icuSources/test/testdata/break_rules/grapheme.txt

   1 #
   2 # Copyright (c) 2016, International Business Machines Corporation and others. All Rights Reserved.
   3
   4 # file: grapheme.txt
   5 #
   6 # Reference Grapheme Break rules for intltest rbbi/RBBIMonkeyTest
   7 #
   8 #
   9 # Note: Rule syntax and the monkey test itself are still a work in progress.
  10 #       They are expected to change with review and the addition of support for rule tailoring.
  11
  12 type = grapheme;      # one of grapheme | word | line | sentence
  13 locale = en;
  14
  15 CR                 = [\u000d];
  16 LF                 = [\u000a];
  17
  18 Control            = [[\p{Grapheme_Cluster_Break = Control}]-[:Block=Tags:]];
  19 Extend             = [[\p{Grapheme_Cluster_Break = Extend}][:Block=Tags:]];
  20 ZWJ                = [\u200D];
  21 Regional_Indicator = [\p{Grapheme_Cluster_Break = Regional_Indicator}];
  22 Prepend            = [];
  23 SpacingMark        = [\p{Grapheme_Cluster_Break = SpacingMark}];
  24
  25 #
  26 # Korean Syllable Definitions
  27 #
  28 L       = [\p{Grapheme_Cluster_Break = L}];
  29 V       = [\p{Grapheme_Cluster_Break = V}];
  30 T       = [\p{Grapheme_Cluster_Break = T}];
  31 LV      = [\p{Grapheme_Cluster_Break = LV}];
  32 LVT     = [\p{Grapheme_Cluster_Break = LVT}];
  33
  34 # Emoji defintions
  35
  36 E_Base             = [\u261D\u26F9\u270A-\u270D\U0001F385\U0001F3C3-\U0001F3C4\U0001F3CA-\U0001F3CB\U0001F442-\U0001F443\U0001F446-\U0001F450\U0001F466-\U0001F469\U0001F46E\U0001F470-\U0001F478\U0001F47C\U0001F481-\U0001F483\U0001F485-\U0001F487\U0001F4AA\U0001F575\U0001F590\U0001F595-\U0001F596\U0001F645-\U0001F647\U0001F64B-\U0001F64F\U0001F6A3\U0001F6B4-\U0001F6B6\U0001F6C0\U0001F918];
  37 E_Modifier         = [\U0001F3FB-\U0001F3FF];
  38 GAZ                = [\u2640\u2642\u2764\U0001F308\U0001F466-\U0001F469\U0001F48B\U0001F5E8];
  39
  40 GB3:     CR LF;
  41 GB4:     (Control | CR | LF) ÷;
  42 GB5:     . ÷ (Control | CR | LF);
  43
  44 GB6:     L (L | V | LV | LVT);
  45 GB7:     (LV | V) (V | T);
  46 GB8:     (LVT | T) T;
  47
  48 GB10:    (E_Base | GAZ) Extend* E_Modifier;
  49 GB9:     . (Extend | ZWJ);
  50
  51 GB9a:    . SpacingMark;
  52 GB9b:    Prepend .;
  53 GB11:    ZWJ GAZ;
  54
  55 # Regional Indicators, split into pairs.
  56 #      Note that a pair of RIs that is not followed by a third RI will fall into
  57 #      the normal rules for Extend, etc.
  58 #
  59 GB12:  Regional_Indicator Regional_Indicator ÷ Regional_Indicator;
  60 GB13:  Regional_Indicator Regional_Indicator;
  61
  62 GB999:     . ÷;