Strings, Char Lists, Graphemes and Codepoints.
Table of Contents
Elixir strings are nothing but a sequence of bytes. Let’s look at an example:
iex> string = <<104,101,108,108,111>> "hello" iex> string <> <<0>> <<104, 101, 108, 108, 111, 0>>
By concatenating the string with the byte
0, IEx displays the string as a binary because it is not a valid string anymore. This trick can help us view the underlying bytes of any string.
NOTE: Using « » syntax we are saying to the compiler that the elements inside those symbols are bytes.
Internally, Elixir strings are represented with a sequence of bytes rather than an array of characters. Elixir also has a char list type (character list). Elixir strings are enclosed with double quotes, while char lists are enclosed with single quotes.
What’s the difference? Each value in a charlist is the Unicode code point of a character whereas in a binary, the codepoints are encoded as UTF-8. Let’s dig in:
iex> 'hełło' [104, 101, 322, 322, 111] iex> "hełło" <> <<0>> <<104, 101, 197, 130, 197, 130, 111, 0>>
322 is the Unicode codepoint for ł but it is encoded in UTF-8 as the two bytes
You can get a character’s code point by using
iex> ?Z 90
This allows you to use the notation
?Z rather than ‘Z’ for a symbol.
When programming in Elixir, we usually use strings, not charlists. The charlist support is mainly included because it is required for some Erlang modules.
For further information, see the official
Getting Started Guide.
Graphemes and Codepoints
Codepoints are just simple Unicode characters which are represented by one or more bytes, depending on the UTF-8 encoding. Characters outside of the US ASCII character set will always encode as more than one byte. For example, Latin characters with a tilde or accents (
á, ñ, è) are typically encoded as two bytes. Characters from Asian languages are often encoded as three or four bytes. Graphemes consist of multiple codepoints that are rendered as a single character.
The String module already provides two functions to obtain them,
codepoints/1. Let’s look at an example:
iex> string = "\u0061\u0301" "á" iex> String.codepoints string ["a", "́"] iex> String.graphemes string ["á"]
Let’s review some of the most important and useful functions of the String module. This lesson will only cover a subset of the available functions. To see a complete set of functions visit the official
Returns the number of Graphemes in the string.
iex> String.length "Hello" 5
Returns a new string replacing a current pattern in the string with some new replacement string.
iex> String.replace("Hello", "e", "a") "Hallo"
Returns a new string repeated n times.
iex> String.duplicate("Oh my ", 3) "Oh my Oh my Oh my "
Returns a list of strings split by a pattern.
iex> String.split("Hello World", " ") ["Hello", "World"]
Let’s walk through a simple exercises to demonstrate we are ready to go with Strings!
A and B are considered anagrams if there’s a way to rearrange A or B making them equal. For example:
- A = super
- B = perus
If we re-arrange the characters on String A, we can get the string B, and vice versa.
So, how could we check if two strings are Anagrams in Elixir? The easiest solution is to just sort the graphemes of each string alphabetically and then check if they both lists are equal. Let’s try that:
defmodule Anagram do def anagrams?(a, b) when is_binary(a) and is_binary(b) do sort_string(a) == sort_string(b) end def sort_string(string) do string |> String.downcase() |> String.graphemes() |> Enum.sort() end end
Let’s first give a watch to
anagrams?/2. We are checking whether the parameters we are receiving are binaries or not. That’s the way we check if a parameter is a String in Elixir.
After that, we are calling a function that orders the string alphabetically. It first converts the string to lowercase and then uses
String.graphemes/1 to get a list of the graphemes in the string. Finally, it pipes that list into
Enum.sort/1. Pretty straight, right?
Let’s check the output on iex:
iex> Anagram.anagrams?("Hello", "ohell") true iex> Anagram.anagrams?("María", "íMara") true iex> Anagram.anagrams?(3, 5) ** (FunctionClauseError) no function clause matching in Anagram.anagrams?/2 The following arguments were given to Anagram.anagrams?/2: # 1 3 # 2 5 iex:11: Anagram.anagrams?/2
As you can see, the last call to
anagrams? caused a FunctionClauseError. This error is telling us that there is no function in our module that meets the pattern of receiving two non-binary arguments, and that’s exactly what we want, to just receive two strings, and nothing more.