ingest.py

import os
import logging
import streamlit as st
from langchain_community.document_loaders import PDFMinerLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import FAISS

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

def create_faiss_index():
    documents = []
    docs_dir = "docs"

    if not os.path.exists(docs_dir):
        st.error(f"The directory '{docs_dir}' does not exist.")
        return

    for root, dirs, files in os.walk(docs_dir):
        for file in files:
            if file.endswith(".pdf"):
                file_path = os.path.join(root, file)
                st.info(f"Loading document: {file_path}")
                try:
                    loader = PDFMinerLoader(file_path)
                    documents.extend(loader.load())
                except Exception as e:
                    st.error(f"Error loading {file_path}: {e}")

    if not documents:
        st.error("No documents were loaded. Check the 'docs' directory and file paths.")
        return

    st.info(f"Loaded {len(documents)} documents.")

    text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
    texts = text_splitter.split_documents(documents)

    if not texts:
        st.error("No text chunks were created. Check the text splitting process.")
        return

    st.info(f"Created {len(texts)} text chunks.")

    try:
        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
    except Exception as e:
        st.error(f"Failed to initialize embeddings: {e}")
        return

    try:
        db = FAISS.from_documents(texts, embeddings)
        st.info(f"Created FAISS index with {len(texts)} vectors")
    except Exception as e:
        st.error(f"Failed to create FAISS index: {e}")
        return

    index_dir = "faiss_index"
    if not os.path.exists(index_dir):
        os.makedirs(index_dir)

    try:
        db.save_local(index_dir)
        st.success(f"FAISS index successfully saved to {index_dir}")
        index_path = os.path.join(index_dir, "index.faiss")
        st.info(f"Index file size: {os.path.getsize(index_path)} bytes")
        st.info(f"Index file permissions: {oct(os.stat(index_path).st_mode)[-3:]}")
    except Exception as e:
        st.error(f"Failed to save FAISS index: {e}")

if __name__ == "__main__":
    create_faiss_index()