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