Crystal ICU

Crystal binding/wrapper to the ICU library

Installation

Debian/Ubuntu

apt-get install -y libicu-dev

macOS

brew install icu4c
brew link --force icu4c

Usage

Add this to your application's shard.yml:

dependencies:
  icu:
    github: olbat/icu.cr

Then require the lib in your Crystal code:

require "icu"

Wrapped components

CharsetDetector, Charset detection (ucsdet.h)
BreakIterator, Text Boundary Analysis (Break Iteration) (ubrk.h)
Transliterator, Text Transformation (Transliteration) (utrans.h)
Normalizer, Normalization (unorm2.h)
Collator, Collation (ucol.h)
StringSearch, String Searching (usearch.h)
Locale, Locales (uloc.h, ulocdata.h)
IDNA, International Domain Names in Applications (uidna.h)
Currencies, Encapsulates information about a currency (ucurr.h)
Region, Territory containment and mapping (uregion.h)
NumberFormatter, Number Formatting/Spellout (unum.h)
DateTimeFormatter Date and Time Formatting (udat.h, udatpg.h, udateintervalformat.h)
Calendar, Calendars (ucal.h)
UniversalTimeScale, Universal Time Scale (utmscale.h)
PluralRules, Plural rules (upluralrules.h)
Bidirectional Algorithm, Bidirectional text handling (ubidi.h)
uregex, Regular Expressions (uregex.h)
ucnv, Codepage Conversion (ucnv.h, ucnvsel.h)

Internals

UChars, UChar conversion routines
UEnum, String Enumeration (uenum.h)
ustring, Strings and Character Iteration (ustring.h, uiter.h)
utext, Abstract Unicode Text API (utext.h)
uset, Sets of Unicode Code Points and Strings (uset.h)
ures, Resource Bundles (ures.h)
usprep, StringPrep (usprep.h)
uspoof, Identifier Spoofing & Confusability (uspoof.h)
udata, Data loading interface (udata.h)

Development

ICU version compatibility

This binding is compatible with the ICU library starting from version 4.8. It has been generated using ICU version 74.2 so every additions in the API from this version will not be present in this binding.

The Crystal wrapper's (the ICU class) class and methods are enabled dynamically at compile-time depending on the version of the ICU library that's installed.

Please make sure that the class/feature you want to use is supported by your version ICU by checking the API documentation.

Regenerate the binding

To be sure that the LibICU binding fits with the version of the lib that's installed, it's possible to regenerate the binding by:

Installing libgen and it's dependencies (cf. libgen's documentation)
Run libgen and the lib_transformer program by running make generate_lib

Note: for some reasons, using LLVM versions >= 16 generate broken bindings. It'll be investigated later. For now, better stick using LLVM versions <= 15 when regenerating the libicu bindings using libgen.

Project's structure

src/lib_icu: this directory contains the binding to the ICU library, it's generated by libgen & a transformation program
src/icu: this directory contains the wrapper to the LibICU crystal lib that eases the usage of the binding
src/icu_info.cr: this small program is used to determine which version of ICU is installed on the system (see #1)
src/lib_transformer.cr: this small program is used to fix the binding generated by libgen (see #3)