Allow <![CDATA[]]> to preserve whitespace in XML text content

[OWL-Varun]

Add a new constructor parameter (an enum XMLTextEscapeStyle with
Standard and CDATA as values) to XMLTextInfosetOutputter. When CDATA is
passed instead of Standard, wrap simple XML elements' text contents in
CDATA brackets to preserve any whitespace they contain.

DAFFODIL-2346-preserve-whitespace-in-dfdl-infoset

[mbeckerle]

+1 Minor fixes to doc strings in XSD.

[tuxji]

Have some questions.

[tuxji]

+1

Looks good except for some inconsistent formatting. Please fix and then you're good to squash and merge.

[tuxji]

Another round of changes suggested by Steve and me.

[mbeckerle]

THere are a very small number of non-resolved conversations still to be fixed.

[tuxji]

We need to update this pull request's title and description since they no longer match what this pull request actually does. We collectively decided not to create a tunable XMLOutputStyle, only pass an enum XMLTextEscapeStyle as a new constructor parameter to XMLTextInfosetOutputter. We decided not to address the question how to let a CLI option change infoset outputter settings in this pull request, even though this pull request's issue, DAFFODIL-2346, originally requested a CLI option to use <!CDATA[]]> to preserve whitespace in XML infosets' simple elements' text content.

Note that another issue, DAFFODIL-2234, also wants a CLI option to not pretty print XML or JSON infosets. We already pass a boolean "pretty" as a constructor parameter to XMLTextInfosetOutputter, but we don't have any way for the CLI to change this constructor parameter's value from true to false in the places where the codebase passes true.

If we still punt the question of how to let the CLI, or TDML Runner, change infoset outputter settings to DAFFODIL-2234 instead of this pull request, then I suggest we edit this pull request's new title and description to:

Allow <!CDATA[]]> to preserve whitespace in XML text content

Add a new constructor parameter (an enum XMLTextEscapeStyle with Standard and CDATA values) to XMLTextInfosetOutputter.  When CDATA is passed instead of the default Standard, wrap simple XML elements' text content in CDATA brackets to preserve any whitespace they contain.

DAFFODIL-2346

Also, I think we should consider adding new XMLTextInfosetOutputter methods which can change infoset outputter settings using "withXXX" calls instead of changing them in constructor parameters. Calling "withXXX" methods is a more upward compatible and extensible mechanism to change objects' settings since we can simply add new "withXXX" methods in the future. A "withXXX" method can also eliminate the need to pass a boolean or enum value since the "withXXX" method's name can describe the new setting as well, such as "withPrettyPrint" or "withCDATA".

[stevedlawrence]

Agreed one the new PR title and commit message.

Also agreed on new withXXX functions for setting parameters, that's a much cleaner API. I'd suggest we do that as a separate PR though.

[stevedlawrence]

+1 👍

[tuxji]

+1

Yes, this pull request is ready for squashing and merging after a fix to some comments.

[mbeckerle]

Feature that can allow for this parameterizing of the infoset outputters is in #912

[stevedlawrence]

I've updated this PR and I believe all issues are resolved, including the latest of only escaping with CDATA if special chars or whitespace exist. Please take one last look at this to make sure it's ready to merge.

[mbeckerle]

+1

I gave this one more quick pass over the latest changes. Looks good to me.

[tuxji]

+1

Also looks good to me, even though a future PR to switch from constructor parameters to withXXX() functions possibly might make the enums unnecessary depending on whether it makes sense to have two differently named no-argument withXXX() methods or still pass an enum to a withXXX() method. Let's merge anyway.