-
Notifications
You must be signed in to change notification settings - Fork 0
/
app.py
168 lines (150 loc) · 7.03 KB
/
app.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
import os.path
import shutil
import time
import gradio as gr
from configs.model_config import *
from langchain.chat_models import ChatOpenAI
from utils.utils import ChatMessageHistory, load_file, VectorStore
from langchain.schema import (
AIMessage,
BaseChatMessageHistory,
BaseMessage,
HumanMessage,
SystemMessage
)
# 初始化记忆池
chat_history = ChatMessageHistory()
chatkn_history = ChatMessageHistory()
# 初始化向量数据库
vector_store = VectorStore()
vector_store.create_vector_store(documents=None)
def move_file(source_path, destination_path):
try:
shutil.move(source_path, destination_path)
print(f"移动文件{str(source_path)}到{str(destination_path)}")
return f"已增加文件{str(destination_path)}至数据库"
except Exception as e:
print(f"移动文件{str(source_path)}失败: {str(e)}")
return f"移动文件{str(source_path)}失败: {str(e)}"
def save_file(files, sentence_size):
"""
建立数据库的回调函数，先读取文件，再
"""
vector_store = VectorStore()
directory_path = os.path.join(VS_ROOT_PATH, "docs")
os.makedirs(directory_path, exist_ok=True) # 创建保存文件的目录
try:
if isinstance(files, list):
result = []
for file in files:
filebasename = os.path.basename(file.name)
file_path = os.path.join(directory_path, filebasename)
result.append(move_file(file.name, file_path))
try:
documents = load_file(file_path, sentence_size=sentence_size)
vector_store.create_vector_store(documents=documents)
except Exception as e:
logger.warning(f"failed to load file {file},{e}")
result = f"failed to load file {file},{e}"
else:
filebasename = os.path.basename(files.name)
file_path = os.path.join(directory_path, filebasename)
result = move_file(files.name, file_path)
try:
documents = load_file(file_path, sentence_size=sentence_size)
vector_store.create_vector_store(documents=documents)
except Exception as e:
logger.warning(f"failed to load file {files},{e}")
result = f"failed to load file {files},{e}"
except Exception as e:
logger.warning("failed to build vector_store")
result = f"failed to add file {str(files)} vector_store"
return result
def chat(query, chat_chatbot, system_prompt, temperature):
if len(chat_history.messages) > 0:
if type(chat_history.messages[0]) is SystemMessage:
if system_prompt:
chat_history.messages[0] = SystemMessage(content=system_prompt)
else:
if system_prompt:
chat_history.messages[0].insert(0, SystemMessage(content=system_prompt))
else:
if system_prompt:
chat_history.add_system_message(system_prompt)
chat_history.add_user_message(query)
chat = ChatOpenAI(temperature=temperature, model_name="gpt-3.5-turbo")
response = chat(chat_history.messages)
chat_history.add_ai_message(response.content)
chat_chatbot.append((query, response.content))
time.sleep(1)
return "", chat_chatbot
def kn_chat(know_ask_input, chat_chatbot, kv_num=4, min_score=0.3):
vector_store = VectorStore()
db = vector_store.db
if db is None:
chat_chatbot.append((know_ask_input, "请先加载或者上传知识库"))
return "", chat_chatbot
docs_and_scores = db.similarity_search_with_score(query=know_ask_input, k=kv_num)
kn_vector = ""
kn = []
# 假设docs_and_scores是db.similarity_search_with_score(query)返回的文档和分数列表
for i, (doc, score) in enumerate(docs_and_scores):
# 将文档的索引，内容，元数据和分数添加到字符串中，用换行符分隔
if score >= min_score:
kn_vector += f"<details><summary>来自文件{os.path.basename(doc.metadata['source'])}</summary>\n\n"
kn_vector += doc.page_content + "\n"
kn_vector += f"Document {i + 1}:\t"
kn_vector += "Score:"
kn_vector += str(f"{score:.3f}") + "\n"
kn_vector += "</details>"
kn.append(f"Document{i + 1}:doc.page_content\n")
query = f"""
Please response in Chinese,
I will ask you questions based on the following context:
- Start of Context -
{kn}
- End of Context-
My question is:“{know_ask_input}"
"""
chatkn_history.add_user_message(query)
chat = ChatOpenAI(model_name="gpt-3.5-turbo")
response = chat(chatkn_history.messages)
chatkn_history.add_ai_message(response.content)
chat_chatbot.append((know_ask_input, response.content + "\n" + kn_vector))
time.sleep(1)
return "", chat_chatbot
with gr.Blocks() as demo:
gr.Markdown("知识库测试")
with gr.Tab("普通聊天模式"):
with gr.Row():
with gr.Column(scale=0.7):
chatbot = gr.Chatbot(label="聊天记录")
ask_input = gr.Textbox(label="提问")
clear = gr.ClearButton([ask_input, chatbot])
with gr.Column(scale=0.3):
with gr.Accordion("参数"):
temperature = gr.Slider(step=0.01, minimum=0, maximum=1, label="temperature", value=0.7)
text_sysprompt_input = gr.Textbox(label="AI人设")
with gr.Tab("知识库问答"):
with gr.Row():
with gr.Column(scale=0.7):
chatbot_kn = gr.Chatbot(label="聊天记录")
know_ask_input = gr.Textbox(label="询问")
clear_kn = gr.ClearButton([know_ask_input, chatbot_kn])
with gr.Column(scale=0.3):
gr.Markdown("分数越高检索到的答案越少")
kv_num = gr.Slider(step=1, minimum=0, maximum=20, label="知识库检索数目", value=4)
min_score = gr.Slider(step=0.01, minimum=0, maximum=1, label="分数阈值", value=0.25)
sentence_size = gr.Slider(step=1, minimum=50, maximum=1000, label="分割长度", value=200)
docs_input = gr.File()
docx_text = gr.Textbox(label="通知栏")
vector_store_creat_button = gr.Button("新建或知识库")
vector_store_delete_button = gr.Button("删除数据库")
with gr.Accordion("Open for More!"):
gr.Markdown("Look at me...")
ask_input.submit(chat, inputs=[ask_input, chatbot, text_sysprompt_input, temperature], outputs=[ask_input, chatbot])
know_ask_input.submit(kn_chat, inputs=[know_ask_input, chatbot_kn, kv_num, min_score],
outputs=[know_ask_input, chatbot_kn])
vector_store_creat_button.click(save_file, inputs=[docs_input, sentence_size], outputs=docx_text)
vector_store_delete_button.click(vector_store.delete_vector_store, outputs=docx_text)
demo.launch(server_name='0.0.0.0', server_port=8888,share=True)