[community] Propose PDFRouterParser and Loader

[pprados]

Load PDFs using different parsers based on the metadata of the PDF or the body of the first page.

The routes are defined as a list of tuples, where each tuple contains the name, a dictionary of metadata and regex pattern and the parser to use. The special key "page1" is to search in the first page with a regexp. Use the route in the correct order, as the first matching route is used.a Add a default route ("default", {}, parser) at the end to catch all PDFs. This code is similar to MimeTypeBasedParser, but on the content of the PDF file.

Sample:

    from langchain_community.document_loaders import PyPDFLoader
    from langchain_community.document_loaders.parsers.pdf import PyMuPDFParser
    from langchain_community.document_loaders.parsers.pdf import PyPDFium2Parser
    from langchain_community.document_loaders.parsers import PDFPlumberParser
    routes = [
        # Name, keys with regex, parser
        ("Microsoft", {"producer": "Microsoft", "creator": "Microsoft"},
        PyMuPDFParser()),
        ("LibreOffice", {"producer": "LibreOffice", }, PDFPlumberParser()),
        ("Xdvipdfmx", {"producer": "xdvipdfmx.*", "page1":"Hello"}, PDFPlumberParser()),
        ("defautl", {}, PyPDFium2Parser())
    ]
    loader = PDFRouterLoader(filename, routes)
    loader.load()

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Skipped Deployment

Name	Status	Preview	Comments	Updated (UTC)
langchain	⬜️ Ignored (Inspect)	Visit Preview		Apr 17, 2025 1:55pm