'''
参考： https://github.com/shroominic/codeinterpreter-api
1. 可以存在本地，然后再调出来。 working.
1. 可以在临时文件夹中读取文件。
1. 可以直接在内存中读出图片。
1. 中文字体成功。
from matplotlib.font_manager import FontProperties
myfont=FontProperties(fname='/Users/yunshi/Downloads/360Data/Data Center/Working-On Task/演讲与培训/2023ChatGPT/Coding/code_interpreter/rawdata/SimHei.ttf')
sns.set_style('whitegrid',{'font.sans-serif':['simhei','Arial']})
1. 解决了account login的问题，主要格式:
## 需要严格的按照yaml文件的格式来定义如下几个字段。
for un, name, pw in zip(users, names, hashed_passwords):
# user_dict = {"name":name,"password":pw}
user_dict = {"name": un, "password": pw}
# credentials["usernames"].update({un:user_dict})
credentials["usernames"].update({un: user_dict})
'''
# TODO：1. Chinese display isssue. 2. account system. 3. local enterprise database.
import database as db
from deta import Deta # pip3 install deta
import requests
from codeinterpreterapi import CodeInterpreterSession, File
import streamlit as st
# from codeinterpreterapi import CodeInterpreterSession
import openai
import os
import matplotlib.pyplot as plt
import pandas as pd
# import csv
import tempfile
from tempfile import NamedTemporaryFile
import pathlib
from pathlib import Path
from matplotlib.font_manager import FontProperties
import seaborn as sns
from time import sleep
import streamlit_authenticator as stauth
import database as db # python文件同目录下的.py程序，直接导入。
import deta
from langchain.chat_models import ChatOpenAI
from llama_index import StorageContext, load_index_from_storage, GPTVectorStoreIndex, LLMPredictor, PromptHelper
from llama_index import ServiceContext, QuestionAnswerPrompt
import sys
import time
import PyPDF2 ## read the local_KB PDF file.
# import localKB_construct
import save_database_info
from datetime import datetime
import pytz
os.environ["OPENAI_API_KEY"] = os.environ['user_token']
openai.api_key = os.environ['user_token']
# os.environ["VERBOSE"] = "True" # 可以看到具体的错误？
#* 如果碰到接口问题，可以启用如下设置。
# openai.proxy = {
# "http": "http://127.0.0.1:7890",
# "https": "http://127.0.0.1:7890"
# }
# layout settings.
st.title("专业版大语言模型商业智能中心")
st.subheader("Artificial Intelligence Backend Center for Professionals")
# clear conversion.
reset_button_key = "reset_button"
reset_button = st.button(label=("扫清世间烦恼，清除所有记录，并开启一轮新对话 ▶"),
key=reset_button_key, use_container_width=True, type="secondary")
if reset_button:
st.session_state.conversation = None
st.session_state.chat_history = None
st.session_state.messages = []
message_placeholder = st.empty()
def clear_all():
st.session_state.conversation = None
st.session_state.chat_history = None
st.session_state.messages = []
message_placeholder = st.empty()
return None
# # with tab2:
# def upload_file(uploaded_file):
# if uploaded_file is not None:
# filename = uploaded_file.name
# # st.write(filename) # print out the whole file name to validate. not to show in the final version.
# try:
# if '.pdf' in filename:
# # pdf_file = PyPDF2.PdfReader(uploaded_file)
# PyPDF2.PdfReader(uploaded_file)
# # st.write(pdf_file.pages[0].extract_text())
# # with st.status('正在为您解析新知识库...', expanded=False, state='running') as status:
# spinner = st.spinner('正在为您解析新知识库...请耐心等待')
# # with st.spinner('正在为您解析新知识库...请耐心等待'):
# with spinner:
# import localKB_construct
# # sleep(3)
# # st.write(upload_file)
# localKB_construct.process_file(uploaded_file)
# st.markdown('新知识库解析成功，可以开始对话！')
# spinner = st.empty()
# # sleep(3)
# # display = []
# else:
# if '.csv' in filename:
# csv_file = pd.read_csv(uploaded_file)
# csv_file.to_csv('./upload.csv', encoding='utf-8', index=False)
# st.write(csv_file[:3]) # 这里只是显示文件，后面需要定位文件所在的绝对路径。
# else:
# xls_file = pd.read_excel(uploaded_file)
# xls_file.to_csv('./upload.csv', index=False)
# st.write(xls_file[:3])
# uploaded_file_name = "File_provided"
# temp_dir = tempfile.TemporaryDirectory()
# # ! working.
# uploaded_file_path = pathlib.Path(temp_dir.name) / uploaded_file_name
# # with open('./upload.csv', 'wb') as output_temporary_file:
# with open(f'./{name}_upload.csv', 'wb') as output_temporary_file:
# # print(f'./{name}_upload.csv')
# # ! 必须用这种格式读入内容，然后才可以写入temporary文件夹中。
# # output_temporary_file.write(uploaded_file.getvalue())
# output_temporary_file.write(uploaded_file.getvalue())
# # st.write(uploaded_file_path) #* 可以查看文件是否真实存在，然后是否可以
# # st.write('Now file saved successfully.')
# except Exception as e:
# st.write(e)
# # uploaded_file_name = "File_provided"
# # temp_dir = tempfile.TemporaryDirectory()
# # # ! working.
# # uploaded_file_path = pathlib.Path(temp_dir.name) / uploaded_file_name
# # # with open('./upload.csv', 'wb') as output_temporary_file:
# # with open(f'./{name}_upload.csv', 'wb') as output_temporary_file:
# # # print(f'./{name}_upload.csv')
# # # ! 必须用这种格式读入内容，然后才可以写入temporary文件夹中。
# # # output_temporary_file.write(uploaded_file.getvalue())
# # output_temporary_file.write(uploaded_file.getvalue())
# # # st.write(uploaded_file_path) # * 可以查看文件是否真实存在，然后是否可以
# # # st.write('Now file saved successfully.')
# return None
bing_search_api_key = os.environ['bing_api_key']
bing_search_endpoint = 'https://api.bing.microsoft.com/v7.0/search'
def search(query):
# Construct a request
# mkt = 'en-EN'
mkt = 'zh-CN'
params = {'q': query, 'mkt': mkt}
headers = {'Ocp-Apim-Subscription-Key': bing_search_api_key}
# Call the API
try:
response = requests.get(bing_search_endpoint,
headers=headers, params=params)
response.raise_for_status()
json = response.json()
return json["webPages"]["value"]
# print("\nJSON Response:\n")
# pprint(response.json())
except Exception as e:
raise e
# openai.api_key = st.secrets["OPENAI_API_KEY"]
async def text_mode():
# Set a default model
if "openai_model" not in st.session_state:
st.session_state["openai_model"] = "gpt-3.5-turbo-16k"
if radio_1 == 'GPT-3.5':
# print('----------'*5)
print('radio_1: GPT-3.5 starts!')
st.session_state["openai_model"] = "gpt-3.5-turbo-16k"
else:
print('radio_1: GPT-4.0 starts!')
st.session_state["openai_model"] = "gpt-4"
# Initialize chat history
if "messages" not in st.session_state:
st.session_state.messages = []
# Display chat messages from history on app rerun
for message in st.session_state.messages:
with st.chat_message(message["role"]):
st.markdown(message["content"])
# Display assistant response in chat message container
# if prompt := st.chat_input("Say something"):
prompt = st.chat_input("Say something")
print('prompt now:', prompt)
print('----------'*5)
# if prompt:
if prompt:
st.session_state.messages.append({"role": "user", "content": prompt})
with st.chat_message("user"):
st.markdown(prompt)
with st.chat_message("assistant"):
message_placeholder = st.empty()
full_response = ""
if radio_2 == '联网模式':
print('联网模式入口，prompt:', prompt)
input_message = prompt
internet_search_result = search(input_message)
search_prompt = [
f"Source:\nTitle: {result['name']}\nURL: {result['url']}\nContent: {result['snippet']}" for result in internet_search_result]
prompt = "基于如下的互联网公开信息, 回答问题:\n\n" + \
"\n\n".join(search_prompt[:3]) + "\n\n问题: " + input_message + \
"你需要注意的是回答问题时必须用提问的语言（如英文或者中文）来提示：'答案基于互联网公开信息。'" + "\n\n答案: " # 限制了只有3个搜索结果。
# prompt = "Use these sources to answer the question:\n\n" + "\n\n".join(search_prompt[0:3]) + "\n\nQuestion: " + input_message + "（注意：回答问题时请提示'以下答案基于互联网公开信息。'）\n\n" + "\n\nAnswer: "
st.session_state.messages.append(
{"role": "user", "content": prompt})
for response in openai.ChatCompletion.create(
model=st.session_state["openai_model"],
messages=[
{"role": m["role"], "content": m["content"]}
for m in st.session_state.messages
],
stream=True,
):
full_response += response.choices[0].delta.get(
"content", "")
message_placeholder.markdown(full_response + "▌")
message_placeholder.markdown(full_response)
st.session_state.messages.append(
{"role": "assistant", "content": full_response})
st.session_state.messages = []
if radio_2 == '核心模式':
print('GPT only starts!!!')
print('messages:', st.session_state['messages'])
for response in openai.ChatCompletion.create(
model=st.session_state["openai_model"],
messages=[
{"role": m["role"], "content": m["content"]}
for m in st.session_state.messages
],
stream=True,
# messages=[{'role': 'system', 'content': 'you are ChatGPT'}, {
# 'role': 'user', 'content': prompt}],
# stream=True,
):
full_response += response.choices[0].delta.get(
"content", "")
message_placeholder.markdown(full_response + "▌")
# print('session completed!')
message_placeholder.markdown(full_response)
st.session_state.messages.append(
{"role": "assistant", "content": full_response})
## load the local_KB PDF file.
# async def localKB_mode():
def localKB_mode(username):
### clear all the prior conversation.
# st.session_state.conversation = None
# st.session_state.chat_history = None
# st.session_state.messages = []
# message_placeholder = st.empty()
clear_all() ## reset the conversation.
print('now starts the local KB version of ChatGPT')
# Initialize chat history
if "messages" not in st.session_state:
st.session_state.messages = []
for message in st.session_state.messages:
with st.chat_message(message["role"]):
st.markdown(message["content"])
# Display assistant response in chat message container
# if prompt := st.chat_input("Say something"):
# prompt = st.chat_input("Say something")
# print('prompt now:', prompt)
# print('----------'*5)
# if prompt:
if prompt := st.chat_input("Say something"):
st.session_state.messages.append({"role": "user", "content": prompt})
with st.chat_message("user"):
st.markdown(prompt)
with st.status('检索中...', expanded=True, state='running') as status:
with st.chat_message("assistant"):
message_placeholder = st.empty()
full_response = ""
# if radio_2 == "知识库模式":
# ! 这里需要重新装载一下storage_context。
QA_PROMPT_TMPL = (
"We have provided context information below. \n"
"---------------------\n"
"{context_str}"
"\n---------------------\n"
"Given all this information, please answer the following questions,"
"You MUST use the SAME language as the question:\n"
"{query_str}\n")
QA_PROMPT = QuestionAnswerPrompt(QA_PROMPT_TMPL)
# print('QA_PROMPT:', QA_PROMPT)
# llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.8, model_name="gpt-3.5-turbo", max_tokens=4024,streaming=True))
# # print('llm_predictor:', llm_predictor)
# prompt_helper = PromptHelper(max_input_size, num_outputs, max_chunk_overlap, chunk_size_limit=chunk_size_limit)
# print('prompt_helper:', prompt_helper)
# service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor, prompt_helper=prompt_helper)
# print('service_context:', service_context)
# # # index = load_index_from_storage(storage_context)
# print("storage_context:", storage_context)
# index = load_index_from_storage(storage_context,service_context=service_context)
storage_context = StorageContext.from_defaults(persist_dir=f"./{username}/")
print('storage_context:',storage_context)
index = load_index_from_storage(storage_context)
# query_engine = index.as_query_engine(streaming=True, similarity_top_k=3, text_qa_template=QA_PROMPT)
query_engine = index.as_query_engine(streaming=True)
# query_engine = index.as_query_engine(streaming=True, text_qa_template=QA_PROMPT)
# query_engine = index.as_query_engine(streaming=False, text_qa_template=QA_PROMPT)
# query_engine = index.as_query_engine()
# reply = query_engine.query(prompt)
llama_index_reply = query_engine.query(prompt)
# full_response += query_engine.query(prompt)
print('local KB reply:', llama_index_reply)
# query_engine.query(prompt).print_response_stream() #* 能在terminal中流式输出。
# for resp in llama_index_reply.response_gen:
# print(resp)
# full_response += resp
# message_placeholder.markdown(full_response + "▌")
message_placeholder.markdown(llama_index_reply)
# st.session_state.messages.append(
# {"role": "assistant", "content": full_response})
# st.session_state.messages = []
# full_response += reply
# full_response = reply
# st.session_state.messages.append(
# {"role": "assistant", "content": full_response})
async def data_mode():
print('数据分析模式启动！')
clear_all() ## reset the conversation.
# uploaded_file_path = './upload.csv'
uploaded_file_path = f'./{username}_upload.csv'
# uploaded_file_path = "/Users/yunshi/Downloads/360Data/Data Center/Working-On Task/演讲与培训/2023ChatGPT/Coding/code_interpreter/test_upload.csv"
print('file path:', uploaded_file_path)
# st.write(f"passed file path in data_mode: {uploaded_file_path}")
# tmp1 = pd.read_csv(uploaded_file_path)
# st.markdown('成功启动数据模式，以下是加载的文件内容')
# st.write(tmp1[:5])
# Initialize chat history
if "messages" not in st.session_state:
st.session_state.messages = []
# Display chat messages from history on app rerun
for message in st.session_state.messages:
with st.chat_message(message["role"]):
st.markdown(message["content"])
# Display assistant response in chat message container
# if prompt := st.chat_input("Say something"):
prompt = st.chat_input("Say something")
print('prompt now:', prompt)
print('----------'*5)
# if prompt:
if prompt:
st.session_state.messages.append({"role": "user", "content": prompt})
with st.chat_message("user"):
st.markdown(prompt)
with st.chat_message("assistant"):
async with CodeInterpreterSession() as session:
# user_request = "对于文件中的'SepalLengthCm’数据给我一个'直方图'，提供图表，并给出分析结果"
#! 可以用设定dpi=300来输出高质量的图表。（注：图的解析度dpi设定为300）
environ_settings = """【背景要求】如果我没有告诉你任何定制化的要求，那么请你按照以下的默认要求来回答：
-------------------------------------------------------------------------
1. 你需要用提问的语言来回答（如：中文提问你就用中文来回答，英文提问你就用英文来回答）。
2. 如果要求你输出图表，那么图的解析度dpi需要设定为600。图尽量使用seaborn库。seaborn库的参数设定：sns.set(rc={'axes.facecolor':'#FFF9ED','figure.facecolor':'#FFF9ED'}, palette='dark'。
3. 图上所有的文字全部翻译成<英文English>来表示。
4. 你回答的文字内容必须尽可能的详细且通俗易懂。
5. 回答时尽可能地展示分析所对应的图表，并提供分析结果。 你需要按如下格式提供内容：
5.1 提供详细且专业的分析结果，提供足够的分析依据。
5.2 给出可能造成这一结果的可能原因有哪些？
以上内容全部用【1/2/3这样的序列号格式】来表达。
-------------------------------------------------------------------------
""" # seaborn中的palette参数可以设定图表的颜色,选项包括：deep, muted, pastel, bright, dark, colorblind，Spectral。更多参数可以参考：https://seaborn.pydata.org/generated/seaborn.color_palette.html。
# uploaded_file_path = upload_file()
user_request = environ_settings + "\n\n" + \
"你需要完成以下任务：\n\n" + prompt + "\n\n" \
f"注：文件位置在 {uploaded_file_path}"
user_request = str(prompt)
print('user_request: \n', user_request)
# 加载上传的文件，主要路径在上面代码中。
files = [File.from_path(str(uploaded_file_path))]
# files = [File.from_path("/Users/yunshi/Downloads/360Data/Data Center/Working-On Task/演讲与培训/2023ChatGPT/Coding/code_interpreter/rawdata/short_csv.csv")]
# st.write(pd.DataFrame(files))
# print('session.__init__', session.__init__)
# print('session', session.__init__)
with st.status('Thinking...', expanded=True, state='running') as status:
# generate the response
# response = await session.generate_response(user_msg=user_request, files=files, detailed_error=True)
response = await session.generate_response(user_msg=user_request, files=files, detailed_error=True)
# output to the user
print("AI: ", response.content)
full_response = response.content
### full_response = "this is full response"
# for file in response.files:
for i, file in enumerate(response.files):
# await file.asave(f"/Users/yunshi/Downloads/360Data/Data Center/Working-On Task/演讲与培训/2023ChatGPT/Coding/code_interpreter/output{i}.png") ##working.
# st.image(file.get_image()) #! working.
# * 注意这里的设定，可以提高图片的精细程度。
st.image(file.get_image(), width=None,
output_format='PNG')
# message_placeholder.markdown(full_response + "▌") ## orignal code.
# message_placeholder.markdown(full_response) ## orignal code.
st.write(full_response)
status.update(label='complete', state='complete')
# TODO: 确认是否要记录所有的full response。
st.session_state.messages.append(
{"role": "assistant", "content": full_response})
await session.astop() # ! 确认需要关闭。
# st.session_state.messages.append({"role": "assistant", "content": full_response})
### authentication with a local yaml file.
import yaml
from yaml.loader import SafeLoader
with open('./config.yaml') as file:
config = yaml.load(file, Loader=SafeLoader)
authenticator = stauth.Authenticate(
config['credentials'],
config['cookie']['name'],
config['cookie']['key'],
config['cookie']['expiry_days'],
config['preauthorized']
)
###'''authentication with a remove cloud-based database.'''
# authentication with a remove cloud-based database.
# 导入云端用户数据库。
# DETA_KEY = "c0zegv33efm_4MBTaoQAn76GzUfsZeKV64Uh9qMY3WZb"
# load_dotenv(".env")
# DETA_KEY = os.getenv("DETA_KEY")
# print(DETA_KEY)
# deta = Deta(DETA_KEY)
# # mybase is the name of the database in Deta. You can change it to any name you want.
# credentials = {"usernames":{}}
# users = []
# email = []
# passwords = []
# names = []
# for row in db.fetch_all_users():
# users.append(row["username"])
# email.append(row["email"])
# names.append(row["key"])
# passwords.append(row["password"])
# hashed_passwords = stauth.Hasher(passwords).generate()
## 需要严格的按照yaml文件的格式来定义如下几个字段。
# for un, name, pw in zip(users, names, hashed_passwords):
# # user_dict = {"name":name,"password":pw}
# user_dict = {"name": un, "password": pw}
# # credentials["usernames"].update({un:user_dict})
# credentials["usernames"].update({un: user_dict})
# ## sign-up模块，未完成。
# database_table = []
# # print(pd.DataFrame(credentials))
# for i in credentials['usernames'].keys():
# # print("i:",i)
# # print("name",credentials['usernames'][i]['name'])
# # print("password",credentials['usernames'][i]['password'])
# database_table.append([i,credentials['usernames'][i]['name'],credentials['usernames'][i]['password']])
# print("database_table:",database_table)
# authenticator = stauth.Authenticate(
# credentials=credentials, cookie_name="joeshi_gpt", key='abcedefg', cookie_expiry_days=30)
# ## sign-up widget，未完成。
# try:
# if authenticator.register_user('新用户注册', preauthorization=False):
# # for list in database_table:
# # db.update_user(username=list[0], name=list[1], password=list[2])
# db.update_user(username=list[-1][0], name=list[-1][1], password=list[-1][2])
# # st.success('User registered successfully')
# st.success('注册成功！')
# except Exception as e:
# st.error(e)
''''''
# user, authentication_status, username = authenticator.login('用户登录', 'main')
user, authentication_status, username = authenticator.login('用户登录', 'sidebar')
# print("name", name, "username", username)
if authentication_status:
with st.sidebar:
st.markdown(
"""
<style>
[data-testid="stSidebar"][aria-expanded="true"]{
min-width: 600px;
max-width: 600px;
}
""",
unsafe_allow_html=True,
)
st.header(f'**欢迎 **{username}** 来到人工智能的世界** ♠')
st.write(f'_Welcome and Hope U Enjoy Staying Here！_')
authenticator.logout('登出', 'sidebar')
## reset password widget
# try:
# if authenticator.reset_password(st.session_state["username"], 'Reset password'):
# st.success('Password modified successfully')
# except Exception as e:
# st.error(e)
# with st.header(body="欢迎"):
# st.markdown("# 欢迎使用大语言模型商业智能中心")
# with st.expander(label=("**重要的使用注意事项**"), expanded=True):
with st.container():
st.markdown("#### 快速上手指南")
# with st.text(body="说明"):
# st.markdown("* 重启一轮新对话时，只需要刷新页面（按Ctrl/Command + R）即可。")
with st.text(body="说明"):
st.markdown("* 为了保护数据与隐私，所有对话均不会被保存，刷新页面立即删除。敬请放心。")
with st.text(body="说明"):
st.markdown("* “GPT-4”回答质量极佳，但速度缓慢、且不支持长文。建议适当使用。")
with st.text(body="说明"):
st.markdown("* “联网模式”和“知识库模式”均基于检索功能，仅限一轮对话，不会保持之前的会话记录。")
with st.text(body="说明"):
st.markdown(
"* “数据模式”暂时只支持1000个单元格以内的数据分析，单元格中的内容不支持中文数据（表头也尽量不使用中文）。一般运行时间在1-5分钟左右，期间需要保持网络畅通。")
with st.text(body="说明"):
st.markdown("* “数据模式”推荐上传csv格式的文件，部分Excel文件容易出现数据不兼容的情况。")
st.markdown("#### 参考资料")
with st.expander(label="**核心模式的专用提示词Prompt示例**", expanded=False):
# with st.subheader(body="提示词Prompt"):
st.code(
body="继续用中文写一篇关于 [文章主题] 的文章，以下列句子开头：[文章开头］。", language='plaintext')
st.code(body="将以下文字概括为 100 个字，使其易于阅读和理解。避免使用复杂的句子结构或技术术语。",
language='plaintext')
st.code(body="给我出一个迪奥2023春季发布会活动策划。", language='plaintext')
st.code(body="帮我按照正式会议结构写一个会邀：主题是xx手机游戏立项会议。", language='plaintext')
st.code(body="帮我写一个车内健康监测全场景落地的项目计划，用表格。", language='plaintext')
st.code(
body="同时掷两枚质地均匀的骰子，则两枚骰子向上的点数之和为 7 的概率是多少。", language='plaintext')
st.code(body="写一篇产品经理的演讲稿，注意使用以下词汇: 赋能,抓手,中台,闭环,落地,漏斗,沉淀,给到,同步,对齐,对标,迭代,拉通,打通,升级,交付,聚焦,倒逼,复盘,梳理,方案,联动,透传,咬合,洞察,渗透,兜底,解耦,耦合,复用,拆解。", language='plaintext')
with st.expander(label="**数据模式的专用提示词Prompt示例**", expanded=False):
# with st.subheader(body="提示词Prompt"):
st.code(body="分析此数据集并绘制一些'有趣的图表'。", language='python')
st.code(
body="对于这个文件中的数据，你需要要找出[X,Y]数据之间的寻找'相关性'。", language='python')
st.code(body="对于这个文件中的[xxx]数据给我一个'整体的分析'。", language='python')
st.code(body="对于[xxx]数据给我一个'直方图'，提供图表，并给出分析结果。", language='python')
st.code(body="对于[xxx]数据给我一个'小提琴图'，并给出分析结果。", language='python')
st.code(
body="对于[X,Y,Z]数据在一个'分布散点图 (stripplot)'，所有的数据在一张图上展现, 并给出分析结果。", language='python')
st.code(body="对于[X，Y]数据，进行'T检验'，你需要展示图表，并给出分析结果。",
language='python')
st.code(body="对于[X，Y]数据给我一个3个类别的'聚类分析'，并给出分析结果。",
language='python')
col1, col2 = st.columns(spec=[1, 2])
radio_2 = col2.radio(label='模式选择', options=[
'核心模式', '联网模式', '知识库模式', '数据模式'], horizontal=True, label_visibility='visible')
radio_1 = col1.radio(label='ChatGPT版本', options=[
'GPT-3.5', 'GPT-4.0'], horizontal=True, label_visibility='visible')
elif authentication_status == False:
st.error('⛔ 用户名或密码错误！')
elif authentication_status == None:
st.warning('⬅ 请先登录！')
### 上传文件的模块
def upload_file(uploaded_file):
if uploaded_file is not None:
filename = uploaded_file.name
# st.write(filename) # print out the whole file name to validate. not to show in the final version.
try:
if '.pdf' in filename:
# pdf_file = PyPDF2.PdfReader(uploaded_file)
PyPDF2.PdfReader(uploaded_file)
# st.write(pdf_file.pages[0].extract_text())
# with st.status('正在为您解析新知识库...', expanded=False, state='running') as status:
spinner = st.spinner('正在为您解析新知识库...请耐心等待')
# with st.spinner('正在为您解析新知识库...请耐心等待'):
with spinner:
import localKB_construct
# st.write(upload_file)
localKB_construct.process_file(uploaded_file, username)
## 在屏幕上展示当前知识库的信息，包括名字和加载日期。
save_database_info.save_database_info(f'./{username}/database_name.csv', filename, str(datetime.now(pytz.timezone('Asia/Shanghai')).strftime("%Y-%m-%d %H:%M")))
st.markdown('新知识库解析成功，请务必刷新页面，然后开启对话 🔃')
# spinner = st.empty()
else:
if '.csv' in filename:
csv_file = pd.read_csv(uploaded_file)
csv_file.to_csv(f'./{username}_upload.csv', encoding='utf-8', index=False)
st.write(csv_file[:3]) # 这里只是显示文件，后面需要定位文件所在的绝对路径。
else:
xls_file = pd.read_excel(uploaded_file)
xls_file.to_csv(f'./{username}_upload.csv', index=False)
st.write(xls_file[:3])
uploaded_file_name = "File_provided"
temp_dir = tempfile.TemporaryDirectory()
# ! working.
# uploaded_file_path = pathlib.Path(temp_dir.name) / uploaded_file_name
# with open('./upload.csv', 'wb') as output_temporary_file:
with open(f'./{username}_upload.csv', 'wb') as output_temporary_file:
# print(f'./{name}_upload.csv')
# ! 必须用这种格式读入内容，然后才可以写入temporary文件夹中。
# output_temporary_file.write(uploaded_file.getvalue())
output_temporary_file.write(uploaded_file.getvalue())
# st.write(uploaded_file_path) #* 可以查看文件是否真实存在，然后是否可以
# st.write('Now file saved successfully.')
except Exception as e:
st.write(e)
## 以下代码是为了解决上传文件后，文件路径和文件名不对的问题。
# uploaded_file_name = "File_provided"
# temp_dir = tempfile.TemporaryDirectory()
# # ! working.
# uploaded_file_path = pathlib.Path(temp_dir.name) / uploaded_file_name
# # with open('./upload.csv', 'wb') as output_temporary_file:
# with open(f'./{name}_upload.csv', 'wb') as output_temporary_file:
# # print(f'./{name}_upload.csv')
# # ! 必须用这种格式读入内容，然后才可以写入temporary文件夹中。
# # output_temporary_file.write(uploaded_file.getvalue())
# output_temporary_file.write(uploaded_file.getvalue())
# # st.write(uploaded_file_path) # * 可以查看文件是否真实存在，然后是否可以
# # st.write('Now file saved successfully.')
return None
if __name__ == "__main__":
import asyncio
try:
if radio_2 == "核心模式":
print(f'radio 选择了 {radio_2}')
# * 也可以用命令执行这个python文件。’streamlit run frontend/app.py‘
asyncio.run(text_mode())
if radio_2 == "联网模式":
print(f'radio 选择了 {radio_2}')
asyncio.run(text_mode())
if radio_2 == "知识库模式":
print(f'radio 选择了 {radio_2}')
path = f'./{username}/vector_store.json'
if os.path.exists(path):
database_info = pd.read_csv(f'./{username}/database_name.csv')
current_database_name = database_info.iloc[-1][0]
current_database_date = database_info.iloc[-1][1]
database_claim = f"当前知识库为：{current_database_name}，创建于{current_database_date}。可以开始提问！"
st.markdown(database_claim)
# st.markdown("注意：系统中已经存在一个知识库，您现在可以开始提问！")
uploaded_file = st.file_uploader(
"选择上传一个新知识库", type=(["pdf"]))
# 默认状态下没有上传文件，None，会报错。需要判断。
if uploaded_file is not None:
# uploaded_file_path = upload_file(uploaded_file)
upload_file(uploaded_file)
localKB_mode(username)
# asyncio.run(localKB_mode())
if radio_2 == "数据模式":
uploaded_file = st.file_uploader(
"选择一个文件", type=(["csv", "xlsx", "xls"]))
# 默认状态下没有上传文件，None，会报错。需要判断。
if uploaded_file is not None:
uploaded_file_path = upload_file(uploaded_file)
asyncio.run(data_mode())
except:
# st.markdown('**请先登录！**')
pass