The Unicode standard

The Unicode Standard is a character encoding that endeavors to support all historical and modern writing system using a single character encoding. Unicode is the de facto character encoding standard accepted by all major computer companies. Many products and standards now require or allow use of Unicode, for example, XML, HTML, Microsoft JScript, Java, Perl, Rust, and C#. While ISO 10646 is the corresponding worldwide de jure standard approved by all ISO members, the two standards include identical character repertoires and binary representations.

Compared to older mechanisms for handling character and string data, Unicode simplifies software localization and improves multilingual text processing. By using Unicode to represent character and string data in your applications, you can enable universal data exchange capabilities and support multiple languages in a single application. The Unicode Standard and associated specifications:

  • Allow any combination of characters, drawn from any combination of scripts and languages, to coexist in a single document.
  • Define multiple encodings of its single character set: UTF-8, UTF-16, and UTF-32. Conversion of data among these encodings is lossless.
  • Define semantics for each character.
  • Standardize script behavior and text rendering.
  • Provide a standard algorithm for bidirectional text.
  • Define cross-mappings to other standards.

Character and script support

Unicode encompasses virtually all characters used widely in computers today. For example, Version 15.0 of the Unicode Standard contains 149,186 characters and supports 161 scripts. In addition to the characters that support scripts, Unicode includes many other types of characters, such as:

  • characters used in publishing
  • mathematical and technical symbols
  • geometric shapes
  • basic dingbats
  • punctuation marks
  • emoji
  • Arabic presentation forms
  • musical symbols

Unicode encoding and planes

The original version of Unicode was designed as a 16-bit encoding, which limited the support to 65,536 (2^16) code points. Version 2.0 of the Unicode Standard increased the range from U+0000..U+10FFFF. These ranges are grouped in planes - 65,536 (2^16) code points per plane. Planes are subdivided into Unicode blocks. Each Unicode block is a contiguous range of code points that share a common purpose, such as supporting a single script.

Plane Range Name Description
0 0000-FFFF Basic Multilingual Plane (BMP)
  • Support for most modern languages including the most common CJK Unified Ideographs
  • Symbols, such as currency
  • Private Use Area
1 10000–1FFFF Supplementary Multilingual Plane (SMP)
  • Non-CJK ideographic historic scripts
  • Modern scripts
  • Symbols, such as musical notation
  • Emoji
2 20000-2FFFF Supplementary Ideographic Plane (SIP) Additional CJK Unified Ideographs, mostly historical, uncommon, or variants
3 30000-3FFFF Tertiary Ideographic Plane (TIP) CJK Unified Ideographs, mostly historical, that aren't encoded in the BMP or SIP
4-13 40000-DFFFF unassigned
14 E0000-EFFFF Supplementary Special-purpose Plane (SSP) Tags and variation sequence selectors
15-16 F0000-10FFFF Supplementary Private Use Area planes (SPUA-A/B) Private Use Area

The Unicode Standard specifies that the complete range of Unicode code points can be converted to unique code unit sequences using one of seven Unicode encoding schemes or Unicode Transformation Formats (UTF). This table compares the Unicode encoding schemes.

Unicode encoding scheme Unit size Byte order Number of bytes (bits) per character Notes
UTF-8 8 bits n/a 1-4 bytes (8-32 bits) BOM can be used to indicate encoding.
UTF-16 16 bits BOM required 2-4 bytes (16-32 bits)
UTF-16BE 16 bits Big-endian 2-4 bytes (16-32 bits)
UTF-16LE 16 bits Little-endian 2-4 bytes (16-32 bits) Default encoding for Windows.
UTF-32 32 bits BOM required 4 bytes (32 bits)
UTF-32BE 32 bits Big-endian 4 bytes (32 bits)
UTF-32LE 32 bits Little-endian 4 bytes (32 bits)

Byte order and byte order marks

UTF-8 is interpreted as a sequence of bytes:

  • The first 128 code points (U+0000..U+007F) are equivalent to the ASCII code points and only require one byte to encode.
  • The range U+0080..U+07FF requires two bytes
  • The range U+0800..U+FFFF (the remainder of the BMP) requires three bytes
  • Code points in planes 1-16 require four bytes

UTF-16 encoding forms encode code points in one or two 16-bit code units. UTF-32 encoding forms encodes code points in a single 32-bit code unit.

UTF-16 and UTF-32 can use little-endian (least-significant byte at the smallest memory address) or big-endian (most-significant byte at the smallest memory address) byte orders. UTF-16 and UTF-32 require a byte order mark (BOM, U+FEFF) at the beginning of the stream or file to indicate the byte order. The byte orders for UTF-16LE, UTF-16BE, UTF-32LE, and UTF-32BE are unambiguous.

When the BOM is encountered at the beginning of the stream or file, the order of the bytes of the BOM indicate the order of the remainder of the stream or file.

  • FE FF (UTF-16) or 00 00 FE FF (UTF-32) indicate big-endian
  • FF FE (UTF-16) or FF FE 00 00 (UTF-32) indicate little-endian

A BOM can also be used with UTF-8, but the BOM is only used as a signature to distinguish, for example, a UTF-8 stream or file from an ASCII stream or file.

Unicode Transformation Formats and file sizes

As UTF-32 requires four bytes for every Unicode code point, it would seem that UTF-32 would always lead to larger file sizes than UTF-16 and UTF-8. However, file size also depends on the code points that the file contains.

  • If the file only contains characters in the range U+0000..U+007F, UTF-8 uses 1 byte per character, UTF-16 uses 2 bytes per character, while UTF-32 uses 4 bytes per character. In other words, the file saved as UTF-32 is 4 times larger than the file saved as UTF-8.
  • If the file only contains characters in the range U+0800..U+D7FF, UTF-8 uses 3 bytes per character, UTF-16 uses 2 bytes per character, while UTF-32 uses 4 bytes per character. In other words, the file saved as UTF-8 is 50% larger than the file saved as UTF-16.
  • If the file only contains characters in the supplementary planes, UTF-8, UTF-16, and UTF-32 all use 4 bytes per character. The file is the same size irrespective of how it's saved.

Surrogate Pairs

UTF-16 uses either one or two 16-bit code units (2 bytes or 4 bytes) to represent the Unicode code points. A single code unit can represent code points in the range U+0000..U+FFFF. For code points in the supplementary planes (range U+10000...U+10FFFF) a pair of code units, called a surrogate pair, are required. The first (high) surrogate is a 16-bit code value in the range U+D800 to U+DBFF. The second (low) surrogate is a 16-bit code value in the range U+DC00 to U+DFFF.

For example 𓅡 (U+13161 EGYPTIAN HIEROGLYPH G029) is a character in the Supplementary Multilingual Plane. The UTF-32 value is 0x00013161. In UTF-16, a surrogate pair is required. The high surrogate is U+D80C while the low surrogate is U+DD61.

Unicode code points in the range U+D800..U+DFFF are reserved for use by UTF-16. When reading a UTF-16 stream, it's unambiguous whether a byte is a lead byte (high surrogate) or a trail byte (low surrogate) of a multi-byte (surrogate pair) character, which is a common challenge with MBCS text.

The Unicode Character Database and character properties

In addition to the encoding of characters, the Unicode Standard also published the Unicode Character Database (UCD). UCD is a catalog of the character and a corresponding set of semantics that describe each character including character type, use, directionality, and combining behavior.

For example, the Unicode character set includes numerous combining characters, such as U+0308 ("¨"), a combining dieresis or umlaut. Unicode can often represent the same glyph in either a composed or a decomposed form: for example, the composed form of "Ä" is the single Unicode code point "Ä" (U+00C4), while its decomposed form is "A" + "¨" (U+0041 U+0308). Unicode doesn't define a composed form for every glyph. For example, the Vietnamese lowercase "o" with circumflex and tilde ("ỗ") is represented by U+006f U+0302 U+0303 (o + Circumflex + Tilde). The composed characters are encoded in the Unicode Standard primarily for round-trip compatibility with other encodings. The Unicode Standard contains strict rules for determining the equivalence of composed characters to combining character sequences.

Unicode characters and text layout

Not all Unicode-based text processing is a matter of simple character-by-character parsing. Complex text-based operations such as hyphenation, line breaking, and glyph formation need to consider the context in which they're being used (the relation to surrounding characters, for instance). The complexity of these operations hinges on language rules and has nothing to do with Unicode as an encoding standard. Instead, the software implementation should define a higher-level protocol for handling these operations.

In contrast, there are unusual characters that have specific semantic rules attached to them; these characters are detailed in The Unicode Standard. Some characters always allow a line break (for example, most spaces), whereas others never do (for instance, nonspacing or nonbreaking characters). Many characters used in Arabic and Hebrew are defined as having a strong or weak directionality. The Unicode Standard defines an algorithm for determining the display order of bidirectional text. It also defines several directional formatting codes as overrides for cases not handled by the implicit bidirectional ordering rules to help create comprehensible bidirectional text.

The code-point positions of Unicode elements don't imply a sort order and the organization of characters within the Unicode code-point positions is often unrelated to language usage. To implement sorting, the application or the platform must handle the mapping of Unicode points to the rules that determine the sort order. The Common Locale Data Repository (CLDR) provides the regional format and sorting information that can be used to implement sorting and collation.

Variation Selection

Although the Unicode standard represents many characters, the forms the glyphs take can vary from font to font and from culture to culture. A simple example is the difference between the following two characters. The first using a serif font while the second is using a sans-serif font.

For these types of difference, there's no need to encode a different code point. This difference is stylistic and is easily recognizable as the same character. However, during the process of encoding some characters there might be cases where characters with the same semantic meaning have different representations for contextual, historical, or stylistic reasons. In those cases, Unicode provides the variation selector method to represent these characters.

There are three sets of variants defined by Unicode:

A variation sequence is defined as a base character followed by a variation selector (U+FE00-U+FE0F and U+E0100-U+E01EF, abbreviated VS1-VS16 and VS17-VS256).

This example shows GREATER-THAN BUT NOT EQUAL TO (U+2269) with and without VS1 (U+FE00). Depending on the fonts available in your browser, the first character should display with a slanted line indicating “not equals” while the second character should display with a vertical line.

For example, the character U+845B (“edible bean; surname”) can be used with a variation selector as defined in the Ideographic Variation Database (IVD):

  • U+845B: as used in the name of the Nishi-Kasai railway station: 西葛西駅

  • U+845B with U+E0100 (VS17): as used in the name of Katsuragi City in Nara prefecture: 葛󠄀城市

Tip

You will need a font that has this character defined for it to display correctly.

For more information about using variation sequences with emoji, see Emoji.