api.py

from flask import Flask, request, jsonify
import logging
from llama_index.core.postprocessor import MetadataReplacementPostProcessor
from llama_index.core import (
    load_index_from_storage,
    StorageContext,
)

app = Flask(__name__)

# Enable debugging for Flask
app.config['DEBUG'] = True

# Configure logging
logging.basicConfig(level=logging.DEBUG)

# Load the storage context and index
SC_retrieved_sentence = StorageContext.from_defaults(persist_dir="./sentence_index")
retrieved_sentence_index = load_index_from_storage(SC_retrieved_sentence)

# Initialize the query engine
sentence_query_engine = retrieved_sentence_index.as_query_engine(
    similarity_top_k=3,
    node_postprocessors=[
        MetadataReplacementPostProcessor(target_metadata_key="window")
    ],
)

# Define the template
template = """
    You are a helpful first aid assistant who has a lot of experience as an Emergency Medical Technician.
    
    You need to go through the knowledge base you have in order to provide the best possible advice to the user. 
    This is important: DO NOT GIVE FALSE INFORMATION IF YOU CAN NOT FIND ANY INFORMATION IN THE KNOWLEDGE BASE. 
    
    When you are providing the answer please follow this pattern:
        1. Do not say anything before and after the response.
        2. The response should only consist of the steps that the user should follow in giving first aid.
        3. The response should be in a JSON format. (e.g. {{"no_steps": 3, "step_1": "What to do as step 01", "step_2": "What to do as step 02", "step_3": "What to do as step 03"}})
        4. The response should be in a single line.
        
    So the emergency I have is {question}
"""

@app.route('/query', methods=['POST'])
def query():
    try:
        data = request.get_json()
        logging.debug(f"Received data: {data}")

        question = data.get('question')
        if not question:
            return jsonify({"error": "No question provided"}), 400

        # Format the template with the question
        formatted_query = template.format(question=question)
        logging.debug(f"Formatted query: {formatted_query}")

        # Query the knowledge base
        sentence_response = sentence_query_engine.query(formatted_query)
        logging.debug(f"Sentence response: {sentence_response}")

        # Return the response
        return jsonify(sentence_response)
    except Exception as e:
        logging.error(f"Error processing request: {e}", exc_info=True)
        return jsonify({"error": str(e)}), 500

if __name__ == '__main__':
    app.run(debug=True)