Add a header and footer to a PDF

A running header is one of those things that is trivial in a word processor and oddly hard once a document has become a PDF. The page is finished — its text is placed at fixed coordinates, and there is no reflow, no margin box and no concept of a header region to drop text into. Anything added has to be drawn on top, in a space the document has hopefully left empty.

That is what this does, and knowing it explains the one setting people wonder about. The margin control is not decoration: it is the distance in from the page edge at which the text is drawn, and because nothing on the page moves out of the way, it is the whole of your control over whether the header lands in white space or across the top line of a paragraph. Documents produced from a word processor almost always have room. Documents produced from a scan, or trimmed to their content, sometimes do not.

Six slots rather than one is the other deliberate decision. Real documents put different things in different places — a matter number top-left and a page count top-right, a confidentiality notice bottom-centre — and a tool with a single text box forces those to be done as three separate passes over the file, each one re-saving the document. Here they are one operation.

{page} and {pages} are filled in as the stamping happens, because their value is different on every page. Anything whose value is constant across the document — a date, a title, a reference — is put into the text before stamping begins, which is why there is a button that inserts today's date rather than a token that means it. The practical difference is that the date is fixed at the moment you make the file, which is what a dated document should want; a token evaluated when the file is opened would silently change what the document says.

The limitation to know about before you start is the font. The standard PDF fonts encode WinAnsi — Latin script, accented characters, most European punctuation — and nothing else. Devanagari, CJK, Cyrillic, Greek, Arabic and Hebrew are outside it. Drawing those needs a font embedded in the output file with the right glyphs and, for Indic script, correct shaping as well, and doing that badly produces text that is reordered or joined incorrectly rather than merely ugly. So this tool checks what you have typed and refuses characters it cannot draw, naming them, instead of accepting the job and writing a page of question marks.

Everything else about the document survives. The page tree is untouched, so bookmarks, links, annotations and form fields come through exactly as they were, and the text you add is real text rather than a picture of text.