blob: 35bd922b8eb9ac2d681777104af5499ead5a5c94 [file] [log] [blame]
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001/*
2 An implementation of Text I/O as defined by PEP 3116 - "New I/O"
Antoine Pitrou24f36292009-03-28 22:16:42 +00003
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00004 Classes defined here: TextIOBase, IncrementalNewlineDecoder, TextIOWrapper.
Antoine Pitrou24f36292009-03-28 22:16:42 +00005
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00006 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
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000016PyDoc_STRVAR(textiobase_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000017 "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
24static PyObject *
25_unsupported(const char *message)
26{
27 PyErr_SetString(IO_STATE->unsupported_operation, message);
28 return NULL;
29}
30
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000031PyDoc_STRVAR(textiobase_detach_doc,
Benjamin Petersond2e0c792009-05-01 20:40:59 +000032 "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
38static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000039textiobase_detach(PyObject *self)
Benjamin Petersond2e0c792009-05-01 20:40:59 +000040{
41 return _unsupported("detach");
42}
43
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000044PyDoc_STRVAR(textiobase_read_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000045 "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
51static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000052textiobase_read(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000053{
54 return _unsupported("read");
55}
56
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000057PyDoc_STRVAR(textiobase_readline_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000058 "Read until newline or EOF.\n"
59 "\n"
60 "Returns an empty string if EOF is hit immediately.\n"
61 );
62
63static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000064textiobase_readline(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000065{
66 return _unsupported("readline");
67}
68
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000069PyDoc_STRVAR(textiobase_write_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000070 "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
75static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000076textiobase_write(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000077{
78 return _unsupported("write");
79}
80
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000081PyDoc_STRVAR(textiobase_encoding_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000082 "Encoding of the text stream.\n"
83 "\n"
84 "Subclasses should override.\n"
85 );
86
87static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000088textiobase_encoding_get(PyObject *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000089{
90 Py_RETURN_NONE;
91}
92
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000093PyDoc_STRVAR(textiobase_newlines_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000094 "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
101static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000102textiobase_newlines_get(PyObject *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000103{
104 Py_RETURN_NONE;
105}
106
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000107PyDoc_STRVAR(textiobase_errors_doc,
Benjamin Peterson0926ad12009-06-06 18:02:12 +0000108 "The error setting of the decoder or encoder.\n"
109 "\n"
110 "Subclasses should override.\n"
111 );
112
113static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000114textiobase_errors_get(PyObject *self, void *context)
Benjamin Peterson0926ad12009-06-06 18:02:12 +0000115{
116 Py_RETURN_NONE;
117}
118
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000119
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000120static PyMethodDef textiobase_methods[] = {
121 {"detach", (PyCFunction)textiobase_detach, METH_NOARGS, textiobase_detach_doc},
122 {"read", textiobase_read, METH_VARARGS, textiobase_read_doc},
123 {"readline", textiobase_readline, METH_VARARGS, textiobase_readline_doc},
124 {"write", textiobase_write, METH_VARARGS, textiobase_write_doc},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000125 {NULL, NULL}
126};
127
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000128static PyGetSetDef textiobase_getset[] = {
129 {"encoding", (getter)textiobase_encoding_get, NULL, textiobase_encoding_doc},
130 {"newlines", (getter)textiobase_newlines_get, NULL, textiobase_newlines_doc},
131 {"errors", (getter)textiobase_errors_get, NULL, textiobase_errors_doc},
Benjamin Peterson1fea3212009-04-19 03:15:20 +0000132 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000133};
134
135PyTypeObject PyTextIOBase_Type = {
136 PyVarObject_HEAD_INIT(NULL, 0)
137 "_io._TextIOBase", /*tp_name*/
138 0, /*tp_basicsize*/
139 0, /*tp_itemsize*/
140 0, /*tp_dealloc*/
141 0, /*tp_print*/
142 0, /*tp_getattr*/
143 0, /*tp_setattr*/
144 0, /*tp_compare */
145 0, /*tp_repr*/
146 0, /*tp_as_number*/
147 0, /*tp_as_sequence*/
148 0, /*tp_as_mapping*/
149 0, /*tp_hash */
150 0, /*tp_call*/
151 0, /*tp_str*/
152 0, /*tp_getattro*/
153 0, /*tp_setattro*/
154 0, /*tp_as_buffer*/
155 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000156 textiobase_doc, /* tp_doc */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000157 0, /* tp_traverse */
158 0, /* tp_clear */
159 0, /* tp_richcompare */
160 0, /* tp_weaklistoffset */
161 0, /* tp_iter */
162 0, /* tp_iternext */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000163 textiobase_methods, /* tp_methods */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000164 0, /* tp_members */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000165 textiobase_getset, /* tp_getset */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000166 &PyIOBase_Type, /* tp_base */
167 0, /* tp_dict */
168 0, /* tp_descr_get */
169 0, /* tp_descr_set */
170 0, /* tp_dictoffset */
171 0, /* tp_init */
172 0, /* tp_alloc */
173 0, /* tp_new */
174};
175
176
177/* IncrementalNewlineDecoder */
178
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000179PyDoc_STRVAR(incrementalnewlinedecoder_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000180 "Codec used when reading a file in universal newlines mode. It wraps\n"
181 "another incremental decoder, translating \\r\\n and \\r into \\n. It also\n"
182 "records the types of newlines encountered. When used with\n"
183 "translate=False, it ensures that the newline sequence is returned in\n"
184 "one piece. When used with decoder=None, it expects unicode strings as\n"
185 "decode input and translates newlines without first invoking an external\n"
186 "decoder.\n"
187 );
188
189typedef struct {
190 PyObject_HEAD
191 PyObject *decoder;
192 PyObject *errors;
Antoine Pitrouca767bd2009-09-21 21:37:02 +0000193 signed int pendingcr: 1;
194 signed int translate: 1;
195 unsigned int seennl: 3;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000196} nldecoder_object;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000197
198static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000199incrementalnewlinedecoder_init(nldecoder_object *self,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000200 PyObject *args, PyObject *kwds)
201{
202 PyObject *decoder;
203 int translate;
204 PyObject *errors = NULL;
205 char *kwlist[] = {"decoder", "translate", "errors", NULL};
206
207 if (!PyArg_ParseTupleAndKeywords(args, kwds, "Oi|O:IncrementalNewlineDecoder",
208 kwlist, &decoder, &translate, &errors))
209 return -1;
210
211 self->decoder = decoder;
212 Py_INCREF(decoder);
213
214 if (errors == NULL) {
215 self->errors = PyUnicode_FromString("strict");
216 if (self->errors == NULL)
217 return -1;
218 }
219 else {
220 Py_INCREF(errors);
221 self->errors = errors;
222 }
223
224 self->translate = translate;
225 self->seennl = 0;
226 self->pendingcr = 0;
227
228 return 0;
229}
230
231static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000232incrementalnewlinedecoder_dealloc(nldecoder_object *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000233{
234 Py_CLEAR(self->decoder);
235 Py_CLEAR(self->errors);
236 Py_TYPE(self)->tp_free((PyObject *)self);
237}
238
239#define SEEN_CR 1
240#define SEEN_LF 2
241#define SEEN_CRLF 4
242#define SEEN_ALL (SEEN_CR | SEEN_LF | SEEN_CRLF)
243
244PyObject *
Antoine Pitrou24f36292009-03-28 22:16:42 +0000245_PyIncrementalNewlineDecoder_decode(PyObject *_self,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000246 PyObject *input, int final)
247{
248 PyObject *output;
249 Py_ssize_t output_len;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000250 nldecoder_object *self = (nldecoder_object *) _self;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000251
252 if (self->decoder == NULL) {
253 PyErr_SetString(PyExc_ValueError,
254 "IncrementalNewlineDecoder.__init__ not called");
255 return NULL;
256 }
257
258 /* decode input (with the eventual \r from a previous pass) */
259 if (self->decoder != Py_None) {
260 output = PyObject_CallMethodObjArgs(self->decoder,
261 _PyIO_str_decode, input, final ? Py_True : Py_False, NULL);
262 }
263 else {
264 output = input;
265 Py_INCREF(output);
266 }
267
268 if (output == NULL)
269 return NULL;
270
271 if (!PyUnicode_Check(output)) {
272 PyErr_SetString(PyExc_TypeError,
273 "decoder should return a string result");
274 goto error;
275 }
276
277 output_len = PyUnicode_GET_SIZE(output);
278 if (self->pendingcr && (final || output_len > 0)) {
279 Py_UNICODE *out;
280 PyObject *modified = PyUnicode_FromUnicode(NULL, output_len + 1);
281 if (modified == NULL)
282 goto error;
283 out = PyUnicode_AS_UNICODE(modified);
284 out[0] = '\r';
285 memcpy(out + 1, PyUnicode_AS_UNICODE(output),
286 output_len * sizeof(Py_UNICODE));
287 Py_DECREF(output);
288 output = modified;
289 self->pendingcr = 0;
290 output_len++;
291 }
292
293 /* retain last \r even when not translating data:
294 * then readline() is sure to get \r\n in one pass
295 */
296 if (!final) {
Antoine Pitrou24f36292009-03-28 22:16:42 +0000297 if (output_len > 0
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000298 && PyUnicode_AS_UNICODE(output)[output_len - 1] == '\r') {
299
300 if (Py_REFCNT(output) == 1) {
301 if (PyUnicode_Resize(&output, output_len - 1) < 0)
302 goto error;
303 }
304 else {
305 PyObject *modified = PyUnicode_FromUnicode(
306 PyUnicode_AS_UNICODE(output),
307 output_len - 1);
308 if (modified == NULL)
309 goto error;
310 Py_DECREF(output);
311 output = modified;
312 }
313 self->pendingcr = 1;
314 }
315 }
316
317 /* Record which newlines are read and do newline translation if desired,
318 all in one pass. */
319 {
320 Py_UNICODE *in_str;
321 Py_ssize_t len;
322 int seennl = self->seennl;
323 int only_lf = 0;
324
325 in_str = PyUnicode_AS_UNICODE(output);
326 len = PyUnicode_GET_SIZE(output);
327
328 if (len == 0)
329 return output;
330
331 /* If, up to now, newlines are consistently \n, do a quick check
332 for the \r *byte* with the libc's optimized memchr.
333 */
334 if (seennl == SEEN_LF || seennl == 0) {
Antoine Pitrou0e941892009-03-06 23:57:20 +0000335 only_lf = (memchr(in_str, '\r', len * sizeof(Py_UNICODE)) == NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000336 }
337
Antoine Pitrou66913e22009-03-06 23:40:56 +0000338 if (only_lf) {
339 /* If not already seen, quick scan for a possible "\n" character.
340 (there's nothing else to be done, even when in translation mode)
341 */
342 if (seennl == 0 &&
343 memchr(in_str, '\n', len * sizeof(Py_UNICODE)) != NULL) {
344 Py_UNICODE *s, *end;
345 s = in_str;
346 end = in_str + len;
347 for (;;) {
348 Py_UNICODE c;
349 /* Fast loop for non-control characters */
350 while (*s > '\n')
351 s++;
352 c = *s++;
353 if (c == '\n') {
354 seennl |= SEEN_LF;
355 break;
356 }
357 if (s > end)
358 break;
359 }
360 }
361 /* Finished: we have scanned for newlines, and none of them
362 need translating */
363 }
364 else if (!self->translate) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000365 Py_UNICODE *s, *end;
Antoine Pitrou66913e22009-03-06 23:40:56 +0000366 /* We have already seen all newline types, no need to scan again */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000367 if (seennl == SEEN_ALL)
368 goto endscan;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000369 s = in_str;
370 end = in_str + len;
371 for (;;) {
372 Py_UNICODE c;
373 /* Fast loop for non-control characters */
374 while (*s > '\r')
375 s++;
376 c = *s++;
377 if (c == '\n')
378 seennl |= SEEN_LF;
379 else if (c == '\r') {
380 if (*s == '\n') {
381 seennl |= SEEN_CRLF;
382 s++;
383 }
384 else
385 seennl |= SEEN_CR;
386 }
387 if (s > end)
388 break;
389 if (seennl == SEEN_ALL)
390 break;
391 }
392 endscan:
393 ;
394 }
Antoine Pitrou66913e22009-03-06 23:40:56 +0000395 else {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000396 PyObject *translated = NULL;
397 Py_UNICODE *out_str;
398 Py_UNICODE *in, *out, *end;
399 if (Py_REFCNT(output) != 1) {
400 /* We could try to optimize this so that we only do a copy
401 when there is something to translate. On the other hand,
402 most decoders should only output non-shared strings, i.e.
403 translation is done in place. */
404 translated = PyUnicode_FromUnicode(NULL, len);
405 if (translated == NULL)
406 goto error;
407 assert(Py_REFCNT(translated) == 1);
408 memcpy(PyUnicode_AS_UNICODE(translated),
409 PyUnicode_AS_UNICODE(output),
410 len * sizeof(Py_UNICODE));
411 }
412 else {
413 translated = output;
414 }
415 out_str = PyUnicode_AS_UNICODE(translated);
416 in = in_str;
417 out = out_str;
418 end = in_str + len;
419 for (;;) {
420 Py_UNICODE c;
421 /* Fast loop for non-control characters */
422 while ((c = *in++) > '\r')
423 *out++ = c;
424 if (c == '\n') {
425 *out++ = c;
426 seennl |= SEEN_LF;
427 continue;
428 }
429 if (c == '\r') {
430 if (*in == '\n') {
431 in++;
432 seennl |= SEEN_CRLF;
433 }
434 else
435 seennl |= SEEN_CR;
436 *out++ = '\n';
437 continue;
438 }
439 if (in > end)
440 break;
441 *out++ = c;
442 }
443 if (translated != output) {
444 Py_DECREF(output);
445 output = translated;
446 }
447 if (out - out_str != len) {
448 if (PyUnicode_Resize(&output, out - out_str) < 0)
449 goto error;
450 }
451 }
452 self->seennl |= seennl;
453 }
454
455 return output;
456
457 error:
458 Py_DECREF(output);
459 return NULL;
460}
461
462static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000463incrementalnewlinedecoder_decode(nldecoder_object *self,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000464 PyObject *args, PyObject *kwds)
465{
466 char *kwlist[] = {"input", "final", NULL};
467 PyObject *input;
468 int final = 0;
469
470 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|i:IncrementalNewlineDecoder",
471 kwlist, &input, &final))
472 return NULL;
473 return _PyIncrementalNewlineDecoder_decode((PyObject *) self, input, final);
474}
475
476static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000477incrementalnewlinedecoder_getstate(nldecoder_object *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000478{
479 PyObject *buffer;
480 unsigned PY_LONG_LONG flag;
481
482 if (self->decoder != Py_None) {
483 PyObject *state = PyObject_CallMethodObjArgs(self->decoder,
484 _PyIO_str_getstate, NULL);
485 if (state == NULL)
486 return NULL;
487 if (!PyArg_Parse(state, "(OK)", &buffer, &flag)) {
488 Py_DECREF(state);
489 return NULL;
490 }
491 Py_INCREF(buffer);
492 Py_DECREF(state);
493 }
494 else {
495 buffer = PyBytes_FromString("");
496 flag = 0;
497 }
498 flag <<= 1;
499 if (self->pendingcr)
500 flag |= 1;
501 return Py_BuildValue("NK", buffer, flag);
502}
503
504static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000505incrementalnewlinedecoder_setstate(nldecoder_object *self, PyObject *state)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000506{
507 PyObject *buffer;
508 unsigned PY_LONG_LONG flag;
509
510 if (!PyArg_Parse(state, "(OK)", &buffer, &flag))
511 return NULL;
512
513 self->pendingcr = (int) flag & 1;
514 flag >>= 1;
515
516 if (self->decoder != Py_None)
517 return PyObject_CallMethod(self->decoder,
518 "setstate", "((OK))", buffer, flag);
519 else
520 Py_RETURN_NONE;
521}
522
523static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000524incrementalnewlinedecoder_reset(nldecoder_object *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000525{
526 self->seennl = 0;
527 self->pendingcr = 0;
528 if (self->decoder != Py_None)
529 return PyObject_CallMethodObjArgs(self->decoder, _PyIO_str_reset, NULL);
530 else
531 Py_RETURN_NONE;
532}
533
534static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000535incrementalnewlinedecoder_newlines_get(nldecoder_object *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000536{
537 switch (self->seennl) {
538 case SEEN_CR:
539 return PyUnicode_FromString("\r");
540 case SEEN_LF:
541 return PyUnicode_FromString("\n");
542 case SEEN_CRLF:
543 return PyUnicode_FromString("\r\n");
544 case SEEN_CR | SEEN_LF:
545 return Py_BuildValue("ss", "\r", "\n");
546 case SEEN_CR | SEEN_CRLF:
547 return Py_BuildValue("ss", "\r", "\r\n");
548 case SEEN_LF | SEEN_CRLF:
549 return Py_BuildValue("ss", "\n", "\r\n");
550 case SEEN_CR | SEEN_LF | SEEN_CRLF:
551 return Py_BuildValue("sss", "\r", "\n", "\r\n");
552 default:
553 Py_RETURN_NONE;
554 }
555
556}
557
558
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000559static PyMethodDef incrementalnewlinedecoder_methods[] = {
560 {"decode", (PyCFunction)incrementalnewlinedecoder_decode, METH_VARARGS|METH_KEYWORDS},
561 {"getstate", (PyCFunction)incrementalnewlinedecoder_getstate, METH_NOARGS},
562 {"setstate", (PyCFunction)incrementalnewlinedecoder_setstate, METH_O},
563 {"reset", (PyCFunction)incrementalnewlinedecoder_reset, METH_NOARGS},
Benjamin Peterson1fea3212009-04-19 03:15:20 +0000564 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000565};
566
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000567static PyGetSetDef incrementalnewlinedecoder_getset[] = {
568 {"newlines", (getter)incrementalnewlinedecoder_newlines_get, NULL, NULL},
Benjamin Peterson1fea3212009-04-19 03:15:20 +0000569 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000570};
571
572PyTypeObject PyIncrementalNewlineDecoder_Type = {
573 PyVarObject_HEAD_INIT(NULL, 0)
574 "_io.IncrementalNewlineDecoder", /*tp_name*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000575 sizeof(nldecoder_object), /*tp_basicsize*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000576 0, /*tp_itemsize*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000577 (destructor)incrementalnewlinedecoder_dealloc, /*tp_dealloc*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000578 0, /*tp_print*/
579 0, /*tp_getattr*/
580 0, /*tp_setattr*/
581 0, /*tp_compare */
582 0, /*tp_repr*/
583 0, /*tp_as_number*/
584 0, /*tp_as_sequence*/
585 0, /*tp_as_mapping*/
586 0, /*tp_hash */
587 0, /*tp_call*/
588 0, /*tp_str*/
589 0, /*tp_getattro*/
590 0, /*tp_setattro*/
591 0, /*tp_as_buffer*/
592 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000593 incrementalnewlinedecoder_doc, /* tp_doc */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000594 0, /* tp_traverse */
595 0, /* tp_clear */
596 0, /* tp_richcompare */
597 0, /*tp_weaklistoffset*/
598 0, /* tp_iter */
599 0, /* tp_iternext */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000600 incrementalnewlinedecoder_methods, /* tp_methods */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000601 0, /* tp_members */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000602 incrementalnewlinedecoder_getset, /* tp_getset */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000603 0, /* tp_base */
604 0, /* tp_dict */
605 0, /* tp_descr_get */
606 0, /* tp_descr_set */
607 0, /* tp_dictoffset */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000608 (initproc)incrementalnewlinedecoder_init, /* tp_init */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000609 0, /* tp_alloc */
610 PyType_GenericNew, /* tp_new */
611};
612
613
614/* TextIOWrapper */
615
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000616PyDoc_STRVAR(textiowrapper_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000617 "Character and line based layer over a BufferedIOBase object, buffer.\n"
618 "\n"
619 "encoding gives the name of the encoding that the stream will be\n"
620 "decoded or encoded with. It defaults to locale.getpreferredencoding.\n"
621 "\n"
622 "errors determines the strictness of encoding and decoding (see the\n"
623 "codecs.register) and defaults to \"strict\".\n"
624 "\n"
625 "newline can be None, '', '\\n', '\\r', or '\\r\\n'. It controls the\n"
626 "handling of line endings. If it is None, universal newlines is\n"
627 "enabled. With this enabled, on input, the lines endings '\\n', '\\r',\n"
628 "or '\\r\\n' are translated to '\\n' before being returned to the\n"
629 "caller. Conversely, on output, '\\n' is translated to the system\n"
630 "default line seperator, os.linesep. If newline is any other of its\n"
631 "legal values, that newline becomes the newline when the file is read\n"
632 "and it is returned untranslated. On output, '\\n' is converted to the\n"
633 "newline.\n"
634 "\n"
635 "If line_buffering is True, a call to flush is implied when a call to\n"
636 "write contains a newline character."
637 );
638
639typedef PyObject *
640 (*encodefunc_t)(PyObject *, PyObject *);
641
642typedef struct
643{
644 PyObject_HEAD
645 int ok; /* initialized? */
Benjamin Petersond2e0c792009-05-01 20:40:59 +0000646 int detached;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000647 Py_ssize_t chunk_size;
648 PyObject *buffer;
649 PyObject *encoding;
650 PyObject *encoder;
651 PyObject *decoder;
652 PyObject *readnl;
653 PyObject *errors;
654 const char *writenl; /* utf-8 encoded, NULL stands for \n */
655 char line_buffering;
656 char readuniversal;
657 char readtranslate;
658 char writetranslate;
659 char seekable;
660 char telling;
Antoine Pitroue033e062010-10-29 10:38:18 +0000661 char deallocating;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000662 /* Specialized encoding func (see below) */
663 encodefunc_t encodefunc;
Antoine Pitroue4501852009-05-14 18:55:55 +0000664 /* Whether or not it's the start of the stream */
665 char encoding_start_of_stream;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000666
667 /* Reads and writes are internally buffered in order to speed things up.
668 However, any read will first flush the write buffer if itsn't empty.
Antoine Pitrou24f36292009-03-28 22:16:42 +0000669
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000670 Please also note that text to be written is first encoded before being
671 buffered. This is necessary so that encoding errors are immediately
672 reported to the caller, but it unfortunately means that the
673 IncrementalEncoder (whose encode() method is always written in Python)
674 becomes a bottleneck for small writes.
675 */
676 PyObject *decoded_chars; /* buffer for text returned from decoder */
677 Py_ssize_t decoded_chars_used; /* offset into _decoded_chars for read() */
678 PyObject *pending_bytes; /* list of bytes objects waiting to be
679 written, or NULL */
680 Py_ssize_t pending_bytes_count;
Antoine Pitrou211b81d2011-02-25 20:27:33 +0000681
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000682 /* snapshot is either None, or a tuple (dec_flags, next_input) where
683 * dec_flags is the second (integer) item of the decoder state and
684 * next_input is the chunk of input bytes that comes next after the
685 * snapshot point. We use this to reconstruct decoder states in tell().
686 */
Antoine Pitrou211b81d2011-02-25 20:27:33 +0000687 PyObject *snapshot;
688 /* Bytes-to-characters ratio for the current chunk. Serves as input for
689 the heuristic in tell(). */
690 double b2cratio;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000691
692 /* Cache raw object if it's a FileIO object */
693 PyObject *raw;
694
695 PyObject *weakreflist;
696 PyObject *dict;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000697} textio;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000698
699
700/* A couple of specialized cases in order to bypass the slow incremental
701 encoding methods for the most popular encodings. */
702
703static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000704ascii_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000705{
706 return PyUnicode_EncodeASCII(PyUnicode_AS_UNICODE(text),
707 PyUnicode_GET_SIZE(text),
708 PyBytes_AS_STRING(self->errors));
709}
710
711static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000712utf16be_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000713{
714 return PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(text),
715 PyUnicode_GET_SIZE(text),
716 PyBytes_AS_STRING(self->errors), 1);
717}
718
719static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000720utf16le_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000721{
722 return PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(text),
723 PyUnicode_GET_SIZE(text),
724 PyBytes_AS_STRING(self->errors), -1);
725}
726
727static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000728utf16_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000729{
Antoine Pitroue4501852009-05-14 18:55:55 +0000730 if (!self->encoding_start_of_stream) {
731 /* Skip the BOM and use native byte ordering */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000732#if defined(WORDS_BIGENDIAN)
Antoine Pitroue4501852009-05-14 18:55:55 +0000733 return utf16be_encode(self, text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000734#else
Antoine Pitroue4501852009-05-14 18:55:55 +0000735 return utf16le_encode(self, text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000736#endif
Antoine Pitroue4501852009-05-14 18:55:55 +0000737 }
738 return PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(text),
739 PyUnicode_GET_SIZE(text),
740 PyBytes_AS_STRING(self->errors), 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000741}
742
Antoine Pitroue4501852009-05-14 18:55:55 +0000743static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000744utf32be_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000745{
746 return PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(text),
747 PyUnicode_GET_SIZE(text),
748 PyBytes_AS_STRING(self->errors), 1);
749}
750
751static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000752utf32le_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000753{
754 return PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(text),
755 PyUnicode_GET_SIZE(text),
756 PyBytes_AS_STRING(self->errors), -1);
757}
758
759static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000760utf32_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000761{
762 if (!self->encoding_start_of_stream) {
763 /* Skip the BOM and use native byte ordering */
764#if defined(WORDS_BIGENDIAN)
765 return utf32be_encode(self, text);
766#else
767 return utf32le_encode(self, text);
768#endif
769 }
770 return PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(text),
771 PyUnicode_GET_SIZE(text),
772 PyBytes_AS_STRING(self->errors), 0);
773}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000774
775static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000776utf8_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000777{
778 return PyUnicode_EncodeUTF8(PyUnicode_AS_UNICODE(text),
779 PyUnicode_GET_SIZE(text),
780 PyBytes_AS_STRING(self->errors));
781}
782
783static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000784latin1_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000785{
786 return PyUnicode_EncodeLatin1(PyUnicode_AS_UNICODE(text),
787 PyUnicode_GET_SIZE(text),
788 PyBytes_AS_STRING(self->errors));
789}
790
791/* Map normalized encoding names onto the specialized encoding funcs */
792
793typedef struct {
794 const char *name;
795 encodefunc_t encodefunc;
796} encodefuncentry;
797
Antoine Pitrou24f36292009-03-28 22:16:42 +0000798static encodefuncentry encodefuncs[] = {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000799 {"ascii", (encodefunc_t) ascii_encode},
800 {"iso8859-1", (encodefunc_t) latin1_encode},
Antoine Pitroue4501852009-05-14 18:55:55 +0000801 {"utf-8", (encodefunc_t) utf8_encode},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000802 {"utf-16-be", (encodefunc_t) utf16be_encode},
803 {"utf-16-le", (encodefunc_t) utf16le_encode},
804 {"utf-16", (encodefunc_t) utf16_encode},
Antoine Pitroue4501852009-05-14 18:55:55 +0000805 {"utf-32-be", (encodefunc_t) utf32be_encode},
806 {"utf-32-le", (encodefunc_t) utf32le_encode},
807 {"utf-32", (encodefunc_t) utf32_encode},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000808 {NULL, NULL}
809};
810
811
812static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000813textiowrapper_init(textio *self, PyObject *args, PyObject *kwds)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000814{
815 char *kwlist[] = {"buffer", "encoding", "errors",
816 "newline", "line_buffering",
817 NULL};
818 PyObject *buffer, *raw;
819 char *encoding = NULL;
820 char *errors = NULL;
821 char *newline = NULL;
822 int line_buffering = 0;
823 _PyIO_State *state = IO_STATE;
824
825 PyObject *res;
826 int r;
827
828 self->ok = 0;
Benjamin Petersond2e0c792009-05-01 20:40:59 +0000829 self->detached = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000830 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|zzzi:fileio",
831 kwlist, &buffer, &encoding, &errors,
832 &newline, &line_buffering))
833 return -1;
834
835 if (newline && newline[0] != '\0'
836 && !(newline[0] == '\n' && newline[1] == '\0')
837 && !(newline[0] == '\r' && newline[1] == '\0')
838 && !(newline[0] == '\r' && newline[1] == '\n' && newline[2] == '\0')) {
839 PyErr_Format(PyExc_ValueError,
840 "illegal newline value: %s", newline);
841 return -1;
842 }
843
844 Py_CLEAR(self->buffer);
845 Py_CLEAR(self->encoding);
846 Py_CLEAR(self->encoder);
847 Py_CLEAR(self->decoder);
848 Py_CLEAR(self->readnl);
849 Py_CLEAR(self->decoded_chars);
850 Py_CLEAR(self->pending_bytes);
851 Py_CLEAR(self->snapshot);
852 Py_CLEAR(self->errors);
853 Py_CLEAR(self->raw);
854 self->decoded_chars_used = 0;
855 self->pending_bytes_count = 0;
856 self->encodefunc = NULL;
Antoine Pitrou211b81d2011-02-25 20:27:33 +0000857 self->b2cratio = 0.0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000858
859 if (encoding == NULL) {
860 /* Try os.device_encoding(fileno) */
861 PyObject *fileno;
862 fileno = PyObject_CallMethod(buffer, "fileno", NULL);
863 /* Ignore only AttributeError and UnsupportedOperation */
864 if (fileno == NULL) {
865 if (PyErr_ExceptionMatches(PyExc_AttributeError) ||
866 PyErr_ExceptionMatches(state->unsupported_operation)) {
867 PyErr_Clear();
868 }
869 else {
870 goto error;
871 }
872 }
873 else {
874 self->encoding = PyObject_CallMethod(state->os_module,
875 "device_encoding",
876 "N", fileno);
877 if (self->encoding == NULL)
878 goto error;
879 else if (!PyUnicode_Check(self->encoding))
880 Py_CLEAR(self->encoding);
881 }
882 }
883 if (encoding == NULL && self->encoding == NULL) {
884 if (state->locale_module == NULL) {
885 state->locale_module = PyImport_ImportModule("locale");
886 if (state->locale_module == NULL)
887 goto catch_ImportError;
888 else
889 goto use_locale;
890 }
891 else {
892 use_locale:
893 self->encoding = PyObject_CallMethod(
894 state->locale_module, "getpreferredencoding", NULL);
895 if (self->encoding == NULL) {
896 catch_ImportError:
897 /*
898 Importing locale can raise a ImportError because of
899 _functools, and locale.getpreferredencoding can raise a
900 ImportError if _locale is not available. These will happen
901 during module building.
902 */
903 if (PyErr_ExceptionMatches(PyExc_ImportError)) {
904 PyErr_Clear();
905 self->encoding = PyUnicode_FromString("ascii");
906 }
907 else
908 goto error;
909 }
910 else if (!PyUnicode_Check(self->encoding))
911 Py_CLEAR(self->encoding);
912 }
913 }
Victor Stinnerf6c57832010-05-19 01:17:01 +0000914 if (self->encoding != NULL) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000915 encoding = _PyUnicode_AsString(self->encoding);
Victor Stinnerf6c57832010-05-19 01:17:01 +0000916 if (encoding == NULL)
917 goto error;
918 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000919 else if (encoding != NULL) {
920 self->encoding = PyUnicode_FromString(encoding);
921 if (self->encoding == NULL)
922 goto error;
923 }
924 else {
925 PyErr_SetString(PyExc_IOError,
926 "could not determine default encoding");
927 }
928
929 if (errors == NULL)
930 errors = "strict";
931 self->errors = PyBytes_FromString(errors);
932 if (self->errors == NULL)
933 goto error;
934
935 self->chunk_size = 8192;
936 self->readuniversal = (newline == NULL || newline[0] == '\0');
937 self->line_buffering = line_buffering;
938 self->readtranslate = (newline == NULL);
939 if (newline) {
940 self->readnl = PyUnicode_FromString(newline);
941 if (self->readnl == NULL)
942 return -1;
943 }
944 self->writetranslate = (newline == NULL || newline[0] != '\0');
945 if (!self->readuniversal && self->readnl) {
946 self->writenl = _PyUnicode_AsString(self->readnl);
Victor Stinnerf6c57832010-05-19 01:17:01 +0000947 if (self->writenl == NULL)
948 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000949 if (!strcmp(self->writenl, "\n"))
950 self->writenl = NULL;
951 }
952#ifdef MS_WINDOWS
953 else
954 self->writenl = "\r\n";
955#endif
956
957 /* Build the decoder object */
958 res = PyObject_CallMethod(buffer, "readable", NULL);
959 if (res == NULL)
960 goto error;
961 r = PyObject_IsTrue(res);
962 Py_DECREF(res);
963 if (r == -1)
964 goto error;
965 if (r == 1) {
966 self->decoder = PyCodec_IncrementalDecoder(
967 encoding, errors);
968 if (self->decoder == NULL)
969 goto error;
970
971 if (self->readuniversal) {
972 PyObject *incrementalDecoder = PyObject_CallFunction(
973 (PyObject *)&PyIncrementalNewlineDecoder_Type,
974 "Oi", self->decoder, (int)self->readtranslate);
975 if (incrementalDecoder == NULL)
976 goto error;
977 Py_CLEAR(self->decoder);
978 self->decoder = incrementalDecoder;
979 }
980 }
981
982 /* Build the encoder object */
983 res = PyObject_CallMethod(buffer, "writable", NULL);
984 if (res == NULL)
985 goto error;
986 r = PyObject_IsTrue(res);
987 Py_DECREF(res);
988 if (r == -1)
989 goto error;
990 if (r == 1) {
991 PyObject *ci;
992 self->encoder = PyCodec_IncrementalEncoder(
993 encoding, errors);
994 if (self->encoder == NULL)
995 goto error;
996 /* Get the normalized named of the codec */
997 ci = _PyCodec_Lookup(encoding);
998 if (ci == NULL)
999 goto error;
1000 res = PyObject_GetAttrString(ci, "name");
1001 Py_DECREF(ci);
Benjamin Peterson2cfca792009-06-06 20:46:48 +00001002 if (res == NULL) {
1003 if (PyErr_ExceptionMatches(PyExc_AttributeError))
1004 PyErr_Clear();
1005 else
1006 goto error;
1007 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001008 else if (PyUnicode_Check(res)) {
1009 encodefuncentry *e = encodefuncs;
1010 while (e->name != NULL) {
1011 if (!PyUnicode_CompareWithASCIIString(res, e->name)) {
1012 self->encodefunc = e->encodefunc;
1013 break;
1014 }
1015 e++;
1016 }
1017 }
1018 Py_XDECREF(res);
1019 }
1020
1021 self->buffer = buffer;
1022 Py_INCREF(buffer);
Antoine Pitrou24f36292009-03-28 22:16:42 +00001023
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001024 if (Py_TYPE(buffer) == &PyBufferedReader_Type ||
1025 Py_TYPE(buffer) == &PyBufferedWriter_Type ||
1026 Py_TYPE(buffer) == &PyBufferedRandom_Type) {
1027 raw = PyObject_GetAttrString(buffer, "raw");
1028 /* Cache the raw FileIO object to speed up 'closed' checks */
Benjamin Peterson2cfca792009-06-06 20:46:48 +00001029 if (raw == NULL) {
1030 if (PyErr_ExceptionMatches(PyExc_AttributeError))
1031 PyErr_Clear();
1032 else
1033 goto error;
1034 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001035 else if (Py_TYPE(raw) == &PyFileIO_Type)
1036 self->raw = raw;
1037 else
1038 Py_DECREF(raw);
1039 }
1040
1041 res = PyObject_CallMethod(buffer, "seekable", NULL);
1042 if (res == NULL)
1043 goto error;
1044 self->seekable = self->telling = PyObject_IsTrue(res);
1045 Py_DECREF(res);
1046
Antoine Pitroue4501852009-05-14 18:55:55 +00001047 self->encoding_start_of_stream = 0;
1048 if (self->seekable && self->encoder) {
1049 PyObject *cookieObj;
1050 int cmp;
1051
1052 self->encoding_start_of_stream = 1;
1053
1054 cookieObj = PyObject_CallMethodObjArgs(buffer, _PyIO_str_tell, NULL);
1055 if (cookieObj == NULL)
1056 goto error;
1057
1058 cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_EQ);
1059 Py_DECREF(cookieObj);
1060 if (cmp < 0) {
1061 goto error;
1062 }
1063
1064 if (cmp == 0) {
1065 self->encoding_start_of_stream = 0;
1066 res = PyObject_CallMethodObjArgs(self->encoder, _PyIO_str_setstate,
1067 _PyIO_zero, NULL);
1068 if (res == NULL)
1069 goto error;
1070 Py_DECREF(res);
1071 }
1072 }
1073
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001074 self->ok = 1;
1075 return 0;
1076
1077 error:
1078 return -1;
1079}
1080
1081static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001082_textiowrapper_clear(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001083{
1084 if (self->ok && _PyIOBase_finalize((PyObject *) self) < 0)
1085 return -1;
1086 self->ok = 0;
1087 Py_CLEAR(self->buffer);
1088 Py_CLEAR(self->encoding);
1089 Py_CLEAR(self->encoder);
1090 Py_CLEAR(self->decoder);
1091 Py_CLEAR(self->readnl);
1092 Py_CLEAR(self->decoded_chars);
1093 Py_CLEAR(self->pending_bytes);
1094 Py_CLEAR(self->snapshot);
1095 Py_CLEAR(self->errors);
1096 Py_CLEAR(self->raw);
1097 return 0;
1098}
1099
1100static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001101textiowrapper_dealloc(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001102{
Antoine Pitroue033e062010-10-29 10:38:18 +00001103 self->deallocating = 1;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001104 if (_textiowrapper_clear(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001105 return;
1106 _PyObject_GC_UNTRACK(self);
1107 if (self->weakreflist != NULL)
1108 PyObject_ClearWeakRefs((PyObject *)self);
1109 Py_CLEAR(self->dict);
1110 Py_TYPE(self)->tp_free((PyObject *)self);
1111}
1112
1113static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001114textiowrapper_traverse(textio *self, visitproc visit, void *arg)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001115{
1116 Py_VISIT(self->buffer);
1117 Py_VISIT(self->encoding);
1118 Py_VISIT(self->encoder);
1119 Py_VISIT(self->decoder);
1120 Py_VISIT(self->readnl);
1121 Py_VISIT(self->decoded_chars);
1122 Py_VISIT(self->pending_bytes);
1123 Py_VISIT(self->snapshot);
1124 Py_VISIT(self->errors);
1125 Py_VISIT(self->raw);
1126
1127 Py_VISIT(self->dict);
1128 return 0;
1129}
1130
1131static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001132textiowrapper_clear(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001133{
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001134 if (_textiowrapper_clear(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001135 return -1;
1136 Py_CLEAR(self->dict);
1137 return 0;
1138}
1139
1140static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001141textiowrapper_closed_get(textio *self, void *context);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001142
1143/* This macro takes some shortcuts to make the common case faster. */
1144#define CHECK_CLOSED(self) \
1145 do { \
1146 int r; \
1147 PyObject *_res; \
1148 if (Py_TYPE(self) == &PyTextIOWrapper_Type) { \
1149 if (self->raw != NULL) \
1150 r = _PyFileIO_closed(self->raw); \
1151 else { \
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001152 _res = textiowrapper_closed_get(self, NULL); \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001153 if (_res == NULL) \
1154 return NULL; \
1155 r = PyObject_IsTrue(_res); \
1156 Py_DECREF(_res); \
1157 if (r < 0) \
1158 return NULL; \
1159 } \
1160 if (r > 0) { \
1161 PyErr_SetString(PyExc_ValueError, \
1162 "I/O operation on closed file."); \
1163 return NULL; \
1164 } \
1165 } \
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001166 else if (_PyIOBase_check_closed((PyObject *)self, Py_True) == NULL) \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001167 return NULL; \
1168 } while (0)
1169
1170#define CHECK_INITIALIZED(self) \
1171 if (self->ok <= 0) { \
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001172 if (self->detached) { \
1173 PyErr_SetString(PyExc_ValueError, \
1174 "underlying buffer has been detached"); \
1175 } else { \
1176 PyErr_SetString(PyExc_ValueError, \
1177 "I/O operation on uninitialized object"); \
1178 } \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001179 return NULL; \
1180 }
1181
1182#define CHECK_INITIALIZED_INT(self) \
1183 if (self->ok <= 0) { \
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001184 if (self->detached) { \
1185 PyErr_SetString(PyExc_ValueError, \
1186 "underlying buffer has been detached"); \
1187 } else { \
1188 PyErr_SetString(PyExc_ValueError, \
1189 "I/O operation on uninitialized object"); \
1190 } \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001191 return -1; \
1192 }
1193
1194
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001195static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001196textiowrapper_detach(textio *self)
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001197{
1198 PyObject *buffer, *res;
1199 CHECK_INITIALIZED(self);
1200 res = PyObject_CallMethodObjArgs((PyObject *)self, _PyIO_str_flush, NULL);
1201 if (res == NULL)
1202 return NULL;
1203 Py_DECREF(res);
1204 buffer = self->buffer;
1205 self->buffer = NULL;
1206 self->detached = 1;
1207 self->ok = 0;
1208 return buffer;
1209}
1210
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001211Py_LOCAL_INLINE(const Py_UNICODE *)
1212findchar(const Py_UNICODE *s, Py_ssize_t size, Py_UNICODE ch)
1213{
1214 /* like wcschr, but doesn't stop at NULL characters */
1215 while (size-- > 0) {
1216 if (*s == ch)
1217 return s;
1218 s++;
1219 }
1220 return NULL;
1221}
1222
Antoine Pitrou24f36292009-03-28 22:16:42 +00001223/* Flush the internal write buffer. This doesn't explicitly flush the
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001224 underlying buffered object, though. */
1225static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001226_textiowrapper_writeflush(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001227{
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001228 PyObject *pending, *b, *ret;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001229
1230 if (self->pending_bytes == NULL)
1231 return 0;
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001232
1233 pending = self->pending_bytes;
1234 Py_INCREF(pending);
1235 self->pending_bytes_count = 0;
1236 Py_CLEAR(self->pending_bytes);
1237
1238 b = _PyBytes_Join(_PyIO_empty_bytes, pending);
1239 Py_DECREF(pending);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001240 if (b == NULL)
1241 return -1;
1242 ret = PyObject_CallMethodObjArgs(self->buffer,
1243 _PyIO_str_write, b, NULL);
1244 Py_DECREF(b);
1245 if (ret == NULL)
1246 return -1;
1247 Py_DECREF(ret);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001248 return 0;
1249}
1250
1251static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001252textiowrapper_write(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001253{
1254 PyObject *ret;
1255 PyObject *text; /* owned reference */
1256 PyObject *b;
1257 Py_ssize_t textlen;
1258 int haslf = 0;
1259 int needflush = 0;
1260
1261 CHECK_INITIALIZED(self);
1262
1263 if (!PyArg_ParseTuple(args, "U:write", &text)) {
1264 return NULL;
1265 }
1266
1267 CHECK_CLOSED(self);
1268
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001269 if (self->encoder == NULL)
1270 return _unsupported("not writable");
Benjamin Peterson81971ea2009-05-14 22:01:31 +00001271
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001272 Py_INCREF(text);
1273
1274 textlen = PyUnicode_GetSize(text);
1275
1276 if ((self->writetranslate && self->writenl != NULL) || self->line_buffering)
1277 if (findchar(PyUnicode_AS_UNICODE(text),
1278 PyUnicode_GET_SIZE(text), '\n'))
1279 haslf = 1;
1280
1281 if (haslf && self->writetranslate && self->writenl != NULL) {
1282 PyObject *newtext = PyObject_CallMethod(
1283 text, "replace", "ss", "\n", self->writenl);
1284 Py_DECREF(text);
1285 if (newtext == NULL)
1286 return NULL;
1287 text = newtext;
1288 }
1289
1290 if (self->line_buffering &&
1291 (haslf ||
1292 findchar(PyUnicode_AS_UNICODE(text),
1293 PyUnicode_GET_SIZE(text), '\r')))
1294 needflush = 1;
1295
1296 /* XXX What if we were just reading? */
Antoine Pitroue4501852009-05-14 18:55:55 +00001297 if (self->encodefunc != NULL) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001298 b = (*self->encodefunc)((PyObject *) self, text);
Antoine Pitroue4501852009-05-14 18:55:55 +00001299 self->encoding_start_of_stream = 0;
1300 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001301 else
1302 b = PyObject_CallMethodObjArgs(self->encoder,
1303 _PyIO_str_encode, text, NULL);
1304 Py_DECREF(text);
1305 if (b == NULL)
1306 return NULL;
1307
1308 if (self->pending_bytes == NULL) {
1309 self->pending_bytes = PyList_New(0);
1310 if (self->pending_bytes == NULL) {
1311 Py_DECREF(b);
1312 return NULL;
1313 }
1314 self->pending_bytes_count = 0;
1315 }
1316 if (PyList_Append(self->pending_bytes, b) < 0) {
1317 Py_DECREF(b);
1318 return NULL;
1319 }
1320 self->pending_bytes_count += PyBytes_GET_SIZE(b);
1321 Py_DECREF(b);
1322 if (self->pending_bytes_count > self->chunk_size || needflush) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001323 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001324 return NULL;
1325 }
Antoine Pitrou24f36292009-03-28 22:16:42 +00001326
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001327 if (needflush) {
1328 ret = PyObject_CallMethodObjArgs(self->buffer, _PyIO_str_flush, NULL);
1329 if (ret == NULL)
1330 return NULL;
1331 Py_DECREF(ret);
1332 }
1333
1334 Py_CLEAR(self->snapshot);
1335
1336 if (self->decoder) {
1337 ret = PyObject_CallMethod(self->decoder, "reset", NULL);
1338 if (ret == NULL)
1339 return NULL;
1340 Py_DECREF(ret);
1341 }
1342
1343 return PyLong_FromSsize_t(textlen);
1344}
1345
1346/* Steal a reference to chars and store it in the decoded_char buffer;
1347 */
1348static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001349textiowrapper_set_decoded_chars(textio *self, PyObject *chars)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001350{
1351 Py_CLEAR(self->decoded_chars);
1352 self->decoded_chars = chars;
1353 self->decoded_chars_used = 0;
1354}
1355
1356static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001357textiowrapper_get_decoded_chars(textio *self, Py_ssize_t n)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001358{
1359 PyObject *chars;
1360 Py_ssize_t avail;
1361
1362 if (self->decoded_chars == NULL)
1363 return PyUnicode_FromStringAndSize(NULL, 0);
1364
1365 avail = (PyUnicode_GET_SIZE(self->decoded_chars)
1366 - self->decoded_chars_used);
1367
1368 assert(avail >= 0);
1369
1370 if (n < 0 || n > avail)
1371 n = avail;
1372
1373 if (self->decoded_chars_used > 0 || n < avail) {
1374 chars = PyUnicode_FromUnicode(
1375 PyUnicode_AS_UNICODE(self->decoded_chars)
1376 + self->decoded_chars_used, n);
1377 if (chars == NULL)
1378 return NULL;
1379 }
1380 else {
1381 chars = self->decoded_chars;
1382 Py_INCREF(chars);
1383 }
1384
1385 self->decoded_chars_used += n;
1386 return chars;
1387}
1388
1389/* Read and decode the next chunk of data from the BufferedReader.
1390 */
1391static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001392textiowrapper_read_chunk(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001393{
1394 PyObject *dec_buffer = NULL;
1395 PyObject *dec_flags = NULL;
1396 PyObject *input_chunk = NULL;
1397 PyObject *decoded_chars, *chunk_size;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001398 Py_ssize_t nbytes, nchars;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001399 int eof;
1400
1401 /* The return value is True unless EOF was reached. The decoded string is
1402 * placed in self._decoded_chars (replacing its previous value). The
1403 * entire input chunk is sent to the decoder, though some of it may remain
1404 * buffered in the decoder, yet to be converted.
1405 */
1406
1407 if (self->decoder == NULL) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001408 _unsupported("not readable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001409 return -1;
1410 }
1411
1412 if (self->telling) {
1413 /* To prepare for tell(), we need to snapshot a point in the file
1414 * where the decoder's input buffer is empty.
1415 */
1416
1417 PyObject *state = PyObject_CallMethodObjArgs(self->decoder,
1418 _PyIO_str_getstate, NULL);
1419 if (state == NULL)
1420 return -1;
1421 /* Given this, we know there was a valid snapshot point
1422 * len(dec_buffer) bytes ago with decoder state (b'', dec_flags).
1423 */
1424 if (PyArg_Parse(state, "(OO)", &dec_buffer, &dec_flags) < 0) {
1425 Py_DECREF(state);
1426 return -1;
1427 }
1428 Py_INCREF(dec_buffer);
1429 Py_INCREF(dec_flags);
1430 Py_DECREF(state);
1431 }
1432
1433 /* Read a chunk, decode it, and put the result in self._decoded_chars. */
1434 chunk_size = PyLong_FromSsize_t(self->chunk_size);
1435 if (chunk_size == NULL)
1436 goto fail;
1437 input_chunk = PyObject_CallMethodObjArgs(self->buffer,
1438 _PyIO_str_read1, chunk_size, NULL);
1439 Py_DECREF(chunk_size);
1440 if (input_chunk == NULL)
1441 goto fail;
1442 assert(PyBytes_Check(input_chunk));
1443
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001444 nbytes = PyBytes_Size(input_chunk);
1445 eof = (nbytes == 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001446
1447 if (Py_TYPE(self->decoder) == &PyIncrementalNewlineDecoder_Type) {
1448 decoded_chars = _PyIncrementalNewlineDecoder_decode(
1449 self->decoder, input_chunk, eof);
1450 }
1451 else {
1452 decoded_chars = PyObject_CallMethodObjArgs(self->decoder,
1453 _PyIO_str_decode, input_chunk, eof ? Py_True : Py_False, NULL);
1454 }
1455
1456 /* TODO sanity check: isinstance(decoded_chars, unicode) */
1457 if (decoded_chars == NULL)
1458 goto fail;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001459 textiowrapper_set_decoded_chars(self, decoded_chars);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001460 nchars = PyUnicode_GET_SIZE(decoded_chars);
1461 if (nchars > 0)
1462 self->b2cratio = (double) nbytes / nchars;
1463 else
1464 self->b2cratio = 0.0;
1465 if (nchars > 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001466 eof = 0;
1467
1468 if (self->telling) {
1469 /* At the snapshot point, len(dec_buffer) bytes before the read, the
1470 * next input to be decoded is dec_buffer + input_chunk.
1471 */
1472 PyObject *next_input = PyNumber_Add(dec_buffer, input_chunk);
1473 if (next_input == NULL)
1474 goto fail;
1475 assert (PyBytes_Check(next_input));
1476 Py_DECREF(dec_buffer);
1477 Py_CLEAR(self->snapshot);
1478 self->snapshot = Py_BuildValue("NN", dec_flags, next_input);
1479 }
1480 Py_DECREF(input_chunk);
1481
1482 return (eof == 0);
1483
1484 fail:
1485 Py_XDECREF(dec_buffer);
1486 Py_XDECREF(dec_flags);
1487 Py_XDECREF(input_chunk);
1488 return -1;
1489}
1490
1491static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001492textiowrapper_read(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001493{
1494 Py_ssize_t n = -1;
1495 PyObject *result = NULL, *chunks = NULL;
1496
1497 CHECK_INITIALIZED(self);
1498
Benjamin Petersonbf5ff762009-12-13 19:25:34 +00001499 if (!PyArg_ParseTuple(args, "|O&:read", &_PyIO_ConvertSsize_t, &n))
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001500 return NULL;
1501
1502 CHECK_CLOSED(self);
1503
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001504 if (self->decoder == NULL)
1505 return _unsupported("not readable");
Benjamin Petersona1b49012009-03-31 23:11:32 +00001506
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001507 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001508 return NULL;
1509
1510 if (n < 0) {
1511 /* Read everything */
1512 PyObject *bytes = PyObject_CallMethod(self->buffer, "read", NULL);
1513 PyObject *decoded;
1514 if (bytes == NULL)
1515 goto fail;
1516 decoded = PyObject_CallMethodObjArgs(self->decoder, _PyIO_str_decode,
1517 bytes, Py_True, NULL);
1518 Py_DECREF(bytes);
1519 if (decoded == NULL)
1520 goto fail;
1521
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001522 result = textiowrapper_get_decoded_chars(self, -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001523
1524 if (result == NULL) {
1525 Py_DECREF(decoded);
1526 return NULL;
1527 }
1528
1529 PyUnicode_AppendAndDel(&result, decoded);
1530 if (result == NULL)
1531 goto fail;
1532
1533 Py_CLEAR(self->snapshot);
1534 return result;
1535 }
1536 else {
1537 int res = 1;
1538 Py_ssize_t remaining = n;
1539
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001540 result = textiowrapper_get_decoded_chars(self, n);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001541 if (result == NULL)
1542 goto fail;
1543 remaining -= PyUnicode_GET_SIZE(result);
1544
1545 /* Keep reading chunks until we have n characters to return */
1546 while (remaining > 0) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001547 res = textiowrapper_read_chunk(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001548 if (res < 0)
1549 goto fail;
1550 if (res == 0) /* EOF */
1551 break;
1552 if (chunks == NULL) {
1553 chunks = PyList_New(0);
1554 if (chunks == NULL)
1555 goto fail;
1556 }
1557 if (PyList_Append(chunks, result) < 0)
1558 goto fail;
1559 Py_DECREF(result);
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001560 result = textiowrapper_get_decoded_chars(self, remaining);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001561 if (result == NULL)
1562 goto fail;
1563 remaining -= PyUnicode_GET_SIZE(result);
1564 }
1565 if (chunks != NULL) {
1566 if (result != NULL && PyList_Append(chunks, result) < 0)
1567 goto fail;
1568 Py_CLEAR(result);
1569 result = PyUnicode_Join(_PyIO_empty_str, chunks);
1570 if (result == NULL)
1571 goto fail;
1572 Py_CLEAR(chunks);
1573 }
1574 return result;
1575 }
1576 fail:
1577 Py_XDECREF(result);
1578 Py_XDECREF(chunks);
1579 return NULL;
1580}
1581
1582
1583/* NOTE: `end` must point to the real end of the Py_UNICODE storage,
1584 that is to the NUL character. Otherwise the function will produce
1585 incorrect results. */
1586static Py_UNICODE *
1587find_control_char(Py_UNICODE *start, Py_UNICODE *end, Py_UNICODE ch)
1588{
1589 Py_UNICODE *s = start;
1590 for (;;) {
1591 while (*s > ch)
1592 s++;
1593 if (*s == ch)
1594 return s;
1595 if (s == end)
1596 return NULL;
1597 s++;
1598 }
1599}
1600
1601Py_ssize_t
1602_PyIO_find_line_ending(
1603 int translated, int universal, PyObject *readnl,
1604 Py_UNICODE *start, Py_UNICODE *end, Py_ssize_t *consumed)
1605{
1606 Py_ssize_t len = end - start;
1607
1608 if (translated) {
1609 /* Newlines are already translated, only search for \n */
1610 Py_UNICODE *pos = find_control_char(start, end, '\n');
1611 if (pos != NULL)
1612 return pos - start + 1;
1613 else {
1614 *consumed = len;
1615 return -1;
1616 }
1617 }
1618 else if (universal) {
1619 /* Universal newline search. Find any of \r, \r\n, \n
1620 * The decoder ensures that \r\n are not split in two pieces
1621 */
1622 Py_UNICODE *s = start;
1623 for (;;) {
1624 Py_UNICODE ch;
1625 /* Fast path for non-control chars. The loop always ends
1626 since the Py_UNICODE storage is NUL-terminated. */
1627 while (*s > '\r')
1628 s++;
1629 if (s >= end) {
1630 *consumed = len;
1631 return -1;
1632 }
1633 ch = *s++;
1634 if (ch == '\n')
1635 return s - start;
1636 if (ch == '\r') {
1637 if (*s == '\n')
1638 return s - start + 1;
1639 else
1640 return s - start;
1641 }
1642 }
1643 }
1644 else {
1645 /* Non-universal mode. */
1646 Py_ssize_t readnl_len = PyUnicode_GET_SIZE(readnl);
1647 Py_UNICODE *nl = PyUnicode_AS_UNICODE(readnl);
1648 if (readnl_len == 1) {
1649 Py_UNICODE *pos = find_control_char(start, end, nl[0]);
1650 if (pos != NULL)
1651 return pos - start + 1;
1652 *consumed = len;
1653 return -1;
1654 }
1655 else {
1656 Py_UNICODE *s = start;
1657 Py_UNICODE *e = end - readnl_len + 1;
1658 Py_UNICODE *pos;
1659 if (e < s)
1660 e = s;
1661 while (s < e) {
1662 Py_ssize_t i;
1663 Py_UNICODE *pos = find_control_char(s, end, nl[0]);
1664 if (pos == NULL || pos >= e)
1665 break;
1666 for (i = 1; i < readnl_len; i++) {
1667 if (pos[i] != nl[i])
1668 break;
1669 }
1670 if (i == readnl_len)
1671 return pos - start + readnl_len;
1672 s = pos + 1;
1673 }
1674 pos = find_control_char(e, end, nl[0]);
1675 if (pos == NULL)
1676 *consumed = len;
1677 else
1678 *consumed = pos - start;
1679 return -1;
1680 }
1681 }
1682}
1683
1684static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001685_textiowrapper_readline(textio *self, Py_ssize_t limit)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001686{
1687 PyObject *line = NULL, *chunks = NULL, *remaining = NULL;
1688 Py_ssize_t start, endpos, chunked, offset_to_buffer;
1689 int res;
1690
1691 CHECK_CLOSED(self);
1692
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001693 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001694 return NULL;
1695
1696 chunked = 0;
1697
1698 while (1) {
1699 Py_UNICODE *ptr;
1700 Py_ssize_t line_len;
1701 Py_ssize_t consumed = 0;
1702
1703 /* First, get some data if necessary */
1704 res = 1;
1705 while (!self->decoded_chars ||
1706 !PyUnicode_GET_SIZE(self->decoded_chars)) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001707 res = textiowrapper_read_chunk(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001708 if (res < 0)
1709 goto error;
1710 if (res == 0)
1711 break;
1712 }
1713 if (res == 0) {
1714 /* end of file */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001715 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001716 Py_CLEAR(self->snapshot);
1717 start = endpos = offset_to_buffer = 0;
1718 break;
1719 }
1720
1721 if (remaining == NULL) {
1722 line = self->decoded_chars;
1723 start = self->decoded_chars_used;
1724 offset_to_buffer = 0;
1725 Py_INCREF(line);
1726 }
1727 else {
1728 assert(self->decoded_chars_used == 0);
1729 line = PyUnicode_Concat(remaining, self->decoded_chars);
1730 start = 0;
1731 offset_to_buffer = PyUnicode_GET_SIZE(remaining);
1732 Py_CLEAR(remaining);
1733 if (line == NULL)
1734 goto error;
1735 }
1736
1737 ptr = PyUnicode_AS_UNICODE(line);
1738 line_len = PyUnicode_GET_SIZE(line);
1739
1740 endpos = _PyIO_find_line_ending(
1741 self->readtranslate, self->readuniversal, self->readnl,
1742 ptr + start, ptr + line_len, &consumed);
1743 if (endpos >= 0) {
1744 endpos += start;
1745 if (limit >= 0 && (endpos - start) + chunked >= limit)
1746 endpos = start + limit - chunked;
1747 break;
1748 }
1749
1750 /* We can put aside up to `endpos` */
1751 endpos = consumed + start;
1752 if (limit >= 0 && (endpos - start) + chunked >= limit) {
1753 /* Didn't find line ending, but reached length limit */
1754 endpos = start + limit - chunked;
1755 break;
1756 }
1757
1758 if (endpos > start) {
1759 /* No line ending seen yet - put aside current data */
1760 PyObject *s;
1761 if (chunks == NULL) {
1762 chunks = PyList_New(0);
1763 if (chunks == NULL)
1764 goto error;
1765 }
1766 s = PyUnicode_FromUnicode(ptr + start, endpos - start);
1767 if (s == NULL)
1768 goto error;
1769 if (PyList_Append(chunks, s) < 0) {
1770 Py_DECREF(s);
1771 goto error;
1772 }
1773 chunked += PyUnicode_GET_SIZE(s);
1774 Py_DECREF(s);
1775 }
1776 /* There may be some remaining bytes we'll have to prepend to the
1777 next chunk of data */
1778 if (endpos < line_len) {
1779 remaining = PyUnicode_FromUnicode(
1780 ptr + endpos, line_len - endpos);
1781 if (remaining == NULL)
1782 goto error;
1783 }
1784 Py_CLEAR(line);
1785 /* We have consumed the buffer */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001786 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001787 }
1788
1789 if (line != NULL) {
1790 /* Our line ends in the current buffer */
1791 self->decoded_chars_used = endpos - offset_to_buffer;
1792 if (start > 0 || endpos < PyUnicode_GET_SIZE(line)) {
1793 if (start == 0 && Py_REFCNT(line) == 1) {
1794 if (PyUnicode_Resize(&line, endpos) < 0)
1795 goto error;
1796 }
1797 else {
1798 PyObject *s = PyUnicode_FromUnicode(
1799 PyUnicode_AS_UNICODE(line) + start, endpos - start);
1800 Py_CLEAR(line);
1801 if (s == NULL)
1802 goto error;
1803 line = s;
1804 }
1805 }
1806 }
1807 if (remaining != NULL) {
1808 if (chunks == NULL) {
1809 chunks = PyList_New(0);
1810 if (chunks == NULL)
1811 goto error;
1812 }
1813 if (PyList_Append(chunks, remaining) < 0)
1814 goto error;
1815 Py_CLEAR(remaining);
1816 }
1817 if (chunks != NULL) {
1818 if (line != NULL && PyList_Append(chunks, line) < 0)
1819 goto error;
1820 Py_CLEAR(line);
1821 line = PyUnicode_Join(_PyIO_empty_str, chunks);
1822 if (line == NULL)
1823 goto error;
1824 Py_DECREF(chunks);
1825 }
1826 if (line == NULL)
1827 line = PyUnicode_FromStringAndSize(NULL, 0);
1828
1829 return line;
1830
1831 error:
1832 Py_XDECREF(chunks);
1833 Py_XDECREF(remaining);
1834 Py_XDECREF(line);
1835 return NULL;
1836}
1837
1838static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001839textiowrapper_readline(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001840{
1841 Py_ssize_t limit = -1;
1842
1843 CHECK_INITIALIZED(self);
1844 if (!PyArg_ParseTuple(args, "|n:readline", &limit)) {
1845 return NULL;
1846 }
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001847 return _textiowrapper_readline(self, limit);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001848}
1849
1850/* Seek and Tell */
1851
1852typedef struct {
1853 Py_off_t start_pos;
1854 int dec_flags;
1855 int bytes_to_feed;
1856 int chars_to_skip;
1857 char need_eof;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001858} cookie_type;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001859
1860/*
1861 To speed up cookie packing/unpacking, we store the fields in a temporary
1862 string and call _PyLong_FromByteArray() or _PyLong_AsByteArray (resp.).
1863 The following macros define at which offsets in the intermediary byte
1864 string the various CookieStruct fields will be stored.
1865 */
1866
1867#define COOKIE_BUF_LEN (sizeof(Py_off_t) + 3 * sizeof(int) + sizeof(char))
1868
1869#if defined(WORDS_BIGENDIAN)
1870
1871# define IS_LITTLE_ENDIAN 0
1872
1873/* We want the least significant byte of start_pos to also be the least
1874 significant byte of the cookie, which means that in big-endian mode we
1875 must copy the fields in reverse order. */
1876
1877# define OFF_START_POS (sizeof(char) + 3 * sizeof(int))
1878# define OFF_DEC_FLAGS (sizeof(char) + 2 * sizeof(int))
1879# define OFF_BYTES_TO_FEED (sizeof(char) + sizeof(int))
1880# define OFF_CHARS_TO_SKIP (sizeof(char))
1881# define OFF_NEED_EOF 0
1882
1883#else
1884
1885# define IS_LITTLE_ENDIAN 1
1886
1887/* Little-endian mode: the least significant byte of start_pos will
1888 naturally end up the least significant byte of the cookie. */
1889
1890# define OFF_START_POS 0
1891# define OFF_DEC_FLAGS (sizeof(Py_off_t))
1892# define OFF_BYTES_TO_FEED (sizeof(Py_off_t) + sizeof(int))
1893# define OFF_CHARS_TO_SKIP (sizeof(Py_off_t) + 2 * sizeof(int))
1894# define OFF_NEED_EOF (sizeof(Py_off_t) + 3 * sizeof(int))
1895
1896#endif
1897
1898static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001899textiowrapper_parse_cookie(cookie_type *cookie, PyObject *cookieObj)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001900{
1901 unsigned char buffer[COOKIE_BUF_LEN];
1902 PyLongObject *cookieLong = (PyLongObject *)PyNumber_Long(cookieObj);
1903 if (cookieLong == NULL)
1904 return -1;
1905
1906 if (_PyLong_AsByteArray(cookieLong, buffer, sizeof(buffer),
1907 IS_LITTLE_ENDIAN, 0) < 0) {
1908 Py_DECREF(cookieLong);
1909 return -1;
1910 }
1911 Py_DECREF(cookieLong);
1912
Antoine Pitrou2db74c22009-03-06 21:49:02 +00001913 memcpy(&cookie->start_pos, buffer + OFF_START_POS, sizeof(cookie->start_pos));
1914 memcpy(&cookie->dec_flags, buffer + OFF_DEC_FLAGS, sizeof(cookie->dec_flags));
1915 memcpy(&cookie->bytes_to_feed, buffer + OFF_BYTES_TO_FEED, sizeof(cookie->bytes_to_feed));
1916 memcpy(&cookie->chars_to_skip, buffer + OFF_CHARS_TO_SKIP, sizeof(cookie->chars_to_skip));
1917 memcpy(&cookie->need_eof, buffer + OFF_NEED_EOF, sizeof(cookie->need_eof));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001918
1919 return 0;
1920}
1921
1922static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001923textiowrapper_build_cookie(cookie_type *cookie)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001924{
1925 unsigned char buffer[COOKIE_BUF_LEN];
1926
Antoine Pitrou2db74c22009-03-06 21:49:02 +00001927 memcpy(buffer + OFF_START_POS, &cookie->start_pos, sizeof(cookie->start_pos));
1928 memcpy(buffer + OFF_DEC_FLAGS, &cookie->dec_flags, sizeof(cookie->dec_flags));
1929 memcpy(buffer + OFF_BYTES_TO_FEED, &cookie->bytes_to_feed, sizeof(cookie->bytes_to_feed));
1930 memcpy(buffer + OFF_CHARS_TO_SKIP, &cookie->chars_to_skip, sizeof(cookie->chars_to_skip));
1931 memcpy(buffer + OFF_NEED_EOF, &cookie->need_eof, sizeof(cookie->need_eof));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001932
1933 return _PyLong_FromByteArray(buffer, sizeof(buffer), IS_LITTLE_ENDIAN, 0);
1934}
1935#undef IS_LITTLE_ENDIAN
1936
1937static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001938_textiowrapper_decoder_setstate(textio *self, cookie_type *cookie)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001939{
1940 PyObject *res;
1941 /* When seeking to the start of the stream, we call decoder.reset()
1942 rather than decoder.getstate().
1943 This is for a few decoders such as utf-16 for which the state value
1944 at start is not (b"", 0) but e.g. (b"", 2) (meaning, in the case of
1945 utf-16, that we are expecting a BOM).
1946 */
1947 if (cookie->start_pos == 0 && cookie->dec_flags == 0)
1948 res = PyObject_CallMethodObjArgs(self->decoder, _PyIO_str_reset, NULL);
1949 else
1950 res = PyObject_CallMethod(self->decoder, "setstate",
1951 "((yi))", "", cookie->dec_flags);
1952 if (res == NULL)
1953 return -1;
1954 Py_DECREF(res);
1955 return 0;
1956}
1957
Antoine Pitroue4501852009-05-14 18:55:55 +00001958static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001959_textiowrapper_encoder_setstate(textio *self, cookie_type *cookie)
Antoine Pitroue4501852009-05-14 18:55:55 +00001960{
1961 PyObject *res;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001962 /* Same as _textiowrapper_decoder_setstate() above. */
Antoine Pitroue4501852009-05-14 18:55:55 +00001963 if (cookie->start_pos == 0 && cookie->dec_flags == 0) {
1964 res = PyObject_CallMethodObjArgs(self->encoder, _PyIO_str_reset, NULL);
1965 self->encoding_start_of_stream = 1;
1966 }
1967 else {
1968 res = PyObject_CallMethodObjArgs(self->encoder, _PyIO_str_setstate,
1969 _PyIO_zero, NULL);
1970 self->encoding_start_of_stream = 0;
1971 }
1972 if (res == NULL)
1973 return -1;
1974 Py_DECREF(res);
1975 return 0;
1976}
1977
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001978static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001979textiowrapper_seek(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001980{
1981 PyObject *cookieObj, *posobj;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001982 cookie_type cookie;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001983 int whence = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001984 PyObject *res;
1985 int cmp;
1986
1987 CHECK_INITIALIZED(self);
1988
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001989 if (!PyArg_ParseTuple(args, "O|i:seek", &cookieObj, &whence))
1990 return NULL;
1991 CHECK_CLOSED(self);
1992
1993 Py_INCREF(cookieObj);
1994
1995 if (!self->seekable) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001996 _unsupported("underlying stream is not seekable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001997 goto fail;
1998 }
1999
2000 if (whence == 1) {
2001 /* seek relative to current position */
Antoine Pitroue4501852009-05-14 18:55:55 +00002002 cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_EQ);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002003 if (cmp < 0)
2004 goto fail;
2005
2006 if (cmp == 0) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002007 _unsupported("can't do nonzero cur-relative seeks");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002008 goto fail;
2009 }
2010
2011 /* Seeking to the current position should attempt to
2012 * sync the underlying buffer with the current position.
2013 */
2014 Py_DECREF(cookieObj);
2015 cookieObj = PyObject_CallMethod((PyObject *)self, "tell", NULL);
2016 if (cookieObj == NULL)
2017 goto fail;
2018 }
2019 else if (whence == 2) {
2020 /* seek relative to end of file */
2021
Antoine Pitroue4501852009-05-14 18:55:55 +00002022 cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_EQ);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002023 if (cmp < 0)
2024 goto fail;
2025
2026 if (cmp == 0) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002027 _unsupported("can't do nonzero end-relative seeks");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002028 goto fail;
2029 }
2030
2031 res = PyObject_CallMethod((PyObject *)self, "flush", NULL);
2032 if (res == NULL)
2033 goto fail;
2034 Py_DECREF(res);
2035
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002036 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002037 Py_CLEAR(self->snapshot);
2038 if (self->decoder) {
2039 res = PyObject_CallMethod(self->decoder, "reset", NULL);
2040 if (res == NULL)
2041 goto fail;
2042 Py_DECREF(res);
2043 }
2044
2045 res = PyObject_CallMethod(self->buffer, "seek", "ii", 0, 2);
2046 Py_XDECREF(cookieObj);
2047 return res;
2048 }
2049 else if (whence != 0) {
2050 PyErr_Format(PyExc_ValueError,
2051 "invalid whence (%d, should be 0, 1 or 2)", whence);
2052 goto fail;
2053 }
2054
Antoine Pitroue4501852009-05-14 18:55:55 +00002055 cmp = PyObject_RichCompareBool(cookieObj, _PyIO_zero, Py_LT);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002056 if (cmp < 0)
2057 goto fail;
2058
2059 if (cmp == 1) {
2060 PyErr_Format(PyExc_ValueError,
2061 "negative seek position %R", cookieObj);
2062 goto fail;
2063 }
2064
2065 res = PyObject_CallMethodObjArgs((PyObject *)self, _PyIO_str_flush, NULL);
2066 if (res == NULL)
2067 goto fail;
2068 Py_DECREF(res);
2069
2070 /* The strategy of seek() is to go back to the safe start point
2071 * and replay the effect of read(chars_to_skip) from there.
2072 */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002073 if (textiowrapper_parse_cookie(&cookie, cookieObj) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002074 goto fail;
2075
2076 /* Seek back to the safe start point. */
2077 posobj = PyLong_FromOff_t(cookie.start_pos);
2078 if (posobj == NULL)
2079 goto fail;
2080 res = PyObject_CallMethodObjArgs(self->buffer,
2081 _PyIO_str_seek, posobj, NULL);
2082 Py_DECREF(posobj);
2083 if (res == NULL)
2084 goto fail;
2085 Py_DECREF(res);
2086
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002087 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002088 Py_CLEAR(self->snapshot);
2089
2090 /* Restore the decoder to its state from the safe start point. */
2091 if (self->decoder) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002092 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002093 goto fail;
2094 }
2095
2096 if (cookie.chars_to_skip) {
2097 /* Just like _read_chunk, feed the decoder and save a snapshot. */
2098 PyObject *input_chunk = PyObject_CallMethod(
2099 self->buffer, "read", "i", cookie.bytes_to_feed);
2100 PyObject *decoded;
2101
2102 if (input_chunk == NULL)
2103 goto fail;
2104
2105 assert (PyBytes_Check(input_chunk));
2106
2107 self->snapshot = Py_BuildValue("iN", cookie.dec_flags, input_chunk);
2108 if (self->snapshot == NULL) {
2109 Py_DECREF(input_chunk);
2110 goto fail;
2111 }
2112
2113 decoded = PyObject_CallMethod(self->decoder, "decode",
2114 "Oi", input_chunk, (int)cookie.need_eof);
2115
2116 if (decoded == NULL)
2117 goto fail;
2118
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002119 textiowrapper_set_decoded_chars(self, decoded);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002120
2121 /* Skip chars_to_skip of the decoded characters. */
2122 if (PyUnicode_GetSize(self->decoded_chars) < cookie.chars_to_skip) {
2123 PyErr_SetString(PyExc_IOError, "can't restore logical file position");
2124 goto fail;
2125 }
2126 self->decoded_chars_used = cookie.chars_to_skip;
2127 }
2128 else {
2129 self->snapshot = Py_BuildValue("iy", cookie.dec_flags, "");
2130 if (self->snapshot == NULL)
2131 goto fail;
2132 }
2133
Antoine Pitroue4501852009-05-14 18:55:55 +00002134 /* Finally, reset the encoder (merely useful for proper BOM handling) */
2135 if (self->encoder) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002136 if (_textiowrapper_encoder_setstate(self, &cookie) < 0)
Antoine Pitroue4501852009-05-14 18:55:55 +00002137 goto fail;
2138 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002139 return cookieObj;
2140 fail:
2141 Py_XDECREF(cookieObj);
2142 return NULL;
2143
2144}
2145
2146static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002147textiowrapper_tell(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002148{
2149 PyObject *res;
2150 PyObject *posobj = NULL;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002151 cookie_type cookie = {0,0,0,0,0};
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002152 PyObject *next_input;
2153 Py_ssize_t chars_to_skip, chars_decoded;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002154 Py_ssize_t skip_bytes, skip_back;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002155 PyObject *saved_state = NULL;
2156 char *input, *input_end;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002157 char *dec_buffer;
2158 Py_ssize_t dec_buffer_len;
2159 int dec_flags;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002160
2161 CHECK_INITIALIZED(self);
2162 CHECK_CLOSED(self);
2163
2164 if (!self->seekable) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002165 _unsupported("underlying stream is not seekable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002166 goto fail;
2167 }
2168 if (!self->telling) {
2169 PyErr_SetString(PyExc_IOError,
2170 "telling position disabled by next() call");
2171 goto fail;
2172 }
2173
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002174 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002175 return NULL;
2176 res = PyObject_CallMethod((PyObject *)self, "flush", NULL);
2177 if (res == NULL)
2178 goto fail;
2179 Py_DECREF(res);
2180
2181 posobj = PyObject_CallMethod(self->buffer, "tell", NULL);
2182 if (posobj == NULL)
2183 goto fail;
2184
2185 if (self->decoder == NULL || self->snapshot == NULL) {
2186 assert (self->decoded_chars == NULL || PyUnicode_GetSize(self->decoded_chars) == 0);
2187 return posobj;
2188 }
2189
2190#if defined(HAVE_LARGEFILE_SUPPORT)
2191 cookie.start_pos = PyLong_AsLongLong(posobj);
2192#else
2193 cookie.start_pos = PyLong_AsLong(posobj);
2194#endif
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002195 Py_DECREF(posobj);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002196 if (PyErr_Occurred())
2197 goto fail;
2198
2199 /* Skip backward to the snapshot point (see _read_chunk). */
2200 if (!PyArg_Parse(self->snapshot, "(iO)", &cookie.dec_flags, &next_input))
2201 goto fail;
2202
2203 assert (PyBytes_Check(next_input));
2204
2205 cookie.start_pos -= PyBytes_GET_SIZE(next_input);
2206
2207 /* How many decoded characters have been used up since the snapshot? */
2208 if (self->decoded_chars_used == 0) {
2209 /* We haven't moved from the snapshot point. */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002210 return textiowrapper_build_cookie(&cookie);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002211 }
2212
2213 chars_to_skip = self->decoded_chars_used;
2214
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002215 /* Decoder state will be restored at the end */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002216 saved_state = PyObject_CallMethodObjArgs(self->decoder,
2217 _PyIO_str_getstate, NULL);
2218 if (saved_state == NULL)
2219 goto fail;
2220
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002221#define DECODER_GETSTATE() do { \
2222 PyObject *_state = PyObject_CallMethodObjArgs(self->decoder, \
2223 _PyIO_str_getstate, NULL); \
2224 if (_state == NULL) \
2225 goto fail; \
2226 if (!PyArg_Parse(_state, "(y#i)", &dec_buffer, &dec_buffer_len, &dec_flags)) { \
2227 Py_DECREF(_state); \
2228 goto fail; \
2229 } \
2230 Py_DECREF(_state); \
2231 } while (0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002232
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002233 /* TODO: replace assert with exception */
2234#define DECODER_DECODE(start, len, res) do { \
2235 PyObject *_decoded = PyObject_CallMethod( \
2236 self->decoder, "decode", "y#", start, len); \
2237 if (_decoded == NULL) \
2238 goto fail; \
2239 assert (PyUnicode_Check(_decoded)); \
2240 res = PyUnicode_GET_SIZE(_decoded); \
2241 Py_DECREF(_decoded); \
2242 } while (0)
2243
2244 /* Fast search for an acceptable start point, close to our
2245 current pos */
2246 skip_bytes = (Py_ssize_t) (self->b2cratio * chars_to_skip);
2247 skip_back = 1;
2248 assert(skip_back <= PyBytes_GET_SIZE(next_input));
2249 input = PyBytes_AS_STRING(next_input);
2250 while (skip_bytes > 0) {
2251 /* Decode up to temptative start point */
2252 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
2253 goto fail;
2254 DECODER_DECODE(input, skip_bytes, chars_decoded);
2255 if (chars_decoded <= chars_to_skip) {
2256 DECODER_GETSTATE();
2257 if (dec_buffer_len == 0) {
2258 /* Before pos and no bytes buffered in decoder => OK */
2259 cookie.dec_flags = dec_flags;
2260 chars_to_skip -= chars_decoded;
2261 break;
2262 }
2263 /* Skip back by buffered amount and reset heuristic */
2264 skip_bytes -= dec_buffer_len;
2265 skip_back = 1;
2266 }
2267 else {
2268 /* We're too far ahead, skip back a bit */
2269 skip_bytes -= skip_back;
2270 skip_back *= 2;
2271 }
2272 }
2273 if (skip_bytes <= 0) {
2274 skip_bytes = 0;
2275 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
2276 goto fail;
2277 }
2278
2279 /* Note our initial start point. */
2280 cookie.start_pos += skip_bytes;
2281 cookie.chars_to_skip = chars_to_skip;
2282 if (chars_to_skip == 0)
2283 goto finally;
2284
2285 /* We should be close to the desired position. Now feed the decoder one
2286 * byte at a time until we reach the `chars_to_skip` target.
2287 * As we go, note the nearest "safe start point" before the current
2288 * location (a point where the decoder has nothing buffered, so seek()
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002289 * can safely start from there and advance to this location).
2290 */
2291 chars_decoded = 0;
2292 input = PyBytes_AS_STRING(next_input);
2293 input_end = input + PyBytes_GET_SIZE(next_input);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002294 input += skip_bytes;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002295 while (input < input_end) {
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002296 Py_ssize_t n;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002297
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002298 DECODER_DECODE(input, 1, n);
2299 /* We got n chars for 1 byte */
2300 chars_decoded += n;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002301 cookie.bytes_to_feed += 1;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002302 DECODER_GETSTATE();
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002303
2304 if (dec_buffer_len == 0 && chars_decoded <= chars_to_skip) {
2305 /* Decoder buffer is empty, so this is a safe start point. */
2306 cookie.start_pos += cookie.bytes_to_feed;
2307 chars_to_skip -= chars_decoded;
2308 cookie.dec_flags = dec_flags;
2309 cookie.bytes_to_feed = 0;
2310 chars_decoded = 0;
2311 }
2312 if (chars_decoded >= chars_to_skip)
2313 break;
2314 input++;
2315 }
2316 if (input == input_end) {
2317 /* We didn't get enough decoded data; signal EOF to get more. */
2318 PyObject *decoded = PyObject_CallMethod(
2319 self->decoder, "decode", "yi", "", /* final = */ 1);
2320 if (decoded == NULL)
2321 goto fail;
2322 assert (PyUnicode_Check(decoded));
2323 chars_decoded += PyUnicode_GET_SIZE(decoded);
2324 Py_DECREF(decoded);
2325 cookie.need_eof = 1;
2326
2327 if (chars_decoded < chars_to_skip) {
2328 PyErr_SetString(PyExc_IOError,
2329 "can't reconstruct logical file position");
2330 goto fail;
2331 }
2332 }
2333
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002334finally:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002335 res = PyObject_CallMethod(self->decoder, "setstate", "(O)", saved_state);
2336 Py_DECREF(saved_state);
2337 if (res == NULL)
2338 return NULL;
2339 Py_DECREF(res);
2340
2341 /* The returned cookie corresponds to the last safe start point. */
2342 cookie.chars_to_skip = Py_SAFE_DOWNCAST(chars_to_skip, Py_ssize_t, int);
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002343 return textiowrapper_build_cookie(&cookie);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002344
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002345fail:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002346 if (saved_state) {
2347 PyObject *type, *value, *traceback;
2348 PyErr_Fetch(&type, &value, &traceback);
2349
2350 res = PyObject_CallMethod(self->decoder, "setstate", "(O)", saved_state);
2351 Py_DECREF(saved_state);
2352 if (res == NULL)
2353 return NULL;
2354 Py_DECREF(res);
2355
2356 PyErr_Restore(type, value, traceback);
2357 }
2358 return NULL;
2359}
2360
2361static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002362textiowrapper_truncate(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002363{
2364 PyObject *pos = Py_None;
2365 PyObject *res;
2366
2367 CHECK_INITIALIZED(self)
2368 if (!PyArg_ParseTuple(args, "|O:truncate", &pos)) {
2369 return NULL;
2370 }
2371
2372 res = PyObject_CallMethodObjArgs((PyObject *) self, _PyIO_str_flush, NULL);
2373 if (res == NULL)
2374 return NULL;
2375 Py_DECREF(res);
2376
Antoine Pitrou905a2ff2010-01-31 22:47:27 +00002377 return PyObject_CallMethodObjArgs(self->buffer, _PyIO_str_truncate, pos, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002378}
2379
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002380static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002381textiowrapper_repr(textio *self)
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002382{
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002383 PyObject *nameobj, *modeobj, *res, *s;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002384
2385 CHECK_INITIALIZED(self);
2386
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002387 res = PyUnicode_FromString("<_io.TextIOWrapper");
2388 if (res == NULL)
2389 return NULL;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002390 nameobj = PyObject_GetAttrString((PyObject *) self, "name");
2391 if (nameobj == NULL) {
2392 if (PyErr_ExceptionMatches(PyExc_AttributeError))
2393 PyErr_Clear();
2394 else
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002395 goto error;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002396 }
2397 else {
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002398 s = PyUnicode_FromFormat(" name=%R", nameobj);
Antoine Pitrou716c4442009-05-23 19:04:03 +00002399 Py_DECREF(nameobj);
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002400 if (s == NULL)
2401 goto error;
2402 PyUnicode_AppendAndDel(&res, s);
2403 if (res == NULL)
2404 return NULL;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002405 }
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002406 modeobj = PyObject_GetAttrString((PyObject *) self, "mode");
2407 if (modeobj == NULL) {
2408 if (PyErr_ExceptionMatches(PyExc_AttributeError))
2409 PyErr_Clear();
2410 else
2411 goto error;
2412 }
2413 else {
2414 s = PyUnicode_FromFormat(" mode=%R", modeobj);
2415 Py_DECREF(modeobj);
2416 if (s == NULL)
2417 goto error;
2418 PyUnicode_AppendAndDel(&res, s);
2419 if (res == NULL)
2420 return NULL;
2421 }
2422 s = PyUnicode_FromFormat("%U encoding=%R>",
2423 res, self->encoding);
2424 Py_DECREF(res);
2425 return s;
2426error:
2427 Py_XDECREF(res);
2428 return NULL;
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002429}
2430
2431
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002432/* Inquiries */
2433
2434static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002435textiowrapper_fileno(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002436{
2437 CHECK_INITIALIZED(self);
2438 return PyObject_CallMethod(self->buffer, "fileno", NULL);
2439}
2440
2441static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002442textiowrapper_seekable(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002443{
2444 CHECK_INITIALIZED(self);
2445 return PyObject_CallMethod(self->buffer, "seekable", NULL);
2446}
2447
2448static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002449textiowrapper_readable(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002450{
2451 CHECK_INITIALIZED(self);
2452 return PyObject_CallMethod(self->buffer, "readable", NULL);
2453}
2454
2455static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002456textiowrapper_writable(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002457{
2458 CHECK_INITIALIZED(self);
2459 return PyObject_CallMethod(self->buffer, "writable", NULL);
2460}
2461
2462static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002463textiowrapper_isatty(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002464{
2465 CHECK_INITIALIZED(self);
2466 return PyObject_CallMethod(self->buffer, "isatty", NULL);
2467}
2468
2469static PyObject *
Antoine Pitrou243757e2010-11-05 21:15:39 +00002470textiowrapper_getstate(textio *self, PyObject *args)
2471{
2472 PyErr_Format(PyExc_TypeError,
2473 "cannot serialize '%s' object", Py_TYPE(self)->tp_name);
2474 return NULL;
2475}
2476
2477static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002478textiowrapper_flush(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002479{
2480 CHECK_INITIALIZED(self);
2481 CHECK_CLOSED(self);
2482 self->telling = self->seekable;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002483 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002484 return NULL;
2485 return PyObject_CallMethod(self->buffer, "flush", NULL);
2486}
2487
2488static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002489textiowrapper_close(textio *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002490{
2491 PyObject *res;
Antoine Pitrou6be88762010-05-03 16:48:20 +00002492 int r;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002493 CHECK_INITIALIZED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002494
Antoine Pitrou6be88762010-05-03 16:48:20 +00002495 res = textiowrapper_closed_get(self, NULL);
2496 if (res == NULL)
2497 return NULL;
2498 r = PyObject_IsTrue(res);
2499 Py_DECREF(res);
2500 if (r < 0)
2501 return NULL;
Victor Stinnerf6c57832010-05-19 01:17:01 +00002502
Antoine Pitrou6be88762010-05-03 16:48:20 +00002503 if (r > 0) {
2504 Py_RETURN_NONE; /* stream already closed */
2505 }
2506 else {
Antoine Pitroue033e062010-10-29 10:38:18 +00002507 if (self->deallocating) {
2508 res = PyObject_CallMethod(self->buffer, "_dealloc_warn", "O", self);
2509 if (res)
2510 Py_DECREF(res);
2511 else
2512 PyErr_Clear();
2513 }
Antoine Pitrou6be88762010-05-03 16:48:20 +00002514 res = PyObject_CallMethod((PyObject *)self, "flush", NULL);
2515 if (res == NULL) {
2516 return NULL;
2517 }
2518 else
2519 Py_DECREF(res);
2520
2521 return PyObject_CallMethod(self->buffer, "close", NULL);
2522 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002523}
2524
2525static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002526textiowrapper_iternext(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002527{
2528 PyObject *line;
2529
2530 CHECK_INITIALIZED(self);
2531
2532 self->telling = 0;
2533 if (Py_TYPE(self) == &PyTextIOWrapper_Type) {
2534 /* Skip method call overhead for speed */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002535 line = _textiowrapper_readline(self, -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002536 }
2537 else {
2538 line = PyObject_CallMethodObjArgs((PyObject *)self,
2539 _PyIO_str_readline, NULL);
2540 if (line && !PyUnicode_Check(line)) {
2541 PyErr_Format(PyExc_IOError,
2542 "readline() should have returned an str object, "
2543 "not '%.200s'", Py_TYPE(line)->tp_name);
2544 Py_DECREF(line);
2545 return NULL;
2546 }
2547 }
2548
2549 if (line == NULL)
2550 return NULL;
2551
2552 if (PyUnicode_GET_SIZE(line) == 0) {
2553 /* Reached EOF or would have blocked */
2554 Py_DECREF(line);
2555 Py_CLEAR(self->snapshot);
2556 self->telling = self->seekable;
2557 return NULL;
2558 }
2559
2560 return line;
2561}
2562
2563static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002564textiowrapper_name_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002565{
2566 CHECK_INITIALIZED(self);
2567 return PyObject_GetAttrString(self->buffer, "name");
2568}
2569
2570static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002571textiowrapper_closed_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002572{
2573 CHECK_INITIALIZED(self);
2574 return PyObject_GetAttr(self->buffer, _PyIO_str_closed);
2575}
2576
2577static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002578textiowrapper_newlines_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002579{
2580 PyObject *res;
2581 CHECK_INITIALIZED(self);
2582 if (self->decoder == NULL)
2583 Py_RETURN_NONE;
2584 res = PyObject_GetAttr(self->decoder, _PyIO_str_newlines);
2585 if (res == NULL) {
Benjamin Peterson2cfca792009-06-06 20:46:48 +00002586 if (PyErr_ExceptionMatches(PyExc_AttributeError)) {
2587 PyErr_Clear();
2588 Py_RETURN_NONE;
2589 }
2590 else {
2591 return NULL;
2592 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002593 }
2594 return res;
2595}
2596
2597static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002598textiowrapper_errors_get(textio *self, void *context)
Benjamin Peterson0926ad12009-06-06 18:02:12 +00002599{
2600 CHECK_INITIALIZED(self);
2601 return PyUnicode_FromString(PyBytes_AS_STRING(self->errors));
2602}
2603
2604static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002605textiowrapper_chunk_size_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002606{
2607 CHECK_INITIALIZED(self);
2608 return PyLong_FromSsize_t(self->chunk_size);
2609}
2610
2611static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002612textiowrapper_chunk_size_set(textio *self, PyObject *arg, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002613{
2614 Py_ssize_t n;
2615 CHECK_INITIALIZED_INT(self);
2616 n = PyNumber_AsSsize_t(arg, PyExc_TypeError);
2617 if (n == -1 && PyErr_Occurred())
2618 return -1;
2619 if (n <= 0) {
2620 PyErr_SetString(PyExc_ValueError,
2621 "a strictly positive integer is required");
2622 return -1;
2623 }
2624 self->chunk_size = n;
2625 return 0;
2626}
2627
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002628static PyMethodDef textiowrapper_methods[] = {
2629 {"detach", (PyCFunction)textiowrapper_detach, METH_NOARGS},
2630 {"write", (PyCFunction)textiowrapper_write, METH_VARARGS},
2631 {"read", (PyCFunction)textiowrapper_read, METH_VARARGS},
2632 {"readline", (PyCFunction)textiowrapper_readline, METH_VARARGS},
2633 {"flush", (PyCFunction)textiowrapper_flush, METH_NOARGS},
2634 {"close", (PyCFunction)textiowrapper_close, METH_NOARGS},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002635
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002636 {"fileno", (PyCFunction)textiowrapper_fileno, METH_NOARGS},
2637 {"seekable", (PyCFunction)textiowrapper_seekable, METH_NOARGS},
2638 {"readable", (PyCFunction)textiowrapper_readable, METH_NOARGS},
2639 {"writable", (PyCFunction)textiowrapper_writable, METH_NOARGS},
2640 {"isatty", (PyCFunction)textiowrapper_isatty, METH_NOARGS},
Antoine Pitrou243757e2010-11-05 21:15:39 +00002641 {"__getstate__", (PyCFunction)textiowrapper_getstate, METH_NOARGS},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002642
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002643 {"seek", (PyCFunction)textiowrapper_seek, METH_VARARGS},
2644 {"tell", (PyCFunction)textiowrapper_tell, METH_NOARGS},
2645 {"truncate", (PyCFunction)textiowrapper_truncate, METH_VARARGS},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002646 {NULL, NULL}
2647};
2648
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002649static PyMemberDef textiowrapper_members[] = {
2650 {"encoding", T_OBJECT, offsetof(textio, encoding), READONLY},
2651 {"buffer", T_OBJECT, offsetof(textio, buffer), READONLY},
2652 {"line_buffering", T_BOOL, offsetof(textio, line_buffering), READONLY},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002653 {NULL}
2654};
2655
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002656static PyGetSetDef textiowrapper_getset[] = {
2657 {"name", (getter)textiowrapper_name_get, NULL, NULL},
2658 {"closed", (getter)textiowrapper_closed_get, NULL, NULL},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002659/* {"mode", (getter)TextIOWrapper_mode_get, NULL, NULL},
2660*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002661 {"newlines", (getter)textiowrapper_newlines_get, NULL, NULL},
2662 {"errors", (getter)textiowrapper_errors_get, NULL, NULL},
2663 {"_CHUNK_SIZE", (getter)textiowrapper_chunk_size_get,
2664 (setter)textiowrapper_chunk_size_set, NULL},
Benjamin Peterson1fea3212009-04-19 03:15:20 +00002665 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002666};
2667
2668PyTypeObject PyTextIOWrapper_Type = {
2669 PyVarObject_HEAD_INIT(NULL, 0)
2670 "_io.TextIOWrapper", /*tp_name*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002671 sizeof(textio), /*tp_basicsize*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002672 0, /*tp_itemsize*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002673 (destructor)textiowrapper_dealloc, /*tp_dealloc*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002674 0, /*tp_print*/
2675 0, /*tp_getattr*/
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002676 0, /*tps_etattr*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002677 0, /*tp_compare */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002678 (reprfunc)textiowrapper_repr,/*tp_repr*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002679 0, /*tp_as_number*/
2680 0, /*tp_as_sequence*/
2681 0, /*tp_as_mapping*/
2682 0, /*tp_hash */
2683 0, /*tp_call*/
2684 0, /*tp_str*/
2685 0, /*tp_getattro*/
2686 0, /*tp_setattro*/
2687 0, /*tp_as_buffer*/
2688 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE
2689 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002690 textiowrapper_doc, /* tp_doc */
2691 (traverseproc)textiowrapper_traverse, /* tp_traverse */
2692 (inquiry)textiowrapper_clear, /* tp_clear */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002693 0, /* tp_richcompare */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002694 offsetof(textio, weakreflist), /*tp_weaklistoffset*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002695 0, /* tp_iter */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002696 (iternextfunc)textiowrapper_iternext, /* tp_iternext */
2697 textiowrapper_methods, /* tp_methods */
2698 textiowrapper_members, /* tp_members */
2699 textiowrapper_getset, /* tp_getset */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002700 0, /* tp_base */
2701 0, /* tp_dict */
2702 0, /* tp_descr_get */
2703 0, /* tp_descr_set */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002704 offsetof(textio, dict), /*tp_dictoffset*/
2705 (initproc)textiowrapper_init, /* tp_init */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002706 0, /* tp_alloc */
2707 PyType_GenericNew, /* tp_new */
2708};