Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 1 | /* |
| 2 | Unicode character type helpers. |
| 3 | |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 4 | Written by Marc-Andre Lemburg (mal@lemburg.com). |
| 5 | Modified for Python 2.0 by Fredrik Lundh (fredrik@pythonware.com) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 6 | |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 7 | Copyright (c) Corporation for National Research Initiatives. |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 8 | |
| 9 | */ |
| 10 | |
| 11 | #include "Python.h" |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 12 | #include "unicodeobject.h" |
| 13 | |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 14 | #define ALPHA_MASK 0x01 |
| 15 | #define DECIMAL_MASK 0x02 |
| 16 | #define DIGIT_MASK 0x04 |
| 17 | #define LOWER_MASK 0x08 |
| 18 | #define LINEBREAK_MASK 0x10 |
| 19 | #define SPACE_MASK 0x20 |
| 20 | #define TITLE_MASK 0x40 |
| 21 | #define UPPER_MASK 0x80 |
Jack Jansen | 56cdce3 | 2000-07-06 13:57:38 +0000 | [diff] [blame] | 22 | |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 23 | typedef struct { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 24 | const Py_UNICODE upper; |
| 25 | const Py_UNICODE lower; |
| 26 | const Py_UNICODE title; |
| 27 | const unsigned char decimal; |
| 28 | const unsigned char digit; |
Hye-Shik Chang | 974ed7c | 2004-06-02 16:49:17 +0000 | [diff] [blame] | 29 | const unsigned short flags; |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 30 | } _PyUnicode_TypeRecord; |
| 31 | |
| 32 | #include "unicodetype_db.h" |
| 33 | |
| 34 | static const _PyUnicode_TypeRecord * |
Fredrik Lundh | ee13dba | 2001-06-26 20:36:12 +0000 | [diff] [blame] | 35 | gettyperecord(Py_UNICODE code) |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 36 | { |
| 37 | int index; |
| 38 | |
Hye-Shik Chang | 7db07e6 | 2003-12-29 01:36:01 +0000 | [diff] [blame] | 39 | #ifdef Py_UNICODE_WIDE |
Martin v. Löwis | 9def6a3 | 2002-10-18 16:11:54 +0000 | [diff] [blame] | 40 | if (code >= 0x110000) |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 41 | index = 0; |
Hye-Shik Chang | 7db07e6 | 2003-12-29 01:36:01 +0000 | [diff] [blame] | 42 | else |
| 43 | #endif |
| 44 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 45 | index = index1[(code>>SHIFT)]; |
| 46 | index = index2[(index<<SHIFT)+(code&((1<<SHIFT)-1))]; |
| 47 | } |
Fredrik Lundh | ee13dba | 2001-06-26 20:36:12 +0000 | [diff] [blame] | 48 | |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 49 | return &_PyUnicode_TypeRecords[index]; |
| 50 | } |
Jack Jansen | 56cdce3 | 2000-07-06 13:57:38 +0000 | [diff] [blame] | 51 | |
Marc-André Lemburg | 2cb94ab | 2005-10-20 19:06:35 +0000 | [diff] [blame] | 52 | /* Returns 1 for Unicode characters having the category 'Zl', 'Zp' or |
| 53 | type 'B', 0 otherwise. */ |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 54 | |
Marc-André Lemburg | 2cb94ab | 2005-10-20 19:06:35 +0000 | [diff] [blame] | 55 | int _PyUnicode_IsLinebreak(register const Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 56 | { |
Marc-André Lemburg | 2cb94ab | 2005-10-20 19:06:35 +0000 | [diff] [blame] | 57 | switch (ch) { |
| 58 | case 0x000A: /* LINE FEED */ |
| 59 | case 0x000D: /* CARRIAGE RETURN */ |
| 60 | case 0x001C: /* FILE SEPARATOR */ |
| 61 | case 0x001D: /* GROUP SEPARATOR */ |
| 62 | case 0x001E: /* RECORD SEPARATOR */ |
| 63 | case 0x0085: /* NEXT LINE */ |
| 64 | case 0x2028: /* LINE SEPARATOR */ |
| 65 | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 66 | return 1; |
| 67 | default: |
| 68 | return 0; |
| 69 | } |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 70 | } |
| 71 | |
| 72 | /* Returns the titlecase Unicode characters corresponding to ch or just |
| 73 | ch if no titlecase mapping is known. */ |
| 74 | |
Martin v. Löwis | ce9b5a5 | 2001-06-27 06:28:56 +0000 | [diff] [blame] | 75 | Py_UNICODE _PyUnicode_ToTitlecase(register Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 76 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 77 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
Martin v. Löwis | edf368c | 2002-10-18 16:40:36 +0000 | [diff] [blame] | 78 | int delta; |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 79 | |
| 80 | if (ctype->title) |
Martin v. Löwis | edf368c | 2002-10-18 16:40:36 +0000 | [diff] [blame] | 81 | delta = ctype->title; |
Martin v. Löwis | ce9b5a5 | 2001-06-27 06:28:56 +0000 | [diff] [blame] | 82 | else |
Martin v. Löwis | edf368c | 2002-10-18 16:40:36 +0000 | [diff] [blame] | 83 | delta = ctype->upper; |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 84 | |
Martin v. Löwis | edf368c | 2002-10-18 16:40:36 +0000 | [diff] [blame] | 85 | if (delta >= 32768) |
| 86 | delta -= 65536; |
| 87 | |
| 88 | return ch + delta; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 89 | } |
| 90 | |
| 91 | /* Returns 1 for Unicode characters having the category 'Lt', 0 |
| 92 | otherwise. */ |
| 93 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 94 | int _PyUnicode_IsTitlecase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 95 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 96 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
| 97 | |
| 98 | return (ctype->flags & TITLE_MASK) != 0; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 99 | } |
| 100 | |
| 101 | /* Returns the integer decimal (0-9) for Unicode characters having |
| 102 | this property, -1 otherwise. */ |
| 103 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 104 | int _PyUnicode_ToDecimalDigit(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 105 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 106 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
| 107 | |
| 108 | return (ctype->flags & DECIMAL_MASK) ? ctype->decimal : -1; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 109 | } |
| 110 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 111 | int _PyUnicode_IsDecimalDigit(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 112 | { |
| 113 | if (_PyUnicode_ToDecimalDigit(ch) < 0) |
| 114 | return 0; |
| 115 | return 1; |
| 116 | } |
| 117 | |
| 118 | /* Returns the integer digit (0-9) for Unicode characters having |
| 119 | this property, -1 otherwise. */ |
| 120 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 121 | int _PyUnicode_ToDigit(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 122 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 123 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
| 124 | |
| 125 | return (ctype->flags & DIGIT_MASK) ? ctype->digit : -1; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 126 | } |
| 127 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 128 | int _PyUnicode_IsDigit(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 129 | { |
| 130 | if (_PyUnicode_ToDigit(ch) < 0) |
| 131 | return 0; |
| 132 | return 1; |
| 133 | } |
| 134 | |
| 135 | /* Returns the numeric value as double for Unicode characters having |
| 136 | this property, -1.0 otherwise. */ |
| 137 | |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 138 | /* TODO: replace with unicodetype_db.h table */ |
| 139 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 140 | double _PyUnicode_ToNumeric(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 141 | { |
| 142 | switch (ch) { |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 143 | case 0x0F33: |
| 144 | return (double) -1 / 2; |
| 145 | case 0x17F0: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 146 | case 0x3007: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 147 | #ifdef Py_UNICODE_WIDE |
| 148 | case 0x1018A: |
| 149 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 150 | return (double) 0; |
| 151 | case 0x09F4: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 152 | case 0x17F1: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 153 | case 0x215F: |
| 154 | case 0x2160: |
| 155 | case 0x2170: |
| 156 | case 0x3021: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 157 | case 0x3192: |
| 158 | case 0x3220: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 159 | case 0x3280: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 160 | #ifdef Py_UNICODE_WIDE |
| 161 | case 0x10107: |
| 162 | case 0x10142: |
| 163 | case 0x10158: |
| 164 | case 0x10159: |
| 165 | case 0x1015A: |
| 166 | case 0x10320: |
| 167 | case 0x103D1: |
| 168 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 169 | return (double) 1; |
| 170 | case 0x00BD: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 171 | case 0x0F2A: |
| 172 | case 0x2CFD: |
| 173 | #ifdef Py_UNICODE_WIDE |
| 174 | case 0x10141: |
| 175 | case 0x10175: |
| 176 | case 0x10176: |
| 177 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 178 | return (double) 1 / 2; |
| 179 | case 0x2153: |
| 180 | return (double) 1 / 3; |
| 181 | case 0x00BC: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 182 | #ifdef Py_UNICODE_WIDE |
| 183 | case 0x10140: |
| 184 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 185 | return (double) 1 / 4; |
| 186 | case 0x2155: |
| 187 | return (double) 1 / 5; |
| 188 | case 0x2159: |
| 189 | return (double) 1 / 6; |
| 190 | case 0x215B: |
| 191 | return (double) 1 / 8; |
| 192 | case 0x0BF0: |
| 193 | case 0x1372: |
| 194 | case 0x2169: |
| 195 | case 0x2179: |
| 196 | case 0x2469: |
| 197 | case 0x247D: |
| 198 | case 0x2491: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 199 | case 0x24FE: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 200 | case 0x277F: |
| 201 | case 0x2789: |
| 202 | case 0x2793: |
| 203 | case 0x3038: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 204 | case 0x3229: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 205 | case 0x3289: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 206 | #ifdef Py_UNICODE_WIDE |
| 207 | case 0x10110: |
| 208 | case 0x10149: |
| 209 | case 0x10150: |
| 210 | case 0x10157: |
| 211 | case 0x10160: |
| 212 | case 0x10161: |
| 213 | case 0x10162: |
| 214 | case 0x10163: |
| 215 | case 0x10164: |
| 216 | case 0x10322: |
| 217 | case 0x103D3: |
| 218 | case 0x10A44: |
| 219 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 220 | return (double) 10; |
| 221 | case 0x0BF1: |
| 222 | case 0x137B: |
| 223 | case 0x216D: |
| 224 | case 0x217D: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 225 | #ifdef Py_UNICODE_WIDE |
| 226 | case 0x10119: |
| 227 | case 0x1014B: |
| 228 | case 0x10152: |
| 229 | case 0x1016A: |
| 230 | case 0x103D5: |
| 231 | case 0x10A46: |
| 232 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 233 | return (double) 100; |
| 234 | case 0x0BF2: |
| 235 | case 0x216F: |
| 236 | case 0x217F: |
| 237 | case 0x2180: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 238 | #ifdef Py_UNICODE_WIDE |
| 239 | case 0x10122: |
| 240 | case 0x1014D: |
| 241 | case 0x10154: |
| 242 | case 0x10171: |
| 243 | case 0x10A47: |
| 244 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 245 | return (double) 1000; |
| 246 | case 0x137C: |
| 247 | case 0x2182: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 248 | #ifdef Py_UNICODE_WIDE |
| 249 | case 0x1012B: |
| 250 | case 0x10155: |
| 251 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 252 | return (double) 10000; |
| 253 | case 0x216A: |
| 254 | case 0x217A: |
| 255 | case 0x246A: |
| 256 | case 0x247E: |
| 257 | case 0x2492: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 258 | case 0x24EB: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 259 | return (double) 11; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 260 | case 0x0F2F: |
| 261 | return (double) 11 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 262 | case 0x216B: |
| 263 | case 0x217B: |
| 264 | case 0x246B: |
| 265 | case 0x247F: |
| 266 | case 0x2493: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 267 | case 0x24EC: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 268 | return (double) 12; |
| 269 | case 0x246C: |
| 270 | case 0x2480: |
| 271 | case 0x2494: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 272 | case 0x24ED: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 273 | return (double) 13; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 274 | case 0x0F30: |
| 275 | return (double) 13 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 276 | case 0x246D: |
| 277 | case 0x2481: |
| 278 | case 0x2495: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 279 | case 0x24EE: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 280 | return (double) 14; |
| 281 | case 0x246E: |
| 282 | case 0x2482: |
| 283 | case 0x2496: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 284 | case 0x24EF: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 285 | return (double) 15; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 286 | case 0x0F31: |
| 287 | return (double) 15 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 288 | case 0x09F9: |
| 289 | case 0x246F: |
| 290 | case 0x2483: |
| 291 | case 0x2497: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 292 | case 0x24F0: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 293 | return (double) 16; |
| 294 | case 0x16EE: |
| 295 | case 0x2470: |
| 296 | case 0x2484: |
| 297 | case 0x2498: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 298 | case 0x24F1: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 299 | return (double) 17; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 300 | case 0x0F32: |
| 301 | return (double) 17 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 302 | case 0x16EF: |
| 303 | case 0x2471: |
| 304 | case 0x2485: |
| 305 | case 0x2499: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 306 | case 0x24F2: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 307 | return (double) 18; |
| 308 | case 0x16F0: |
| 309 | case 0x2472: |
| 310 | case 0x2486: |
| 311 | case 0x249A: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 312 | case 0x24F3: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 313 | return (double) 19; |
| 314 | case 0x09F5: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 315 | case 0x17F2: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 316 | case 0x2161: |
| 317 | case 0x2171: |
| 318 | case 0x3022: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 319 | case 0x3193: |
| 320 | case 0x3221: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 321 | case 0x3281: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 322 | #ifdef Py_UNICODE_WIDE |
| 323 | case 0x10108: |
| 324 | case 0x1015B: |
| 325 | case 0x1015C: |
| 326 | case 0x1015D: |
| 327 | case 0x1015E: |
| 328 | case 0x103D2: |
| 329 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 330 | return (double) 2; |
| 331 | case 0x2154: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 332 | #ifdef Py_UNICODE_WIDE |
| 333 | case 0x10177: |
| 334 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 335 | return (double) 2 / 3; |
| 336 | case 0x2156: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 337 | return (double) 2 / 5; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 338 | case 0x1373: |
| 339 | case 0x2473: |
| 340 | case 0x2487: |
| 341 | case 0x249B: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 342 | case 0x24F4: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 343 | case 0x3039: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 344 | #ifdef Py_UNICODE_WIDE |
| 345 | case 0x10111: |
| 346 | case 0x103D4: |
| 347 | case 0x10A45: |
| 348 | #endif |
| 349 | return (double) 20; |
| 350 | #ifdef Py_UNICODE_WIDE |
| 351 | case 0x1011A: |
| 352 | return (double) 200; |
| 353 | case 0x10123: |
| 354 | return (double) 2000; |
| 355 | case 0x1012C: |
| 356 | return (double) 20000; |
| 357 | #endif |
| 358 | case 0x3251: |
| 359 | return (double) 21; |
| 360 | case 0x3252: |
| 361 | return (double) 22; |
| 362 | case 0x3253: |
| 363 | return (double) 23; |
| 364 | case 0x3254: |
| 365 | return (double) 24; |
| 366 | case 0x3255: |
| 367 | return (double) 25; |
| 368 | case 0x3256: |
| 369 | return (double) 26; |
| 370 | case 0x3257: |
| 371 | return (double) 27; |
| 372 | case 0x3258: |
| 373 | return (double) 28; |
| 374 | case 0x3259: |
| 375 | return (double) 29; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 376 | case 0x09F6: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 377 | case 0x17F3: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 378 | case 0x2162: |
| 379 | case 0x2172: |
| 380 | case 0x3023: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 381 | case 0x3194: |
| 382 | case 0x3222: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 383 | case 0x3282: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 384 | #ifdef Py_UNICODE_WIDE |
| 385 | case 0x10109: |
| 386 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 387 | return (double) 3; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 388 | case 0x0F2B: |
| 389 | return (double) 3 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 390 | case 0x00BE: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 391 | #ifdef Py_UNICODE_WIDE |
| 392 | case 0x10178: |
| 393 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 394 | return (double) 3 / 4; |
| 395 | case 0x2157: |
| 396 | return (double) 3 / 5; |
| 397 | case 0x215C: |
| 398 | return (double) 3 / 8; |
| 399 | case 0x1374: |
| 400 | case 0x303A: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 401 | case 0x325A: |
| 402 | #ifdef Py_UNICODE_WIDE |
| 403 | case 0x10112: |
| 404 | case 0x10165: |
| 405 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 406 | return (double) 30; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 407 | #ifdef Py_UNICODE_WIDE |
| 408 | case 0x1011B: |
| 409 | case 0x1016B: |
| 410 | return (double) 300; |
| 411 | case 0x10124: |
| 412 | return (double) 3000; |
| 413 | case 0x1012D: |
| 414 | return (double) 30000; |
| 415 | #endif |
| 416 | case 0x325B: |
| 417 | return (double) 31; |
| 418 | case 0x325C: |
| 419 | return (double) 32; |
| 420 | case 0x325D: |
| 421 | return (double) 33; |
| 422 | case 0x325E: |
| 423 | return (double) 34; |
| 424 | case 0x325F: |
| 425 | return (double) 35; |
| 426 | case 0x32B1: |
| 427 | return (double) 36; |
| 428 | case 0x32B2: |
| 429 | return (double) 37; |
| 430 | case 0x32B3: |
| 431 | return (double) 38; |
| 432 | case 0x32B4: |
| 433 | return (double) 39; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 434 | case 0x09F7: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 435 | case 0x17F4: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 436 | case 0x2163: |
| 437 | case 0x2173: |
| 438 | case 0x3024: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 439 | case 0x3195: |
| 440 | case 0x3223: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 441 | case 0x3283: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 442 | #ifdef Py_UNICODE_WIDE |
| 443 | case 0x1010A: |
| 444 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 445 | return (double) 4; |
| 446 | case 0x2158: |
| 447 | return (double) 4 / 5; |
| 448 | case 0x1375: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 449 | case 0x32B5: |
| 450 | #ifdef Py_UNICODE_WIDE |
| 451 | case 0x10113: |
| 452 | #endif |
| 453 | return (double) 40; |
| 454 | #ifdef Py_UNICODE_WIDE |
| 455 | case 0x1011C: |
| 456 | return (double) 400; |
| 457 | case 0x10125: |
| 458 | return (double) 4000; |
| 459 | case 0x1012E: |
| 460 | return (double) 40000; |
| 461 | #endif |
| 462 | case 0x32B6: |
| 463 | return (double) 41; |
| 464 | case 0x32B7: |
| 465 | return (double) 42; |
| 466 | case 0x32B8: |
| 467 | return (double) 43; |
| 468 | case 0x32B9: |
| 469 | return (double) 44; |
| 470 | case 0x32BA: |
| 471 | return (double) 45; |
| 472 | case 0x32BB: |
| 473 | return (double) 46; |
| 474 | case 0x32BC: |
| 475 | return (double) 47; |
| 476 | case 0x32BD: |
| 477 | return (double) 48; |
| 478 | case 0x32BE: |
| 479 | return (double) 49; |
| 480 | case 0x17F5: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 481 | case 0x2164: |
| 482 | case 0x2174: |
| 483 | case 0x3025: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 484 | case 0x3224: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 485 | case 0x3284: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 486 | #ifdef Py_UNICODE_WIDE |
| 487 | case 0x1010B: |
| 488 | case 0x10143: |
| 489 | case 0x10148: |
| 490 | case 0x1014F: |
| 491 | case 0x1015F: |
| 492 | case 0x10173: |
| 493 | case 0x10321: |
| 494 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 495 | return (double) 5; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 496 | case 0x0F2C: |
| 497 | return (double) 5 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 498 | case 0x215A: |
| 499 | return (double) 5 / 6; |
| 500 | case 0x215D: |
| 501 | return (double) 5 / 8; |
| 502 | case 0x1376: |
| 503 | case 0x216C: |
| 504 | case 0x217C: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 505 | case 0x32BF: |
| 506 | #ifdef Py_UNICODE_WIDE |
| 507 | case 0x10114: |
| 508 | case 0x10144: |
| 509 | case 0x1014A: |
| 510 | case 0x10151: |
| 511 | case 0x10166: |
| 512 | case 0x10167: |
| 513 | case 0x10168: |
| 514 | case 0x10169: |
| 515 | case 0x10174: |
| 516 | case 0x10323: |
| 517 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 518 | return (double) 50; |
| 519 | case 0x216E: |
| 520 | case 0x217E: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 521 | #ifdef Py_UNICODE_WIDE |
| 522 | case 0x1011D: |
| 523 | case 0x10145: |
| 524 | case 0x1014C: |
| 525 | case 0x10153: |
| 526 | case 0x1016C: |
| 527 | case 0x1016D: |
| 528 | case 0x1016E: |
| 529 | case 0x1016F: |
| 530 | case 0x10170: |
| 531 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 532 | return (double) 500; |
| 533 | case 0x2181: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 534 | #ifdef Py_UNICODE_WIDE |
| 535 | case 0x10126: |
| 536 | case 0x10146: |
| 537 | case 0x1014E: |
| 538 | case 0x10172: |
| 539 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 540 | return (double) 5000; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 541 | #ifdef Py_UNICODE_WIDE |
| 542 | case 0x1012F: |
| 543 | case 0x10147: |
| 544 | case 0x10156: |
| 545 | return (double) 50000; |
| 546 | #endif |
| 547 | case 0x17F6: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 548 | case 0x2165: |
| 549 | case 0x2175: |
| 550 | case 0x3026: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 551 | case 0x3225: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 552 | case 0x3285: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 553 | #ifdef Py_UNICODE_WIDE |
| 554 | case 0x1010C: |
| 555 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 556 | return (double) 6; |
| 557 | case 0x1377: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 558 | #ifdef Py_UNICODE_WIDE |
| 559 | case 0x10115: |
| 560 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 561 | return (double) 60; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 562 | #ifdef Py_UNICODE_WIDE |
| 563 | case 0x1011E: |
| 564 | return (double) 600; |
| 565 | case 0x10127: |
| 566 | return (double) 6000; |
| 567 | case 0x10130: |
| 568 | return (double) 60000; |
| 569 | #endif |
| 570 | case 0x17F7: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 571 | case 0x2166: |
| 572 | case 0x2176: |
| 573 | case 0x3027: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 574 | case 0x3226: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 575 | case 0x3286: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 576 | #ifdef Py_UNICODE_WIDE |
| 577 | case 0x1010D: |
| 578 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 579 | return (double) 7; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 580 | case 0x0F2D: |
| 581 | return (double) 7 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 582 | case 0x215E: |
| 583 | return (double) 7 / 8; |
| 584 | case 0x1378: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 585 | #ifdef Py_UNICODE_WIDE |
| 586 | case 0x10116: |
| 587 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 588 | return (double) 70; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 589 | #ifdef Py_UNICODE_WIDE |
| 590 | case 0x1011F: |
| 591 | return (double) 700; |
| 592 | case 0x10128: |
| 593 | return (double) 7000; |
| 594 | case 0x10131: |
| 595 | return (double) 70000; |
| 596 | #endif |
| 597 | case 0x17F8: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 598 | case 0x2167: |
| 599 | case 0x2177: |
| 600 | case 0x3028: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 601 | case 0x3227: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 602 | case 0x3287: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 603 | #ifdef Py_UNICODE_WIDE |
| 604 | case 0x1010E: |
| 605 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 606 | return (double) 8; |
| 607 | case 0x1379: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 608 | #ifdef Py_UNICODE_WIDE |
| 609 | case 0x10117: |
| 610 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 611 | return (double) 80; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 612 | #ifdef Py_UNICODE_WIDE |
| 613 | case 0x10120: |
| 614 | return (double) 800; |
| 615 | case 0x10129: |
| 616 | return (double) 8000; |
| 617 | case 0x10132: |
| 618 | return (double) 80000; |
| 619 | #endif |
| 620 | case 0x17F9: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 621 | case 0x2168: |
| 622 | case 0x2178: |
| 623 | case 0x3029: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 624 | case 0x3228: |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 625 | case 0x3288: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 626 | #ifdef Py_UNICODE_WIDE |
| 627 | case 0x1010F: |
| 628 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 629 | return (double) 9; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 630 | case 0x0F2E: |
| 631 | return (double) 9 / 2; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 632 | case 0x137A: |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 633 | #ifdef Py_UNICODE_WIDE |
| 634 | case 0x10118: |
| 635 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 636 | return (double) 90; |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 637 | #ifdef Py_UNICODE_WIDE |
| 638 | case 0x10121: |
| 639 | case 0x1034A: |
| 640 | return (double) 900; |
| 641 | case 0x1012A: |
| 642 | return (double) 9000; |
| 643 | case 0x10133: |
| 644 | return (double) 90000; |
| 645 | #endif |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 646 | default: |
| 647 | return (double) _PyUnicode_ToDigit(ch); |
| 648 | } |
| 649 | } |
| 650 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 651 | int _PyUnicode_IsNumeric(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 652 | { |
Martin v. Löwis | d004fc8 | 2006-05-27 08:36:52 +0000 | [diff] [blame^] | 653 | return _PyUnicode_ToNumeric(ch) != -1.0; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 654 | } |
| 655 | |
| 656 | #ifndef WANT_WCTYPE_FUNCTIONS |
| 657 | |
Guido van Rossum | dc742b3 | 2000-04-11 15:39:02 +0000 | [diff] [blame] | 658 | /* Returns 1 for Unicode characters having the bidirectional type |
| 659 | 'WS', 'B' or 'S' or the category 'Zs', 0 otherwise. */ |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 660 | |
Marc-André Lemburg | 2cb94ab | 2005-10-20 19:06:35 +0000 | [diff] [blame] | 661 | int _PyUnicode_IsWhitespace(register const Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 662 | { |
Marc-André Lemburg | 2cb94ab | 2005-10-20 19:06:35 +0000 | [diff] [blame] | 663 | switch (ch) { |
| 664 | case 0x0009: /* HORIZONTAL TABULATION */ |
| 665 | case 0x000A: /* LINE FEED */ |
| 666 | case 0x000B: /* VERTICAL TABULATION */ |
| 667 | case 0x000C: /* FORM FEED */ |
| 668 | case 0x000D: /* CARRIAGE RETURN */ |
| 669 | case 0x001C: /* FILE SEPARATOR */ |
| 670 | case 0x001D: /* GROUP SEPARATOR */ |
| 671 | case 0x001E: /* RECORD SEPARATOR */ |
| 672 | case 0x001F: /* UNIT SEPARATOR */ |
| 673 | case 0x0020: /* SPACE */ |
| 674 | case 0x0085: /* NEXT LINE */ |
| 675 | case 0x00A0: /* NO-BREAK SPACE */ |
| 676 | case 0x1680: /* OGHAM SPACE MARK */ |
| 677 | case 0x2000: /* EN QUAD */ |
| 678 | case 0x2001: /* EM QUAD */ |
| 679 | case 0x2002: /* EN SPACE */ |
| 680 | case 0x2003: /* EM SPACE */ |
| 681 | case 0x2004: /* THREE-PER-EM SPACE */ |
| 682 | case 0x2005: /* FOUR-PER-EM SPACE */ |
| 683 | case 0x2006: /* SIX-PER-EM SPACE */ |
| 684 | case 0x2007: /* FIGURE SPACE */ |
| 685 | case 0x2008: /* PUNCTUATION SPACE */ |
| 686 | case 0x2009: /* THIN SPACE */ |
| 687 | case 0x200A: /* HAIR SPACE */ |
| 688 | case 0x200B: /* ZERO WIDTH SPACE */ |
| 689 | case 0x2028: /* LINE SEPARATOR */ |
| 690 | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 691 | case 0x202F: /* NARROW NO-BREAK SPACE */ |
| 692 | case 0x205F: /* MEDIUM MATHEMATICAL SPACE */ |
| 693 | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 694 | return 1; |
| 695 | default: |
| 696 | return 0; |
| 697 | } |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 698 | } |
| 699 | |
| 700 | /* Returns 1 for Unicode characters having the category 'Ll', 0 |
| 701 | otherwise. */ |
| 702 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 703 | int _PyUnicode_IsLowercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 704 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 705 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
| 706 | |
| 707 | return (ctype->flags & LOWER_MASK) != 0; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 708 | } |
| 709 | |
| 710 | /* Returns 1 for Unicode characters having the category 'Lu', 0 |
| 711 | otherwise. */ |
| 712 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 713 | int _PyUnicode_IsUppercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 714 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 715 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
| 716 | |
| 717 | return (ctype->flags & UPPER_MASK) != 0; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 718 | } |
| 719 | |
| 720 | /* Returns the uppercase Unicode characters corresponding to ch or just |
| 721 | ch if no uppercase mapping is known. */ |
| 722 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 723 | Py_UNICODE _PyUnicode_ToUppercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 724 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 725 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
Martin v. Löwis | edf368c | 2002-10-18 16:40:36 +0000 | [diff] [blame] | 726 | int delta = ctype->upper; |
| 727 | if (delta >= 32768) |
| 728 | delta -= 65536; |
| 729 | return ch + delta; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 730 | } |
| 731 | |
| 732 | /* Returns the lowercase Unicode characters corresponding to ch or just |
| 733 | ch if no lowercase mapping is known. */ |
| 734 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 735 | Py_UNICODE _PyUnicode_ToLowercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 736 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 737 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
Martin v. Löwis | edf368c | 2002-10-18 16:40:36 +0000 | [diff] [blame] | 738 | int delta = ctype->lower; |
| 739 | if (delta >= 32768) |
| 740 | delta -= 65536; |
| 741 | return ch + delta; |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 742 | } |
| 743 | |
Marc-André Lemburg | f3938f5 | 2000-07-05 09:48:59 +0000 | [diff] [blame] | 744 | /* Returns 1 for Unicode characters having the category 'Ll', 'Lu', 'Lt', |
| 745 | 'Lo' or 'Lm', 0 otherwise. */ |
| 746 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 747 | int _PyUnicode_IsAlpha(Py_UNICODE ch) |
Marc-André Lemburg | f3938f5 | 2000-07-05 09:48:59 +0000 | [diff] [blame] | 748 | { |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 749 | const _PyUnicode_TypeRecord *ctype = gettyperecord(ch); |
Marc-André Lemburg | f3938f5 | 2000-07-05 09:48:59 +0000 | [diff] [blame] | 750 | |
Fredrik Lundh | 9e7dd4c | 2000-09-25 21:48:13 +0000 | [diff] [blame] | 751 | return (ctype->flags & ALPHA_MASK) != 0; |
Marc-André Lemburg | f3938f5 | 2000-07-05 09:48:59 +0000 | [diff] [blame] | 752 | } |
| 753 | |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 754 | #else |
| 755 | |
| 756 | /* Export the interfaces using the wchar_t type for portability |
| 757 | reasons: */ |
| 758 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 759 | int _PyUnicode_IsWhitespace(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 760 | { |
| 761 | return iswspace(ch); |
| 762 | } |
| 763 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 764 | int _PyUnicode_IsLowercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 765 | { |
| 766 | return iswlower(ch); |
| 767 | } |
| 768 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 769 | int _PyUnicode_IsUppercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 770 | { |
| 771 | return iswupper(ch); |
| 772 | } |
| 773 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 774 | Py_UNICODE _PyUnicode_ToLowercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 775 | { |
| 776 | return towlower(ch); |
| 777 | } |
| 778 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 779 | Py_UNICODE _PyUnicode_ToUppercase(Py_UNICODE ch) |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 780 | { |
| 781 | return towupper(ch); |
| 782 | } |
| 783 | |
Fredrik Lundh | 72b0685 | 2001-06-27 22:08:26 +0000 | [diff] [blame] | 784 | int _PyUnicode_IsAlpha(Py_UNICODE ch) |
Marc-André Lemburg | f3938f5 | 2000-07-05 09:48:59 +0000 | [diff] [blame] | 785 | { |
| 786 | return iswalpha(ch); |
| 787 | } |
| 788 | |
Guido van Rossum | 603484d | 2000-03-10 22:52:46 +0000 | [diff] [blame] | 789 | #endif |