Pod::Simple::XHTML: better fallback when HTML::Entities isn't installed

[xenu]

This commit changes the default set of escaped characters in the fallback code to be the same as in HTML::Entities.

Fixes #188

[xenu]

Added a test for this. Probably won't work under Perl 5.6, too bad.

[xenu]

Fixed tests, they failed on most CI configurations, because I forgot to update the number of skipped tests in the case when HTML::Entities isn't installed.

[khwilliamson]

I would like the test to work on EBCDIC. Fortunately this is trivial to do. Just change the non-ASCII character to B6, and change the supporting text to correspond. One possibility would be. "The pilcrow, \xb6, is used to mark the beginning of a new paragraph" B6 is the only non-ASCII character that has the same meaning in Latin1
and EBCDIC

[khwilliamson]

I went searching, and discovered HTML5::Entities which knows many more definitions than plain HTML::Entities. Should that be the default, with the plain being the first fallback?

I also found Pod::Escapes, which ships with core since 5.12,, and automatically handles EBCDIC back to 5.7.3 which is good enough, and is available back to as far as we would ever need. Should this be another fallback?

[xenu]

I went searching, and discovered HTML5::Entities which knows many more definitions than plain HTML::Entities. Should that be the default, with the plain being the first fallback?

I'm not a fan of adding new dependencies to this module, I don't think it's worth it.

Honestly, I really don't think entity handling in Pod::Simple::XHTML has to be super good. For encoding, it doesn't make any difference if we output † or †, both work just as well. For decoding, it's more important to know the proper entity names, but decoding is used only in what I reckon is a rather obscure part of the code (guts of literal HTML handling).

I also found Pod::Escapes, which ships with core since 5.12,, and automatically handles EBCDIC back to 5.7.3 which is good enough, and is available back to as far as we would ever need. Should this be another fallback?

It's not a perfect fit, it doesn't provide a function equivalent to encode_entities of HTML::Entities. You could kinda use it to roll your own, by reversing the %Name2character_number hash it provides and then removing the non-standard entities it includes (specifically: "lchevron", "rchevron"). It feels icky, especially the last step. And like before, I'm not sure if it's worth it.