← All modules

About forty small string routines that the core does not have — substring-before and -after, accent stripping, n-grams, abbreviations, whitespace introspection — each one a sub rather than a method.

Version

0.0.40 zef:lizmat

Depends

none beyond the core

License

Artistic-2.0

Its own test suite

3 files, green

Checked

2026-09-16 against Raku++ 3.28.0 and Rakudo 2026.08

Where it lives

raku.land · source

Install it #

$ rakupp install String::Utils

zef install String::Utils writes the same store; either installer leaves the module usable by both engines.

What it is for #

Every one of these is a line or three of Raku you have written before. The value is not that they are hard; it is that they are named, so a program says before($path, '/') instead of $path.substr(0, $path.index('/')) and stops being wrong when the separator is absent.

The distribution exports about forty of them. This page covers the families rather than the list.

Cutting a string at a marker #

use String::Utils;

my $path = 'lib/String/Utils.rakumod';
say 'before   : ', before($path, '/').raku;
say 'after    : ', after($path, '/').raku;
say 'between  : ', between($path, 'lib/', '.rakumod').raku;
say '';
say 'stem           : ', stem('archive.tar.gz').raku;
say 'stem, 1 part   : ', stem('archive.tar.gz', 1).raku;
say 'shorten        : ', shorten('a rather long sentence', 10).raku;
say 'nomark         : ', nomark('çédille naïve').raku;
say 'is-sha1        : ', is-sha1('0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33');
say '';
say 'ngram("raku", 2) : ', ngram('raku', 2).join(' ');
say 'word-at          : ', word-at('the quick brown', 4).raku;
say 'abbrev("length") : ', abbrev('length').sort(*.key).map({ .key }).join(' ');

before   : "lib"
after    : "String/Utils.rakumod"
between  : "String/Utils"

stem           : "archive"
stem, 1 part   : "archive.tar"
shorten        : "a ra…tence"
nomark         : "cedille naive"
is-sha1        : False

ngram("raku", 2) : ra ak ku
word-at          : (4, 5, 1)
abbrev("length") : l le len leng lengt length

before, after and between return Nil when the marker is not there, which is the behaviour that makes them worth using: the .substr/.index version returns nonsense or throws, and the difference only shows up on the input you did not test.

stem strips extensions and takes a count — one call for archive and one for archive.tar, which is the distinction .extension cannot make. shorten elides in the middle with …, keeping both ends, because the ends are where the information is in a path or an identifier.

nomark strips diacritics without a normalisation table lookup on your side, which is what you want before a case-insensitive compare or a slug.

Sequences and prefixes #

ngram slices a string into overlapping runs — the input to a similarity measure or a cheap index. word-at answers where the word under a character offset starts and how long it is, which is the primitive an editor needs and nothing in core provides.

abbrev returns every unambiguous prefix of a word, as a Hash mapping prefix to full word. The hash is the reason the example sorts before printing: the return value has no order, so a program that joins it directly prints something different on every run, and on two engines at once.

Where the two engines differ #

Nowhere. Every routine above produced identical output on Raku++ and Rakudo, and all three test files pass on both.

The one thing to watch is the abbrev ordering above, and it is not an engine difference — it is a Hash being asked to behave like a list. Sort it, or index it, but do not iterate it and expect a sequence. That is also the reason this page's example is stable under the site's two-runs-per-engine check, and an earlier draft of it was not.