Spaces:
Sleeping
Sleeping
File size: 1,823 Bytes
0e17e2d |
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 |
import os
import tempfile
import streamlit as st
from langchain.embeddings import HuggingFaceEmbeddings
from langchain_community.document_loaders import (
Docx2txtLoader,
PyPDFLoader,
TextLoader,
UnstructuredEPubLoader,
)
from langchain_community.vectorstores import DocArrayInMemorySearch
from langchain_text_splitters import RecursiveCharacterTextSplitter
EMBEDDING_MODEL_NAME = "all-MiniLM-L6-v2"
@st.cache_resource(ttl="1h")
def configure_retriever(files):
# Read documents
docs = []
temp_dir = tempfile.TemporaryDirectory()
for file in files:
temp_filepath = os.path.join(temp_dir.name, file.name)
with open(temp_filepath, "wb") as f:
f.write(file.getvalue())
_, extension = os.path.splitext(temp_filepath)
# Load the file using the appropriate loader
if extension == ".pdf":
loader = PyPDFLoader(temp_filepath)
elif extension == ".docx":
loader = Docx2txtLoader(temp_filepath)
elif extension == ".txt":
loader = TextLoader(temp_filepath)
elif extension == ".epub":
loader = UnstructuredEPubLoader(temp_filepath)
else:
st.write("This document format is not supported!")
return None
docs.extend(loader.load())
# Split documents
text_splitter = RecursiveCharacterTextSplitter(chunk_size=1500, chunk_overlap=200)
splits = text_splitter.split_documents(docs)
# Create embeddings and store in vectordb
embeddings = HuggingFaceEmbeddings(model_name=EMBEDDING_MODEL_NAME)
vectordb = DocArrayInMemorySearch.from_documents(splits, embeddings)
# Define retriever
retriever = vectordb.as_retriever(
search_type="mmr", search_kwargs={"k": 2, "fetch_k": 4}
)
return retriever
|