tag.js: in "text equation" there might not be target elements

[pkra]

In

https://github.com/AmerMathSoc/ams-xml-to-html/blob/4da69c39d51d15cad31a7becdaf39f59776b5bcc/lib/elements/tag.js#L24

the comment says to assume that each tag has a wrapping target.

However, mcom3734 actually does not.

[Surprisingly, this has not caused issues because instead, the parent is the disp-formula which has an ID and we do not preserve those IDs in HTML.]

[pkra]

A quick search showed no other unexpected parents of tag elements -- just tex-math, target, and exactly once disp-formula (for mcom3734's text equation without target).

[pkra]

It seems more transparent to avoid unwrapping target

texml-to-html/lib/elements/target.js

Lines 14 to 17 in 4da69c3

	// NOTE: inside a "text equation" we pass it through; the tag child will pick up the ID.
	if (xmlnode.parentNode.tagName === 'disp-formula' && xmlnode.parentNode.getAttribute('content-type') === 'text') {
	this.passThrough(htmlParentNode, xmlnode);
	return;

Then its output would naturally become the target (duh!) and tag would not need to add an ID it never had.

However, then we need to prepend target (or make a breaking change and let downstream re-arrange -- which seems unnecessary since we have tags on the left everywhere).

[pkra]

On a related note, I've filed AmerMathSoc/texml#231 to see if we can move the parens handling upstream.

[pkra]

However, then we need to prepend target (or make a breaking change and let downstream re-arrange -- which seems unnecessary since we have tags on the left everywhere).

Random thought: a tag without target would also need to be prepended. If we always prepend a tag, could it cause trouble when there's a target?

[pkra]

I've filed AmerMathSoc/texml#232 in which case we get to simplify this in the future.