Emacs: turning parsed XML/HTML nodes back into text (encoding/printing/writing it out)

TL;DR you might want shr-dom-print.

I wanted to write this down because it's kind of non-obvious.

To parse XML / HTML text into nodes, there are libxml-parse-xml-region and libxml-parse-html-region functions which return nodes that can be manipulated with, say, functions from dom.el. But after you got the right parse tree and want to write it out back into a text with XML tags, neither xml.c nor xml.el offer an option for that.

xml.el has an xml-print function, but it seems to expect a parse tree that strictly follows XML (and fails on text nodes, for instance). This may be no fault of xml.el (my need is more with processing loose XML-ish text; so, basically, SGML).

The “parse tree → strings” action has a bunch of names, and in Elisp where everything is its own library it's not even consistent within the language:

• json.el calls it json-encode,
• json.c calls it json-serialize (on my system there's also undo-tree, projectile, markdown.el)
• yaml.el calls it yaml-encode,
• xml.el calls it xml-print,
• for Elisp s-expressions there are print or prin1 (but you can also use format)

…so it's kind of hard to search for.

The function that best matched my uses was shr-dom-print, which takes a parse tree accepted by dom.el and writes its XML representation into the current buffer. To write/print/encode a parse tree into valid XML, maybe shr-dom-to-xml also helps.