private-gpt/ingest.py

128 lines
4.5 KiB
Python
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

import os
import glob
from typing import List
from dotenv import load_dotenv
from langchain.document_loaders import (
CSVLoader,
EverNoteLoader,
PDFMinerLoader,
TextLoader,
UnstructuredEmailLoader,
UnstructuredEPubLoader,
UnstructuredHTMLLoader,
UnstructuredMarkdownLoader,
UnstructuredODTLoader,
UnstructuredPowerPointLoader,
UnstructuredWordDocumentLoader,
)
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.docstore.document import Document
from constants import CHROMA_SETTINGS
load_dotenv()
# Load environment variables
persist_directory = os.environ.get('PERSIST_DIRECTORY')
source_directory = os.environ.get('SOURCE_DIRECTORY', 'source_documents')
embeddings_model_name = os.environ.get('EMBEDDINGS_MODEL_NAME')
chunk_size = 500
chunk_overlap = 50
# Map file extensions to document loaders and their arguments
LOADER_MAPPING = {
".csv": (CSVLoader, {}),
# ".docx": (Docx2txtLoader, {}),
".docx": (UnstructuredWordDocumentLoader, {}),
".enex": (EverNoteLoader, {}),
".eml": (UnstructuredEmailLoader, {}),
".epub": (UnstructuredEPubLoader, {}),
".html": (UnstructuredHTMLLoader, {}),
".md": (UnstructuredMarkdownLoader, {}),
".odt": (UnstructuredODTLoader, {}),
".pdf": (PDFMinerLoader, {}),
".pptx": (UnstructuredPowerPointLoader, {}),
".txt": (TextLoader, {"encoding": "utf8"}),
# Add more mappings for other file extensions and loaders as needed
}
load_dotenv()
def load_single_document(file_path: str) -> Document:
ext = "." + file_path.rsplit(".", 1)[-1]
if ext in LOADER_MAPPING:
loader_class, loader_args = LOADER_MAPPING[ext]
loader = loader_class(file_path, **loader_args)
return loader.load()[0]
raise ValueError(f"Unsupported file extension '{ext}'")
def load_documents(source_dir: str, ignored_files: List[str] = []) -> List[Document]:
"""
Loads all documents from the source documents directory, ignoring specified files
"""
all_files = []
for ext in LOADER_MAPPING:
all_files.extend(
glob.glob(os.path.join(source_dir, f"**/*{ext}"), recursive=True)
)
filtered_files = [file_path for file_path in all_files if file_path not in ignored_files]
return [load_single_document(file_path) for file_path in filtered_files]
def process_documents(ignored_files: List[str] = []) -> List[Document]:
"""
Load documents and split in chunks
"""
print(f"Loading documents from {source_directory}")
documents = load_documents(source_directory, ignored_files)
if not documents:
print("No new documents to load")
exit(0)
print(f"Loaded {len(documents)} new documents from {source_directory}")
text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
texts = text_splitter.split_documents(documents)
print(f"Split into {len(texts)} chunks of text (max. 500 tokens each)")
return texts
def does_vectorstore_exist(persist_directory: str) -> bool:
"""
Checks if vectorstore exists
"""
if os.path.exists(os.path.join(persist_directory, 'index')):
if os.path.exists(os.path.join(persist_directory, 'chroma-collections.parquet')) and os.path.exists(os.path.join(persist_directory, 'chroma-embeddings.parquet')):
list_index_files = glob.glob(os.path.join(persist_directory, 'index/*.bin'))
list_index_files += glob.glob(os.path.join(persist_directory, 'index/*.pkl'))
if len(list_index_files) == 4:
return True
return False
def main():
# Create embeddings
embeddings = HuggingFaceEmbeddings(model_name=embeddings_model_name)
if does_vectorstore_exist(persist_directory):
# Update and store locally vectorstore
print(f"Appending to existing vectorstore at {persist_directory}")
db = Chroma(persist_directory=persist_directory, embedding_function=embeddings, client_settings=CHROMA_SETTINGS)
collection = db.get()
texts = process_documents([metadata['source'] for metadata in collection['metadatas']])
db.add_documents(texts)
else:
# Create and store locally vectorstore
print("Creating new vectorstore")
texts = process_documents()
db = Chroma.from_documents(texts, embeddings, persist_directory=persist_directory, client_settings=CHROMA_SETTINGS)
db.persist()
db = None
if __name__ == "__main__":
main()