A regular expression that matches just a fixed set of constant strings is wasteful of performance and is
hard on maintainers. It is much more readable and often faster to use "eq" or a hash to match such
strings.
# Bad
my $is_file_function = $token =~ m/\A (?: open | close | read ) \z/xms;
# Faster and more readable
my $is_file_function = $token eq 'open' ||
$token eq 'close' ||
$token eq 'read';
For larger numbers of strings, a hash is superior:
# Bad
my $is_perl_keyword =
$token =~ m/\A (?: chomp | chop | chr | crypt | hex | index
lc | lcfirst | length | oct | ord | ... ) \z/xms;
# Better
Readonly::Hash my %PERL_KEYWORDS => map {$_ => 1} qw(
chomp chop chr crypt hex index lc lcfirst length oct ord ...
);
my $is_perl_keyword = $PERL_KEYWORD{$token};
Conway also suggests using lc() instead of a case-insensitive match.
VARIANTS
This policy detects both grouped and non-grouped strings. The grouping may or may not be capturing. The
grouped body may or may not be alternating. "\A" and "\z" are always considered anchoring which "^" and
"$" are considered anchoring is the "m" regexp option is not in use. Thus, all of these are violations:
m/^foo$/;
m/\A foo \z/x;
m/\A foo \z/xm;
m/\A(foo)\z/;
m/\A(?:foo)\z/;
m/\A(foo|bar)\z/;
m/\A(?:foo|bar)\z/;
Furthermore, this policy detects violations in "m//", "s///" and "qr//" constructs, as you would expect.