marigen_api / custon_generic_loader.py
jameszokah's picture
Synced repo using 'sync_with_huggingface' Github Action
1897f56 verified
from typing import AsyncIterable, List
from langchain_core.document_loaders.blob_loaders import BlobLoader
from langchain_community.document_loaders.generic import GenericLoader
from langchain_core.documents import Document
from langchain_community.document_loaders.base import BaseBlobParser
# Extend the base GenericLoader class
class CustomGenericLoader(GenericLoader):
def __init__(self, blob_loader: BlobLoader, blob_parser: BaseBlobParser):
super().__init__(blob_loader, blob_parser)
async def async_load(self) -> AsyncIterable[Document]:
"""
Asynchronously loads and parses documents from blobs.
"""
async for blob in self.blob_loader.yield_blobs():
document = self.blob_parser.lazy_parse(blob)
yield document
async def lazy_load(self) -> AsyncIterable[Document]:
"""
A lazy asynchronous load method that can be overridden for more custom behavior.
"""
async for document in self.async_load():
yield document
async def load_all(self) -> List[Document]:
"""
Asynchronously loads all documents and returns them as a list.
"""
documents = []
async for document in self.async_load():
documents.append(document)
return documents