app.py

import os

import gradio as gr

from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_community.utilities.sql_database import SQLDatabase
from langchain_community.agent_toolkits import create_sql_agent

from langchain_openai import AzureChatOpenAI

ccms_db_loc = 'ccms.db'

ccms_db = SQLDatabase.from_uri(f"sqlite:///{ccms_db_loc}")

gpt4o_azure = AzureChatOpenAI(
    model_name='gpt-4o-mini',
    api_key=os.environ["AZURE_OPENAI_KEY"],
    azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"],
    api_version="2024-02-01",
    temperature=0
)

context = ccms_db.get_context()
database_schema = context['table_info']

system_message = f"""You are a SQLite expert agent designed to interact with a SQLite database.
Given an input question, create a syntactically correct SQLite query to run, then look at the results of the query and return the answer.
Unless the user specifies a specific number of examples they wish to obtain, always limit your query to at most 5 results using the LIMIT clause as per SQLite. You can order the results to return the most informative data in the database..
You can order the results by a relevant column to return the most interesting examples in the database.
Never query for all columns from a table. You must query only the columns that are needed to answer the question. Wrap each column name in double quotes (") to denote them as delimited identifiers.
You have access to tools for interacting with the database.
Only use the given tools. Only use the information returned by the tools to construct your final answer.
You MUST double check your query before executing it. If you get an error while executing a query, rewrite the query and try again.

DO NOT make any DML statements (INSERT, UPDATE, DELETE, DROP etc.) to the database.

If the question does not seem related to the database, just return "I don't know" as the answer.

Only use the following tables:
{database_schema}
"""

full_prompt = ChatPromptTemplate.from_messages(
    [
        ("system", system_message),
        ("human", '{input}'),
        MessagesPlaceholder("agent_scratchpad")
    ]
)

sqlite_agent = create_sql_agent(
    llm=gpt4o_azure,
    db=ccms_db,
    prompt=full_prompt,
    agent_type="openai-tools",
    agent_executor_kwargs={'handle_parsing_errors':True},
    max_iterations=10,
    verbose=True
)

def predict(user_input):

    try:
        response = sqlite_agent.invoke(user_input)

        prediction = response['output']

    except Exception as e:
        prediction = e
        
    return prediction

# UI

textbox = gr.Textbox(placeholder="Enter your query here", lines=6)
schema = 'The schema for the database is presented below: \n <img src="https://cdn-uploads.huggingface.co/production/uploads/64118e60756b9e455c7eddd6/S1alVt_D88qatd-N4Dkjd.png" > \n<img src="https://cdn-uploads.huggingface.co/production/uploads/64118e60756b9e455c7eddd6/81ggHEjrt6wFrMyXJtHVS.png" > (Source: https://github.com/shrivastavasatyam/Credit-Card-Management-System)'

demo = gr.Interface(
    inputs=textbox, fn=predict, outputs="text",
    title="Query a Credit Card Database",
    description="This web API presents an interface to ask questions on information stored in a credit card database.",
    article=schema,
    examples=[
        ["Who are the top 5 merchants by total transactions?", ""],
        ["Which are the top 5 cities with the highest spend and what is their percentage contribution to overall spends?", ""],
        ["Which is the highest spend month and amount for each card type?", ""],
        ["Which was the city with the lowest percentage spend for the Gold card type?", ""],
        ["What was the percentage contribution of spends by females for each card type?", ""],
        ["Which city has the highest spend to transaction ratio on weekends?", ""],
        ["Which was the city to reach 500 transactions the fastest?", ""]
    ],
    cache_examples=False,
    theme=gr.themes.Base(),
    concurrency_limit=8
)

demo.queue()
demo.launch(auth=("demouser", os.getenv('PASSWD')))