y///#

Transliterate characters in a string. y/// is a synonym for tr/// — the two are identical in every respect.

The y spelling is provided for sed devotees, whose muscle memory produces y/abc/ABC/ rather than tr/abc/ABC/. Perl’s parser accepts either; the resulting op is the same. Use whichever spelling reads better in the surrounding code — if the file already quotes a sed script, y/// often fits; in fresh Perl, tr/// is more common.

Synopsis#

y/SEARCHLIST/REPLACEMENTLIST/cdsr
$var =~ y/SEARCHLIST/REPLACEMENTLIST/cdsr
y/SEARCHLIST/REPLACEMENTLIST/      # transliterates $_

What you get back#

Same as tr///:

Without /r: the number of characters matched (and replaced or deleted), after modifying the target in place.
With /r: a new string with the transliteration applied; the original is untouched.

Examples#

Classic sed-style uppercase transliteration on $_:

while (<>) {
    y/a-z/A-Z/;
    print;
}

Counting vowels without modifying the string, using /r to discard the transliterated copy:

my $vowels = ($text =~ y/aeiouAEIOU//) ;   # count in place is fine too
my $count  = () = $text =~ /[aeiouAEIOU]/g;

Strip everything except digits from a string (non-destructive copy via /r, complement + delete):

my $digits = $phone =~ y/0-9//cdr;         # "(415) 555-1212" → "4155551212"

Rot13 — the textbook y/// one-liner:

(my $coded = $plain) =~ y/A-Za-z/N-ZA-Mn-za-m/;

Edge cases#

See tr/// for the full list. Two worth repeating here:

No variable interpolation. y/$x/$y/ transliterates the literal characters $, x against $, y — not the contents of those variables. This matches tr/// and is the single most common sed-to-Perl translation bug.
y'...'...' disables ranges. With single-quote delimiters, - is literal: y'a-z'A-Z' transliterates only a, -, z against A, -, Z.

Differences from upstream#

Fully compatible with upstream Perl 5.42.