blob: f08d14e18b4021870a7c5d89113e86d2f52cc1f5 [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"
Victor Stinner4a21e572020-04-15 02:35:41 +020011#include "pycore_interp.h" // PyInterpreterState.fs_codec
Victor Stinner37834132020-10-27 17:12:53 +010012#include "pycore_long.h" // _PyLong_GetZero()
Victor Stinner710e8262020-10-31 01:02:09 +010013#include "pycore_fileutils.h" // _Py_GetLocaleEncoding()
Victor Stinnerbcda8f12018-11-21 22:27:47 +010014#include "pycore_object.h"
Victor Stinner4a21e572020-04-15 02:35:41 +020015#include "pycore_pystate.h" // _PyInterpreterState_GET()
16#include "structmember.h" // PyMemberDef
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000017#include "_iomodule.h"
18
Serhiy Storchakaf24131f2015-04-16 11:19:43 +030019/*[clinic input]
20module _io
21class _io.IncrementalNewlineDecoder "nldecoder_object *" "&PyIncrementalNewlineDecoder_Type"
22class _io.TextIOWrapper "textio *" "&TextIOWrapper_TYpe"
23[clinic start generated code]*/
24/*[clinic end generated code: output=da39a3ee5e6b4b0d input=2097a4fc85670c26]*/
25
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020026_Py_IDENTIFIER(close);
27_Py_IDENTIFIER(_dealloc_warn);
28_Py_IDENTIFIER(decode);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020029_Py_IDENTIFIER(fileno);
30_Py_IDENTIFIER(flush);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020031_Py_IDENTIFIER(isatty);
Martin v. Löwis767046a2011-10-14 15:35:36 +020032_Py_IDENTIFIER(mode);
33_Py_IDENTIFIER(name);
34_Py_IDENTIFIER(raw);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020035_Py_IDENTIFIER(read);
36_Py_IDENTIFIER(readable);
37_Py_IDENTIFIER(replace);
38_Py_IDENTIFIER(reset);
39_Py_IDENTIFIER(seek);
40_Py_IDENTIFIER(seekable);
41_Py_IDENTIFIER(setstate);
INADA Naoki507434f2017-12-21 09:59:53 +090042_Py_IDENTIFIER(strict);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020043_Py_IDENTIFIER(tell);
44_Py_IDENTIFIER(writable);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +020045
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000046/* TextIOBase */
47
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000048PyDoc_STRVAR(textiobase_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000049 "Base class for text I/O.\n"
50 "\n"
51 "This class provides a character and line based interface to stream\n"
52 "I/O. There is no readinto method because Python's character strings\n"
53 "are immutable. There is no public constructor.\n"
54 );
55
56static PyObject *
57_unsupported(const char *message)
58{
Antoine Pitrou712cb732013-12-21 15:51:54 +010059 _PyIO_State *state = IO_STATE();
60 if (state != NULL)
61 PyErr_SetString(state->unsupported_operation, message);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000062 return NULL;
63}
64
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000065PyDoc_STRVAR(textiobase_detach_doc,
Benjamin Petersond2e0c792009-05-01 20:40:59 +000066 "Separate the underlying buffer from the TextIOBase and return it.\n"
67 "\n"
68 "After the underlying buffer has been detached, the TextIO is in an\n"
69 "unusable state.\n"
70 );
71
72static PyObject *
Siddhesh Poyarekar55edd0c2018-04-30 00:29:33 +053073textiobase_detach(PyObject *self, PyObject *Py_UNUSED(ignored))
Benjamin Petersond2e0c792009-05-01 20:40:59 +000074{
75 return _unsupported("detach");
76}
77
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000078PyDoc_STRVAR(textiobase_read_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000079 "Read at most n characters from stream.\n"
80 "\n"
81 "Read from underlying buffer until we have n characters or we hit EOF.\n"
82 "If n is negative or omitted, read until EOF.\n"
83 );
84
85static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000086textiobase_read(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000087{
88 return _unsupported("read");
89}
90
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000091PyDoc_STRVAR(textiobase_readline_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000092 "Read until newline or EOF.\n"
93 "\n"
94 "Returns an empty string if EOF is hit immediately.\n"
95 );
96
97static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000098textiobase_readline(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000099{
100 return _unsupported("readline");
101}
102
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000103PyDoc_STRVAR(textiobase_write_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000104 "Write string to stream.\n"
105 "Returns the number of characters written (which is always equal to\n"
106 "the length of the string).\n"
107 );
108
109static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000110textiobase_write(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000111{
112 return _unsupported("write");
113}
114
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000115PyDoc_STRVAR(textiobase_encoding_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000116 "Encoding of the text stream.\n"
117 "\n"
118 "Subclasses should override.\n"
119 );
120
121static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000122textiobase_encoding_get(PyObject *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000123{
124 Py_RETURN_NONE;
125}
126
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000127PyDoc_STRVAR(textiobase_newlines_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000128 "Line endings translated so far.\n"
129 "\n"
130 "Only line endings translated during reading are considered.\n"
131 "\n"
132 "Subclasses should override.\n"
133 );
134
135static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000136textiobase_newlines_get(PyObject *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000137{
138 Py_RETURN_NONE;
139}
140
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000141PyDoc_STRVAR(textiobase_errors_doc,
Benjamin Peterson0926ad12009-06-06 18:02:12 +0000142 "The error setting of the decoder or encoder.\n"
143 "\n"
144 "Subclasses should override.\n"
145 );
146
147static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000148textiobase_errors_get(PyObject *self, void *context)
Benjamin Peterson0926ad12009-06-06 18:02:12 +0000149{
150 Py_RETURN_NONE;
151}
152
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000153
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000154static PyMethodDef textiobase_methods[] = {
Siddhesh Poyarekar55edd0c2018-04-30 00:29:33 +0530155 {"detach", textiobase_detach, METH_NOARGS, textiobase_detach_doc},
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000156 {"read", textiobase_read, METH_VARARGS, textiobase_read_doc},
157 {"readline", textiobase_readline, METH_VARARGS, textiobase_readline_doc},
158 {"write", textiobase_write, METH_VARARGS, textiobase_write_doc},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000159 {NULL, NULL}
160};
161
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000162static PyGetSetDef textiobase_getset[] = {
163 {"encoding", (getter)textiobase_encoding_get, NULL, textiobase_encoding_doc},
164 {"newlines", (getter)textiobase_newlines_get, NULL, textiobase_newlines_doc},
165 {"errors", (getter)textiobase_errors_get, NULL, textiobase_errors_doc},
Benjamin Peterson1fea3212009-04-19 03:15:20 +0000166 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000167};
168
169PyTypeObject PyTextIOBase_Type = {
170 PyVarObject_HEAD_INIT(NULL, 0)
171 "_io._TextIOBase", /*tp_name*/
172 0, /*tp_basicsize*/
173 0, /*tp_itemsize*/
174 0, /*tp_dealloc*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +0200175 0, /*tp_vectorcall_offset*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000176 0, /*tp_getattr*/
177 0, /*tp_setattr*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +0200178 0, /*tp_as_async*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000179 0, /*tp_repr*/
180 0, /*tp_as_number*/
181 0, /*tp_as_sequence*/
182 0, /*tp_as_mapping*/
183 0, /*tp_hash */
184 0, /*tp_call*/
185 0, /*tp_str*/
186 0, /*tp_getattro*/
187 0, /*tp_setattro*/
188 0, /*tp_as_buffer*/
Antoine Pitrouada319b2019-05-29 22:12:38 +0200189 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000190 textiobase_doc, /* tp_doc */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000191 0, /* tp_traverse */
192 0, /* tp_clear */
193 0, /* tp_richcompare */
194 0, /* tp_weaklistoffset */
195 0, /* tp_iter */
196 0, /* tp_iternext */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000197 textiobase_methods, /* tp_methods */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000198 0, /* tp_members */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000199 textiobase_getset, /* tp_getset */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000200 &PyIOBase_Type, /* tp_base */
201 0, /* tp_dict */
202 0, /* tp_descr_get */
203 0, /* tp_descr_set */
204 0, /* tp_dictoffset */
205 0, /* tp_init */
206 0, /* tp_alloc */
207 0, /* tp_new */
Antoine Pitrou796564c2013-07-30 19:59:21 +0200208 0, /* tp_free */
209 0, /* tp_is_gc */
210 0, /* tp_bases */
211 0, /* tp_mro */
212 0, /* tp_cache */
213 0, /* tp_subclasses */
214 0, /* tp_weaklist */
215 0, /* tp_del */
216 0, /* tp_version_tag */
217 0, /* tp_finalize */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000218};
219
220
221/* IncrementalNewlineDecoder */
222
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000223typedef struct {
224 PyObject_HEAD
225 PyObject *decoder;
226 PyObject *errors;
Victor Stinner7d7e7752014-06-17 23:31:25 +0200227 unsigned int pendingcr: 1;
228 unsigned int translate: 1;
Antoine Pitrouca767bd2009-09-21 21:37:02 +0000229 unsigned int seennl: 3;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000230} nldecoder_object;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000231
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300232/*[clinic input]
233_io.IncrementalNewlineDecoder.__init__
234 decoder: object
235 translate: int
236 errors: object(c_default="NULL") = "strict"
237
238Codec used when reading a file in universal newlines mode.
239
240It wraps another incremental decoder, translating \r\n and \r into \n.
241It also records the types of newlines encountered. When used with
242translate=False, it ensures that the newline sequence is returned in
243one piece. When used with decoder=None, it expects unicode strings as
244decode input and translates newlines without first invoking an external
245decoder.
246[clinic start generated code]*/
247
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000248static int
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300249_io_IncrementalNewlineDecoder___init___impl(nldecoder_object *self,
250 PyObject *decoder, int translate,
251 PyObject *errors)
252/*[clinic end generated code: output=fbd04d443e764ec2 input=89db6b19c6b126bf]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000253{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000254 self->decoder = decoder;
255 Py_INCREF(decoder);
256
257 if (errors == NULL) {
INADA Naoki507434f2017-12-21 09:59:53 +0900258 self->errors = _PyUnicode_FromId(&PyId_strict);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000259 if (self->errors == NULL)
260 return -1;
261 }
262 else {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000263 self->errors = errors;
264 }
INADA Naoki507434f2017-12-21 09:59:53 +0900265 Py_INCREF(self->errors);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000266
Xiang Zhangb08746b2018-10-31 19:49:16 +0800267 self->translate = translate ? 1 : 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000268 self->seennl = 0;
269 self->pendingcr = 0;
270
271 return 0;
272}
273
274static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000275incrementalnewlinedecoder_dealloc(nldecoder_object *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000276{
277 Py_CLEAR(self->decoder);
278 Py_CLEAR(self->errors);
279 Py_TYPE(self)->tp_free((PyObject *)self);
280}
281
Serhiy Storchaka94dc6732013-02-03 17:03:31 +0200282static int
283check_decoded(PyObject *decoded)
284{
285 if (decoded == NULL)
286 return -1;
287 if (!PyUnicode_Check(decoded)) {
288 PyErr_Format(PyExc_TypeError,
289 "decoder should return a string result, not '%.200s'",
290 Py_TYPE(decoded)->tp_name);
291 Py_DECREF(decoded);
292 return -1;
293 }
Serhiy Storchakad03ce4a2013-02-03 17:07:32 +0200294 if (PyUnicode_READY(decoded) < 0) {
295 Py_DECREF(decoded);
296 return -1;
297 }
Serhiy Storchaka94dc6732013-02-03 17:03:31 +0200298 return 0;
299}
300
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000301#define SEEN_CR 1
302#define SEEN_LF 2
303#define SEEN_CRLF 4
304#define SEEN_ALL (SEEN_CR | SEEN_LF | SEEN_CRLF)
305
306PyObject *
Antoine Pitrou09fcb722013-10-23 19:20:21 +0200307_PyIncrementalNewlineDecoder_decode(PyObject *myself,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000308 PyObject *input, int final)
309{
310 PyObject *output;
311 Py_ssize_t output_len;
Antoine Pitrou09fcb722013-10-23 19:20:21 +0200312 nldecoder_object *self = (nldecoder_object *) myself;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000313
314 if (self->decoder == NULL) {
315 PyErr_SetString(PyExc_ValueError,
316 "IncrementalNewlineDecoder.__init__ not called");
317 return NULL;
318 }
319
320 /* decode input (with the eventual \r from a previous pass) */
321 if (self->decoder != Py_None) {
322 output = PyObject_CallMethodObjArgs(self->decoder,
323 _PyIO_str_decode, input, final ? Py_True : Py_False, NULL);
324 }
325 else {
326 output = input;
327 Py_INCREF(output);
328 }
329
Serhiy Storchaka94dc6732013-02-03 17:03:31 +0200330 if (check_decoded(output) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000331 return NULL;
332
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200333 output_len = PyUnicode_GET_LENGTH(output);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000334 if (self->pendingcr && (final || output_len > 0)) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200335 /* Prefix output with CR */
336 int kind;
337 PyObject *modified;
338 char *out;
339
340 modified = PyUnicode_New(output_len + 1,
341 PyUnicode_MAX_CHAR_VALUE(output));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000342 if (modified == NULL)
343 goto error;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200344 kind = PyUnicode_KIND(modified);
345 out = PyUnicode_DATA(modified);
Serhiy Storchakacd8295f2020-04-11 10:48:40 +0300346 PyUnicode_WRITE(kind, out, 0, '\r');
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200347 memcpy(out + kind, PyUnicode_DATA(output), kind * output_len);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000348 Py_DECREF(output);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200349 output = modified; /* output remains ready */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000350 self->pendingcr = 0;
351 output_len++;
352 }
353
354 /* retain last \r even when not translating data:
355 * then readline() is sure to get \r\n in one pass
356 */
357 if (!final) {
Antoine Pitrou24f36292009-03-28 22:16:42 +0000358 if (output_len > 0
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200359 && PyUnicode_READ_CHAR(output, output_len - 1) == '\r')
360 {
361 PyObject *modified = PyUnicode_Substring(output, 0, output_len -1);
362 if (modified == NULL)
363 goto error;
364 Py_DECREF(output);
365 output = modified;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000366 self->pendingcr = 1;
367 }
368 }
369
370 /* Record which newlines are read and do newline translation if desired,
371 all in one pass. */
372 {
Serhiy Storchakacd8295f2020-04-11 10:48:40 +0300373 const void *in_str;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000374 Py_ssize_t len;
375 int seennl = self->seennl;
376 int only_lf = 0;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200377 int kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000378
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200379 in_str = PyUnicode_DATA(output);
380 len = PyUnicode_GET_LENGTH(output);
381 kind = PyUnicode_KIND(output);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000382
383 if (len == 0)
384 return output;
385
386 /* If, up to now, newlines are consistently \n, do a quick check
387 for the \r *byte* with the libc's optimized memchr.
388 */
389 if (seennl == SEEN_LF || seennl == 0) {
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200390 only_lf = (memchr(in_str, '\r', kind * len) == NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000391 }
392
Antoine Pitrou66913e22009-03-06 23:40:56 +0000393 if (only_lf) {
394 /* If not already seen, quick scan for a possible "\n" character.
395 (there's nothing else to be done, even when in translation mode)
396 */
397 if (seennl == 0 &&
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200398 memchr(in_str, '\n', kind * len) != NULL) {
Antoine Pitrouc28e2e52011-11-13 03:53:42 +0100399 if (kind == PyUnicode_1BYTE_KIND)
400 seennl |= SEEN_LF;
401 else {
402 Py_ssize_t i = 0;
403 for (;;) {
404 Py_UCS4 c;
405 /* Fast loop for non-control characters */
406 while (PyUnicode_READ(kind, in_str, i) > '\n')
407 i++;
408 c = PyUnicode_READ(kind, in_str, i++);
409 if (c == '\n') {
410 seennl |= SEEN_LF;
411 break;
412 }
413 if (i >= len)
414 break;
Antoine Pitrou66913e22009-03-06 23:40:56 +0000415 }
Antoine Pitrou66913e22009-03-06 23:40:56 +0000416 }
417 }
418 /* Finished: we have scanned for newlines, and none of them
419 need translating */
420 }
421 else if (!self->translate) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200422 Py_ssize_t i = 0;
Antoine Pitrou66913e22009-03-06 23:40:56 +0000423 /* We have already seen all newline types, no need to scan again */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000424 if (seennl == SEEN_ALL)
425 goto endscan;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000426 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200427 Py_UCS4 c;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000428 /* Fast loop for non-control characters */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200429 while (PyUnicode_READ(kind, in_str, i) > '\r')
430 i++;
431 c = PyUnicode_READ(kind, in_str, i++);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000432 if (c == '\n')
433 seennl |= SEEN_LF;
434 else if (c == '\r') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200435 if (PyUnicode_READ(kind, in_str, i) == '\n') {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000436 seennl |= SEEN_CRLF;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200437 i++;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000438 }
439 else
440 seennl |= SEEN_CR;
441 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200442 if (i >= len)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000443 break;
444 if (seennl == SEEN_ALL)
445 break;
446 }
447 endscan:
448 ;
449 }
Antoine Pitrou66913e22009-03-06 23:40:56 +0000450 else {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200451 void *translated;
452 int kind = PyUnicode_KIND(output);
Serhiy Storchakacd8295f2020-04-11 10:48:40 +0300453 const void *in_str = PyUnicode_DATA(output);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200454 Py_ssize_t in, out;
455 /* XXX: Previous in-place translation here is disabled as
456 resizing is not possible anymore */
457 /* We could try to optimize this so that we only do a copy
458 when there is something to translate. On the other hand,
459 we already know there is a \r byte, so chances are high
460 that something needs to be done. */
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200461 translated = PyMem_Malloc(kind * len);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200462 if (translated == NULL) {
463 PyErr_NoMemory();
464 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000465 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200466 in = out = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000467 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200468 Py_UCS4 c;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000469 /* Fast loop for non-control characters */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200470 while ((c = PyUnicode_READ(kind, in_str, in++)) > '\r')
471 PyUnicode_WRITE(kind, translated, out++, c);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000472 if (c == '\n') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200473 PyUnicode_WRITE(kind, translated, out++, c);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000474 seennl |= SEEN_LF;
475 continue;
476 }
477 if (c == '\r') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200478 if (PyUnicode_READ(kind, in_str, in) == '\n') {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000479 in++;
480 seennl |= SEEN_CRLF;
481 }
482 else
483 seennl |= SEEN_CR;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200484 PyUnicode_WRITE(kind, translated, out++, '\n');
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000485 continue;
486 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200487 if (in > len)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000488 break;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200489 PyUnicode_WRITE(kind, translated, out++, c);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000490 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200491 Py_DECREF(output);
492 output = PyUnicode_FromKindAndData(kind, translated, out);
Antoine Pitrouc1b0bfd2011-11-12 22:34:28 +0100493 PyMem_Free(translated);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200494 if (!output)
Ross Lagerwall0f9eec12012-04-07 07:09:57 +0200495 return NULL;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000496 }
497 self->seennl |= seennl;
498 }
499
500 return output;
501
502 error:
503 Py_DECREF(output);
504 return NULL;
505}
506
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300507/*[clinic input]
508_io.IncrementalNewlineDecoder.decode
509 input: object
Serhiy Storchaka202fda52017-03-12 10:10:47 +0200510 final: bool(accept={int}) = False
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300511[clinic start generated code]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000512
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300513static PyObject *
514_io_IncrementalNewlineDecoder_decode_impl(nldecoder_object *self,
515 PyObject *input, int final)
Serhiy Storchaka202fda52017-03-12 10:10:47 +0200516/*[clinic end generated code: output=0d486755bb37a66e input=a4ea97f26372d866]*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300517{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000518 return _PyIncrementalNewlineDecoder_decode((PyObject *) self, input, final);
519}
520
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300521/*[clinic input]
522_io.IncrementalNewlineDecoder.getstate
523[clinic start generated code]*/
524
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000525static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300526_io_IncrementalNewlineDecoder_getstate_impl(nldecoder_object *self)
527/*[clinic end generated code: output=f0d2c9c136f4e0d0 input=f8ff101825e32e7f]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000528{
529 PyObject *buffer;
Benjamin Peterson47ff0732016-09-08 09:15:54 -0700530 unsigned long long flag;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000531
532 if (self->decoder != Py_None) {
Petr Viktorinffd97532020-02-11 17:46:57 +0100533 PyObject *state = PyObject_CallMethodNoArgs(self->decoder,
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200534 _PyIO_str_getstate);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000535 if (state == NULL)
536 return NULL;
Oren Milman13614e32017-08-24 19:51:24 +0300537 if (!PyTuple_Check(state)) {
538 PyErr_SetString(PyExc_TypeError,
539 "illegal decoder state");
540 Py_DECREF(state);
541 return NULL;
542 }
543 if (!PyArg_ParseTuple(state, "OK;illegal decoder state",
544 &buffer, &flag))
545 {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000546 Py_DECREF(state);
547 return NULL;
548 }
549 Py_INCREF(buffer);
550 Py_DECREF(state);
551 }
552 else {
553 buffer = PyBytes_FromString("");
554 flag = 0;
555 }
556 flag <<= 1;
557 if (self->pendingcr)
558 flag |= 1;
559 return Py_BuildValue("NK", buffer, flag);
560}
561
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300562/*[clinic input]
563_io.IncrementalNewlineDecoder.setstate
564 state: object
565 /
566[clinic start generated code]*/
567
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000568static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300569_io_IncrementalNewlineDecoder_setstate(nldecoder_object *self,
570 PyObject *state)
571/*[clinic end generated code: output=c10c622508b576cb input=c53fb505a76dbbe2]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000572{
573 PyObject *buffer;
Benjamin Peterson47ff0732016-09-08 09:15:54 -0700574 unsigned long long flag;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000575
Oren Milman1d1d3e92017-08-20 18:35:36 +0300576 if (!PyTuple_Check(state)) {
577 PyErr_SetString(PyExc_TypeError, "state argument must be a tuple");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000578 return NULL;
Oren Milman1d1d3e92017-08-20 18:35:36 +0300579 }
580 if (!PyArg_ParseTuple(state, "OK;setstate(): illegal state argument",
581 &buffer, &flag))
582 {
583 return NULL;
584 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000585
Victor Stinner7d7e7752014-06-17 23:31:25 +0200586 self->pendingcr = (int) (flag & 1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000587 flag >>= 1;
588
589 if (self->decoder != Py_None)
Martin v. Löwisafe55bb2011-10-09 10:38:36 +0200590 return _PyObject_CallMethodId(self->decoder,
591 &PyId_setstate, "((OK))", buffer, flag);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000592 else
593 Py_RETURN_NONE;
594}
595
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300596/*[clinic input]
597_io.IncrementalNewlineDecoder.reset
598[clinic start generated code]*/
599
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000600static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300601_io_IncrementalNewlineDecoder_reset_impl(nldecoder_object *self)
602/*[clinic end generated code: output=32fa40c7462aa8ff input=728678ddaea776df]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000603{
604 self->seennl = 0;
605 self->pendingcr = 0;
606 if (self->decoder != Py_None)
Petr Viktorinffd97532020-02-11 17:46:57 +0100607 return PyObject_CallMethodNoArgs(self->decoder, _PyIO_str_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000608 else
609 Py_RETURN_NONE;
610}
611
612static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000613incrementalnewlinedecoder_newlines_get(nldecoder_object *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000614{
615 switch (self->seennl) {
616 case SEEN_CR:
617 return PyUnicode_FromString("\r");
618 case SEEN_LF:
619 return PyUnicode_FromString("\n");
620 case SEEN_CRLF:
621 return PyUnicode_FromString("\r\n");
622 case SEEN_CR | SEEN_LF:
623 return Py_BuildValue("ss", "\r", "\n");
624 case SEEN_CR | SEEN_CRLF:
625 return Py_BuildValue("ss", "\r", "\r\n");
626 case SEEN_LF | SEEN_CRLF:
627 return Py_BuildValue("ss", "\n", "\r\n");
628 case SEEN_CR | SEEN_LF | SEEN_CRLF:
629 return Py_BuildValue("sss", "\r", "\n", "\r\n");
630 default:
631 Py_RETURN_NONE;
632 }
633
634}
635
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000636/* TextIOWrapper */
637
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000638typedef PyObject *
639 (*encodefunc_t)(PyObject *, PyObject *);
640
641typedef struct
642{
643 PyObject_HEAD
644 int ok; /* initialized? */
Benjamin Petersond2e0c792009-05-01 20:40:59 +0000645 int detached;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000646 Py_ssize_t chunk_size;
647 PyObject *buffer;
648 PyObject *encoding;
649 PyObject *encoder;
650 PyObject *decoder;
651 PyObject *readnl;
652 PyObject *errors;
INADA Naoki507434f2017-12-21 09:59:53 +0900653 const char *writenl; /* ASCII-encoded; NULL stands for \n */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000654 char line_buffering;
Antoine Pitroue96ec682011-07-23 21:46:35 +0200655 char write_through;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000656 char readuniversal;
657 char readtranslate;
658 char writetranslate;
659 char seekable;
Antoine Pitroue96ec682011-07-23 21:46:35 +0200660 char has_read1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000661 char telling;
Antoine Pitrou796564c2013-07-30 19:59:21 +0200662 char finalizing;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000663 /* Specialized encoding func (see below) */
664 encodefunc_t encodefunc;
Antoine Pitroue4501852009-05-14 18:55:55 +0000665 /* Whether or not it's the start of the stream */
666 char encoding_start_of_stream;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000667
668 /* Reads and writes are internally buffered in order to speed things up.
669 However, any read will first flush the write buffer if itsn't empty.
Antoine Pitrou24f36292009-03-28 22:16:42 +0000670
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000671 Please also note that text to be written is first encoded before being
672 buffered. This is necessary so that encoding errors are immediately
673 reported to the caller, but it unfortunately means that the
674 IncrementalEncoder (whose encode() method is always written in Python)
675 becomes a bottleneck for small writes.
676 */
677 PyObject *decoded_chars; /* buffer for text returned from decoder */
678 Py_ssize_t decoded_chars_used; /* offset into _decoded_chars for read() */
Inada Naokibfba8c32019-05-16 15:03:20 +0900679 PyObject *pending_bytes; // data waiting to be written.
680 // ascii unicode, bytes, or list of them.
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000681 Py_ssize_t pending_bytes_count;
Antoine Pitrou211b81d2011-02-25 20:27:33 +0000682
Oren Milman13614e32017-08-24 19:51:24 +0300683 /* snapshot is either NULL, or a tuple (dec_flags, next_input) where
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000684 * dec_flags is the second (integer) item of the decoder state and
685 * next_input is the chunk of input bytes that comes next after the
686 * snapshot point. We use this to reconstruct decoder states in tell().
687 */
Antoine Pitrou211b81d2011-02-25 20:27:33 +0000688 PyObject *snapshot;
689 /* Bytes-to-characters ratio for the current chunk. Serves as input for
690 the heuristic in tell(). */
691 double b2cratio;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000692
693 /* Cache raw object if it's a FileIO object */
694 PyObject *raw;
695
696 PyObject *weakreflist;
697 PyObject *dict;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000698} textio;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000699
Zackery Spytz23db9352018-06-29 04:14:58 -0600700static void
701textiowrapper_set_decoded_chars(textio *self, PyObject *chars);
702
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000703/* A couple of specialized cases in order to bypass the slow incremental
704 encoding methods for the most popular encodings. */
705
706static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000707ascii_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000708{
INADA Naoki507434f2017-12-21 09:59:53 +0900709 return _PyUnicode_AsASCIIString(text, PyUnicode_AsUTF8(self->errors));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000710}
711
712static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000713utf16be_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000714{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100715 return _PyUnicode_EncodeUTF16(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900716 PyUnicode_AsUTF8(self->errors), 1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000717}
718
719static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000720utf16le_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000721{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100722 return _PyUnicode_EncodeUTF16(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900723 PyUnicode_AsUTF8(self->errors), -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000724}
725
726static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000727utf16_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000728{
Antoine Pitroue4501852009-05-14 18:55:55 +0000729 if (!self->encoding_start_of_stream) {
730 /* Skip the BOM and use native byte ordering */
Christian Heimes743e0cd2012-10-17 23:52:17 +0200731#if PY_BIG_ENDIAN
Antoine Pitroue4501852009-05-14 18:55:55 +0000732 return utf16be_encode(self, text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000733#else
Antoine Pitroue4501852009-05-14 18:55:55 +0000734 return utf16le_encode(self, text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000735#endif
Antoine Pitroue4501852009-05-14 18:55:55 +0000736 }
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100737 return _PyUnicode_EncodeUTF16(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900738 PyUnicode_AsUTF8(self->errors), 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000739}
740
Antoine Pitroue4501852009-05-14 18:55:55 +0000741static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000742utf32be_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000743{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100744 return _PyUnicode_EncodeUTF32(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900745 PyUnicode_AsUTF8(self->errors), 1);
Antoine Pitroue4501852009-05-14 18:55:55 +0000746}
747
748static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000749utf32le_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000750{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100751 return _PyUnicode_EncodeUTF32(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900752 PyUnicode_AsUTF8(self->errors), -1);
Antoine Pitroue4501852009-05-14 18:55:55 +0000753}
754
755static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000756utf32_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000757{
758 if (!self->encoding_start_of_stream) {
759 /* Skip the BOM and use native byte ordering */
Christian Heimes743e0cd2012-10-17 23:52:17 +0200760#if PY_BIG_ENDIAN
Antoine Pitroue4501852009-05-14 18:55:55 +0000761 return utf32be_encode(self, text);
762#else
763 return utf32le_encode(self, text);
764#endif
765 }
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100766 return _PyUnicode_EncodeUTF32(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900767 PyUnicode_AsUTF8(self->errors), 0);
Antoine Pitroue4501852009-05-14 18:55:55 +0000768}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000769
770static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000771utf8_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000772{
INADA Naoki507434f2017-12-21 09:59:53 +0900773 return _PyUnicode_AsUTF8String(text, PyUnicode_AsUTF8(self->errors));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000774}
775
776static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000777latin1_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000778{
INADA Naoki507434f2017-12-21 09:59:53 +0900779 return _PyUnicode_AsLatin1String(text, PyUnicode_AsUTF8(self->errors));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000780}
781
Inada Naokibfba8c32019-05-16 15:03:20 +0900782// Return true when encoding can be skipped when text is ascii.
783static inline int
784is_asciicompat_encoding(encodefunc_t f)
785{
786 return f == (encodefunc_t) ascii_encode
787 || f == (encodefunc_t) latin1_encode
788 || f == (encodefunc_t) utf8_encode;
789}
790
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000791/* Map normalized encoding names onto the specialized encoding funcs */
792
793typedef struct {
794 const char *name;
795 encodefunc_t encodefunc;
796} encodefuncentry;
797
Serhiy Storchaka2d06e842015-12-25 19:53:18 +0200798static const 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
INADA Naoki507434f2017-12-21 09:59:53 +0900811static int
812validate_newline(const char *newline)
813{
814 if (newline && newline[0] != '\0'
815 && !(newline[0] == '\n' && newline[1] == '\0')
816 && !(newline[0] == '\r' && newline[1] == '\0')
817 && !(newline[0] == '\r' && newline[1] == '\n' && newline[2] == '\0')) {
818 PyErr_Format(PyExc_ValueError,
819 "illegal newline value: %s", newline);
820 return -1;
821 }
822 return 0;
823}
824
825static int
826set_newline(textio *self, const char *newline)
827{
828 PyObject *old = self->readnl;
829 if (newline == NULL) {
830 self->readnl = NULL;
831 }
832 else {
833 self->readnl = PyUnicode_FromString(newline);
834 if (self->readnl == NULL) {
835 self->readnl = old;
836 return -1;
837 }
838 }
839 self->readuniversal = (newline == NULL || newline[0] == '\0');
840 self->readtranslate = (newline == NULL);
841 self->writetranslate = (newline == NULL || newline[0] != '\0');
842 if (!self->readuniversal && self->readnl != NULL) {
843 // validate_newline() accepts only ASCII newlines.
844 assert(PyUnicode_KIND(self->readnl) == PyUnicode_1BYTE_KIND);
845 self->writenl = (const char *)PyUnicode_1BYTE_DATA(self->readnl);
846 if (strcmp(self->writenl, "\n") == 0) {
847 self->writenl = NULL;
848 }
849 }
850 else {
851#ifdef MS_WINDOWS
852 self->writenl = "\r\n";
853#else
854 self->writenl = NULL;
855#endif
856 }
857 Py_XDECREF(old);
858 return 0;
859}
860
861static int
862_textiowrapper_set_decoder(textio *self, PyObject *codec_info,
863 const char *errors)
864{
865 PyObject *res;
866 int r;
867
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200868 res = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_readable);
INADA Naoki507434f2017-12-21 09:59:53 +0900869 if (res == NULL)
870 return -1;
871
872 r = PyObject_IsTrue(res);
873 Py_DECREF(res);
874 if (r == -1)
875 return -1;
876
877 if (r != 1)
878 return 0;
879
880 Py_CLEAR(self->decoder);
881 self->decoder = _PyCodecInfo_GetIncrementalDecoder(codec_info, errors);
882 if (self->decoder == NULL)
883 return -1;
884
885 if (self->readuniversal) {
Serhiy Storchaka1f21eaa2019-09-01 12:16:51 +0300886 PyObject *incrementalDecoder = PyObject_CallFunctionObjArgs(
INADA Naoki507434f2017-12-21 09:59:53 +0900887 (PyObject *)&PyIncrementalNewlineDecoder_Type,
Serhiy Storchaka1f21eaa2019-09-01 12:16:51 +0300888 self->decoder, self->readtranslate ? Py_True : Py_False, NULL);
INADA Naoki507434f2017-12-21 09:59:53 +0900889 if (incrementalDecoder == NULL)
890 return -1;
891 Py_CLEAR(self->decoder);
892 self->decoder = incrementalDecoder;
893 }
894
895 return 0;
896}
897
898static PyObject*
899_textiowrapper_decode(PyObject *decoder, PyObject *bytes, int eof)
900{
901 PyObject *chars;
902
Andy Lesterdffe4c02020-03-04 07:15:20 -0600903 if (Py_IS_TYPE(decoder, &PyIncrementalNewlineDecoder_Type))
INADA Naoki507434f2017-12-21 09:59:53 +0900904 chars = _PyIncrementalNewlineDecoder_decode(decoder, bytes, eof);
905 else
906 chars = PyObject_CallMethodObjArgs(decoder, _PyIO_str_decode, bytes,
907 eof ? Py_True : Py_False, NULL);
908
909 if (check_decoded(chars) < 0)
910 // check_decoded already decreases refcount
911 return NULL;
912
913 return chars;
914}
915
916static int
917_textiowrapper_set_encoder(textio *self, PyObject *codec_info,
918 const char *errors)
919{
920 PyObject *res;
921 int r;
922
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200923 res = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_writable);
INADA Naoki507434f2017-12-21 09:59:53 +0900924 if (res == NULL)
925 return -1;
926
927 r = PyObject_IsTrue(res);
928 Py_DECREF(res);
929 if (r == -1)
930 return -1;
931
932 if (r != 1)
933 return 0;
934
935 Py_CLEAR(self->encoder);
936 self->encodefunc = NULL;
937 self->encoder = _PyCodecInfo_GetIncrementalEncoder(codec_info, errors);
938 if (self->encoder == NULL)
939 return -1;
940
941 /* Get the normalized named of the codec */
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200942 if (_PyObject_LookupAttrId(codec_info, &PyId_name, &res) < 0) {
943 return -1;
INADA Naoki507434f2017-12-21 09:59:53 +0900944 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200945 if (res != NULL && PyUnicode_Check(res)) {
INADA Naoki507434f2017-12-21 09:59:53 +0900946 const encodefuncentry *e = encodefuncs;
947 while (e->name != NULL) {
948 if (_PyUnicode_EqualToASCIIString(res, e->name)) {
949 self->encodefunc = e->encodefunc;
950 break;
951 }
952 e++;
953 }
954 }
955 Py_XDECREF(res);
956
957 return 0;
958}
959
960static int
961_textiowrapper_fix_encoder_state(textio *self)
962{
963 if (!self->seekable || !self->encoder) {
964 return 0;
965 }
966
967 self->encoding_start_of_stream = 1;
968
Petr Viktorinffd97532020-02-11 17:46:57 +0100969 PyObject *cookieObj = PyObject_CallMethodNoArgs(
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200970 self->buffer, _PyIO_str_tell);
INADA Naoki507434f2017-12-21 09:59:53 +0900971 if (cookieObj == NULL) {
972 return -1;
973 }
974
Victor Stinner37834132020-10-27 17:12:53 +0100975 int cmp = PyObject_RichCompareBool(cookieObj, _PyLong_GetZero(), Py_EQ);
INADA Naoki507434f2017-12-21 09:59:53 +0900976 Py_DECREF(cookieObj);
977 if (cmp < 0) {
978 return -1;
979 }
980
981 if (cmp == 0) {
982 self->encoding_start_of_stream = 0;
Petr Viktorinffd97532020-02-11 17:46:57 +0100983 PyObject *res = PyObject_CallMethodOneArg(
Victor Stinner37834132020-10-27 17:12:53 +0100984 self->encoder, _PyIO_str_setstate, _PyLong_GetZero());
INADA Naoki507434f2017-12-21 09:59:53 +0900985 if (res == NULL) {
986 return -1;
987 }
988 Py_DECREF(res);
989 }
990
991 return 0;
992}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000993
Victor Stinner22eb6892019-06-26 00:51:05 +0200994static int
995io_check_errors(PyObject *errors)
996{
997 assert(errors != NULL && errors != Py_None);
998
Victor Stinner81a7be32020-04-14 15:14:01 +0200999 PyInterpreterState *interp = _PyInterpreterState_GET();
Victor Stinner22eb6892019-06-26 00:51:05 +02001000#ifndef Py_DEBUG
1001 /* In release mode, only check in development mode (-X dev) */
Victor Stinnerda7933e2020-04-13 03:04:28 +02001002 if (!_PyInterpreterState_GetConfig(interp)->dev_mode) {
Victor Stinner22eb6892019-06-26 00:51:05 +02001003 return 0;
1004 }
1005#else
1006 /* Always check in debug mode */
1007#endif
1008
1009 /* Avoid calling PyCodec_LookupError() before the codec registry is ready:
1010 before_PyUnicode_InitEncodings() is called. */
Victor Stinner3d17c042020-05-14 01:48:38 +02001011 if (!interp->unicode.fs_codec.encoding) {
Victor Stinner22eb6892019-06-26 00:51:05 +02001012 return 0;
1013 }
1014
1015 Py_ssize_t name_length;
1016 const char *name = PyUnicode_AsUTF8AndSize(errors, &name_length);
1017 if (name == NULL) {
1018 return -1;
1019 }
1020 if (strlen(name) != (size_t)name_length) {
1021 PyErr_SetString(PyExc_ValueError, "embedded null character in errors");
1022 return -1;
1023 }
1024 PyObject *handler = PyCodec_LookupError(name);
1025 if (handler != NULL) {
1026 Py_DECREF(handler);
1027 return 0;
1028 }
1029 return -1;
1030}
1031
1032
1033
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001034/*[clinic input]
1035_io.TextIOWrapper.__init__
1036 buffer: object
Serhiy Storchaka279f4462019-09-14 12:24:05 +03001037 encoding: str(accept={str, NoneType}) = None
INADA Naoki507434f2017-12-21 09:59:53 +09001038 errors: object = None
Serhiy Storchaka279f4462019-09-14 12:24:05 +03001039 newline: str(accept={str, NoneType}) = None
Serhiy Storchaka202fda52017-03-12 10:10:47 +02001040 line_buffering: bool(accept={int}) = False
1041 write_through: bool(accept={int}) = False
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001042
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001043Character and line based layer over a BufferedIOBase object, buffer.
1044
1045encoding gives the name of the encoding that the stream will be
1046decoded or encoded with. It defaults to locale.getpreferredencoding(False).
1047
1048errors determines the strictness of encoding and decoding (see
1049help(codecs.Codec) or the documentation for codecs.register) and
1050defaults to "strict".
1051
1052newline controls how line endings are handled. It can be None, '',
1053'\n', '\r', and '\r\n'. It works as follows:
1054
1055* On input, if newline is None, universal newlines mode is
1056 enabled. Lines in the input can end in '\n', '\r', or '\r\n', and
1057 these are translated into '\n' before being returned to the
1058 caller. If it is '', universal newline mode is enabled, but line
1059 endings are returned to the caller untranslated. If it has any of
1060 the other legal values, input lines are only terminated by the given
1061 string, and the line ending is returned to the caller untranslated.
1062
1063* On output, if newline is None, any '\n' characters written are
1064 translated to the system default line separator, os.linesep. If
1065 newline is '' or '\n', no translation takes place. If newline is any
1066 of the other legal values, any '\n' characters written are translated
1067 to the given string.
1068
1069If line_buffering is True, a call to flush is implied when a call to
1070write contains a newline character.
1071[clinic start generated code]*/
1072
1073static int
1074_io_TextIOWrapper___init___impl(textio *self, PyObject *buffer,
INADA Naoki507434f2017-12-21 09:59:53 +09001075 const char *encoding, PyObject *errors,
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001076 const char *newline, int line_buffering,
1077 int write_through)
Serhiy Storchaka279f4462019-09-14 12:24:05 +03001078/*[clinic end generated code: output=72267c0c01032ed2 input=77d8696d1a1f460b]*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001079{
1080 PyObject *raw, *codec_info = NULL;
1081 _PyIO_State *state = NULL;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001082 PyObject *res;
1083 int r;
1084
1085 self->ok = 0;
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001086 self->detached = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001087
INADA Naoki507434f2017-12-21 09:59:53 +09001088 if (errors == Py_None) {
1089 errors = _PyUnicode_FromId(&PyId_strict); /* borrowed */
INADA Naoki4856b0f2017-12-24 10:29:19 +09001090 if (errors == NULL) {
1091 return -1;
1092 }
INADA Naoki507434f2017-12-21 09:59:53 +09001093 }
1094 else if (!PyUnicode_Check(errors)) {
1095 // Check 'errors' argument here because Argument Clinic doesn't support
1096 // 'str(accept={str, NoneType})' converter.
1097 PyErr_Format(
1098 PyExc_TypeError,
1099 "TextIOWrapper() argument 'errors' must be str or None, not %.50s",
Victor Stinnerdaa97562020-02-07 03:37:06 +01001100 Py_TYPE(errors)->tp_name);
INADA Naoki507434f2017-12-21 09:59:53 +09001101 return -1;
1102 }
Victor Stinner22eb6892019-06-26 00:51:05 +02001103 else if (io_check_errors(errors)) {
1104 return -1;
1105 }
INADA Naoki507434f2017-12-21 09:59:53 +09001106
1107 if (validate_newline(newline) < 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001108 return -1;
1109 }
1110
1111 Py_CLEAR(self->buffer);
1112 Py_CLEAR(self->encoding);
1113 Py_CLEAR(self->encoder);
1114 Py_CLEAR(self->decoder);
1115 Py_CLEAR(self->readnl);
1116 Py_CLEAR(self->decoded_chars);
1117 Py_CLEAR(self->pending_bytes);
1118 Py_CLEAR(self->snapshot);
1119 Py_CLEAR(self->errors);
1120 Py_CLEAR(self->raw);
1121 self->decoded_chars_used = 0;
1122 self->pending_bytes_count = 0;
1123 self->encodefunc = NULL;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001124 self->b2cratio = 0.0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001125
1126 if (encoding == NULL) {
1127 /* Try os.device_encoding(fileno) */
1128 PyObject *fileno;
Antoine Pitrou712cb732013-12-21 15:51:54 +01001129 state = IO_STATE();
1130 if (state == NULL)
1131 goto error;
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001132 fileno = _PyObject_CallMethodIdNoArgs(buffer, &PyId_fileno);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001133 /* Ignore only AttributeError and UnsupportedOperation */
1134 if (fileno == NULL) {
1135 if (PyErr_ExceptionMatches(PyExc_AttributeError) ||
1136 PyErr_ExceptionMatches(state->unsupported_operation)) {
1137 PyErr_Clear();
1138 }
1139 else {
1140 goto error;
1141 }
1142 }
1143 else {
Serhiy Storchaka78980432013-01-15 01:12:17 +02001144 int fd = _PyLong_AsInt(fileno);
Brett Cannonefb00c02012-02-29 18:31:31 -05001145 Py_DECREF(fileno);
1146 if (fd == -1 && PyErr_Occurred()) {
1147 goto error;
1148 }
1149
1150 self->encoding = _Py_device_encoding(fd);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001151 if (self->encoding == NULL)
1152 goto error;
1153 else if (!PyUnicode_Check(self->encoding))
1154 Py_CLEAR(self->encoding);
1155 }
1156 }
1157 if (encoding == NULL && self->encoding == NULL) {
Victor Stinner82458b62020-11-01 20:59:35 +01001158 self->encoding = _Py_GetLocaleEncodingObject();
Antoine Pitrou932ff832013-08-01 21:04:50 +02001159 if (self->encoding == NULL) {
Victor Stinner710e8262020-10-31 01:02:09 +01001160 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001161 }
Victor Stinner710e8262020-10-31 01:02:09 +01001162 assert(PyUnicode_Check(self->encoding));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001163 }
Victor Stinnerf6c57832010-05-19 01:17:01 +00001164 if (self->encoding != NULL) {
Serhiy Storchaka06515832016-11-20 09:13:07 +02001165 encoding = PyUnicode_AsUTF8(self->encoding);
Victor Stinnerf6c57832010-05-19 01:17:01 +00001166 if (encoding == NULL)
1167 goto error;
1168 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001169 else if (encoding != NULL) {
1170 self->encoding = PyUnicode_FromString(encoding);
1171 if (self->encoding == NULL)
1172 goto error;
1173 }
1174 else {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03001175 PyErr_SetString(PyExc_OSError,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001176 "could not determine default encoding");
Serhiy Storchakad6238a72017-09-24 02:49:58 +03001177 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001178 }
1179
Nick Coghlana9b15242014-02-04 22:11:18 +10001180 /* Check we have been asked for a real text encoding */
1181 codec_info = _PyCodec_LookupTextEncoding(encoding, "codecs.open()");
1182 if (codec_info == NULL) {
1183 Py_CLEAR(self->encoding);
1184 goto error;
1185 }
1186
1187 /* XXX: Failures beyond this point have the potential to leak elements
1188 * of the partially constructed object (like self->encoding)
1189 */
1190
INADA Naoki507434f2017-12-21 09:59:53 +09001191 Py_INCREF(errors);
1192 self->errors = errors;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001193 self->chunk_size = 8192;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001194 self->line_buffering = line_buffering;
Antoine Pitroue96ec682011-07-23 21:46:35 +02001195 self->write_through = write_through;
INADA Naoki507434f2017-12-21 09:59:53 +09001196 if (set_newline(self, newline) < 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001197 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001198 }
1199
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001200 self->buffer = buffer;
1201 Py_INCREF(buffer);
Antoine Pitrou24f36292009-03-28 22:16:42 +00001202
INADA Naoki507434f2017-12-21 09:59:53 +09001203 /* Build the decoder object */
1204 if (_textiowrapper_set_decoder(self, codec_info, PyUnicode_AsUTF8(errors)) != 0)
1205 goto error;
1206
1207 /* Build the encoder object */
1208 if (_textiowrapper_set_encoder(self, codec_info, PyUnicode_AsUTF8(errors)) != 0)
1209 goto error;
1210
1211 /* Finished sorting out the codec details */
1212 Py_CLEAR(codec_info);
1213
Andy Lesterdffe4c02020-03-04 07:15:20 -06001214 if (Py_IS_TYPE(buffer, &PyBufferedReader_Type) ||
1215 Py_IS_TYPE(buffer, &PyBufferedWriter_Type) ||
1216 Py_IS_TYPE(buffer, &PyBufferedRandom_Type))
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001217 {
1218 if (_PyObject_LookupAttrId(buffer, &PyId_raw, &raw) < 0)
1219 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001220 /* Cache the raw FileIO object to speed up 'closed' checks */
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001221 if (raw != NULL) {
Andy Lesterdffe4c02020-03-04 07:15:20 -06001222 if (Py_IS_TYPE(raw, &PyFileIO_Type))
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001223 self->raw = raw;
Benjamin Peterson2cfca792009-06-06 20:46:48 +00001224 else
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001225 Py_DECREF(raw);
Benjamin Peterson2cfca792009-06-06 20:46:48 +00001226 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001227 }
1228
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001229 res = _PyObject_CallMethodIdNoArgs(buffer, &PyId_seekable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001230 if (res == NULL)
1231 goto error;
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001232 r = PyObject_IsTrue(res);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001233 Py_DECREF(res);
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001234 if (r < 0)
1235 goto error;
1236 self->seekable = self->telling = r;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001237
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001238 r = _PyObject_LookupAttr(buffer, _PyIO_str_read1, &res);
1239 if (r < 0) {
Serhiy Storchaka4d9aec02018-01-16 18:34:21 +02001240 goto error;
1241 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001242 Py_XDECREF(res);
1243 self->has_read1 = r;
Antoine Pitroue96ec682011-07-23 21:46:35 +02001244
Antoine Pitroue4501852009-05-14 18:55:55 +00001245 self->encoding_start_of_stream = 0;
INADA Naoki507434f2017-12-21 09:59:53 +09001246 if (_textiowrapper_fix_encoder_state(self) < 0) {
1247 goto error;
Antoine Pitroue4501852009-05-14 18:55:55 +00001248 }
1249
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001250 self->ok = 1;
1251 return 0;
1252
1253 error:
Nick Coghlana9b15242014-02-04 22:11:18 +10001254 Py_XDECREF(codec_info);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001255 return -1;
1256}
1257
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001258/* Return *default_value* if ob is None, 0 if ob is false, 1 if ob is true,
1259 * -1 on error.
1260 */
1261static int
1262convert_optional_bool(PyObject *obj, int default_value)
1263{
1264 long v;
1265 if (obj == Py_None) {
1266 v = default_value;
1267 }
1268 else {
1269 v = PyLong_AsLong(obj);
1270 if (v == -1 && PyErr_Occurred())
1271 return -1;
1272 }
1273 return v != 0;
1274}
1275
INADA Naoki507434f2017-12-21 09:59:53 +09001276static int
1277textiowrapper_change_encoding(textio *self, PyObject *encoding,
1278 PyObject *errors, int newline_changed)
1279{
1280 /* Use existing settings where new settings are not specified */
1281 if (encoding == Py_None && errors == Py_None && !newline_changed) {
1282 return 0; // no change
1283 }
1284
1285 if (encoding == Py_None) {
1286 encoding = self->encoding;
1287 if (errors == Py_None) {
1288 errors = self->errors;
1289 }
1290 }
1291 else if (errors == Py_None) {
1292 errors = _PyUnicode_FromId(&PyId_strict);
INADA Naoki4856b0f2017-12-24 10:29:19 +09001293 if (errors == NULL) {
1294 return -1;
1295 }
INADA Naoki507434f2017-12-21 09:59:53 +09001296 }
1297
1298 const char *c_errors = PyUnicode_AsUTF8(errors);
1299 if (c_errors == NULL) {
1300 return -1;
1301 }
1302
1303 // Create new encoder & decoder
1304 PyObject *codec_info = _PyCodec_LookupTextEncoding(
1305 PyUnicode_AsUTF8(encoding), "codecs.open()");
1306 if (codec_info == NULL) {
1307 return -1;
1308 }
1309 if (_textiowrapper_set_decoder(self, codec_info, c_errors) != 0 ||
1310 _textiowrapper_set_encoder(self, codec_info, c_errors) != 0) {
1311 Py_DECREF(codec_info);
1312 return -1;
1313 }
1314 Py_DECREF(codec_info);
1315
1316 Py_INCREF(encoding);
1317 Py_INCREF(errors);
1318 Py_SETREF(self->encoding, encoding);
1319 Py_SETREF(self->errors, errors);
1320
1321 return _textiowrapper_fix_encoder_state(self);
1322}
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001323
1324/*[clinic input]
1325_io.TextIOWrapper.reconfigure
1326 *
INADA Naoki507434f2017-12-21 09:59:53 +09001327 encoding: object = None
1328 errors: object = None
1329 newline as newline_obj: object(c_default="NULL") = None
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001330 line_buffering as line_buffering_obj: object = None
1331 write_through as write_through_obj: object = None
1332
1333Reconfigure the text stream with new parameters.
1334
1335This also does an implicit stream flush.
1336
1337[clinic start generated code]*/
1338
1339static PyObject *
INADA Naoki507434f2017-12-21 09:59:53 +09001340_io_TextIOWrapper_reconfigure_impl(textio *self, PyObject *encoding,
1341 PyObject *errors, PyObject *newline_obj,
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001342 PyObject *line_buffering_obj,
1343 PyObject *write_through_obj)
INADA Naoki507434f2017-12-21 09:59:53 +09001344/*[clinic end generated code: output=52b812ff4b3d4b0f input=671e82136e0f5822]*/
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001345{
1346 int line_buffering;
1347 int write_through;
INADA Naoki507434f2017-12-21 09:59:53 +09001348 const char *newline = NULL;
1349
1350 /* Check if something is in the read buffer */
1351 if (self->decoded_chars != NULL) {
1352 if (encoding != Py_None || errors != Py_None || newline_obj != NULL) {
Serhiy Storchaka34fd4c22018-11-05 16:20:25 +02001353 _unsupported("It is not possible to set the encoding or newline "
INADA Naoki507434f2017-12-21 09:59:53 +09001354 "of stream after the first read");
1355 return NULL;
1356 }
1357 }
1358
1359 if (newline_obj != NULL && newline_obj != Py_None) {
1360 newline = PyUnicode_AsUTF8(newline_obj);
1361 if (newline == NULL || validate_newline(newline) < 0) {
1362 return NULL;
1363 }
1364 }
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001365
1366 line_buffering = convert_optional_bool(line_buffering_obj,
1367 self->line_buffering);
1368 write_through = convert_optional_bool(write_through_obj,
1369 self->write_through);
1370 if (line_buffering < 0 || write_through < 0) {
1371 return NULL;
1372 }
INADA Naoki507434f2017-12-21 09:59:53 +09001373
Petr Viktorinffd97532020-02-11 17:46:57 +01001374 PyObject *res = PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001375 if (res == NULL) {
1376 return NULL;
1377 }
INADA Naoki507434f2017-12-21 09:59:53 +09001378 Py_DECREF(res);
1379 self->b2cratio = 0;
1380
1381 if (newline_obj != NULL && set_newline(self, newline) < 0) {
1382 return NULL;
1383 }
1384
1385 if (textiowrapper_change_encoding(
1386 self, encoding, errors, newline_obj != NULL) < 0) {
1387 return NULL;
1388 }
1389
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001390 self->line_buffering = line_buffering;
1391 self->write_through = write_through;
1392 Py_RETURN_NONE;
1393}
1394
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001395static int
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001396textiowrapper_clear(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001397{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001398 self->ok = 0;
1399 Py_CLEAR(self->buffer);
1400 Py_CLEAR(self->encoding);
1401 Py_CLEAR(self->encoder);
1402 Py_CLEAR(self->decoder);
1403 Py_CLEAR(self->readnl);
1404 Py_CLEAR(self->decoded_chars);
1405 Py_CLEAR(self->pending_bytes);
1406 Py_CLEAR(self->snapshot);
1407 Py_CLEAR(self->errors);
1408 Py_CLEAR(self->raw);
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001409
1410 Py_CLEAR(self->dict);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001411 return 0;
1412}
1413
1414static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001415textiowrapper_dealloc(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001416{
Antoine Pitrou796564c2013-07-30 19:59:21 +02001417 self->finalizing = 1;
1418 if (_PyIOBase_finalize((PyObject *) self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001419 return;
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001420 self->ok = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001421 _PyObject_GC_UNTRACK(self);
1422 if (self->weakreflist != NULL)
1423 PyObject_ClearWeakRefs((PyObject *)self);
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001424 textiowrapper_clear(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001425 Py_TYPE(self)->tp_free((PyObject *)self);
1426}
1427
1428static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001429textiowrapper_traverse(textio *self, visitproc visit, void *arg)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001430{
1431 Py_VISIT(self->buffer);
1432 Py_VISIT(self->encoding);
1433 Py_VISIT(self->encoder);
1434 Py_VISIT(self->decoder);
1435 Py_VISIT(self->readnl);
1436 Py_VISIT(self->decoded_chars);
1437 Py_VISIT(self->pending_bytes);
1438 Py_VISIT(self->snapshot);
1439 Py_VISIT(self->errors);
1440 Py_VISIT(self->raw);
1441
1442 Py_VISIT(self->dict);
1443 return 0;
1444}
1445
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001446static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001447textiowrapper_closed_get(textio *self, void *context);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001448
1449/* This macro takes some shortcuts to make the common case faster. */
1450#define CHECK_CLOSED(self) \
1451 do { \
1452 int r; \
1453 PyObject *_res; \
Andy Lesterdffe4c02020-03-04 07:15:20 -06001454 if (Py_IS_TYPE(self, &PyTextIOWrapper_Type)) { \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001455 if (self->raw != NULL) \
1456 r = _PyFileIO_closed(self->raw); \
1457 else { \
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001458 _res = textiowrapper_closed_get(self, NULL); \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001459 if (_res == NULL) \
1460 return NULL; \
1461 r = PyObject_IsTrue(_res); \
1462 Py_DECREF(_res); \
1463 if (r < 0) \
1464 return NULL; \
1465 } \
1466 if (r > 0) { \
1467 PyErr_SetString(PyExc_ValueError, \
1468 "I/O operation on closed file."); \
1469 return NULL; \
1470 } \
1471 } \
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001472 else if (_PyIOBase_check_closed((PyObject *)self, Py_True) == NULL) \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001473 return NULL; \
1474 } while (0)
1475
1476#define CHECK_INITIALIZED(self) \
1477 if (self->ok <= 0) { \
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001478 PyErr_SetString(PyExc_ValueError, \
1479 "I/O operation on uninitialized object"); \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001480 return NULL; \
1481 }
1482
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001483#define CHECK_ATTACHED(self) \
1484 CHECK_INITIALIZED(self); \
1485 if (self->detached) { \
1486 PyErr_SetString(PyExc_ValueError, \
1487 "underlying buffer has been detached"); \
1488 return NULL; \
1489 }
1490
1491#define CHECK_ATTACHED_INT(self) \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001492 if (self->ok <= 0) { \
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001493 PyErr_SetString(PyExc_ValueError, \
1494 "I/O operation on uninitialized object"); \
1495 return -1; \
1496 } else if (self->detached) { \
1497 PyErr_SetString(PyExc_ValueError, \
1498 "underlying buffer has been detached"); \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001499 return -1; \
1500 }
1501
1502
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001503/*[clinic input]
1504_io.TextIOWrapper.detach
1505[clinic start generated code]*/
1506
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001507static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001508_io_TextIOWrapper_detach_impl(textio *self)
1509/*[clinic end generated code: output=7ba3715cd032d5f2 input=e5a71fbda9e1d9f9]*/
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001510{
1511 PyObject *buffer, *res;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001512 CHECK_ATTACHED(self);
Petr Viktorinffd97532020-02-11 17:46:57 +01001513 res = PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001514 if (res == NULL)
1515 return NULL;
1516 Py_DECREF(res);
1517 buffer = self->buffer;
1518 self->buffer = NULL;
1519 self->detached = 1;
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001520 return buffer;
1521}
1522
Antoine Pitrou24f36292009-03-28 22:16:42 +00001523/* Flush the internal write buffer. This doesn't explicitly flush the
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001524 underlying buffered object, though. */
1525static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001526_textiowrapper_writeflush(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001527{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001528 if (self->pending_bytes == NULL)
1529 return 0;
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001530
Inada Naokibfba8c32019-05-16 15:03:20 +09001531 PyObject *pending = self->pending_bytes;
1532 PyObject *b;
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001533
Inada Naokibfba8c32019-05-16 15:03:20 +09001534 if (PyBytes_Check(pending)) {
1535 b = pending;
1536 Py_INCREF(b);
1537 }
1538 else if (PyUnicode_Check(pending)) {
1539 assert(PyUnicode_IS_ASCII(pending));
1540 assert(PyUnicode_GET_LENGTH(pending) == self->pending_bytes_count);
1541 b = PyBytes_FromStringAndSize(
1542 PyUnicode_DATA(pending), PyUnicode_GET_LENGTH(pending));
1543 if (b == NULL) {
1544 return -1;
1545 }
1546 }
1547 else {
1548 assert(PyList_Check(pending));
1549 b = PyBytes_FromStringAndSize(NULL, self->pending_bytes_count);
1550 if (b == NULL) {
1551 return -1;
1552 }
1553
1554 char *buf = PyBytes_AsString(b);
1555 Py_ssize_t pos = 0;
1556
1557 for (Py_ssize_t i = 0; i < PyList_GET_SIZE(pending); i++) {
1558 PyObject *obj = PyList_GET_ITEM(pending, i);
1559 char *src;
1560 Py_ssize_t len;
1561 if (PyUnicode_Check(obj)) {
1562 assert(PyUnicode_IS_ASCII(obj));
1563 src = PyUnicode_DATA(obj);
1564 len = PyUnicode_GET_LENGTH(obj);
1565 }
1566 else {
1567 assert(PyBytes_Check(obj));
1568 if (PyBytes_AsStringAndSize(obj, &src, &len) < 0) {
1569 Py_DECREF(b);
1570 return -1;
1571 }
1572 }
1573 memcpy(buf + pos, src, len);
1574 pos += len;
1575 }
1576 assert(pos == self->pending_bytes_count);
1577 }
1578
1579 self->pending_bytes_count = 0;
1580 self->pending_bytes = NULL;
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001581 Py_DECREF(pending);
Inada Naokibfba8c32019-05-16 15:03:20 +09001582
1583 PyObject *ret;
Gregory P. Smithb9817b02013-02-01 13:03:39 -08001584 do {
Petr Viktorinffd97532020-02-11 17:46:57 +01001585 ret = PyObject_CallMethodOneArg(self->buffer, _PyIO_str_write, b);
Gregory P. Smithb9817b02013-02-01 13:03:39 -08001586 } while (ret == NULL && _PyIO_trap_eintr());
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001587 Py_DECREF(b);
1588 if (ret == NULL)
1589 return -1;
1590 Py_DECREF(ret);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001591 return 0;
1592}
1593
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001594/*[clinic input]
1595_io.TextIOWrapper.write
1596 text: unicode
1597 /
1598[clinic start generated code]*/
1599
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001600static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001601_io_TextIOWrapper_write_impl(textio *self, PyObject *text)
1602/*[clinic end generated code: output=d2deb0d50771fcec input=fdf19153584a0e44]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001603{
1604 PyObject *ret;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001605 PyObject *b;
1606 Py_ssize_t textlen;
1607 int haslf = 0;
Antoine Pitrouc644e7c2014-05-09 00:24:50 +02001608 int needflush = 0, text_needflush = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001609
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001610 if (PyUnicode_READY(text) == -1)
1611 return NULL;
1612
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001613 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001614 CHECK_CLOSED(self);
1615
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001616 if (self->encoder == NULL)
1617 return _unsupported("not writable");
Benjamin Peterson81971ea2009-05-14 22:01:31 +00001618
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001619 Py_INCREF(text);
1620
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001621 textlen = PyUnicode_GET_LENGTH(text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001622
1623 if ((self->writetranslate && self->writenl != NULL) || self->line_buffering)
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001624 if (PyUnicode_FindChar(text, '\n', 0, PyUnicode_GET_LENGTH(text), 1) != -1)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001625 haslf = 1;
1626
1627 if (haslf && self->writetranslate && self->writenl != NULL) {
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001628 PyObject *newtext = _PyObject_CallMethodId(
1629 text, &PyId_replace, "ss", "\n", self->writenl);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001630 Py_DECREF(text);
1631 if (newtext == NULL)
1632 return NULL;
1633 text = newtext;
1634 }
1635
Antoine Pitroue96ec682011-07-23 21:46:35 +02001636 if (self->write_through)
Antoine Pitrouc644e7c2014-05-09 00:24:50 +02001637 text_needflush = 1;
1638 if (self->line_buffering &&
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001639 (haslf ||
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001640 PyUnicode_FindChar(text, '\r', 0, PyUnicode_GET_LENGTH(text), 1) != -1))
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001641 needflush = 1;
1642
1643 /* XXX What if we were just reading? */
Antoine Pitroue4501852009-05-14 18:55:55 +00001644 if (self->encodefunc != NULL) {
Inada Naokibfba8c32019-05-16 15:03:20 +09001645 if (PyUnicode_IS_ASCII(text) && is_asciicompat_encoding(self->encodefunc)) {
1646 b = text;
1647 Py_INCREF(b);
1648 }
1649 else {
1650 b = (*self->encodefunc)((PyObject *) self, text);
1651 }
Antoine Pitroue4501852009-05-14 18:55:55 +00001652 self->encoding_start_of_stream = 0;
1653 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001654 else
Petr Viktorinffd97532020-02-11 17:46:57 +01001655 b = PyObject_CallMethodOneArg(self->encoder, _PyIO_str_encode, text);
Inada Naokibfba8c32019-05-16 15:03:20 +09001656
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001657 Py_DECREF(text);
1658 if (b == NULL)
1659 return NULL;
Inada Naokibfba8c32019-05-16 15:03:20 +09001660 if (b != text && !PyBytes_Check(b)) {
Oren Milmana5b4ea12017-08-25 21:14:54 +03001661 PyErr_Format(PyExc_TypeError,
1662 "encoder should return a bytes object, not '%.200s'",
1663 Py_TYPE(b)->tp_name);
1664 Py_DECREF(b);
1665 return NULL;
1666 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001667
Inada Naokibfba8c32019-05-16 15:03:20 +09001668 Py_ssize_t bytes_len;
1669 if (b == text) {
1670 bytes_len = PyUnicode_GET_LENGTH(b);
1671 }
1672 else {
1673 bytes_len = PyBytes_GET_SIZE(b);
1674 }
1675
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001676 if (self->pending_bytes == NULL) {
Inada Naokibfba8c32019-05-16 15:03:20 +09001677 self->pending_bytes_count = 0;
1678 self->pending_bytes = b;
1679 }
1680 else if (!PyList_CheckExact(self->pending_bytes)) {
1681 PyObject *list = PyList_New(2);
1682 if (list == NULL) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001683 Py_DECREF(b);
1684 return NULL;
1685 }
Inada Naokibfba8c32019-05-16 15:03:20 +09001686 PyList_SET_ITEM(list, 0, self->pending_bytes);
1687 PyList_SET_ITEM(list, 1, b);
1688 self->pending_bytes = list;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001689 }
Inada Naokibfba8c32019-05-16 15:03:20 +09001690 else {
1691 if (PyList_Append(self->pending_bytes, b) < 0) {
1692 Py_DECREF(b);
1693 return NULL;
1694 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001695 Py_DECREF(b);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001696 }
Inada Naokibfba8c32019-05-16 15:03:20 +09001697
1698 self->pending_bytes_count += bytes_len;
Antoine Pitrouc644e7c2014-05-09 00:24:50 +02001699 if (self->pending_bytes_count > self->chunk_size || needflush ||
1700 text_needflush) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001701 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001702 return NULL;
1703 }
Antoine Pitrou24f36292009-03-28 22:16:42 +00001704
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001705 if (needflush) {
Petr Viktorinffd97532020-02-11 17:46:57 +01001706 ret = PyObject_CallMethodNoArgs(self->buffer, _PyIO_str_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001707 if (ret == NULL)
1708 return NULL;
1709 Py_DECREF(ret);
1710 }
1711
Zackery Spytz23db9352018-06-29 04:14:58 -06001712 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001713 Py_CLEAR(self->snapshot);
1714
1715 if (self->decoder) {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001716 ret = _PyObject_CallMethodIdNoArgs(self->decoder, &PyId_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001717 if (ret == NULL)
1718 return NULL;
1719 Py_DECREF(ret);
1720 }
1721
1722 return PyLong_FromSsize_t(textlen);
1723}
1724
1725/* Steal a reference to chars and store it in the decoded_char buffer;
1726 */
1727static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001728textiowrapper_set_decoded_chars(textio *self, PyObject *chars)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001729{
Serhiy Storchaka48842712016-04-06 09:45:48 +03001730 Py_XSETREF(self->decoded_chars, chars);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001731 self->decoded_chars_used = 0;
1732}
1733
1734static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001735textiowrapper_get_decoded_chars(textio *self, Py_ssize_t n)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001736{
1737 PyObject *chars;
1738 Py_ssize_t avail;
1739
1740 if (self->decoded_chars == NULL)
1741 return PyUnicode_FromStringAndSize(NULL, 0);
1742
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001743 /* decoded_chars is guaranteed to be "ready". */
1744 avail = (PyUnicode_GET_LENGTH(self->decoded_chars)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001745 - self->decoded_chars_used);
1746
1747 assert(avail >= 0);
1748
1749 if (n < 0 || n > avail)
1750 n = avail;
1751
1752 if (self->decoded_chars_used > 0 || n < avail) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001753 chars = PyUnicode_Substring(self->decoded_chars,
1754 self->decoded_chars_used,
1755 self->decoded_chars_used + n);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001756 if (chars == NULL)
1757 return NULL;
1758 }
1759 else {
1760 chars = self->decoded_chars;
1761 Py_INCREF(chars);
1762 }
1763
1764 self->decoded_chars_used += n;
1765 return chars;
1766}
1767
1768/* Read and decode the next chunk of data from the BufferedReader.
1769 */
1770static int
Antoine Pitroue5324562011-11-19 00:39:01 +01001771textiowrapper_read_chunk(textio *self, Py_ssize_t size_hint)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001772{
1773 PyObject *dec_buffer = NULL;
1774 PyObject *dec_flags = NULL;
1775 PyObject *input_chunk = NULL;
Antoine Pitroub8503892014-04-29 10:14:02 +02001776 Py_buffer input_chunk_buf;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001777 PyObject *decoded_chars, *chunk_size;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001778 Py_ssize_t nbytes, nchars;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001779 int eof;
1780
1781 /* The return value is True unless EOF was reached. The decoded string is
1782 * placed in self._decoded_chars (replacing its previous value). The
1783 * entire input chunk is sent to the decoder, though some of it may remain
1784 * buffered in the decoder, yet to be converted.
1785 */
1786
1787 if (self->decoder == NULL) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001788 _unsupported("not readable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001789 return -1;
1790 }
1791
1792 if (self->telling) {
1793 /* To prepare for tell(), we need to snapshot a point in the file
1794 * where the decoder's input buffer is empty.
1795 */
Petr Viktorinffd97532020-02-11 17:46:57 +01001796 PyObject *state = PyObject_CallMethodNoArgs(self->decoder,
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001797 _PyIO_str_getstate);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001798 if (state == NULL)
1799 return -1;
1800 /* Given this, we know there was a valid snapshot point
1801 * len(dec_buffer) bytes ago with decoder state (b'', dec_flags).
1802 */
Oren Milmanba7d7362017-08-29 11:58:27 +03001803 if (!PyTuple_Check(state)) {
1804 PyErr_SetString(PyExc_TypeError,
1805 "illegal decoder state");
1806 Py_DECREF(state);
1807 return -1;
1808 }
1809 if (!PyArg_ParseTuple(state,
1810 "OO;illegal decoder state", &dec_buffer, &dec_flags))
1811 {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001812 Py_DECREF(state);
1813 return -1;
1814 }
Antoine Pitroub8503892014-04-29 10:14:02 +02001815
1816 if (!PyBytes_Check(dec_buffer)) {
1817 PyErr_Format(PyExc_TypeError,
Oren Milmanba7d7362017-08-29 11:58:27 +03001818 "illegal decoder state: the first item should be a "
1819 "bytes object, not '%.200s'",
Antoine Pitroub8503892014-04-29 10:14:02 +02001820 Py_TYPE(dec_buffer)->tp_name);
1821 Py_DECREF(state);
1822 return -1;
1823 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001824 Py_INCREF(dec_buffer);
1825 Py_INCREF(dec_flags);
1826 Py_DECREF(state);
1827 }
1828
1829 /* Read a chunk, decode it, and put the result in self._decoded_chars. */
Antoine Pitroue5324562011-11-19 00:39:01 +01001830 if (size_hint > 0) {
Victor Stinnerf8facac2011-11-22 02:30:47 +01001831 size_hint = (Py_ssize_t)(Py_MAX(self->b2cratio, 1.0) * size_hint);
Antoine Pitroue5324562011-11-19 00:39:01 +01001832 }
1833 chunk_size = PyLong_FromSsize_t(Py_MAX(self->chunk_size, size_hint));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001834 if (chunk_size == NULL)
1835 goto fail;
Antoine Pitroub8503892014-04-29 10:14:02 +02001836
Petr Viktorinffd97532020-02-11 17:46:57 +01001837 input_chunk = PyObject_CallMethodOneArg(self->buffer,
Antoine Pitroue96ec682011-07-23 21:46:35 +02001838 (self->has_read1 ? _PyIO_str_read1: _PyIO_str_read),
Jeroen Demeyer59ad1102019-07-11 10:59:05 +02001839 chunk_size);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001840 Py_DECREF(chunk_size);
1841 if (input_chunk == NULL)
1842 goto fail;
Antoine Pitroub8503892014-04-29 10:14:02 +02001843
1844 if (PyObject_GetBuffer(input_chunk, &input_chunk_buf, 0) != 0) {
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001845 PyErr_Format(PyExc_TypeError,
Antoine Pitroub8503892014-04-29 10:14:02 +02001846 "underlying %s() should have returned a bytes-like object, "
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001847 "not '%.200s'", (self->has_read1 ? "read1": "read"),
1848 Py_TYPE(input_chunk)->tp_name);
1849 goto fail;
1850 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001851
Antoine Pitroub8503892014-04-29 10:14:02 +02001852 nbytes = input_chunk_buf.len;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001853 eof = (nbytes == 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001854
INADA Naoki507434f2017-12-21 09:59:53 +09001855 decoded_chars = _textiowrapper_decode(self->decoder, input_chunk, eof);
1856 PyBuffer_Release(&input_chunk_buf);
1857 if (decoded_chars == NULL)
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001858 goto fail;
INADA Naoki507434f2017-12-21 09:59:53 +09001859
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001860 textiowrapper_set_decoded_chars(self, decoded_chars);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001861 nchars = PyUnicode_GET_LENGTH(decoded_chars);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001862 if (nchars > 0)
1863 self->b2cratio = (double) nbytes / nchars;
1864 else
1865 self->b2cratio = 0.0;
1866 if (nchars > 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001867 eof = 0;
1868
1869 if (self->telling) {
1870 /* At the snapshot point, len(dec_buffer) bytes before the read, the
1871 * next input to be decoded is dec_buffer + input_chunk.
1872 */
Antoine Pitroub8503892014-04-29 10:14:02 +02001873 PyObject *next_input = dec_buffer;
1874 PyBytes_Concat(&next_input, input_chunk);
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03001875 dec_buffer = NULL; /* Reference lost to PyBytes_Concat */
Antoine Pitroub8503892014-04-29 10:14:02 +02001876 if (next_input == NULL) {
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001877 goto fail;
1878 }
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03001879 PyObject *snapshot = Py_BuildValue("NN", dec_flags, next_input);
1880 if (snapshot == NULL) {
1881 dec_flags = NULL;
1882 goto fail;
1883 }
1884 Py_XSETREF(self->snapshot, snapshot);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001885 }
1886 Py_DECREF(input_chunk);
1887
1888 return (eof == 0);
1889
1890 fail:
1891 Py_XDECREF(dec_buffer);
1892 Py_XDECREF(dec_flags);
1893 Py_XDECREF(input_chunk);
1894 return -1;
1895}
1896
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001897/*[clinic input]
1898_io.TextIOWrapper.read
Serhiy Storchaka762bf402017-03-30 09:15:31 +03001899 size as n: Py_ssize_t(accept={int, NoneType}) = -1
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001900 /
1901[clinic start generated code]*/
1902
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001903static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001904_io_TextIOWrapper_read_impl(textio *self, Py_ssize_t n)
Serhiy Storchaka762bf402017-03-30 09:15:31 +03001905/*[clinic end generated code: output=7e651ce6cc6a25a6 input=123eecbfe214aeb8]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001906{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001907 PyObject *result = NULL, *chunks = NULL;
1908
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001909 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001910 CHECK_CLOSED(self);
1911
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001912 if (self->decoder == NULL)
1913 return _unsupported("not readable");
Benjamin Petersona1b49012009-03-31 23:11:32 +00001914
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001915 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001916 return NULL;
1917
1918 if (n < 0) {
1919 /* Read everything */
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001920 PyObject *bytes = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_read);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001921 PyObject *decoded;
1922 if (bytes == NULL)
1923 goto fail;
Victor Stinnerfd821132011-05-25 22:01:33 +02001924
Andy Lesterdffe4c02020-03-04 07:15:20 -06001925 if (Py_IS_TYPE(self->decoder, &PyIncrementalNewlineDecoder_Type))
Victor Stinnerfd821132011-05-25 22:01:33 +02001926 decoded = _PyIncrementalNewlineDecoder_decode(self->decoder,
1927 bytes, 1);
1928 else
1929 decoded = PyObject_CallMethodObjArgs(
1930 self->decoder, _PyIO_str_decode, bytes, Py_True, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001931 Py_DECREF(bytes);
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001932 if (check_decoded(decoded) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001933 goto fail;
1934
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001935 result = textiowrapper_get_decoded_chars(self, -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001936
1937 if (result == NULL) {
1938 Py_DECREF(decoded);
1939 return NULL;
1940 }
1941
1942 PyUnicode_AppendAndDel(&result, decoded);
1943 if (result == NULL)
1944 goto fail;
1945
Zackery Spytz23db9352018-06-29 04:14:58 -06001946 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001947 Py_CLEAR(self->snapshot);
1948 return result;
1949 }
1950 else {
1951 int res = 1;
1952 Py_ssize_t remaining = n;
1953
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001954 result = textiowrapper_get_decoded_chars(self, n);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001955 if (result == NULL)
1956 goto fail;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001957 if (PyUnicode_READY(result) == -1)
1958 goto fail;
1959 remaining -= PyUnicode_GET_LENGTH(result);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001960
1961 /* Keep reading chunks until we have n characters to return */
1962 while (remaining > 0) {
Antoine Pitroue5324562011-11-19 00:39:01 +01001963 res = textiowrapper_read_chunk(self, remaining);
Gregory P. Smith51359922012-06-23 23:55:39 -07001964 if (res < 0) {
1965 /* NOTE: PyErr_SetFromErrno() calls PyErr_CheckSignals()
1966 when EINTR occurs so we needn't do it ourselves. */
1967 if (_PyIO_trap_eintr()) {
1968 continue;
1969 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001970 goto fail;
Gregory P. Smith51359922012-06-23 23:55:39 -07001971 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001972 if (res == 0) /* EOF */
1973 break;
1974 if (chunks == NULL) {
1975 chunks = PyList_New(0);
1976 if (chunks == NULL)
1977 goto fail;
1978 }
Antoine Pitroue5324562011-11-19 00:39:01 +01001979 if (PyUnicode_GET_LENGTH(result) > 0 &&
1980 PyList_Append(chunks, result) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001981 goto fail;
1982 Py_DECREF(result);
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001983 result = textiowrapper_get_decoded_chars(self, remaining);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001984 if (result == NULL)
1985 goto fail;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001986 remaining -= PyUnicode_GET_LENGTH(result);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001987 }
1988 if (chunks != NULL) {
1989 if (result != NULL && PyList_Append(chunks, result) < 0)
1990 goto fail;
Serhiy Storchaka48842712016-04-06 09:45:48 +03001991 Py_XSETREF(result, PyUnicode_Join(_PyIO_empty_str, chunks));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001992 if (result == NULL)
1993 goto fail;
1994 Py_CLEAR(chunks);
1995 }
1996 return result;
1997 }
1998 fail:
1999 Py_XDECREF(result);
2000 Py_XDECREF(chunks);
2001 return NULL;
2002}
2003
2004
Victor Stinnerf7b8cb62011-09-29 03:28:17 +02002005/* NOTE: `end` must point to the real end of the Py_UCS4 storage,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002006 that is to the NUL character. Otherwise the function will produce
2007 incorrect results. */
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002008static const char *
2009find_control_char(int kind, const char *s, const char *end, Py_UCS4 ch)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002010{
Antoine Pitrouc28e2e52011-11-13 03:53:42 +01002011 if (kind == PyUnicode_1BYTE_KIND) {
2012 assert(ch < 256);
Andy Lestere6be9b52020-02-11 20:28:35 -06002013 return (char *) memchr((const void *) s, (char) ch, end - s);
Antoine Pitrouc28e2e52011-11-13 03:53:42 +01002014 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002015 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002016 while (PyUnicode_READ(kind, s, 0) > ch)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002017 s += kind;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002018 if (PyUnicode_READ(kind, s, 0) == ch)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002019 return s;
2020 if (s == end)
2021 return NULL;
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002022 s += kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002023 }
2024}
2025
2026Py_ssize_t
2027_PyIO_find_line_ending(
2028 int translated, int universal, PyObject *readnl,
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002029 int kind, const char *start, const char *end, Py_ssize_t *consumed)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002030{
Andy Lestere6be9b52020-02-11 20:28:35 -06002031 Py_ssize_t len = (end - start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002032
2033 if (translated) {
2034 /* Newlines are already translated, only search for \n */
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002035 const char *pos = find_control_char(kind, start, end, '\n');
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002036 if (pos != NULL)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002037 return (pos - start)/kind + 1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002038 else {
2039 *consumed = len;
2040 return -1;
2041 }
2042 }
2043 else if (universal) {
2044 /* Universal newline search. Find any of \r, \r\n, \n
2045 * The decoder ensures that \r\n are not split in two pieces
2046 */
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002047 const char *s = start;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002048 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002049 Py_UCS4 ch;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002050 /* Fast path for non-control chars. The loop always ends
Victor Stinnerf7b8cb62011-09-29 03:28:17 +02002051 since the Unicode string is NUL-terminated. */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002052 while (PyUnicode_READ(kind, s, 0) > '\r')
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002053 s += kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002054 if (s >= end) {
2055 *consumed = len;
2056 return -1;
2057 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002058 ch = PyUnicode_READ(kind, s, 0);
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002059 s += kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002060 if (ch == '\n')
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002061 return (s - start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002062 if (ch == '\r') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002063 if (PyUnicode_READ(kind, s, 0) == '\n')
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002064 return (s - start)/kind + 1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002065 else
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002066 return (s - start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002067 }
2068 }
2069 }
2070 else {
2071 /* Non-universal mode. */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002072 Py_ssize_t readnl_len = PyUnicode_GET_LENGTH(readnl);
Serhiy Storchakacd8295f2020-04-11 10:48:40 +03002073 const Py_UCS1 *nl = PyUnicode_1BYTE_DATA(readnl);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002074 /* Assume that readnl is an ASCII character. */
2075 assert(PyUnicode_KIND(readnl) == PyUnicode_1BYTE_KIND);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002076 if (readnl_len == 1) {
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002077 const char *pos = find_control_char(kind, start, end, nl[0]);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002078 if (pos != NULL)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002079 return (pos - start)/kind + 1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002080 *consumed = len;
2081 return -1;
2082 }
2083 else {
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002084 const char *s = start;
2085 const char *e = end - (readnl_len - 1)*kind;
2086 const char *pos;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002087 if (e < s)
2088 e = s;
2089 while (s < e) {
2090 Py_ssize_t i;
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002091 const char *pos = find_control_char(kind, s, end, nl[0]);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002092 if (pos == NULL || pos >= e)
2093 break;
2094 for (i = 1; i < readnl_len; i++) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002095 if (PyUnicode_READ(kind, pos, i) != nl[i])
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002096 break;
2097 }
2098 if (i == readnl_len)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002099 return (pos - start)/kind + readnl_len;
2100 s = pos + kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002101 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002102 pos = find_control_char(kind, e, end, nl[0]);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002103 if (pos == NULL)
2104 *consumed = len;
2105 else
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002106 *consumed = (pos - start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002107 return -1;
2108 }
2109 }
2110}
2111
2112static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002113_textiowrapper_readline(textio *self, Py_ssize_t limit)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002114{
2115 PyObject *line = NULL, *chunks = NULL, *remaining = NULL;
2116 Py_ssize_t start, endpos, chunked, offset_to_buffer;
2117 int res;
2118
2119 CHECK_CLOSED(self);
2120
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002121 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002122 return NULL;
2123
2124 chunked = 0;
2125
2126 while (1) {
Serhiy Storchakacd8295f2020-04-11 10:48:40 +03002127 const char *ptr;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002128 Py_ssize_t line_len;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002129 int kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002130 Py_ssize_t consumed = 0;
2131
2132 /* First, get some data if necessary */
2133 res = 1;
2134 while (!self->decoded_chars ||
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002135 !PyUnicode_GET_LENGTH(self->decoded_chars)) {
Antoine Pitroue5324562011-11-19 00:39:01 +01002136 res = textiowrapper_read_chunk(self, 0);
Gregory P. Smith51359922012-06-23 23:55:39 -07002137 if (res < 0) {
2138 /* NOTE: PyErr_SetFromErrno() calls PyErr_CheckSignals()
2139 when EINTR occurs so we needn't do it ourselves. */
2140 if (_PyIO_trap_eintr()) {
2141 continue;
2142 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002143 goto error;
Gregory P. Smith51359922012-06-23 23:55:39 -07002144 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002145 if (res == 0)
2146 break;
2147 }
2148 if (res == 0) {
2149 /* end of file */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002150 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002151 Py_CLEAR(self->snapshot);
2152 start = endpos = offset_to_buffer = 0;
2153 break;
2154 }
2155
2156 if (remaining == NULL) {
2157 line = self->decoded_chars;
2158 start = self->decoded_chars_used;
2159 offset_to_buffer = 0;
2160 Py_INCREF(line);
2161 }
2162 else {
2163 assert(self->decoded_chars_used == 0);
2164 line = PyUnicode_Concat(remaining, self->decoded_chars);
2165 start = 0;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002166 offset_to_buffer = PyUnicode_GET_LENGTH(remaining);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002167 Py_CLEAR(remaining);
2168 if (line == NULL)
2169 goto error;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002170 if (PyUnicode_READY(line) == -1)
2171 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002172 }
2173
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002174 ptr = PyUnicode_DATA(line);
2175 line_len = PyUnicode_GET_LENGTH(line);
2176 kind = PyUnicode_KIND(line);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002177
2178 endpos = _PyIO_find_line_ending(
2179 self->readtranslate, self->readuniversal, self->readnl,
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002180 kind,
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002181 ptr + kind * start,
2182 ptr + kind * line_len,
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002183 &consumed);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002184 if (endpos >= 0) {
2185 endpos += start;
2186 if (limit >= 0 && (endpos - start) + chunked >= limit)
2187 endpos = start + limit - chunked;
2188 break;
2189 }
2190
2191 /* We can put aside up to `endpos` */
2192 endpos = consumed + start;
2193 if (limit >= 0 && (endpos - start) + chunked >= limit) {
2194 /* Didn't find line ending, but reached length limit */
2195 endpos = start + limit - chunked;
2196 break;
2197 }
2198
2199 if (endpos > start) {
2200 /* No line ending seen yet - put aside current data */
2201 PyObject *s;
2202 if (chunks == NULL) {
2203 chunks = PyList_New(0);
2204 if (chunks == NULL)
2205 goto error;
2206 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002207 s = PyUnicode_Substring(line, start, endpos);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002208 if (s == NULL)
2209 goto error;
2210 if (PyList_Append(chunks, s) < 0) {
2211 Py_DECREF(s);
2212 goto error;
2213 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002214 chunked += PyUnicode_GET_LENGTH(s);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002215 Py_DECREF(s);
2216 }
2217 /* There may be some remaining bytes we'll have to prepend to the
2218 next chunk of data */
2219 if (endpos < line_len) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002220 remaining = PyUnicode_Substring(line, endpos, line_len);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002221 if (remaining == NULL)
2222 goto error;
2223 }
2224 Py_CLEAR(line);
2225 /* We have consumed the buffer */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002226 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002227 }
2228
2229 if (line != NULL) {
2230 /* Our line ends in the current buffer */
2231 self->decoded_chars_used = endpos - offset_to_buffer;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002232 if (start > 0 || endpos < PyUnicode_GET_LENGTH(line)) {
2233 PyObject *s = PyUnicode_Substring(line, start, endpos);
2234 Py_CLEAR(line);
2235 if (s == NULL)
2236 goto error;
2237 line = s;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002238 }
2239 }
2240 if (remaining != NULL) {
2241 if (chunks == NULL) {
2242 chunks = PyList_New(0);
2243 if (chunks == NULL)
2244 goto error;
2245 }
2246 if (PyList_Append(chunks, remaining) < 0)
2247 goto error;
2248 Py_CLEAR(remaining);
2249 }
2250 if (chunks != NULL) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002251 if (line != NULL) {
2252 if (PyList_Append(chunks, line) < 0)
2253 goto error;
2254 Py_DECREF(line);
2255 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002256 line = PyUnicode_Join(_PyIO_empty_str, chunks);
2257 if (line == NULL)
2258 goto error;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002259 Py_CLEAR(chunks);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002260 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002261 if (line == NULL) {
2262 Py_INCREF(_PyIO_empty_str);
2263 line = _PyIO_empty_str;
2264 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002265
2266 return line;
2267
2268 error:
2269 Py_XDECREF(chunks);
2270 Py_XDECREF(remaining);
2271 Py_XDECREF(line);
2272 return NULL;
2273}
2274
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002275/*[clinic input]
2276_io.TextIOWrapper.readline
2277 size: Py_ssize_t = -1
2278 /
2279[clinic start generated code]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002280
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002281static PyObject *
2282_io_TextIOWrapper_readline_impl(textio *self, Py_ssize_t size)
2283/*[clinic end generated code: output=344afa98804e8b25 input=56c7172483b36db6]*/
2284{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002285 CHECK_ATTACHED(self);
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002286 return _textiowrapper_readline(self, size);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002287}
2288
2289/* Seek and Tell */
2290
2291typedef struct {
2292 Py_off_t start_pos;
2293 int dec_flags;
2294 int bytes_to_feed;
2295 int chars_to_skip;
2296 char need_eof;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002297} cookie_type;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002298
2299/*
2300 To speed up cookie packing/unpacking, we store the fields in a temporary
2301 string and call _PyLong_FromByteArray() or _PyLong_AsByteArray (resp.).
2302 The following macros define at which offsets in the intermediary byte
2303 string the various CookieStruct fields will be stored.
2304 */
2305
2306#define COOKIE_BUF_LEN (sizeof(Py_off_t) + 3 * sizeof(int) + sizeof(char))
2307
Christian Heimes743e0cd2012-10-17 23:52:17 +02002308#if PY_BIG_ENDIAN
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002309/* We want the least significant byte of start_pos to also be the least
2310 significant byte of the cookie, which means that in big-endian mode we
2311 must copy the fields in reverse order. */
2312
2313# define OFF_START_POS (sizeof(char) + 3 * sizeof(int))
2314# define OFF_DEC_FLAGS (sizeof(char) + 2 * sizeof(int))
2315# define OFF_BYTES_TO_FEED (sizeof(char) + sizeof(int))
2316# define OFF_CHARS_TO_SKIP (sizeof(char))
2317# define OFF_NEED_EOF 0
2318
2319#else
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002320/* Little-endian mode: the least significant byte of start_pos will
2321 naturally end up the least significant byte of the cookie. */
2322
2323# define OFF_START_POS 0
2324# define OFF_DEC_FLAGS (sizeof(Py_off_t))
2325# define OFF_BYTES_TO_FEED (sizeof(Py_off_t) + sizeof(int))
2326# define OFF_CHARS_TO_SKIP (sizeof(Py_off_t) + 2 * sizeof(int))
2327# define OFF_NEED_EOF (sizeof(Py_off_t) + 3 * sizeof(int))
2328
2329#endif
2330
2331static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002332textiowrapper_parse_cookie(cookie_type *cookie, PyObject *cookieObj)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002333{
2334 unsigned char buffer[COOKIE_BUF_LEN];
2335 PyLongObject *cookieLong = (PyLongObject *)PyNumber_Long(cookieObj);
2336 if (cookieLong == NULL)
2337 return -1;
2338
2339 if (_PyLong_AsByteArray(cookieLong, buffer, sizeof(buffer),
Christian Heimes743e0cd2012-10-17 23:52:17 +02002340 PY_LITTLE_ENDIAN, 0) < 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002341 Py_DECREF(cookieLong);
2342 return -1;
2343 }
2344 Py_DECREF(cookieLong);
2345
Antoine Pitrou2db74c22009-03-06 21:49:02 +00002346 memcpy(&cookie->start_pos, buffer + OFF_START_POS, sizeof(cookie->start_pos));
2347 memcpy(&cookie->dec_flags, buffer + OFF_DEC_FLAGS, sizeof(cookie->dec_flags));
2348 memcpy(&cookie->bytes_to_feed, buffer + OFF_BYTES_TO_FEED, sizeof(cookie->bytes_to_feed));
2349 memcpy(&cookie->chars_to_skip, buffer + OFF_CHARS_TO_SKIP, sizeof(cookie->chars_to_skip));
2350 memcpy(&cookie->need_eof, buffer + OFF_NEED_EOF, sizeof(cookie->need_eof));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002351
2352 return 0;
2353}
2354
2355static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002356textiowrapper_build_cookie(cookie_type *cookie)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002357{
2358 unsigned char buffer[COOKIE_BUF_LEN];
2359
Antoine Pitrou2db74c22009-03-06 21:49:02 +00002360 memcpy(buffer + OFF_START_POS, &cookie->start_pos, sizeof(cookie->start_pos));
2361 memcpy(buffer + OFF_DEC_FLAGS, &cookie->dec_flags, sizeof(cookie->dec_flags));
2362 memcpy(buffer + OFF_BYTES_TO_FEED, &cookie->bytes_to_feed, sizeof(cookie->bytes_to_feed));
2363 memcpy(buffer + OFF_CHARS_TO_SKIP, &cookie->chars_to_skip, sizeof(cookie->chars_to_skip));
2364 memcpy(buffer + OFF_NEED_EOF, &cookie->need_eof, sizeof(cookie->need_eof));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002365
Christian Heimes743e0cd2012-10-17 23:52:17 +02002366 return _PyLong_FromByteArray(buffer, sizeof(buffer),
2367 PY_LITTLE_ENDIAN, 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002368}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002369
2370static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002371_textiowrapper_decoder_setstate(textio *self, cookie_type *cookie)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002372{
2373 PyObject *res;
2374 /* When seeking to the start of the stream, we call decoder.reset()
2375 rather than decoder.getstate().
2376 This is for a few decoders such as utf-16 for which the state value
2377 at start is not (b"", 0) but e.g. (b"", 2) (meaning, in the case of
2378 utf-16, that we are expecting a BOM).
2379 */
2380 if (cookie->start_pos == 0 && cookie->dec_flags == 0)
Petr Viktorinffd97532020-02-11 17:46:57 +01002381 res = PyObject_CallMethodNoArgs(self->decoder, _PyIO_str_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002382 else
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002383 res = _PyObject_CallMethodId(self->decoder, &PyId_setstate,
2384 "((yi))", "", cookie->dec_flags);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002385 if (res == NULL)
2386 return -1;
2387 Py_DECREF(res);
2388 return 0;
2389}
2390
Antoine Pitroue4501852009-05-14 18:55:55 +00002391static int
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002392_textiowrapper_encoder_reset(textio *self, int start_of_stream)
Antoine Pitroue4501852009-05-14 18:55:55 +00002393{
2394 PyObject *res;
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002395 if (start_of_stream) {
Petr Viktorinffd97532020-02-11 17:46:57 +01002396 res = PyObject_CallMethodNoArgs(self->encoder, _PyIO_str_reset);
Antoine Pitroue4501852009-05-14 18:55:55 +00002397 self->encoding_start_of_stream = 1;
2398 }
2399 else {
Petr Viktorinffd97532020-02-11 17:46:57 +01002400 res = PyObject_CallMethodOneArg(self->encoder, _PyIO_str_setstate,
Victor Stinner37834132020-10-27 17:12:53 +01002401 _PyLong_GetZero());
Antoine Pitroue4501852009-05-14 18:55:55 +00002402 self->encoding_start_of_stream = 0;
2403 }
2404 if (res == NULL)
2405 return -1;
2406 Py_DECREF(res);
2407 return 0;
2408}
2409
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002410static int
2411_textiowrapper_encoder_setstate(textio *self, cookie_type *cookie)
2412{
2413 /* Same as _textiowrapper_decoder_setstate() above. */
2414 return _textiowrapper_encoder_reset(
2415 self, cookie->start_pos == 0 && cookie->dec_flags == 0);
2416}
2417
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002418/*[clinic input]
2419_io.TextIOWrapper.seek
2420 cookie as cookieObj: object
2421 whence: int = 0
2422 /
2423[clinic start generated code]*/
2424
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002425static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002426_io_TextIOWrapper_seek_impl(textio *self, PyObject *cookieObj, int whence)
2427/*[clinic end generated code: output=0a15679764e2d04d input=0458abeb3d7842be]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002428{
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002429 PyObject *posobj;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002430 cookie_type cookie;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002431 PyObject *res;
2432 int cmp;
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002433 PyObject *snapshot;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002434
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002435 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002436 CHECK_CLOSED(self);
2437
2438 Py_INCREF(cookieObj);
2439
2440 if (!self->seekable) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002441 _unsupported("underlying stream is not seekable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002442 goto fail;
2443 }
2444
Victor Stinner37834132020-10-27 17:12:53 +01002445 PyObject *zero = _PyLong_GetZero(); // borrowed reference
2446
ngie-eign848037c2019-03-02 23:28:26 -08002447 switch (whence) {
2448 case SEEK_CUR:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002449 /* seek relative to current position */
Victor Stinner37834132020-10-27 17:12:53 +01002450 cmp = PyObject_RichCompareBool(cookieObj, zero, Py_EQ);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002451 if (cmp < 0)
2452 goto fail;
2453
2454 if (cmp == 0) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002455 _unsupported("can't do nonzero cur-relative seeks");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002456 goto fail;
2457 }
2458
2459 /* Seeking to the current position should attempt to
2460 * sync the underlying buffer with the current position.
2461 */
2462 Py_DECREF(cookieObj);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002463 cookieObj = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_tell);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002464 if (cookieObj == NULL)
2465 goto fail;
Inada Naoki8c17d922019-03-04 01:22:39 +09002466 break;
2467
ngie-eign848037c2019-03-02 23:28:26 -08002468 case SEEK_END:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002469 /* seek relative to end of file */
Victor Stinner37834132020-10-27 17:12:53 +01002470 cmp = PyObject_RichCompareBool(cookieObj, zero, Py_EQ);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002471 if (cmp < 0)
2472 goto fail;
2473
2474 if (cmp == 0) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002475 _unsupported("can't do nonzero end-relative seeks");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002476 goto fail;
2477 }
2478
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002479 res = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002480 if (res == NULL)
2481 goto fail;
2482 Py_DECREF(res);
2483
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002484 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002485 Py_CLEAR(self->snapshot);
2486 if (self->decoder) {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002487 res = _PyObject_CallMethodIdNoArgs(self->decoder, &PyId_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002488 if (res == NULL)
2489 goto fail;
2490 Py_DECREF(res);
2491 }
2492
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002493 res = _PyObject_CallMethodId(self->buffer, &PyId_seek, "ii", 0, 2);
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002494 Py_CLEAR(cookieObj);
2495 if (res == NULL)
2496 goto fail;
2497 if (self->encoder) {
2498 /* If seek() == 0, we are at the start of stream, otherwise not */
Victor Stinner37834132020-10-27 17:12:53 +01002499 cmp = PyObject_RichCompareBool(res, zero, Py_EQ);
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002500 if (cmp < 0 || _textiowrapper_encoder_reset(self, cmp)) {
2501 Py_DECREF(res);
2502 goto fail;
2503 }
2504 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002505 return res;
Inada Naoki8c17d922019-03-04 01:22:39 +09002506
ngie-eign848037c2019-03-02 23:28:26 -08002507 case SEEK_SET:
2508 break;
Inada Naoki8c17d922019-03-04 01:22:39 +09002509
ngie-eign848037c2019-03-02 23:28:26 -08002510 default:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002511 PyErr_Format(PyExc_ValueError,
ngie-eign848037c2019-03-02 23:28:26 -08002512 "invalid whence (%d, should be %d, %d or %d)", whence,
2513 SEEK_SET, SEEK_CUR, SEEK_END);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002514 goto fail;
2515 }
2516
Victor Stinner37834132020-10-27 17:12:53 +01002517 cmp = PyObject_RichCompareBool(cookieObj, zero, Py_LT);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002518 if (cmp < 0)
2519 goto fail;
2520
2521 if (cmp == 1) {
2522 PyErr_Format(PyExc_ValueError,
2523 "negative seek position %R", cookieObj);
2524 goto fail;
2525 }
2526
Petr Viktorinffd97532020-02-11 17:46:57 +01002527 res = PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002528 if (res == NULL)
2529 goto fail;
2530 Py_DECREF(res);
2531
2532 /* The strategy of seek() is to go back to the safe start point
2533 * and replay the effect of read(chars_to_skip) from there.
2534 */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002535 if (textiowrapper_parse_cookie(&cookie, cookieObj) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002536 goto fail;
2537
2538 /* Seek back to the safe start point. */
2539 posobj = PyLong_FromOff_t(cookie.start_pos);
2540 if (posobj == NULL)
2541 goto fail;
Petr Viktorinffd97532020-02-11 17:46:57 +01002542 res = PyObject_CallMethodOneArg(self->buffer, _PyIO_str_seek, posobj);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002543 Py_DECREF(posobj);
2544 if (res == NULL)
2545 goto fail;
2546 Py_DECREF(res);
2547
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002548 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002549 Py_CLEAR(self->snapshot);
2550
2551 /* Restore the decoder to its state from the safe start point. */
2552 if (self->decoder) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002553 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002554 goto fail;
2555 }
2556
2557 if (cookie.chars_to_skip) {
2558 /* Just like _read_chunk, feed the decoder and save a snapshot. */
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002559 PyObject *input_chunk = _PyObject_CallMethodId(
2560 self->buffer, &PyId_read, "i", cookie.bytes_to_feed);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002561 PyObject *decoded;
2562
2563 if (input_chunk == NULL)
2564 goto fail;
2565
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02002566 if (!PyBytes_Check(input_chunk)) {
2567 PyErr_Format(PyExc_TypeError,
2568 "underlying read() should have returned a bytes "
2569 "object, not '%.200s'",
2570 Py_TYPE(input_chunk)->tp_name);
2571 Py_DECREF(input_chunk);
2572 goto fail;
2573 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002574
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002575 snapshot = Py_BuildValue("iN", cookie.dec_flags, input_chunk);
2576 if (snapshot == NULL) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002577 goto fail;
2578 }
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002579 Py_XSETREF(self->snapshot, snapshot);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002580
Serhiy Storchaka1f21eaa2019-09-01 12:16:51 +03002581 decoded = _PyObject_CallMethodIdObjArgs(self->decoder, &PyId_decode,
2582 input_chunk, cookie.need_eof ? Py_True : Py_False, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002583
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02002584 if (check_decoded(decoded) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002585 goto fail;
2586
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002587 textiowrapper_set_decoded_chars(self, decoded);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002588
2589 /* Skip chars_to_skip of the decoded characters. */
Victor Stinner9e30aa52011-11-21 02:49:52 +01002590 if (PyUnicode_GetLength(self->decoded_chars) < cookie.chars_to_skip) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03002591 PyErr_SetString(PyExc_OSError, "can't restore logical file position");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002592 goto fail;
2593 }
2594 self->decoded_chars_used = cookie.chars_to_skip;
2595 }
2596 else {
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002597 snapshot = Py_BuildValue("iy", cookie.dec_flags, "");
2598 if (snapshot == NULL)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002599 goto fail;
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002600 Py_XSETREF(self->snapshot, snapshot);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002601 }
2602
Antoine Pitroue4501852009-05-14 18:55:55 +00002603 /* Finally, reset the encoder (merely useful for proper BOM handling) */
2604 if (self->encoder) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002605 if (_textiowrapper_encoder_setstate(self, &cookie) < 0)
Antoine Pitroue4501852009-05-14 18:55:55 +00002606 goto fail;
2607 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002608 return cookieObj;
2609 fail:
2610 Py_XDECREF(cookieObj);
2611 return NULL;
2612
2613}
2614
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002615/*[clinic input]
2616_io.TextIOWrapper.tell
2617[clinic start generated code]*/
2618
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002619static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002620_io_TextIOWrapper_tell_impl(textio *self)
2621/*[clinic end generated code: output=4f168c08bf34ad5f input=9a2caf88c24f9ddf]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002622{
2623 PyObject *res;
2624 PyObject *posobj = NULL;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002625 cookie_type cookie = {0,0,0,0,0};
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002626 PyObject *next_input;
2627 Py_ssize_t chars_to_skip, chars_decoded;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002628 Py_ssize_t skip_bytes, skip_back;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002629 PyObject *saved_state = NULL;
Serhiy Storchaka8f87eef2020-04-12 14:58:27 +03002630 const char *input, *input_end;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002631 Py_ssize_t dec_buffer_len;
2632 int dec_flags;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002633
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002634 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002635 CHECK_CLOSED(self);
2636
2637 if (!self->seekable) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002638 _unsupported("underlying stream is not seekable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002639 goto fail;
2640 }
2641 if (!self->telling) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03002642 PyErr_SetString(PyExc_OSError,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002643 "telling position disabled by next() call");
2644 goto fail;
2645 }
2646
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002647 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002648 return NULL;
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002649 res = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002650 if (res == NULL)
2651 goto fail;
2652 Py_DECREF(res);
2653
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002654 posobj = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_tell);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002655 if (posobj == NULL)
2656 goto fail;
2657
2658 if (self->decoder == NULL || self->snapshot == NULL) {
Victor Stinner9e30aa52011-11-21 02:49:52 +01002659 assert (self->decoded_chars == NULL || PyUnicode_GetLength(self->decoded_chars) == 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002660 return posobj;
2661 }
2662
2663#if defined(HAVE_LARGEFILE_SUPPORT)
2664 cookie.start_pos = PyLong_AsLongLong(posobj);
2665#else
2666 cookie.start_pos = PyLong_AsLong(posobj);
2667#endif
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002668 Py_DECREF(posobj);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002669 if (PyErr_Occurred())
2670 goto fail;
2671
2672 /* Skip backward to the snapshot point (see _read_chunk). */
Oren Milman13614e32017-08-24 19:51:24 +03002673 assert(PyTuple_Check(self->snapshot));
Serhiy Storchakabb72c472015-04-19 20:38:19 +03002674 if (!PyArg_ParseTuple(self->snapshot, "iO", &cookie.dec_flags, &next_input))
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002675 goto fail;
2676
2677 assert (PyBytes_Check(next_input));
2678
2679 cookie.start_pos -= PyBytes_GET_SIZE(next_input);
2680
2681 /* How many decoded characters have been used up since the snapshot? */
2682 if (self->decoded_chars_used == 0) {
2683 /* We haven't moved from the snapshot point. */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002684 return textiowrapper_build_cookie(&cookie);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002685 }
2686
2687 chars_to_skip = self->decoded_chars_used;
2688
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002689 /* Decoder state will be restored at the end */
Petr Viktorinffd97532020-02-11 17:46:57 +01002690 saved_state = PyObject_CallMethodNoArgs(self->decoder,
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002691 _PyIO_str_getstate);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002692 if (saved_state == NULL)
2693 goto fail;
2694
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002695#define DECODER_GETSTATE() do { \
Serhiy Storchaka008d88b2015-05-06 09:53:07 +03002696 PyObject *dec_buffer; \
Petr Viktorinffd97532020-02-11 17:46:57 +01002697 PyObject *_state = PyObject_CallMethodNoArgs(self->decoder, \
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002698 _PyIO_str_getstate); \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002699 if (_state == NULL) \
2700 goto fail; \
Oren Milman13614e32017-08-24 19:51:24 +03002701 if (!PyTuple_Check(_state)) { \
2702 PyErr_SetString(PyExc_TypeError, \
2703 "illegal decoder state"); \
2704 Py_DECREF(_state); \
2705 goto fail; \
2706 } \
2707 if (!PyArg_ParseTuple(_state, "Oi;illegal decoder state", \
2708 &dec_buffer, &dec_flags)) \
2709 { \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002710 Py_DECREF(_state); \
2711 goto fail; \
2712 } \
Serhiy Storchaka008d88b2015-05-06 09:53:07 +03002713 if (!PyBytes_Check(dec_buffer)) { \
2714 PyErr_Format(PyExc_TypeError, \
Oren Milmanba7d7362017-08-29 11:58:27 +03002715 "illegal decoder state: the first item should be a " \
2716 "bytes object, not '%.200s'", \
Serhiy Storchaka008d88b2015-05-06 09:53:07 +03002717 Py_TYPE(dec_buffer)->tp_name); \
2718 Py_DECREF(_state); \
2719 goto fail; \
2720 } \
2721 dec_buffer_len = PyBytes_GET_SIZE(dec_buffer); \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002722 Py_DECREF(_state); \
2723 } while (0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002724
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002725#define DECODER_DECODE(start, len, res) do { \
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002726 PyObject *_decoded = _PyObject_CallMethodId( \
2727 self->decoder, &PyId_decode, "y#", start, len); \
Serhiy Storchakad03ce4a2013-02-03 17:07:32 +02002728 if (check_decoded(_decoded) < 0) \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002729 goto fail; \
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002730 res = PyUnicode_GET_LENGTH(_decoded); \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002731 Py_DECREF(_decoded); \
2732 } while (0)
2733
2734 /* Fast search for an acceptable start point, close to our
2735 current pos */
2736 skip_bytes = (Py_ssize_t) (self->b2cratio * chars_to_skip);
2737 skip_back = 1;
2738 assert(skip_back <= PyBytes_GET_SIZE(next_input));
2739 input = PyBytes_AS_STRING(next_input);
2740 while (skip_bytes > 0) {
2741 /* Decode up to temptative start point */
2742 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
2743 goto fail;
2744 DECODER_DECODE(input, skip_bytes, chars_decoded);
2745 if (chars_decoded <= chars_to_skip) {
2746 DECODER_GETSTATE();
2747 if (dec_buffer_len == 0) {
2748 /* Before pos and no bytes buffered in decoder => OK */
2749 cookie.dec_flags = dec_flags;
2750 chars_to_skip -= chars_decoded;
2751 break;
2752 }
2753 /* Skip back by buffered amount and reset heuristic */
2754 skip_bytes -= dec_buffer_len;
2755 skip_back = 1;
2756 }
2757 else {
2758 /* We're too far ahead, skip back a bit */
2759 skip_bytes -= skip_back;
2760 skip_back *= 2;
2761 }
2762 }
2763 if (skip_bytes <= 0) {
2764 skip_bytes = 0;
2765 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
2766 goto fail;
2767 }
2768
2769 /* Note our initial start point. */
2770 cookie.start_pos += skip_bytes;
Victor Stinner9a282972013-06-24 23:01:33 +02002771 cookie.chars_to_skip = Py_SAFE_DOWNCAST(chars_to_skip, Py_ssize_t, int);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002772 if (chars_to_skip == 0)
2773 goto finally;
2774
2775 /* We should be close to the desired position. Now feed the decoder one
2776 * byte at a time until we reach the `chars_to_skip` target.
2777 * As we go, note the nearest "safe start point" before the current
2778 * location (a point where the decoder has nothing buffered, so seek()
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002779 * can safely start from there and advance to this location).
2780 */
2781 chars_decoded = 0;
2782 input = PyBytes_AS_STRING(next_input);
2783 input_end = input + PyBytes_GET_SIZE(next_input);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002784 input += skip_bytes;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002785 while (input < input_end) {
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002786 Py_ssize_t n;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002787
Serhiy Storchakaec67d182013-08-20 20:04:47 +03002788 DECODER_DECODE(input, (Py_ssize_t)1, n);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002789 /* We got n chars for 1 byte */
2790 chars_decoded += n;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002791 cookie.bytes_to_feed += 1;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002792 DECODER_GETSTATE();
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002793
2794 if (dec_buffer_len == 0 && chars_decoded <= chars_to_skip) {
2795 /* Decoder buffer is empty, so this is a safe start point. */
2796 cookie.start_pos += cookie.bytes_to_feed;
2797 chars_to_skip -= chars_decoded;
2798 cookie.dec_flags = dec_flags;
2799 cookie.bytes_to_feed = 0;
2800 chars_decoded = 0;
2801 }
2802 if (chars_decoded >= chars_to_skip)
2803 break;
2804 input++;
2805 }
2806 if (input == input_end) {
2807 /* We didn't get enough decoded data; signal EOF to get more. */
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002808 PyObject *decoded = _PyObject_CallMethodId(
Serhiy Storchaka1f21eaa2019-09-01 12:16:51 +03002809 self->decoder, &PyId_decode, "yO", "", /* final = */ Py_True);
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02002810 if (check_decoded(decoded) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002811 goto fail;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002812 chars_decoded += PyUnicode_GET_LENGTH(decoded);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002813 Py_DECREF(decoded);
2814 cookie.need_eof = 1;
2815
2816 if (chars_decoded < chars_to_skip) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03002817 PyErr_SetString(PyExc_OSError,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002818 "can't reconstruct logical file position");
2819 goto fail;
2820 }
2821 }
2822
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002823finally:
Jeroen Demeyer59ad1102019-07-11 10:59:05 +02002824 res = _PyObject_CallMethodIdOneArg(self->decoder, &PyId_setstate, saved_state);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002825 Py_DECREF(saved_state);
2826 if (res == NULL)
2827 return NULL;
2828 Py_DECREF(res);
2829
2830 /* The returned cookie corresponds to the last safe start point. */
2831 cookie.chars_to_skip = Py_SAFE_DOWNCAST(chars_to_skip, Py_ssize_t, int);
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002832 return textiowrapper_build_cookie(&cookie);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002833
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002834fail:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002835 if (saved_state) {
2836 PyObject *type, *value, *traceback;
2837 PyErr_Fetch(&type, &value, &traceback);
Jeroen Demeyer59ad1102019-07-11 10:59:05 +02002838 res = _PyObject_CallMethodIdOneArg(self->decoder, &PyId_setstate, saved_state);
Serhiy Storchaka04d09eb2015-03-30 09:58:41 +03002839 _PyErr_ChainExceptions(type, value, traceback);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002840 Py_DECREF(saved_state);
Serhiy Storchaka04d09eb2015-03-30 09:58:41 +03002841 Py_XDECREF(res);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002842 }
2843 return NULL;
2844}
2845
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002846/*[clinic input]
2847_io.TextIOWrapper.truncate
2848 pos: object = None
2849 /
2850[clinic start generated code]*/
2851
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002852static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002853_io_TextIOWrapper_truncate_impl(textio *self, PyObject *pos)
2854/*[clinic end generated code: output=90ec2afb9bb7745f input=56ec8baa65aea377]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002855{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002856 PyObject *res;
2857
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002858 CHECK_ATTACHED(self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002859
Petr Viktorinffd97532020-02-11 17:46:57 +01002860 res = PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002861 if (res == NULL)
2862 return NULL;
2863 Py_DECREF(res);
2864
Petr Viktorinffd97532020-02-11 17:46:57 +01002865 return PyObject_CallMethodOneArg(self->buffer, _PyIO_str_truncate, pos);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002866}
2867
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002868static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002869textiowrapper_repr(textio *self)
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002870{
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002871 PyObject *nameobj, *modeobj, *res, *s;
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002872 int status;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002873
2874 CHECK_INITIALIZED(self);
2875
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002876 res = PyUnicode_FromString("<_io.TextIOWrapper");
2877 if (res == NULL)
2878 return NULL;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002879
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002880 status = Py_ReprEnter((PyObject *)self);
2881 if (status != 0) {
2882 if (status > 0) {
2883 PyErr_Format(PyExc_RuntimeError,
2884 "reentrant call inside %s.__repr__",
2885 Py_TYPE(self)->tp_name);
2886 }
2887 goto error;
2888 }
Serhiy Storchakab235a1b2019-08-29 09:25:22 +03002889 if (_PyObject_LookupAttrId((PyObject *) self, &PyId_name, &nameobj) < 0) {
2890 if (!PyErr_ExceptionMatches(PyExc_ValueError)) {
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002891 goto error;
Serhiy Storchakab235a1b2019-08-29 09:25:22 +03002892 }
2893 /* Ignore ValueError raised if the underlying stream was detached */
2894 PyErr_Clear();
Antoine Pitrou716c4442009-05-23 19:04:03 +00002895 }
Serhiy Storchakab235a1b2019-08-29 09:25:22 +03002896 if (nameobj != NULL) {
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002897 s = PyUnicode_FromFormat(" name=%R", nameobj);
Antoine Pitrou716c4442009-05-23 19:04:03 +00002898 Py_DECREF(nameobj);
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002899 if (s == NULL)
2900 goto error;
2901 PyUnicode_AppendAndDel(&res, s);
2902 if (res == NULL)
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002903 goto error;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002904 }
Serhiy Storchakab235a1b2019-08-29 09:25:22 +03002905 if (_PyObject_LookupAttrId((PyObject *) self, &PyId_mode, &modeobj) < 0) {
2906 goto error;
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002907 }
Serhiy Storchakab235a1b2019-08-29 09:25:22 +03002908 if (modeobj != NULL) {
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002909 s = PyUnicode_FromFormat(" mode=%R", modeobj);
2910 Py_DECREF(modeobj);
2911 if (s == NULL)
2912 goto error;
2913 PyUnicode_AppendAndDel(&res, s);
2914 if (res == NULL)
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002915 goto error;
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002916 }
2917 s = PyUnicode_FromFormat("%U encoding=%R>",
2918 res, self->encoding);
2919 Py_DECREF(res);
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002920 if (status == 0) {
2921 Py_ReprLeave((PyObject *)self);
2922 }
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002923 return s;
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002924
2925 error:
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002926 Py_XDECREF(res);
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002927 if (status == 0) {
2928 Py_ReprLeave((PyObject *)self);
2929 }
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002930 return NULL;
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002931}
2932
2933
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002934/* Inquiries */
2935
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002936/*[clinic input]
2937_io.TextIOWrapper.fileno
2938[clinic start generated code]*/
2939
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002940static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002941_io_TextIOWrapper_fileno_impl(textio *self)
2942/*[clinic end generated code: output=21490a4c3da13e6c input=c488ca83d0069f9b]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002943{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002944 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002945 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_fileno);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002946}
2947
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002948/*[clinic input]
2949_io.TextIOWrapper.seekable
2950[clinic start generated code]*/
2951
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002952static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002953_io_TextIOWrapper_seekable_impl(textio *self)
2954/*[clinic end generated code: output=ab223dbbcffc0f00 input=8b005ca06e1fca13]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002955{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002956 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002957 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_seekable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002958}
2959
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002960/*[clinic input]
2961_io.TextIOWrapper.readable
2962[clinic start generated code]*/
2963
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002964static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002965_io_TextIOWrapper_readable_impl(textio *self)
2966/*[clinic end generated code: output=72ff7ba289a8a91b input=0704ea7e01b0d3eb]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002967{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002968 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002969 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_readable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002970}
2971
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002972/*[clinic input]
2973_io.TextIOWrapper.writable
2974[clinic start generated code]*/
2975
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002976static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002977_io_TextIOWrapper_writable_impl(textio *self)
2978/*[clinic end generated code: output=a728c71790d03200 input=c41740bc9d8636e8]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002979{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002980 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002981 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_writable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002982}
2983
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002984/*[clinic input]
2985_io.TextIOWrapper.isatty
2986[clinic start generated code]*/
2987
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002988static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002989_io_TextIOWrapper_isatty_impl(textio *self)
2990/*[clinic end generated code: output=12be1a35bace882e input=fb68d9f2c99bbfff]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002991{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002992 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002993 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_isatty);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002994}
2995
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002996/*[clinic input]
2997_io.TextIOWrapper.flush
2998[clinic start generated code]*/
2999
Antoine Pitrou243757e2010-11-05 21:15:39 +00003000static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003001_io_TextIOWrapper_flush_impl(textio *self)
3002/*[clinic end generated code: output=59de9165f9c2e4d2 input=928c60590694ab85]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003003{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003004 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003005 CHECK_CLOSED(self);
3006 self->telling = self->seekable;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003007 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003008 return NULL;
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003009 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003010}
3011
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003012/*[clinic input]
3013_io.TextIOWrapper.close
3014[clinic start generated code]*/
3015
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003016static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003017_io_TextIOWrapper_close_impl(textio *self)
3018/*[clinic end generated code: output=056ccf8b4876e4f4 input=9c2114315eae1948]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003019{
3020 PyObject *res;
Antoine Pitrou6be88762010-05-03 16:48:20 +00003021 int r;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003022 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003023
Antoine Pitrou6be88762010-05-03 16:48:20 +00003024 res = textiowrapper_closed_get(self, NULL);
3025 if (res == NULL)
3026 return NULL;
3027 r = PyObject_IsTrue(res);
3028 Py_DECREF(res);
3029 if (r < 0)
3030 return NULL;
Victor Stinnerf6c57832010-05-19 01:17:01 +00003031
Antoine Pitrou6be88762010-05-03 16:48:20 +00003032 if (r > 0) {
3033 Py_RETURN_NONE; /* stream already closed */
3034 }
3035 else {
Benjamin Peterson68623612012-12-20 11:53:11 -06003036 PyObject *exc = NULL, *val, *tb;
Antoine Pitrou796564c2013-07-30 19:59:21 +02003037 if (self->finalizing) {
Jeroen Demeyer59ad1102019-07-11 10:59:05 +02003038 res = _PyObject_CallMethodIdOneArg(self->buffer,
3039 &PyId__dealloc_warn,
3040 (PyObject *)self);
Antoine Pitroue033e062010-10-29 10:38:18 +00003041 if (res)
3042 Py_DECREF(res);
3043 else
3044 PyErr_Clear();
3045 }
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003046 res = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_flush);
Benjamin Peterson68623612012-12-20 11:53:11 -06003047 if (res == NULL)
3048 PyErr_Fetch(&exc, &val, &tb);
Antoine Pitrou6be88762010-05-03 16:48:20 +00003049 else
3050 Py_DECREF(res);
3051
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003052 res = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_close);
Benjamin Peterson68623612012-12-20 11:53:11 -06003053 if (exc != NULL) {
Serhiy Storchakae2bd2a72014-10-08 22:31:52 +03003054 _PyErr_ChainExceptions(exc, val, tb);
3055 Py_CLEAR(res);
Benjamin Peterson68623612012-12-20 11:53:11 -06003056 }
3057 return res;
Antoine Pitrou6be88762010-05-03 16:48:20 +00003058 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003059}
3060
3061static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003062textiowrapper_iternext(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003063{
3064 PyObject *line;
3065
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003066 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003067
3068 self->telling = 0;
Andy Lesterdffe4c02020-03-04 07:15:20 -06003069 if (Py_IS_TYPE(self, &PyTextIOWrapper_Type)) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003070 /* Skip method call overhead for speed */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003071 line = _textiowrapper_readline(self, -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003072 }
3073 else {
Petr Viktorinffd97532020-02-11 17:46:57 +01003074 line = PyObject_CallMethodNoArgs((PyObject *)self,
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003075 _PyIO_str_readline);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003076 if (line && !PyUnicode_Check(line)) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03003077 PyErr_Format(PyExc_OSError,
Serhiy Storchakab6a9c972016-04-17 09:39:28 +03003078 "readline() should have returned a str object, "
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003079 "not '%.200s'", Py_TYPE(line)->tp_name);
3080 Py_DECREF(line);
3081 return NULL;
3082 }
3083 }
3084
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02003085 if (line == NULL || PyUnicode_READY(line) == -1)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003086 return NULL;
3087
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02003088 if (PyUnicode_GET_LENGTH(line) == 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003089 /* Reached EOF or would have blocked */
3090 Py_DECREF(line);
3091 Py_CLEAR(self->snapshot);
3092 self->telling = self->seekable;
3093 return NULL;
3094 }
3095
3096 return line;
3097}
3098
3099static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003100textiowrapper_name_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003101{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003102 CHECK_ATTACHED(self);
Martin v. Löwis767046a2011-10-14 15:35:36 +02003103 return _PyObject_GetAttrId(self->buffer, &PyId_name);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003104}
3105
3106static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003107textiowrapper_closed_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003108{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003109 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003110 return PyObject_GetAttr(self->buffer, _PyIO_str_closed);
3111}
3112
3113static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003114textiowrapper_newlines_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003115{
3116 PyObject *res;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003117 CHECK_ATTACHED(self);
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003118 if (self->decoder == NULL ||
3119 _PyObject_LookupAttr(self->decoder, _PyIO_str_newlines, &res) == 0)
3120 {
Serhiy Storchaka4d9aec02018-01-16 18:34:21 +02003121 Py_RETURN_NONE;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003122 }
3123 return res;
3124}
3125
3126static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003127textiowrapper_errors_get(textio *self, void *context)
Benjamin Peterson0926ad12009-06-06 18:02:12 +00003128{
3129 CHECK_INITIALIZED(self);
INADA Naoki507434f2017-12-21 09:59:53 +09003130 Py_INCREF(self->errors);
3131 return self->errors;
Benjamin Peterson0926ad12009-06-06 18:02:12 +00003132}
3133
3134static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003135textiowrapper_chunk_size_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003136{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003137 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003138 return PyLong_FromSsize_t(self->chunk_size);
3139}
3140
3141static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003142textiowrapper_chunk_size_set(textio *self, PyObject *arg, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003143{
3144 Py_ssize_t n;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003145 CHECK_ATTACHED_INT(self);
Zackery Spytz842acaa2018-12-17 07:52:45 -07003146 if (arg == NULL) {
3147 PyErr_SetString(PyExc_AttributeError, "cannot delete attribute");
3148 return -1;
3149 }
Antoine Pitroucb4ae812011-07-13 21:07:49 +02003150 n = PyNumber_AsSsize_t(arg, PyExc_ValueError);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003151 if (n == -1 && PyErr_Occurred())
3152 return -1;
3153 if (n <= 0) {
3154 PyErr_SetString(PyExc_ValueError,
3155 "a strictly positive integer is required");
3156 return -1;
3157 }
3158 self->chunk_size = n;
3159 return 0;
3160}
3161
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003162#include "clinic/textio.c.h"
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003163
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003164static PyMethodDef incrementalnewlinedecoder_methods[] = {
3165 _IO_INCREMENTALNEWLINEDECODER_DECODE_METHODDEF
3166 _IO_INCREMENTALNEWLINEDECODER_GETSTATE_METHODDEF
3167 _IO_INCREMENTALNEWLINEDECODER_SETSTATE_METHODDEF
3168 _IO_INCREMENTALNEWLINEDECODER_RESET_METHODDEF
3169 {NULL}
3170};
3171
3172static PyGetSetDef incrementalnewlinedecoder_getset[] = {
3173 {"newlines", (getter)incrementalnewlinedecoder_newlines_get, NULL, NULL},
3174 {NULL}
3175};
3176
3177PyTypeObject PyIncrementalNewlineDecoder_Type = {
3178 PyVarObject_HEAD_INIT(NULL, 0)
3179 "_io.IncrementalNewlineDecoder", /*tp_name*/
3180 sizeof(nldecoder_object), /*tp_basicsize*/
3181 0, /*tp_itemsize*/
3182 (destructor)incrementalnewlinedecoder_dealloc, /*tp_dealloc*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003183 0, /*tp_vectorcall_offset*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003184 0, /*tp_getattr*/
3185 0, /*tp_setattr*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003186 0, /*tp_as_async*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003187 0, /*tp_repr*/
3188 0, /*tp_as_number*/
3189 0, /*tp_as_sequence*/
3190 0, /*tp_as_mapping*/
3191 0, /*tp_hash */
3192 0, /*tp_call*/
3193 0, /*tp_str*/
3194 0, /*tp_getattro*/
3195 0, /*tp_setattro*/
3196 0, /*tp_as_buffer*/
3197 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
3198 _io_IncrementalNewlineDecoder___init____doc__, /* tp_doc */
3199 0, /* tp_traverse */
3200 0, /* tp_clear */
3201 0, /* tp_richcompare */
3202 0, /*tp_weaklistoffset*/
3203 0, /* tp_iter */
3204 0, /* tp_iternext */
3205 incrementalnewlinedecoder_methods, /* tp_methods */
3206 0, /* tp_members */
3207 incrementalnewlinedecoder_getset, /* tp_getset */
3208 0, /* tp_base */
3209 0, /* tp_dict */
3210 0, /* tp_descr_get */
3211 0, /* tp_descr_set */
3212 0, /* tp_dictoffset */
3213 _io_IncrementalNewlineDecoder___init__, /* tp_init */
3214 0, /* tp_alloc */
3215 PyType_GenericNew, /* tp_new */
3216};
3217
3218
3219static PyMethodDef textiowrapper_methods[] = {
3220 _IO_TEXTIOWRAPPER_DETACH_METHODDEF
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02003221 _IO_TEXTIOWRAPPER_RECONFIGURE_METHODDEF
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003222 _IO_TEXTIOWRAPPER_WRITE_METHODDEF
3223 _IO_TEXTIOWRAPPER_READ_METHODDEF
3224 _IO_TEXTIOWRAPPER_READLINE_METHODDEF
3225 _IO_TEXTIOWRAPPER_FLUSH_METHODDEF
3226 _IO_TEXTIOWRAPPER_CLOSE_METHODDEF
3227
3228 _IO_TEXTIOWRAPPER_FILENO_METHODDEF
3229 _IO_TEXTIOWRAPPER_SEEKABLE_METHODDEF
3230 _IO_TEXTIOWRAPPER_READABLE_METHODDEF
3231 _IO_TEXTIOWRAPPER_WRITABLE_METHODDEF
3232 _IO_TEXTIOWRAPPER_ISATTY_METHODDEF
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003233
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003234 _IO_TEXTIOWRAPPER_SEEK_METHODDEF
3235 _IO_TEXTIOWRAPPER_TELL_METHODDEF
3236 _IO_TEXTIOWRAPPER_TRUNCATE_METHODDEF
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003237 {NULL, NULL}
3238};
3239
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003240static PyMemberDef textiowrapper_members[] = {
3241 {"encoding", T_OBJECT, offsetof(textio, encoding), READONLY},
3242 {"buffer", T_OBJECT, offsetof(textio, buffer), READONLY},
3243 {"line_buffering", T_BOOL, offsetof(textio, line_buffering), READONLY},
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02003244 {"write_through", T_BOOL, offsetof(textio, write_through), READONLY},
Antoine Pitrou796564c2013-07-30 19:59:21 +02003245 {"_finalizing", T_BOOL, offsetof(textio, finalizing), 0},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003246 {NULL}
3247};
3248
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003249static PyGetSetDef textiowrapper_getset[] = {
3250 {"name", (getter)textiowrapper_name_get, NULL, NULL},
3251 {"closed", (getter)textiowrapper_closed_get, NULL, NULL},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003252/* {"mode", (getter)TextIOWrapper_mode_get, NULL, NULL},
3253*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003254 {"newlines", (getter)textiowrapper_newlines_get, NULL, NULL},
3255 {"errors", (getter)textiowrapper_errors_get, NULL, NULL},
3256 {"_CHUNK_SIZE", (getter)textiowrapper_chunk_size_get,
3257 (setter)textiowrapper_chunk_size_set, NULL},
Benjamin Peterson1fea3212009-04-19 03:15:20 +00003258 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003259};
3260
3261PyTypeObject PyTextIOWrapper_Type = {
3262 PyVarObject_HEAD_INIT(NULL, 0)
3263 "_io.TextIOWrapper", /*tp_name*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003264 sizeof(textio), /*tp_basicsize*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003265 0, /*tp_itemsize*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003266 (destructor)textiowrapper_dealloc, /*tp_dealloc*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003267 0, /*tp_vectorcall_offset*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003268 0, /*tp_getattr*/
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00003269 0, /*tps_etattr*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003270 0, /*tp_as_async*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003271 (reprfunc)textiowrapper_repr,/*tp_repr*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003272 0, /*tp_as_number*/
3273 0, /*tp_as_sequence*/
3274 0, /*tp_as_mapping*/
3275 0, /*tp_hash */
3276 0, /*tp_call*/
3277 0, /*tp_str*/
3278 0, /*tp_getattro*/
3279 0, /*tp_setattro*/
3280 0, /*tp_as_buffer*/
3281 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE
Antoine Pitrouada319b2019-05-29 22:12:38 +02003282 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003283 _io_TextIOWrapper___init____doc__, /* tp_doc */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003284 (traverseproc)textiowrapper_traverse, /* tp_traverse */
3285 (inquiry)textiowrapper_clear, /* tp_clear */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003286 0, /* tp_richcompare */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003287 offsetof(textio, weakreflist), /*tp_weaklistoffset*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003288 0, /* tp_iter */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003289 (iternextfunc)textiowrapper_iternext, /* tp_iternext */
3290 textiowrapper_methods, /* tp_methods */
3291 textiowrapper_members, /* tp_members */
3292 textiowrapper_getset, /* tp_getset */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003293 0, /* tp_base */
3294 0, /* tp_dict */
3295 0, /* tp_descr_get */
3296 0, /* tp_descr_set */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003297 offsetof(textio, dict), /*tp_dictoffset*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003298 _io_TextIOWrapper___init__, /* tp_init */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003299 0, /* tp_alloc */
3300 PyType_GenericNew, /* tp_new */
Antoine Pitrou796564c2013-07-30 19:59:21 +02003301 0, /* tp_free */
3302 0, /* tp_is_gc */
3303 0, /* tp_bases */
3304 0, /* tp_mro */
3305 0, /* tp_cache */
3306 0, /* tp_subclasses */
3307 0, /* tp_weaklist */
3308 0, /* tp_del */
3309 0, /* tp_version_tag */
3310 0, /* tp_finalize */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003311};