface (if needed)
+interface = gr.Interface(
+    fn=process_query,
+    inputs=["text", "state"],
+    outputs="text",
 )
 
 if __name__ == "__main__":
-    demo.launch(debug=True)
+    interface.launch()