diff options
Diffstat (limited to 'i18npool/source/breakiterator/data/char_in.txt')
-rw-r--r-- | i18npool/source/breakiterator/data/char_in.txt | 48 |
1 files changed, 48 insertions, 0 deletions
diff --git a/i18npool/source/breakiterator/data/char_in.txt b/i18npool/source/breakiterator/data/char_in.txt new file mode 100644 index 000000000000..72c4a44720cd --- /dev/null +++ b/i18npool/source/breakiterator/data/char_in.txt @@ -0,0 +1,48 @@ +# +# Copyright (C) 2002-2003, International Business Machines Corporation and others. +# All Rights Reserved. +# +# file: char.txt +# +# ICU Character Break Rules, also known as Grapheme Cluster Boundaries +# See Unicode Standard Annex #29. +# These rules are based on TR29 Version 4.0.0 +# + +# +# Character Class Definitions. +# The names are those from TR29. +# +$CR = \r; +$LF = \n; +$Control = [[:Zl:] [:Zp:] [:Cc:] [:Cf:]]; + +# add Japanese Half Width voicing marks to $Extend +$VoiceMarks = [\uff9e\uff9f]; +$cmcextend = [ \u0903 \u093e-\u0940 \u0949-\u094C \u09bf-\u09c0 \u09c7-\u09c8 \u09cb-\u09cc \u0bc1-\u0bc2 \u0bc6-\u0bc8 \u0bca-\u0bcc \u0c01-\u0c03 \u0c41-\u0c44]; +$Extend = [[:Grapheme_Extend = TRUE:] $VoiceMarks $cmcextend]; + +# +# Korean Syllable Definitions +# +$L = [:Hangul_Syllable_Type = L:]; +$V = [:Hangul_Syllable_Type = V:]; +$T = [:Hangul_Syllable_Type = T:]; + +$LV = [:Hangul_Syllable_Type = LV:]; +$LVT = [:Hangul_Syllable_Type = LVT:]; + +$HangulSyllable = $L+ | ($L* ($LV? $V+ | $LV | $LVT) $T*) | $T+; + +# +# Forward Break Rules +# +$CR $LF; +([^$Control] | $HangulSyllable) $Extend*; +.; + + +# +# Reverse Rule, back up to the beginning of some preceding grapheme cluster. +# +! ($Extend | $V | $T )* ($LF $CR | ($LV | $LVT)?$L* | .); |