PyLMKit(9)：ChatTable与你的表格聊天，表格问答

功能介绍

与你的结构化数据聊天：支持主流数据库、表格型excel等数据！

• ChatDB：支持数据库问答
• ChatTable：支持txt,excel,csv等pandas dataframe表格的问答

1.下载安装

pip install pylmkit -U
pip install pandasql

2.ChatTable实现样例

样例数据下载：（1000行员工信息）
employees.csv
可以在python运行，也可以streamlit Web运行。文件名为demo.py

import streamlit as st
from pylmkit.web.webui import ChatDBWebUI
from dotenv import load_dotenv
from pylmkit.app.chatdb import ChatTable
from pylmkit.llms import ChatTongyi
load_dotenv()


file_path = r'C:\Users\xxx\Desktop\pylmkit\test_datasets\employees.csv'
include_table_column_comments = {
    "employees": {
        "name": "雇员名单表",
        "columns": {
            "emp_no": "雇员ID",
            "birth_date": "出生日期",
            "first_name": "名字",
            "last_name": "姓氏",
            "hire_date": "雇佣日期",
            "gender": "性别",
        }
    }
}

# python中运行
# chattable = ChatTable(
#     model=ChatTongyi(),
#     table_paths=file_path,
#     include_table_column_comments=include_table_column_comments
# )
# res = chattable.invoke(
#     question="性别为M的员工有多少人？"
# )
# print(res)


# web中运行
if 'chattable' not in st.session_state:
    st.session_state.chattable = ChatTable(
        model=ChatTongyi(),
        table_paths=file_path,
        include_table_column_comments=include_table_column_comments
    )
web = ChatDBWebUI(language='zh')

web.run(
    obj=st.session_state.chattable.invoke,
    input_param=[
        {"name": "question", "label": "输入", "type": "chat"},
        {"name": "max_rollback_num", "label": "最大回滚数", "type": "int", "value": 5},
        {"name": "return_dict", "label": "返回字典", "type": "bool", "value": False},
    ],
    output_param=[
        {'label': '结果', 'name': 'ai', 'type': 'chat'},
    ]
)

然后在该路径下的终端运行下面命令：

streamlit run demo.py

默认会打开浏览器网页，也可以通过网址访问： http://localhost:8501/

GitHub
GitHub - 52phm/pylmkit: PyLMKit: 帮助用户快速构建实用的大模型应用