Characters can consist of multiple code points

Remarks

An Unicode code point, what programmers often think of one character, often corresponds to what the user thinks is one character. Sometimes however a “character” is made up of multiple code points, as the examples above show.

This means that operations like slicing a string, or getting a character at a given index may not work as expected. For instance the 4th character of the string "Café" is 'e' (without the accent). Similarly, clipping the string to length 4 will remove the accent.

The technical term for such a group of code points is a grapheme cluster. See UAX #29: Unicode Text Segmentation

Diacritics

A letter with a diacritic may be represented with the letter, and a combining modifier letter. You normally think of as one character, but it's really 2 code points:

  • U+0065 — LATIN SMALL LETTER E
  • U+0301 — COMBINING ACUTE ACCENT

Similarly = c + ¸, and = a + ˚

combined forms

To complicate matters, there is often a code point for the composed form as well:

"Café" = 'C' + 'a' + 'f' + 'e' + '´'
"Café" = 'C' + 'a' + 'f' + 'é'

Although these strings look the same, they are not equal, and they don't even have the same length (5 and 4 respectively).

Emoji and flags

A lot of emoji consist of more than one code point.

  • 🇯🇵: A flag is defined as a pair of "regional symbol indicator letters" (🇯 + 🇵)
  • 🙋🏿: Some emoji may be followed by a skin tone modifier: 🙋 + 🏿
  • 😀︎ or 😀️: Windows 10 allows you to specify if an emoji is colored or black/white by appending a variation selector (U+FE0E or U+FE0F)
  • 👨‍👩‍👧‍👦: a family. Encoded by joining the emoji for boy, girl, woman and man (👦, 👧, 👩, 👨) together with zero-width joiners (U+200D). On platforms which support it, this is rendered as an emoji of a family with two kids.

Zalgo Text

There is this thing called Zalgo Text which pushes this to the extreme. Here is the first grapheme cluster of the example. It consists of 15 code points: the Latin letter H and 14 combining marks.

 

H̡̫̤̤̣͉̤ͭ̓̓̇͗̎̀

 

Although this doesn't show up in normal text, it shows that a “character” really can consist of an arbitrary number of code points