| Subject: | Avoid regex recompilation |
The attached patch provides a measurable performance improvement by
avoiding recompilation of regexps.
| Subject: | o.diff |
--- SAC.pm~ 2008-07-05 22:44:20.000000000 +0300
+++ SAC.pm 2010-01-23 13:55:04.000000000 +0200
@@ -382,7 +382,7 @@
elsif ($css =~ s/^\s*\@page\s+//i) {
warn "[SAC] parsing page\n" if DEBUG;
# grab the name and pseudo-page if they're there
- $css =~ s/^($RE_IDENT)?\s*(?::($RE_IDENT))?//;
+ $css =~ s/^($RE_IDENT)?\s*(?::($RE_IDENT))?//o;
my ($name,$pseudo) = ($1,$2);
# parse the block
@@ -401,7 +401,7 @@
warn "[SAC] parsing unknown at rule ($css)\n" if DEBUG;
# take off the @rule first
- $css =~ s/^\s*(\@$RE_IDENT\s*)//;
+ $css =~ s/^\s*(\@$RE_IDENT\s*)//o;
$at_rule .= $1;
# then grab whatever is not a block
@@ -493,7 +493,7 @@
return unless $$css =~ s/^\@charset\s+//i;
# extract the string
- if ($$css =~ s/^($RE_STRING)\s*;//) {
+ if ($$css =~ s/^($RE_STRING)\s*;//o) {
my $charset = $1;
$charset =~ s/^(?:'|")//; #"
$charset =~ s/(?:'|")$//; #'
@@ -526,14 +526,14 @@
# first get the uri
my $uri;
if ($$css =~ s/^url\(//) {
- $$css =~ s/^((?:$RE_STRING)|([^\)]*))\s*//;
+ $$css =~ s/^((?:$RE_STRING)|([^\)]*))\s*//o;
$uri = $1;
$uri =~ s/^(?:'|")//; # "
$uri =~ s/(?:'|")$//; # '
$$css =~ s/^\)//;
}
else {
- $$css =~ s/^($RE_STRING)//;
+ $$css =~ s/^($RE_STRING)//o;
$uri = $1;
$uri =~ s/^(?:'|")//; #"
$uri =~ s/(?:'|")$//; #'
@@ -574,20 +574,20 @@
while ($$css =~ s/^\s*\@namespace\s+//i) {
my ($prefix,$uri);
# first get the prefix
- if ($$css !~ /^url\(/ and $$css =~ s/^($RE_IDENT)\s+//) {
+ if ($$css !~ /^url\(/ and $$css =~ s/^($RE_IDENT)\s+//o) {
$prefix = $1;
}
# then get the uri
if ($$css =~ s/^url\(//) {
- $$css =~ s/^((?:$RE_STRING)|([^\)]*))\s*//;
+ $$css =~ s/^((?:$RE_STRING)|([^\)]*))\s*//o;
$uri = $1;
$uri =~ s/^(?:'|")//; # "
$uri =~ s/(?:'|")$//; # '
$$css =~ s/^\)//;
}
else {
- $$css =~ s/^($RE_STRING)//;
+ $$css =~ s/^($RE_STRING)//o;
$uri = $1;
$uri =~ s/^(?:'|")//; #"
$uri =~ s/(?:'|")$//; #'
@@ -630,7 +630,7 @@
my $css = shift;
# test for the right content and return a list if found
- return [] unless $$css =~ s/^\s*($RE_IDENT(?:\s*,\s*$RE_IDENT)*)//;
+ return [] unless $$css =~ s/^\s*($RE_IDENT(?:\s*,\s*$RE_IDENT)*)//o;
return [map { $_ =~ s/^\s+//; $_ =~ s/\s+$//; $_; } split /,/, $1];
}
#---------------------------------------------------------------------#
@@ -749,7 +749,7 @@
}
# element
- elsif ($$css =~ s/^(?:($RE_IDENT|\*)?(\|))?($RE_IDENT|\*)//) {
+ elsif ($$css =~ s/^(?:($RE_IDENT|\*)?(\|))?($RE_IDENT|\*)//o) {
# create element selector
my ($ns,$lname);
$lname = ($3 eq '*')?undef:$3;
@@ -775,17 +775,17 @@
}
# hash id
- elsif ($$css =~ s/^#($RE_NAME)//) {
+ elsif ($$css =~ s/^#($RE_NAME)//o) {
push @tokens, $sac->[_cf_]->create_id_condition($1);
}
# dot class
- elsif ($$css =~ s/^\.($RE_IDENT)//) {
+ elsif ($$css =~ s/^\.($RE_IDENT)//o) {
push @tokens, $sac->[_cf_]->create_class_condition(undef,$1);
}
# CSS3 pseudo-elements
- elsif ($$css =~ s/^::($RE_IDENT)//) {
+ elsif ($$css =~ s/^::($RE_IDENT)//o) {
push @tokens, $sac->[_sf_]->create_pseudo_element_selector(undef,$1);
}
@@ -799,7 +799,7 @@
$attr =~ s/\s*\]$//;
# get the attr lname and ns
- $attr =~ s/^(?:($RE_IDENT|\*)?(\|))?($RE_IDENT|\*)//;
+ $attr =~ s/^(?:($RE_IDENT|\*)?(\|))?($RE_IDENT|\*)//o;
my ($ns,$lname);
$lname = ($3 eq '*')?undef:$3;
if (defined $2 and $2 eq '|') {
@@ -864,7 +864,7 @@
# :pseudo()
elsif (
- ($args,$$css,$func) = Text::Balanced::extract_bracketed($$css,q/()'"/,qr/:$RE_IDENT/)
+ ($args,$$css,$func) = Text::Balanced::extract_bracketed($$css,q/()'"/,qr/:$RE_IDENT/o)
and
length $func
) {
@@ -910,7 +910,7 @@
}
# :pseudo (not a function)
- elsif ($$css =~ s/^\:($RE_IDENT)//) {
+ elsif ($$css =~ s/^\:($RE_IDENT)//o) {
# root
if (lc($1) eq 'root') {
@@ -1153,7 +1153,7 @@
$$css =~ s/^\s*//;
while (length $$css) {
# the property
- $$css =~ s/^(-?$RE_IDENT)\s*//; # includes the - prefix
+ $$css =~ s/^(-?$RE_IDENT)\s*//o; # includes the - prefix
my $prop = $1;
$sac->parse_comments($css);
@@ -1247,21 +1247,21 @@
# lengths and assoc
elsif ($$css =~ s/^((?:\+|-)?$RE_NUM)
(em|ex|px|cm|mm|in|pt|pc|deg|rad|grad|ms|s|hz|khz|%)
- //xi) {
+ //xio) {
$value = $1;
$text = lc $2;
$type = $DIM_MAP{$text};
}
# dimension
- elsif ($$css =~ s/^((?:\+|-)?$RE_NUM)($RE_IDENT)//) {
+ elsif ($$css =~ s/^((?:\+|-)?$RE_NUM)($RE_IDENT)//o) {
$value = $1;
$text = lc $2;
$type = DIMENSION;
}
# number
- elsif ($$css =~ s/^((?:\+|-)?$RE_NUM)//) {
+ elsif ($$css =~ s/^((?:\+|-)?$RE_NUM)//o) {
$value = $1;
$text = 'number';
if ($value =~ m/\./) {
@@ -1273,7 +1273,7 @@
}
# unicode range
- elsif ($$css =~ s/^($RE_RANGE)//) {
+ elsif ($$css =~ s/^($RE_RANGE)//o) {
$value = $1;
$text = 'unicode-range';
$type = UNICODERANGE;
@@ -1288,11 +1288,11 @@
# functions
# elsif (
-# ($value,$$css,$text) = Text::Balanced::extract_bracketed($$css,q/()'"/,qr/$RE_IDENT/)
+# ($value,$$css,$text) = Text::Balanced::extract_bracketed($$css,q/()'"/,qr/$RE_IDENT/o)
# and
# length $text
# ) {
- elsif ($$css =~ s/^($RE_IDENT)\(//) {
+ elsif ($$css =~ s/^($RE_IDENT)\(//o) {
# cleanup the func and args
# $text = lc $text;
@@ -1313,14 +1313,14 @@
}
# ident
- elsif ($$css =~ s/^($RE_IDENT)//) {
+ elsif ($$css =~ s/^($RE_IDENT)//o) {
$value = $1;
$text = 'ident';
$type = IDENT;
}
# string
- elsif ($$css =~ s/^($RE_STRING)//) {
+ elsif ($$css =~ s/^($RE_STRING)//o) {
$value = $1;
$value =~ s/^(?:"|')//; #"
$value =~ s/(?:"|')$//; #"