First of all, I'm guessing there's no point in using /i.
As for solution, alternation is an option.
/
\p{Script_Extensions=Arabic}
| \p{Script_Extensions=Armenian}
| \p{Script_Extensions=Bengali}
| ...
| \p{Script_Extensions=Thaana}
| \p{Script_Extensions=Thai}
| \p{Script_Extensions=Tibetan}
/x
Alternation allows us to use whitespace, but a faster solution would be a character class.
/
[\p{Script_Extensions=Arabic}\p{Script_Extensions=Armenian}\p{Script_Extensions=Bengali}...\p{Script_Extensions=Thaana}\p{Script_Extensions=Thai}\p{Script_Extensions=Tibetan}]
/x
However, just 3 of the 41 properties takes up the entire width of the screen. As with the other solutions presented here, nothing stops you from building the pattern dynamically.
my $class_body =
join '',
map "\\p{Script_Extensions=$_}",
qw(
Arabic Armenian Bengali
...
Thaana Thai Tibetan
);
/[$class_body]/
But there's an alternative: (?[...])
use experimental qw( regex_sets );
/
(?[ \p{Script_Extensions=Arabic}
+ \p{Script_Extensions=Armenian}
+ \p{Script_Extensions=Bengali}
+ ...
+ \p{Script_Extensions=Thaana}
+ \p{Script_Extensions=Thai}
+ \p{Script_Extensions=Tibetan}
])
/x
It's still experimental, but I consider it stable and likely to make it into language.
Oh I like the suggestion of using the character class- much cleaner. And also map to build up the regex. Doing a benchmark on those 2 comes up good: my new method:
test1: 105 wallclock secs (92.75 usr + 0.06 sys = 92.81 CPU) @ 107747.01/s (n=10000000)- char class method:test2: 23 wallclock secs (10.76 usr + 0.07 sys = 10.83 CPU) @ 923361.03/s (n=10000000)and also, good shout about
/i- but sure why I put that on there as it makes bugger all difference when looking at whole sets!