import os
from langchain.chains import RetrievalQA
from langchain.llms import OpenAI
from langchain.document_loaders import PyPDFLoader
from langchain.indexes import VectorstoreIndexCreator
from langchain.text_splitter import CharacterTextSplitter
from langchain.embeddings import OpenAIEmbeddings
from langchain.vectorstores import Chroma
import gradio as gr
import tempfile

#定义查询函数qa
def qa(file, openaikey, query, chain_type, k):
    os.environ["OPENAI_API_KEY"] = openaikey

    # load document 加载PDF文件
    loader = PyPDFLoader(file.name)
    documents = loader.load()
    # split the documents into chunks 将PDF文件分割成小块
    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
    texts = text_splitter.split_documents(documents)
    # select which embeddings we want to use 使用 OpenAI 的embeddings模型为每个文本块创建一个向量嵌入
    embeddings = OpenAIEmbeddings()
    # create the vectorestore to use as the index 创建一个向量存储VectorStore，用于后续的搜索。
    db = Chroma.from_documents(texts, embeddings)
    # expose this index in a retriever interface 使用这个向量存储VectorStore创建一个检索器retriever
    retriever = db.as_retriever(
        search_type="similarity", search_kwargs={"k": k})
    # create a chain to answer questions 然后使用这个检索器和 OpenAI 的模型创建一个问答链来回答问题。
    qa = RetrievalQA.from_chain_type(
        llm=OpenAI(), chain_type=chain_type, retriever=retriever, return_source_documents=True)
    result = qa({"query": query})
    print(result['result'])
    return result["result"]


iface = gr.Interface(
    fn=qa, 
    inputs=[
        gr.inputs.File(label="上传PDF"),
        gr.inputs.Textbox(label="OpenAI API Key"),
        gr.inputs.Textbox(label="你的问题"),
        #longchain的文档documents分析功能的不同类型，具体见https://python.langchain.com.cn/docs/modules/chains/document/的解释
        gr.inputs.Dropdown(choices=['stuff', 'map_reduce', "refine", "map_rerank"], label="Chain type"),
        gr.inputs.Slider(minimum=1, maximum=5, default=2, label="Number of relevant chunks"),
    ], 
    outputs="text",
    title="你可以问我关于你上传的PDF文件的任何信息！",
    description="1) 上传一个PDF文件. 2)输入你的OpenAI API key.这将产生费用  3) 输入问题然后点击运行."
)

iface.launch()