• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1#
2# Copyright (C) 2016 and later: Unicode, Inc. and others.
3# License & terms of use: http://www.unicode.org/copyright.html
4# Copyright (C) 2002-2016, International Business Machines Corporation
5# and others. All Rights Reserved.
6#
7# file:  word.txt
8#
9# ICU Word Break Rules
10#      See Unicode Standard Annex #29.
11#      These rules are based on UAX #29 Revision 34 for Unicode Version 12.0
12#
13# Note:  Updates to word.txt will usually need to be merged into
14#        word_POSIX.txt and word_fi_sv.txt also.
15
16##############################################################################
17#
18#  Character class definitions from TR 29
19#
20##############################################################################
21
22!!chain;
23!!quoted_literals_only;
24
25
26#
27#  Character Class Definitions.
28#
29
30$Han                = [:Han:];
31
32$CR                 = [\p{Word_Break = CR}];
33$LF                 = [\p{Word_Break = LF}];
34$Newline            = [\p{Word_Break = Newline}];
35$Extend             = [\p{Word_Break = Extend}-$Han];
36$ZWJ                = [\p{Word_Break = ZWJ}];
37$Regional_Indicator = [\p{Word_Break = Regional_Indicator}];
38$Format             = [\p{Word_Break = Format}];
39$Katakana           = [\p{Word_Break = Katakana}];
40$Hebrew_Letter      = [\p{Word_Break = Hebrew_Letter}];
41$ALetter            = [\p{Word_Break = ALetter}];
42$Single_Quote       = [\p{Word_Break = Single_Quote}];
43$Double_Quote       = [\p{Word_Break = Double_Quote}];
44$MidNumLet          = [\p{Word_Break = MidNumLet}];
45$MidLetter          = [\p{Word_Break = MidLetter} - [\: \uFE55 \uFF1A]];
46$MidNum             = [\p{Word_Break = MidNum}];
47$Numeric            = [\p{Word_Break = Numeric}];
48$ExtendNumLet       = [\p{Word_Break = ExtendNumLet}];
49$WSegSpace          = [\p{Word_Break = WSegSpace}];
50$Extended_Pict      = [\p{Extended_Pictographic}];
51
52$Hiragana           = [:Hiragana:];
53$Ideographic        = [\p{Ideographic}];
54
55
56#   Dictionary character set, for triggering language-based break engines. Currently
57#   limited to LineBreak=Complex_Context. Note that this set only works in Unicode
58#   5.0 or later as the definition of Complex_Context was corrected to include all
59#   characters requiring dictionary break.
60
61$Control        = [\p{Grapheme_Cluster_Break = Control}];
62$HangulSyllable = [\uac00-\ud7a3];
63$ComplexContext = [:LineBreak = Complex_Context:];
64$KanaKanji      = [$Han $Hiragana $Katakana];
65$dictionaryCJK  = [$KanaKanji $HangulSyllable];
66$dictionary     = [$ComplexContext $dictionaryCJK];
67
68# TODO: check if handling of katakana in dictionary makes rules incorrect/void
69
70# leave CJK scripts out of ALetterPlus
71$ALetterPlus  = [$ALetter-$dictionaryCJK [$ComplexContext-$Extend-$Control]];
72
73
74## -------------------------------------------------
75
76# Rule 3 - CR x LF
77#
78$CR $LF;
79
80# Rule 3c   Do not break within emoji zwj sequences.
81#             ZWJ ×  \p{Extended_Pictographic}.  Precedes WB4, so no intervening Extend chars allowed.
82#
83$ZWJ $Extended_Pict;
84
85# Rule 3d - Keep horizontal whitespace together.
86#
87$WSegSpace $WSegSpace;
88
89# Rule 4 - ignore Format and Extend characters, except when they appear at the beginning
90#          of a region of Text.
91
92$ExFm  = [$Extend $Format $ZWJ];
93
94^$ExFm+;            # This rule fires only when there are format or extend characters at the
95                    # start of text, or immediately following another boundary. It groups them, in
96                    # the event there are more than one.
97
98[^$CR $LF $Newline $ExFm] $ExFm*;   # This rule rule attaches trailing format/extends to words,
99                                    # with no special rule status value.
100
101$Numeric $ExFm* {100};              # This group of rules also attach trailing format/extends, but
102$ALetterPlus $ExFm* {200};          # with rule status set based on the word's final base character.
103$HangulSyllable {200};
104$Hebrew_Letter $ExFm* {200};
105$Katakana $ExFm* {400};             # note:  these status values override those from rule 5
106$Hiragana $ExFm* {400};             #        by virtue of being numerically larger.
107$Ideographic $ExFm* {400};          #
108
109#
110# rule 5
111#    Do not break between most letters.
112#
113($ALetterPlus | $Hebrew_Letter) $ExFm* ($ALetterPlus | $Hebrew_Letter);
114
115# rule 6 and 7
116($ALetterPlus | $Hebrew_Letter)  $ExFm* ($MidLetter | $MidNumLet | $Single_Quote) $ExFm* ($ALetterPlus | $Hebrew_Letter) {200};
117
118# rule 7a
119$Hebrew_Letter $ExFm* $Single_Quote {200};
120
121# rule 7b and 7c
122$Hebrew_Letter $ExFm* $Double_Quote $ExFm* $Hebrew_Letter;
123
124# rule 8
125
126$Numeric $ExFm* $Numeric;
127
128# rule 9
129
130($ALetterPlus | $Hebrew_Letter)  $ExFm* $Numeric;
131
132# rule 10
133
134$Numeric $ExFm* ($ALetterPlus | $Hebrew_Letter);
135
136# rule 11 and 12
137
138$Numeric $ExFm* ($MidNum | $MidNumLet | $Single_Quote) $ExFm* $Numeric;
139
140# rule 13
141# to be consistent with $KanaKanji $KanaKanhi, changed
142# from 300 to 400.
143# See also TestRuleStatus in intltest/rbbiapts.cpp
144$Katakana $ExFm*  $Katakana {400};
145
146# rule 13a/b
147
148$ALetterPlus   $ExFm* $ExtendNumLet {200};    #  (13a)
149$Hebrew_Letter $ExFm* $ExtendNumLet {200};    #  (13a)
150$Numeric       $ExFm* $ExtendNumLet {100};    #  (13a)
151$Katakana      $ExFm* $ExtendNumLet {400};    #  (13a)
152$ExtendNumLet  $ExFm* $ExtendNumLet {200};    #  (13a)
153
154$ExtendNumLet  $ExFm* $ALetterPlus  {200};    #  (13b)
155$ExtendNumLet  $ExFm* $Hebrew_Letter {200};    #  (13b)
156$ExtendNumLet  $ExFm* $Numeric      {100};    #  (13b)
157$ExtendNumLet  $ExFm* $Katakana     {400};    #  (13b)
158
159# rules 15 - 17
160#    Pairs of Regional Indicators stay together.
161#    With incoming rule chaining disabled by ^, this rule will match exactly two of them.
162#    No other rule begins with a Regional_Indicator, so chaining cannot extend the match.
163#
164^$Regional_Indicator $ExFm* $Regional_Indicator;
165
166# special handling for CJK characters: chain for later dictionary segmentation
167$HangulSyllable $HangulSyllable {200};
168$KanaKanji $KanaKanji {400}; # different rule status if both kana and kanji found
169
170# Rule 999
171#     Match a single code point if no other rule applies.
172.;
173