http://qs321.pair.com?node_id=11112988

LanX has asked for the wisdom of the Perl Monks concerning the following question:

Hi

Task

I need a regex to transform wiki markup surrounding words to html, * to <b> etc.

my problem is that */_ could be combined at word boundaries, see the following example

DB<66> $_=$wiki; tf();tf();tf() ; print "'$wiki' \n=>\n'$_'" '_*one /two/*_ _*three /four/*_ _*five /six/*_' => '<u><b>one <i>two</i></b></u> <u><b>three /four/</b></u> <u><b>five <i +>six</i></b></u>' DB<67>
'_*one /two/*_ _*three /four/*_ _*five /six/*_'
=>
'one two three /four/ five six'

as you can see I have to run the tf() transformation thrice

DB<40> %h = ( '*'=>'b', '/' => 'i' , '_' => 'u' ) DB<59> sub tf { s{ $pre ([_*/]) (.*?) \2 $post}{$1<$h{$2}>$3</$h{$2} +>$4}xg } DB<62> $pre = qr/(^|\s|>)/ DB<63> $post = qr/($|\s|<)/ DB<65> $wiki='_*one /two/*_ _*three /four/*_ _*five /six/*_'

Question

Is there a way to make it a one-run transformation?

Trouble is that /g continues after the inserted replacement, here underline

I was experimenting with lookaround-assertions and \G and couldn't get it done.

Approaches

The only ways I can (theoretically) think of so far are

NB: It's a more theoretical question because running tf() three times doesn't pose problems.

UPDATE:

I just noticed a bug, since four wasn't expanded.

&tf has to be better written with a lookbehind which doesn't consume the next whitespace

DB<90> sub tf { s{ $pre ([_*/]) (.*?) \2 (?=$post)}{$1<$h{$2}>$3</$h +{$2}>}xg }

I'll update an SSCCE soon.

Cheers Rolf
(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery