1 /************************************************************************/ 2 /* */ 3 /* UCD General Categories. */ 4 /* */ 5 /************************************************************************/ 6 7 /** 8 * To build the masks. Do not directly refer to these values. 9 */ 10 typedef enum UCDGeneralCategoryGroup 11 { 12 UCDgc_L= 1, /* Letter */ 13 UCDgc_M, /* Mark */ 14 UCDgc_N, /* Number */ 15 UCDgc_P, /* Punctuation */ 16 UCDgc_S, /* Symbol */ 17 UCDgc_Z, /* Separator */ 18 UCDgc_C, /* Other */ 19 20 UCDgc__COUNT 21 } UCDGeneralCategoryGroup; 22 23 # define UCDgc_X_SHIFT 5 24 # define UCDgc_X_MASK 0xe0 25 # define UCDgc_XX_MASK 0xf0 26 27 typedef enum UCDGeneralCategory 28 { 29 /********** 30 * Offset for the letters. Do not directly refer to this value. 31 * Note that the order is different from that in the Unicode 32 * standard to reserve a bit for the LC mask. 33 */ 34 UCDgc_L_MASK= ( UCDgc_L << UCDgc_X_SHIFT ), 35 /** 36 * Modifier_Letter: a modifier letter 37 */ 38 UCDgc_Lm, 39 /** 40 * Other_Letter: other letters, including syllables and ideographs 41 */ 42 UCDgc_Lo, 43 /********** 44 * Offset for the cased letters. Do not directly refer to 45 * this value. 46 */ 47 UCDgc_LC_MASK= ( ( UCDgc_L << UCDgc_X_SHIFT ) | 0x10 ), 48 /** 49 * Uppercase_Letter: an uppercase letter 50 */ 51 UCDgc_Lu, 52 /** 53 * Lowercase_Letter: a lowercase letter 54 */ 55 UCDgc_Ll, 56 /** 57 * Titlecase_Letter: 58 * a digraphic character, with first part uppercase 59 */ 60 UCDgc_Lt, 61 62 /********** 63 * Offset for the marks. Do not directly refer to this value. 64 */ 65 UCDgc_M_MASK= ( UCDgc_M << UCDgc_X_SHIFT ), 66 /** 67 * Nonspacing_Mark: a nonspacing combining mark (zero advance width) 68 */ 69 UCDgc_Mn, 70 /** 71 * Spacing_Mark: a spacing combining mark (positive advance width) 72 */ 73 UCDgc_Mc, 74 /** 75 * Enclosing_Mark: an enclosing combining mark 76 */ 77 UCDgc_Me, 78 79 /********** 80 * Offset for the numbers. Do not directly refer to this value. 81 */ 82 UCDgc_N_MASK= ( UCDgc_N << UCDgc_X_SHIFT ), 83 /** 84 * Decimal_Number: a decimal digit 85 */ 86 UCDgc_Nd, 87 /** 88 * Letter_Number: a letterlike numeric character 89 */ 90 UCDgc_Nl, 91 /** 92 * Other_Number: a numeric character of other type 93 */ 94 UCDgc_No, 95 96 /********** 97 * Offset for the punctuation. Do not directly refer to this value. 98 */ 99 UCDgc_P_MASK= ( UCDgc_P << UCDgc_X_SHIFT ), 100 /** 101 * Connector_Punctuation: a connecting punctuation mark, like a tie 102 */ 103 UCDgc_Pc, 104 /** 105 * Dash_Punctuation: a dash or hyphen punctuation mark 106 */ 107 UCDgc_Pd, 108 /** 109 * Open_Punctuation: an opening punctuation mark (of a pair) 110 */ 111 UCDgc_Ps, 112 /** 113 * Close_Punctuation: a closing punctuation mark (of a pair) 114 */ 115 UCDgc_Pe, 116 /** 117 * Initial_Punctuation: an initial quotation mark 118 */ 119 UCDgc_Pi, 120 /** 121 * Final_Punctuation: a final quotation mark 122 */ 123 UCDgc_Pf, 124 /** 125 * Other_Punctuation: a punctuation mark of other type 126 */ 127 UCDgc_Po, 128 129 /********** 130 * Offset for the signs. Do not directly refer to this value. 131 */ 132 UCDgc_S_MASK= ( UCDgc_S << UCDgc_X_SHIFT ), 133 /** 134 * Math_Symbol: a symbol of mathematical use 135 */ 136 UCDgc_Sm, 137 /** 138 * Currency_Symbol: a currency sign 139 */ 140 UCDgc_Sc, 141 /** 142 * Modifier_Symbol: a non-letterlike modifier symbol 143 */ 144 UCDgc_Sk, 145 /** 146 * Other_Symbol: a symbol of other type 147 */ 148 UCDgc_So, 149 150 /********** 151 * Offset for the spaces. Do not directly refer to this value. 152 */ 153 UCDgc_Z_MASK= ( UCDgc_Z << UCDgc_X_SHIFT ), 154 /** 155 * Space_Separator: a space character (of various non-zero widths) 156 */ 157 UCDgc_Zs, 158 /** 159 * Line_Separator: U+2028 LINE SEPARATOR only 160 */ 161 UCDgc_Zl, 162 /** 163 * Paragraph_Separator: U+2029 PARAGRAPH SEPARATOR only 164 */ 165 UCDgc_Zp, 166 167 /********** 168 * Offset for the other categories. Do not directly refer to 169 * this value. 170 */ 171 UCDgc_C_MASK= ( UCDgc_C << UCDgc_X_SHIFT ), 172 /** 173 * Control: a C0 or C1 control code 174 */ 175 UCDgc_Cc, 176 /** 177 * Format: a format control character 178 */ 179 UCDgc_Cf, 180 /** 181 * Surrogate: a surrogate code point 182 */ 183 UCDgc_Cs, 184 /** 185 * Private_Use: a private-use character 186 */ 187 UCDgc_Co, 188 /** 189 * Unassigned: a reserved unassigned code point or a noncharacter 190 */ 191 UCDgc_Cn 192 } UCDGeneralCategory; 193 194 /************************************************************************/ 195 /* */ 196 /* Routine declarations */ 197 /* */ 198 /************************************************************************/ 199 200 extern int ucdGeneralCategory( int sym ); 201 extern const char * ucdGeneralCategoryStr( int sym ); 202 203 /************************************************************************/ 204 /* */ 205 /* Probing macros */ 206 /* */ 207 /************************************************************************/ 208 209 # define ucdIsL(c) ( ( ucdGeneralCategory((c)) & UCDgc_X_MASK ) == UCDgc_L_MASK ) 210 # define ucdIsLm(c) ( ucdGeneralCategory((c)) == UCDgc_Lm ) 211 # define ucdIsLo(c) ( ucdGeneralCategory((c)) == UCDgc_Lo ) 212 # define ucdIsLC(c) ( ( ucdGeneralCategory((c)) & UCDgc_XX_MASK ) == UCDgc_LC_MASK ) 213 # define ucdIsLu(c) ( ucdGeneralCategory((c)) == UCDgc_Lu ) 214 # define ucdIsLl(c) ( ucdGeneralCategory((c)) == UCDgc_Ll ) 215 # define ucdIsLt(c) ( ucdGeneralCategory((c)) == UCDgc_Lt ) 216 217 # define ucdIsM(c) ( ( ucdGeneralCategory((c)) & UCDgc_X_MASK ) == UCDgc_M_MASK ) 218 # define ucdIsMn(c) ( ucdGeneralCategory((c)) == UCDgc_Mn ) 219 # define ucdIsMc(c) ( ucdGeneralCategory((c)) == UCDgc_Mc ) 220 # define ucdIsMe(c) ( ucdGeneralCategory((c)) == UCDgc_Me ) 221 222 # define ucdIsN(c) ( ( ucdGeneralCategory((c)) & UCDgc_X_MASK ) == UCDgc_N_MASK ) 223 # define ucdIsNd(c) ( ucdGeneralCategory((c)) == UCDgc_Nd ) 224 # define ucdIsNl(c) ( ucdGeneralCategory((c)) == UCDgc_Nl ) 225 # define ucdIsNo(c) ( ucdGeneralCategory((c)) == UCDgc_No ) 226 227 # define ucdIsP(c) ( ( ucdGeneralCategory((c)) & UCDgc_X_MASK ) == UCDgc_P_MASK ) 228 # define ucdIsPc(c) ( ucdGeneralCategory((c)) == UCDgc_Pc ) 229 # define ucdIsPd(c) ( ucdGeneralCategory((c)) == UCDgc_Pd ) 230 # define ucdIsPs(c) ( ucdGeneralCategory((c)) == UCDgc_Ps ) 231 # define ucdIsPe(c) ( ucdGeneralCategory((c)) == UCDgc_Pe ) 232 # define ucdIsPi(c) ( ucdGeneralCategory((c)) == UCDgc_Pi ) 233 # define ucdIsPf(c) ( ucdGeneralCategory((c)) == UCDgc_Pf ) 234 # define ucdIsPo(c) ( ucdGeneralCategory((c)) == UCDgc_Po ) 235 236 # define ucdIsS(c) ( ( ucdGeneralCategory((c)) & UCDgc_X_MASK ) == UCDgc_S_MASK ) 237 # define ucdIsSm(c) ( ucdGeneralCategory((c)) == UCDgc_Sm ) 238 # define ucdIsSc(c) ( ucdGeneralCategory((c)) == UCDgc_Sc ) 239 # define ucdIsSk(c) ( ucdGeneralCategory((c)) == UCDgc_Sk ) 240 # define ucdIsSo(c) ( ucdGeneralCategory((c)) == UCDgc_So ) 241 242 # define ucdIsZ(c) ( ( ucdGeneralCategory((c)) & UCDgc_X_MASK ) == UCDgc_Z_MASK ) 243 # define ucdIsZs(c) ( ucdGeneralCategory((c)) == UCDgc_Zs ) 244 # define ucdIsZl(c) ( ucdGeneralCategory((c)) == UCDgc_Zl ) 245 # define ucdIsZp(c) ( ucdGeneralCategory((c)) == UCDgc_Zp ) 246 247 # define ucdIsC(c) ( ( ucdGeneralCategory((c)) & UCDgc_X_MASK ) == UCDgc_C_MASK ) 248 # define ucdIsCc(c) ( ucdGeneralCategory((c)) == UCDgc_Cc ) 249 # define ucdIsCf(c) ( ucdGeneralCategory((c)) == UCDgc_Cf ) 250 # define ucdIsCs(c) ( ucdGeneralCategory((c)) == UCDgc_Cs ) 251 # define ucdIsCo(c) ( ucdGeneralCategory((c)) == UCDgc_Co ) 252 # define ucdIsCn(c) ( ucdGeneralCategory((c)) == UCDgc_Cn ) 253 254