Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1 | /* |
| 2 | An implementation of Text I/O as defined by PEP 3116 - "New I/O" |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 3 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 4 | Classes defined here: TextIOBase, IncrementalNewlineDecoder, TextIOWrapper. |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 5 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 6 | Written by Amaury Forgeot d'Arc and Antoine Pitrou |
| 7 | */ |
| 8 | |
| 9 | #define PY_SSIZE_T_CLEAN |
| 10 | #include "Python.h" |
| 11 | #include "structmember.h" |
| 12 | #include "_iomodule.h" |
| 13 | |
| 14 | /* TextIOBase */ |
| 15 | |
| 16 | PyDoc_STRVAR(TextIOBase_doc, |
| 17 | "Base class for text I/O.\n" |
| 18 | "\n" |
| 19 | "This class provides a character and line based interface to stream\n" |
| 20 | "I/O. There is no readinto method because Python's character strings\n" |
| 21 | "are immutable. There is no public constructor.\n" |
| 22 | ); |
| 23 | |
| 24 | static PyObject * |
| 25 | _unsupported(const char *message) |
| 26 | { |
| 27 | PyErr_SetString(IO_STATE->unsupported_operation, message); |
| 28 | return NULL; |
| 29 | } |
| 30 | |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 31 | PyDoc_STRVAR(TextIOBase_detach_doc, |
| 32 | "Separate the underlying buffer from the TextIOBase and return it.\n" |
| 33 | "\n" |
| 34 | "After the underlying buffer has been detached, the TextIO is in an\n" |
| 35 | "unusable state.\n" |
| 36 | ); |
| 37 | |
| 38 | static PyObject * |
| 39 | TextIOBase_detach(PyObject *self) |
| 40 | { |
| 41 | return _unsupported("detach"); |
| 42 | } |
| 43 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 44 | PyDoc_STRVAR(TextIOBase_read_doc, |
| 45 | "Read at most n characters from stream.\n" |
| 46 | "\n" |
| 47 | "Read from underlying buffer until we have n characters or we hit EOF.\n" |
| 48 | "If n is negative or omitted, read until EOF.\n" |
| 49 | ); |
| 50 | |
| 51 | static PyObject * |
| 52 | TextIOBase_read(PyObject *self, PyObject *args) |
| 53 | { |
| 54 | return _unsupported("read"); |
| 55 | } |
| 56 | |
| 57 | PyDoc_STRVAR(TextIOBase_readline_doc, |
| 58 | "Read until newline or EOF.\n" |
| 59 | "\n" |
| 60 | "Returns an empty string if EOF is hit immediately.\n" |
| 61 | ); |
| 62 | |
| 63 | static PyObject * |
| 64 | TextIOBase_readline(PyObject *self, PyObject *args) |
| 65 | { |
| 66 | return _unsupported("readline"); |
| 67 | } |
| 68 | |
| 69 | PyDoc_STRVAR(TextIOBase_write_doc, |
| 70 | "Write string to stream.\n" |
| 71 | "Returns the number of characters written (which is always equal to\n" |
| 72 | "the length of the string).\n" |
| 73 | ); |
| 74 | |
| 75 | static PyObject * |
| 76 | TextIOBase_write(PyObject *self, PyObject *args) |
| 77 | { |
| 78 | return _unsupported("write"); |
| 79 | } |
| 80 | |
| 81 | PyDoc_STRVAR(TextIOBase_encoding_doc, |
| 82 | "Encoding of the text stream.\n" |
| 83 | "\n" |
| 84 | "Subclasses should override.\n" |
| 85 | ); |
| 86 | |
| 87 | static PyObject * |
| 88 | TextIOBase_encoding_get(PyObject *self, void *context) |
| 89 | { |
| 90 | Py_RETURN_NONE; |
| 91 | } |
| 92 | |
| 93 | PyDoc_STRVAR(TextIOBase_newlines_doc, |
| 94 | "Line endings translated so far.\n" |
| 95 | "\n" |
| 96 | "Only line endings translated during reading are considered.\n" |
| 97 | "\n" |
| 98 | "Subclasses should override.\n" |
| 99 | ); |
| 100 | |
| 101 | static PyObject * |
| 102 | TextIOBase_newlines_get(PyObject *self, void *context) |
| 103 | { |
| 104 | Py_RETURN_NONE; |
| 105 | } |
| 106 | |
| 107 | |
| 108 | static PyMethodDef TextIOBase_methods[] = { |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 109 | {"detach", (PyCFunction)TextIOBase_detach, METH_NOARGS, TextIOBase_detach_doc}, |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 110 | {"read", TextIOBase_read, METH_VARARGS, TextIOBase_read_doc}, |
| 111 | {"readline", TextIOBase_readline, METH_VARARGS, TextIOBase_readline_doc}, |
| 112 | {"write", TextIOBase_write, METH_VARARGS, TextIOBase_write_doc}, |
| 113 | {NULL, NULL} |
| 114 | }; |
| 115 | |
| 116 | static PyGetSetDef TextIOBase_getset[] = { |
| 117 | {"encoding", (getter)TextIOBase_encoding_get, NULL, TextIOBase_encoding_doc}, |
| 118 | {"newlines", (getter)TextIOBase_newlines_get, NULL, TextIOBase_newlines_doc}, |
Benjamin Peterson | 1fea321 | 2009-04-19 03:15:20 +0000 | [diff] [blame] | 119 | {NULL} |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 120 | }; |
| 121 | |
| 122 | PyTypeObject PyTextIOBase_Type = { |
| 123 | PyVarObject_HEAD_INIT(NULL, 0) |
| 124 | "_io._TextIOBase", /*tp_name*/ |
| 125 | 0, /*tp_basicsize*/ |
| 126 | 0, /*tp_itemsize*/ |
| 127 | 0, /*tp_dealloc*/ |
| 128 | 0, /*tp_print*/ |
| 129 | 0, /*tp_getattr*/ |
| 130 | 0, /*tp_setattr*/ |
| 131 | 0, /*tp_compare */ |
| 132 | 0, /*tp_repr*/ |
| 133 | 0, /*tp_as_number*/ |
| 134 | 0, /*tp_as_sequence*/ |
| 135 | 0, /*tp_as_mapping*/ |
| 136 | 0, /*tp_hash */ |
| 137 | 0, /*tp_call*/ |
| 138 | 0, /*tp_str*/ |
| 139 | 0, /*tp_getattro*/ |
| 140 | 0, /*tp_setattro*/ |
| 141 | 0, /*tp_as_buffer*/ |
| 142 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/ |
| 143 | TextIOBase_doc, /* tp_doc */ |
| 144 | 0, /* tp_traverse */ |
| 145 | 0, /* tp_clear */ |
| 146 | 0, /* tp_richcompare */ |
| 147 | 0, /* tp_weaklistoffset */ |
| 148 | 0, /* tp_iter */ |
| 149 | 0, /* tp_iternext */ |
| 150 | TextIOBase_methods, /* tp_methods */ |
| 151 | 0, /* tp_members */ |
| 152 | TextIOBase_getset, /* tp_getset */ |
| 153 | &PyIOBase_Type, /* tp_base */ |
| 154 | 0, /* tp_dict */ |
| 155 | 0, /* tp_descr_get */ |
| 156 | 0, /* tp_descr_set */ |
| 157 | 0, /* tp_dictoffset */ |
| 158 | 0, /* tp_init */ |
| 159 | 0, /* tp_alloc */ |
| 160 | 0, /* tp_new */ |
| 161 | }; |
| 162 | |
| 163 | |
| 164 | /* IncrementalNewlineDecoder */ |
| 165 | |
| 166 | PyDoc_STRVAR(IncrementalNewlineDecoder_doc, |
| 167 | "Codec used when reading a file in universal newlines mode. It wraps\n" |
| 168 | "another incremental decoder, translating \\r\\n and \\r into \\n. It also\n" |
| 169 | "records the types of newlines encountered. When used with\n" |
| 170 | "translate=False, it ensures that the newline sequence is returned in\n" |
| 171 | "one piece. When used with decoder=None, it expects unicode strings as\n" |
| 172 | "decode input and translates newlines without first invoking an external\n" |
| 173 | "decoder.\n" |
| 174 | ); |
| 175 | |
| 176 | typedef struct { |
| 177 | PyObject_HEAD |
| 178 | PyObject *decoder; |
| 179 | PyObject *errors; |
| 180 | int pendingcr:1; |
| 181 | int translate:1; |
| 182 | unsigned int seennl:3; |
| 183 | } PyNewLineDecoderObject; |
| 184 | |
| 185 | static int |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 186 | IncrementalNewlineDecoder_init(PyNewLineDecoderObject *self, |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 187 | PyObject *args, PyObject *kwds) |
| 188 | { |
| 189 | PyObject *decoder; |
| 190 | int translate; |
| 191 | PyObject *errors = NULL; |
| 192 | char *kwlist[] = {"decoder", "translate", "errors", NULL}; |
| 193 | |
| 194 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "Oi|O:IncrementalNewlineDecoder", |
| 195 | kwlist, &decoder, &translate, &errors)) |
| 196 | return -1; |
| 197 | |
| 198 | self->decoder = decoder; |
| 199 | Py_INCREF(decoder); |
| 200 | |
| 201 | if (errors == NULL) { |
| 202 | self->errors = PyUnicode_FromString("strict"); |
| 203 | if (self->errors == NULL) |
| 204 | return -1; |
| 205 | } |
| 206 | else { |
| 207 | Py_INCREF(errors); |
| 208 | self->errors = errors; |
| 209 | } |
| 210 | |
| 211 | self->translate = translate; |
| 212 | self->seennl = 0; |
| 213 | self->pendingcr = 0; |
| 214 | |
| 215 | return 0; |
| 216 | } |
| 217 | |
| 218 | static void |
| 219 | IncrementalNewlineDecoder_dealloc(PyNewLineDecoderObject *self) |
| 220 | { |
| 221 | Py_CLEAR(self->decoder); |
| 222 | Py_CLEAR(self->errors); |
| 223 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 224 | } |
| 225 | |
| 226 | #define SEEN_CR 1 |
| 227 | #define SEEN_LF 2 |
| 228 | #define SEEN_CRLF 4 |
| 229 | #define SEEN_ALL (SEEN_CR | SEEN_LF | SEEN_CRLF) |
| 230 | |
| 231 | PyObject * |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 232 | _PyIncrementalNewlineDecoder_decode(PyObject *_self, |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 233 | PyObject *input, int final) |
| 234 | { |
| 235 | PyObject *output; |
| 236 | Py_ssize_t output_len; |
| 237 | PyNewLineDecoderObject *self = (PyNewLineDecoderObject *) _self; |
| 238 | |
| 239 | if (self->decoder == NULL) { |
| 240 | PyErr_SetString(PyExc_ValueError, |
| 241 | "IncrementalNewlineDecoder.__init__ not called"); |
| 242 | return NULL; |
| 243 | } |
| 244 | |
| 245 | /* decode input (with the eventual \r from a previous pass) */ |
| 246 | if (self->decoder != Py_None) { |
| 247 | output = PyObject_CallMethodObjArgs(self->decoder, |
| 248 | _PyIO_str_decode, input, final ? Py_True : Py_False, NULL); |
| 249 | } |
| 250 | else { |
| 251 | output = input; |
| 252 | Py_INCREF(output); |
| 253 | } |
| 254 | |
| 255 | if (output == NULL) |
| 256 | return NULL; |
| 257 | |
| 258 | if (!PyUnicode_Check(output)) { |
| 259 | PyErr_SetString(PyExc_TypeError, |
| 260 | "decoder should return a string result"); |
| 261 | goto error; |
| 262 | } |
| 263 | |
| 264 | output_len = PyUnicode_GET_SIZE(output); |
| 265 | if (self->pendingcr && (final || output_len > 0)) { |
| 266 | Py_UNICODE *out; |
| 267 | PyObject *modified = PyUnicode_FromUnicode(NULL, output_len + 1); |
| 268 | if (modified == NULL) |
| 269 | goto error; |
| 270 | out = PyUnicode_AS_UNICODE(modified); |
| 271 | out[0] = '\r'; |
| 272 | memcpy(out + 1, PyUnicode_AS_UNICODE(output), |
| 273 | output_len * sizeof(Py_UNICODE)); |
| 274 | Py_DECREF(output); |
| 275 | output = modified; |
| 276 | self->pendingcr = 0; |
| 277 | output_len++; |
| 278 | } |
| 279 | |
| 280 | /* retain last \r even when not translating data: |
| 281 | * then readline() is sure to get \r\n in one pass |
| 282 | */ |
| 283 | if (!final) { |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 284 | if (output_len > 0 |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 285 | && PyUnicode_AS_UNICODE(output)[output_len - 1] == '\r') { |
| 286 | |
| 287 | if (Py_REFCNT(output) == 1) { |
| 288 | if (PyUnicode_Resize(&output, output_len - 1) < 0) |
| 289 | goto error; |
| 290 | } |
| 291 | else { |
| 292 | PyObject *modified = PyUnicode_FromUnicode( |
| 293 | PyUnicode_AS_UNICODE(output), |
| 294 | output_len - 1); |
| 295 | if (modified == NULL) |
| 296 | goto error; |
| 297 | Py_DECREF(output); |
| 298 | output = modified; |
| 299 | } |
| 300 | self->pendingcr = 1; |
| 301 | } |
| 302 | } |
| 303 | |
| 304 | /* Record which newlines are read and do newline translation if desired, |
| 305 | all in one pass. */ |
| 306 | { |
| 307 | Py_UNICODE *in_str; |
| 308 | Py_ssize_t len; |
| 309 | int seennl = self->seennl; |
| 310 | int only_lf = 0; |
| 311 | |
| 312 | in_str = PyUnicode_AS_UNICODE(output); |
| 313 | len = PyUnicode_GET_SIZE(output); |
| 314 | |
| 315 | if (len == 0) |
| 316 | return output; |
| 317 | |
| 318 | /* If, up to now, newlines are consistently \n, do a quick check |
| 319 | for the \r *byte* with the libc's optimized memchr. |
| 320 | */ |
| 321 | if (seennl == SEEN_LF || seennl == 0) { |
Antoine Pitrou | 0e94189 | 2009-03-06 23:57:20 +0000 | [diff] [blame] | 322 | only_lf = (memchr(in_str, '\r', len * sizeof(Py_UNICODE)) == NULL); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 323 | } |
| 324 | |
Antoine Pitrou | 66913e2 | 2009-03-06 23:40:56 +0000 | [diff] [blame] | 325 | if (only_lf) { |
| 326 | /* If not already seen, quick scan for a possible "\n" character. |
| 327 | (there's nothing else to be done, even when in translation mode) |
| 328 | */ |
| 329 | if (seennl == 0 && |
| 330 | memchr(in_str, '\n', len * sizeof(Py_UNICODE)) != NULL) { |
| 331 | Py_UNICODE *s, *end; |
| 332 | s = in_str; |
| 333 | end = in_str + len; |
| 334 | for (;;) { |
| 335 | Py_UNICODE c; |
| 336 | /* Fast loop for non-control characters */ |
| 337 | while (*s > '\n') |
| 338 | s++; |
| 339 | c = *s++; |
| 340 | if (c == '\n') { |
| 341 | seennl |= SEEN_LF; |
| 342 | break; |
| 343 | } |
| 344 | if (s > end) |
| 345 | break; |
| 346 | } |
| 347 | } |
| 348 | /* Finished: we have scanned for newlines, and none of them |
| 349 | need translating */ |
| 350 | } |
| 351 | else if (!self->translate) { |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 352 | Py_UNICODE *s, *end; |
Antoine Pitrou | 66913e2 | 2009-03-06 23:40:56 +0000 | [diff] [blame] | 353 | /* We have already seen all newline types, no need to scan again */ |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 354 | if (seennl == SEEN_ALL) |
| 355 | goto endscan; |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 356 | s = in_str; |
| 357 | end = in_str + len; |
| 358 | for (;;) { |
| 359 | Py_UNICODE c; |
| 360 | /* Fast loop for non-control characters */ |
| 361 | while (*s > '\r') |
| 362 | s++; |
| 363 | c = *s++; |
| 364 | if (c == '\n') |
| 365 | seennl |= SEEN_LF; |
| 366 | else if (c == '\r') { |
| 367 | if (*s == '\n') { |
| 368 | seennl |= SEEN_CRLF; |
| 369 | s++; |
| 370 | } |
| 371 | else |
| 372 | seennl |= SEEN_CR; |
| 373 | } |
| 374 | if (s > end) |
| 375 | break; |
| 376 | if (seennl == SEEN_ALL) |
| 377 | break; |
| 378 | } |
| 379 | endscan: |
| 380 | ; |
| 381 | } |
Antoine Pitrou | 66913e2 | 2009-03-06 23:40:56 +0000 | [diff] [blame] | 382 | else { |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 383 | PyObject *translated = NULL; |
| 384 | Py_UNICODE *out_str; |
| 385 | Py_UNICODE *in, *out, *end; |
| 386 | if (Py_REFCNT(output) != 1) { |
| 387 | /* We could try to optimize this so that we only do a copy |
| 388 | when there is something to translate. On the other hand, |
| 389 | most decoders should only output non-shared strings, i.e. |
| 390 | translation is done in place. */ |
| 391 | translated = PyUnicode_FromUnicode(NULL, len); |
| 392 | if (translated == NULL) |
| 393 | goto error; |
| 394 | assert(Py_REFCNT(translated) == 1); |
| 395 | memcpy(PyUnicode_AS_UNICODE(translated), |
| 396 | PyUnicode_AS_UNICODE(output), |
| 397 | len * sizeof(Py_UNICODE)); |
| 398 | } |
| 399 | else { |
| 400 | translated = output; |
| 401 | } |
| 402 | out_str = PyUnicode_AS_UNICODE(translated); |
| 403 | in = in_str; |
| 404 | out = out_str; |
| 405 | end = in_str + len; |
| 406 | for (;;) { |
| 407 | Py_UNICODE c; |
| 408 | /* Fast loop for non-control characters */ |
| 409 | while ((c = *in++) > '\r') |
| 410 | *out++ = c; |
| 411 | if (c == '\n') { |
| 412 | *out++ = c; |
| 413 | seennl |= SEEN_LF; |
| 414 | continue; |
| 415 | } |
| 416 | if (c == '\r') { |
| 417 | if (*in == '\n') { |
| 418 | in++; |
| 419 | seennl |= SEEN_CRLF; |
| 420 | } |
| 421 | else |
| 422 | seennl |= SEEN_CR; |
| 423 | *out++ = '\n'; |
| 424 | continue; |
| 425 | } |
| 426 | if (in > end) |
| 427 | break; |
| 428 | *out++ = c; |
| 429 | } |
| 430 | if (translated != output) { |
| 431 | Py_DECREF(output); |
| 432 | output = translated; |
| 433 | } |
| 434 | if (out - out_str != len) { |
| 435 | if (PyUnicode_Resize(&output, out - out_str) < 0) |
| 436 | goto error; |
| 437 | } |
| 438 | } |
| 439 | self->seennl |= seennl; |
| 440 | } |
| 441 | |
| 442 | return output; |
| 443 | |
| 444 | error: |
| 445 | Py_DECREF(output); |
| 446 | return NULL; |
| 447 | } |
| 448 | |
| 449 | static PyObject * |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 450 | IncrementalNewlineDecoder_decode(PyNewLineDecoderObject *self, |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 451 | PyObject *args, PyObject *kwds) |
| 452 | { |
| 453 | char *kwlist[] = {"input", "final", NULL}; |
| 454 | PyObject *input; |
| 455 | int final = 0; |
| 456 | |
| 457 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|i:IncrementalNewlineDecoder", |
| 458 | kwlist, &input, &final)) |
| 459 | return NULL; |
| 460 | return _PyIncrementalNewlineDecoder_decode((PyObject *) self, input, final); |
| 461 | } |
| 462 | |
| 463 | static PyObject * |
| 464 | IncrementalNewlineDecoder_getstate(PyNewLineDecoderObject *self, PyObject *args) |
| 465 | { |
| 466 | PyObject *buffer; |
| 467 | unsigned PY_LONG_LONG flag; |
| 468 | |
| 469 | if (self->decoder != Py_None) { |
| 470 | PyObject *state = PyObject_CallMethodObjArgs(self->decoder, |
| 471 | _PyIO_str_getstate, NULL); |
| 472 | if (state == NULL) |
| 473 | return NULL; |
| 474 | if (!PyArg_Parse(state, "(OK)", &buffer, &flag)) { |
| 475 | Py_DECREF(state); |
| 476 | return NULL; |
| 477 | } |
| 478 | Py_INCREF(buffer); |
| 479 | Py_DECREF(state); |
| 480 | } |
| 481 | else { |
| 482 | buffer = PyBytes_FromString(""); |
| 483 | flag = 0; |
| 484 | } |
| 485 | flag <<= 1; |
| 486 | if (self->pendingcr) |
| 487 | flag |= 1; |
| 488 | return Py_BuildValue("NK", buffer, flag); |
| 489 | } |
| 490 | |
| 491 | static PyObject * |
| 492 | IncrementalNewlineDecoder_setstate(PyNewLineDecoderObject *self, PyObject *state) |
| 493 | { |
| 494 | PyObject *buffer; |
| 495 | unsigned PY_LONG_LONG flag; |
| 496 | |
| 497 | if (!PyArg_Parse(state, "(OK)", &buffer, &flag)) |
| 498 | return NULL; |
| 499 | |
| 500 | self->pendingcr = (int) flag & 1; |
| 501 | flag >>= 1; |
| 502 | |
| 503 | if (self->decoder != Py_None) |
| 504 | return PyObject_CallMethod(self->decoder, |
| 505 | "setstate", "((OK))", buffer, flag); |
| 506 | else |
| 507 | Py_RETURN_NONE; |
| 508 | } |
| 509 | |
| 510 | static PyObject * |
| 511 | IncrementalNewlineDecoder_reset(PyNewLineDecoderObject *self, PyObject *args) |
| 512 | { |
| 513 | self->seennl = 0; |
| 514 | self->pendingcr = 0; |
| 515 | if (self->decoder != Py_None) |
| 516 | return PyObject_CallMethodObjArgs(self->decoder, _PyIO_str_reset, NULL); |
| 517 | else |
| 518 | Py_RETURN_NONE; |
| 519 | } |
| 520 | |
| 521 | static PyObject * |
| 522 | IncrementalNewlineDecoder_newlines_get(PyNewLineDecoderObject *self, void *context) |
| 523 | { |
| 524 | switch (self->seennl) { |
| 525 | case SEEN_CR: |
| 526 | return PyUnicode_FromString("\r"); |
| 527 | case SEEN_LF: |
| 528 | return PyUnicode_FromString("\n"); |
| 529 | case SEEN_CRLF: |
| 530 | return PyUnicode_FromString("\r\n"); |
| 531 | case SEEN_CR | SEEN_LF: |
| 532 | return Py_BuildValue("ss", "\r", "\n"); |
| 533 | case SEEN_CR | SEEN_CRLF: |
| 534 | return Py_BuildValue("ss", "\r", "\r\n"); |
| 535 | case SEEN_LF | SEEN_CRLF: |
| 536 | return Py_BuildValue("ss", "\n", "\r\n"); |
| 537 | case SEEN_CR | SEEN_LF | SEEN_CRLF: |
| 538 | return Py_BuildValue("sss", "\r", "\n", "\r\n"); |
| 539 | default: |
| 540 | Py_RETURN_NONE; |
| 541 | } |
| 542 | |
| 543 | } |
| 544 | |
| 545 | |
| 546 | static PyMethodDef IncrementalNewlineDecoder_methods[] = { |
| 547 | {"decode", (PyCFunction)IncrementalNewlineDecoder_decode, METH_VARARGS|METH_KEYWORDS}, |
| 548 | {"getstate", (PyCFunction)IncrementalNewlineDecoder_getstate, METH_NOARGS}, |
| 549 | {"setstate", (PyCFunction)IncrementalNewlineDecoder_setstate, METH_O}, |
| 550 | {"reset", (PyCFunction)IncrementalNewlineDecoder_reset, METH_NOARGS}, |
Benjamin Peterson | 1fea321 | 2009-04-19 03:15:20 +0000 | [diff] [blame] | 551 | {NULL} |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 552 | }; |
| 553 | |
| 554 | static PyGetSetDef IncrementalNewlineDecoder_getset[] = { |
| 555 | {"newlines", (getter)IncrementalNewlineDecoder_newlines_get, NULL, NULL}, |
Benjamin Peterson | 1fea321 | 2009-04-19 03:15:20 +0000 | [diff] [blame] | 556 | {NULL} |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 557 | }; |
| 558 | |
| 559 | PyTypeObject PyIncrementalNewlineDecoder_Type = { |
| 560 | PyVarObject_HEAD_INIT(NULL, 0) |
| 561 | "_io.IncrementalNewlineDecoder", /*tp_name*/ |
| 562 | sizeof(PyNewLineDecoderObject), /*tp_basicsize*/ |
| 563 | 0, /*tp_itemsize*/ |
| 564 | (destructor)IncrementalNewlineDecoder_dealloc, /*tp_dealloc*/ |
| 565 | 0, /*tp_print*/ |
| 566 | 0, /*tp_getattr*/ |
| 567 | 0, /*tp_setattr*/ |
| 568 | 0, /*tp_compare */ |
| 569 | 0, /*tp_repr*/ |
| 570 | 0, /*tp_as_number*/ |
| 571 | 0, /*tp_as_sequence*/ |
| 572 | 0, /*tp_as_mapping*/ |
| 573 | 0, /*tp_hash */ |
| 574 | 0, /*tp_call*/ |
| 575 | 0, /*tp_str*/ |
| 576 | 0, /*tp_getattro*/ |
| 577 | 0, /*tp_setattro*/ |
| 578 | 0, /*tp_as_buffer*/ |
| 579 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/ |
| 580 | IncrementalNewlineDecoder_doc, /* tp_doc */ |
| 581 | 0, /* tp_traverse */ |
| 582 | 0, /* tp_clear */ |
| 583 | 0, /* tp_richcompare */ |
| 584 | 0, /*tp_weaklistoffset*/ |
| 585 | 0, /* tp_iter */ |
| 586 | 0, /* tp_iternext */ |
| 587 | IncrementalNewlineDecoder_methods, /* tp_methods */ |
| 588 | 0, /* tp_members */ |
| 589 | IncrementalNewlineDecoder_getset, /* tp_getset */ |
| 590 | 0, /* tp_base */ |
| 591 | 0, /* tp_dict */ |
| 592 | 0, /* tp_descr_get */ |
| 593 | 0, /* tp_descr_set */ |
| 594 | 0, /* tp_dictoffset */ |
| 595 | (initproc)IncrementalNewlineDecoder_init, /* tp_init */ |
| 596 | 0, /* tp_alloc */ |
| 597 | PyType_GenericNew, /* tp_new */ |
| 598 | }; |
| 599 | |
| 600 | |
| 601 | /* TextIOWrapper */ |
| 602 | |
| 603 | PyDoc_STRVAR(TextIOWrapper_doc, |
| 604 | "Character and line based layer over a BufferedIOBase object, buffer.\n" |
| 605 | "\n" |
| 606 | "encoding gives the name of the encoding that the stream will be\n" |
| 607 | "decoded or encoded with. It defaults to locale.getpreferredencoding.\n" |
| 608 | "\n" |
| 609 | "errors determines the strictness of encoding and decoding (see the\n" |
| 610 | "codecs.register) and defaults to \"strict\".\n" |
| 611 | "\n" |
| 612 | "newline can be None, '', '\\n', '\\r', or '\\r\\n'. It controls the\n" |
| 613 | "handling of line endings. If it is None, universal newlines is\n" |
| 614 | "enabled. With this enabled, on input, the lines endings '\\n', '\\r',\n" |
| 615 | "or '\\r\\n' are translated to '\\n' before being returned to the\n" |
| 616 | "caller. Conversely, on output, '\\n' is translated to the system\n" |
| 617 | "default line seperator, os.linesep. If newline is any other of its\n" |
| 618 | "legal values, that newline becomes the newline when the file is read\n" |
| 619 | "and it is returned untranslated. On output, '\\n' is converted to the\n" |
| 620 | "newline.\n" |
| 621 | "\n" |
| 622 | "If line_buffering is True, a call to flush is implied when a call to\n" |
| 623 | "write contains a newline character." |
| 624 | ); |
| 625 | |
| 626 | typedef PyObject * |
| 627 | (*encodefunc_t)(PyObject *, PyObject *); |
| 628 | |
| 629 | typedef struct |
| 630 | { |
| 631 | PyObject_HEAD |
| 632 | int ok; /* initialized? */ |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 633 | int detached; |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 634 | Py_ssize_t chunk_size; |
| 635 | PyObject *buffer; |
| 636 | PyObject *encoding; |
| 637 | PyObject *encoder; |
| 638 | PyObject *decoder; |
| 639 | PyObject *readnl; |
| 640 | PyObject *errors; |
| 641 | const char *writenl; /* utf-8 encoded, NULL stands for \n */ |
| 642 | char line_buffering; |
| 643 | char readuniversal; |
| 644 | char readtranslate; |
| 645 | char writetranslate; |
| 646 | char seekable; |
| 647 | char telling; |
| 648 | /* Specialized encoding func (see below) */ |
| 649 | encodefunc_t encodefunc; |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 650 | /* Whether or not it's the start of the stream */ |
| 651 | char encoding_start_of_stream; |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 652 | |
| 653 | /* Reads and writes are internally buffered in order to speed things up. |
| 654 | However, any read will first flush the write buffer if itsn't empty. |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 655 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 656 | Please also note that text to be written is first encoded before being |
| 657 | buffered. This is necessary so that encoding errors are immediately |
| 658 | reported to the caller, but it unfortunately means that the |
| 659 | IncrementalEncoder (whose encode() method is always written in Python) |
| 660 | becomes a bottleneck for small writes. |
| 661 | */ |
| 662 | PyObject *decoded_chars; /* buffer for text returned from decoder */ |
| 663 | Py_ssize_t decoded_chars_used; /* offset into _decoded_chars for read() */ |
| 664 | PyObject *pending_bytes; /* list of bytes objects waiting to be |
| 665 | written, or NULL */ |
| 666 | Py_ssize_t pending_bytes_count; |
| 667 | PyObject *snapshot; |
| 668 | /* snapshot is either None, or a tuple (dec_flags, next_input) where |
| 669 | * dec_flags is the second (integer) item of the decoder state and |
| 670 | * next_input is the chunk of input bytes that comes next after the |
| 671 | * snapshot point. We use this to reconstruct decoder states in tell(). |
| 672 | */ |
| 673 | |
| 674 | /* Cache raw object if it's a FileIO object */ |
| 675 | PyObject *raw; |
| 676 | |
| 677 | PyObject *weakreflist; |
| 678 | PyObject *dict; |
| 679 | } PyTextIOWrapperObject; |
| 680 | |
| 681 | |
| 682 | /* A couple of specialized cases in order to bypass the slow incremental |
| 683 | encoding methods for the most popular encodings. */ |
| 684 | |
| 685 | static PyObject * |
| 686 | ascii_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 687 | { |
| 688 | return PyUnicode_EncodeASCII(PyUnicode_AS_UNICODE(text), |
| 689 | PyUnicode_GET_SIZE(text), |
| 690 | PyBytes_AS_STRING(self->errors)); |
| 691 | } |
| 692 | |
| 693 | static PyObject * |
| 694 | utf16be_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 695 | { |
| 696 | return PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(text), |
| 697 | PyUnicode_GET_SIZE(text), |
| 698 | PyBytes_AS_STRING(self->errors), 1); |
| 699 | } |
| 700 | |
| 701 | static PyObject * |
| 702 | utf16le_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 703 | { |
| 704 | return PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(text), |
| 705 | PyUnicode_GET_SIZE(text), |
| 706 | PyBytes_AS_STRING(self->errors), -1); |
| 707 | } |
| 708 | |
| 709 | static PyObject * |
| 710 | utf16_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 711 | { |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 712 | if (!self->encoding_start_of_stream) { |
| 713 | /* Skip the BOM and use native byte ordering */ |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 714 | #if defined(WORDS_BIGENDIAN) |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 715 | return utf16be_encode(self, text); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 716 | #else |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 717 | return utf16le_encode(self, text); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 718 | #endif |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 719 | } |
| 720 | return PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(text), |
| 721 | PyUnicode_GET_SIZE(text), |
| 722 | PyBytes_AS_STRING(self->errors), 0); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 723 | } |
| 724 | |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 725 | static PyObject * |
| 726 | utf32be_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 727 | { |
| 728 | return PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(text), |
| 729 | PyUnicode_GET_SIZE(text), |
| 730 | PyBytes_AS_STRING(self->errors), 1); |
| 731 | } |
| 732 | |
| 733 | static PyObject * |
| 734 | utf32le_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 735 | { |
| 736 | return PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(text), |
| 737 | PyUnicode_GET_SIZE(text), |
| 738 | PyBytes_AS_STRING(self->errors), -1); |
| 739 | } |
| 740 | |
| 741 | static PyObject * |
| 742 | utf32_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 743 | { |
| 744 | if (!self->encoding_start_of_stream) { |
| 745 | /* Skip the BOM and use native byte ordering */ |
| 746 | #if defined(WORDS_BIGENDIAN) |
| 747 | return utf32be_encode(self, text); |
| 748 | #else |
| 749 | return utf32le_encode(self, text); |
| 750 | #endif |
| 751 | } |
| 752 | return PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(text), |
| 753 | PyUnicode_GET_SIZE(text), |
| 754 | PyBytes_AS_STRING(self->errors), 0); |
| 755 | } |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 756 | |
| 757 | static PyObject * |
| 758 | utf8_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 759 | { |
| 760 | return PyUnicode_EncodeUTF8(PyUnicode_AS_UNICODE(text), |
| 761 | PyUnicode_GET_SIZE(text), |
| 762 | PyBytes_AS_STRING(self->errors)); |
| 763 | } |
| 764 | |
| 765 | static PyObject * |
| 766 | latin1_encode(PyTextIOWrapperObject *self, PyObject *text) |
| 767 | { |
| 768 | return PyUnicode_EncodeLatin1(PyUnicode_AS_UNICODE(text), |
| 769 | PyUnicode_GET_SIZE(text), |
| 770 | PyBytes_AS_STRING(self->errors)); |
| 771 | } |
| 772 | |
| 773 | /* Map normalized encoding names onto the specialized encoding funcs */ |
| 774 | |
| 775 | typedef struct { |
| 776 | const char *name; |
| 777 | encodefunc_t encodefunc; |
| 778 | } encodefuncentry; |
| 779 | |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 780 | static encodefuncentry encodefuncs[] = { |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 781 | {"ascii", (encodefunc_t) ascii_encode}, |
| 782 | {"iso8859-1", (encodefunc_t) latin1_encode}, |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 783 | {"utf-8", (encodefunc_t) utf8_encode}, |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 784 | {"utf-16-be", (encodefunc_t) utf16be_encode}, |
| 785 | {"utf-16-le", (encodefunc_t) utf16le_encode}, |
| 786 | {"utf-16", (encodefunc_t) utf16_encode}, |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 787 | {"utf-32-be", (encodefunc_t) utf32be_encode}, |
| 788 | {"utf-32-le", (encodefunc_t) utf32le_encode}, |
| 789 | {"utf-32", (encodefunc_t) utf32_encode}, |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 790 | {NULL, NULL} |
| 791 | }; |
| 792 | |
| 793 | |
| 794 | static int |
| 795 | TextIOWrapper_init(PyTextIOWrapperObject *self, PyObject *args, PyObject *kwds) |
| 796 | { |
| 797 | char *kwlist[] = {"buffer", "encoding", "errors", |
| 798 | "newline", "line_buffering", |
| 799 | NULL}; |
| 800 | PyObject *buffer, *raw; |
| 801 | char *encoding = NULL; |
| 802 | char *errors = NULL; |
| 803 | char *newline = NULL; |
| 804 | int line_buffering = 0; |
| 805 | _PyIO_State *state = IO_STATE; |
| 806 | |
| 807 | PyObject *res; |
| 808 | int r; |
| 809 | |
| 810 | self->ok = 0; |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 811 | self->detached = 0; |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 812 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|zzzi:fileio", |
| 813 | kwlist, &buffer, &encoding, &errors, |
| 814 | &newline, &line_buffering)) |
| 815 | return -1; |
| 816 | |
| 817 | if (newline && newline[0] != '\0' |
| 818 | && !(newline[0] == '\n' && newline[1] == '\0') |
| 819 | && !(newline[0] == '\r' && newline[1] == '\0') |
| 820 | && !(newline[0] == '\r' && newline[1] == '\n' && newline[2] == '\0')) { |
| 821 | PyErr_Format(PyExc_ValueError, |
| 822 | "illegal newline value: %s", newline); |
| 823 | return -1; |
| 824 | } |
| 825 | |
| 826 | Py_CLEAR(self->buffer); |
| 827 | Py_CLEAR(self->encoding); |
| 828 | Py_CLEAR(self->encoder); |
| 829 | Py_CLEAR(self->decoder); |
| 830 | Py_CLEAR(self->readnl); |
| 831 | Py_CLEAR(self->decoded_chars); |
| 832 | Py_CLEAR(self->pending_bytes); |
| 833 | Py_CLEAR(self->snapshot); |
| 834 | Py_CLEAR(self->errors); |
| 835 | Py_CLEAR(self->raw); |
| 836 | self->decoded_chars_used = 0; |
| 837 | self->pending_bytes_count = 0; |
| 838 | self->encodefunc = NULL; |
| 839 | |
| 840 | if (encoding == NULL) { |
| 841 | /* Try os.device_encoding(fileno) */ |
| 842 | PyObject *fileno; |
| 843 | fileno = PyObject_CallMethod(buffer, "fileno", NULL); |
| 844 | /* Ignore only AttributeError and UnsupportedOperation */ |
| 845 | if (fileno == NULL) { |
| 846 | if (PyErr_ExceptionMatches(PyExc_AttributeError) || |
| 847 | PyErr_ExceptionMatches(state->unsupported_operation)) { |
| 848 | PyErr_Clear(); |
| 849 | } |
| 850 | else { |
| 851 | goto error; |
| 852 | } |
| 853 | } |
| 854 | else { |
| 855 | self->encoding = PyObject_CallMethod(state->os_module, |
| 856 | "device_encoding", |
| 857 | "N", fileno); |
| 858 | if (self->encoding == NULL) |
| 859 | goto error; |
| 860 | else if (!PyUnicode_Check(self->encoding)) |
| 861 | Py_CLEAR(self->encoding); |
| 862 | } |
| 863 | } |
| 864 | if (encoding == NULL && self->encoding == NULL) { |
| 865 | if (state->locale_module == NULL) { |
| 866 | state->locale_module = PyImport_ImportModule("locale"); |
| 867 | if (state->locale_module == NULL) |
| 868 | goto catch_ImportError; |
| 869 | else |
| 870 | goto use_locale; |
| 871 | } |
| 872 | else { |
| 873 | use_locale: |
| 874 | self->encoding = PyObject_CallMethod( |
| 875 | state->locale_module, "getpreferredencoding", NULL); |
| 876 | if (self->encoding == NULL) { |
| 877 | catch_ImportError: |
| 878 | /* |
| 879 | Importing locale can raise a ImportError because of |
| 880 | _functools, and locale.getpreferredencoding can raise a |
| 881 | ImportError if _locale is not available. These will happen |
| 882 | during module building. |
| 883 | */ |
| 884 | if (PyErr_ExceptionMatches(PyExc_ImportError)) { |
| 885 | PyErr_Clear(); |
| 886 | self->encoding = PyUnicode_FromString("ascii"); |
| 887 | } |
| 888 | else |
| 889 | goto error; |
| 890 | } |
| 891 | else if (!PyUnicode_Check(self->encoding)) |
| 892 | Py_CLEAR(self->encoding); |
| 893 | } |
| 894 | } |
| 895 | if (self->encoding != NULL) |
| 896 | encoding = _PyUnicode_AsString(self->encoding); |
| 897 | else if (encoding != NULL) { |
| 898 | self->encoding = PyUnicode_FromString(encoding); |
| 899 | if (self->encoding == NULL) |
| 900 | goto error; |
| 901 | } |
| 902 | else { |
| 903 | PyErr_SetString(PyExc_IOError, |
| 904 | "could not determine default encoding"); |
| 905 | } |
| 906 | |
| 907 | if (errors == NULL) |
| 908 | errors = "strict"; |
| 909 | self->errors = PyBytes_FromString(errors); |
| 910 | if (self->errors == NULL) |
| 911 | goto error; |
| 912 | |
| 913 | self->chunk_size = 8192; |
| 914 | self->readuniversal = (newline == NULL || newline[0] == '\0'); |
| 915 | self->line_buffering = line_buffering; |
| 916 | self->readtranslate = (newline == NULL); |
| 917 | if (newline) { |
| 918 | self->readnl = PyUnicode_FromString(newline); |
| 919 | if (self->readnl == NULL) |
| 920 | return -1; |
| 921 | } |
| 922 | self->writetranslate = (newline == NULL || newline[0] != '\0'); |
| 923 | if (!self->readuniversal && self->readnl) { |
| 924 | self->writenl = _PyUnicode_AsString(self->readnl); |
| 925 | if (!strcmp(self->writenl, "\n")) |
| 926 | self->writenl = NULL; |
| 927 | } |
| 928 | #ifdef MS_WINDOWS |
| 929 | else |
| 930 | self->writenl = "\r\n"; |
| 931 | #endif |
| 932 | |
| 933 | /* Build the decoder object */ |
| 934 | res = PyObject_CallMethod(buffer, "readable", NULL); |
| 935 | if (res == NULL) |
| 936 | goto error; |
| 937 | r = PyObject_IsTrue(res); |
| 938 | Py_DECREF(res); |
| 939 | if (r == -1) |
| 940 | goto error; |
| 941 | if (r == 1) { |
| 942 | self->decoder = PyCodec_IncrementalDecoder( |
| 943 | encoding, errors); |
| 944 | if (self->decoder == NULL) |
| 945 | goto error; |
| 946 | |
| 947 | if (self->readuniversal) { |
| 948 | PyObject *incrementalDecoder = PyObject_CallFunction( |
| 949 | (PyObject *)&PyIncrementalNewlineDecoder_Type, |
| 950 | "Oi", self->decoder, (int)self->readtranslate); |
| 951 | if (incrementalDecoder == NULL) |
| 952 | goto error; |
| 953 | Py_CLEAR(self->decoder); |
| 954 | self->decoder = incrementalDecoder; |
| 955 | } |
| 956 | } |
| 957 | |
| 958 | /* Build the encoder object */ |
| 959 | res = PyObject_CallMethod(buffer, "writable", NULL); |
| 960 | if (res == NULL) |
| 961 | goto error; |
| 962 | r = PyObject_IsTrue(res); |
| 963 | Py_DECREF(res); |
| 964 | if (r == -1) |
| 965 | goto error; |
| 966 | if (r == 1) { |
| 967 | PyObject *ci; |
| 968 | self->encoder = PyCodec_IncrementalEncoder( |
| 969 | encoding, errors); |
| 970 | if (self->encoder == NULL) |
| 971 | goto error; |
| 972 | /* Get the normalized named of the codec */ |
| 973 | ci = _PyCodec_Lookup(encoding); |
| 974 | if (ci == NULL) |
| 975 | goto error; |
| 976 | res = PyObject_GetAttrString(ci, "name"); |
| 977 | Py_DECREF(ci); |
| 978 | if (res == NULL) |
| 979 | PyErr_Clear(); |
| 980 | else if (PyUnicode_Check(res)) { |
| 981 | encodefuncentry *e = encodefuncs; |
| 982 | while (e->name != NULL) { |
| 983 | if (!PyUnicode_CompareWithASCIIString(res, e->name)) { |
| 984 | self->encodefunc = e->encodefunc; |
| 985 | break; |
| 986 | } |
| 987 | e++; |
| 988 | } |
| 989 | } |
| 990 | Py_XDECREF(res); |
| 991 | } |
| 992 | |
| 993 | self->buffer = buffer; |
| 994 | Py_INCREF(buffer); |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 995 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 996 | if (Py_TYPE(buffer) == &PyBufferedReader_Type || |
| 997 | Py_TYPE(buffer) == &PyBufferedWriter_Type || |
| 998 | Py_TYPE(buffer) == &PyBufferedRandom_Type) { |
| 999 | raw = PyObject_GetAttrString(buffer, "raw"); |
| 1000 | /* Cache the raw FileIO object to speed up 'closed' checks */ |
| 1001 | if (raw == NULL) |
| 1002 | PyErr_Clear(); |
| 1003 | else if (Py_TYPE(raw) == &PyFileIO_Type) |
| 1004 | self->raw = raw; |
| 1005 | else |
| 1006 | Py_DECREF(raw); |
| 1007 | } |
| 1008 | |
| 1009 | res = PyObject_CallMethod(buffer, "seekable", NULL); |
| 1010 | if (res == NULL) |
| 1011 | goto error; |
| 1012 | self->seekable = self->telling = PyObject_IsTrue(res); |
| 1013 | Py_DECREF(res); |
| 1014 | |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 1015 | self->encoding_start_of_stream = 0; |
| 1016 | if (self->seekable && self->encoder) { |
| 1017 | PyObject *cookieObj; |
| 1018 | int cmp; |
| 1019 | |
| 1020 | self->encoding_start_of_stream = 1; |
| 1021 | |
| 1022 | cookieObj = PyObject_CallMethodObjArgs(buffer, _PyIO_str_tell, NULL); |
| 1023 | if (cookieObj == NULL) |
| 1024 | goto error; |
| 1025 | |
| 1026 | cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_EQ); |
| 1027 | Py_DECREF(cookieObj); |
| 1028 | if (cmp < 0) { |
| 1029 | goto error; |
| 1030 | } |
| 1031 | |
| 1032 | if (cmp == 0) { |
| 1033 | self->encoding_start_of_stream = 0; |
| 1034 | res = PyObject_CallMethodObjArgs(self->encoder, _PyIO_str_setstate, |
| 1035 | _PyIO_zero, NULL); |
| 1036 | if (res == NULL) |
| 1037 | goto error; |
| 1038 | Py_DECREF(res); |
| 1039 | } |
| 1040 | } |
| 1041 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1042 | self->ok = 1; |
| 1043 | return 0; |
| 1044 | |
| 1045 | error: |
| 1046 | return -1; |
| 1047 | } |
| 1048 | |
| 1049 | static int |
| 1050 | _TextIOWrapper_clear(PyTextIOWrapperObject *self) |
| 1051 | { |
| 1052 | if (self->ok && _PyIOBase_finalize((PyObject *) self) < 0) |
| 1053 | return -1; |
| 1054 | self->ok = 0; |
| 1055 | Py_CLEAR(self->buffer); |
| 1056 | Py_CLEAR(self->encoding); |
| 1057 | Py_CLEAR(self->encoder); |
| 1058 | Py_CLEAR(self->decoder); |
| 1059 | Py_CLEAR(self->readnl); |
| 1060 | Py_CLEAR(self->decoded_chars); |
| 1061 | Py_CLEAR(self->pending_bytes); |
| 1062 | Py_CLEAR(self->snapshot); |
| 1063 | Py_CLEAR(self->errors); |
| 1064 | Py_CLEAR(self->raw); |
| 1065 | return 0; |
| 1066 | } |
| 1067 | |
| 1068 | static void |
| 1069 | TextIOWrapper_dealloc(PyTextIOWrapperObject *self) |
| 1070 | { |
| 1071 | if (_TextIOWrapper_clear(self) < 0) |
| 1072 | return; |
| 1073 | _PyObject_GC_UNTRACK(self); |
| 1074 | if (self->weakreflist != NULL) |
| 1075 | PyObject_ClearWeakRefs((PyObject *)self); |
| 1076 | Py_CLEAR(self->dict); |
| 1077 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 1078 | } |
| 1079 | |
| 1080 | static int |
| 1081 | TextIOWrapper_traverse(PyTextIOWrapperObject *self, visitproc visit, void *arg) |
| 1082 | { |
| 1083 | Py_VISIT(self->buffer); |
| 1084 | Py_VISIT(self->encoding); |
| 1085 | Py_VISIT(self->encoder); |
| 1086 | Py_VISIT(self->decoder); |
| 1087 | Py_VISIT(self->readnl); |
| 1088 | Py_VISIT(self->decoded_chars); |
| 1089 | Py_VISIT(self->pending_bytes); |
| 1090 | Py_VISIT(self->snapshot); |
| 1091 | Py_VISIT(self->errors); |
| 1092 | Py_VISIT(self->raw); |
| 1093 | |
| 1094 | Py_VISIT(self->dict); |
| 1095 | return 0; |
| 1096 | } |
| 1097 | |
| 1098 | static int |
| 1099 | TextIOWrapper_clear(PyTextIOWrapperObject *self) |
| 1100 | { |
| 1101 | if (_TextIOWrapper_clear(self) < 0) |
| 1102 | return -1; |
| 1103 | Py_CLEAR(self->dict); |
| 1104 | return 0; |
| 1105 | } |
| 1106 | |
| 1107 | static PyObject * |
| 1108 | TextIOWrapper_closed_get(PyTextIOWrapperObject *self, void *context); |
| 1109 | |
| 1110 | /* This macro takes some shortcuts to make the common case faster. */ |
| 1111 | #define CHECK_CLOSED(self) \ |
| 1112 | do { \ |
| 1113 | int r; \ |
| 1114 | PyObject *_res; \ |
| 1115 | if (Py_TYPE(self) == &PyTextIOWrapper_Type) { \ |
| 1116 | if (self->raw != NULL) \ |
| 1117 | r = _PyFileIO_closed(self->raw); \ |
| 1118 | else { \ |
| 1119 | _res = TextIOWrapper_closed_get(self, NULL); \ |
| 1120 | if (_res == NULL) \ |
| 1121 | return NULL; \ |
| 1122 | r = PyObject_IsTrue(_res); \ |
| 1123 | Py_DECREF(_res); \ |
| 1124 | if (r < 0) \ |
| 1125 | return NULL; \ |
| 1126 | } \ |
| 1127 | if (r > 0) { \ |
| 1128 | PyErr_SetString(PyExc_ValueError, \ |
| 1129 | "I/O operation on closed file."); \ |
| 1130 | return NULL; \ |
| 1131 | } \ |
| 1132 | } \ |
| 1133 | else if (_PyIOBase_checkClosed((PyObject *)self, Py_True) == NULL) \ |
| 1134 | return NULL; \ |
| 1135 | } while (0) |
| 1136 | |
| 1137 | #define CHECK_INITIALIZED(self) \ |
| 1138 | if (self->ok <= 0) { \ |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 1139 | if (self->detached) { \ |
| 1140 | PyErr_SetString(PyExc_ValueError, \ |
| 1141 | "underlying buffer has been detached"); \ |
| 1142 | } else { \ |
| 1143 | PyErr_SetString(PyExc_ValueError, \ |
| 1144 | "I/O operation on uninitialized object"); \ |
| 1145 | } \ |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1146 | return NULL; \ |
| 1147 | } |
| 1148 | |
| 1149 | #define CHECK_INITIALIZED_INT(self) \ |
| 1150 | if (self->ok <= 0) { \ |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 1151 | if (self->detached) { \ |
| 1152 | PyErr_SetString(PyExc_ValueError, \ |
| 1153 | "underlying buffer has been detached"); \ |
| 1154 | } else { \ |
| 1155 | PyErr_SetString(PyExc_ValueError, \ |
| 1156 | "I/O operation on uninitialized object"); \ |
| 1157 | } \ |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1158 | return -1; \ |
| 1159 | } |
| 1160 | |
| 1161 | |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 1162 | static PyObject * |
| 1163 | TextIOWrapper_detach(PyTextIOWrapperObject *self) |
| 1164 | { |
| 1165 | PyObject *buffer, *res; |
| 1166 | CHECK_INITIALIZED(self); |
| 1167 | res = PyObject_CallMethodObjArgs((PyObject *)self, _PyIO_str_flush, NULL); |
| 1168 | if (res == NULL) |
| 1169 | return NULL; |
| 1170 | Py_DECREF(res); |
| 1171 | buffer = self->buffer; |
| 1172 | self->buffer = NULL; |
| 1173 | self->detached = 1; |
| 1174 | self->ok = 0; |
| 1175 | return buffer; |
| 1176 | } |
| 1177 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1178 | Py_LOCAL_INLINE(const Py_UNICODE *) |
| 1179 | findchar(const Py_UNICODE *s, Py_ssize_t size, Py_UNICODE ch) |
| 1180 | { |
| 1181 | /* like wcschr, but doesn't stop at NULL characters */ |
| 1182 | while (size-- > 0) { |
| 1183 | if (*s == ch) |
| 1184 | return s; |
| 1185 | s++; |
| 1186 | } |
| 1187 | return NULL; |
| 1188 | } |
| 1189 | |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 1190 | /* Flush the internal write buffer. This doesn't explicitly flush the |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1191 | underlying buffered object, though. */ |
| 1192 | static int |
| 1193 | _TextIOWrapper_writeflush(PyTextIOWrapperObject *self) |
| 1194 | { |
| 1195 | PyObject *b, *ret; |
| 1196 | |
| 1197 | if (self->pending_bytes == NULL) |
| 1198 | return 0; |
| 1199 | b = _PyBytes_Join(_PyIO_empty_bytes, self->pending_bytes); |
| 1200 | if (b == NULL) |
| 1201 | return -1; |
| 1202 | ret = PyObject_CallMethodObjArgs(self->buffer, |
| 1203 | _PyIO_str_write, b, NULL); |
| 1204 | Py_DECREF(b); |
| 1205 | if (ret == NULL) |
| 1206 | return -1; |
| 1207 | Py_DECREF(ret); |
| 1208 | Py_CLEAR(self->pending_bytes); |
| 1209 | self->pending_bytes_count = 0; |
| 1210 | return 0; |
| 1211 | } |
| 1212 | |
| 1213 | static PyObject * |
| 1214 | TextIOWrapper_write(PyTextIOWrapperObject *self, PyObject *args) |
| 1215 | { |
| 1216 | PyObject *ret; |
| 1217 | PyObject *text; /* owned reference */ |
| 1218 | PyObject *b; |
| 1219 | Py_ssize_t textlen; |
| 1220 | int haslf = 0; |
| 1221 | int needflush = 0; |
| 1222 | |
| 1223 | CHECK_INITIALIZED(self); |
| 1224 | |
| 1225 | if (!PyArg_ParseTuple(args, "U:write", &text)) { |
| 1226 | return NULL; |
| 1227 | } |
| 1228 | |
| 1229 | CHECK_CLOSED(self); |
| 1230 | |
Benjamin Peterson | 81971ea | 2009-05-14 22:01:31 +0000 | [diff] [blame] | 1231 | if (self->encoder == NULL) { |
| 1232 | PyErr_SetString(PyExc_IOError, "not writable"); |
| 1233 | return NULL; |
| 1234 | } |
| 1235 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1236 | Py_INCREF(text); |
| 1237 | |
| 1238 | textlen = PyUnicode_GetSize(text); |
| 1239 | |
| 1240 | if ((self->writetranslate && self->writenl != NULL) || self->line_buffering) |
| 1241 | if (findchar(PyUnicode_AS_UNICODE(text), |
| 1242 | PyUnicode_GET_SIZE(text), '\n')) |
| 1243 | haslf = 1; |
| 1244 | |
| 1245 | if (haslf && self->writetranslate && self->writenl != NULL) { |
| 1246 | PyObject *newtext = PyObject_CallMethod( |
| 1247 | text, "replace", "ss", "\n", self->writenl); |
| 1248 | Py_DECREF(text); |
| 1249 | if (newtext == NULL) |
| 1250 | return NULL; |
| 1251 | text = newtext; |
| 1252 | } |
| 1253 | |
| 1254 | if (self->line_buffering && |
| 1255 | (haslf || |
| 1256 | findchar(PyUnicode_AS_UNICODE(text), |
| 1257 | PyUnicode_GET_SIZE(text), '\r'))) |
| 1258 | needflush = 1; |
| 1259 | |
| 1260 | /* XXX What if we were just reading? */ |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 1261 | if (self->encodefunc != NULL) { |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1262 | b = (*self->encodefunc)((PyObject *) self, text); |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 1263 | self->encoding_start_of_stream = 0; |
| 1264 | } |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1265 | else |
| 1266 | b = PyObject_CallMethodObjArgs(self->encoder, |
| 1267 | _PyIO_str_encode, text, NULL); |
| 1268 | Py_DECREF(text); |
| 1269 | if (b == NULL) |
| 1270 | return NULL; |
| 1271 | |
| 1272 | if (self->pending_bytes == NULL) { |
| 1273 | self->pending_bytes = PyList_New(0); |
| 1274 | if (self->pending_bytes == NULL) { |
| 1275 | Py_DECREF(b); |
| 1276 | return NULL; |
| 1277 | } |
| 1278 | self->pending_bytes_count = 0; |
| 1279 | } |
| 1280 | if (PyList_Append(self->pending_bytes, b) < 0) { |
| 1281 | Py_DECREF(b); |
| 1282 | return NULL; |
| 1283 | } |
| 1284 | self->pending_bytes_count += PyBytes_GET_SIZE(b); |
| 1285 | Py_DECREF(b); |
| 1286 | if (self->pending_bytes_count > self->chunk_size || needflush) { |
| 1287 | if (_TextIOWrapper_writeflush(self) < 0) |
| 1288 | return NULL; |
| 1289 | } |
Antoine Pitrou | 24f3629 | 2009-03-28 22:16:42 +0000 | [diff] [blame] | 1290 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1291 | if (needflush) { |
| 1292 | ret = PyObject_CallMethodObjArgs(self->buffer, _PyIO_str_flush, NULL); |
| 1293 | if (ret == NULL) |
| 1294 | return NULL; |
| 1295 | Py_DECREF(ret); |
| 1296 | } |
| 1297 | |
| 1298 | Py_CLEAR(self->snapshot); |
| 1299 | |
| 1300 | if (self->decoder) { |
| 1301 | ret = PyObject_CallMethod(self->decoder, "reset", NULL); |
| 1302 | if (ret == NULL) |
| 1303 | return NULL; |
| 1304 | Py_DECREF(ret); |
| 1305 | } |
| 1306 | |
| 1307 | return PyLong_FromSsize_t(textlen); |
| 1308 | } |
| 1309 | |
| 1310 | /* Steal a reference to chars and store it in the decoded_char buffer; |
| 1311 | */ |
| 1312 | static void |
| 1313 | TextIOWrapper_set_decoded_chars(PyTextIOWrapperObject *self, PyObject *chars) |
| 1314 | { |
| 1315 | Py_CLEAR(self->decoded_chars); |
| 1316 | self->decoded_chars = chars; |
| 1317 | self->decoded_chars_used = 0; |
| 1318 | } |
| 1319 | |
| 1320 | static PyObject * |
| 1321 | TextIOWrapper_get_decoded_chars(PyTextIOWrapperObject *self, Py_ssize_t n) |
| 1322 | { |
| 1323 | PyObject *chars; |
| 1324 | Py_ssize_t avail; |
| 1325 | |
| 1326 | if (self->decoded_chars == NULL) |
| 1327 | return PyUnicode_FromStringAndSize(NULL, 0); |
| 1328 | |
| 1329 | avail = (PyUnicode_GET_SIZE(self->decoded_chars) |
| 1330 | - self->decoded_chars_used); |
| 1331 | |
| 1332 | assert(avail >= 0); |
| 1333 | |
| 1334 | if (n < 0 || n > avail) |
| 1335 | n = avail; |
| 1336 | |
| 1337 | if (self->decoded_chars_used > 0 || n < avail) { |
| 1338 | chars = PyUnicode_FromUnicode( |
| 1339 | PyUnicode_AS_UNICODE(self->decoded_chars) |
| 1340 | + self->decoded_chars_used, n); |
| 1341 | if (chars == NULL) |
| 1342 | return NULL; |
| 1343 | } |
| 1344 | else { |
| 1345 | chars = self->decoded_chars; |
| 1346 | Py_INCREF(chars); |
| 1347 | } |
| 1348 | |
| 1349 | self->decoded_chars_used += n; |
| 1350 | return chars; |
| 1351 | } |
| 1352 | |
| 1353 | /* Read and decode the next chunk of data from the BufferedReader. |
| 1354 | */ |
| 1355 | static int |
| 1356 | TextIOWrapper_read_chunk(PyTextIOWrapperObject *self) |
| 1357 | { |
| 1358 | PyObject *dec_buffer = NULL; |
| 1359 | PyObject *dec_flags = NULL; |
| 1360 | PyObject *input_chunk = NULL; |
| 1361 | PyObject *decoded_chars, *chunk_size; |
| 1362 | int eof; |
| 1363 | |
| 1364 | /* The return value is True unless EOF was reached. The decoded string is |
| 1365 | * placed in self._decoded_chars (replacing its previous value). The |
| 1366 | * entire input chunk is sent to the decoder, though some of it may remain |
| 1367 | * buffered in the decoder, yet to be converted. |
| 1368 | */ |
| 1369 | |
| 1370 | if (self->decoder == NULL) { |
Benjamin Peterson | 81971ea | 2009-05-14 22:01:31 +0000 | [diff] [blame] | 1371 | PyErr_SetString(PyExc_IOError, "not readable"); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1372 | return -1; |
| 1373 | } |
| 1374 | |
| 1375 | if (self->telling) { |
| 1376 | /* To prepare for tell(), we need to snapshot a point in the file |
| 1377 | * where the decoder's input buffer is empty. |
| 1378 | */ |
| 1379 | |
| 1380 | PyObject *state = PyObject_CallMethodObjArgs(self->decoder, |
| 1381 | _PyIO_str_getstate, NULL); |
| 1382 | if (state == NULL) |
| 1383 | return -1; |
| 1384 | /* Given this, we know there was a valid snapshot point |
| 1385 | * len(dec_buffer) bytes ago with decoder state (b'', dec_flags). |
| 1386 | */ |
| 1387 | if (PyArg_Parse(state, "(OO)", &dec_buffer, &dec_flags) < 0) { |
| 1388 | Py_DECREF(state); |
| 1389 | return -1; |
| 1390 | } |
| 1391 | Py_INCREF(dec_buffer); |
| 1392 | Py_INCREF(dec_flags); |
| 1393 | Py_DECREF(state); |
| 1394 | } |
| 1395 | |
| 1396 | /* Read a chunk, decode it, and put the result in self._decoded_chars. */ |
| 1397 | chunk_size = PyLong_FromSsize_t(self->chunk_size); |
| 1398 | if (chunk_size == NULL) |
| 1399 | goto fail; |
| 1400 | input_chunk = PyObject_CallMethodObjArgs(self->buffer, |
| 1401 | _PyIO_str_read1, chunk_size, NULL); |
| 1402 | Py_DECREF(chunk_size); |
| 1403 | if (input_chunk == NULL) |
| 1404 | goto fail; |
| 1405 | assert(PyBytes_Check(input_chunk)); |
| 1406 | |
| 1407 | eof = (PyBytes_Size(input_chunk) == 0); |
| 1408 | |
| 1409 | if (Py_TYPE(self->decoder) == &PyIncrementalNewlineDecoder_Type) { |
| 1410 | decoded_chars = _PyIncrementalNewlineDecoder_decode( |
| 1411 | self->decoder, input_chunk, eof); |
| 1412 | } |
| 1413 | else { |
| 1414 | decoded_chars = PyObject_CallMethodObjArgs(self->decoder, |
| 1415 | _PyIO_str_decode, input_chunk, eof ? Py_True : Py_False, NULL); |
| 1416 | } |
| 1417 | |
| 1418 | /* TODO sanity check: isinstance(decoded_chars, unicode) */ |
| 1419 | if (decoded_chars == NULL) |
| 1420 | goto fail; |
| 1421 | TextIOWrapper_set_decoded_chars(self, decoded_chars); |
| 1422 | if (PyUnicode_GET_SIZE(decoded_chars) > 0) |
| 1423 | eof = 0; |
| 1424 | |
| 1425 | if (self->telling) { |
| 1426 | /* At the snapshot point, len(dec_buffer) bytes before the read, the |
| 1427 | * next input to be decoded is dec_buffer + input_chunk. |
| 1428 | */ |
| 1429 | PyObject *next_input = PyNumber_Add(dec_buffer, input_chunk); |
| 1430 | if (next_input == NULL) |
| 1431 | goto fail; |
| 1432 | assert (PyBytes_Check(next_input)); |
| 1433 | Py_DECREF(dec_buffer); |
| 1434 | Py_CLEAR(self->snapshot); |
| 1435 | self->snapshot = Py_BuildValue("NN", dec_flags, next_input); |
| 1436 | } |
| 1437 | Py_DECREF(input_chunk); |
| 1438 | |
| 1439 | return (eof == 0); |
| 1440 | |
| 1441 | fail: |
| 1442 | Py_XDECREF(dec_buffer); |
| 1443 | Py_XDECREF(dec_flags); |
| 1444 | Py_XDECREF(input_chunk); |
| 1445 | return -1; |
| 1446 | } |
| 1447 | |
| 1448 | static PyObject * |
| 1449 | TextIOWrapper_read(PyTextIOWrapperObject *self, PyObject *args) |
| 1450 | { |
| 1451 | Py_ssize_t n = -1; |
| 1452 | PyObject *result = NULL, *chunks = NULL; |
| 1453 | |
| 1454 | CHECK_INITIALIZED(self); |
| 1455 | |
| 1456 | if (!PyArg_ParseTuple(args, "|n:read", &n)) |
| 1457 | return NULL; |
| 1458 | |
| 1459 | CHECK_CLOSED(self); |
| 1460 | |
Benjamin Peterson | a1b4901 | 2009-03-31 23:11:32 +0000 | [diff] [blame] | 1461 | if (self->decoder == NULL) { |
| 1462 | PyErr_SetString(PyExc_IOError, "not readable"); |
| 1463 | return NULL; |
| 1464 | } |
| 1465 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1466 | if (_TextIOWrapper_writeflush(self) < 0) |
| 1467 | return NULL; |
| 1468 | |
| 1469 | if (n < 0) { |
| 1470 | /* Read everything */ |
| 1471 | PyObject *bytes = PyObject_CallMethod(self->buffer, "read", NULL); |
| 1472 | PyObject *decoded; |
| 1473 | if (bytes == NULL) |
| 1474 | goto fail; |
| 1475 | decoded = PyObject_CallMethodObjArgs(self->decoder, _PyIO_str_decode, |
| 1476 | bytes, Py_True, NULL); |
| 1477 | Py_DECREF(bytes); |
| 1478 | if (decoded == NULL) |
| 1479 | goto fail; |
| 1480 | |
| 1481 | result = TextIOWrapper_get_decoded_chars(self, -1); |
| 1482 | |
| 1483 | if (result == NULL) { |
| 1484 | Py_DECREF(decoded); |
| 1485 | return NULL; |
| 1486 | } |
| 1487 | |
| 1488 | PyUnicode_AppendAndDel(&result, decoded); |
| 1489 | if (result == NULL) |
| 1490 | goto fail; |
| 1491 | |
| 1492 | Py_CLEAR(self->snapshot); |
| 1493 | return result; |
| 1494 | } |
| 1495 | else { |
| 1496 | int res = 1; |
| 1497 | Py_ssize_t remaining = n; |
| 1498 | |
| 1499 | result = TextIOWrapper_get_decoded_chars(self, n); |
| 1500 | if (result == NULL) |
| 1501 | goto fail; |
| 1502 | remaining -= PyUnicode_GET_SIZE(result); |
| 1503 | |
| 1504 | /* Keep reading chunks until we have n characters to return */ |
| 1505 | while (remaining > 0) { |
| 1506 | res = TextIOWrapper_read_chunk(self); |
| 1507 | if (res < 0) |
| 1508 | goto fail; |
| 1509 | if (res == 0) /* EOF */ |
| 1510 | break; |
| 1511 | if (chunks == NULL) { |
| 1512 | chunks = PyList_New(0); |
| 1513 | if (chunks == NULL) |
| 1514 | goto fail; |
| 1515 | } |
| 1516 | if (PyList_Append(chunks, result) < 0) |
| 1517 | goto fail; |
| 1518 | Py_DECREF(result); |
| 1519 | result = TextIOWrapper_get_decoded_chars(self, remaining); |
| 1520 | if (result == NULL) |
| 1521 | goto fail; |
| 1522 | remaining -= PyUnicode_GET_SIZE(result); |
| 1523 | } |
| 1524 | if (chunks != NULL) { |
| 1525 | if (result != NULL && PyList_Append(chunks, result) < 0) |
| 1526 | goto fail; |
| 1527 | Py_CLEAR(result); |
| 1528 | result = PyUnicode_Join(_PyIO_empty_str, chunks); |
| 1529 | if (result == NULL) |
| 1530 | goto fail; |
| 1531 | Py_CLEAR(chunks); |
| 1532 | } |
| 1533 | return result; |
| 1534 | } |
| 1535 | fail: |
| 1536 | Py_XDECREF(result); |
| 1537 | Py_XDECREF(chunks); |
| 1538 | return NULL; |
| 1539 | } |
| 1540 | |
| 1541 | |
| 1542 | /* NOTE: `end` must point to the real end of the Py_UNICODE storage, |
| 1543 | that is to the NUL character. Otherwise the function will produce |
| 1544 | incorrect results. */ |
| 1545 | static Py_UNICODE * |
| 1546 | find_control_char(Py_UNICODE *start, Py_UNICODE *end, Py_UNICODE ch) |
| 1547 | { |
| 1548 | Py_UNICODE *s = start; |
| 1549 | for (;;) { |
| 1550 | while (*s > ch) |
| 1551 | s++; |
| 1552 | if (*s == ch) |
| 1553 | return s; |
| 1554 | if (s == end) |
| 1555 | return NULL; |
| 1556 | s++; |
| 1557 | } |
| 1558 | } |
| 1559 | |
| 1560 | Py_ssize_t |
| 1561 | _PyIO_find_line_ending( |
| 1562 | int translated, int universal, PyObject *readnl, |
| 1563 | Py_UNICODE *start, Py_UNICODE *end, Py_ssize_t *consumed) |
| 1564 | { |
| 1565 | Py_ssize_t len = end - start; |
| 1566 | |
| 1567 | if (translated) { |
| 1568 | /* Newlines are already translated, only search for \n */ |
| 1569 | Py_UNICODE *pos = find_control_char(start, end, '\n'); |
| 1570 | if (pos != NULL) |
| 1571 | return pos - start + 1; |
| 1572 | else { |
| 1573 | *consumed = len; |
| 1574 | return -1; |
| 1575 | } |
| 1576 | } |
| 1577 | else if (universal) { |
| 1578 | /* Universal newline search. Find any of \r, \r\n, \n |
| 1579 | * The decoder ensures that \r\n are not split in two pieces |
| 1580 | */ |
| 1581 | Py_UNICODE *s = start; |
| 1582 | for (;;) { |
| 1583 | Py_UNICODE ch; |
| 1584 | /* Fast path for non-control chars. The loop always ends |
| 1585 | since the Py_UNICODE storage is NUL-terminated. */ |
| 1586 | while (*s > '\r') |
| 1587 | s++; |
| 1588 | if (s >= end) { |
| 1589 | *consumed = len; |
| 1590 | return -1; |
| 1591 | } |
| 1592 | ch = *s++; |
| 1593 | if (ch == '\n') |
| 1594 | return s - start; |
| 1595 | if (ch == '\r') { |
| 1596 | if (*s == '\n') |
| 1597 | return s - start + 1; |
| 1598 | else |
| 1599 | return s - start; |
| 1600 | } |
| 1601 | } |
| 1602 | } |
| 1603 | else { |
| 1604 | /* Non-universal mode. */ |
| 1605 | Py_ssize_t readnl_len = PyUnicode_GET_SIZE(readnl); |
| 1606 | Py_UNICODE *nl = PyUnicode_AS_UNICODE(readnl); |
| 1607 | if (readnl_len == 1) { |
| 1608 | Py_UNICODE *pos = find_control_char(start, end, nl[0]); |
| 1609 | if (pos != NULL) |
| 1610 | return pos - start + 1; |
| 1611 | *consumed = len; |
| 1612 | return -1; |
| 1613 | } |
| 1614 | else { |
| 1615 | Py_UNICODE *s = start; |
| 1616 | Py_UNICODE *e = end - readnl_len + 1; |
| 1617 | Py_UNICODE *pos; |
| 1618 | if (e < s) |
| 1619 | e = s; |
| 1620 | while (s < e) { |
| 1621 | Py_ssize_t i; |
| 1622 | Py_UNICODE *pos = find_control_char(s, end, nl[0]); |
| 1623 | if (pos == NULL || pos >= e) |
| 1624 | break; |
| 1625 | for (i = 1; i < readnl_len; i++) { |
| 1626 | if (pos[i] != nl[i]) |
| 1627 | break; |
| 1628 | } |
| 1629 | if (i == readnl_len) |
| 1630 | return pos - start + readnl_len; |
| 1631 | s = pos + 1; |
| 1632 | } |
| 1633 | pos = find_control_char(e, end, nl[0]); |
| 1634 | if (pos == NULL) |
| 1635 | *consumed = len; |
| 1636 | else |
| 1637 | *consumed = pos - start; |
| 1638 | return -1; |
| 1639 | } |
| 1640 | } |
| 1641 | } |
| 1642 | |
| 1643 | static PyObject * |
| 1644 | _TextIOWrapper_readline(PyTextIOWrapperObject *self, Py_ssize_t limit) |
| 1645 | { |
| 1646 | PyObject *line = NULL, *chunks = NULL, *remaining = NULL; |
| 1647 | Py_ssize_t start, endpos, chunked, offset_to_buffer; |
| 1648 | int res; |
| 1649 | |
| 1650 | CHECK_CLOSED(self); |
| 1651 | |
| 1652 | if (_TextIOWrapper_writeflush(self) < 0) |
| 1653 | return NULL; |
| 1654 | |
| 1655 | chunked = 0; |
| 1656 | |
| 1657 | while (1) { |
| 1658 | Py_UNICODE *ptr; |
| 1659 | Py_ssize_t line_len; |
| 1660 | Py_ssize_t consumed = 0; |
| 1661 | |
| 1662 | /* First, get some data if necessary */ |
| 1663 | res = 1; |
| 1664 | while (!self->decoded_chars || |
| 1665 | !PyUnicode_GET_SIZE(self->decoded_chars)) { |
| 1666 | res = TextIOWrapper_read_chunk(self); |
| 1667 | if (res < 0) |
| 1668 | goto error; |
| 1669 | if (res == 0) |
| 1670 | break; |
| 1671 | } |
| 1672 | if (res == 0) { |
| 1673 | /* end of file */ |
| 1674 | TextIOWrapper_set_decoded_chars(self, NULL); |
| 1675 | Py_CLEAR(self->snapshot); |
| 1676 | start = endpos = offset_to_buffer = 0; |
| 1677 | break; |
| 1678 | } |
| 1679 | |
| 1680 | if (remaining == NULL) { |
| 1681 | line = self->decoded_chars; |
| 1682 | start = self->decoded_chars_used; |
| 1683 | offset_to_buffer = 0; |
| 1684 | Py_INCREF(line); |
| 1685 | } |
| 1686 | else { |
| 1687 | assert(self->decoded_chars_used == 0); |
| 1688 | line = PyUnicode_Concat(remaining, self->decoded_chars); |
| 1689 | start = 0; |
| 1690 | offset_to_buffer = PyUnicode_GET_SIZE(remaining); |
| 1691 | Py_CLEAR(remaining); |
| 1692 | if (line == NULL) |
| 1693 | goto error; |
| 1694 | } |
| 1695 | |
| 1696 | ptr = PyUnicode_AS_UNICODE(line); |
| 1697 | line_len = PyUnicode_GET_SIZE(line); |
| 1698 | |
| 1699 | endpos = _PyIO_find_line_ending( |
| 1700 | self->readtranslate, self->readuniversal, self->readnl, |
| 1701 | ptr + start, ptr + line_len, &consumed); |
| 1702 | if (endpos >= 0) { |
| 1703 | endpos += start; |
| 1704 | if (limit >= 0 && (endpos - start) + chunked >= limit) |
| 1705 | endpos = start + limit - chunked; |
| 1706 | break; |
| 1707 | } |
| 1708 | |
| 1709 | /* We can put aside up to `endpos` */ |
| 1710 | endpos = consumed + start; |
| 1711 | if (limit >= 0 && (endpos - start) + chunked >= limit) { |
| 1712 | /* Didn't find line ending, but reached length limit */ |
| 1713 | endpos = start + limit - chunked; |
| 1714 | break; |
| 1715 | } |
| 1716 | |
| 1717 | if (endpos > start) { |
| 1718 | /* No line ending seen yet - put aside current data */ |
| 1719 | PyObject *s; |
| 1720 | if (chunks == NULL) { |
| 1721 | chunks = PyList_New(0); |
| 1722 | if (chunks == NULL) |
| 1723 | goto error; |
| 1724 | } |
| 1725 | s = PyUnicode_FromUnicode(ptr + start, endpos - start); |
| 1726 | if (s == NULL) |
| 1727 | goto error; |
| 1728 | if (PyList_Append(chunks, s) < 0) { |
| 1729 | Py_DECREF(s); |
| 1730 | goto error; |
| 1731 | } |
| 1732 | chunked += PyUnicode_GET_SIZE(s); |
| 1733 | Py_DECREF(s); |
| 1734 | } |
| 1735 | /* There may be some remaining bytes we'll have to prepend to the |
| 1736 | next chunk of data */ |
| 1737 | if (endpos < line_len) { |
| 1738 | remaining = PyUnicode_FromUnicode( |
| 1739 | ptr + endpos, line_len - endpos); |
| 1740 | if (remaining == NULL) |
| 1741 | goto error; |
| 1742 | } |
| 1743 | Py_CLEAR(line); |
| 1744 | /* We have consumed the buffer */ |
| 1745 | TextIOWrapper_set_decoded_chars(self, NULL); |
| 1746 | } |
| 1747 | |
| 1748 | if (line != NULL) { |
| 1749 | /* Our line ends in the current buffer */ |
| 1750 | self->decoded_chars_used = endpos - offset_to_buffer; |
| 1751 | if (start > 0 || endpos < PyUnicode_GET_SIZE(line)) { |
| 1752 | if (start == 0 && Py_REFCNT(line) == 1) { |
| 1753 | if (PyUnicode_Resize(&line, endpos) < 0) |
| 1754 | goto error; |
| 1755 | } |
| 1756 | else { |
| 1757 | PyObject *s = PyUnicode_FromUnicode( |
| 1758 | PyUnicode_AS_UNICODE(line) + start, endpos - start); |
| 1759 | Py_CLEAR(line); |
| 1760 | if (s == NULL) |
| 1761 | goto error; |
| 1762 | line = s; |
| 1763 | } |
| 1764 | } |
| 1765 | } |
| 1766 | if (remaining != NULL) { |
| 1767 | if (chunks == NULL) { |
| 1768 | chunks = PyList_New(0); |
| 1769 | if (chunks == NULL) |
| 1770 | goto error; |
| 1771 | } |
| 1772 | if (PyList_Append(chunks, remaining) < 0) |
| 1773 | goto error; |
| 1774 | Py_CLEAR(remaining); |
| 1775 | } |
| 1776 | if (chunks != NULL) { |
| 1777 | if (line != NULL && PyList_Append(chunks, line) < 0) |
| 1778 | goto error; |
| 1779 | Py_CLEAR(line); |
| 1780 | line = PyUnicode_Join(_PyIO_empty_str, chunks); |
| 1781 | if (line == NULL) |
| 1782 | goto error; |
| 1783 | Py_DECREF(chunks); |
| 1784 | } |
| 1785 | if (line == NULL) |
| 1786 | line = PyUnicode_FromStringAndSize(NULL, 0); |
| 1787 | |
| 1788 | return line; |
| 1789 | |
| 1790 | error: |
| 1791 | Py_XDECREF(chunks); |
| 1792 | Py_XDECREF(remaining); |
| 1793 | Py_XDECREF(line); |
| 1794 | return NULL; |
| 1795 | } |
| 1796 | |
| 1797 | static PyObject * |
| 1798 | TextIOWrapper_readline(PyTextIOWrapperObject *self, PyObject *args) |
| 1799 | { |
| 1800 | Py_ssize_t limit = -1; |
| 1801 | |
| 1802 | CHECK_INITIALIZED(self); |
| 1803 | if (!PyArg_ParseTuple(args, "|n:readline", &limit)) { |
| 1804 | return NULL; |
| 1805 | } |
| 1806 | return _TextIOWrapper_readline(self, limit); |
| 1807 | } |
| 1808 | |
| 1809 | /* Seek and Tell */ |
| 1810 | |
| 1811 | typedef struct { |
| 1812 | Py_off_t start_pos; |
| 1813 | int dec_flags; |
| 1814 | int bytes_to_feed; |
| 1815 | int chars_to_skip; |
| 1816 | char need_eof; |
| 1817 | } CookieStruct; |
| 1818 | |
| 1819 | /* |
| 1820 | To speed up cookie packing/unpacking, we store the fields in a temporary |
| 1821 | string and call _PyLong_FromByteArray() or _PyLong_AsByteArray (resp.). |
| 1822 | The following macros define at which offsets in the intermediary byte |
| 1823 | string the various CookieStruct fields will be stored. |
| 1824 | */ |
| 1825 | |
| 1826 | #define COOKIE_BUF_LEN (sizeof(Py_off_t) + 3 * sizeof(int) + sizeof(char)) |
| 1827 | |
| 1828 | #if defined(WORDS_BIGENDIAN) |
| 1829 | |
| 1830 | # define IS_LITTLE_ENDIAN 0 |
| 1831 | |
| 1832 | /* We want the least significant byte of start_pos to also be the least |
| 1833 | significant byte of the cookie, which means that in big-endian mode we |
| 1834 | must copy the fields in reverse order. */ |
| 1835 | |
| 1836 | # define OFF_START_POS (sizeof(char) + 3 * sizeof(int)) |
| 1837 | # define OFF_DEC_FLAGS (sizeof(char) + 2 * sizeof(int)) |
| 1838 | # define OFF_BYTES_TO_FEED (sizeof(char) + sizeof(int)) |
| 1839 | # define OFF_CHARS_TO_SKIP (sizeof(char)) |
| 1840 | # define OFF_NEED_EOF 0 |
| 1841 | |
| 1842 | #else |
| 1843 | |
| 1844 | # define IS_LITTLE_ENDIAN 1 |
| 1845 | |
| 1846 | /* Little-endian mode: the least significant byte of start_pos will |
| 1847 | naturally end up the least significant byte of the cookie. */ |
| 1848 | |
| 1849 | # define OFF_START_POS 0 |
| 1850 | # define OFF_DEC_FLAGS (sizeof(Py_off_t)) |
| 1851 | # define OFF_BYTES_TO_FEED (sizeof(Py_off_t) + sizeof(int)) |
| 1852 | # define OFF_CHARS_TO_SKIP (sizeof(Py_off_t) + 2 * sizeof(int)) |
| 1853 | # define OFF_NEED_EOF (sizeof(Py_off_t) + 3 * sizeof(int)) |
| 1854 | |
| 1855 | #endif |
| 1856 | |
| 1857 | static int |
| 1858 | TextIOWrapper_parseCookie(CookieStruct *cookie, PyObject *cookieObj) |
| 1859 | { |
| 1860 | unsigned char buffer[COOKIE_BUF_LEN]; |
| 1861 | PyLongObject *cookieLong = (PyLongObject *)PyNumber_Long(cookieObj); |
| 1862 | if (cookieLong == NULL) |
| 1863 | return -1; |
| 1864 | |
| 1865 | if (_PyLong_AsByteArray(cookieLong, buffer, sizeof(buffer), |
| 1866 | IS_LITTLE_ENDIAN, 0) < 0) { |
| 1867 | Py_DECREF(cookieLong); |
| 1868 | return -1; |
| 1869 | } |
| 1870 | Py_DECREF(cookieLong); |
| 1871 | |
Antoine Pitrou | 2db74c2 | 2009-03-06 21:49:02 +0000 | [diff] [blame] | 1872 | memcpy(&cookie->start_pos, buffer + OFF_START_POS, sizeof(cookie->start_pos)); |
| 1873 | memcpy(&cookie->dec_flags, buffer + OFF_DEC_FLAGS, sizeof(cookie->dec_flags)); |
| 1874 | memcpy(&cookie->bytes_to_feed, buffer + OFF_BYTES_TO_FEED, sizeof(cookie->bytes_to_feed)); |
| 1875 | memcpy(&cookie->chars_to_skip, buffer + OFF_CHARS_TO_SKIP, sizeof(cookie->chars_to_skip)); |
| 1876 | memcpy(&cookie->need_eof, buffer + OFF_NEED_EOF, sizeof(cookie->need_eof)); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1877 | |
| 1878 | return 0; |
| 1879 | } |
| 1880 | |
| 1881 | static PyObject * |
| 1882 | TextIOWrapper_buildCookie(CookieStruct *cookie) |
| 1883 | { |
| 1884 | unsigned char buffer[COOKIE_BUF_LEN]; |
| 1885 | |
Antoine Pitrou | 2db74c2 | 2009-03-06 21:49:02 +0000 | [diff] [blame] | 1886 | memcpy(buffer + OFF_START_POS, &cookie->start_pos, sizeof(cookie->start_pos)); |
| 1887 | memcpy(buffer + OFF_DEC_FLAGS, &cookie->dec_flags, sizeof(cookie->dec_flags)); |
| 1888 | memcpy(buffer + OFF_BYTES_TO_FEED, &cookie->bytes_to_feed, sizeof(cookie->bytes_to_feed)); |
| 1889 | memcpy(buffer + OFF_CHARS_TO_SKIP, &cookie->chars_to_skip, sizeof(cookie->chars_to_skip)); |
| 1890 | memcpy(buffer + OFF_NEED_EOF, &cookie->need_eof, sizeof(cookie->need_eof)); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1891 | |
| 1892 | return _PyLong_FromByteArray(buffer, sizeof(buffer), IS_LITTLE_ENDIAN, 0); |
| 1893 | } |
| 1894 | #undef IS_LITTLE_ENDIAN |
| 1895 | |
| 1896 | static int |
| 1897 | _TextIOWrapper_decoder_setstate(PyTextIOWrapperObject *self, |
| 1898 | CookieStruct *cookie) |
| 1899 | { |
| 1900 | PyObject *res; |
| 1901 | /* When seeking to the start of the stream, we call decoder.reset() |
| 1902 | rather than decoder.getstate(). |
| 1903 | This is for a few decoders such as utf-16 for which the state value |
| 1904 | at start is not (b"", 0) but e.g. (b"", 2) (meaning, in the case of |
| 1905 | utf-16, that we are expecting a BOM). |
| 1906 | */ |
| 1907 | if (cookie->start_pos == 0 && cookie->dec_flags == 0) |
| 1908 | res = PyObject_CallMethodObjArgs(self->decoder, _PyIO_str_reset, NULL); |
| 1909 | else |
| 1910 | res = PyObject_CallMethod(self->decoder, "setstate", |
| 1911 | "((yi))", "", cookie->dec_flags); |
| 1912 | if (res == NULL) |
| 1913 | return -1; |
| 1914 | Py_DECREF(res); |
| 1915 | return 0; |
| 1916 | } |
| 1917 | |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 1918 | static int |
| 1919 | _TextIOWrapper_encoder_setstate(PyTextIOWrapperObject *self, |
| 1920 | CookieStruct *cookie) |
| 1921 | { |
| 1922 | PyObject *res; |
| 1923 | /* Same as _TextIOWrapper_decoder_setstate() above. */ |
| 1924 | if (cookie->start_pos == 0 && cookie->dec_flags == 0) { |
| 1925 | res = PyObject_CallMethodObjArgs(self->encoder, _PyIO_str_reset, NULL); |
| 1926 | self->encoding_start_of_stream = 1; |
| 1927 | } |
| 1928 | else { |
| 1929 | res = PyObject_CallMethodObjArgs(self->encoder, _PyIO_str_setstate, |
| 1930 | _PyIO_zero, NULL); |
| 1931 | self->encoding_start_of_stream = 0; |
| 1932 | } |
| 1933 | if (res == NULL) |
| 1934 | return -1; |
| 1935 | Py_DECREF(res); |
| 1936 | return 0; |
| 1937 | } |
| 1938 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1939 | static PyObject * |
| 1940 | TextIOWrapper_seek(PyTextIOWrapperObject *self, PyObject *args) |
| 1941 | { |
| 1942 | PyObject *cookieObj, *posobj; |
| 1943 | CookieStruct cookie; |
| 1944 | int whence = 0; |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1945 | PyObject *res; |
| 1946 | int cmp; |
| 1947 | |
| 1948 | CHECK_INITIALIZED(self); |
| 1949 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1950 | if (!PyArg_ParseTuple(args, "O|i:seek", &cookieObj, &whence)) |
| 1951 | return NULL; |
| 1952 | CHECK_CLOSED(self); |
| 1953 | |
| 1954 | Py_INCREF(cookieObj); |
| 1955 | |
| 1956 | if (!self->seekable) { |
| 1957 | PyErr_SetString(PyExc_IOError, |
| 1958 | "underlying stream is not seekable"); |
| 1959 | goto fail; |
| 1960 | } |
| 1961 | |
| 1962 | if (whence == 1) { |
| 1963 | /* seek relative to current position */ |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 1964 | cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_EQ); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1965 | if (cmp < 0) |
| 1966 | goto fail; |
| 1967 | |
| 1968 | if (cmp == 0) { |
| 1969 | PyErr_SetString(PyExc_IOError, |
| 1970 | "can't do nonzero cur-relative seeks"); |
| 1971 | goto fail; |
| 1972 | } |
| 1973 | |
| 1974 | /* Seeking to the current position should attempt to |
| 1975 | * sync the underlying buffer with the current position. |
| 1976 | */ |
| 1977 | Py_DECREF(cookieObj); |
| 1978 | cookieObj = PyObject_CallMethod((PyObject *)self, "tell", NULL); |
| 1979 | if (cookieObj == NULL) |
| 1980 | goto fail; |
| 1981 | } |
| 1982 | else if (whence == 2) { |
| 1983 | /* seek relative to end of file */ |
| 1984 | |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 1985 | cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_EQ); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 1986 | if (cmp < 0) |
| 1987 | goto fail; |
| 1988 | |
| 1989 | if (cmp == 0) { |
| 1990 | PyErr_SetString(PyExc_IOError, |
| 1991 | "can't do nonzero end-relative seeks"); |
| 1992 | goto fail; |
| 1993 | } |
| 1994 | |
| 1995 | res = PyObject_CallMethod((PyObject *)self, "flush", NULL); |
| 1996 | if (res == NULL) |
| 1997 | goto fail; |
| 1998 | Py_DECREF(res); |
| 1999 | |
| 2000 | TextIOWrapper_set_decoded_chars(self, NULL); |
| 2001 | Py_CLEAR(self->snapshot); |
| 2002 | if (self->decoder) { |
| 2003 | res = PyObject_CallMethod(self->decoder, "reset", NULL); |
| 2004 | if (res == NULL) |
| 2005 | goto fail; |
| 2006 | Py_DECREF(res); |
| 2007 | } |
| 2008 | |
| 2009 | res = PyObject_CallMethod(self->buffer, "seek", "ii", 0, 2); |
| 2010 | Py_XDECREF(cookieObj); |
| 2011 | return res; |
| 2012 | } |
| 2013 | else if (whence != 0) { |
| 2014 | PyErr_Format(PyExc_ValueError, |
| 2015 | "invalid whence (%d, should be 0, 1 or 2)", whence); |
| 2016 | goto fail; |
| 2017 | } |
| 2018 | |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 2019 | cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_LT); |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2020 | if (cmp < 0) |
| 2021 | goto fail; |
| 2022 | |
| 2023 | if (cmp == 1) { |
| 2024 | PyErr_Format(PyExc_ValueError, |
| 2025 | "negative seek position %R", cookieObj); |
| 2026 | goto fail; |
| 2027 | } |
| 2028 | |
| 2029 | res = PyObject_CallMethodObjArgs((PyObject *)self, _PyIO_str_flush, NULL); |
| 2030 | if (res == NULL) |
| 2031 | goto fail; |
| 2032 | Py_DECREF(res); |
| 2033 | |
| 2034 | /* The strategy of seek() is to go back to the safe start point |
| 2035 | * and replay the effect of read(chars_to_skip) from there. |
| 2036 | */ |
| 2037 | if (TextIOWrapper_parseCookie(&cookie, cookieObj) < 0) |
| 2038 | goto fail; |
| 2039 | |
| 2040 | /* Seek back to the safe start point. */ |
| 2041 | posobj = PyLong_FromOff_t(cookie.start_pos); |
| 2042 | if (posobj == NULL) |
| 2043 | goto fail; |
| 2044 | res = PyObject_CallMethodObjArgs(self->buffer, |
| 2045 | _PyIO_str_seek, posobj, NULL); |
| 2046 | Py_DECREF(posobj); |
| 2047 | if (res == NULL) |
| 2048 | goto fail; |
| 2049 | Py_DECREF(res); |
| 2050 | |
| 2051 | TextIOWrapper_set_decoded_chars(self, NULL); |
| 2052 | Py_CLEAR(self->snapshot); |
| 2053 | |
| 2054 | /* Restore the decoder to its state from the safe start point. */ |
| 2055 | if (self->decoder) { |
| 2056 | if (_TextIOWrapper_decoder_setstate(self, &cookie) < 0) |
| 2057 | goto fail; |
| 2058 | } |
| 2059 | |
| 2060 | if (cookie.chars_to_skip) { |
| 2061 | /* Just like _read_chunk, feed the decoder and save a snapshot. */ |
| 2062 | PyObject *input_chunk = PyObject_CallMethod( |
| 2063 | self->buffer, "read", "i", cookie.bytes_to_feed); |
| 2064 | PyObject *decoded; |
| 2065 | |
| 2066 | if (input_chunk == NULL) |
| 2067 | goto fail; |
| 2068 | |
| 2069 | assert (PyBytes_Check(input_chunk)); |
| 2070 | |
| 2071 | self->snapshot = Py_BuildValue("iN", cookie.dec_flags, input_chunk); |
| 2072 | if (self->snapshot == NULL) { |
| 2073 | Py_DECREF(input_chunk); |
| 2074 | goto fail; |
| 2075 | } |
| 2076 | |
| 2077 | decoded = PyObject_CallMethod(self->decoder, "decode", |
| 2078 | "Oi", input_chunk, (int)cookie.need_eof); |
| 2079 | |
| 2080 | if (decoded == NULL) |
| 2081 | goto fail; |
| 2082 | |
| 2083 | TextIOWrapper_set_decoded_chars(self, decoded); |
| 2084 | |
| 2085 | /* Skip chars_to_skip of the decoded characters. */ |
| 2086 | if (PyUnicode_GetSize(self->decoded_chars) < cookie.chars_to_skip) { |
| 2087 | PyErr_SetString(PyExc_IOError, "can't restore logical file position"); |
| 2088 | goto fail; |
| 2089 | } |
| 2090 | self->decoded_chars_used = cookie.chars_to_skip; |
| 2091 | } |
| 2092 | else { |
| 2093 | self->snapshot = Py_BuildValue("iy", cookie.dec_flags, ""); |
| 2094 | if (self->snapshot == NULL) |
| 2095 | goto fail; |
| 2096 | } |
| 2097 | |
Antoine Pitrou | e450185 | 2009-05-14 18:55:55 +0000 | [diff] [blame] | 2098 | /* Finally, reset the encoder (merely useful for proper BOM handling) */ |
| 2099 | if (self->encoder) { |
| 2100 | if (_TextIOWrapper_encoder_setstate(self, &cookie) < 0) |
| 2101 | goto fail; |
| 2102 | } |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2103 | return cookieObj; |
| 2104 | fail: |
| 2105 | Py_XDECREF(cookieObj); |
| 2106 | return NULL; |
| 2107 | |
| 2108 | } |
| 2109 | |
| 2110 | static PyObject * |
| 2111 | TextIOWrapper_tell(PyTextIOWrapperObject *self, PyObject *args) |
| 2112 | { |
| 2113 | PyObject *res; |
| 2114 | PyObject *posobj = NULL; |
| 2115 | CookieStruct cookie = {0,0,0,0,0}; |
| 2116 | PyObject *next_input; |
| 2117 | Py_ssize_t chars_to_skip, chars_decoded; |
| 2118 | PyObject *saved_state = NULL; |
| 2119 | char *input, *input_end; |
| 2120 | |
| 2121 | CHECK_INITIALIZED(self); |
| 2122 | CHECK_CLOSED(self); |
| 2123 | |
| 2124 | if (!self->seekable) { |
| 2125 | PyErr_SetString(PyExc_IOError, |
| 2126 | "underlying stream is not seekable"); |
| 2127 | goto fail; |
| 2128 | } |
| 2129 | if (!self->telling) { |
| 2130 | PyErr_SetString(PyExc_IOError, |
| 2131 | "telling position disabled by next() call"); |
| 2132 | goto fail; |
| 2133 | } |
| 2134 | |
| 2135 | if (_TextIOWrapper_writeflush(self) < 0) |
| 2136 | return NULL; |
| 2137 | res = PyObject_CallMethod((PyObject *)self, "flush", NULL); |
| 2138 | if (res == NULL) |
| 2139 | goto fail; |
| 2140 | Py_DECREF(res); |
| 2141 | |
| 2142 | posobj = PyObject_CallMethod(self->buffer, "tell", NULL); |
| 2143 | if (posobj == NULL) |
| 2144 | goto fail; |
| 2145 | |
| 2146 | if (self->decoder == NULL || self->snapshot == NULL) { |
| 2147 | assert (self->decoded_chars == NULL || PyUnicode_GetSize(self->decoded_chars) == 0); |
| 2148 | return posobj; |
| 2149 | } |
| 2150 | |
| 2151 | #if defined(HAVE_LARGEFILE_SUPPORT) |
| 2152 | cookie.start_pos = PyLong_AsLongLong(posobj); |
| 2153 | #else |
| 2154 | cookie.start_pos = PyLong_AsLong(posobj); |
| 2155 | #endif |
| 2156 | if (PyErr_Occurred()) |
| 2157 | goto fail; |
| 2158 | |
| 2159 | /* Skip backward to the snapshot point (see _read_chunk). */ |
| 2160 | if (!PyArg_Parse(self->snapshot, "(iO)", &cookie.dec_flags, &next_input)) |
| 2161 | goto fail; |
| 2162 | |
| 2163 | assert (PyBytes_Check(next_input)); |
| 2164 | |
| 2165 | cookie.start_pos -= PyBytes_GET_SIZE(next_input); |
| 2166 | |
| 2167 | /* How many decoded characters have been used up since the snapshot? */ |
| 2168 | if (self->decoded_chars_used == 0) { |
| 2169 | /* We haven't moved from the snapshot point. */ |
| 2170 | Py_DECREF(posobj); |
| 2171 | return TextIOWrapper_buildCookie(&cookie); |
| 2172 | } |
| 2173 | |
| 2174 | chars_to_skip = self->decoded_chars_used; |
| 2175 | |
| 2176 | /* Starting from the snapshot position, we will walk the decoder |
| 2177 | * forward until it gives us enough decoded characters. |
| 2178 | */ |
| 2179 | saved_state = PyObject_CallMethodObjArgs(self->decoder, |
| 2180 | _PyIO_str_getstate, NULL); |
| 2181 | if (saved_state == NULL) |
| 2182 | goto fail; |
| 2183 | |
| 2184 | /* Note our initial start point. */ |
| 2185 | if (_TextIOWrapper_decoder_setstate(self, &cookie) < 0) |
| 2186 | goto fail; |
| 2187 | |
| 2188 | /* Feed the decoder one byte at a time. As we go, note the |
| 2189 | * nearest "safe start point" before the current location |
| 2190 | * (a point where the decoder has nothing buffered, so seek() |
| 2191 | * can safely start from there and advance to this location). |
| 2192 | */ |
| 2193 | chars_decoded = 0; |
| 2194 | input = PyBytes_AS_STRING(next_input); |
| 2195 | input_end = input + PyBytes_GET_SIZE(next_input); |
| 2196 | while (input < input_end) { |
| 2197 | PyObject *state; |
| 2198 | char *dec_buffer; |
| 2199 | Py_ssize_t dec_buffer_len; |
| 2200 | int dec_flags; |
| 2201 | |
| 2202 | PyObject *decoded = PyObject_CallMethod( |
| 2203 | self->decoder, "decode", "y#", input, 1); |
| 2204 | if (decoded == NULL) |
| 2205 | goto fail; |
| 2206 | assert (PyUnicode_Check(decoded)); |
| 2207 | chars_decoded += PyUnicode_GET_SIZE(decoded); |
| 2208 | Py_DECREF(decoded); |
| 2209 | |
| 2210 | cookie.bytes_to_feed += 1; |
| 2211 | |
| 2212 | state = PyObject_CallMethodObjArgs(self->decoder, |
| 2213 | _PyIO_str_getstate, NULL); |
| 2214 | if (state == NULL) |
| 2215 | goto fail; |
| 2216 | if (!PyArg_Parse(state, "(y#i)", &dec_buffer, &dec_buffer_len, &dec_flags)) { |
| 2217 | Py_DECREF(state); |
| 2218 | goto fail; |
| 2219 | } |
| 2220 | Py_DECREF(state); |
| 2221 | |
| 2222 | if (dec_buffer_len == 0 && chars_decoded <= chars_to_skip) { |
| 2223 | /* Decoder buffer is empty, so this is a safe start point. */ |
| 2224 | cookie.start_pos += cookie.bytes_to_feed; |
| 2225 | chars_to_skip -= chars_decoded; |
| 2226 | cookie.dec_flags = dec_flags; |
| 2227 | cookie.bytes_to_feed = 0; |
| 2228 | chars_decoded = 0; |
| 2229 | } |
| 2230 | if (chars_decoded >= chars_to_skip) |
| 2231 | break; |
| 2232 | input++; |
| 2233 | } |
| 2234 | if (input == input_end) { |
| 2235 | /* We didn't get enough decoded data; signal EOF to get more. */ |
| 2236 | PyObject *decoded = PyObject_CallMethod( |
| 2237 | self->decoder, "decode", "yi", "", /* final = */ 1); |
| 2238 | if (decoded == NULL) |
| 2239 | goto fail; |
| 2240 | assert (PyUnicode_Check(decoded)); |
| 2241 | chars_decoded += PyUnicode_GET_SIZE(decoded); |
| 2242 | Py_DECREF(decoded); |
| 2243 | cookie.need_eof = 1; |
| 2244 | |
| 2245 | if (chars_decoded < chars_to_skip) { |
| 2246 | PyErr_SetString(PyExc_IOError, |
| 2247 | "can't reconstruct logical file position"); |
| 2248 | goto fail; |
| 2249 | } |
| 2250 | } |
| 2251 | |
| 2252 | /* finally */ |
| 2253 | Py_XDECREF(posobj); |
| 2254 | res = PyObject_CallMethod(self->decoder, "setstate", "(O)", saved_state); |
| 2255 | Py_DECREF(saved_state); |
| 2256 | if (res == NULL) |
| 2257 | return NULL; |
| 2258 | Py_DECREF(res); |
| 2259 | |
| 2260 | /* The returned cookie corresponds to the last safe start point. */ |
| 2261 | cookie.chars_to_skip = Py_SAFE_DOWNCAST(chars_to_skip, Py_ssize_t, int); |
| 2262 | return TextIOWrapper_buildCookie(&cookie); |
| 2263 | |
| 2264 | fail: |
| 2265 | Py_XDECREF(posobj); |
| 2266 | if (saved_state) { |
| 2267 | PyObject *type, *value, *traceback; |
| 2268 | PyErr_Fetch(&type, &value, &traceback); |
| 2269 | |
| 2270 | res = PyObject_CallMethod(self->decoder, "setstate", "(O)", saved_state); |
| 2271 | Py_DECREF(saved_state); |
| 2272 | if (res == NULL) |
| 2273 | return NULL; |
| 2274 | Py_DECREF(res); |
| 2275 | |
| 2276 | PyErr_Restore(type, value, traceback); |
| 2277 | } |
| 2278 | return NULL; |
| 2279 | } |
| 2280 | |
| 2281 | static PyObject * |
| 2282 | TextIOWrapper_truncate(PyTextIOWrapperObject *self, PyObject *args) |
| 2283 | { |
| 2284 | PyObject *pos = Py_None; |
| 2285 | PyObject *res; |
| 2286 | |
| 2287 | CHECK_INITIALIZED(self) |
| 2288 | if (!PyArg_ParseTuple(args, "|O:truncate", &pos)) { |
| 2289 | return NULL; |
| 2290 | } |
| 2291 | |
| 2292 | res = PyObject_CallMethodObjArgs((PyObject *) self, _PyIO_str_flush, NULL); |
| 2293 | if (res == NULL) |
| 2294 | return NULL; |
| 2295 | Py_DECREF(res); |
| 2296 | |
| 2297 | if (pos != Py_None) { |
| 2298 | res = PyObject_CallMethodObjArgs((PyObject *) self, |
| 2299 | _PyIO_str_seek, pos, NULL); |
| 2300 | if (res == NULL) |
| 2301 | return NULL; |
| 2302 | Py_DECREF(res); |
| 2303 | } |
| 2304 | |
| 2305 | return PyObject_CallMethodObjArgs(self->buffer, _PyIO_str_truncate, NULL); |
| 2306 | } |
| 2307 | |
Benjamin Peterson | c4c0eae | 2009-03-09 00:07:03 +0000 | [diff] [blame] | 2308 | static PyObject * |
| 2309 | TextIOWrapper_repr(PyTextIOWrapperObject *self) |
| 2310 | { |
| 2311 | CHECK_INITIALIZED(self); |
| 2312 | return PyUnicode_FromFormat("<TextIOWrapper encoding=%S>", self->encoding); |
| 2313 | } |
| 2314 | |
| 2315 | |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2316 | /* Inquiries */ |
| 2317 | |
| 2318 | static PyObject * |
| 2319 | TextIOWrapper_fileno(PyTextIOWrapperObject *self, PyObject *args) |
| 2320 | { |
| 2321 | CHECK_INITIALIZED(self); |
| 2322 | return PyObject_CallMethod(self->buffer, "fileno", NULL); |
| 2323 | } |
| 2324 | |
| 2325 | static PyObject * |
| 2326 | TextIOWrapper_seekable(PyTextIOWrapperObject *self, PyObject *args) |
| 2327 | { |
| 2328 | CHECK_INITIALIZED(self); |
| 2329 | return PyObject_CallMethod(self->buffer, "seekable", NULL); |
| 2330 | } |
| 2331 | |
| 2332 | static PyObject * |
| 2333 | TextIOWrapper_readable(PyTextIOWrapperObject *self, PyObject *args) |
| 2334 | { |
| 2335 | CHECK_INITIALIZED(self); |
| 2336 | return PyObject_CallMethod(self->buffer, "readable", NULL); |
| 2337 | } |
| 2338 | |
| 2339 | static PyObject * |
| 2340 | TextIOWrapper_writable(PyTextIOWrapperObject *self, PyObject *args) |
| 2341 | { |
| 2342 | CHECK_INITIALIZED(self); |
| 2343 | return PyObject_CallMethod(self->buffer, "writable", NULL); |
| 2344 | } |
| 2345 | |
| 2346 | static PyObject * |
| 2347 | TextIOWrapper_isatty(PyTextIOWrapperObject *self, PyObject *args) |
| 2348 | { |
| 2349 | CHECK_INITIALIZED(self); |
| 2350 | return PyObject_CallMethod(self->buffer, "isatty", NULL); |
| 2351 | } |
| 2352 | |
| 2353 | static PyObject * |
| 2354 | TextIOWrapper_flush(PyTextIOWrapperObject *self, PyObject *args) |
| 2355 | { |
| 2356 | CHECK_INITIALIZED(self); |
| 2357 | CHECK_CLOSED(self); |
| 2358 | self->telling = self->seekable; |
| 2359 | if (_TextIOWrapper_writeflush(self) < 0) |
| 2360 | return NULL; |
| 2361 | return PyObject_CallMethod(self->buffer, "flush", NULL); |
| 2362 | } |
| 2363 | |
| 2364 | static PyObject * |
| 2365 | TextIOWrapper_close(PyTextIOWrapperObject *self, PyObject *args) |
| 2366 | { |
| 2367 | PyObject *res; |
| 2368 | CHECK_INITIALIZED(self); |
| 2369 | res = PyObject_CallMethod((PyObject *)self, "flush", NULL); |
| 2370 | if (res == NULL) { |
| 2371 | /* If flush() fails, just give up */ |
| 2372 | PyErr_Clear(); |
| 2373 | } |
| 2374 | else |
| 2375 | Py_DECREF(res); |
| 2376 | |
| 2377 | return PyObject_CallMethod(self->buffer, "close", NULL); |
| 2378 | } |
| 2379 | |
| 2380 | static PyObject * |
| 2381 | TextIOWrapper_iternext(PyTextIOWrapperObject *self) |
| 2382 | { |
| 2383 | PyObject *line; |
| 2384 | |
| 2385 | CHECK_INITIALIZED(self); |
| 2386 | |
| 2387 | self->telling = 0; |
| 2388 | if (Py_TYPE(self) == &PyTextIOWrapper_Type) { |
| 2389 | /* Skip method call overhead for speed */ |
| 2390 | line = _TextIOWrapper_readline(self, -1); |
| 2391 | } |
| 2392 | else { |
| 2393 | line = PyObject_CallMethodObjArgs((PyObject *)self, |
| 2394 | _PyIO_str_readline, NULL); |
| 2395 | if (line && !PyUnicode_Check(line)) { |
| 2396 | PyErr_Format(PyExc_IOError, |
| 2397 | "readline() should have returned an str object, " |
| 2398 | "not '%.200s'", Py_TYPE(line)->tp_name); |
| 2399 | Py_DECREF(line); |
| 2400 | return NULL; |
| 2401 | } |
| 2402 | } |
| 2403 | |
| 2404 | if (line == NULL) |
| 2405 | return NULL; |
| 2406 | |
| 2407 | if (PyUnicode_GET_SIZE(line) == 0) { |
| 2408 | /* Reached EOF or would have blocked */ |
| 2409 | Py_DECREF(line); |
| 2410 | Py_CLEAR(self->snapshot); |
| 2411 | self->telling = self->seekable; |
| 2412 | return NULL; |
| 2413 | } |
| 2414 | |
| 2415 | return line; |
| 2416 | } |
| 2417 | |
| 2418 | static PyObject * |
| 2419 | TextIOWrapper_name_get(PyTextIOWrapperObject *self, void *context) |
| 2420 | { |
| 2421 | CHECK_INITIALIZED(self); |
| 2422 | return PyObject_GetAttrString(self->buffer, "name"); |
| 2423 | } |
| 2424 | |
| 2425 | static PyObject * |
| 2426 | TextIOWrapper_closed_get(PyTextIOWrapperObject *self, void *context) |
| 2427 | { |
| 2428 | CHECK_INITIALIZED(self); |
| 2429 | return PyObject_GetAttr(self->buffer, _PyIO_str_closed); |
| 2430 | } |
| 2431 | |
| 2432 | static PyObject * |
| 2433 | TextIOWrapper_newlines_get(PyTextIOWrapperObject *self, void *context) |
| 2434 | { |
| 2435 | PyObject *res; |
| 2436 | CHECK_INITIALIZED(self); |
| 2437 | if (self->decoder == NULL) |
| 2438 | Py_RETURN_NONE; |
| 2439 | res = PyObject_GetAttr(self->decoder, _PyIO_str_newlines); |
| 2440 | if (res == NULL) { |
| 2441 | PyErr_Clear(); |
| 2442 | Py_RETURN_NONE; |
| 2443 | } |
| 2444 | return res; |
| 2445 | } |
| 2446 | |
| 2447 | static PyObject * |
| 2448 | TextIOWrapper_chunk_size_get(PyTextIOWrapperObject *self, void *context) |
| 2449 | { |
| 2450 | CHECK_INITIALIZED(self); |
| 2451 | return PyLong_FromSsize_t(self->chunk_size); |
| 2452 | } |
| 2453 | |
| 2454 | static int |
| 2455 | TextIOWrapper_chunk_size_set(PyTextIOWrapperObject *self, |
| 2456 | PyObject *arg, void *context) |
| 2457 | { |
| 2458 | Py_ssize_t n; |
| 2459 | CHECK_INITIALIZED_INT(self); |
| 2460 | n = PyNumber_AsSsize_t(arg, PyExc_TypeError); |
| 2461 | if (n == -1 && PyErr_Occurred()) |
| 2462 | return -1; |
| 2463 | if (n <= 0) { |
| 2464 | PyErr_SetString(PyExc_ValueError, |
| 2465 | "a strictly positive integer is required"); |
| 2466 | return -1; |
| 2467 | } |
| 2468 | self->chunk_size = n; |
| 2469 | return 0; |
| 2470 | } |
| 2471 | |
| 2472 | static PyMethodDef TextIOWrapper_methods[] = { |
Benjamin Peterson | d2e0c79 | 2009-05-01 20:40:59 +0000 | [diff] [blame] | 2473 | {"detach", (PyCFunction)TextIOWrapper_detach, METH_NOARGS}, |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2474 | {"write", (PyCFunction)TextIOWrapper_write, METH_VARARGS}, |
| 2475 | {"read", (PyCFunction)TextIOWrapper_read, METH_VARARGS}, |
| 2476 | {"readline", (PyCFunction)TextIOWrapper_readline, METH_VARARGS}, |
| 2477 | {"flush", (PyCFunction)TextIOWrapper_flush, METH_NOARGS}, |
| 2478 | {"close", (PyCFunction)TextIOWrapper_close, METH_NOARGS}, |
| 2479 | |
| 2480 | {"fileno", (PyCFunction)TextIOWrapper_fileno, METH_NOARGS}, |
| 2481 | {"seekable", (PyCFunction)TextIOWrapper_seekable, METH_NOARGS}, |
| 2482 | {"readable", (PyCFunction)TextIOWrapper_readable, METH_NOARGS}, |
| 2483 | {"writable", (PyCFunction)TextIOWrapper_writable, METH_NOARGS}, |
| 2484 | {"isatty", (PyCFunction)TextIOWrapper_isatty, METH_NOARGS}, |
| 2485 | |
| 2486 | {"seek", (PyCFunction)TextIOWrapper_seek, METH_VARARGS}, |
| 2487 | {"tell", (PyCFunction)TextIOWrapper_tell, METH_NOARGS}, |
| 2488 | {"truncate", (PyCFunction)TextIOWrapper_truncate, METH_VARARGS}, |
| 2489 | {NULL, NULL} |
| 2490 | }; |
| 2491 | |
| 2492 | static PyMemberDef TextIOWrapper_members[] = { |
| 2493 | {"encoding", T_OBJECT, offsetof(PyTextIOWrapperObject, encoding), READONLY}, |
| 2494 | {"buffer", T_OBJECT, offsetof(PyTextIOWrapperObject, buffer), READONLY}, |
| 2495 | {"line_buffering", T_BOOL, offsetof(PyTextIOWrapperObject, line_buffering), READONLY}, |
| 2496 | {NULL} |
| 2497 | }; |
| 2498 | |
| 2499 | static PyGetSetDef TextIOWrapper_getset[] = { |
| 2500 | {"name", (getter)TextIOWrapper_name_get, NULL, NULL}, |
| 2501 | {"closed", (getter)TextIOWrapper_closed_get, NULL, NULL}, |
| 2502 | /* {"mode", (getter)TextIOWrapper_mode_get, NULL, NULL}, |
| 2503 | */ |
| 2504 | {"newlines", (getter)TextIOWrapper_newlines_get, NULL, NULL}, |
| 2505 | {"_CHUNK_SIZE", (getter)TextIOWrapper_chunk_size_get, |
| 2506 | (setter)TextIOWrapper_chunk_size_set, NULL}, |
Benjamin Peterson | 1fea321 | 2009-04-19 03:15:20 +0000 | [diff] [blame] | 2507 | {NULL} |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2508 | }; |
| 2509 | |
| 2510 | PyTypeObject PyTextIOWrapper_Type = { |
| 2511 | PyVarObject_HEAD_INIT(NULL, 0) |
| 2512 | "_io.TextIOWrapper", /*tp_name*/ |
| 2513 | sizeof(PyTextIOWrapperObject), /*tp_basicsize*/ |
| 2514 | 0, /*tp_itemsize*/ |
| 2515 | (destructor)TextIOWrapper_dealloc, /*tp_dealloc*/ |
| 2516 | 0, /*tp_print*/ |
| 2517 | 0, /*tp_getattr*/ |
Benjamin Peterson | c4c0eae | 2009-03-09 00:07:03 +0000 | [diff] [blame] | 2518 | 0, /*tps_etattr*/ |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2519 | 0, /*tp_compare */ |
Benjamin Peterson | c4c0eae | 2009-03-09 00:07:03 +0000 | [diff] [blame] | 2520 | (reprfunc)TextIOWrapper_repr,/*tp_repr*/ |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2521 | 0, /*tp_as_number*/ |
| 2522 | 0, /*tp_as_sequence*/ |
| 2523 | 0, /*tp_as_mapping*/ |
| 2524 | 0, /*tp_hash */ |
| 2525 | 0, /*tp_call*/ |
| 2526 | 0, /*tp_str*/ |
| 2527 | 0, /*tp_getattro*/ |
| 2528 | 0, /*tp_setattro*/ |
| 2529 | 0, /*tp_as_buffer*/ |
| 2530 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE |
| 2531 | | Py_TPFLAGS_HAVE_GC, /*tp_flags*/ |
| 2532 | TextIOWrapper_doc, /* tp_doc */ |
| 2533 | (traverseproc)TextIOWrapper_traverse, /* tp_traverse */ |
| 2534 | (inquiry)TextIOWrapper_clear, /* tp_clear */ |
| 2535 | 0, /* tp_richcompare */ |
| 2536 | offsetof(PyTextIOWrapperObject, weakreflist), /*tp_weaklistoffset*/ |
| 2537 | 0, /* tp_iter */ |
| 2538 | (iternextfunc)TextIOWrapper_iternext, /* tp_iternext */ |
| 2539 | TextIOWrapper_methods, /* tp_methods */ |
| 2540 | TextIOWrapper_members, /* tp_members */ |
| 2541 | TextIOWrapper_getset, /* tp_getset */ |
| 2542 | 0, /* tp_base */ |
| 2543 | 0, /* tp_dict */ |
| 2544 | 0, /* tp_descr_get */ |
| 2545 | 0, /* tp_descr_set */ |
| 2546 | offsetof(PyTextIOWrapperObject, dict), /*tp_dictoffset*/ |
| 2547 | (initproc)TextIOWrapper_init, /* tp_init */ |
| 2548 | 0, /* tp_alloc */ |
| 2549 | PyType_GenericNew, /* tp_new */ |
| 2550 | }; |