Converter Generation Helper Script

[nikitawootten-nist]

Write a single bash script that calls Maven to produce the converter XSLTs from this repository, with minimal change to interfaces or underlying calls. (If it calls Saxon twice, so be it.)

As arguments it accepts a Metaschema module source XML with valid imports; a name part; and a path.

The script writes two converters (XSLTs) to the path named, one in each direction (XML/JSON), with appropriate names constructed from the name part.

Longer term work: refactor this to run under XProc -- goes with #41 and #42. This will involve some XSLT rework as well (to externalize some wrapper logic now only in the XSLT pipeline. The advantages will be (a) more efficient runtime (speed/memory), and (b) easier integration into further pipeline optimizations.

A goal for this Issue is thus to set the ground for that one, producing a follow-on Issue prior to #41.

[wendellpiez]

The XSLT rewrite/shakeout to make converter generators work under XProc -- as well as XSLT -- could (should) also address this bug in the runtime interface: usnistgov/OSCAL#1849

[wendellpiez]

The script to borrow from is https://github.com/usnistgov/metaschema-xslt/blob/develop/src/schema-gen/mvn-xsd-schema-xsl.sh. As noted above, result file naming should follow the current OSCAL schema naming convention.

[wendellpiez]

Working here: #44 (see the src/converter-gen folder).

These are ready for testing.

Note that the script that generates the pair of XSLTs must put them in the same place - a shortcoming, for now, to be remedied by the followup work here (the XProc).

Each of the scripts that generates a single converter lets you designate the path and file name where you want it to be produced.