File size: 6,113 Bytes
5cbeaec |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 |
import os
import time
import gradio as gr
import openai
from langdetect import detect
from gtts import gTTS
from pdfminer.high_level import extract_text
#any vector server should work, trying pinecone first
import pinecone
#langchain part
import spacy
import tiktoken
from langchain.llms import OpenAI
from langchain.text_splitter import SpacyTextSplitter
from langchain.document_loaders import TextLoader
from langchain.document_loaders import DirectoryLoader
from langchain.indexes import VectorstoreIndexCreator
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import Pinecone
openai.api_key = os.environ['OPENAI_API_KEY']
pinecone_key = os.environ['PINECONE_API_KEY']
pinecone_environment='us-west1-gcp-free'
user_db = {os.environ['username1']: os.environ['password1']}
messages = [{"role": "system", "content": 'You are a helpful assistant.'}]
#load up spacy
nlp = spacy.load("en_core_web_sm")
def init_pinecone():
pinecone.init(api_key=pinecone_key, environment=pinecone_environment)
return
def process_file(index_name, dir):
init_pinecone()
#using openai embedding hence dim = 1536
pinecone.create_index(index_name, dimension=1536, metric="cosine")
#time.sleep(5)
embeddings = OpenAIEmbeddings(openai_api_key=os.environ['OPENAI_API_KEY'])
splter = SpacyTextSplitter(chunk_size=1000,chunk_overlap=200)
for doc in dir:
loader = TextLoader(doc.name , encoding='utf8')
content = loader.load()
split_text = splter.split_documents(content)
for text in split_text:
Pinecone.from_documents([text], embeddings, index_name=index_name)
#pipeline='zh_core_web_sm'
return
def list_pinecone():
init_pinecone()
return pinecone.list_indexes()
def show_pinecone(index_name):
init_pinecone()
#return pinecone.describe_index(index_name)
index = pinecone.Index(index_name)
stats = index.describe_index_stats()
return stats
def delete_pinecone(index_name):
init_pinecone()
pinecone.delete_index(index_name)
return
def roleChoice(role):
global messages
messages = [{"role": "system", "content": role}]
return "role:" + role
def talk2file(index_name, text):
global messages
#same as filesearch
init_pinecone()
embeddings = OpenAIEmbeddings(openai_api_key=os.environ['OPENAI_API_KEY'])
docsearch = Pinecone.from_existing_index(index_name, embeddings)
docs = docsearch.similarity_search(text)
prompt = text + ", 根据以下文本: \n\n" + docs[0].page_content
messages.append({"role": "user", "content": prompt})
response = openai.ChatCompletion.create(model="gpt-3.5-turbo", messages=messages)
system_message = response["choices"][0]["message"]
messages.append(system_message)
chats = ""
for msg in messages:
if msg['role'] != 'system':
chats += msg['role'] + ": " + msg['content'] + "\n\n"
return chats
def fileSearch(index_name, prompt):
global messages
init_pinecone()
embeddings = OpenAIEmbeddings(openai_api_key=os.environ['OPENAI_API_KEY'])
docsearch = Pinecone.from_existing_index(index_name, embeddings)
docs = docsearch.similarity_search(prompt)
return "Content in file: \n\n" + docs[0].page_content + "\n\n"
def clear():
global messages
messages = [{"role": "system", "content": 'You are a helpful technology assistant.'}]
return
def show():
global messages
chats = ""
for msg in messages:
if msg['role'] != 'system':
chats += msg['role'] + ": " + msg['content'] + "\n\n"
return chats
with gr.Blocks() as chatHistory:
gr.Markdown("Click the Clear button below to remove all the chat history.")
clear_btn = gr.Button("Clear")
clear_btn.click(fn=clear, inputs=None, outputs=None, queue=False)
gr.Markdown("Click the Display button below to show all the chat history.")
show_out = gr.Textbox()
show_btn = gr.Button("Display")
show_btn.click(fn=show, inputs=None, outputs=show_out, queue=False)
#pinecone tools
with gr.Blocks() as pinecone_tools:
pinecone_list = gr.Textbox()
list = gr.Button(value="List")
list.click(fn=list_pinecone, inputs=None, outputs=pinecone_list, queue=False)
pinecone_delete_name = gr.Textbox()
delete = gr.Button(value="Delete")
delete.click(fn=delete_pinecone, inputs=pinecone_delete_name, outputs=None, queue=False)
pinecone_show_name = gr.Textbox()
pinecone_info = gr.Textbox()
show = gr.Button(value="Show")
show.click(fn=show_pinecone, inputs=pinecone_show_name, outputs=pinecone_info, queue=False)
role = gr.Interface(fn=roleChoice, inputs="text", outputs="text", description = "Choose your GPT roles, e.g. You are a helpful technology assistant. 你是一位 IT 架构师。 你是一位开发者关系顾问。你是一位机器学习工程师。你是一位高级 C++ 开发人员 ")
text = gr.Interface(fn=talk2file, inputs=["text", "text"], outputs="text")
vector_server = gr.Interface(fn=process_file, inputs=["text", gr.inputs.File(file_count="directory")], outputs="text")
#audio = gr.Interface(fn=audioGPT, inputs=gr.Audio(source="microphone", type="filepath"), outputs="text")
#siri = gr.Interface(fn=siriGPT, inputs=gr.Audio(source="microphone", type="filepath"), outputs = "audio")
file = gr.Interface(fn=fileSearch, inputs=["text", "text"], outputs="text", description = "Enter file name and prompt")
demo = gr.TabbedInterface([role, text, file, vector_server, pinecone_tools, chatHistory], [ "roleChoice", "Talk2File", "FileSearch", "VectorServer", "PineconeTools", "ChatHistory"])
if __name__ == "__main__":
demo.launch(enable_queue=False, auth=lambda u, p: user_db.get(u) == p,
auth_message="This is not designed to be used publicly as it links to a personal openAI API. However, you can copy my code and create your own multi-functional ChatGPT with your unique ID and password by utilizing the 'Repository secrets' feature in huggingface.")
#demo.launch() |