app.py

# -*-coding:utf-8 -*-
import gradio as gr
import os
import json
from glob import glob
import requests
from langchain import FAISS
from langchain.embeddings import CohereEmbeddings, OpenAIEmbeddings
from langchain import VectorDBQA
from langchain.chat_models import ChatOpenAI
from prompts import MyTemplate
from build_index.run import process_files
from langchain.prompts.chat import (
    ChatPromptTemplate,
    SystemMessagePromptTemplate,
    HumanMessagePromptTemplate,
)
from langchain.prompts import PromptTemplate
from langchain.chains.llm import LLMChain
from langchain.chains.combine_documents.map_reduce import MapReduceDocumentsChain
from langchain.chains import QAGenerationChain
from langchain.chains.combine_documents.stuff import StuffDocumentsChain

# Streaming endpoint
API_URL = "https://api.openai.com/v1/chat/completions"
cohere_key = '5IRbILAbjTI0VcqTsktBfKsr13Lych9iBAFbLpkj'
faiss_store = './output/'
global doc_search


def process(files, openai_api_key, max_tokens, n_sample):
    """
    对文档处理进行摘要，构建问题，构建文档索引
    """
    os.environ['OPENAI_API_KEY'] = openai_api_key
    print('Displaying uploading files ')
    print(glob('/tmp/*'))
    docs = process_files([i.name for i in files], 'openai', max_tokens)
    print('Display Faiss index')
    print(glob('./output/*'))
    question = get_question(docs, openai_api_key, max_tokens, n_sample)
    summary = get_summary(docs, openai_api_key, max_tokens, n_sample)
    return question, summary


def get_question(docs, openai_api_key, max_tokens, n_sample=5):
    q_list = []
    llm = ChatOpenAI(openai_api_key=openai_api_key, max_tokens=max_tokens, temperature=0)
    # 基于文档进行QA生成
    prompt = ChatPromptTemplate.from_messages(
        [
            SystemMessagePromptTemplate.from_template(MyTemplate['qa_sys_template']),
            HumanMessagePromptTemplate.from_template(MyTemplate['qa_user_template']),
        ]
    )
    chain = QAGenerationChain.from_llm(llm, prompt=prompt)
    print('Generating Question from template')
    for i in range(n_sample):
        qa = chain.run(docs[i].page_content)[0]
        print(qa)
        q_list.append(f"问题{i + 1}: {qa['question']}")
    return '\n'.join(q_list)


def get_summary(docs, openai_api_key, max_tokens, n_sample=5, verbose=None):
    llm = ChatOpenAI(openai_api_key=openai_api_key, max_tokens=max_tokens)
    # chain = load_summarize_chain(llm, chain_type="map_reduce")
    # summary = chain.run(docs[:n_sample])
    print('Generating Summary from template')
    map_prompt = PromptTemplate(template=MyTemplate['summary_template'], input_variables=["text"])
    combine_prompt = PromptTemplate(template=MyTemplate['summary_template'], input_variables=["text"])
    map_chain = LLMChain(llm=llm, prompt=map_prompt, verbose=verbose)
    reduce_chain = LLMChain(llm=llm, prompt=combine_prompt, verbose=verbose)
    combine_document_chain = StuffDocumentsChain(
        llm_chain=reduce_chain,
        document_variable_name='text',
        verbose=verbose,
    )
    chain = MapReduceDocumentsChain(
        llm_chain=map_chain,
        combine_document_chain=combine_document_chain,
        document_variable_name='text',
        collapse_document_chain=None,
        verbose=verbose
    )
    summary = chain.run(docs[:n_sample])
    print(summary)
    return summary


def predict(inputs, openai_api_key, max_tokens, chat_counter, chatbot=[], history=[]):
    print(f"chat_counter - {chat_counter}")
    print(f'Histroy - {history}')  # History: Original Input and Output in flatten list
    print(f'chatbot - {chatbot}')  # Chat Bot： 上一轮回复的[[user, AI]]

    history.append(inputs)
    if doc_search is None:
        print(f'loading faiss store from {faiss_store}')
        docsearch = FAISS.load_local(faiss_store, OpenAIEmbeddings(openai_api_key=openai_api_key))
    else:
        print('faiss already loaded')
    # 构建模板
    llm = ChatOpenAI(openai_api_key=openai_api_key, max_tokens=max_tokens)
    messages_combine = [
        SystemMessagePromptTemplate.from_template(MyTemplate['chat_combine_template']),
        HumanMessagePromptTemplate.from_template("{question}")
    ]
    p_chat_combine = ChatPromptTemplate.from_messages(messages_combine)
    messages_reduce = [
        SystemMessagePromptTemplate.from_template(MyTemplate['chat_reduce_template']),
        HumanMessagePromptTemplate.from_template("{question}")
    ]
    p_chat_reduce = ChatPromptTemplate.from_messages(messages_reduce)
    chain = VectorDBQA.from_chain_type(llm=llm, chain_type="map_reduce", vectorstore=docsearch,
                                       k=4,
                                       chain_type_kwargs={"question_prompt": p_chat_reduce,
                                                          "combine_prompt": p_chat_combine}
                                       )
    result = chain({"query": inputs})
    print(result)
    result = result['result']
    # 生成返回值
    history.append(result)
    chat = [(history[i], history[i + 1]) for i in range(0, len(history) - 1, 2)]
    chat_counter += 1
    yield chat, history, chat_counter


def reset_textbox():
    return gr.update(value='')


with gr.Blocks(css="""#col_container {width: 1000px; margin-left: auto; margin-right: auto;}
                #chatbot {height: 520px; overflow: auto;}""") as demo:
    gr.HTML("""<h1 align="center">🚀Smart Doc Reader🚀</h1>""")
    with gr.Column(elem_id="col_container"):
        openai_api_key = gr.Textbox(type='password', label="输入 API Key")

        with gr.Accordion("Parameters", open=True):
            with gr.Row():
                max_tokens = gr.Slider(minimum=100, maximum=2000, value=1000, step=100, interactive=True,
                                       label="字数")
                chat_counter = gr.Number(value=0, precision=0, label='对话轮数')
                n_sample = gr.Slider(minimum=3, maximum=5, value=3, step=1, interactive=True,
                                     label="问题数")

        # 输入文件，进行摘要和问题生成
        with gr.Row():
            with gr.Column():
                files = gr.File(file_count="multiple", file_types=[".pdf"], label='上传pdf文件')
                run = gr.Button('文档内容解读')

            with gr.Column():
                summary = gr.Textbox(type='text', label="一眼看尽 - 文档概览")
                question = gr.Textbox(type='text', label='推荐问题 - 问别的也行哟')

        chatbot = gr.Chatbot(elem_id='chatbot')
        inputs = gr.Textbox(placeholder="这篇文档是关于什么的", label="针对文档你有哪些问题？")
        state = gr.State([])

        with gr.Row():
            clear = gr.Button("清空")
            start = gr.Button("提问")

    run.click(process, [files, openai_api_key, max_tokens, n_sample], [question, summary])
    inputs.submit(predict,
                  [inputs, openai_api_key, max_tokens, chat_counter, chatbot, state],
                  [chatbot, state, chat_counter], )
    start.click(predict,
                [inputs, openai_api_key, max_tokens, chat_counter, chatbot, state],
                [chatbot, state, chat_counter], )

    # 每次对话结束都重置对话
    clear.click(reset_textbox, [], [inputs], queue=False)
    inputs.submit(reset_textbox, [], [inputs])
    demo.queue().launch(debug=True)