langchain-pdfluent

A LangChain document loader for PDFluent. It reads a PDF with PDFluent's structured extraction and hands LangChain one Document per paragraph, heading, list item or table, in reading order, with the page and box each came from in its metadata so an answer can cite its source.

Install

pip install langchain-pdfluent

It needs Python 3.9 or later and pdfluent 1.1.0 or later, which pip pulls in automatically. It also depends on langchain-core (MIT).

Use

Blocks mode

One document per block, a table as one document, in reading order:

from langchain_pdfluent import PDFluentLoader

loader = PDFluentLoader("report.pdf")
docs = loader.load()

print(len(docs), "chunks")
print(docs[0].page_content)
print(docs[0].metadata["page"])
print(docs[0].metadata["bbox"])
print(docs[0].metadata["kind"])

Markdown mode

One document per page, as markdown:

from langchain_pdfluent import PDFluentLoader

loader = PDFluentLoader("report.pdf", mode="markdown")
pages = loader.load()

print(len(pages), "pages")
print(pages[0].page_content[:200])
print(pages[0].metadata["page"])
print(pages[1].metadata["page"])
print(pages[0].metadata["kind"])

Options:

Metadata

bbox, toc_items and tables are lists. A vector store that only accepts scalar metadata, such as Chroma, needs filter_complex_metadata from langchain_community over the documents first.

Licence

AGPL-3.0 or a commercial licence, at your option; see LICENSE and LICENSE-COMMERCIAL. The commercial licence is sold yearly and self-service at https://pdfluent.com/sdk/pricing.