app.py

import gradio as gr
import requests
import json
from decouple import Config

# Function to interact with Vectara API
def query_vectara(question, chat_history, uploaded_file):
    # Handle file upload to Vectara
    customer_id = config('CUSTOMER_ID')  # Read from .env file
    corpus_id = config('CORPUS_ID')  # Read from .env file
    api_key = config('API_KEY')  # Read from .env file
    url = f"https://api.vectara.io/v1/upload?c={customer_id}&o={corpus_id}"

    post_headers = {
        "x-api-key": api_key,
        "customer-id": customer_id
    }

    files = {
        "file": (uploaded_file.name, uploaded_file),
        "doc_metadata": (None, json.dumps({"metadata_key": "metadata_value"})),  # Replace with your metadata
    }
    response = requests.post(url, files=files, verify=True, headers=post_headers)

    if response.status_code == 200:
        upload_status = "File uploaded successfully"
    else:
        upload_status = "Failed to upload the file"

    # Get the user's message from the chat history
    user_message = chat_history[-1][0]

    query_body = {
        "query": [
            {
                "query": user_message,  # Use the user's message as the query
                "start": 0,
                "numResults": 10,
                "corpusKey": [
                    {
                        "customerId": customer_id,
                        "corpusId": corpus_id,
                        "lexicalInterpolationConfig": {"lambda": 0.025}
                    }
                ]
            }
        ]
    }

    api_endpoint = "https://api.vectara.io/v1/query"
    return f"{upload_status}\n\nResponse from Vectara API: {response.text}"

# from vectara_retriever import VectaraRetriever  # Assuming VectaraRetriever is in vectara_retriever.py

# Function to interact with Vectara API
# def query_vectara(question, chat_history, uploaded_file):
#    # Handle file upload to Vectara
#
#    # Get the user's message from the chat history
#    user_message = chat_history[-1][0]
#
#    # Create a VectaraRetriever instance
#    retriever = VectaraRetriever(index=your_vectara_index)  # Replace with your VectaraIndex instance
#
#    # Create a QueryBundle with the user's message
#    query_bundle = QueryBundle(query_str=user_message)
#
#    # Retrieve the top k most similar nodes
#    top_k_nodes = retriever._retrieve(query_bundle)
#
#    # Format the nodes for display
#    responses = "\n".join([f"{node.node.text} (score: {node.score})" for node in top_k_nodes])
#
#    return f"{upload_status}\n\nResponse from Vectara API: {responses}"
    
# Create a Gradio ChatInterface with a text input, a file upload input, and a text output
iface = gr.Interface(
    fn=query_vectara,
    inputs=[gr.Textbox(label="Input Text"), gr.File(label="Upload a file")],
    outputs=gr.Textbox(label="Output Text"),
    title="Vectara Chatbot",
    description="Ask me anything using the Vectara API!"
)

iface.launch()