ICML filter: Opening angled bracket in opening CDATA tag must not be converted to XML entity

[Martn Wunderl]

When trying to generate an XLIFF 2.0 file from the attached ICML using the Okapi ICML filter, the resulting XLIFF file is not a valid XML file. The XML parser will throw an error similar to:

"The character sequence "]]>" must not appear in content unless used to mark the end of a CDATA section."

The root cause seems to be that the opening angled bracket in an opening CDATA tag gets converted to an XML entity. So, in the ICML you have "<![CDATA[..." and in the resulting XLIFF you get "<![CDATA[...". Therefore, the closing CDATA tags are not associated with an opening CDATA tag anymore and the XLIFF is not valid XML.

[Martn Wunderl]

Assigning this to myself doesn't work, but I'll look into it.

[Chase Tingley]

Strangely, using M30, the XLIFF 1.2 it produces seems ok. ie, the output of this command is valid XML:

tikal.sh -fc okf_icml -x 278802.icml

Is there a filter option or something that would make it extract different content for you, or is this really somehow tied to the XLIFF 2.0 serialization?

[Martn Wunderl]

Fixed with 6bb9d6efb0cde66c3b72597d6b60a40680594d4e in the XLIFF2 toolkit. Need to update Okapi to use 1.1.5 as soon as the release is out. Also, added new Unit test for this issue that needs to activated (cf. d0eab28dfd9a0e9aa139b7b6094b8e1782080ae3).

[Martn Wunderl]

Fixed with 6bb9d6efb0cde66c3b72597d6b60a40680594d4e in the XLIFF2 toolkit. Need to update Okapi to use 1.1.5 as soon as the release is out.