Data Object Layer for PDF data
To install: pip install pdfdol
Get a dict-like object to list and read the pdfs of a folder, as text:
>>> from pdfdol import PdfFilesReader
>>> from pdfdol.tests import get_test_pdf_folder
>>> folder_path = get_test_pdf_folder()
>>> pdfs = PdfFilesReader(folder_path)
>>> sorted(pdfs)
['sample_pdf_1', 'sample_pdf_2']
>>> assert pdfs['sample_pdf_2'] == [
... 'Page 1\nThis is a sample text for testing Python PDF tools.'
... ]
See that the values of a PdfFilesReader are lists of pages.
If you need strings (i.e. all the pages together) you can add a decoder like so:
fromdolimportadd_decoderpage_separator='---------------------'pdfs=add_decoder(pdfs, decoder=page_separator.join)If you need this at the level of the class, just do this:
fromdolimportadd_decoderpage_separator='---------------------'FilesReader=add_decoder(PdfFilesReader, decoder=page_separator.join)
# and thenpdfs=FilesReader(folder_path)
# ...If you need to concatinate a bunch of pdfs together, you can do so in many ways. Here's one:
fromdolimportFilesfrompdfdolimportconcat_pdfss=Files('~/Downloads/cosmograph_documentation_pdfs/')
concat_pdfs(s, key_order=sorted)pdfdol natively converts images, HTML, and Markdown to PDF.
For additional formats -- EPUB, MOBI, DOCX, ODT, DJVU, RTF, and
many more --
install Calibre, which provides the
ebook-convert command-line tool.
pdfdol does not depend on Calibre; it auto-detects the tool at runtime and uses it only for formats that have no built-in converter.
Installing ebook-convert:
| Platform | Command |
|---|---|
| macOS (Homebrew) | brew install --cask calibre |
| Debian / Ubuntu | sudo apt install calibre |
| Fedora / RHEL | sudo dnf install calibre |
| Linux servers (headless) | sudo -v && wget -nv -O- https://download.calibre-ebook.com/linux-installer.sh | sudo sh /dev/stdin |
The Linux CLI installer is self-contained (no GUI/X11 needed) and recommended for servers. See https://calibre-ebook.com/download for all options.
frompdfdolimportebook_convert_to_pdf, find_ebook_convert# Check whether Calibre is availableiffind_ebook_convert():
pdf_bytes=ebook_convert_to_pdf("book.epub")You can also go through the usual get_pdf entry point -- it will
automatically route to ebook-convert when it recognises the file extension:
frompdfdolimportget_pdfpdf_bytes=get_pdf("book.epub") # returns PDF bytesget_pdf("book.epub", egress="book.pdf") # saves to fileDRM-protected ebooks (Amazon KFX/AZW, Adobe-DRM EPUB) are encrypted, so neither Calibre nor pdfdol can read them; the DRM has to be removed first, where you are legally entitled to do so. Image-only sources (CBZ/CBR, DJVU, scanned PDF) do convert, but carry no text layer -- extracting their words needs OCR, which pdfdol does not do.
pdfdol maintains a format converter registry that maps file extensions to converter functions. You can register your own:
frompdfdolimportregister_format_converter, supported_extensionsdefmy_custom_converter(source):
"""source is a filepath (str) or raw bytes; must return PDF bytes."""
...
register_format_converter('.xyz', my_custom_converter)
# See everything that's currently supportedprint(supported_extensions())Example with a URL
pdf_data=get_pdf("https://pypi.org", src_kind="url")
print("Got PDF data of length:", len(pdf_data))Example with HTML content
html_content="<html><body><h1>Hello, PDF!</h1></body></html>"pdf_data=get_pdf(html_content, src_kind="html")
print("Got PDF data of length:", len(pdf_data))Example saving to file
filepath=get_pdf("https://pypi.org", egress="output.pdf", src_kind="url")
print("PDF saved to:", filepath)