Convert HTML to PDF

This project intends to replace wkhtmltopdf with a similar executable to convert HTML pages into PDF files.

wkhtmltopdf is a convenient and fast tool for converting files to PDF. Unfortunately, it uses the discontinued QtWeb rendering engine. There are quite a few JavaScript and CSS features which it cannot handle.

This tool is a simple NodeJS wrapper script that evokes a headless chromium, to use it's "print-to-pdf" feature to render the HTML page as a PDF.

Usage

This executable is meant for streaming, so it reads the raw HTML from STDIN and writes the PDF file to STDOUT:

htmltopdf -i /path/to/input.html -o /path/to/output.pdf

Commandline Flags

Options:
  --help           Show help                                           [boolean]
  --version        Show version number                                 [boolean]
  --landscape, -l                                     [boolean] [default: false]
  --pagesize, -s
       [choices: "Letter", "Legal", "Tabloid", "Ledger", "A0", "A1", "A2", "A3",
                                           "A4", "A5", "A6"] [default: "Letter"]
  --footer, -f     File with HTML snippet to use as the footer.         [string]
  --input, -i      File to use as HTML input (instead of STDIN).        [string]
  --output, -o     File to write PDF to (instead of STDOUT).            [string]

Future Improvements

Refactor out temporary files

Currently this program dumps the HTML into a temporary file, before opening it in chromium. Ideally, no files would be created and the HTML text would be passed directly to chromium via page.setContent(html); however, at this time setContent doesn't trigger any lifecycle events, so it's not possible to know when the page is done being loaded (and ready to print).

This may be splitting hairs, as our server stores temporary files in tmpfs, which is entirely in RAM anyway.

Replace with an online service

Every time this executable runs, it has to spin up a fresh instance of chromium, and free it once it's finished. Considering that rendering PDFs is a core part of the Spark application, it would be better to have an always-running, microservice version of this app that keeps chromium running. It could then accepts jobs from Spark via some interface.