icuSources/data/brkitr/char.txt

   1 #
   2 #   Copyright (C) 2002-2013, International Business Machines Corporation and others.
   3 #       All Rights Reserved.
   4 #
   5 #   file:  char.txt
   6 #
   7 #   ICU Character Break Rules, also known as Grapheme Cluster Boundaries
   8 #      See Unicode Standard Annex #29.
   9 #      These rules are based on UAX #29 Revision 20 for Unicode Version 6.2
  10 #
  11
  12 #
  13 #  Character Class Definitions.
  14 #
  15 $CR          = [\p{Grapheme_Cluster_Break = CR}];
  16 $LF          = [\p{Grapheme_Cluster_Break = LF}];
  17 $Control     = [\p{Grapheme_Cluster_Break = Control}];
  18 # TODO: Restore if the Prepend set becomes non-empty again: $Prepend     = [\p{Grapheme_Cluster_Break = Prepend}];
  19 $Extend      = [\p{Grapheme_Cluster_Break = Extend}];
  20 $SpacingMark = [\p{Grapheme_Cluster_Break = SpacingMark}];
  21
  22 $RI_A        = \U0001F1E6;  # Trail ERTU
  23 $RI_B        = \U0001F1E7;  # Trail EGR
  24 $RI_C        = \U0001F1E8;  # Trail AHLNZ
  25 $RI_D        = \U0001F1E9;  # Trail EK
  26 $RI_E        = \U0001F1EA;  # Trail GS
  27 $RI_F        = \U0001F1EB;  # Trail IR
  28 $RI_G        = \U0001F1EC;  # Trail BR
  29 $RI_H        = \U0001F1ED;  # Trail KU
  30 $RI_I        = \U0001F1EE;  # Trail DLNT
  31 $RI_J        = \U0001F1EF;  # Trail OP
  32 $RI_K        = \U0001F1F0;  # Trail R
  33 $RI_L        = \U0001F1F1;  # Trail B
  34 $RI_M        = \U0001F1F2;  # Trail OXY
  35 $RI_N        = \U0001F1F3;  # Trail LO
  36 $RI_P        = \U0001F1F5;  # Trail LT
  37 $RI_R        = \U0001F1F7;  # Trail OU
  38 $RI_S        = \U0001F1F8;  # Trail AEGK
  39 $RI_T        = \U0001F1F9;  # Trail HRW
  40 $RI_U        = \U0001F1FA;  # Trail AS
  41 $RI_V        = \U0001F1FB;      # Trail N
  42
  43 $RI_A_End    = [\U0001F1EA \U0001F1F7 \U0001F1F9 \U0001F1FA];   # ERTU
  44 $RI_B_End    = [\U0001F1EA \U0001F1EC \U0001F1F7];              # EGR
  45 $RI_C_End    = [\U0001F1E6 \U0001F1ED \U0001F1F1 \U0001F1F3 \U0001F1FF]; # AHLNZ
  46 $RI_D_End    = [\U0001F1EA \U0001F1F0];                         # EK
  47 $RI_E_End    = [\U0001F1EC \U0001F1F8];                         # GS
  48 $RI_F_End    = [\U0001F1EE \U0001F1F7];                         # IR
  49 $RI_G_End    = [\U0001F1E7 \U0001F1F7];                         # BR
  50 $RI_H_End    = [\U0001F1F0 \U0001F1FA];                         # KU
  51 $RI_I_End    = [\U0001F1E9 \U0001F1F1 \U0001F1F3 \U0001F1F9];   # DLNT
  52 $RI_J_End    = [\U0001F1F4 \U0001F1F5];                         # OP
  53 $RI_K_End    = \U0001F1F7;                                      # R
  54 $RI_L_End    = \U0001F1E7;                                      # B
  55 $RI_M_End    = [\U0001F1F4 \U0001F1FD \U0001F1FE];              # OXY
  56 $RI_N_End    = [\U0001F1F1 \U0001F1F4];                         # LO
  57 $RI_P_End    = [\U0001F1F1 \U0001F1F9];                         # LT
  58 $RI_R_End    = [\U0001F1F4 \U0001F1FA];                         # OU
  59 $RI_S_End    = [\U0001F1E6 \U0001F1EA \U0001F1EC \U0001F1F0];   # AEGK
  60 $RI_T_End    = [\U0001F1ED \U0001F1F7 \U0001F1FC];              # HRW
  61 $RI_U_End    = [\U0001F1E6 \U0001F1F8];                         # AS
  62 $RI_V_End    = \U0001F1F3;                                      # N
  63
  64 #
  65 # Korean Syllable Definitions
  66 #
  67 $L       = [\p{Grapheme_Cluster_Break = L}];
  68 $V       = [\p{Grapheme_Cluster_Break = V}];
  69 $T       = [\p{Grapheme_Cluster_Break = T}];
  70
  71 $LV      = [\p{Grapheme_Cluster_Break = LV}];
  72 $LVT     = [\p{Grapheme_Cluster_Break = LVT}];
  73
  74
  75 ## -------------------------------------------------
  76 !!chain;
  77
  78 !!forward;
  79
  80 $CR $LF;
  81
  82 $L ($L | $V | $LV | $LVT);
  83 ($LV | $V) ($V | $T);
  84 ($LVT | $T) $T;
  85
  86 $RI_A $RI_A_End;
  87 $RI_B $RI_B_End;
  88 $RI_C $RI_C_End;
  89 $RI_D $RI_D_End;
  90 $RI_E $RI_E_End;
  91 $RI_F $RI_F_End;
  92 $RI_G $RI_G_End;
  93 $RI_H $RI_H_End;
  94 $RI_I $RI_I_End;
  95 $RI_J $RI_J_End;
  96 $RI_K $RI_K_End;
  97 $RI_L $RI_L_End;
  98 $RI_M $RI_M_End;
  99 $RI_N $RI_N_End;
 100 $RI_P $RI_P_End;
 101 $RI_R $RI_R_End;
 102 $RI_S $RI_S_End;
 103 $RI_T $RI_T_End;
 104 $RI_U $RI_U_End;
 105 $RI_V $RI_V_End;
 106
 107 [^$Control $CR $LF] $Extend;
 108
 109 [^$Control $CR $LF] $SpacingMark;
 110 # TODO: Restore if the Prepend set becomes non-empty again: $Prepend [^$Control $CR $LF];
 111
 112
 113 ## -------------------------------------------------
 114
 115 !!reverse;
 116 $LF $CR;
 117 ($L | $V | $LV | $LVT) $L;
 118 ($V | $T) ($LV | $V);
 119 $T ($LVT | $T);
 120
 121 $RI_A_End $RI_A;
 122 $RI_B_End $RI_B;
 123 $RI_C_End $RI_C;
 124 $RI_D_End $RI_D;
 125 $RI_E_End $RI_E;
 126 $RI_F_End $RI_F;
 127 $RI_G_End $RI_G;
 128 $RI_H_End $RI_H;
 129 $RI_I_End $RI_I;
 130 $RI_J_End $RI_J;
 131 $RI_K_End $RI_K;
 132 $RI_L_End $RI_L;
 133 $RI_M_End $RI_M;
 134 $RI_N_End $RI_N;
 135 $RI_P_End $RI_P;
 136 $RI_R_End $RI_R;
 137 $RI_S_End $RI_S;
 138 $RI_T_End $RI_T;
 139 $RI_U_End $RI_U;
 140 $RI_V_End $RI_V;
 141
 142 $Extend      [^$Control $CR $LF];
 143 $SpacingMark [^$Control $CR $LF];
 144 # TODO: Restore if the Prepend set becomes non-empty again: [^$Control $CR $LF] $Prepend;
 145
 146
 147 ## -------------------------------------------------
 148 #  We don't logically need safe char break rules, but if we don't provide any at all
 149 #  the engine for preceding() and following() will fall back to the
 150 #  old style inefficient algorithm.
 151
 152 !!safe_reverse;
 153 $LF $CR;
 154
 155 ## -------------------------------------------------
 156
 157 !!safe_forward;
 158 $CR $LF;
 159