[Unicode]  Technical Reports
 

Proposed Update Unicode® Standard Annex #29

Unicode Text Segmentation

Version Unicode 12.0.0
Editors Mark Davis (markdavis@google.com)
Date 2018-11-01
This Version http://www.unicode.org/reports/tr29/tr29-34.html
Previous Version http://www.unicode.org/reports/tr29/tr29-33.html
Latest Version http://www.unicode.org/reports/tr29/
Latest Proposed Update http://www.unicode.org/reports/tr29/proposed.html
Revision 34

Summary

This annex describes guidelines for determining default segmentation boundaries between certain significant text elements: grapheme clusters (“user-perceived characters”), words, and sentences. For line boundaries, see [UAX14] .

Status

This is a draft document which may be updated, replaced, or superseded by other documents at any time. Publication does not imply endorsement by the Unicode Consortium. This is not a stable document; it is inappropriate to cite this document as other than a work in progress.

A Unicode Standard Annex (UAX) forms an integral part of the Unicode Standard, but is published online as a separate document. The Unicode Standard may require conformance to normative content in a Unicode Standard Annex, if so specified in the Conformance chapter of that version of the Unicode Standard. The version number of a UAX document corresponds to the version of the Unicode Standard of which it forms a part.

Please submit corrigenda and other comments with the online reporting form [Feedback]. Related information that is useful in understanding this annex is found in Unicode Standard Annex #41, “Common References for Unicode Standard Annexes.” For the latest version of the Unicode Standard, see [Unicode]. For a list of current Unicode Technical Reports, see [Reports]. For more information about versions of the Unicode Standard, see [Versions]. For any errata which may apply to this annex, see [Errata].

Contents


1 Introduction

This annex describes guidelines for determining default boundaries between certain significant text elements: user-perceived characters, words, and sentences. The process of boundary determination is also called segmentation.

A string of Unicode-encoded text often needs to be broken up into text elements programmatically. Common examples of text elements include what users think of as characters, words, lines (more precisely, where line breaks are allowed), and sentences. The precise determination of text elements may vary according to orthographic conventions for a given script or language. The goal of matching user perceptions cannot always be met exactly because the text alone does not always contain enough information to unambiguously decide boundaries. For example, the period (U+002E FULL STOP) is used ambiguously, sometimes for end-of-sentence purposes, sometimes for abbreviations, and sometimes for numbers. In most cases, however, programmatic text boundaries can match user perceptions quite closely, although sometimes the best that can be done is not to surprise the user.

Rather than concentrate on algorithmically searching for text elements (often called segments), a simpler and more useful computation instead detects the boundaries (or breaks) between those text elements. The determination of those boundaries is often critical to performance, so it is important to be able to make such a determination as quickly as possible. (For a general discussion of text elements, see Chapter 2, General Structure, of [Unicode].)

The default boundary determination mechanism specified in this annex provides a straightforward and efficient way to determine some of the most significant boundaries in text: user-perceived characters, words, and sentences. Boundaries used in line breaking (also called word wrapping) are defined in [UAX14].

The sheer number of characters in the Unicode Standard, together with its representational power, place requirements on both the specification of text element boundaries and the underlying implementation. The specification needs to allow the designation of large sets of characters sharing the same characteristics (for example, uppercase letters), while the implementation must provide quick access and matches to those large sets. The mechanism also must handle special features of the Unicode Standard, such as nonspacing marks and conjoining jamos.

The default boundary determination builds upon the uniform character representation of the Unicode Standard, while handling the large number of characters and special features such as nonspacing marks and conjoining jamos in an effective manner. As this mechanism lends itself to a completely data-driven implementation, it can be tailored to particular orthographic conventions or user preferences without recoding.

As in other Unicode algorithms, these specifications provide a logical description of the processes: implementations can achieve the same results without using code or data that follows these rules step-by-step. In particular, many production-grade implementations will use a state-table approach. In that case, the performance does not depend on the complexity or number of rules. Rather, performance is only affected by the number of characters that may match after the boundary position in a rule that applies.

1.1 Notation

A boundary specification summarizes boundary property values used in that specification, then lists the rules for boundary determinations in terms of those property values. The summary is provided as a list, where each element of the list is one of the following:

For example, the following is such a list:

General_Category = Line_Separator, or
General_Category = Paragraph_Separator, or
General_Category = Control, or
General_Category = Format
and not U+000D CARRIAGE RETURN (CR)
and not U+000A LINE FEED (LF)
and not U+200C ZERO WIDTH NON-JOINER (ZWNJ)
and not U+200D ZERO WIDTH JOINER (ZWJ)

In the table assigning the boundary property values, all of the values are intended to be disjoint except for the special value Any. In case of conflict, rows higher in the table have precedence in terms of assigning property values to characters. Data files containing explicit assignments of the property values are found in [Props].

Boundary determination is specified in terms of an ordered list of rules, indicating the status of a boundary position. The rules are numbered for reference and are applied in sequence to determine whether there is a boundary at any given offset. That is, there is an implicit “otherwise” at the front of each rule following the first. The rules are processed from top to bottom. As soon as a rule matches and produces a boundary status (boundary or no boundary) for that offset, the process is terminated.

Each rule consists of a left side, a boundary symbol (see Table 1), and a right side. Either of the sides can be empty. The left and right sides use the boundary property values in regular expressions. The regular expression syntax used is a simplified version of the format supplied in Unicode Technical Standard #18, Unicode Regular Expressions [UTS18].

Table 1. Boundary Symbols

÷ Boundary (allow break here)
× No boundary (do not allow break here)
Treat whatever on the left side as if it were what is on the right side

An open-box symbol (“␣”) is used to indicate a space in examples.

1.2 Rule Constraints

These rules are constrained in three ways, to make implementations significantly simpler and more efficient. These constraints have not been found to be limitations for natural language use. In particular, the rules are formulated so that they can be efficiently implemented, such as with a deterministic finite-state machine based on a small number of property values.

  1. Single boundaries. Each rule has exactly one boundary position. This restriction is more a limitation on the specification methods, because a rule with multiple boundaries could be expressed instead as multiple rules. For example:
    • “a b ÷ c d ÷ e f” could be broken into two rules “a b ÷ c d e f” and “a b c d ÷ e f”
    • “a b × c d × e f” could be broken into two rules “a b × c d e f” and “a b c d × e f”
  2. Limited negation. Negation of expressions is limited to instances that resolve to a match against single characters, such as “¬(OLetter | Upper | Lower | Sep)”.
  3. Ignore degenerates. No special provisions are made to get marginally better behavior for degenerate cases that never occur in practice, such as an A followed by an Indic combining mark.
  4. Script boundaries. Script boundaries are treated as degenerate cases in these rules, so the string “aquaφοβία” is treated as a single word, and the sequence ‘a’ + ‘ ि’ as a single grapheme cluster. However, implementations are free to customize boundary testing to break at script boundaries, which may be especially useful for grapheme clusters. When this is done, the Common/Inherited values need to be handled properly, and the Script_Extensions property should be used instead of the Script property alone.

2 Conformance

There are many different ways to divide text elements corresponding to user-perceived characters, words, and sentences, and the Unicode Standard does not restrict the ways in which implementations can produce these divisions.

This specification defines default mechanisms; more sophisticated implementations can and should tailor them for particular locales or environments. For example, reliable detection of word boundaries in languages such as Thai, Lao, Chinese, or Japanese requires the use of dictionary lookup, analogous to English hyphenation. An implementation therefore may need to provide means to override or subclass the default mechanisms described in this annex. Note that tailoring can either add boundary positions or remove boundary positions, compared to the defaults specified here.

Notes:

To maintain canonical equivalence, all of the following specifications are defined on text normalized in form NFD, as defined in Unicode Standard Annex #15, “Unicode Normalization Forms” [UAX15]. A boundary exists in text not normalized in form NFD if and only if it would occur at the corresponding position in NFD text. However, the default rules have been written to provide equivalent results for non-NFD text and can be applied directly. Even in the case of tailored rules, the requirement to use NFD is only a logical specification; in practice, implementations can avoid normalization and achieve the same results. For more information, see Section 6, Implementation Notes.

3 Grapheme Cluster Boundaries

It is important to recognize that what the user thinks of as a “character”—a basic unit of a writing system for a language—may not be just a single Unicode code point. Instead, that basic unit may be made up of multiple Unicode code points. To avoid ambiguity with the computer use of the term character, this is called a user-perceived character. For example, “G” + grave-accent is a user-perceived character: users think of it as a single character, yet is actually represented by two Unicode code points. These user-perceived characters are approximated by what is called a grapheme cluster, which can be determined programmatically.

Grapheme cluster boundaries are important for collation, regular expressions, UI interactions, segmentation for vertical text, identification of boundaries for first-letter styling, and counting “character” positions within text. Word boundaries, line boundaries, and sentence boundaries should not occur within a grapheme cluster: in other words, a grapheme cluster should be an atomic unit with respect to the process of determining these other boundaries.

As far as a user is concerned, the underlying representation of text is not important, but it is important that an editing interface present a uniform implementation of what the user thinks of as characters. Grapheme clusters can be treated as units, by default, for processes such as the formatting of drop caps, as well as the implementation of text selection, arrow key movement or backspacing through text, and so forth. For example, when a grapheme cluster is represented internally by a character sequence consisting of base character + accents, then using the right arrow key would skip from the start of the base character to the end of the last accent.

This document defines a default specification for grapheme clusters. It may be customized for particular languages, operations, or other situations. For example, arrow key movement could be tailored by language, or could use knowledge specific to particular fonts to move in a more granular manner, in circumstances where it would be useful to edit individual components. This could apply, for example, to the complex editorial requirements for the Northern Thai script Tai Tham (Lanna). Similarly, editing a grapheme cluster element by element may be preferable in some circumstances. For example, on a given system the backspace key might delete by code point, while the delete key may delete an entire cluster.

Moreover, there is not a one-to-one relationship between grapheme clusters and keys on a keyboard. A single key on a keyboard may correspond to a whole grapheme cluster, a part of a grapheme cluster, or a sequence of more than one grapheme cluster.

Grapheme clusters can only provide an approximation of where to put cursors. Detailed cursor placement depends on the text editing framework. The text editing framework determines where the edges of glyphs are, and how they correspond to the underlying characters, based on information supplied by the lower-level text rendering engine and font. For example, the text editing framework must know if a digraph is represented as a single glyph in the font, and therefore may not be able to position a cursor at the proper position separating its two components. That framework must also be able to determine display representation in cases where two glyphs overlap—this is true generally when a character is displayed together with a subsequent nonspacing mark, but must also be determined in detail for complex script rendering. For cursor placement, grapheme clusters boundaries can only supply an approximate guide for cursor placement using least-common-denominator fonts for the script.

In those relatively rare circumstances where programmers need to supply end users with user-perceived character counts, the counts should correspond to the number of segments delimited by grapheme cluster boundaries. Grapheme clusters may also be used in searching and matching; for more information, see Unicode Technical Standard #10, “Unicode Collation Algorithm” [UTS10], and Unicode Technical Standard #18, “Unicode Regular Expressions” [UTS18].

The Unicode Standard provides default algorithms for determining grapheme cluster boundaries, with two variants: legacy grapheme clusters and extended grapheme clusters. The most appropriate variant depends on the language and operation involved. However, the extended grapheme cluster boundaries are recommended for general processing, while the legacy grapheme cluster boundaries are maintained primarily for backwards compatibility with earlier versions of this specification.

These algorithms can be adapted to produce tailored grapheme clusters for specific locales or other customizations, such as the contractions used in collation tailoring tables. In Table 1a are some examples of the differences between these concepts. The tailored examples are only for illustration: what constitutes a grapheme cluster will depend on the customizations used by the particular tailoring in question.

Table 1a. Sample Grapheme Clusters

Ex Characters Comments
Grapheme clusters (both legacy and extended)
0067 ( g ) LATIN SMALL LETTER G
0308 ( ◌̈ ) COMBINING DIAERESIS
combining character sequences
AC01 ( 각 ) HANGUL SYLLABLE GAG Hangul syllables such as gag (which may be a single character, or a sequence of conjoining jamos)
1100 ( ᄀ ) HANGUL CHOSEONG KIYEOK
1161 ( ᅡ ) HANGUL JUNGSEONG A
11A8 ( ᆨ ) HANGUL JONGSEONG KIYEOK
0E01 ( ก ) THAI CHARACTER KO KAI Thai ko
Extended grapheme clusters
நி 0BA8 ( ந ) TAMIL LETTER NA
0BBF ( ி ) TAMIL VOWEL SIGN I
Tamil ni
0E40 ( เ ) THAI CHARACTER SARA E Thai e
กำ 0E01 ( ก ) THAI CHARACTER KO KAI
0E33 ( ำ ) THAI CHARACTER SARA AM
Thai kam
षि 0937 ( ष ) DEVANAGARI LETTER SSA
093F ( ि ) DEVANAGARI VOWEL SIGN I
Devanagari ssi
Legacy grapheme clusters
0E33 ( ำ ) THAI CHARACTER SARA AM Thai am
0937 ( ष ) DEVANAGARI LETTER SSA Devanagari ssa
ि 093F ( ि ) DEVANAGARI VOWEL SIGN I Devanagari i
Tailored grapheme clusters
ch 0063 ( c ) LATIN SMALL LETTER C
0068 ( h ) LATIN SMALL LETTER H
Slovak ch digraph
006B ( k ) LATIN SMALL LETTER K
02B7 ( ʷ ) MODIFIER LETTER SMALL W
sequence with letter modifier
क्षि 0915 ( क ) DEVANAGARI LETTER KA
094D ( ् ) DEVANAGARI SIGN VIRAMA
0937 ( ष ) DEVANAGARI LETTER SSA
093F ( ि ) DEVANAGARI VOWEL SIGN I
Devanagari kshi

See also: Where is my Character?, and the UCD file NamedSequences.txt [Data34].

A legacy grapheme cluster is defined as a base (such as A or カ) followed by zero or more continuing characters. One way to think of this is as a sequence of characters that form a “stack”.

The base can be single characters, or be any sequence of Hangul Jamo characters that form a Hangul Syllable, as defined by D133 in The Unicode Standard, or be a pair of Regional_Indicator (RI) characters. For more information about RI characters, see [UTS51].

The continuing characters include nonspacing marks, the Join_Controls (U+200C ZERO WIDTH NON-JOINER and U+200D ZERO WIDTH JOINER) used in Indic languages, and a few spacing combining marks to ensure canonical equivalence. Additional cases need to be added for completeness, so that any string of text can be divided up into a sequence of grapheme clusters. Some of these may be degenerate cases, such as a control code, or an isolated combining mark.

An extended grapheme cluster is the same as a legacy grapheme cluster, with the addition of some other characters. The continuing characters are extended to include all spacing combining marks, such as the spacing (but dependent) vowel signs in Indic scripts. For example, this includes U+093F ( ि ) DEVANAGARI VOWEL SIGN I. The extended grapheme clusters should be used in implementations in preference to legacy grapheme clusters, because they provide better results for Indic scripts such as Tamil or Devanagari in which editing by orthographic syllable is typically preferred. For scripts such as Thai, Lao, and certain other Southeast Asian scripts, editing by visual unit is typically preferred, so for those scripts the behavior of extended grapheme clusters is similar to (but not identical to) the behavior of legacy grapheme clusters.

For the rules defining the boundaries for grapheme clusters, see Section 3.1. For more information on the composition of Hangul syllables, see Chapter 3, Conformance, of [Unicode].

Note: The boundary between default Unicode grapheme clusters can be determined by just the two adjacent characters. See Section 7, Testing, for a chart showing the interactions of pairs of characters.

A key feature of default Unicode grapheme clusters (both legacy and extended) is that they remain unchanged across all canonically equivalent forms of the underlying text. Thus the boundaries remain unchanged whether the text is in NFC or NFD. Using a grapheme cluster as the fundamental unit of matching thus provides a very clear and easily explained basis for canonically equivalent matching. This is important for applications from searching to regular expressions.

Another key feature is that default Unicode grapheme clusters are atomic units with respect to the process of determining the Unicode default word, and sentence boundaries. They are usually—but not always—atomic units with respect to line boundaries: there are exceptions due to the special handling of spaces. For more information, see Section 9.2 Legacy Support for Space Character as Base for Combining Marks in [UAX14].

Grapheme clusters can be tailored to meet further requirements. Such tailoring is permitted, but the possible rules are outside of the scope of this document. One example of such a tailoring would be for the aksaras, or orthographic syllables, used in many Indic scripts. Aksaras usually consist of a consonant, sometimes with an inherent vowel and sometimes followed by an explicit, dependent vowel whose rendering may end up on any side of the consonant letter base. Extended grapheme clusters include such simple combinations.

However, aksaras may also include one or more additional prefixed consonants, typically with a virama (halant) character between each pair of consonants in the sequence. Such consonant cluster aksaras are not incorporated into the default rules for extended grapheme clusters, in part because not all such sequences are considered to be single “characters” by users. Indic scripts vary considerably in how they handle the rendering of such aksaras—in some cases stacking them up into combined forms known as consonant conjuncts, and in other cases stringing them out horizontally, with visible renditions of the halant on each consonant in the sequence. There is even greater variability in how the typical liquid consonants (or “medials”), ya, ra, la, and wa, are handled for display in combinations in aksaras. So tailorings for aksaras may need to be script-, language-, font-, or context-specific to be useful.

Note: Font-based information may be required to determine the appropriate unit to use for UI purposes, such as identification of boundaries for first-letter paragraph styling. For example, such a unit could be a ligature formed of two grapheme clusters, such as لا (Arabic lam + alef).

The Unicode definitions of grapheme clusters are defaults: not meant to exclude the use of more sophisticated definitions of tailored grapheme clusters where appropriate. Such definitions may more precisely match the user expectations within individual languages for given processes. For example, “ch” may be considered a grapheme cluster in Slovak, for processes such as collation. The default definitions are, however, designed to provide a much more accurate match to overall user expectations for what the user perceives of as characters than is provided by individual Unicode code points.

Note: The default Unicode grapheme clusters were previously referred to as “locale-independent graphemes.” The term cluster is used to emphasize that the term grapheme is used differently in linguistics. For simplicity and to align terminology with Unicode Technical Standard #10, “Unicode Collation Algorithm” [UTS10], the terms default and tailored are preferred over locale-independent and locale-dependent, respectively.

Display of Grapheme Clusters. Grapheme clusters are not the same as ligatures. For example, the grapheme cluster “ch” in Slovak is not normally a ligature and, conversely, the ligature “fi” is not a grapheme cluster. Default grapheme clusters do not necessarily reflect text display. For example, the sequence <f, i> may be displayed as a single glyph on the screen, but would still be two grapheme clusters.

For information on the matching of grapheme clusters with regular expressions, see Unicode Technical Standard #18, “Unicode Regular Expressions” [UTS18].

Degenerate Cases. The default specifications are designed to be simple to implement, and provide an algorithmic determination of grapheme clusters. However, they do not have to cover edge cases that will not occur in practice. For the purpose of segmentation, they may also include degenerate cases that are not thought of as grapheme clusters, such as an isolated control character or combining mark. In this, they differ from the combining character sequences and extended combining character sequences defined in [Unicode]. In addition, Unassigned (Cn) code points and Private_Use (Co) characters are given property values that anticipate potential usage.

Combining Character Sequences and Grapheme Clusters. For comparison, Table 1b shows the relationship between combining character sequences and grapheme clusters, using regex notation. Note that given alternates (X|Y), the first match is taken. The simple identifiers starting with lowercase are variables that are defined in Table 1c; those starting with uppercase letters are Grapheme_Cluster_Break Property Values defined in Table 2.

Table 1b. Combining Character Sequences and Grapheme Clusters

Term Regex Notes
combining character sequence ccs-base? ccs-extend+ A single base character is not a combining character sequence. However, a single combining mark is a (degenerate) combining character sequence.
extended combining character sequence extended_base? ccs-extend+ extended_base includes Hangul Syllables
legacy grapheme cluster crlf
| Control
| legacy-core legacy-postcore*
A single base character is a grapheme cluster. Degenerate cases include any isolated non-base characters, and non-base characters like controls.
extended grapheme cluster crlf
| Control
| precore* core postcore*
Extended grapheme clusters add prepending and spacing marks.

Table 1b uses several symbols defined in Table 1c. Square brackets and \p{...} are used to indicate sets of characters, using the normal UnicodeSet notion.

Table 1c. Regex Definitions

ccs-base := [\p{L}\p{N}\p{P}\p{S}\p{Zs}]
ccs-extend := [\p{M}\p{Join_Control}]
extended_base := ccs-base
| hangul-syllable
crlf := CR LF
legacy-core := hangul-syllable
| ri-sequence
| xpicto-sequence
| [^Control CR LF]
legacy-postcore := [Extend ZWJ]
core := hangul-syllable
| ri-sequence
| xpicto-sequence
| [^Control CR LF]
postcore := [Extend ZWJ SpacingMark]
precore := Prepend
RI-Sequence := RI RI
hangul-syllable := L* (V+ | LV V* | LVT) T*
| L+
| T+
xpicto-sequence := \p{Extended_Pictographic} (Extend* ZWJ \p{Extended_Pictographic})*

3.1 Default Grapheme Cluster Boundary Specification

The following is a general specification for grapheme cluster boundaries—language-specific rules in [CLDR] should be used where available.

The Grapheme_Cluster_Break property value assignments are explicitly listed in the corresponding data file in [Props]. The values in that file are the normative property values.

For illustration, property values are summarized in Table 2, but the lists of characters are illustrative.

Table 2. Grapheme_Cluster_Break Property Values

Value Summary List of Characters
CR U+000D CARRIAGE RETURN (CR)
LF U+000A LINE FEED (LF)
Control General_Category = Line_Separator, or
General_Category = Paragraph_Separator, or
General_Category = Control, or
General_Category = Unassigned and Default_Ignorable_Code_Point, or
General_Category = Surrogate, or
General_Category = Format
and not U+000D CARRIAGE RETURN
and not U+000A LINE FEED
and not U+200C ZERO WIDTH NON-JOINER (ZWNJ)
and not U+200D ZERO WIDTH JOINER (ZWJ)
Extend Grapheme_Extend = Yes, or
Emoji_Modifier=Yes in emoji-data.txt (See [UTS51])
This includes:
General_Category = Nonspacing_Mark
General_Category = Enclosing_Mark
U+200C ZERO WIDTH NON-JOINER
plus a few General_Category = Spacing_Mark needed for canonical equivalence.
ZWJ U+200D ZERO WIDTH JOINER
Regional_Indicator (RI) Regional_Indicator = Yes

This consists of the range:
U+1F1E6 REGIONAL INDICATOR SYMBOL LETTER A
..U+1F1FF REGIONAL INDICATOR SYMBOL LETTER Z
Prepend Indic_Syllabic_Category = Consonant_Preceding_Repha, or
Indic_Syllabic_Category = Consonant_Prefixed, or
Prepended_Concatenation_Mark = Yes
SpacingMark Grapheme_Cluster_Break ≠ Extend, and
General_Category = Spacing_Mark, or
any of the following (which have General_Category = Other_Letter):
U+0E33 ( ำ ) THAI CHARACTER SARA AM
U+0EB3 ( ຳ ) LAO VOWEL SIGN AM

Exceptions: The following (which have General_Category = Spacing_Mark and would otherwise be included) are specifically excluded:
U+102B ( ါ ) MYANMAR VOWEL SIGN TALL AA
U+102C ( ာ ) MYANMAR VOWEL SIGN AA
U+1038 ( း ) MYANMAR SIGN VISARGA
U+1062 ( ၢ ) MYANMAR VOWEL SIGN SGAW KAREN EU
..U+1064 ( ၤ ) MYANMAR TONE MARK SGAW KAREN KE PHO
U+1067 ( ၧ ) MYANMAR VOWEL SIGN WESTERN PWO KAREN EU
..U+106D ( ၭ ) MYANMAR SIGN WESTERN PWO KAREN TONE-5
U+1083 ( ႃ ) MYANMAR VOWEL SIGN SHAN AA
U+1087 ( ႇ ) MYANMAR SIGN SHAN TONE-2
..U+108C ( ႌ ) MYANMAR SIGN SHAN COUNCIL TONE-3
U+108F ( ႏ ) MYANMAR SIGN RUMAI PALAUNG TONE-5
U+109A ( ႚ ) MYANMAR SIGN KHAMTI TONE-1
..U+109C ( ႜ ) MYANMAR VOWEL SIGN AITON A
U+1A61 ( ᩡ ) TAI THAM VOWEL SIGN A
U+1A63 ( ᩣ ) TAI THAM VOWEL SIGN AA
U+1A64 ( ᩤ ) TAI THAM VOWEL SIGN TALL AA
U+AA7B ( ꩻ ) MYANMAR SIGN PAO KAREN TONE
U+AA7D ( ꩽ ) MYANMAR SIGN TAI LAING TONE-5
U+11720 ( 𑜠 ) AHOM VOWEL SIGN A
U+11721 ( 𑜡 ) AHOM VOWEL SIGN AA
L Hangul_Syllable_Type=L, such as:
U+1100 ( ᄀ ) HANGUL CHOSEONG KIYEOK
U+115F ( ) HANGUL CHOSEONG FILLER
U+A960 ( ꥠ ) HANGUL CHOSEONG TIKEUT-MIEUM
U+A97C ( ꥼ ) HANGUL CHOSEONG SSANGYEORINHIEUH
V Hangul_Syllable_Type=V, such as:
U+1160 ( ) HANGUL JUNGSEONG FILLER
U+11A2 ( ᆢ ) HANGUL JUNGSEONG SSANGARAEA
U+D7B0 ( ힰ ) HANGUL JUNGSEONG O-YEO
U+D7C6 ( ퟆ ) HANGUL JUNGSEONG ARAEA-E
T Hangul_Syllable_Type=T, such as:
U+11A8 ( ᆨ ) HANGUL JONGSEONG KIYEOK
U+11F9 ( ᇹ ) HANGUL JONGSEONG YEORINHIEUH
U+D7CB ( ퟋ ) HANGUL JONGSEONG NIEUN-RIEUL
U+D7FB ( ퟻ ) HANGUL JONGSEONG PHIEUPH-THIEUTH
LV Hangul_Syllable_Type=LV, that is:
U+AC00 ( 가 ) HANGUL SYLLABLE GA
U+AC1C ( 개 ) HANGUL SYLLABLE GAE
U+AC38 ( 갸 ) HANGUL SYLLABLE GYA
...
LVT Hangul_Syllable_Type=LVT, that is:
U+AC01 ( 각 ) HANGUL SYLLABLE GAG
U+AC02 ( 갂 ) HANGUL SYLLABLE GAGG
U+AC03 ( 갃 ) HANGUL SYLLABLE GAGS
U+AC04 ( 간 ) HANGUL SYLLABLE GAN
...
E_Base This value is obsolete and unused.
E_Modifier This value is obsolete and unused.
Glue_After_Zwj This value is obsolete and unused.
E_Base_GAZ (EBG) This value is obsolete and unused.
Any This is not a property value; it is used in the rules to represent any code point.

3.1.1 Grapheme Cluster Boundary Rules

The same rules are used for the two variants of grapheme clusters, except the rules GB9a and GB9b. The following table shows the differences, which are also marked on the rules themselves. The extended rules are recommended, except where the legacy variant is required for a specific environment.

Grapheme Cluster Variant Includes Excludes
LG: legacy grapheme clusters   GB9a, GB9b
​EG: ​extended grapheme clusters GB9a, GB9b  

When citing the Unicode definition of grapheme clusters, it must be clear which of the two alternatives are being specified: extended versus legacy.

Break at the start and end of text, unless the text is empty.
GB1 sot ÷ Any
GB2 Any ÷ eot
Do not break between a CR and LF. Otherwise, break before and after controls.
GB3 CR × LF
GB4 (Control | CR | LF) ÷  
GB5 ÷ (Control | CR | LF)
Do not break Hangul syllable sequences.
GB6 L × (L | V | LV | LVT)
GB7 (LV | V) × (V | T)
GB8 (LVT | T) × T
Do not break before extending characters or ZWJ.
GB9   × (Extend | ZWJ)
The GB9a and GB9b rules only apply to extended grapheme clusters:
Do not break before SpacingMarks, or after Prepend characters.
GB9a   × SpacingMark
GB9b Prepend ×  
Do not break within emoji modifier sequences or emoji zwj sequences.
GB11 \p{Extended_Pictographic} Extend* ZWJ × \p{Extended_Pictographic}
The \p{Extended_Pictographic} values are provided as a part of the Emoji data in [UTS51].
Do not break within emoji flag sequences. That is, do not break between regional indicator (RI) symbols if there is an odd number of RI characters before the break point.
GB12 sot (RI RI)* RI × RI
GB13 [^RI] (RI RI)* RI × RI
Otherwise, break everywhere.
GB999 Any ÷ Any

Notes:

4 Word Boundaries

Word boundaries are used in a number of different contexts. The most familiar ones are selection (double-click mouse selection or “move to next word” control-arrow keys) and the dialog option “Whole Word Search” for search and replace. They are also used in database queries, to determine whether elements are within a certain number of words of one another. Searching may also use word boundaries in determining matching items. Word boundaries are not restricted to whitespace and punctuation. Indeed, some languages do not use spaces at all.

Figure 1 gives an example of word boundaries, marked in the sample text with vertical bars. In the following discussion, search terms are indicated by enclosing them in square brackets for clarity. Spaces are indicated with the open-box symbol “␣”, and the matching parts between the search terms and target text are emphasized in color.

Figure 1. Word Boundaries

The   quick   ( brown )   fox   can’t   jump   32.3   feet ,   right ?

Boundaries such as those flanking the words in Figure 1 are the boundaries that users would expect, for example, when searching for a term in the target text using Whole Word Search mode. In that mode there is a match if—in addition to a matching sequence of characters—there are word boundaries in the target text on both sides of the search term. In the sample target text in Figure 1, Whole Word Search would have results such as the following:

To allow for such matches that users would expect, there are word breaks by default between most characters that are not normally considered parts of words, such as punctuation and spaces.

Word boundaries can also be used in intelligent cut and paste. With this feature, if the user cuts a selection of text on word boundaries, adjacent spaces are collapsed to a single space. For example, cutting “quick” from “The␣quick␣fox” would leave “The␣ ␣fox”. Intelligent cut and paste collapses this text to “The␣fox”. However, spaces need to be handled separately: cutting the center space from “The␣ ␣ ␣fox” probably should not collapse the remaining two spaces to one.

Proximity tests in searching determines whether, for example, “quick” is within three words of “fox”. That is done with the above boundaries by ignoring any words that contain only whitespace, punctuation, and similar characters, as in Figure 2. Thus, for proximity, “fox” is within three words of “quick”. This same technique can be used for “get next/previous word” commands or keyboard arrow keys. Letters are not the only characters that can be used to determine the “significant” words; different implementations may include other types of characters such as digits or perform other analysis of the characters.

Figure 2. Extracted Words

The quick brown fox can’t jump 32.3 feet right

Word boundaries are related to line boundaries, but are distinct: there are some word boundaries that are not line boundaries, and vice versa. A line boundary is usually a word boundary, but there are exceptions such as a word containing a SHY (soft hyphen): it will break across lines, yet is a single word.

As with the other default specifications, implementations may override (tailor) the results to meet the requirements of different environments or particular languages. For some languages, it may also be necessary to have different tailored word break rules for selection versus Whole Word Search.

In particular, the characters with the Line_Break property values of Contingent_Break (CB), Complex_Context (SA/Southeast Asian), and Unknown (XX) are assigned Word_Break property values based on criteria outside of the scope of this annex. That means that satisfactory treatment of languages like Chinese or Thai requires special handling.

4.1 Default Word Boundary Specification

The following is a general specification for word boundaries—language-specific rules in [CLDR] should be used where available.

The Word_Break property value assignments are explicitly listed in the corresponding data file in [Props]. The values in that file are the normative property values.

For illustration, property values are summarized in Table 3, but the lists of characters are illustrative.

Table 3. Word_Break Property Values

Value Summary List of Characters
CR U+000D CARRIAGE RETURN (CR)
LF U+000A LINE FEED (LF)
Newline U+000B LINE TABULATION
U+000C FORM FEED (FF)
U+0085 NEXT LINE (NEL)
U+2028 LINE SEPARATOR
U+2029 PARAGRAPH SEPARATOR
Extend Grapheme_Extend = Yes, or
General_Category = Spacing_Mark, or
Emoji_Modifier=Yes in emoji-data.txt (See [UTS51])
and not U+200D ZERO WIDTH JOINER (ZWJ)
ZWJ U+200D ZERO WIDTH JOINER
Regional_Indicator (RI) Regional_Indicator = Yes

This consists of the range:
U+1F1E6 REGIONAL INDICATOR SYMBOL LETTER A
..U+1F1FF REGIONAL INDICATOR SYMBOL LETTER Z
Format General_Category = Format
and not U+200B ZERO WIDTH SPACE (ZWSP)
and not U+200C ZERO WIDTH NON-JOINER (ZWNJ)
and not U+200D ZERO WIDTH JOINER (ZWJ)
Katakana Script = KATAKANA, or
any of the following:

U+3031 ( 〱 ) VERTICAL KANA REPEAT MARK
U+3032 ( 〲 ) VERTICAL KANA REPEAT WITH VOICED SOUND MARK
U+3033 ( 〳 ) VERTICAL KANA REPEAT MARK UPPER HALF
U+3034 ( 〴 ) VERTICAL KANA REPEAT WITH VOICED SOUND MARK UPPER HALF
U+3035 ( 〵 ) VERTICAL KANA REPEAT MARK LOWER HALF
U+309B ( ゛ ) KATAKANA-HIRAGANA VOICED SOUND MARK
U+309C ( ゜ ) KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK
U+30A0 ( ゠ ) KATAKANA-HIRAGANA DOUBLE HYPHEN
U+30FC ( ー ) KATAKANA-HIRAGANA PROLONGED SOUND MARK
U+FF70 ( ー ) HALFWIDTH KATAKANA-HIRAGANA PROLONGED SOUND MARK
Hebrew_Letter Script = Hebrew
and General_Category = Other_Letter
ALetter Alphabetic = Yes, or
any of the following characters:
U+02C2 ( ˂ ) MODIFIER LETTER LEFT ARROWHEAD
..U+02C5 ( ˅ ) MODIFIER LETTER DOWN ARROWHEAD
U+02D2 ( ˒ ) MODIFIER LETTER CENTRED RIGHT HALF RING
..U+02D7 ( ˗ ) MODIFIER LETTER MINUS SIGN
U+02DE ( ˞ ) MODIFIER LETTER RHOTIC HOOK
U+02DF ( ˟ ) MODIFIER LETTER CROSS ACCENT
U+02ED ( ˭ ) MODIFIER LETTER UNASPIRATED
U+02EF ( ˯ ) MODIFIER LETTER LOW DOWN ARROWHEAD
..U+02FF ( ˿ ) MODIFIER LETTER LOW LEFT ARROW
U+05F3 ( ׳ ) HEBREW PUNCTUATION GERESH
U+A720 ( ꜠ ) MODIFIER LETTER STRESS AND HIGH TONE
U+A721 ( ꜡ ) MODIFIER LETTER STRESS AND LOW TONE
U+A789 ( ꞉ ) MODIFIER LETTER COLON
U+A78A ( ꞊ ) MODIFIER LETTER SHORT EQUALS SIGN
U+AB5B ( ꭛ ) MODIFIER BREVE WITH INVERTED BREVE
U+055B ( ՛ ) ARMENIAN EMPHASIS MARK
U+055C ( ՜ ) ARMENIAN EXCLAMATION MARK
U+055E ( ՞ ) ARMENIAN QUESTION MARK
and Ideographic = No
and Word_Break ≠ Katakana
and Line_Break ≠ Complex_Context (SA)
and Script ≠ Hiragana
and Word_Break ≠ Extend
and Word_Break ≠ Hebrew_Letter
Single_Quote U+0027 ( ' ) APOSTROPHE
Double_Quote U+0022 ( " ) QUOTATION MARK
MidNumLet U+002E ( . ) FULL STOP
U+2018 ( ‘ ) LEFT SINGLE QUOTATION MARK
U+2019 ( ’ ) RIGHT SINGLE QUOTATION MARK
U+2024 ( ․ ) ONE DOT LEADER
U+FE52 ( ﹒ ) SMALL FULL STOP
U+FF07 ( ' ) FULLWIDTH APOSTROPHE
U+FF0E ( . ) FULLWIDTH FULL STOP
MidLetter U+00B7 ( · ) MIDDLE DOT
U+0387 ( · ) GREEK ANO TELEIA
U+05F4 ( ״ ) HEBREW PUNCTUATION GERSHAYIM
U+2027 ( ‧ ) HYPHENATION POINT
U+003A ( : ) COLON (used in Swedish)
U+FE13 ( ︓ ) PRESENTATION FORM FOR VERTICAL COLON
U+FE55 ( ﹕ ) SMALL COLON
U+FF1A ( : ) FULLWIDTH COLON
MidNum Line_Break = Infix_Numeric, or
any of the following:
U+066C ( ٬ ) ARABIC THOUSANDS SEPARATOR
U+FE50 ( ﹐ ) SMALL COMMA
U+FE54 ( ﹔ ) SMALL SEMICOLON
U+FF0C ( , ) FULLWIDTH COMMA
U+FF1B ( ; ) FULLWIDTH SEMICOLON
and not U+003A ( : ) COLON
and not U+FE13 ( ︓ ) PRESENTATION FORM FOR VERTICAL COLON
and not U+002E ( . ) FULL STOP
Numeric Line_Break = Numeric
and not U+066C ( ٬ ) ARABIC THOUSANDS SEPARATOR
ExtendNumLet General_Category = Connector_Punctuation, or
U+202F NARROW NO-BREAK SPACE (NNBSP)
E_Base This value is obsolete and unused.
E_Modifier This value is obsolete and unused.
Glue_After_Zwj This value is obsolete and unused.
E_Base_GAZ (EBG) This value is obsolete and unused.
WSegSpace General_Category = Zs
and not Linebreak = Glue
Any This is not a property value; it is used in the rules to represent any code point.

 

4.1.1 Word Boundary Rules

The table of word boundary rules uses the macro values listed in Table 3a. Each macro represents a repeated union of the basic Word_Break property values and is shown in boldface to distinguish it from the basic property values.

Table 3a. Word_Break Rule Macros

Macro Represents
AHLetter (ALetter | Hebrew_Letter)
MidNumLetQ (MidNumLet | Single_Quote)

 

Break at the start and end of text, unless the text is empty.
WB1 sot ÷ Any
WB2 Any ÷ eot
Do not break within CRLF.
WB3 CR × LF
Otherwise break before and after Newlines (including CR and LF)
WB3a (Newline | CR | LF) ÷  
WB3b   ÷ (Newline | CR | LF)
Do not break within emoji zwj sequences.
WB3c ZWJ × \p{Extended_Pictographic}
The \p{Extended_Pictographic} values are provided as a part of the Emoji data in [UTS51].
Keep horizontal whitespace together.
WB3d WSegSpace × WSegSpace
Ignore Format and Extend characters, except after sot, CR, LF, and Newline. (See Section 6.2, Replacing Ignore Rules.) This also has the effect of: Any × (Format | Extend | ZWJ)
WB4 X (Extend | Format | ZWJ)* X
Do not break between most letters.
WB5 AHLetter × AHLetter
Do not break letters across certain punctuation.
WB6 AHLetter × (MidLetter | MidNumLetQ) AHLetter
WB7 AHLetter (MidLetter | MidNumLetQ) × AHLetter
WB7a Hebrew_Letter × Single_Quote
WB7b Hebrew_Letter × Double_Quote Hebrew_Letter
WB7c Hebrew_Letter Double_Quote × Hebrew_Letter
Do not break within sequences of digits, or digits adjacent to letters (“3a”, or “A3”).
WB8 Numeric × Numeric
WB9 AHLetter × Numeric
WB10 Numeric × AHLetter
Do not break within sequences, such as “3.2” or “3,456.789”.
WB11 Numeric (MidNum | MidNumLetQ) × Numeric
WB12 Numeric × (MidNum | MidNumLetQ) Numeric
Do not break between Katakana.
WB13 Katakana × Katakana
Do not break from extenders.
WB13a (AHLetter | Numeric | Katakana | ExtendNumLet) × ExtendNumLet
WB13b ExtendNumLet × (AHLetter | Numeric | Katakana)
Do not break within emoji flag sequences. That is, do not break between regional indicator (RI) symbols if there is an odd number of RI characters before the break point.
WB15 sot (RI RI)* RI × RI
WB16 [^RI] (RI RI)* RI × RI
Otherwise, break everywhere (including around ideographs).
WB999 Any ÷ Any

Notes:

4.2 Name Validation

Related to word determination is the issue of personal name validation. Implementations sometimes need to validate fields in which personal names are entered. The goal is to distinguish between characters like those in “James Smith-Faley, Jr.” and those in “!#@♥≠”. It is important to be reasonably lenient, because users need to be able to add legitimate names, like “di Silva”, even if the names contain characters such as space. Typically, these personal name validations should not be language-specific; someone might be using a Web site in one language while his name is in a different language, for example. A basic set of name validation characters consists the characters allowed in words according to the above definition, plus a number of exceptional characters:

Basic Name Validation Characters

This is only a basic set of validation characters; in particular, the following points should be kept in mind:

5 Sentence Boundaries

Sentence boundaries are often used for triple-click or some other method of selecting or iterating through blocks of text that are larger than single words. They are also used to determine whether words occur within the same sentence in database queries.

Plain text provides inadequate information for determining good sentence boundaries. Periods can signal the end of a sentence, indicate abbreviations, or be used for decimal points, for example. Without much more sophisticated analysis, one cannot distinguish between the two following examples of the sequence <?, ”, space, uppercase-letter>. In the first example, they mark the end of a sentence, while in the second they do not.

He said, “Are you going?”  John shook his head.

“Are you going?” John asked.

Without analyzing the text semantically, it is impossible to be certain which of these usages is intended (and sometimes ambiguities still remain). However, in most cases a straightforward mechanism works well.

Note: As with the other default specifications, implementations are free to override (tailor) the results to meet the requirements of different environments or particular languages. For example, locale-sensitive boundary suppression specifications can be expressed in LDML [UTS35]. Specific sentence boundary suppressions are available in the Common Locale Data Repository [CLDR] and may be used to improve the quality of boundary analysis.

5.1 Default Sentence Boundary Specification

The following is a general specification for sentence boundaries—language-specific rules in [CLDR] should be used where available.

The Sentence_Break property value assignments are explicitly listed in the corresponding data file in [Props]. The values in that file are the normative property values.

For illustration, property values are summarized in Table 4, but the lists of characters are illustrative.

Table 4. Sentence_Break Property Values

Value Summary List of Characters
CR U+000D CARRIAGE RETURN (CR)
LF U+000A LINE FEED (LF)
Extend Grapheme_Extend = Yes, or
U+200D ZERO WIDTH JOINER (ZWJ), or
General_Category = Spacing_Mark
Sep U+0085 NEXT LINE (NEL)
U+2028 LINE SEPARATOR
U+2029 PARAGRAPH SEPARATOR
Format General_Category = Format
and not U+200C ZERO WIDTH NON-JOINER (ZWNJ)
and not U+200D ZERO WIDTH JOINER (ZWJ)
Sp White_Space = Yes
and Sentence_Break ≠ Sep
and Sentence_Break ≠ CR
and Sentence_Break ≠ LF
Lower Lowercase = Yes
and Grapheme_Extend = No
and not in the ranges
U+10D0..U+10FA (ა..ჺ) GEORGIAN LETTER AN .. GEORGIAN LETTER AIN
— Mkhedruli
U+10FD..U+10FF (ჽ..ჿ) GEORGIAN LETTER AEN .. GEORGIAN LETTER LABIAL SIGN

Review Note: The above exclusion is because Georgian cased letters do not consistently mark sentence boundaries. This exclusion does not include the following Lowercase Georgian characters — feedback is requested as to whether it should.
U+2D00..U+2D25 (ⴀ..ⴥ) GEORGIAN SMALL LETTER AN .. GEORGIAN SMALL LETTER HOE
— Small letters (Khutsuri)
U+2D27 GEORGIAN SMALL LETTER YN
U+2D2D GEORGIAN SMALL LETTER AEN
Upper General_Category = Titlecase_Letter, or
Uppercase = Yes
and not in the ranges
U+1C90..U+1CBA (Ა..Ჺ) GEORGIAN MTAVRULI CAPITAL LETTER AN .. GEORGIAN MTAVRULI CAPITAL LETTER AIN
— Capital letters (Mtavruli)
U+1CBD..U+1CBF (Ჽ..Ჿ) GEORGIAN MTAVRULI CAPITAL LETTER AEN .. GEORGIAN MTAVRULI CAPITAL LETTER LABIAL SIGN
OLetter Alphabetic = Yes, or
U+00A0 NO-BREAK SPACE (NBSP), or
U+05F3 ( ׳ ) HEBREW PUNCTUATION GERESH
and Lower = No
and Upper = No
and Sentence_Break ≠ Extend
Numeric Line_Break = Numeric
ATerm U+002E ( . ) FULL STOP
U+2024 ( ․ ) ONE DOT LEADER
U+FE52 ( ﹒ ) SMALL FULL STOP
U+FF0E ( . ) FULLWIDTH FULL STOP
SContinue U+002C ( , ) COMMA
U+002D ( - ) HYPHEN-MINUS
U+003A ( : ) COLON
U+055D ( ՝ ) ARMENIAN COMMA
U+060C ( ، ) ARABIC COMMA
U+060D ( ‎؍‎ ) ARABIC DATE SEPARATOR
U+07F8 ( ߸ ) NKO COMMA
U+1802 ( ᠂ ) MONGOLIAN COMMA
U+1808 ( ᠈ ) MONGOLIAN MANCHU COMMA
U+2013 ( – ) EN DASH
U+2014 ( — ) EM DASH
U+3001 ( 、 ) IDEOGRAPHIC COMMA
U+FE10 ( ︐ ) PRESENTATION FORM FOR VERTICAL COMMA
U+FE11 ( ︑ ) PRESENTATION FORM FOR VERTICAL IDEOGRAPHIC COMMA
U+FE13 ( ︓ ) PRESENTATION FORM FOR VERTICAL COLON
U+FE31 ( ︱ ) PRESENTATION FORM FOR VERTICAL EM DASH
U+FE32 ( ︲ ) PRESENTATION FORM FOR VERTICAL EN DASH
U+FE50 ( ﹐ ) SMALL COMMA
U+FE51 ( ﹑ ) SMALL IDEOGRAPHIC COMMA
U+FE55 ( ﹕ ) SMALL COLON
U+FE58 ( ﹘ ) SMALL EM DASH
U+FE63 ( ﹣ ) SMALL HYPHEN-MINUS
U+FF0C ( , ) FULLWIDTH COMMA
U+FF0D ( - ) FULLWIDTH HYPHEN-MINUS
U+FF1A ( : ) FULLWIDTH COLON
U+FF64 ( 、 ) HALFWIDTH IDEOGRAPHIC COMMA
STerm Sentence_Terminal = Yes
Close General_Category = Open_Punctuation, or
General_Category = Close_Punctuation, or
Line_Break = Quotation
and not U+05F3 ( ׳ ) HEBREW PUNCTUATION GERESH
and ATerm = No
and STerm = No
Any This is not a property value; it is used in the rules to represent any code point.

 

5.1.1 Sentence Boundary Rules

The table of sentence boundary rules uses the macro values listed in Table 4a. Each macro represents a repeated union of the basic Sentence_Break property values and is shown in boldface to distinguish it from the basic property values.

Table 4a. Sentence_Break Rule Macros

Macro Represents
ParaSep (Sep | CR | LF)
SATerm (STerm | ATerm)

 

Break at the start and end of text, unless the text is empty.
SB1 sot ÷ Any
SB2 Any ÷ eot
Do not break within CRLF.
SB3 CR × LF
Break after paragraph separators.
SB4 ParaSep ÷  
Ignore Format and Extend characters, except after sot, ParaSep, and within CRLF. (See Section 6.2, Replacing Ignore Rules.) This also has the effect of: Any × (Format | Extend)
SB5 X (Extend | Format)* X
Do not break after full stop in certain contexts. [See note below.]
SB6 ATerm × Numeric
SB7 (Upper | Lower) ATerm × Upper
SB8 ATerm Close* Sp* × ( ¬(OLetter | Upper | Lower | ParaSep | SATerm) )* Lower
SB8a SATerm Close* Sp* × (SContinue | SATerm)
Break after sentence terminators, but include closing punctuation, trailing spaces, and any paragraph separator. [See note below.]
SB9 SATerm Close* × (Close | Sp | ParaSep)
SB10 SATerm Close* Sp* × (Sp | ParaSep)
SB11 SATerm Close* Sp* ParaSep? ÷  
Otherwise, do not break.
SB998 Any × Any

Notes:

Figure 3. Forbidden Breaks on “.”

c. d
3. 4
U. S.
... the resp.  leaders are ...
... etc.)’  ‘(the ...

Figure 4. Allowed Breaks on “.”

She said “See spot run.”  John shook his head. ...
... etc. 它们指...
...理数字. 它们指...

 

6 Implementation Notes

6.1 Normalization

The boundary specifications are stated in terms of text normalized according to Normalization Form NFD (see Unicode Standard Annex #15, “Unicode Normalization Forms” [UAX15]). In practice, normalization of the input is not required. To ensure that the same results are returned for canonically equivalent text (that is, the same boundary positions will be found, although those may be represented by different offsets), the grapheme cluster boundary specification has the following features:

The specification also avoids certain problems by explicitly assigning the Extend property value to certain characters, such as U+09BE ( া ) BENGALI VOWEL SIGN AA, to deal with particular compositions.

The other default boundary specifications never break within grapheme clusters, and they always use a consistent property value for each grapheme cluster as a whole.

6.2 Replacing Ignore Rules

An important rule for the default word and sentence specifications ignores Extend and Format characters. The main purpose of this rule is to always treat a grapheme cluster as a single character—that is, as if it were simply the first character of the cluster. Both word and sentence specifications do not distinguish between L, V, T, LV, and LVT: thus it does not matter whether there is a sequence of these or a single one. In addition, there is a specific rule to disallow breaking within CRLF. Thus ignoring Extend is sufficient to disallow breaking within a grapheme cluster. Format characters are also ignored by default, because these characters are normally irrelevant to such boundaries.

The “Ignore” rule is then equivalent to making the following changes in the rules:

Replace the “Ignore” rule by the following, to disallow breaks within sequences (except after CRLF and related characters):
Original   Modified
X (Extend | Format)*→X (¬Sep) × (Extend | Format)
In all subsequent rules, insert (Extend | Format)* after every boundary property value, except in negations (such as ¬(OLetter | Upper ...). (It is not necessary to do this after the final property, on the right side of the break symbol.) For example:
Original   Modified
X Y × Z W X (Extend | Format)* Y (Extend | Format)* × Z (Extend | Format)* W
X Y × X (Extend | Format)* Y (Extend | Format)* ×
An alternate expression that resolves to a single character is treated as a whole. For example:
Original   Modified
(STerm | ATerm) (STerm | ATerm) (Extend | Format)*
This is not interpreted as:
  (STerm (Extend | Format)* | ATerm (Extend | Format)*)

Note: Where the “Ignore” rule uses a different set, such as (Extend | Format | ZWJ) instead of (Extend | Format), the corresponding changes would be made in the above replacements.

The “Ignore” rules should not be overridden by tailorings, with the possible exception of remapping some of the Format characters to other classes.

6.3 State Machines

The rules for grapheme clusters can be easily converted into a regular expression, as in Table 1b, Combining Character Sequences and Grapheme Clusters. It must be evaluated starting at a known boundary (such as the start of the text), and it will determine the next boundary position. The resulting regular expression can also be used to generate fast, deterministic finite-state machines that will recognize all the same boundaries that the rules do.

The conversion into a regular expression is very straightforward for grapheme cluster boundaries. It is not as easy to convert the word and sentence boundaries, nor the more complex line boundaries [UAX14]. However, it is possible to also convert their rules into fast, deterministic finite-state machines that will recognize all the same boundaries that the rules do. The implementation of text segmentation in the ICU library follows that strategy.

For more information on Unicode Regular Expressions, see Unicode Technical Standard #18, “Unicode Regular Expressions” [UTS18].

6.4 Random Access

Random access introduces a further complication. When iterating through a string from beginning to end, a regular expression or state machine works well. From each boundary to find the next boundary is very fast. By constructing a state table for the reverse direction from the same specification of the rules, reverse iteration is possible.

However, suppose that the user wants to iterate starting at a random point in the text, or detect whether a random point in the text is a boundary. If the starting point does not provide enough context to allow the correct set of rules to be applied, then one could fail to find a valid boundary point. For example, suppose a user clicked after the first space after the question mark in “Are␣you␣there?␣ ␣No,␣I’m␣not”. On a forward iteration searching for a sentence boundary, one would fail to find the boundary before the “N”, because the “?” had not been seen yet.

A second set of rules to determine a “safe” starting point provides a solution. Iterate backward with this second set of rules until a safe starting point is located, then iterate forward from there. Iterate forward to find boundaries that were located between the safe point and the starting point; discard these. The desired boundary is the first one that is not less than the starting point. The safe rules must be designed so that they function correctly no matter what the starting point is, so they have to be conservative in terms of finding boundaries, and only find those boundaries that can be determined by a small context (a few neighboring characters).

Figure 5. Random Access

random access diagram

This process would represent a significant performance cost if it had to be performed on every search. However, this functionality can be wrapped up in an iterator object, which preserves the information regarding whether it currently is at a valid boundary point. Only if it is reset to an arbitrary location in the text is this extra backup processing performed. The iterator may even cache local values that it has already traversed.

6.5 Tailoring

Rule-based implementation can also be combined with a code-based or table-based tailoring mechanism. For typical state machine implementations, for example, a Unicode character is typically passed to a mapping table that maps characters to boundary property values. This mapping can use an efficient mechanism such as a trie. Once a boundary property value is produced, it is passed to the state machine.

The simplest customization is to adjust the values coming out of the character mapping table. For example, to mark the appropriate quotation marks for a given language as having the sentence boundary property value Close, artificial property values can be introduced for different quotation marks. A table can be applied after the main mapping table to map those artificial character property values to the real ones. To change languages, a different small table is substituted. The only real cost is then an extra array lookup.

For code-based tailoring a different special range of property values can be added. The state machine is set up so that any special property value causes the state machine to halt and return a particular exception value. When this exception value is detected, the higher-level process can call specialized code according to whatever the exceptional value is. This can all be encapsulated so that it is transparent to the caller.

For example, Thai characters can be mapped to a special property value. When the state machine halts for one of these values, then a Thai word break implementation is invoked internally, to produce boundaries within the subsequent string of Thai characters. These boundaries can then be cached so that subsequent calls for next or previous boundaries merely return the cached values. Similarly Lao characters can be mapped to a different special property value, causing a different implementation to be invoked.

7 Testing

There is no requirement that Unicode-conformant implementations implement these default boundaries. As with the other default specifications, implementations are also free to override (tailor) the results to meet the requirements of different environments or particular languages. For those who do implement the default boundaries as specified in this annex, and wish to check that that their implementation matches that specification, three test files have been made available in [Tests29].

These tests cannot be exhaustive, because of the large number of possible combinations; but they do provide samples that test all pairs of property values, using a representative character for each value, plus certain other sequences.

A sample HTML file is also available for each that shows various combinations in chart form, in [Charts29]. The header cells of the chart consist of a property value, followed by a representative code point number. The body cells in the chart show the break status: whether a break occurs between the row property value and the column property value. If the browser supports tool-tips, then hovering the mouse over the code point number will show the character name, General_Category, Line_Break, and Script property values. Hovering over the break status will display the number of the rule responsible for that status.

Note: Testing two adjacent characters is insufficient for determining a boundary, except for the case of the default grapheme clusters.

The chart may be followed by some test cases. These test cases consist of various strings with the break status between each pair of characters shown by blue lines for breaks and by whitespace for non-breaks. Hovering over each character (with tool-tips enabled) shows the character name and property value; hovering over the break status shows the number of the rule responsible for that status.

Due to the way they have been mechanically processed for generation, the test rules do not match the rules in this annex precisely. In particular:

  1. The rules are cast into a more regex-style.
  2. The rules “sot ÷”, “÷ eot”, and “÷ Any” are added mechanically and have artificial numbers.
  3. The rules are given decimal numbers without prefix, so rules such as WB13a are given a number using tenths, such as 13.1.
  4. Where a rule has multiple parts (lines), each one is numbered using hundredths, such as
    • 21.01) × $BA
    • 21.02) × $HY
    • ...
  5. Any “treat as” or “ignore” rules are handled as discussed in this annex, and thus reflected in a transformation of the rules not visible in the tests.

The mapping from the rule numbering in this annex to the numbering for the test rules is summarized in Table 5.

Table 5. Numbering of Rules

Rule in This Annex Test Rule Comment
xx1 0.2 sot (start of text)
xx2 0.3 eot (end of text)
SB8a 8.1 Letter style
WB13a 13.1
WB13b 13.2
GB999 999.0 Any
WB999

Note: Rule numbers may change between versions of this annex.

8 Hangul Syllable Boundary Determination

In rendering, a sequence of jamos is displayed as a series of syllable blocks. The following rules specify how to divide up an arbitrary sequence of jamos (including nonstandard sequences) into these syllable blocks. The symbols L, V, T, LV, LVT represent the corresponding Hangul_Syllable_Type property values; the symbol M for combining marks.

The precomposed Hangul syllables are of two types: LV or LVT. In determining the syllable boundaries, the LV behave as if they were a sequence of jamo L V, and the LVT behave as if they were a sequence of jamo L V T.

Within any sequence of characters, a syllable break never occurs between the pairs of characters shown in Table 6. In all cases other than those shown in Table 6, a syllable break occurs before and after any jamo or precomposed Hangul syllable. As for other characters, any combining mark between two conjoining jamos prevents the jamos from forming a syllable block.

Table 6. Hangul Syllable No-Break Rules

Do Not Break Between Examples
L L, V, LV or LVT L × L
L × V
L × LV
L × LVT
V or LV V or T V × V
V × T
LV × V
LV × T
T or LVT T T × T
LVT × T
Jamo, LV or LVT Combining marks L × M
V × M
T × M
LV × M
LVT × M

Even in Normalization Form NFC, a syllable block may contain a precomposed Hangul syllable in the middle. An example is L LVT T. Each well-formed modern Hangul syllable, however, can be represented in the form L V T? (that is one L, one V and optionally one T) and consists of a single encoded character in NFC.

For information on the behavior of Hangul compatibility jamos in syllables, see Section 18.6, Hangul of [Unicode].

8.1 Standard Korean Syllables

Using regular expression notation, a canonically decomposed standard Korean syllable block is of the following form:

L+ V+ T*

Arbitrary standard Korean syllable blocks have a somewhat more complex form because they include any canonically equivalent sequence, thus including precomposed Korean syllables. The regular expressions for them have the following form:

(L+ V+ T*) | (L* LV V* T*) | (L* LVT T*)

All standard Korean syllable blocks used in modern Korean are of the form <L V T> or <L V> and have equivalent, single-character precomposed forms.

Old Korean characters are represented by a series of conjoining jamos. While the Unicode Standard allows for two L, V, or T characters as part of a syllable, KS X 1026-1 only allows single instances. Implementations that need to conform to KS X 1026-1 can tailor the default rules in Section 3.1  Default Grapheme Cluster Boundary Specification accordingly.

8.2 Transforming into Standard Korean Syllables

A sequence of jamos that do not all match the regular expression for a standard Korean syllable block can be transformed into a sequence of standard Korean syllable blocks by the correct insertion of choseong fillers (Lf ) and jungseong fillers (Vf ). This transformation of a string of text into standard Korean syllables is performed by determining the syllable breaks as explained in the earlier subsection “Hangul Syllable Boundaries,” then inserting one or two fillers as necessary to transform each syllable into a standard Korean syllable as shown in Figure 6.

Figure 6. Inserting Fillers

L [^V] → L Vf [^V]
[^L] V → [^L] Lf V
[^V] T → [^V] Lf Vf T

In Figure 6, [^X] indicates a character that is not X, or the absence of a character.

In Table 7, the first row shows syllable breaks in a standard sequence, the second row shows syllable breaks in a nonstandard sequence, and the third row shows how the sequence in the second row could be transformed into standard form by inserting fillers into each syllable. Syllable breaks are shown by middle dots “·”.

Table 7. Korean Syllable Break Examples

No. Sequence   Sequence with Syllable Breaks Marked
1 LVTLVLVLVf Lf VLf Vf T LVT · LV · LV · LVf · Lf V · Lf Vf T
2 LLTTVVTTVVLLVV LL · TT · VVTT · VV · LLVV
3 LLTTVVTTVVLLVV LLVf · Lf Vf TT · Lf VVTT · Lf VV · LLVV

Acknowledgments

Mark Davis is the author of the initial version and has added to and maintained the text of this annex. Laurențiu Iancu assisted in updating it for Versions 7.0 through 10.0.

Thanks to Julie Allen, Asmus Freytag, Manish Goregaokar, Andy Heninger, Ted Hopp, Tsuyoshi Ito, Martin Hosken, Michael Kaplan, Eric Mader, Otto Stolz, Steve Tolkin, Ken Whistler, and Karl Williamson for their feedback on this annex, including earlier versions.

References

For references for this annex, see Unicode Standard Annex #41, “Common References for Unicode Standard Annexes.”

Modifications

The following summarizes modifications from the previous published version of this annex.

Revision 34

Revision 33

Revision 32 being a proposed update, only changes between revisions 31 and 33 are noted here.

Modifications for previous versions are listed in those respective versions.