A software bug in modules with OSIS Glosses

I just tried an experiment using diatheke on a module that I'm currently developing.

Here is a verse as compiled into the module using imp2vs

$$$Numbers 13:1
<w gloss="(13:2)">​</w> Et l’Éternel parla à Moïse, en disant:

NB. The w element was wrapped around a zero width space, just for the sake of the experiment.

This diatheke command was used on the module:

diatheke -b FreMartin1707 -o g -k Num 13:1 >gloss.test.log

Examination of the log file has this:

Numbers 13:1: ​ <2)> Et l’Éternel parla à Moïse, en disant:

So we see that diatheke wraps a gloss between < and >

Yet clearly there's a serious bug in how the gloss text is processed.

Instead of "(13:2)" all I got was the portion AFTER the colon, "2)".

It would seem that the colon character has caused the gloss to be misparsed!

It's definitely the colon where the bug occurs in the gloss text, not the number of characters.

As a colon is the delimiter within lemma attributes (eg "strong:nnn" "robinson:xxx"), I guess the code for lemmas was copied more or less unaltered to the code for glosses.

In theory, there no reason why gloss text should exclude any character, even punctuation.
In terms of syntax, that's what makes the distinction between a lemma and a gloss.

A lemma has to use either an implicitly defined or explicitly specified reference work.
A gloss has no such requirement.

It's just text in general, as in the example on our wiki page.
i.e. In the line exported from the module JapMeiji.