One HTML file, printed as one page or two

The host of the buoy needed a sheet: what the node is, what it draws, what it needs from the buoy, on one page, with a fuller pre-read behind it. One HTML file does all three. On the web it is the full page. A script in the head adds a class to the root when the URL has ?sheet, and that class hides the second half and the link to the PDF. Headless Chrome prints the URL twice, once with the query and once without. poppler then reports the page count and size and lists the embedded fonts, so a build that spills to three pages or ships a system font fails in the terminal. The same script renders both PDFs to PNG and screenshots the web page at desktop and phone width.

The footers are page margin boxes, which Chrome has had since 131: an @page rule with bottom-left and bottom-right content, the right one printing counter(page) of counter(pages) in the mono face at 8px. Nothing in the body knows what page it is on.

Keeping the full version on two pages took three passes. Print sizes, tighter table padding, break-inside avoid on the table and the cards. Then a blank third page that no content explained. Chrome adds one for a trailing margin at the end of the document, so in print the last child of main and of every list has no bottom margin, and the root and body have height auto.

The blank page had nothing on it, so nothing in the CSS pointed at it.