Это может выглядеть уродливо, но
my $quoted = qr/'[^']+'/; # ' fix Stackoverflow highlighting
my %_extract_wanted_cache;
sub extract_wanted_memo {
my($string, $index) = @_;
$string =~ ($_extract_wanted_cache{$index} ||=
qr/^(?:.*?$quoted.*?){@{[$index-1]}}($quoted)/)
? $1
: ();
}
бенчмаркинг предполагает, что это может стоить
sub extract_wanted {
my($string, $index) = @_;
$string =~ /^(?:.*?$quoted.*?){@{[$index-1]}}($quoted)/
? $1
: ();
}
sub extract_wanted_gindex {
my($string, $index) = @_;
($string =~ /$quoted/g)[$index-1];
}
use Benchmark;
timethese -1 => {
nocache => sub { extract_wanted $string2, 3 },
memoize => sub { extract_wanted_memo $string2, 3 },
index => sub { extract_wanted_gindex $string2, 3 },
nocache_fail => sub { extract_wanted $string2, 100 },
memoize_fail => sub { extract_wanted_memo $string2, 100 },
index_fail => sub { extract_wanted_gindex $string2, 100 },
}
Результаты:
Benchmark:
running
index, index_fail, memoize, memoize_fail, nocache, nocache_fail
for at least 1 CPU seconds
...
index: 1 w/c secs (1.04 usr + 0.00 sys = 1.04 CPU) @183794.23/s (n=191146)
index_fail: 1 w/c secs (1.03 usr + 0.00 sys = 1.03 CPU) @185578.64/s (n=191146)
memoize: 1 w/c secs (1.00 usr + 0.00 sys = 1.00 CPU) @264664.00/s (n=264664)
memoize_fail: 0 w/c secs (1.03 usr + 0.00 sys = 1.03 CPU) @835106.80/s (n=860160)
nocache: 0 w/c secs (1.03 usr + 0.00 sys = 1.03 CPU) @196495.15/s (n=202390)
nocache_fail: 2 w/c secs (1.03 usr + 0.00 sys = 1.03 CPU) @445390.29/s (n=458752)