xref: /aoo41x/main/i18npool/source/breakiterator/data/dict_word.txt (revision cdf0e10c)

#
#   Copyright (C) 2002-2003, International Business Machines Corporation and others.
#       All Rights Reserved.
#
#   file:  dict_word.txt
#
#   ICU Word Break Rules
#      See Unicode Standard Annex #29.
#      These rules are based on Version 4.0.0, dated 2003-04-17
#



####################################################################################
#
#  Character class definitions from TR 29
#
####################################################################################
$Katakana  = [[:Script = KATAKANA:] [:name = KATAKANA-HIRAGANA PROLONGED SOUND MARK:]
                                   [:name = HALFWIDTH KATAKANA-HIRAGANA PROLONGED SOUND MARK:]
                                   [:name = HALFWIDTH KATAKANA VOICED SOUND MARK:]
                                   [:name = HALFWIDTH KATAKANA SEMI-VOICED SOUND MARK:]];

$Ideographic = [:Ideographic:];
$Hangul = [:Script = HANGUL:];

$ALetter   = [\u0002 [:Alphabetic:] [:name= COMMERCIAL AT:] [:name= HEBREW PUNCTUATION GERESH:]
                           - $Ideographic
                           - $Katakana
                           - $Hangul
                           - [:Script = Thai:]
                           - [:Script = Lao:]
                           - [:Script = Hiragana:]];

$MidLetter = [[:name = APOSTROPHE:] [:name = GRAVE ACCENT:] \u0084 [:name = SOFT HYPHEN:] [:name = MIDDLE DOT:] [:name = GREEK TONOS:] [:name= FULL STOP:]
              [:name = HEBREW PUNCTUATION GERSHAYIM:] [:name = DOUBLE VERTICAL LINE:] [:name = LEFT SINGLE QUOTATION MARK:]
              [:name = RIGHT SINGLE QUOTATION MARK:] [:name = HYPHENATION POINT:] [:name = PRIME:]
              [:name = HYPHEN-MINUS:] ];

$SufixLetter = [:name= FULL STOP:];


$MidNum    = [[:LineBreak = Infix_Numeric:] [:name= COMMERCIAL AT:] \u0084 [:name = GREEK TONOS:] [:name = ARABIC DECIMAL SEPARATOR:]
             [:name = LEFT SINGLE QUOTATION MARK:] [:name = RIGHT SINGLE QUOTATION MARK:] [:name = SINGLE HIGH-REVERSED-9 QUOTATION MARK:]
             [:name = PRIME:]];
$Numeric   = [:LineBreak = Numeric:];


$TheZWSP = \u200b;

#
#  Character Class Definitions.
#    The names are those from TR29.
#
$CR         = \u000d;
$LF         = \u000a;
$Control    = [[[:Zl:] [:Zp:] [:Cc:] [:Cf:]] - $TheZWSP];
$Extend     = [[:Grapheme_Extend = TRUE:]];




####################################################################################
#
#  Word Break Rules.    Definitions and Rules specific to word break begin Here.
#
####################################################################################

$Format    = [[:Cf:] - $TheZWSP];



# Rule 3:  Treat a grapheme cluster as if it were a single character.
#          Hangul Syllables are easier to deal with here than they are in Grapheme Clusters
#          because we don't need to find the boundaries between adjacent syllables -
#          they won't be word boundaries.
#


#
#  "Extended"  definitions.  Grapheme Cluster + Format Chars, treated like the base char.
#
$ALetterEx    = $ALetter   $Extend*;
$NumericEx    = $Numeric   $Extend*;
$MidNumEx     = $MidNum    $Extend*;
$MidLetterEx  = $MidLetter $Extend*;
$SufixLetterEx= $SufixLetter $Extend*;
$KatakanaEx   = $Katakana  $Extend*;
$IdeographicEx= $Ideographic  $Extend*;
$HangulEx = $Hangul  $Extend*;
$FormatEx     = $Format    $Extend*;


#
#  Numbers.  Rules 8, 11, 12 form the TR.
#
$NumberSequence = $NumericEx ($FormatEx* $MidNumEx? $FormatEx* $NumericEx)*;
$NumberSequence {100};

#
#  Words.  Alpha-numerics.  Rule 5, 6, 7, 9, 10
#     - must include at least one letter.
#     - may include both letters and numbers.
#     - may include  MideLetter, MidNumber punctuation.
#
$LetterSequence = $ALetterEx ($FormatEx* $MidLetterEx? $FormatEx* $ALetterEx)*;     # rules #6, #7
($NumberSequence $FormatEx*)? $LetterSequence ($FormatEx* ($NumberSequence | $LetterSequence))* $SufixLetterEx? {200};

[[:P:][:S:]]*;

#
#  Do not break between Katakana.   Rule #13.
#
$KatakanaEx ($FormatEx* $KatakanaEx)* {300};
[:Hiragana:] $Extend* {300};

#
#  Ideographic Characters.  Stand by themselves as words.
#                           Separated from the "Everything Else" rule, below, only so that they
#                           can be tagged with a return value.   TODO:  is this what we want?
#
$IdeographicEx ($FormatEx* $IdeographicEx)* {400};
$HangulEx ($FormatEx* $HangulEx)* {400};

#
#  Everything Else, with no tag.
#                   Non-Control chars combine with $Extend (combining) chars.
#                   Controls are do not.
#
[^$Control [:Ideographic:]] $Extend*;
$CR $LF;

#
#  Reverse Rules.   Back up over any of the chars that can group together.
#                   (Reverse rules do not need to be exact; they can back up  too far,
#                   but must back up at least enough, and must stop on a boundary.)
#

# NonStarters are the set of all characters that can appear at the 2nd - nth position of
#    a word.   (They may also be the first.)   The reverse rule skips over these, until it
#    reaches something that can only be the start (and probably only) char in a "word".
#    A space or punctuation meets the test.
#
$NonStarters = [$Numeric $ALetter $Katakana $Ideographic $Hangul [:P:] [:S:] $MidLetter $MidNum $SufixLetter $Extend $Format];

#!.*;
! ($NonStarters* | \n \r) .;