optimized macro parser for perl 5.10

This commit is contained in:
Graham Knop 2010-02-05 18:06:31 -06:00
parent a8e7ec090b
commit 1c27279d6b
2 changed files with 46 additions and 18 deletions

View file

@ -42,25 +42,52 @@ These functions are available from this package:
=cut
#-------------------------------------------------------------------
my $parenthesis;
$parenthesis = qr{
\( # Start with '(',
(?: # Followed by
(?>\\[()]) # Escaped parenthesis
| # or
(?>[^()]) # Non-parenthesis
| # or
(??{ $parenthesis }) # a balanced parenthesis block
)* # zero or more times
\) # Ending with ')'
}x;
my $macro_re;
BEGIN {
if ( eval { require 5.010 } ) {
$macro_re = eval <<'END_REGEX';
qr{
( # capture #1 - entire macro call
\^ # start with carat
([-a-zA-Z0-9_@#/*]{1,64}) # capture #2 - macro name
( # capture #3 - parenthesis
\( # start with open parenthesis
(?: # followed by
(?> [^()] ) # non-parenthesis
| # or
(?>\\[()]) # Escaped parenthesis
| # or
(?3) # a balanced parenthesis block (recursive)
)* # zero or more times
\) # ending with closing parenthesis
)?
; # End with a semicolon.
)
}msx;
END_REGEX
}
else {
my $parenthesis;
$parenthesis = qr{
\( # Start with '(',
(?: # Followed by
(?>\\[()]) # Escaped parenthesis
| # or
(?>[^()]) # Non-parenthesis
| # or
(??{ $parenthesis }) # a balanced parenthesis block
)* # zero or more times
\) # Ending with ')'
}x;
my $macro_re = qr{
(\^ # Start with carat
([-a-zA-Z0-9_@#/*]{1,64}) # And one or more non-macro characters -tagged-
((??{ $parenthesis })?) # a balanced parenthesis block
;) # End with a semicolon.
}msx;
$macro_re = qr{
(\^ # Start with carat
([-a-zA-Z0-9_@#/*]{1,64}) # And one or more non-macro characters -tagged-
((??{ $parenthesis })?) # a balanced parenthesis block
;) # End with a semicolon.
}msx;
}
}
=head2 filter ( html )