dism.py

import nltk
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer

# Download necessary NLTK data
nltk.download('punkt')

# Load the OmniParser model and tokenizer
model_name = "microsoft/OmniParser"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

# SQLBot function to generate SQL commands
def generate_sql_with_omnparser(query):
    # Encode the input query
    inputs = tokenizer.encode(query, return_tensors="pt")

    # Generate SQL command
    outputs = model.generate(inputs, max_length=50)
    
    # Decode the generated SQL command
    sql_command = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return sql_command

# SQLBot function with personality
def sqlbot(query):
    gpt_sql_command = generate_sql_with_omnparser(query)
    
    # Adding personality to the bot
    if "create table" in query.lower():
        response = f"Alright, rolling up my sleeves to create that table for you! Here it is:\n{gpt_sql_command}"
    elif "select" in query.lower():
        response = f"Got it! Fetching the data you need:\n{gpt_sql_command}"
    elif "show tables" in query.lower():
        response = f"Let me show you all the tables you've got:\n{gpt_sql_command}"
    elif "insert" in query.lower():
        response = f"Great! Adding new records as requested:\n{gpt_sql_command}"
    elif "update" in query.lower():
        response = f"Time to make some updates! Here you go:\n{gpt_sql_command}"
    elif "delete" in query.lower():
        response = f"Okay, we're deleting those records:\n{gpt_sql_command}"
    else:
        response = f"Here's what I found:\n{gpt_sql_command}"
    
    return response

# Example usage
user_query = "Create table employees with name age department"
generated_sql = sqlbot(user_query)
print(generated_sql)

# Another example usage
user_query = "Insert into users (name, age) values ('Alice', 30)"
generated_sql = sqlbot(user_query)
print(generated_sql)