pdfpipe-python

Official Python SDK for PDFPipe. Turn HTML or a URL into a PDF with one call, backed by a real browser engine.

pip install pdfpipe-python

Quick start

from pdfpipe import PDFPipe

pdfpipe = PDFPipe(api_key="pp_live_...")

# From HTML
pdf = pdfpipe.from_html("<h1>Invoice #4012</h1>", {"format": "A4"})
with open("invoice.pdf", "wb") as f:
    f.write(pdf)

# From a URL
pdf = pdfpipe.from_url("https://example.com")
with open("example.pdf", "wb") as f:
    f.write(pdf)

Get an API key at pdfpipe.xyz. Keys look like pp_live_....

API

PDFPipe(api_key, base_url="https://api.pdfpipe.xyz", session=None)

from_html(html, options=None) -> bytes

from_url(url, options=None) -> bytes

Both return the raw PDF bytes. options is a dict, all keys optional:

Running headers and footers

header_html and footer_html render on every page. Five class names are substituted per page as the document is laid out:

pdf = pdfpipe.from_html(
    html,
    {
        "header_html": '<div style="font-size:9px;width:100%;text-align:center">Q3 Report</div>',
        "footer_html": '<div style="font-size:9px;width:100%;text-align:center">'
                       'Page <span class="pageNumber"></span> of <span class="totalPages"></span></div>',
    },
)

Inline your styles. The header and footer are laid out separately from the page body, so a stylesheet in your HTML does not reach them, and they inherit a very small default font size.

If you leave margin at its 1cm default, the top or bottom margin widens to 2cm automatically so the header or footer has room. Set margin yourself if you want different spacing.

The other three

tabular_nums forces font-variant-numeric: tabular-nums, so digits share one width. Use it whenever a column of numbers has to line up, which is most invoices and statements.

deduplicate_images merges identical image XObjects after rendering. It only pays off when the same image repeats across pages, a logo in a running header being the usual case, and it costs a little post-processing time.

pdf_a writes PDF/A-1b XMP metadata on a best-effort basis. It marks the file for archival tooling but does not guarantee full conformance, which would also require tagged structure and an embedded ICC profile.

Errors

Non-2xx responses raise PDFPipeError with a .status attribute and a clear message.

from pdfpipe import PDFPipe, PDFPipeError

try:
    pdf = pdfpipe.from_url("https://example.com", {"timeout_ms": 45000})
except PDFPipeError as e:
    print(e.status, str(e))

License

MIT