RE: [q-lang-users] silly regex wonderings

[David v. <da...@pu...>]

OK, you've got me playing with these regexes now.

Based on the regsplit example, a regex search and replace is:
	snr Opts OldRE NewRE S = strcat
		(regex Opts OldRE S (regskip ++ `NewRE)
		++ [regskip]);

i.e. instead of splitting at a match we're inserting a new regex.

Then
	embolden2 = snr "g" "\\*([^\\*]*)\\*" '(tag "bold" (reg 1));

Note the quote and unquote to delay evaluation of (reg 1) etc.

==> embolden2 "this is some *bold* text, and a lone * at the end"
"this is some <bold>bold</bold> text, and a lone * at the end"


Slightly more usable:
	tagify Delim Tag = snr "g" (Delim++"([^"++Delim++"]*)"++Delim)
					'(tag Tag (reg 1));
	embolden3 = tagify "\\*" "bold";
	underline = tagify "_" "underline";

Then we can write:
==> underline (embolden3 "This has *bold _underlining_*!")
"This has <bold>bold <underline>underlining</underline></bold>!"

The problem is 
==> underline (embolden3 "This has *bold _underlining*_ !")
"This has <bold>bold <underline>underlining</bold></underline> !"

And this is malformed html.
I'm torn between two instincts here:
 - A clean way to do it would be to process your wikitext, build a full
   parse tree and transform this to html, but
 - this doesn't seem in the wiki spirit, where I suspect the ideal would
   be to capture the entire conversion to html in a single regex.

Cheers,
David.