confluence_qa.py

from langchain.document_loaders import ConfluenceLoader
from langchain.text_splitter import CharacterTextSplitter, TokenTextSplitter,RecursiveCharacterTextSplitter,SentenceTransformersTokenTextSplitter
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM,pipeline
from langchain.llms.huggingface_pipeline import HuggingFacePipeline
from langchain.prompts import PromptTemplate
from langchain.chains import RetrievalQA
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import Chroma

class ConfluenceQA:
    def init_embeddings(self) -> None:
        self.embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
    
    def define_model(self) -> None:
        tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-large")
        model = AutoModelForSeq2SeqLM.from_pretrained("google/flan-t5-large")
        pipe = pipeline("text2text-generation", model=model, tokenizer=tokenizer, max_new_tokens=1024, truncation=True)
        self.llm = HuggingFacePipeline(pipeline = pipe,model_kwargs={"temperature": 0, "max_length": 1024})

    def store_in_vector_db(self) -> None:
        persist_directory = self.config.get("persist_directory",None)
        confluence_url = self.config.get("confluence_url",None)
        username = self.config.get("username",None)
        api_key = self.config.get("api_key",None)
        space_key = self.config.get("space_key",None)  
        include_attachment = self.config.get("include_attachment", True)      
        loader = ConfluenceLoader(
            url=confluence_url, username=username, api_key=api_key
        )
        documents = loader.load(include_attachments=include_attachment, limit=100, space_key=space_key)
        # text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=10)
        # documents = text_splitter.split_documents(documents)
        # print(documents)
        text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=10)
        documents = text_splitter.split_documents(documents)
        # text_splitter = SentenceTransformersTokenTextSplitter(chunk_overlap=10)
        # documents = text_splitter.split_documents(documents)
        # print(documents)
        text_splitter = TokenTextSplitter(chunk_size=1000, chunk_overlap=10, encoding_name="cl100k_base")  # This the encoding for text-embedding-ada-002
        documents = text_splitter.split_documents(documents)
        self.db = Chroma.from_documents(documents, self.embeddings)


    def retrieve_qa_chain(self) -> None:
        template = """Use the following pieces of context to answer the question at the end. If you don't know the answer, just say that you don't know, don't try to make up an answer.
        {context}
        Question: {question}
        Helpful Answer:"""
        QA_CHAIN_PROMPT = PromptTemplate(
            template=template, input_variables=["context", "question"]
        )
        chain_type_kwargs = {"prompt": QA_CHAIN_PROMPT}
        self.qa = RetrievalQA.from_chain_type(llm=self.llm, chain_type="stuff", retriever=self.db.as_retriever(search_kwargs={"k":4}), chain_type_kwargs=chain_type_kwargs)

    def __init__(self,config:dict = {}) -> None:
        self.db=None
        self.embeddings=None
        self.llm=None
        self.config=config
        self.qa=None

    def qa_bot(self, query:str):
        result = self.qa.run(query)
        return result