1. 安装MySQL和PyMySQL
安装MySQL
# 在Ubuntu/Debian上安装
sudo apt update
sudo apt install mysql-server
sudo mysql_secure_installation
# 启动MySQL服务
sudo systemctl start mysql
sudo systemctl enable mysql
安装PyMySQL
pip install pymysql
使用 apt
安装 MySQL 后,默认情况下 root 用户没有密码 ,但需要通过 sudo
权限访问。
如果希望设置密码(推荐)
使用 mysql_secure_installation
运行以下命令交互式设置密码:
sudo mysql_secure_installation
按照提示:
-
选择密码强度验证策略(通常选
0
跳过) -
输入新密码并确认
-
后续选项建议全部选
Y
(移除匿名用户、禁止远程 root 登录等)
用 sudo 登录 MySQL
python
sudo mysql -u root
检查 MySQL 用户认证方式
登录 MySQL 后,执行:
python
SELECT user, host, plugin FROM mysql.user WHERE user='root';
修改 root 用户认证方式为密码
假设你已经用 sudo mysql 进入了 MySQL,执行:
python
ALTER USER 'root'@'localhost' IDENTIFIED WITH mysql_native_password BY '12345678';
FLUSH PRIVILEGES;
创建数据库和表
python
import pymysql
# 替换为你的MySQL root密码
MYSQL_PASSWORD = 'your_root_password'
connection = pymysql.connect(
host='localhost',
user='root',
password='12345678'
)
try:
with connection.cursor() as cursor:
# 创建数据库
cursor.execute("CREATE DATABASE IF NOT EXISTS qwen_demo")
cursor.execute("USE qwen_demo")
# 创建产品表
cursor.execute("""
CREATE TABLE IF NOT EXISTS products (
id INT AUTO_INCREMENT PRIMARY KEY,
name VARCHAR(100),
category VARCHAR(50),
price DECIMAL(10,2),
stock INT
)
""")
# 插入示例数据
cursor.execute("""
INSERT INTO products (name, category, price, stock)
VALUES
('笔记本电脑', '电子产品', 5999.00, 50),
('智能手机', '电子产品', 3999.00, 100),
('平板电脑', '电子产品', 2999.00, 30),
('办公椅', '家具', 899.00, 20),
('书桌', '家具', 1299.00, 15)
""")
connection.commit()
print("数据库和表创建成功,示例数据已插入!")
finally:
connection.close()

2. 部署Qwen3-0.5B模型
python
pip install transformers torch sentencepiece
python
from transformers import AutoModelForCausalLM, AutoTokenizer
model_path = "Qwen/Qwen1.5-0.5B"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path, device_map="auto")
def generate_sql_from_nl(query):
prompt = f"""将以下中文问题转换为SQL查询语句。只返回SQL语句,不要有其他解释或说明。
数据库表结构:
表名:products
字段:id, name, category, price, stock
问题:{query}
SQL:"""
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=200)
sql = tokenizer.decode(outputs[0], skip_special_tokens=True)
# 提取SQL部分
sql = sql.split("SQL:")[-1].strip()
return sql
测试代码:
python
from transformers import AutoModelForCausalLM, AutoTokenizer
model_path = "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-1.5B-Instruct"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path, device_map="auto")
def generate_sql_from_nl(query):
prompt = f"""将以下中文问题转换为SQL查询语句。只返回SQL语句,不要有其他解释或说明。
数据库表结构:
表名:products
字段:id, name, category, price, stock
问题:{query}
SQL:"""
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=300)
sql = tokenizer.decode(outputs[0], skip_special_tokens=True)
# 提取SQL部分
sql = sql.split("SQL:")[-1].strip()
return sql
if __name__ == "__main__":
query = "查询所有价格大于100的产品"
sql = generate_sql_from_nl(query)
print("问题:", query)
print("SQL:", sql)
3. 使用Flask部署API
python
pip install flask flask-cors
创建 app.py
:
python
from flask import Flask, request, jsonify
from flask_cors import CORS
import pymysql
from qwen_model import generate_sql_from_nl # 假设上面的Qwen代码保存在qwen_model.py
app = Flask(__name__)
CORS(app)
# MySQL配置
db_config = {
'host': 'localhost',
'user': 'root',
'password': 'your_password',
'database': 'qwen_demo',
'charset': 'utf8mb4',
'cursorclass': pymysql.cursors.DictCursor
}
@app.route('/api/query', methods=['POST'])
def handle_query():
data = request.json
user_query = data.get('query')
if not user_query:
return jsonify({'error': 'No query provided'}), 400
try:
# 生成SQL
sql = generate_sql_from_nl(user_query)
# 执行SQL
connection = pymysql.connect(**db_config)
with connection.cursor() as cursor:
cursor.execute(sql)
result = cursor.fetchall()
return jsonify({
'sql': sql,
'result': result
})
except Exception as e:
return jsonify({'error': str(e)}), 500
if __name__ == '__main__':
app.run(host='0.0.0.0', port=5000)
启动Flask服务:
python app.py
python
from flask import Flask, request, jsonify
from flask_cors import CORS
import pymysql
from qwen_model import generate_sql_from_nl # 假设上面的Qwen代码保存在qwen_model.py
app = Flask(__name__)
CORS(app)
# MySQL配置
db_config = {
'host': 'localhost',
'user': 'root',
'password': '12345678',
'database': 'qwen_demo',
'charset': 'utf8mb4',
'cursorclass': pymysql.cursors.DictCursor
}
@app.route('/api/query', methods=['POST'])
def handle_query():
data = request.json
user_query = data.get('query')
if not user_query:
return jsonify({'error': 'No query provided'}), 400
try:
# 生成SQL
sql = generate_sql_from_nl(user_query)
# 执行SQL
connection = pymysql.connect(**db_config)
with connection.cursor() as cursor:
cursor.execute(sql)
result = cursor.fetchall()
return jsonify({
'sql': sql,
'result': result
})
except Exception as e:
return jsonify({'error': str(e)}), 500
if __name__ == '__main__':
app.run(host='0.0.0.0', port=5000)
test_api.py
python
import requests
url = "http://127.0.0.1:5000/api/query"
data = {
"query": "价格大于3000的产品" # 这里可以换成你想测试的自然语言问题
}
response = requests.post(url, json=data)
print("Status Code:", response.status_code)
print("Response:", response.json())
4. 在Dify中创建工作流
-
登录Dify平台
-
创建一个新的工作流
-
添加以下节点:
节点1: 用户输入
-
类型:输入节点
-
配置:接收用户的中文查询
节点2: 调用Flask API
-
类型:HTTP请求节点
-
配置:
-
方法: POST
-
Headers:
- Content-Type: application/json
-
Body:
{ "query": "{{input.query}}" }
节点3: 结果格式化
-
类型:JavaScript处理节点
-
代码:
function formatResult(data) {
const result = data.result;
if (result.length === 0) return "没有找到匹配的结果";
let output = "查询结果:\\n";
result.forEach(item => {
output += `名称: ${item.name}, 类别: ${item.category}, 价格: ${item.price}, 库存: ${item.stock}\\n`;
});
return {
sql: data.sql,
result: output
};
}
return formatResult(input);
节点4: 输出结果
-
类型:输出节点
-
配置:显示格式化后的结果