Spaces:
Sleeping
Sleeping
nileshhanotia
commited on
Commit
•
39e6004
1
Parent(s):
17ee535
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,208 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
3 |
+
import torch
|
4 |
+
from functools import lru_cache
|
5 |
+
import json
|
6 |
+
import mysql.connector
|
7 |
+
from mysql.connector import Error
|
8 |
+
import os
|
9 |
+
import sys
|
10 |
+
from datetime import datetime
|
11 |
+
import time
|
12 |
+
|
13 |
+
# Enable GPU if available
|
14 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
15 |
+
|
16 |
+
# Database configuration
|
17 |
+
DB_CONFIG = {
|
18 |
+
'host': 'sql12.freemysqlhosting.net',
|
19 |
+
'database': 'sql12740625',
|
20 |
+
'user': 'sql12740625',
|
21 |
+
'password': 'QGG9kdrE4g',
|
22 |
+
'port': 3306,
|
23 |
+
'pool_size': 5,
|
24 |
+
'pool_reset_session': True
|
25 |
+
}
|
26 |
+
|
27 |
+
# Global variables for model and tokenizer
|
28 |
+
GLOBAL_MODEL = None
|
29 |
+
GLOBAL_TOKENIZER = None
|
30 |
+
|
31 |
+
def initialize_model():
|
32 |
+
"""Initialize model and tokenizer globally"""
|
33 |
+
global GLOBAL_MODEL, GLOBAL_TOKENIZER
|
34 |
+
st.write("Initializing model and tokenizer...")
|
35 |
+
start_time = time.time()
|
36 |
+
|
37 |
+
model_name_sql = "premai-io/prem-1B-SQL"
|
38 |
+
GLOBAL_TOKENIZER = AutoTokenizer.from_pretrained(model_name_sql)
|
39 |
+
GLOBAL_MODEL = AutoModelForCausalLM.from_pretrained(
|
40 |
+
model_name_sql,
|
41 |
+
torch_dtype=torch.float32, # Use float32 for CPU
|
42 |
+
).to(device)
|
43 |
+
|
44 |
+
# Set model to evaluation mode
|
45 |
+
GLOBAL_MODEL.eval()
|
46 |
+
|
47 |
+
st.write(f"Model initialization took {time.time() - start_time:.2f} seconds")
|
48 |
+
|
49 |
+
def test_db_connection():
|
50 |
+
"""Test database connection with timeout"""
|
51 |
+
try:
|
52 |
+
connection = mysql.connector.connect(
|
53 |
+
**DB_CONFIG,
|
54 |
+
connect_timeout=10
|
55 |
+
)
|
56 |
+
if connection.is_connected():
|
57 |
+
db_info = connection.get_server_info()
|
58 |
+
cursor = connection.cursor()
|
59 |
+
cursor.execute("SELECT DATABASE();")
|
60 |
+
db_name = cursor.fetchone()[0]
|
61 |
+
cursor.close()
|
62 |
+
connection.close()
|
63 |
+
return True, f"Successfully connected to MySQL Server version {db_info}\nDatabase: {db_name}"
|
64 |
+
except Error as e:
|
65 |
+
return False, f"Error connecting to MySQL database: {e}"
|
66 |
+
return False, "Unable to establish database connection"
|
67 |
+
|
68 |
+
def get_db_connection():
|
69 |
+
"""Get database connection from pool"""
|
70 |
+
return mysql.connector.connect(**DB_CONFIG)
|
71 |
+
|
72 |
+
def execute_query(query):
|
73 |
+
"""Execute SQL query with timeout and connection pooling"""
|
74 |
+
connection = None
|
75 |
+
try:
|
76 |
+
connection = get_db_connection()
|
77 |
+
cursor = connection.cursor(dictionary=True, buffered=True)
|
78 |
+
cursor.execute(query)
|
79 |
+
results = cursor.fetchall()
|
80 |
+
return results
|
81 |
+
except Error as e:
|
82 |
+
return f"Error executing query: {e}"
|
83 |
+
finally:
|
84 |
+
if connection and connection.is_connected():
|
85 |
+
cursor.close()
|
86 |
+
connection.close()
|
87 |
+
|
88 |
+
def generate_sql(natural_language_query):
|
89 |
+
"""Generate SQL query with performance optimizations"""
|
90 |
+
try:
|
91 |
+
start_time = time.time()
|
92 |
+
|
93 |
+
schema_info = """
|
94 |
+
CREATE TABLE sales (
|
95 |
+
pizza_id DECIMAL(8,2) PRIMARY KEY,
|
96 |
+
order_id DECIMAL(8,2),
|
97 |
+
pizza_name_id VARCHAR(14),
|
98 |
+
quantity DECIMAL(4,2),
|
99 |
+
order_date DATE,
|
100 |
+
order_time VARCHAR(8),
|
101 |
+
unit_price DECIMAL(5,2),
|
102 |
+
total_price DECIMAL(5,2),
|
103 |
+
pizza_size VARCHAR(3),
|
104 |
+
pizza_category VARCHAR(7),
|
105 |
+
pizza_ingredients VARCHAR(97),
|
106 |
+
pizza_name VARCHAR(42)
|
107 |
+
);
|
108 |
+
"""
|
109 |
+
|
110 |
+
prompt = f"""### Task: Generate a SQL query to answer the following question.
|
111 |
+
### Database Schema:
|
112 |
+
{schema_info}
|
113 |
+
### Question: {natural_language_query}
|
114 |
+
### SQL Query:"""
|
115 |
+
|
116 |
+
inputs = GLOBAL_TOKENIZER(
|
117 |
+
prompt,
|
118 |
+
return_tensors="pt",
|
119 |
+
padding=True,
|
120 |
+
truncation=True,
|
121 |
+
max_length=512,
|
122 |
+
return_attention_mask=True
|
123 |
+
)
|
124 |
+
inputs = {k: v.to(device) for k, v in inputs.items()}
|
125 |
+
|
126 |
+
with torch.no_grad():
|
127 |
+
outputs = GLOBAL_MODEL.generate(
|
128 |
+
input_ids=inputs["input_ids"],
|
129 |
+
attention_mask=inputs["attention_mask"],
|
130 |
+
max_length=256,
|
131 |
+
temperature=0.1,
|
132 |
+
do_sample=True,
|
133 |
+
top_p=0.95,
|
134 |
+
num_return_sequences=1,
|
135 |
+
pad_token_id=GLOBAL_TOKENIZER.eos_token_id,
|
136 |
+
)
|
137 |
+
|
138 |
+
generated_query = GLOBAL_TOKENIZER.decode(outputs[0], skip_special_tokens=True)
|
139 |
+
sql_query = generated_query.split("### SQL Query:")[-1].strip()
|
140 |
+
|
141 |
+
st.write(f"SQL generation took {time.time() - start_time:.2f} seconds")
|
142 |
+
return sql_query
|
143 |
+
|
144 |
+
except Exception as e:
|
145 |
+
return f"Error generating SQL query: {str(e)}"
|
146 |
+
|
147 |
+
def format_result(query_result):
|
148 |
+
"""Format query results efficiently"""
|
149 |
+
if isinstance(query_result, str) and "Error" in query_result:
|
150 |
+
return query_result
|
151 |
+
|
152 |
+
if not query_result:
|
153 |
+
return "No results found."
|
154 |
+
|
155 |
+
# Use list comprehension for better performance
|
156 |
+
if len(query_result) == 1:
|
157 |
+
return "\n".join(f"{k}: {v}" for k, v in query_result[0].items())
|
158 |
+
|
159 |
+
results = [f"Found {len(query_result)} results:\n"]
|
160 |
+
for i, row in enumerate(query_result[:5], 1):
|
161 |
+
results.append(f"Result {i}:")
|
162 |
+
results.extend(f"{k}: {v}" for k, v in row.items())
|
163 |
+
results.append("")
|
164 |
+
|
165 |
+
if len(query_result) > 5:
|
166 |
+
results.append(f"(Showing first 5 of {len(query_result)} results)")
|
167 |
+
|
168 |
+
return "\n".join(results)
|
169 |
+
|
170 |
+
def main():
|
171 |
+
"""Main function with Streamlit UI components"""
|
172 |
+
st.title("Natural Language to SQL Query")
|
173 |
+
st.write("Ask questions about pizza sales data in plain English.")
|
174 |
+
|
175 |
+
# Test and display database connection status
|
176 |
+
db_success, db_message = test_db_connection()
|
177 |
+
st.write(db_message)
|
178 |
+
|
179 |
+
if not db_success:
|
180 |
+
st.write("Could not connect to the database. Exiting.")
|
181 |
+
return
|
182 |
+
|
183 |
+
# Initialize model
|
184 |
+
initialize_model()
|
185 |
+
|
186 |
+
# Input field for natural language query
|
187 |
+
natural_language_query = st.text_input("Enter your question", placeholder="e.g., What were the total sales for each pizza category?")
|
188 |
+
|
189 |
+
if st.button("Generate and Execute Query"):
|
190 |
+
if natural_language_query:
|
191 |
+
# Generate SQL query
|
192 |
+
sql_query = generate_sql(natural_language_query)
|
193 |
+
st.write("Generated SQL Query:", sql_query)
|
194 |
+
|
195 |
+
# Execute the generated query
|
196 |
+
query_result = execute_query(sql_query)
|
197 |
+
formatted_result = format_result(query_result)
|
198 |
+
|
199 |
+
st.write("Query Result:")
|
200 |
+
st.code(json.dumps(query_result, indent=2))
|
201 |
+
|
202 |
+
st.write("Human-Readable Response:")
|
203 |
+
st.text(formatted_result)
|
204 |
+
else:
|
205 |
+
st.write("Please enter a query.")
|
206 |
+
|
207 |
+
if __name__ == "__main__":
|
208 |
+
main()
|