If an xliff contains a > inside an attribute value of a ph tag, the import fails.

The segment that causes the problem looks like this:

<source>
<ph id="5" x="& lt;lvlIndex value=& quot;%1& quot;{}/>">{5}
</ph>.
</source>

I put a space after the &, because otherwise Jira displays the value for which the entity stands instead of the entity.

Please see linked TS issue for an example xliff file.

Solution:

since the problem could not be fixed by a modified regular expression, an optional pre-parsing of the XML is introduced which normalizes the XML - so that all > characters are properly encoded.