src/doc_qa_1.py

from langchain.vectorstores import FAISS
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.text_splitter import CharacterTextSplitter
from langchain.docstore.document import Document
from transformers import pipeline
from langchain.chains.question_answering import load_qa_chain
import os

# Step 1: Load QA pipeline (don't wrap in HuggingFacePipeline)
embeddings = HuggingFaceEmbeddings(model_name="intfloat/multilingual-e5-small")
qa_pipeline = pipeline("question-answering", model="deepset/xlm-roberta-base-squad2")
multi_directory_path=r'tmp/'

def docs_vector_index():
    from langchain.document_loaders import DirectoryLoader
    # Define a directory path
    directory_path = r"C:\Users\savni\PycharmProjects\DocsSearchEngine\tmp"

    # Create the DirectoryLoader, specifying loaders for each file type
    loader = DirectoryLoader(
        directory_path,
        glob="**/*",  # This pattern loads all files; modify as needed

    )
    docs = loader.load()

    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=1024, chunk_overlap=100, separators=[" ", ",", "\n", "."]
    )
    print(docs)
    docs_chunks = text_splitter.split_documents(docs)

    print(f"docs_chunks length: {len(docs_chunks)}")
    print('********************docs_chunks',docs_chunks)
    if len(docs_chunks)>0:
        db = FAISS.from_documents(docs_chunks, embeddings)
        return db
    else:
        return ''



def run_custom_qa(question, retrieved_docs):
    context = " ".join([doc.page_content for doc in retrieved_docs])
    output = qa_pipeline(question=question, context=context)
    return output #output["answer"]

# # Step 6: Ask question
# question = "東京大学はいつ設立されましたか？"
# relevant_docs = retriever.get_relevant_documents(question)
# answer = run_custom_qa(question, relevant_docs)
#
# print("Answer:", answer)

def doc_qa(query, db):
    print("*************************custom qa doc_qa",query)
    retriever = db.as_retriever()
    relevant_docs = retriever.get_relevant_documents(query)
    response=run_custom_qa(query, relevant_docs)
    print('response', response)
    return response