Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 1 | /* ------------------------------------------------------------------------ |
| 2 | |
| 3 | unicodedata -- Provides access to the Unicode 3.0 data base. |
| 4 | |
| 5 | Data was extracted from the Unicode 3.0 UnicodeData.txt file. |
| 6 | |
Fredrik Lundh | cfcea49 | 2000-09-25 08:07:06 +0000 | [diff] [blame] | 7 | Written by Marc-Andre Lemburg (mal@lemburg.com). |
| 8 | Modified for Python 2.0 by Fredrik Lundh (fredrik@pythonware.com) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 9 | |
Fredrik Lundh | cfcea49 | 2000-09-25 08:07:06 +0000 | [diff] [blame] | 10 | Copyright (c) Corporation for National Research Initiatives. |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 11 | |
| 12 | ------------------------------------------------------------------------ */ |
| 13 | |
| 14 | #include "Python.h" |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 15 | |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 16 | typedef struct { |
| 17 | const unsigned char category; /* index into |
| 18 | _PyUnicode_CategoryNames */ |
| 19 | const unsigned char combining; /* combining class value 0 - 255 */ |
| 20 | const unsigned char bidirectional; /* index into |
| 21 | _PyUnicode_BidirectionalNames */ |
| 22 | const unsigned char mirrored; /* true if mirrored in bidir mode */ |
| 23 | } _PyUnicode_DatabaseRecord; |
| 24 | |
| 25 | /* data file generated by Tools/unicode/makeunicodedata.py */ |
| 26 | #include "unicodedata_db.h" |
| 27 | |
| 28 | static const _PyUnicode_DatabaseRecord* |
| 29 | getrecord(PyUnicodeObject* v) |
| 30 | { |
| 31 | int code; |
| 32 | int index; |
| 33 | |
| 34 | code = (int) *PyUnicode_AS_UNICODE(v); |
| 35 | |
| 36 | if (code < 0 || code >= 65536) |
| 37 | index = 0; |
| 38 | else { |
| 39 | index = index1[(code>>SHIFT)]; |
| 40 | index = index2[(index<<SHIFT)+(code&((1<<SHIFT)-1))]; |
| 41 | } |
| 42 | |
| 43 | return &_PyUnicode_Database_Records[index]; |
| 44 | } |
| 45 | |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 46 | /* --- Module API --------------------------------------------------------- */ |
| 47 | |
| 48 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 49 | unicodedata_decimal(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 50 | { |
| 51 | PyUnicodeObject *v; |
| 52 | PyObject *defobj = NULL; |
| 53 | long rc; |
| 54 | |
| 55 | if (!PyArg_ParseTuple(args, "O!|O:decimal", |
| 56 | &PyUnicode_Type, &v, &defobj)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 57 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 58 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 59 | PyErr_SetString(PyExc_TypeError, |
| 60 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 61 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 62 | } |
| 63 | rc = Py_UNICODE_TODECIMAL(*PyUnicode_AS_UNICODE(v)); |
| 64 | if (rc < 0) { |
| 65 | if (defobj == NULL) { |
| 66 | PyErr_SetString(PyExc_ValueError, |
| 67 | "not a decimal"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 68 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 69 | } |
| 70 | else { |
| 71 | Py_INCREF(defobj); |
| 72 | return defobj; |
| 73 | } |
| 74 | } |
| 75 | return PyInt_FromLong(rc); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 76 | } |
| 77 | |
| 78 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 79 | unicodedata_digit(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 80 | { |
| 81 | PyUnicodeObject *v; |
| 82 | PyObject *defobj = NULL; |
| 83 | long rc; |
| 84 | |
| 85 | if (!PyArg_ParseTuple(args, "O!|O:digit", |
| 86 | &PyUnicode_Type, &v, &defobj)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 87 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 88 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 89 | PyErr_SetString(PyExc_TypeError, |
| 90 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 91 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 92 | } |
| 93 | rc = Py_UNICODE_TODIGIT(*PyUnicode_AS_UNICODE(v)); |
| 94 | if (rc < 0) { |
| 95 | if (defobj == NULL) { |
| 96 | PyErr_SetString(PyExc_ValueError, |
| 97 | "not a digit"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 98 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 99 | } |
| 100 | else { |
| 101 | Py_INCREF(defobj); |
| 102 | return defobj; |
| 103 | } |
| 104 | } |
| 105 | return PyInt_FromLong(rc); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 106 | } |
| 107 | |
| 108 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 109 | unicodedata_numeric(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 110 | { |
| 111 | PyUnicodeObject *v; |
| 112 | PyObject *defobj = NULL; |
| 113 | double rc; |
| 114 | |
| 115 | if (!PyArg_ParseTuple(args, "O!|O:numeric", |
| 116 | &PyUnicode_Type, &v, &defobj)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 117 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 118 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 119 | PyErr_SetString(PyExc_TypeError, |
| 120 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 121 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 122 | } |
| 123 | rc = Py_UNICODE_TONUMERIC(*PyUnicode_AS_UNICODE(v)); |
| 124 | if (rc < 0) { |
| 125 | if (defobj == NULL) { |
| 126 | PyErr_SetString(PyExc_ValueError, |
| 127 | "not a numeric character"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 128 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 129 | } |
| 130 | else { |
| 131 | Py_INCREF(defobj); |
| 132 | return defobj; |
| 133 | } |
| 134 | } |
| 135 | return PyFloat_FromDouble(rc); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 136 | } |
| 137 | |
| 138 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 139 | unicodedata_category(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 140 | { |
| 141 | PyUnicodeObject *v; |
| 142 | int index; |
| 143 | |
| 144 | if (!PyArg_ParseTuple(args, "O!:category", |
| 145 | &PyUnicode_Type, &v)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 146 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 147 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 148 | PyErr_SetString(PyExc_TypeError, |
| 149 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 150 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 151 | } |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 152 | index = (int) getrecord(v)->category; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 153 | return PyString_FromString(_PyUnicode_CategoryNames[index]); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 154 | } |
| 155 | |
| 156 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 157 | unicodedata_bidirectional(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 158 | { |
| 159 | PyUnicodeObject *v; |
| 160 | int index; |
| 161 | |
| 162 | if (!PyArg_ParseTuple(args, "O!:bidirectional", |
| 163 | &PyUnicode_Type, &v)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 164 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 165 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 166 | PyErr_SetString(PyExc_TypeError, |
| 167 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 168 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 169 | } |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 170 | index = (int) getrecord(v)->bidirectional; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 171 | return PyString_FromString(_PyUnicode_BidirectionalNames[index]); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 172 | } |
| 173 | |
| 174 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 175 | unicodedata_combining(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 176 | { |
| 177 | PyUnicodeObject *v; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 178 | |
| 179 | if (!PyArg_ParseTuple(args, "O!:combining", |
| 180 | &PyUnicode_Type, &v)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 181 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 182 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 183 | PyErr_SetString(PyExc_TypeError, |
| 184 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 185 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 186 | } |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 187 | return PyInt_FromLong((int) getrecord(v)->combining); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 188 | } |
| 189 | |
| 190 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 191 | unicodedata_mirrored(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 192 | { |
| 193 | PyUnicodeObject *v; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 194 | |
| 195 | if (!PyArg_ParseTuple(args, "O!:mirrored", |
| 196 | &PyUnicode_Type, &v)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 197 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 198 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 199 | PyErr_SetString(PyExc_TypeError, |
| 200 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 201 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 202 | } |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 203 | return PyInt_FromLong((int) getrecord(v)->mirrored); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 204 | } |
| 205 | |
| 206 | static PyObject * |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 207 | unicodedata_decomposition(PyObject *self, PyObject *args) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 208 | { |
| 209 | PyUnicodeObject *v; |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 210 | char decomp[256]; |
| 211 | int code, index, count, i; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 212 | |
| 213 | if (!PyArg_ParseTuple(args, "O!:decomposition", |
| 214 | &PyUnicode_Type, &v)) |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 215 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 216 | if (PyUnicode_GET_SIZE(v) != 1) { |
| 217 | PyErr_SetString(PyExc_TypeError, |
| 218 | "need a single Unicode character as parameter"); |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 219 | return NULL; |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 220 | } |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 221 | |
| 222 | code = (int) *PyUnicode_AS_UNICODE(v); |
| 223 | |
| 224 | if (code < 0 || code >= 65536) |
| 225 | index = 0; |
| 226 | else { |
| 227 | index = decomp_index1[(code>>DECOMP_SHIFT)]; |
| 228 | index = decomp_index2[(index<<DECOMP_SHIFT)+ |
| 229 | (code&((1<<DECOMP_SHIFT)-1))]; |
| 230 | } |
| 231 | |
| 232 | /* high byte is of hex bytes (usually one or two), low byte |
| 233 | is prefix code (from*/ |
| 234 | count = decomp_data[index] >> 8; |
| 235 | |
| 236 | /* XXX: could allocate the PyString up front instead |
| 237 | (strlen(prefix) + 5 * count + 1 bytes) */ |
| 238 | |
| 239 | /* copy prefix */ |
| 240 | i = strlen(decomp_prefix[decomp_data[index] & 255]); |
| 241 | memcpy(decomp, decomp_prefix[decomp_data[index] & 255], i); |
| 242 | |
| 243 | while (count-- > 0) { |
| 244 | if (i) |
| 245 | decomp[i++] = ' '; |
| 246 | sprintf(decomp + i, "%04X", decomp_data[++index]); |
| 247 | i += strlen(decomp + i); |
| 248 | } |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 249 | |
Fredrik Lundh | 7b7dd10 | 2001-01-21 22:41:08 +0000 | [diff] [blame] | 250 | decomp[i] = '\0'; |
| 251 | |
| 252 | return PyString_FromString(decomp); |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 253 | } |
| 254 | |
| 255 | /* XXX Add doc strings. */ |
| 256 | |
| 257 | static PyMethodDef unicodedata_functions[] = { |
| 258 | {"decimal", unicodedata_decimal, 1}, |
| 259 | {"digit", unicodedata_digit, 1}, |
| 260 | {"numeric", unicodedata_numeric, 1}, |
| 261 | {"category", unicodedata_category, 1}, |
| 262 | {"bidirectional", unicodedata_bidirectional, 1}, |
| 263 | {"combining", unicodedata_combining, 1}, |
| 264 | {"mirrored", unicodedata_mirrored, 1}, |
| 265 | {"decomposition", unicodedata_decomposition, 1}, |
| 266 | {NULL, NULL} /* sentinel */ |
| 267 | }; |
| 268 | |
| 269 | DL_EXPORT(void) |
Thomas Wouters | f3f33dc | 2000-07-21 06:00:07 +0000 | [diff] [blame] | 270 | initunicodedata(void) |
Guido van Rossum | 2a70a3a | 2000-03-10 23:10:21 +0000 | [diff] [blame] | 271 | { |
| 272 | Py_InitModule("unicodedata", unicodedata_functions); |
| 273 | } |