Implement atransform_documents() for MarkdownifyTransformer (in langchain_community.document_transformers); a completed implementation is linked in this discussion

[rparkr]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

Feature: implement async document transformer for MarkdownifyTransformer

Currently, the MarkdownifyTransformer raises a NotImplementedError if a user calls the atransform_documents() method:

langchain/libs/community/langchain_community/document_transformers/markdownify.py

Lines 78 to 83 in 3b956b3

	async def atransform_documents(
	self,
	documents: Sequence[Document],
	**kwargs: Any,
	) -> Sequence[Document]:
	raise NotImplementedError

I am requesting an asynchronous implementation for document transformation to enable use with asynchronous frameworks or other async workflows.

Motivation

I would expect the atransform_documents() method to work, since it is defined in the docs and the description does not indicate that the method is not implemented.

Along with an enhanced user experience, providing this implementation would enable use with asynchronous frameworks or workflows.

Proposal (If applicable)

I have implemented a working solution here, which I have used for a personal project. I welcome any comments or updates to my solution. I also still need to write unit tests prior to submitting a PR for this.

https://github.com/rparkr/langchain/blob/rparkr/async-markdownify-transform/libs/community/langchain_community/document_transformers/markdownify.py#L107-L122