Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1 | /* ------------------------------------------------------------------------ |
| 2 | |
| 3 | _codecs -- Provides access to the codec registry and the builtin |
| 4 | codecs. |
| 5 | |
| 6 | This module should never be imported directly. The standard library |
| 7 | module "codecs" wraps this builtin module for use within Python. |
| 8 | |
| 9 | The codec registry is accessible via: |
| 10 | |
| 11 | register(search_function) -> None |
| 12 | |
| 13 | lookup(encoding) -> (encoder, decoder, stream_reader, stream_writer) |
| 14 | |
| 15 | The builtin Unicode codecs use the following interface: |
| 16 | |
| 17 | <encoding>_encode(Unicode_object[,errors='strict']) -> |
| 18 | (string object, bytes consumed) |
| 19 | |
| 20 | <encoding>_decode(char_buffer_obj[,errors='strict']) -> |
| 21 | (Unicode object, bytes consumed) |
| 22 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 23 | <encoding>_encode() interfaces also accept non-Unicode object as |
| 24 | input. The objects are then converted to Unicode using |
| 25 | PyUnicode_FromObject() prior to applying the conversion. |
| 26 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 27 | These <encoding>s are available: utf_8, unicode_escape, |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 28 | raw_unicode_escape, unicode_internal, latin_1, ascii (7-bit), |
| 29 | mbcs (on win32). |
| 30 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 31 | |
| 32 | Written by Marc-Andre Lemburg (mal@lemburg.com). |
| 33 | |
Guido van Rossum | 16b1ad9 | 2000-08-03 16:24:25 +0000 | [diff] [blame] | 34 | Copyright (c) Corporation for National Research Initiatives. |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 35 | |
| 36 | ------------------------------------------------------------------------ */ |
| 37 | |
| 38 | #include "Python.h" |
| 39 | |
| 40 | /* --- Registry ----------------------------------------------------------- */ |
| 41 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 42 | PyDoc_STRVAR(register__doc__, |
| 43 | "register(search_function)\n\ |
| 44 | \n\ |
| 45 | Register a codec search function. Search functions are expected to take\n\ |
| 46 | one argument, the encoding name in all lower case letters, and return\n\ |
| 47 | a tuple of functions (encoder, decoder, stream_reader, stream_writer)."); |
| 48 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 49 | static |
| 50 | PyObject *codecregister(PyObject *self, PyObject *args) |
| 51 | { |
| 52 | PyObject *search_function; |
| 53 | |
| 54 | if (!PyArg_ParseTuple(args, "O:register", &search_function)) |
| 55 | goto onError; |
| 56 | |
| 57 | if (PyCodec_Register(search_function)) |
| 58 | goto onError; |
| 59 | |
| 60 | Py_INCREF(Py_None); |
| 61 | return Py_None; |
| 62 | |
| 63 | onError: |
| 64 | return NULL; |
| 65 | } |
| 66 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 67 | PyDoc_STRVAR(lookup__doc__, |
| 68 | "lookup(encoding) -> (encoder, decoder, stream_reader, stream_writer)\n\ |
| 69 | \n\ |
| 70 | Looks up a codec tuple in the Python codec registry and returns\n\ |
| 71 | a tuple of functions."); |
| 72 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 73 | static |
| 74 | PyObject *codeclookup(PyObject *self, PyObject *args) |
| 75 | { |
| 76 | char *encoding; |
| 77 | |
| 78 | if (!PyArg_ParseTuple(args, "s:lookup", &encoding)) |
| 79 | goto onError; |
| 80 | |
| 81 | return _PyCodec_Lookup(encoding); |
| 82 | |
| 83 | onError: |
| 84 | return NULL; |
| 85 | } |
| 86 | |
| 87 | /* --- Helpers ------------------------------------------------------------ */ |
| 88 | |
| 89 | static |
| 90 | PyObject *codec_tuple(PyObject *unicode, |
| 91 | int len) |
| 92 | { |
| 93 | PyObject *v,*w; |
| 94 | |
| 95 | if (unicode == NULL) |
| 96 | return NULL; |
| 97 | v = PyTuple_New(2); |
| 98 | if (v == NULL) { |
| 99 | Py_DECREF(unicode); |
| 100 | return NULL; |
| 101 | } |
| 102 | PyTuple_SET_ITEM(v,0,unicode); |
| 103 | w = PyInt_FromLong(len); |
| 104 | if (w == NULL) { |
| 105 | Py_DECREF(v); |
| 106 | return NULL; |
| 107 | } |
| 108 | PyTuple_SET_ITEM(v,1,w); |
| 109 | return v; |
| 110 | } |
| 111 | |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 112 | /* --- String codecs ------------------------------------------------------ */ |
| 113 | static PyObject * |
| 114 | escape_decode(PyObject *self, |
| 115 | PyObject *args) |
| 116 | { |
| 117 | const char *errors = NULL; |
| 118 | const char *data; |
| 119 | int size; |
| 120 | |
| 121 | if (!PyArg_ParseTuple(args, "s#|z:escape_decode", |
| 122 | &data, &size, &errors)) |
| 123 | return NULL; |
| 124 | return codec_tuple(PyString_DecodeEscape(data, size, errors, 0, NULL), |
| 125 | size); |
| 126 | } |
| 127 | |
| 128 | static PyObject * |
| 129 | escape_encode(PyObject *self, |
| 130 | PyObject *args) |
| 131 | { |
| 132 | PyObject *str; |
| 133 | const char *errors = NULL; |
| 134 | char *buf; |
| 135 | int len; |
| 136 | |
| 137 | if (!PyArg_ParseTuple(args, "O!|z:escape_encode", |
| 138 | &PyString_Type, &str, &errors)) |
| 139 | return NULL; |
| 140 | |
| 141 | str = PyString_Repr(str, 0); |
| 142 | if (!str) |
| 143 | return NULL; |
| 144 | |
| 145 | /* The string will be quoted. Unquote, similar to unicode-escape. */ |
| 146 | buf = PyString_AS_STRING (str); |
| 147 | len = PyString_GET_SIZE (str); |
| 148 | memmove(buf, buf+1, len-2); |
| 149 | _PyString_Resize(&str, len-2); |
| 150 | |
| 151 | return codec_tuple(str, PyString_Size(str)); |
| 152 | } |
| 153 | |
| 154 | #ifdef Py_USING_UNICODE |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 155 | /* --- Decoder ------------------------------------------------------------ */ |
| 156 | |
| 157 | static PyObject * |
| 158 | unicode_internal_decode(PyObject *self, |
| 159 | PyObject *args) |
| 160 | { |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 161 | PyObject *obj; |
| 162 | const char *errors = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 163 | const char *data; |
| 164 | int size; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 165 | |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 166 | if (!PyArg_ParseTuple(args, "O|z:unicode_internal_decode", |
| 167 | &obj, &errors)) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 168 | return NULL; |
| 169 | |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 170 | if (PyUnicode_Check(obj)) |
| 171 | return codec_tuple(obj, PyUnicode_GET_SIZE(obj)); |
| 172 | else { |
| 173 | if (PyObject_AsReadBuffer(obj, (const void **)&data, &size)) |
| 174 | return NULL; |
| 175 | return codec_tuple(PyUnicode_FromUnicode((Py_UNICODE *)data, |
| 176 | size / sizeof(Py_UNICODE)), |
| 177 | size); |
| 178 | } |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 179 | } |
| 180 | |
| 181 | static PyObject * |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 182 | utf_7_decode(PyObject *self, |
| 183 | PyObject *args) |
| 184 | { |
| 185 | const char *data; |
| 186 | int size; |
| 187 | const char *errors = NULL; |
| 188 | |
| 189 | if (!PyArg_ParseTuple(args, "t#|z:utf_7_decode", |
| 190 | &data, &size, &errors)) |
| 191 | return NULL; |
| 192 | |
| 193 | return codec_tuple(PyUnicode_DecodeUTF7(data, size, errors), |
| 194 | size); |
| 195 | } |
| 196 | |
| 197 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 198 | utf_8_decode(PyObject *self, |
| 199 | PyObject *args) |
| 200 | { |
| 201 | const char *data; |
| 202 | int size; |
| 203 | const char *errors = NULL; |
| 204 | |
| 205 | if (!PyArg_ParseTuple(args, "t#|z:utf_8_decode", |
| 206 | &data, &size, &errors)) |
| 207 | return NULL; |
| 208 | |
| 209 | return codec_tuple(PyUnicode_DecodeUTF8(data, size, errors), |
| 210 | size); |
| 211 | } |
| 212 | |
| 213 | static PyObject * |
| 214 | utf_16_decode(PyObject *self, |
| 215 | PyObject *args) |
| 216 | { |
| 217 | const char *data; |
| 218 | int size; |
| 219 | const char *errors = NULL; |
| 220 | int byteorder = 0; |
| 221 | |
| 222 | if (!PyArg_ParseTuple(args, "t#|z:utf_16_decode", |
| 223 | &data, &size, &errors)) |
| 224 | return NULL; |
| 225 | return codec_tuple(PyUnicode_DecodeUTF16(data, size, errors, &byteorder), |
| 226 | size); |
| 227 | } |
| 228 | |
| 229 | static PyObject * |
| 230 | utf_16_le_decode(PyObject *self, |
| 231 | PyObject *args) |
| 232 | { |
| 233 | const char *data; |
| 234 | int size; |
| 235 | const char *errors = NULL; |
| 236 | int byteorder = -1; |
| 237 | |
| 238 | if (!PyArg_ParseTuple(args, "t#|z:utf_16_le_decode", |
| 239 | &data, &size, &errors)) |
| 240 | return NULL; |
| 241 | return codec_tuple(PyUnicode_DecodeUTF16(data, size, errors, &byteorder), |
| 242 | size); |
| 243 | } |
| 244 | |
| 245 | static PyObject * |
| 246 | utf_16_be_decode(PyObject *self, |
| 247 | PyObject *args) |
| 248 | { |
| 249 | const char *data; |
| 250 | int size; |
| 251 | const char *errors = NULL; |
| 252 | int byteorder = 1; |
| 253 | |
| 254 | if (!PyArg_ParseTuple(args, "t#|z:utf_16_be_decode", |
| 255 | &data, &size, &errors)) |
| 256 | return NULL; |
| 257 | return codec_tuple(PyUnicode_DecodeUTF16(data, size, errors, &byteorder), |
| 258 | size); |
| 259 | } |
| 260 | |
| 261 | /* This non-standard version also provides access to the byteorder |
| 262 | parameter of the builtin UTF-16 codec. |
| 263 | |
| 264 | It returns a tuple (unicode, bytesread, byteorder) with byteorder |
| 265 | being the value in effect at the end of data. |
| 266 | |
| 267 | */ |
| 268 | |
| 269 | static PyObject * |
| 270 | utf_16_ex_decode(PyObject *self, |
| 271 | PyObject *args) |
| 272 | { |
| 273 | const char *data; |
| 274 | int size; |
| 275 | const char *errors = NULL; |
| 276 | int byteorder = 0; |
| 277 | PyObject *unicode, *tuple; |
| 278 | |
| 279 | if (!PyArg_ParseTuple(args, "t#|zi:utf_16_ex_decode", |
| 280 | &data, &size, &errors, &byteorder)) |
| 281 | return NULL; |
| 282 | |
| 283 | unicode = PyUnicode_DecodeUTF16(data, size, errors, &byteorder); |
| 284 | if (unicode == NULL) |
| 285 | return NULL; |
| 286 | tuple = Py_BuildValue("Oii", unicode, size, byteorder); |
| 287 | Py_DECREF(unicode); |
| 288 | return tuple; |
| 289 | } |
| 290 | |
| 291 | static PyObject * |
| 292 | unicode_escape_decode(PyObject *self, |
| 293 | PyObject *args) |
| 294 | { |
| 295 | const char *data; |
| 296 | int size; |
| 297 | const char *errors = NULL; |
| 298 | |
| 299 | if (!PyArg_ParseTuple(args, "t#|z:unicode_escape_decode", |
| 300 | &data, &size, &errors)) |
| 301 | return NULL; |
| 302 | |
| 303 | return codec_tuple(PyUnicode_DecodeUnicodeEscape(data, size, errors), |
| 304 | size); |
| 305 | } |
| 306 | |
| 307 | static PyObject * |
| 308 | raw_unicode_escape_decode(PyObject *self, |
| 309 | PyObject *args) |
| 310 | { |
| 311 | const char *data; |
| 312 | int size; |
| 313 | const char *errors = NULL; |
| 314 | |
| 315 | if (!PyArg_ParseTuple(args, "t#|z:raw_unicode_escape_decode", |
| 316 | &data, &size, &errors)) |
| 317 | return NULL; |
| 318 | |
| 319 | return codec_tuple(PyUnicode_DecodeRawUnicodeEscape(data, size, errors), |
| 320 | size); |
| 321 | } |
| 322 | |
| 323 | static PyObject * |
| 324 | latin_1_decode(PyObject *self, |
| 325 | PyObject *args) |
| 326 | { |
| 327 | const char *data; |
| 328 | int size; |
| 329 | const char *errors = NULL; |
| 330 | |
| 331 | if (!PyArg_ParseTuple(args, "t#|z:latin_1_decode", |
| 332 | &data, &size, &errors)) |
| 333 | return NULL; |
| 334 | |
| 335 | return codec_tuple(PyUnicode_DecodeLatin1(data, size, errors), |
| 336 | size); |
| 337 | } |
| 338 | |
| 339 | static PyObject * |
| 340 | ascii_decode(PyObject *self, |
| 341 | PyObject *args) |
| 342 | { |
| 343 | const char *data; |
| 344 | int size; |
| 345 | const char *errors = NULL; |
| 346 | |
| 347 | if (!PyArg_ParseTuple(args, "t#|z:ascii_decode", |
| 348 | &data, &size, &errors)) |
| 349 | return NULL; |
| 350 | |
| 351 | return codec_tuple(PyUnicode_DecodeASCII(data, size, errors), |
| 352 | size); |
| 353 | } |
| 354 | |
| 355 | static PyObject * |
| 356 | charmap_decode(PyObject *self, |
| 357 | PyObject *args) |
| 358 | { |
| 359 | const char *data; |
| 360 | int size; |
| 361 | const char *errors = NULL; |
| 362 | PyObject *mapping = NULL; |
| 363 | |
| 364 | if (!PyArg_ParseTuple(args, "t#|zO:charmap_decode", |
| 365 | &data, &size, &errors, &mapping)) |
| 366 | return NULL; |
| 367 | if (mapping == Py_None) |
| 368 | mapping = NULL; |
| 369 | |
| 370 | return codec_tuple(PyUnicode_DecodeCharmap(data, size, mapping, errors), |
| 371 | size); |
| 372 | } |
| 373 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 374 | #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 375 | |
| 376 | static PyObject * |
| 377 | mbcs_decode(PyObject *self, |
| 378 | PyObject *args) |
| 379 | { |
| 380 | const char *data; |
| 381 | int size; |
| 382 | const char *errors = NULL; |
| 383 | |
| 384 | if (!PyArg_ParseTuple(args, "t#|z:mbcs_decode", |
| 385 | &data, &size, &errors)) |
| 386 | return NULL; |
| 387 | |
| 388 | return codec_tuple(PyUnicode_DecodeMBCS(data, size, errors), |
| 389 | size); |
| 390 | } |
| 391 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 392 | #endif /* MS_WINDOWS */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 393 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 394 | /* --- Encoder ------------------------------------------------------------ */ |
| 395 | |
| 396 | static PyObject * |
| 397 | readbuffer_encode(PyObject *self, |
| 398 | PyObject *args) |
| 399 | { |
| 400 | const char *data; |
| 401 | int size; |
| 402 | const char *errors = NULL; |
| 403 | |
| 404 | if (!PyArg_ParseTuple(args, "s#|z:readbuffer_encode", |
| 405 | &data, &size, &errors)) |
| 406 | return NULL; |
| 407 | |
| 408 | return codec_tuple(PyString_FromStringAndSize(data, size), |
| 409 | size); |
| 410 | } |
| 411 | |
| 412 | static PyObject * |
| 413 | charbuffer_encode(PyObject *self, |
| 414 | PyObject *args) |
| 415 | { |
| 416 | const char *data; |
| 417 | int size; |
| 418 | const char *errors = NULL; |
| 419 | |
| 420 | if (!PyArg_ParseTuple(args, "t#|z:charbuffer_encode", |
| 421 | &data, &size, &errors)) |
| 422 | return NULL; |
| 423 | |
| 424 | return codec_tuple(PyString_FromStringAndSize(data, size), |
| 425 | size); |
| 426 | } |
| 427 | |
| 428 | static PyObject * |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 429 | unicode_internal_encode(PyObject *self, |
| 430 | PyObject *args) |
| 431 | { |
| 432 | PyObject *obj; |
| 433 | const char *errors = NULL; |
| 434 | const char *data; |
| 435 | int size; |
| 436 | |
| 437 | if (!PyArg_ParseTuple(args, "O|z:unicode_internal_encode", |
| 438 | &obj, &errors)) |
| 439 | return NULL; |
| 440 | |
| 441 | if (PyUnicode_Check(obj)) { |
| 442 | data = PyUnicode_AS_DATA(obj); |
| 443 | size = PyUnicode_GET_DATA_SIZE(obj); |
| 444 | return codec_tuple(PyString_FromStringAndSize(data, size), |
| 445 | size); |
| 446 | } |
| 447 | else { |
| 448 | if (PyObject_AsReadBuffer(obj, (const void **)&data, &size)) |
| 449 | return NULL; |
| 450 | return codec_tuple(PyString_FromStringAndSize(data, size), |
| 451 | size); |
| 452 | } |
| 453 | } |
| 454 | |
| 455 | static PyObject * |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 456 | utf_7_encode(PyObject *self, |
| 457 | PyObject *args) |
| 458 | { |
| 459 | PyObject *str, *v; |
| 460 | const char *errors = NULL; |
| 461 | |
| 462 | if (!PyArg_ParseTuple(args, "O|z:utf_7_encode", |
| 463 | &str, &errors)) |
| 464 | return NULL; |
| 465 | |
| 466 | str = PyUnicode_FromObject(str); |
| 467 | if (str == NULL) |
| 468 | return NULL; |
| 469 | v = codec_tuple(PyUnicode_EncodeUTF7(PyUnicode_AS_UNICODE(str), |
| 470 | PyUnicode_GET_SIZE(str), |
| 471 | 0, |
| 472 | 0, |
| 473 | errors), |
| 474 | PyUnicode_GET_SIZE(str)); |
| 475 | Py_DECREF(str); |
| 476 | return v; |
| 477 | } |
| 478 | |
| 479 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 480 | utf_8_encode(PyObject *self, |
| 481 | PyObject *args) |
| 482 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 483 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 484 | const char *errors = NULL; |
| 485 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 486 | if (!PyArg_ParseTuple(args, "O|z:utf_8_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 487 | &str, &errors)) |
| 488 | return NULL; |
| 489 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 490 | str = PyUnicode_FromObject(str); |
| 491 | if (str == NULL) |
| 492 | return NULL; |
| 493 | v = codec_tuple(PyUnicode_EncodeUTF8(PyUnicode_AS_UNICODE(str), |
| 494 | PyUnicode_GET_SIZE(str), |
| 495 | errors), |
| 496 | PyUnicode_GET_SIZE(str)); |
| 497 | Py_DECREF(str); |
| 498 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 499 | } |
| 500 | |
| 501 | /* This version provides access to the byteorder parameter of the |
| 502 | builtin UTF-16 codecs as optional third argument. It defaults to 0 |
| 503 | which means: use the native byte order and prepend the data with a |
| 504 | BOM mark. |
| 505 | |
| 506 | */ |
| 507 | |
| 508 | static PyObject * |
| 509 | utf_16_encode(PyObject *self, |
| 510 | PyObject *args) |
| 511 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 512 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 513 | const char *errors = NULL; |
| 514 | int byteorder = 0; |
| 515 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 516 | if (!PyArg_ParseTuple(args, "O|zi:utf_16_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 517 | &str, &errors, &byteorder)) |
| 518 | return NULL; |
| 519 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 520 | str = PyUnicode_FromObject(str); |
| 521 | if (str == NULL) |
| 522 | return NULL; |
| 523 | v = codec_tuple(PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(str), |
| 524 | PyUnicode_GET_SIZE(str), |
| 525 | errors, |
| 526 | byteorder), |
| 527 | PyUnicode_GET_SIZE(str)); |
| 528 | Py_DECREF(str); |
| 529 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 530 | } |
| 531 | |
| 532 | static PyObject * |
| 533 | utf_16_le_encode(PyObject *self, |
| 534 | PyObject *args) |
| 535 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 536 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 537 | const char *errors = NULL; |
| 538 | |
Marc-André Lemburg | 4157dd5 | 2001-06-17 18:32:36 +0000 | [diff] [blame] | 539 | if (!PyArg_ParseTuple(args, "O|z:utf_16_le_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 540 | &str, &errors)) |
| 541 | return NULL; |
| 542 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 543 | str = PyUnicode_FromObject(str); |
| 544 | if (str == NULL) |
| 545 | return NULL; |
| 546 | v = codec_tuple(PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(str), |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 547 | PyUnicode_GET_SIZE(str), |
| 548 | errors, |
| 549 | -1), |
| 550 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 551 | Py_DECREF(str); |
| 552 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 553 | } |
| 554 | |
| 555 | static PyObject * |
| 556 | utf_16_be_encode(PyObject *self, |
| 557 | PyObject *args) |
| 558 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 559 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 560 | const char *errors = NULL; |
| 561 | |
Marc-André Lemburg | 4157dd5 | 2001-06-17 18:32:36 +0000 | [diff] [blame] | 562 | if (!PyArg_ParseTuple(args, "O|z:utf_16_be_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 563 | &str, &errors)) |
| 564 | return NULL; |
| 565 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 566 | str = PyUnicode_FromObject(str); |
| 567 | if (str == NULL) |
| 568 | return NULL; |
| 569 | v = codec_tuple(PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(str), |
| 570 | PyUnicode_GET_SIZE(str), |
| 571 | errors, |
| 572 | +1), |
| 573 | PyUnicode_GET_SIZE(str)); |
| 574 | Py_DECREF(str); |
| 575 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 576 | } |
| 577 | |
| 578 | static PyObject * |
| 579 | unicode_escape_encode(PyObject *self, |
| 580 | PyObject *args) |
| 581 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 582 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 583 | const char *errors = NULL; |
| 584 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 585 | if (!PyArg_ParseTuple(args, "O|z:unicode_escape_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 586 | &str, &errors)) |
| 587 | return NULL; |
| 588 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 589 | str = PyUnicode_FromObject(str); |
| 590 | if (str == NULL) |
| 591 | return NULL; |
| 592 | v = codec_tuple(PyUnicode_EncodeUnicodeEscape(PyUnicode_AS_UNICODE(str), |
| 593 | PyUnicode_GET_SIZE(str)), |
| 594 | PyUnicode_GET_SIZE(str)); |
| 595 | Py_DECREF(str); |
| 596 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 597 | } |
| 598 | |
| 599 | static PyObject * |
| 600 | raw_unicode_escape_encode(PyObject *self, |
| 601 | PyObject *args) |
| 602 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 603 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 604 | const char *errors = NULL; |
| 605 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 606 | if (!PyArg_ParseTuple(args, "O|z:raw_unicode_escape_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 607 | &str, &errors)) |
| 608 | return NULL; |
| 609 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 610 | str = PyUnicode_FromObject(str); |
| 611 | if (str == NULL) |
| 612 | return NULL; |
| 613 | v = codec_tuple(PyUnicode_EncodeRawUnicodeEscape( |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 614 | PyUnicode_AS_UNICODE(str), |
| 615 | PyUnicode_GET_SIZE(str)), |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 616 | PyUnicode_GET_SIZE(str)); |
| 617 | Py_DECREF(str); |
| 618 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 619 | } |
| 620 | |
| 621 | static PyObject * |
| 622 | latin_1_encode(PyObject *self, |
| 623 | PyObject *args) |
| 624 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 625 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 626 | const char *errors = NULL; |
| 627 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 628 | if (!PyArg_ParseTuple(args, "O|z:latin_1_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 629 | &str, &errors)) |
| 630 | return NULL; |
| 631 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 632 | str = PyUnicode_FromObject(str); |
| 633 | if (str == NULL) |
| 634 | return NULL; |
| 635 | v = codec_tuple(PyUnicode_EncodeLatin1( |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 636 | PyUnicode_AS_UNICODE(str), |
| 637 | PyUnicode_GET_SIZE(str), |
| 638 | errors), |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 639 | PyUnicode_GET_SIZE(str)); |
| 640 | Py_DECREF(str); |
| 641 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 642 | } |
| 643 | |
| 644 | static PyObject * |
| 645 | ascii_encode(PyObject *self, |
| 646 | PyObject *args) |
| 647 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 648 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 649 | const char *errors = NULL; |
| 650 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 651 | if (!PyArg_ParseTuple(args, "O|z:ascii_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 652 | &str, &errors)) |
| 653 | return NULL; |
| 654 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 655 | str = PyUnicode_FromObject(str); |
| 656 | if (str == NULL) |
| 657 | return NULL; |
| 658 | v = codec_tuple(PyUnicode_EncodeASCII( |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 659 | PyUnicode_AS_UNICODE(str), |
| 660 | PyUnicode_GET_SIZE(str), |
| 661 | errors), |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 662 | PyUnicode_GET_SIZE(str)); |
| 663 | Py_DECREF(str); |
| 664 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 665 | } |
| 666 | |
| 667 | static PyObject * |
| 668 | charmap_encode(PyObject *self, |
| 669 | PyObject *args) |
| 670 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 671 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 672 | const char *errors = NULL; |
| 673 | PyObject *mapping = NULL; |
| 674 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 675 | if (!PyArg_ParseTuple(args, "O|zO:charmap_encode", |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 676 | &str, &errors, &mapping)) |
| 677 | return NULL; |
| 678 | if (mapping == Py_None) |
| 679 | mapping = NULL; |
| 680 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 681 | str = PyUnicode_FromObject(str); |
| 682 | if (str == NULL) |
| 683 | return NULL; |
| 684 | v = codec_tuple(PyUnicode_EncodeCharmap( |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 685 | PyUnicode_AS_UNICODE(str), |
| 686 | PyUnicode_GET_SIZE(str), |
| 687 | mapping, |
| 688 | errors), |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 689 | PyUnicode_GET_SIZE(str)); |
| 690 | Py_DECREF(str); |
| 691 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 692 | } |
| 693 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 694 | #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 695 | |
| 696 | static PyObject * |
| 697 | mbcs_encode(PyObject *self, |
| 698 | PyObject *args) |
| 699 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 700 | PyObject *str, *v; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 701 | const char *errors = NULL; |
| 702 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 703 | if (!PyArg_ParseTuple(args, "O|z:mbcs_encode", |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 704 | &str, &errors)) |
| 705 | return NULL; |
| 706 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 707 | str = PyUnicode_FromObject(str); |
| 708 | if (str == NULL) |
| 709 | return NULL; |
| 710 | v = codec_tuple(PyUnicode_EncodeMBCS( |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 711 | PyUnicode_AS_UNICODE(str), |
| 712 | PyUnicode_GET_SIZE(str), |
| 713 | errors), |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 714 | PyUnicode_GET_SIZE(str)); |
| 715 | Py_DECREF(str); |
| 716 | return v; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 717 | } |
| 718 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 719 | #endif /* MS_WINDOWS */ |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 720 | #endif /* Py_USING_UNICODE */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 721 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 722 | /* --- Error handler registry --------------------------------------------- */ |
| 723 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 724 | PyDoc_STRVAR(register_error__doc__, |
| 725 | "register_error(errors, handler)\n\ |
| 726 | \n\ |
| 727 | Register the specified error handler under the name\n\ |
| 728 | errors. handler must be a callable object, that\n\ |
| 729 | will be called with an exception instance containing\n\ |
| 730 | information about the location of the encoding/decoding\n\ |
| 731 | error and must return a (replacement, new position) tuple."); |
| 732 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 733 | static PyObject *register_error(PyObject *self, PyObject *args) |
| 734 | { |
| 735 | const char *name; |
| 736 | PyObject *handler; |
| 737 | |
| 738 | if (!PyArg_ParseTuple(args, "sO:register_error", |
| 739 | &name, &handler)) |
| 740 | return NULL; |
| 741 | if (PyCodec_RegisterError(name, handler)) |
| 742 | return NULL; |
| 743 | Py_INCREF(Py_None); |
| 744 | return Py_None; |
| 745 | } |
| 746 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 747 | PyDoc_STRVAR(lookup_error__doc__, |
| 748 | "lookup_error(errors) -> handler\n\ |
| 749 | \n\ |
| 750 | Return the error handler for the specified error handling name\n\ |
| 751 | or raise a LookupError, if no handler exists under this name."); |
| 752 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 753 | static PyObject *lookup_error(PyObject *self, PyObject *args) |
| 754 | { |
| 755 | const char *name; |
| 756 | |
| 757 | if (!PyArg_ParseTuple(args, "s:lookup_error", |
| 758 | &name)) |
| 759 | return NULL; |
| 760 | return PyCodec_LookupError(name); |
| 761 | } |
| 762 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 763 | /* --- Module API --------------------------------------------------------- */ |
| 764 | |
| 765 | static PyMethodDef _codecs_functions[] = { |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 766 | {"register", codecregister, METH_VARARGS, |
| 767 | register__doc__}, |
| 768 | {"lookup", codeclookup, METH_VARARGS, |
| 769 | lookup__doc__}, |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 770 | {"escape_encode", escape_encode, METH_VARARGS}, |
| 771 | {"escape_decode", escape_decode, METH_VARARGS}, |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 772 | #ifdef Py_USING_UNICODE |
Martin v. Löwis | 43b936d | 2002-01-17 23:15:58 +0000 | [diff] [blame] | 773 | {"utf_8_encode", utf_8_encode, METH_VARARGS}, |
| 774 | {"utf_8_decode", utf_8_decode, METH_VARARGS}, |
| 775 | {"utf_7_encode", utf_7_encode, METH_VARARGS}, |
| 776 | {"utf_7_decode", utf_7_decode, METH_VARARGS}, |
| 777 | {"utf_16_encode", utf_16_encode, METH_VARARGS}, |
| 778 | {"utf_16_le_encode", utf_16_le_encode, METH_VARARGS}, |
| 779 | {"utf_16_be_encode", utf_16_be_encode, METH_VARARGS}, |
| 780 | {"utf_16_decode", utf_16_decode, METH_VARARGS}, |
| 781 | {"utf_16_le_decode", utf_16_le_decode, METH_VARARGS}, |
| 782 | {"utf_16_be_decode", utf_16_be_decode, METH_VARARGS}, |
| 783 | {"utf_16_ex_decode", utf_16_ex_decode, METH_VARARGS}, |
| 784 | {"unicode_escape_encode", unicode_escape_encode, METH_VARARGS}, |
| 785 | {"unicode_escape_decode", unicode_escape_decode, METH_VARARGS}, |
| 786 | {"unicode_internal_encode", unicode_internal_encode, METH_VARARGS}, |
| 787 | {"unicode_internal_decode", unicode_internal_decode, METH_VARARGS}, |
| 788 | {"raw_unicode_escape_encode", raw_unicode_escape_encode, METH_VARARGS}, |
| 789 | {"raw_unicode_escape_decode", raw_unicode_escape_decode, METH_VARARGS}, |
| 790 | {"latin_1_encode", latin_1_encode, METH_VARARGS}, |
| 791 | {"latin_1_decode", latin_1_decode, METH_VARARGS}, |
| 792 | {"ascii_encode", ascii_encode, METH_VARARGS}, |
| 793 | {"ascii_decode", ascii_decode, METH_VARARGS}, |
| 794 | {"charmap_encode", charmap_encode, METH_VARARGS}, |
| 795 | {"charmap_decode", charmap_decode, METH_VARARGS}, |
| 796 | {"readbuffer_encode", readbuffer_encode, METH_VARARGS}, |
| 797 | {"charbuffer_encode", charbuffer_encode, METH_VARARGS}, |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 798 | #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) |
Martin v. Löwis | 43b936d | 2002-01-17 23:15:58 +0000 | [diff] [blame] | 799 | {"mbcs_encode", mbcs_encode, METH_VARARGS}, |
| 800 | {"mbcs_decode", mbcs_decode, METH_VARARGS}, |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 801 | #endif |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 802 | #endif /* Py_USING_UNICODE */ |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 803 | {"register_error", register_error, METH_VARARGS, |
| 804 | register_error__doc__}, |
| 805 | {"lookup_error", lookup_error, METH_VARARGS, |
| 806 | lookup_error__doc__}, |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 807 | {NULL, NULL} /* sentinel */ |
| 808 | }; |
| 809 | |
Mark Hammond | fe51c6d | 2002-08-02 02:27:13 +0000 | [diff] [blame] | 810 | PyMODINIT_FUNC |
Thomas Wouters | f3f33dc | 2000-07-21 06:00:07 +0000 | [diff] [blame] | 811 | init_codecs(void) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 812 | { |
| 813 | Py_InitModule("_codecs", _codecs_functions); |
| 814 | } |