Unicode

Functions to return information about Unicode codepoints.

Elixir strings are UTF-8-encoded Unicode binaries. This is a flexible and complete encoding scheme for the worlds many scripts, characters and emjois. However since it's a variable length encoding (using between one and four bytes for UTF-8) it is harder to use high-performance byte-oriented functions to decompose strings.

Since checking strings and codepoints for certain attributes - like whether they are upper case, or symbols, or whitespace - is a common occurrence, a performant approach to such detection is useful.

It is tempting to assume the use of US ASCII encoding and checking only for characters in that range. For example it is very common to see code in Elixir checking codepoint in ?a..?z to check for lowercase alphabetic characters. When the underlying programming language has no canonical form for a string beyond bytes this may be considered acceptable - the programmer is defining the script domain as he or she sees fit.

However since Elixir strings are declared to be UTF-8 encoded Unicode strings it seems appropriate to make it easier to determine the characteristics of codepoints (and strings) using this standard.

The Elixir standard library does not provide introspection beyond that required to support casing (String.downcase/1, String.upcase/1, String.capitalize/1). This library aims to fill in the blanks a little bit.

Unicode version

As of unicode version 1.20.0 published on September 11th, 2024, Unicode 16.0 forms the underlying data.

Additional Unicode libraries

ex_unicode provides basic introspection of Unicode codepoints and strings. Additional libraries (either released or in development) build upon this library):

• unicode_set implements functions to parse and match on unicode sets

• unicode_guards is a simple library implementing common function guards using unicode_set and unicode

• unicode_string is a library to implement efficient string splitting into words and sentences based upon the Unicode Segementation algorithm.

• unicode_transform implements the Unicode transform specification.

Unicode Functions

The following is a partial list of functions included in the library. See the documentation for the relevant module for further information:

Codepoint ranges

These functions return the codepoints as list of 2-tuples for the given property:

• Unicode.Block.blocks/0
• Unicode.Script.scripts/0
• Unicode.GeneralCategory.categories/0
• Unicode.CombiningClass.combining_classes/0
• Unicode.GraphemeBreak.grapheme_breaks/0
• Unicode.LineBreak.line_breaks/0
• Unicode.SentenceBreak.sentence_breaks/0
• Unicode.IndicSyllabicCategory.indic_syllabic_categories/0
• Unicode.Property.properties/0

Introspection of codepoints and strings

The following functions return the block, script and category for codepoints and strings:

• Unicode.script/1

  iex> Unicode.script ?ä
  :latin
  
  iex> Unicode.script ?خ
  :arabic
  
  iex> Unicode.script ?अ
  :devanagari

• Unicode.block/1

  iex> Unicode.block ?ä
  :latin_1_supplement
  
  iex> Unicode.block ?A
  :basic_latin
  
  iex> Unicode.block "äA"
  [:latin_1_supplement, :basic_latin]

• Unicode.category/1

  iex> Unicode.category ?ä
  :Ll
  iex> Unicode.category ?A
  :Lu
  iex> Unicode.category ?🧐
  :So

• Unicode.properties/1

  iex> Unicode.properties 0x1bf0
  [
    :alphabetic,
    :case_ignorable,
    :grapheme_extend,
    :id_continue,
    :other_alphabetic,
    :xid_continue
  ]
  
  iex> Unicode.properties ?A
  [
    :alphabetic,
    :ascii_hex_digit,
    :cased,
    :changes_when_casefolded,
    :changes_when_casemapped,
    :changes_when_lowercased,
    :grapheme_base,
    :hex_digit,
    :id_continue,
    :id_start,
    :uppercase,
    :xid_continue,
    :xid_start
  ]
  
  iex> Unicode.properties ?+
  [:grapheme_base, :math, :pattern_syntax]
  
  iex> Unicode.properties "a1+"
  [
    [
      :alphabetic,
      :ascii_hex_digit,
      :cased,
      :changes_when_casemapped,
      :changes_when_titlecased,
      :changes_when_uppercased,
      :grapheme_base,
      :hex_digit,
      :id_continue,
      :id_start,
      :lowercase,
      :xid_continue,
      :xid_start
    ],
    [
      :ascii_hex_digit,
      :emoji,
      :grapheme_base,
      :hex_digit,
      :id_continue,
      :xid_continue
    ],
    [:grapheme_base, :math, :pattern_syntax]
  ]

Character classes

These functions help filter codepoints and strings based upon their properties. They return a boolean result.

• Unicode.alphabetic?/1
• Unicode.alphanumeric?/1
• Unicode.digits?/1
• Unicode.numeric?/1
• Unicode.emoji?/1
• Unicode.math?/1
• Unicode.cased?/1
• Unicode.lowercase?/1
• Unicode.uppercase?/1

Any known property can be called as a function Unicode.Property.<property_name>(codepoint_or_string) or Unicode.Property.<property_name>?(codepoint_or_string) to return a boolean.

Transformations

The function Unicode.unaccent/1 attempts to transform a Unicode string into a subset of the Latin-1 alphabet by removing diacritical marks from text. It is not a full transformation (which will be available in the upcoming unicode_transform library.)

Recognition

The information functions are heavily inspired by @qqwy's elixir-unicode package and compatibility with some of the api is represented by including some of the doctests from that package. Originally published under the :unicode package name on hex, this original work is now replaced with this library code.

Installation

The package can be installed by adding unicode to your list of dependencies in mix.exs:

def deps do
  [
    {:unicode, "~> 1.20"}
  ]
end

The docs can be found at https://hexdocs.pm/unicode.