blob: ed1dc005c69a174dcd792f6582edf83d6088839e [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 Stinnerbcda8f12018-11-21 22:27:47 +010011#include "pycore_object.h"
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000012#include "structmember.h"
13#include "_iomodule.h"
14
Serhiy Storchakaf24131f2015-04-16 11:19:43 +030015/*[clinic input]
16module _io
17class _io.IncrementalNewlineDecoder "nldecoder_object *" "&PyIncrementalNewlineDecoder_Type"
18class _io.TextIOWrapper "textio *" "&TextIOWrapper_TYpe"
19[clinic start generated code]*/
20/*[clinic end generated code: output=da39a3ee5e6b4b0d input=2097a4fc85670c26]*/
21
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020022_Py_IDENTIFIER(close);
23_Py_IDENTIFIER(_dealloc_warn);
24_Py_IDENTIFIER(decode);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020025_Py_IDENTIFIER(fileno);
26_Py_IDENTIFIER(flush);
27_Py_IDENTIFIER(getpreferredencoding);
28_Py_IDENTIFIER(isatty);
Martin v. Löwis767046a2011-10-14 15:35:36 +020029_Py_IDENTIFIER(mode);
30_Py_IDENTIFIER(name);
31_Py_IDENTIFIER(raw);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020032_Py_IDENTIFIER(read);
33_Py_IDENTIFIER(readable);
34_Py_IDENTIFIER(replace);
35_Py_IDENTIFIER(reset);
36_Py_IDENTIFIER(seek);
37_Py_IDENTIFIER(seekable);
38_Py_IDENTIFIER(setstate);
INADA Naoki507434f2017-12-21 09:59:53 +090039_Py_IDENTIFIER(strict);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +020040_Py_IDENTIFIER(tell);
41_Py_IDENTIFIER(writable);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +020042
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000043/* TextIOBase */
44
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000045PyDoc_STRVAR(textiobase_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000046 "Base class for text I/O.\n"
47 "\n"
48 "This class provides a character and line based interface to stream\n"
49 "I/O. There is no readinto method because Python's character strings\n"
50 "are immutable. There is no public constructor.\n"
51 );
52
53static PyObject *
54_unsupported(const char *message)
55{
Antoine Pitrou712cb732013-12-21 15:51:54 +010056 _PyIO_State *state = IO_STATE();
57 if (state != NULL)
58 PyErr_SetString(state->unsupported_operation, message);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000059 return NULL;
60}
61
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000062PyDoc_STRVAR(textiobase_detach_doc,
Benjamin Petersond2e0c792009-05-01 20:40:59 +000063 "Separate the underlying buffer from the TextIOBase and return it.\n"
64 "\n"
65 "After the underlying buffer has been detached, the TextIO is in an\n"
66 "unusable state.\n"
67 );
68
69static PyObject *
Siddhesh Poyarekar55edd0c2018-04-30 00:29:33 +053070textiobase_detach(PyObject *self, PyObject *Py_UNUSED(ignored))
Benjamin Petersond2e0c792009-05-01 20:40:59 +000071{
72 return _unsupported("detach");
73}
74
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000075PyDoc_STRVAR(textiobase_read_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000076 "Read at most n characters from stream.\n"
77 "\n"
78 "Read from underlying buffer until we have n characters or we hit EOF.\n"
79 "If n is negative or omitted, read until EOF.\n"
80 );
81
82static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000083textiobase_read(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000084{
85 return _unsupported("read");
86}
87
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000088PyDoc_STRVAR(textiobase_readline_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000089 "Read until newline or EOF.\n"
90 "\n"
91 "Returns an empty string if EOF is hit immediately.\n"
92 );
93
94static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000095textiobase_readline(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000096{
97 return _unsupported("readline");
98}
99
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000100PyDoc_STRVAR(textiobase_write_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000101 "Write string to stream.\n"
102 "Returns the number of characters written (which is always equal to\n"
103 "the length of the string).\n"
104 );
105
106static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000107textiobase_write(PyObject *self, PyObject *args)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000108{
109 return _unsupported("write");
110}
111
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000112PyDoc_STRVAR(textiobase_encoding_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000113 "Encoding of the text stream.\n"
114 "\n"
115 "Subclasses should override.\n"
116 );
117
118static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000119textiobase_encoding_get(PyObject *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000120{
121 Py_RETURN_NONE;
122}
123
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000124PyDoc_STRVAR(textiobase_newlines_doc,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000125 "Line endings translated so far.\n"
126 "\n"
127 "Only line endings translated during reading are considered.\n"
128 "\n"
129 "Subclasses should override.\n"
130 );
131
132static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000133textiobase_newlines_get(PyObject *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000134{
135 Py_RETURN_NONE;
136}
137
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000138PyDoc_STRVAR(textiobase_errors_doc,
Benjamin Peterson0926ad12009-06-06 18:02:12 +0000139 "The error setting of the decoder or encoder.\n"
140 "\n"
141 "Subclasses should override.\n"
142 );
143
144static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000145textiobase_errors_get(PyObject *self, void *context)
Benjamin Peterson0926ad12009-06-06 18:02:12 +0000146{
147 Py_RETURN_NONE;
148}
149
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000150
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000151static PyMethodDef textiobase_methods[] = {
Siddhesh Poyarekar55edd0c2018-04-30 00:29:33 +0530152 {"detach", textiobase_detach, METH_NOARGS, textiobase_detach_doc},
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000153 {"read", textiobase_read, METH_VARARGS, textiobase_read_doc},
154 {"readline", textiobase_readline, METH_VARARGS, textiobase_readline_doc},
155 {"write", textiobase_write, METH_VARARGS, textiobase_write_doc},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000156 {NULL, NULL}
157};
158
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000159static PyGetSetDef textiobase_getset[] = {
160 {"encoding", (getter)textiobase_encoding_get, NULL, textiobase_encoding_doc},
161 {"newlines", (getter)textiobase_newlines_get, NULL, textiobase_newlines_doc},
162 {"errors", (getter)textiobase_errors_get, NULL, textiobase_errors_doc},
Benjamin Peterson1fea3212009-04-19 03:15:20 +0000163 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000164};
165
166PyTypeObject PyTextIOBase_Type = {
167 PyVarObject_HEAD_INIT(NULL, 0)
168 "_io._TextIOBase", /*tp_name*/
169 0, /*tp_basicsize*/
170 0, /*tp_itemsize*/
171 0, /*tp_dealloc*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +0200172 0, /*tp_vectorcall_offset*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000173 0, /*tp_getattr*/
174 0, /*tp_setattr*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +0200175 0, /*tp_as_async*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000176 0, /*tp_repr*/
177 0, /*tp_as_number*/
178 0, /*tp_as_sequence*/
179 0, /*tp_as_mapping*/
180 0, /*tp_hash */
181 0, /*tp_call*/
182 0, /*tp_str*/
183 0, /*tp_getattro*/
184 0, /*tp_setattro*/
185 0, /*tp_as_buffer*/
Antoine Pitrouada319b2019-05-29 22:12:38 +0200186 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000187 textiobase_doc, /* tp_doc */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000188 0, /* tp_traverse */
189 0, /* tp_clear */
190 0, /* tp_richcompare */
191 0, /* tp_weaklistoffset */
192 0, /* tp_iter */
193 0, /* tp_iternext */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000194 textiobase_methods, /* tp_methods */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000195 0, /* tp_members */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000196 textiobase_getset, /* tp_getset */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000197 &PyIOBase_Type, /* tp_base */
198 0, /* tp_dict */
199 0, /* tp_descr_get */
200 0, /* tp_descr_set */
201 0, /* tp_dictoffset */
202 0, /* tp_init */
203 0, /* tp_alloc */
204 0, /* tp_new */
Antoine Pitrou796564c2013-07-30 19:59:21 +0200205 0, /* tp_free */
206 0, /* tp_is_gc */
207 0, /* tp_bases */
208 0, /* tp_mro */
209 0, /* tp_cache */
210 0, /* tp_subclasses */
211 0, /* tp_weaklist */
212 0, /* tp_del */
213 0, /* tp_version_tag */
214 0, /* tp_finalize */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000215};
216
217
218/* IncrementalNewlineDecoder */
219
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000220typedef struct {
221 PyObject_HEAD
222 PyObject *decoder;
223 PyObject *errors;
Victor Stinner7d7e7752014-06-17 23:31:25 +0200224 unsigned int pendingcr: 1;
225 unsigned int translate: 1;
Antoine Pitrouca767bd2009-09-21 21:37:02 +0000226 unsigned int seennl: 3;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000227} nldecoder_object;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000228
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300229/*[clinic input]
230_io.IncrementalNewlineDecoder.__init__
231 decoder: object
232 translate: int
233 errors: object(c_default="NULL") = "strict"
234
235Codec used when reading a file in universal newlines mode.
236
237It wraps another incremental decoder, translating \r\n and \r into \n.
238It also records the types of newlines encountered. When used with
239translate=False, it ensures that the newline sequence is returned in
240one piece. When used with decoder=None, it expects unicode strings as
241decode input and translates newlines without first invoking an external
242decoder.
243[clinic start generated code]*/
244
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000245static int
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300246_io_IncrementalNewlineDecoder___init___impl(nldecoder_object *self,
247 PyObject *decoder, int translate,
248 PyObject *errors)
249/*[clinic end generated code: output=fbd04d443e764ec2 input=89db6b19c6b126bf]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000250{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000251 self->decoder = decoder;
252 Py_INCREF(decoder);
253
254 if (errors == NULL) {
INADA Naoki507434f2017-12-21 09:59:53 +0900255 self->errors = _PyUnicode_FromId(&PyId_strict);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000256 if (self->errors == NULL)
257 return -1;
258 }
259 else {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000260 self->errors = errors;
261 }
INADA Naoki507434f2017-12-21 09:59:53 +0900262 Py_INCREF(self->errors);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000263
Xiang Zhangb08746b2018-10-31 19:49:16 +0800264 self->translate = translate ? 1 : 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000265 self->seennl = 0;
266 self->pendingcr = 0;
267
268 return 0;
269}
270
271static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000272incrementalnewlinedecoder_dealloc(nldecoder_object *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000273{
274 Py_CLEAR(self->decoder);
275 Py_CLEAR(self->errors);
276 Py_TYPE(self)->tp_free((PyObject *)self);
277}
278
Serhiy Storchaka94dc6732013-02-03 17:03:31 +0200279static int
280check_decoded(PyObject *decoded)
281{
282 if (decoded == NULL)
283 return -1;
284 if (!PyUnicode_Check(decoded)) {
285 PyErr_Format(PyExc_TypeError,
286 "decoder should return a string result, not '%.200s'",
287 Py_TYPE(decoded)->tp_name);
288 Py_DECREF(decoded);
289 return -1;
290 }
Serhiy Storchakad03ce4a2013-02-03 17:07:32 +0200291 if (PyUnicode_READY(decoded) < 0) {
292 Py_DECREF(decoded);
293 return -1;
294 }
Serhiy Storchaka94dc6732013-02-03 17:03:31 +0200295 return 0;
296}
297
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000298#define SEEN_CR 1
299#define SEEN_LF 2
300#define SEEN_CRLF 4
301#define SEEN_ALL (SEEN_CR | SEEN_LF | SEEN_CRLF)
302
303PyObject *
Antoine Pitrou09fcb722013-10-23 19:20:21 +0200304_PyIncrementalNewlineDecoder_decode(PyObject *myself,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000305 PyObject *input, int final)
306{
307 PyObject *output;
308 Py_ssize_t output_len;
Antoine Pitrou09fcb722013-10-23 19:20:21 +0200309 nldecoder_object *self = (nldecoder_object *) myself;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000310
311 if (self->decoder == NULL) {
312 PyErr_SetString(PyExc_ValueError,
313 "IncrementalNewlineDecoder.__init__ not called");
314 return NULL;
315 }
316
317 /* decode input (with the eventual \r from a previous pass) */
318 if (self->decoder != Py_None) {
319 output = PyObject_CallMethodObjArgs(self->decoder,
320 _PyIO_str_decode, input, final ? Py_True : Py_False, NULL);
321 }
322 else {
323 output = input;
324 Py_INCREF(output);
325 }
326
Serhiy Storchaka94dc6732013-02-03 17:03:31 +0200327 if (check_decoded(output) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000328 return NULL;
329
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200330 output_len = PyUnicode_GET_LENGTH(output);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000331 if (self->pendingcr && (final || output_len > 0)) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200332 /* Prefix output with CR */
333 int kind;
334 PyObject *modified;
335 char *out;
336
337 modified = PyUnicode_New(output_len + 1,
338 PyUnicode_MAX_CHAR_VALUE(output));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000339 if (modified == NULL)
340 goto error;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200341 kind = PyUnicode_KIND(modified);
342 out = PyUnicode_DATA(modified);
343 PyUnicode_WRITE(kind, PyUnicode_DATA(modified), 0, '\r');
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200344 memcpy(out + kind, PyUnicode_DATA(output), kind * output_len);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000345 Py_DECREF(output);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200346 output = modified; /* output remains ready */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000347 self->pendingcr = 0;
348 output_len++;
349 }
350
351 /* retain last \r even when not translating data:
352 * then readline() is sure to get \r\n in one pass
353 */
354 if (!final) {
Antoine Pitrou24f36292009-03-28 22:16:42 +0000355 if (output_len > 0
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200356 && PyUnicode_READ_CHAR(output, output_len - 1) == '\r')
357 {
358 PyObject *modified = PyUnicode_Substring(output, 0, output_len -1);
359 if (modified == NULL)
360 goto error;
361 Py_DECREF(output);
362 output = modified;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000363 self->pendingcr = 1;
364 }
365 }
366
367 /* Record which newlines are read and do newline translation if desired,
368 all in one pass. */
369 {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200370 void *in_str;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000371 Py_ssize_t len;
372 int seennl = self->seennl;
373 int only_lf = 0;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200374 int kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000375
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200376 in_str = PyUnicode_DATA(output);
377 len = PyUnicode_GET_LENGTH(output);
378 kind = PyUnicode_KIND(output);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000379
380 if (len == 0)
381 return output;
382
383 /* If, up to now, newlines are consistently \n, do a quick check
384 for the \r *byte* with the libc's optimized memchr.
385 */
386 if (seennl == SEEN_LF || seennl == 0) {
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200387 only_lf = (memchr(in_str, '\r', kind * len) == NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000388 }
389
Antoine Pitrou66913e22009-03-06 23:40:56 +0000390 if (only_lf) {
391 /* If not already seen, quick scan for a possible "\n" character.
392 (there's nothing else to be done, even when in translation mode)
393 */
394 if (seennl == 0 &&
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200395 memchr(in_str, '\n', kind * len) != NULL) {
Antoine Pitrouc28e2e52011-11-13 03:53:42 +0100396 if (kind == PyUnicode_1BYTE_KIND)
397 seennl |= SEEN_LF;
398 else {
399 Py_ssize_t i = 0;
400 for (;;) {
401 Py_UCS4 c;
402 /* Fast loop for non-control characters */
403 while (PyUnicode_READ(kind, in_str, i) > '\n')
404 i++;
405 c = PyUnicode_READ(kind, in_str, i++);
406 if (c == '\n') {
407 seennl |= SEEN_LF;
408 break;
409 }
410 if (i >= len)
411 break;
Antoine Pitrou66913e22009-03-06 23:40:56 +0000412 }
Antoine Pitrou66913e22009-03-06 23:40:56 +0000413 }
414 }
415 /* Finished: we have scanned for newlines, and none of them
416 need translating */
417 }
418 else if (!self->translate) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200419 Py_ssize_t i = 0;
Antoine Pitrou66913e22009-03-06 23:40:56 +0000420 /* We have already seen all newline types, no need to scan again */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000421 if (seennl == SEEN_ALL)
422 goto endscan;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000423 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200424 Py_UCS4 c;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000425 /* Fast loop for non-control characters */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200426 while (PyUnicode_READ(kind, in_str, i) > '\r')
427 i++;
428 c = PyUnicode_READ(kind, in_str, i++);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000429 if (c == '\n')
430 seennl |= SEEN_LF;
431 else if (c == '\r') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200432 if (PyUnicode_READ(kind, in_str, i) == '\n') {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000433 seennl |= SEEN_CRLF;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200434 i++;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000435 }
436 else
437 seennl |= SEEN_CR;
438 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200439 if (i >= len)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000440 break;
441 if (seennl == SEEN_ALL)
442 break;
443 }
444 endscan:
445 ;
446 }
Antoine Pitrou66913e22009-03-06 23:40:56 +0000447 else {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200448 void *translated;
449 int kind = PyUnicode_KIND(output);
450 void *in_str = PyUnicode_DATA(output);
451 Py_ssize_t in, out;
452 /* XXX: Previous in-place translation here is disabled as
453 resizing is not possible anymore */
454 /* We could try to optimize this so that we only do a copy
455 when there is something to translate. On the other hand,
456 we already know there is a \r byte, so chances are high
457 that something needs to be done. */
Martin v. Löwisc47adb02011-10-07 20:55:35 +0200458 translated = PyMem_Malloc(kind * len);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200459 if (translated == NULL) {
460 PyErr_NoMemory();
461 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000462 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200463 in = out = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000464 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200465 Py_UCS4 c;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000466 /* Fast loop for non-control characters */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200467 while ((c = PyUnicode_READ(kind, in_str, in++)) > '\r')
468 PyUnicode_WRITE(kind, translated, out++, c);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000469 if (c == '\n') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200470 PyUnicode_WRITE(kind, translated, out++, c);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000471 seennl |= SEEN_LF;
472 continue;
473 }
474 if (c == '\r') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200475 if (PyUnicode_READ(kind, in_str, in) == '\n') {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000476 in++;
477 seennl |= SEEN_CRLF;
478 }
479 else
480 seennl |= SEEN_CR;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200481 PyUnicode_WRITE(kind, translated, out++, '\n');
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000482 continue;
483 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200484 if (in > len)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000485 break;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200486 PyUnicode_WRITE(kind, translated, out++, c);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000487 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200488 Py_DECREF(output);
489 output = PyUnicode_FromKindAndData(kind, translated, out);
Antoine Pitrouc1b0bfd2011-11-12 22:34:28 +0100490 PyMem_Free(translated);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200491 if (!output)
Ross Lagerwall0f9eec12012-04-07 07:09:57 +0200492 return NULL;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000493 }
494 self->seennl |= seennl;
495 }
496
497 return output;
498
499 error:
500 Py_DECREF(output);
501 return NULL;
502}
503
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300504/*[clinic input]
505_io.IncrementalNewlineDecoder.decode
506 input: object
Serhiy Storchaka202fda52017-03-12 10:10:47 +0200507 final: bool(accept={int}) = False
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300508[clinic start generated code]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000509
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300510static PyObject *
511_io_IncrementalNewlineDecoder_decode_impl(nldecoder_object *self,
512 PyObject *input, int final)
Serhiy Storchaka202fda52017-03-12 10:10:47 +0200513/*[clinic end generated code: output=0d486755bb37a66e input=a4ea97f26372d866]*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300514{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000515 return _PyIncrementalNewlineDecoder_decode((PyObject *) self, input, final);
516}
517
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300518/*[clinic input]
519_io.IncrementalNewlineDecoder.getstate
520[clinic start generated code]*/
521
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000522static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300523_io_IncrementalNewlineDecoder_getstate_impl(nldecoder_object *self)
524/*[clinic end generated code: output=f0d2c9c136f4e0d0 input=f8ff101825e32e7f]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000525{
526 PyObject *buffer;
Benjamin Peterson47ff0732016-09-08 09:15:54 -0700527 unsigned long long flag;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000528
529 if (self->decoder != Py_None) {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200530 PyObject *state = _PyObject_CallMethodNoArgs(self->decoder,
531 _PyIO_str_getstate);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000532 if (state == NULL)
533 return NULL;
Oren Milman13614e32017-08-24 19:51:24 +0300534 if (!PyTuple_Check(state)) {
535 PyErr_SetString(PyExc_TypeError,
536 "illegal decoder state");
537 Py_DECREF(state);
538 return NULL;
539 }
540 if (!PyArg_ParseTuple(state, "OK;illegal decoder state",
541 &buffer, &flag))
542 {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000543 Py_DECREF(state);
544 return NULL;
545 }
546 Py_INCREF(buffer);
547 Py_DECREF(state);
548 }
549 else {
550 buffer = PyBytes_FromString("");
551 flag = 0;
552 }
553 flag <<= 1;
554 if (self->pendingcr)
555 flag |= 1;
556 return Py_BuildValue("NK", buffer, flag);
557}
558
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300559/*[clinic input]
560_io.IncrementalNewlineDecoder.setstate
561 state: object
562 /
563[clinic start generated code]*/
564
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000565static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300566_io_IncrementalNewlineDecoder_setstate(nldecoder_object *self,
567 PyObject *state)
568/*[clinic end generated code: output=c10c622508b576cb input=c53fb505a76dbbe2]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000569{
570 PyObject *buffer;
Benjamin Peterson47ff0732016-09-08 09:15:54 -0700571 unsigned long long flag;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000572
Oren Milman1d1d3e92017-08-20 18:35:36 +0300573 if (!PyTuple_Check(state)) {
574 PyErr_SetString(PyExc_TypeError, "state argument must be a tuple");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000575 return NULL;
Oren Milman1d1d3e92017-08-20 18:35:36 +0300576 }
577 if (!PyArg_ParseTuple(state, "OK;setstate(): illegal state argument",
578 &buffer, &flag))
579 {
580 return NULL;
581 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000582
Victor Stinner7d7e7752014-06-17 23:31:25 +0200583 self->pendingcr = (int) (flag & 1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000584 flag >>= 1;
585
586 if (self->decoder != Py_None)
Martin v. Löwisafe55bb2011-10-09 10:38:36 +0200587 return _PyObject_CallMethodId(self->decoder,
588 &PyId_setstate, "((OK))", buffer, flag);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000589 else
590 Py_RETURN_NONE;
591}
592
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300593/*[clinic input]
594_io.IncrementalNewlineDecoder.reset
595[clinic start generated code]*/
596
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000597static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +0300598_io_IncrementalNewlineDecoder_reset_impl(nldecoder_object *self)
599/*[clinic end generated code: output=32fa40c7462aa8ff input=728678ddaea776df]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000600{
601 self->seennl = 0;
602 self->pendingcr = 0;
603 if (self->decoder != Py_None)
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200604 return _PyObject_CallMethodNoArgs(self->decoder, _PyIO_str_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000605 else
606 Py_RETURN_NONE;
607}
608
609static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000610incrementalnewlinedecoder_newlines_get(nldecoder_object *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000611{
612 switch (self->seennl) {
613 case SEEN_CR:
614 return PyUnicode_FromString("\r");
615 case SEEN_LF:
616 return PyUnicode_FromString("\n");
617 case SEEN_CRLF:
618 return PyUnicode_FromString("\r\n");
619 case SEEN_CR | SEEN_LF:
620 return Py_BuildValue("ss", "\r", "\n");
621 case SEEN_CR | SEEN_CRLF:
622 return Py_BuildValue("ss", "\r", "\r\n");
623 case SEEN_LF | SEEN_CRLF:
624 return Py_BuildValue("ss", "\n", "\r\n");
625 case SEEN_CR | SEEN_LF | SEEN_CRLF:
626 return Py_BuildValue("sss", "\r", "\n", "\r\n");
627 default:
628 Py_RETURN_NONE;
629 }
630
631}
632
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000633/* TextIOWrapper */
634
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000635typedef PyObject *
636 (*encodefunc_t)(PyObject *, PyObject *);
637
638typedef struct
639{
640 PyObject_HEAD
641 int ok; /* initialized? */
Benjamin Petersond2e0c792009-05-01 20:40:59 +0000642 int detached;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000643 Py_ssize_t chunk_size;
644 PyObject *buffer;
645 PyObject *encoding;
646 PyObject *encoder;
647 PyObject *decoder;
648 PyObject *readnl;
649 PyObject *errors;
INADA Naoki507434f2017-12-21 09:59:53 +0900650 const char *writenl; /* ASCII-encoded; NULL stands for \n */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000651 char line_buffering;
Antoine Pitroue96ec682011-07-23 21:46:35 +0200652 char write_through;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000653 char readuniversal;
654 char readtranslate;
655 char writetranslate;
656 char seekable;
Antoine Pitroue96ec682011-07-23 21:46:35 +0200657 char has_read1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000658 char telling;
Antoine Pitrou796564c2013-07-30 19:59:21 +0200659 char finalizing;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000660 /* Specialized encoding func (see below) */
661 encodefunc_t encodefunc;
Antoine Pitroue4501852009-05-14 18:55:55 +0000662 /* Whether or not it's the start of the stream */
663 char encoding_start_of_stream;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000664
665 /* Reads and writes are internally buffered in order to speed things up.
666 However, any read will first flush the write buffer if itsn't empty.
Antoine Pitrou24f36292009-03-28 22:16:42 +0000667
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000668 Please also note that text to be written is first encoded before being
669 buffered. This is necessary so that encoding errors are immediately
670 reported to the caller, but it unfortunately means that the
671 IncrementalEncoder (whose encode() method is always written in Python)
672 becomes a bottleneck for small writes.
673 */
674 PyObject *decoded_chars; /* buffer for text returned from decoder */
675 Py_ssize_t decoded_chars_used; /* offset into _decoded_chars for read() */
Inada Naokibfba8c32019-05-16 15:03:20 +0900676 PyObject *pending_bytes; // data waiting to be written.
677 // ascii unicode, bytes, or list of them.
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000678 Py_ssize_t pending_bytes_count;
Antoine Pitrou211b81d2011-02-25 20:27:33 +0000679
Oren Milman13614e32017-08-24 19:51:24 +0300680 /* snapshot is either NULL, or a tuple (dec_flags, next_input) where
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000681 * dec_flags is the second (integer) item of the decoder state and
682 * next_input is the chunk of input bytes that comes next after the
683 * snapshot point. We use this to reconstruct decoder states in tell().
684 */
Antoine Pitrou211b81d2011-02-25 20:27:33 +0000685 PyObject *snapshot;
686 /* Bytes-to-characters ratio for the current chunk. Serves as input for
687 the heuristic in tell(). */
688 double b2cratio;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000689
690 /* Cache raw object if it's a FileIO object */
691 PyObject *raw;
692
693 PyObject *weakreflist;
694 PyObject *dict;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000695} textio;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000696
Zackery Spytz23db9352018-06-29 04:14:58 -0600697static void
698textiowrapper_set_decoded_chars(textio *self, PyObject *chars);
699
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000700/* A couple of specialized cases in order to bypass the slow incremental
701 encoding methods for the most popular encodings. */
702
703static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000704ascii_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000705{
INADA Naoki507434f2017-12-21 09:59:53 +0900706 return _PyUnicode_AsASCIIString(text, PyUnicode_AsUTF8(self->errors));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000707}
708
709static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000710utf16be_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000711{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100712 return _PyUnicode_EncodeUTF16(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900713 PyUnicode_AsUTF8(self->errors), 1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000714}
715
716static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000717utf16le_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000718{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100719 return _PyUnicode_EncodeUTF16(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900720 PyUnicode_AsUTF8(self->errors), -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000721}
722
723static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000724utf16_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000725{
Antoine Pitroue4501852009-05-14 18:55:55 +0000726 if (!self->encoding_start_of_stream) {
727 /* Skip the BOM and use native byte ordering */
Christian Heimes743e0cd2012-10-17 23:52:17 +0200728#if PY_BIG_ENDIAN
Antoine Pitroue4501852009-05-14 18:55:55 +0000729 return utf16be_encode(self, text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000730#else
Antoine Pitroue4501852009-05-14 18:55:55 +0000731 return utf16le_encode(self, text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000732#endif
Antoine Pitroue4501852009-05-14 18:55:55 +0000733 }
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100734 return _PyUnicode_EncodeUTF16(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900735 PyUnicode_AsUTF8(self->errors), 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000736}
737
Antoine Pitroue4501852009-05-14 18:55:55 +0000738static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000739utf32be_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000740{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100741 return _PyUnicode_EncodeUTF32(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900742 PyUnicode_AsUTF8(self->errors), 1);
Antoine Pitroue4501852009-05-14 18:55:55 +0000743}
744
745static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000746utf32le_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000747{
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100748 return _PyUnicode_EncodeUTF32(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900749 PyUnicode_AsUTF8(self->errors), -1);
Antoine Pitroue4501852009-05-14 18:55:55 +0000750}
751
752static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000753utf32_encode(textio *self, PyObject *text)
Antoine Pitroue4501852009-05-14 18:55:55 +0000754{
755 if (!self->encoding_start_of_stream) {
756 /* Skip the BOM and use native byte ordering */
Christian Heimes743e0cd2012-10-17 23:52:17 +0200757#if PY_BIG_ENDIAN
Antoine Pitroue4501852009-05-14 18:55:55 +0000758 return utf32be_encode(self, text);
759#else
760 return utf32le_encode(self, text);
761#endif
762 }
Antoine Pitrou5c398e82011-11-13 04:11:37 +0100763 return _PyUnicode_EncodeUTF32(text,
INADA Naoki507434f2017-12-21 09:59:53 +0900764 PyUnicode_AsUTF8(self->errors), 0);
Antoine Pitroue4501852009-05-14 18:55:55 +0000765}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000766
767static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000768utf8_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000769{
INADA Naoki507434f2017-12-21 09:59:53 +0900770 return _PyUnicode_AsUTF8String(text, PyUnicode_AsUTF8(self->errors));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000771}
772
773static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000774latin1_encode(textio *self, PyObject *text)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000775{
INADA Naoki507434f2017-12-21 09:59:53 +0900776 return _PyUnicode_AsLatin1String(text, PyUnicode_AsUTF8(self->errors));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000777}
778
Inada Naokibfba8c32019-05-16 15:03:20 +0900779// Return true when encoding can be skipped when text is ascii.
780static inline int
781is_asciicompat_encoding(encodefunc_t f)
782{
783 return f == (encodefunc_t) ascii_encode
784 || f == (encodefunc_t) latin1_encode
785 || f == (encodefunc_t) utf8_encode;
786}
787
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000788/* Map normalized encoding names onto the specialized encoding funcs */
789
790typedef struct {
791 const char *name;
792 encodefunc_t encodefunc;
793} encodefuncentry;
794
Serhiy Storchaka2d06e842015-12-25 19:53:18 +0200795static const encodefuncentry encodefuncs[] = {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000796 {"ascii", (encodefunc_t) ascii_encode},
797 {"iso8859-1", (encodefunc_t) latin1_encode},
Antoine Pitroue4501852009-05-14 18:55:55 +0000798 {"utf-8", (encodefunc_t) utf8_encode},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000799 {"utf-16-be", (encodefunc_t) utf16be_encode},
800 {"utf-16-le", (encodefunc_t) utf16le_encode},
801 {"utf-16", (encodefunc_t) utf16_encode},
Antoine Pitroue4501852009-05-14 18:55:55 +0000802 {"utf-32-be", (encodefunc_t) utf32be_encode},
803 {"utf-32-le", (encodefunc_t) utf32le_encode},
804 {"utf-32", (encodefunc_t) utf32_encode},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000805 {NULL, NULL}
806};
807
INADA Naoki507434f2017-12-21 09:59:53 +0900808static int
809validate_newline(const char *newline)
810{
811 if (newline && newline[0] != '\0'
812 && !(newline[0] == '\n' && newline[1] == '\0')
813 && !(newline[0] == '\r' && newline[1] == '\0')
814 && !(newline[0] == '\r' && newline[1] == '\n' && newline[2] == '\0')) {
815 PyErr_Format(PyExc_ValueError,
816 "illegal newline value: %s", newline);
817 return -1;
818 }
819 return 0;
820}
821
822static int
823set_newline(textio *self, const char *newline)
824{
825 PyObject *old = self->readnl;
826 if (newline == NULL) {
827 self->readnl = NULL;
828 }
829 else {
830 self->readnl = PyUnicode_FromString(newline);
831 if (self->readnl == NULL) {
832 self->readnl = old;
833 return -1;
834 }
835 }
836 self->readuniversal = (newline == NULL || newline[0] == '\0');
837 self->readtranslate = (newline == NULL);
838 self->writetranslate = (newline == NULL || newline[0] != '\0');
839 if (!self->readuniversal && self->readnl != NULL) {
840 // validate_newline() accepts only ASCII newlines.
841 assert(PyUnicode_KIND(self->readnl) == PyUnicode_1BYTE_KIND);
842 self->writenl = (const char *)PyUnicode_1BYTE_DATA(self->readnl);
843 if (strcmp(self->writenl, "\n") == 0) {
844 self->writenl = NULL;
845 }
846 }
847 else {
848#ifdef MS_WINDOWS
849 self->writenl = "\r\n";
850#else
851 self->writenl = NULL;
852#endif
853 }
854 Py_XDECREF(old);
855 return 0;
856}
857
858static int
859_textiowrapper_set_decoder(textio *self, PyObject *codec_info,
860 const char *errors)
861{
862 PyObject *res;
863 int r;
864
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200865 res = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_readable);
INADA Naoki507434f2017-12-21 09:59:53 +0900866 if (res == NULL)
867 return -1;
868
869 r = PyObject_IsTrue(res);
870 Py_DECREF(res);
871 if (r == -1)
872 return -1;
873
874 if (r != 1)
875 return 0;
876
877 Py_CLEAR(self->decoder);
878 self->decoder = _PyCodecInfo_GetIncrementalDecoder(codec_info, errors);
879 if (self->decoder == NULL)
880 return -1;
881
882 if (self->readuniversal) {
883 PyObject *incrementalDecoder = PyObject_CallFunction(
884 (PyObject *)&PyIncrementalNewlineDecoder_Type,
885 "Oi", self->decoder, (int)self->readtranslate);
886 if (incrementalDecoder == NULL)
887 return -1;
888 Py_CLEAR(self->decoder);
889 self->decoder = incrementalDecoder;
890 }
891
892 return 0;
893}
894
895static PyObject*
896_textiowrapper_decode(PyObject *decoder, PyObject *bytes, int eof)
897{
898 PyObject *chars;
899
900 if (Py_TYPE(decoder) == &PyIncrementalNewlineDecoder_Type)
901 chars = _PyIncrementalNewlineDecoder_decode(decoder, bytes, eof);
902 else
903 chars = PyObject_CallMethodObjArgs(decoder, _PyIO_str_decode, bytes,
904 eof ? Py_True : Py_False, NULL);
905
906 if (check_decoded(chars) < 0)
907 // check_decoded already decreases refcount
908 return NULL;
909
910 return chars;
911}
912
913static int
914_textiowrapper_set_encoder(textio *self, PyObject *codec_info,
915 const char *errors)
916{
917 PyObject *res;
918 int r;
919
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200920 res = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_writable);
INADA Naoki507434f2017-12-21 09:59:53 +0900921 if (res == NULL)
922 return -1;
923
924 r = PyObject_IsTrue(res);
925 Py_DECREF(res);
926 if (r == -1)
927 return -1;
928
929 if (r != 1)
930 return 0;
931
932 Py_CLEAR(self->encoder);
933 self->encodefunc = NULL;
934 self->encoder = _PyCodecInfo_GetIncrementalEncoder(codec_info, errors);
935 if (self->encoder == NULL)
936 return -1;
937
938 /* Get the normalized named of the codec */
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200939 if (_PyObject_LookupAttrId(codec_info, &PyId_name, &res) < 0) {
940 return -1;
INADA Naoki507434f2017-12-21 09:59:53 +0900941 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200942 if (res != NULL && PyUnicode_Check(res)) {
INADA Naoki507434f2017-12-21 09:59:53 +0900943 const encodefuncentry *e = encodefuncs;
944 while (e->name != NULL) {
945 if (_PyUnicode_EqualToASCIIString(res, e->name)) {
946 self->encodefunc = e->encodefunc;
947 break;
948 }
949 e++;
950 }
951 }
952 Py_XDECREF(res);
953
954 return 0;
955}
956
957static int
958_textiowrapper_fix_encoder_state(textio *self)
959{
960 if (!self->seekable || !self->encoder) {
961 return 0;
962 }
963
964 self->encoding_start_of_stream = 1;
965
Jeroen Demeyer762f93f2019-07-08 10:19:25 +0200966 PyObject *cookieObj = _PyObject_CallMethodNoArgs(
967 self->buffer, _PyIO_str_tell);
INADA Naoki507434f2017-12-21 09:59:53 +0900968 if (cookieObj == NULL) {
969 return -1;
970 }
971
972 int cmp = PyObject_RichCompareBool(cookieObj, _PyLong_Zero, Py_EQ);
973 Py_DECREF(cookieObj);
974 if (cmp < 0) {
975 return -1;
976 }
977
978 if (cmp == 0) {
979 self->encoding_start_of_stream = 0;
980 PyObject *res = PyObject_CallMethodObjArgs(
981 self->encoder, _PyIO_str_setstate, _PyLong_Zero, NULL);
982 if (res == NULL) {
983 return -1;
984 }
985 Py_DECREF(res);
986 }
987
988 return 0;
989}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000990
Victor Stinner22eb6892019-06-26 00:51:05 +0200991static int
992io_check_errors(PyObject *errors)
993{
994 assert(errors != NULL && errors != Py_None);
995
996 PyInterpreterState *interp = _PyInterpreterState_GET_UNSAFE();
997#ifndef Py_DEBUG
998 /* In release mode, only check in development mode (-X dev) */
999 if (!interp->config.dev_mode) {
1000 return 0;
1001 }
1002#else
1003 /* Always check in debug mode */
1004#endif
1005
1006 /* Avoid calling PyCodec_LookupError() before the codec registry is ready:
1007 before_PyUnicode_InitEncodings() is called. */
1008 if (!interp->fs_codec.encoding) {
1009 return 0;
1010 }
1011
1012 Py_ssize_t name_length;
1013 const char *name = PyUnicode_AsUTF8AndSize(errors, &name_length);
1014 if (name == NULL) {
1015 return -1;
1016 }
1017 if (strlen(name) != (size_t)name_length) {
1018 PyErr_SetString(PyExc_ValueError, "embedded null character in errors");
1019 return -1;
1020 }
1021 PyObject *handler = PyCodec_LookupError(name);
1022 if (handler != NULL) {
1023 Py_DECREF(handler);
1024 return 0;
1025 }
1026 return -1;
1027}
1028
1029
1030
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001031/*[clinic input]
1032_io.TextIOWrapper.__init__
1033 buffer: object
Larry Hastingsdbfdc382015-05-04 06:59:46 -07001034 encoding: str(accept={str, NoneType}) = NULL
INADA Naoki507434f2017-12-21 09:59:53 +09001035 errors: object = None
Larry Hastingsdbfdc382015-05-04 06:59:46 -07001036 newline: str(accept={str, NoneType}) = NULL
Serhiy Storchaka202fda52017-03-12 10:10:47 +02001037 line_buffering: bool(accept={int}) = False
1038 write_through: bool(accept={int}) = False
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001039
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001040Character and line based layer over a BufferedIOBase object, buffer.
1041
1042encoding gives the name of the encoding that the stream will be
1043decoded or encoded with. It defaults to locale.getpreferredencoding(False).
1044
1045errors determines the strictness of encoding and decoding (see
1046help(codecs.Codec) or the documentation for codecs.register) and
1047defaults to "strict".
1048
1049newline controls how line endings are handled. It can be None, '',
1050'\n', '\r', and '\r\n'. It works as follows:
1051
1052* On input, if newline is None, universal newlines mode is
1053 enabled. Lines in the input can end in '\n', '\r', or '\r\n', and
1054 these are translated into '\n' before being returned to the
1055 caller. If it is '', universal newline mode is enabled, but line
1056 endings are returned to the caller untranslated. If it has any of
1057 the other legal values, input lines are only terminated by the given
1058 string, and the line ending is returned to the caller untranslated.
1059
1060* On output, if newline is None, any '\n' characters written are
1061 translated to the system default line separator, os.linesep. If
1062 newline is '' or '\n', no translation takes place. If newline is any
1063 of the other legal values, any '\n' characters written are translated
1064 to the given string.
1065
1066If line_buffering is True, a call to flush is implied when a call to
1067write contains a newline character.
1068[clinic start generated code]*/
1069
1070static int
1071_io_TextIOWrapper___init___impl(textio *self, PyObject *buffer,
INADA Naoki507434f2017-12-21 09:59:53 +09001072 const char *encoding, PyObject *errors,
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001073 const char *newline, int line_buffering,
1074 int write_through)
INADA Naoki507434f2017-12-21 09:59:53 +09001075/*[clinic end generated code: output=72267c0c01032ed2 input=1c5dd5d78bfcc675]*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001076{
1077 PyObject *raw, *codec_info = NULL;
1078 _PyIO_State *state = NULL;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001079 PyObject *res;
1080 int r;
1081
1082 self->ok = 0;
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001083 self->detached = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001084
INADA Naoki507434f2017-12-21 09:59:53 +09001085 if (errors == Py_None) {
1086 errors = _PyUnicode_FromId(&PyId_strict); /* borrowed */
INADA Naoki4856b0f2017-12-24 10:29:19 +09001087 if (errors == NULL) {
1088 return -1;
1089 }
INADA Naoki507434f2017-12-21 09:59:53 +09001090 }
1091 else if (!PyUnicode_Check(errors)) {
1092 // Check 'errors' argument here because Argument Clinic doesn't support
1093 // 'str(accept={str, NoneType})' converter.
1094 PyErr_Format(
1095 PyExc_TypeError,
1096 "TextIOWrapper() argument 'errors' must be str or None, not %.50s",
1097 errors->ob_type->tp_name);
1098 return -1;
1099 }
Victor Stinner22eb6892019-06-26 00:51:05 +02001100 else if (io_check_errors(errors)) {
1101 return -1;
1102 }
INADA Naoki507434f2017-12-21 09:59:53 +09001103
1104 if (validate_newline(newline) < 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001105 return -1;
1106 }
1107
1108 Py_CLEAR(self->buffer);
1109 Py_CLEAR(self->encoding);
1110 Py_CLEAR(self->encoder);
1111 Py_CLEAR(self->decoder);
1112 Py_CLEAR(self->readnl);
1113 Py_CLEAR(self->decoded_chars);
1114 Py_CLEAR(self->pending_bytes);
1115 Py_CLEAR(self->snapshot);
1116 Py_CLEAR(self->errors);
1117 Py_CLEAR(self->raw);
1118 self->decoded_chars_used = 0;
1119 self->pending_bytes_count = 0;
1120 self->encodefunc = NULL;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001121 self->b2cratio = 0.0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001122
1123 if (encoding == NULL) {
1124 /* Try os.device_encoding(fileno) */
1125 PyObject *fileno;
Antoine Pitrou712cb732013-12-21 15:51:54 +01001126 state = IO_STATE();
1127 if (state == NULL)
1128 goto error;
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001129 fileno = _PyObject_CallMethodIdNoArgs(buffer, &PyId_fileno);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001130 /* Ignore only AttributeError and UnsupportedOperation */
1131 if (fileno == NULL) {
1132 if (PyErr_ExceptionMatches(PyExc_AttributeError) ||
1133 PyErr_ExceptionMatches(state->unsupported_operation)) {
1134 PyErr_Clear();
1135 }
1136 else {
1137 goto error;
1138 }
1139 }
1140 else {
Serhiy Storchaka78980432013-01-15 01:12:17 +02001141 int fd = _PyLong_AsInt(fileno);
Brett Cannonefb00c02012-02-29 18:31:31 -05001142 Py_DECREF(fileno);
1143 if (fd == -1 && PyErr_Occurred()) {
1144 goto error;
1145 }
1146
1147 self->encoding = _Py_device_encoding(fd);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001148 if (self->encoding == NULL)
1149 goto error;
1150 else if (!PyUnicode_Check(self->encoding))
1151 Py_CLEAR(self->encoding);
1152 }
1153 }
1154 if (encoding == NULL && self->encoding == NULL) {
Antoine Pitrou932ff832013-08-01 21:04:50 +02001155 PyObject *locale_module = _PyIO_get_locale_module(state);
1156 if (locale_module == NULL)
1157 goto catch_ImportError;
Victor Stinner61bdb0d2016-12-09 15:39:28 +01001158 self->encoding = _PyObject_CallMethodIdObjArgs(
1159 locale_module, &PyId_getpreferredencoding, Py_False, NULL);
Antoine Pitrou932ff832013-08-01 21:04:50 +02001160 Py_DECREF(locale_module);
1161 if (self->encoding == NULL) {
1162 catch_ImportError:
1163 /*
Martin Panter7462b6492015-11-02 03:37:02 +00001164 Importing locale can raise an ImportError because of
1165 _functools, and locale.getpreferredencoding can raise an
Antoine Pitrou932ff832013-08-01 21:04:50 +02001166 ImportError if _locale is not available. These will happen
1167 during module building.
1168 */
1169 if (PyErr_ExceptionMatches(PyExc_ImportError)) {
1170 PyErr_Clear();
1171 self->encoding = PyUnicode_FromString("ascii");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001172 }
Antoine Pitrou932ff832013-08-01 21:04:50 +02001173 else
1174 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001175 }
Antoine Pitrou932ff832013-08-01 21:04:50 +02001176 else if (!PyUnicode_Check(self->encoding))
1177 Py_CLEAR(self->encoding);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001178 }
Victor Stinnerf6c57832010-05-19 01:17:01 +00001179 if (self->encoding != NULL) {
Serhiy Storchaka06515832016-11-20 09:13:07 +02001180 encoding = PyUnicode_AsUTF8(self->encoding);
Victor Stinnerf6c57832010-05-19 01:17:01 +00001181 if (encoding == NULL)
1182 goto error;
1183 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001184 else if (encoding != NULL) {
1185 self->encoding = PyUnicode_FromString(encoding);
1186 if (self->encoding == NULL)
1187 goto error;
1188 }
1189 else {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03001190 PyErr_SetString(PyExc_OSError,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001191 "could not determine default encoding");
Serhiy Storchakad6238a72017-09-24 02:49:58 +03001192 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001193 }
1194
Nick Coghlana9b15242014-02-04 22:11:18 +10001195 /* Check we have been asked for a real text encoding */
1196 codec_info = _PyCodec_LookupTextEncoding(encoding, "codecs.open()");
1197 if (codec_info == NULL) {
1198 Py_CLEAR(self->encoding);
1199 goto error;
1200 }
1201
1202 /* XXX: Failures beyond this point have the potential to leak elements
1203 * of the partially constructed object (like self->encoding)
1204 */
1205
INADA Naoki507434f2017-12-21 09:59:53 +09001206 Py_INCREF(errors);
1207 self->errors = errors;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001208 self->chunk_size = 8192;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001209 self->line_buffering = line_buffering;
Antoine Pitroue96ec682011-07-23 21:46:35 +02001210 self->write_through = write_through;
INADA Naoki507434f2017-12-21 09:59:53 +09001211 if (set_newline(self, newline) < 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001212 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001213 }
1214
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001215 self->buffer = buffer;
1216 Py_INCREF(buffer);
Antoine Pitrou24f36292009-03-28 22:16:42 +00001217
INADA Naoki507434f2017-12-21 09:59:53 +09001218 /* Build the decoder object */
1219 if (_textiowrapper_set_decoder(self, codec_info, PyUnicode_AsUTF8(errors)) != 0)
1220 goto error;
1221
1222 /* Build the encoder object */
1223 if (_textiowrapper_set_encoder(self, codec_info, PyUnicode_AsUTF8(errors)) != 0)
1224 goto error;
1225
1226 /* Finished sorting out the codec details */
1227 Py_CLEAR(codec_info);
1228
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001229 if (Py_TYPE(buffer) == &PyBufferedReader_Type ||
1230 Py_TYPE(buffer) == &PyBufferedWriter_Type ||
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001231 Py_TYPE(buffer) == &PyBufferedRandom_Type)
1232 {
1233 if (_PyObject_LookupAttrId(buffer, &PyId_raw, &raw) < 0)
1234 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001235 /* Cache the raw FileIO object to speed up 'closed' checks */
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001236 if (raw != NULL) {
1237 if (Py_TYPE(raw) == &PyFileIO_Type)
1238 self->raw = raw;
Benjamin Peterson2cfca792009-06-06 20:46:48 +00001239 else
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001240 Py_DECREF(raw);
Benjamin Peterson2cfca792009-06-06 20:46:48 +00001241 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001242 }
1243
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001244 res = _PyObject_CallMethodIdNoArgs(buffer, &PyId_seekable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001245 if (res == NULL)
1246 goto error;
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001247 r = PyObject_IsTrue(res);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001248 Py_DECREF(res);
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001249 if (r < 0)
1250 goto error;
1251 self->seekable = self->telling = r;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001252
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001253 r = _PyObject_LookupAttr(buffer, _PyIO_str_read1, &res);
1254 if (r < 0) {
Serhiy Storchaka4d9aec02018-01-16 18:34:21 +02001255 goto error;
1256 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001257 Py_XDECREF(res);
1258 self->has_read1 = r;
Antoine Pitroue96ec682011-07-23 21:46:35 +02001259
Antoine Pitroue4501852009-05-14 18:55:55 +00001260 self->encoding_start_of_stream = 0;
INADA Naoki507434f2017-12-21 09:59:53 +09001261 if (_textiowrapper_fix_encoder_state(self) < 0) {
1262 goto error;
Antoine Pitroue4501852009-05-14 18:55:55 +00001263 }
1264
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001265 self->ok = 1;
1266 return 0;
1267
1268 error:
Nick Coghlana9b15242014-02-04 22:11:18 +10001269 Py_XDECREF(codec_info);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001270 return -1;
1271}
1272
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001273/* Return *default_value* if ob is None, 0 if ob is false, 1 if ob is true,
1274 * -1 on error.
1275 */
1276static int
1277convert_optional_bool(PyObject *obj, int default_value)
1278{
1279 long v;
1280 if (obj == Py_None) {
1281 v = default_value;
1282 }
1283 else {
1284 v = PyLong_AsLong(obj);
1285 if (v == -1 && PyErr_Occurred())
1286 return -1;
1287 }
1288 return v != 0;
1289}
1290
INADA Naoki507434f2017-12-21 09:59:53 +09001291static int
1292textiowrapper_change_encoding(textio *self, PyObject *encoding,
1293 PyObject *errors, int newline_changed)
1294{
1295 /* Use existing settings where new settings are not specified */
1296 if (encoding == Py_None && errors == Py_None && !newline_changed) {
1297 return 0; // no change
1298 }
1299
1300 if (encoding == Py_None) {
1301 encoding = self->encoding;
1302 if (errors == Py_None) {
1303 errors = self->errors;
1304 }
1305 }
1306 else if (errors == Py_None) {
1307 errors = _PyUnicode_FromId(&PyId_strict);
INADA Naoki4856b0f2017-12-24 10:29:19 +09001308 if (errors == NULL) {
1309 return -1;
1310 }
INADA Naoki507434f2017-12-21 09:59:53 +09001311 }
1312
1313 const char *c_errors = PyUnicode_AsUTF8(errors);
1314 if (c_errors == NULL) {
1315 return -1;
1316 }
1317
1318 // Create new encoder & decoder
1319 PyObject *codec_info = _PyCodec_LookupTextEncoding(
1320 PyUnicode_AsUTF8(encoding), "codecs.open()");
1321 if (codec_info == NULL) {
1322 return -1;
1323 }
1324 if (_textiowrapper_set_decoder(self, codec_info, c_errors) != 0 ||
1325 _textiowrapper_set_encoder(self, codec_info, c_errors) != 0) {
1326 Py_DECREF(codec_info);
1327 return -1;
1328 }
1329 Py_DECREF(codec_info);
1330
1331 Py_INCREF(encoding);
1332 Py_INCREF(errors);
1333 Py_SETREF(self->encoding, encoding);
1334 Py_SETREF(self->errors, errors);
1335
1336 return _textiowrapper_fix_encoder_state(self);
1337}
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001338
1339/*[clinic input]
1340_io.TextIOWrapper.reconfigure
1341 *
INADA Naoki507434f2017-12-21 09:59:53 +09001342 encoding: object = None
1343 errors: object = None
1344 newline as newline_obj: object(c_default="NULL") = None
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001345 line_buffering as line_buffering_obj: object = None
1346 write_through as write_through_obj: object = None
1347
1348Reconfigure the text stream with new parameters.
1349
1350This also does an implicit stream flush.
1351
1352[clinic start generated code]*/
1353
1354static PyObject *
INADA Naoki507434f2017-12-21 09:59:53 +09001355_io_TextIOWrapper_reconfigure_impl(textio *self, PyObject *encoding,
1356 PyObject *errors, PyObject *newline_obj,
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001357 PyObject *line_buffering_obj,
1358 PyObject *write_through_obj)
INADA Naoki507434f2017-12-21 09:59:53 +09001359/*[clinic end generated code: output=52b812ff4b3d4b0f input=671e82136e0f5822]*/
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001360{
1361 int line_buffering;
1362 int write_through;
INADA Naoki507434f2017-12-21 09:59:53 +09001363 const char *newline = NULL;
1364
1365 /* Check if something is in the read buffer */
1366 if (self->decoded_chars != NULL) {
1367 if (encoding != Py_None || errors != Py_None || newline_obj != NULL) {
Serhiy Storchaka34fd4c22018-11-05 16:20:25 +02001368 _unsupported("It is not possible to set the encoding or newline "
INADA Naoki507434f2017-12-21 09:59:53 +09001369 "of stream after the first read");
1370 return NULL;
1371 }
1372 }
1373
1374 if (newline_obj != NULL && newline_obj != Py_None) {
1375 newline = PyUnicode_AsUTF8(newline_obj);
1376 if (newline == NULL || validate_newline(newline) < 0) {
1377 return NULL;
1378 }
1379 }
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001380
1381 line_buffering = convert_optional_bool(line_buffering_obj,
1382 self->line_buffering);
1383 write_through = convert_optional_bool(write_through_obj,
1384 self->write_through);
1385 if (line_buffering < 0 || write_through < 0) {
1386 return NULL;
1387 }
INADA Naoki507434f2017-12-21 09:59:53 +09001388
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001389 PyObject *res = _PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001390 if (res == NULL) {
1391 return NULL;
1392 }
INADA Naoki507434f2017-12-21 09:59:53 +09001393 Py_DECREF(res);
1394 self->b2cratio = 0;
1395
1396 if (newline_obj != NULL && set_newline(self, newline) < 0) {
1397 return NULL;
1398 }
1399
1400 if (textiowrapper_change_encoding(
1401 self, encoding, errors, newline_obj != NULL) < 0) {
1402 return NULL;
1403 }
1404
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02001405 self->line_buffering = line_buffering;
1406 self->write_through = write_through;
1407 Py_RETURN_NONE;
1408}
1409
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001410static int
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001411textiowrapper_clear(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001412{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001413 self->ok = 0;
1414 Py_CLEAR(self->buffer);
1415 Py_CLEAR(self->encoding);
1416 Py_CLEAR(self->encoder);
1417 Py_CLEAR(self->decoder);
1418 Py_CLEAR(self->readnl);
1419 Py_CLEAR(self->decoded_chars);
1420 Py_CLEAR(self->pending_bytes);
1421 Py_CLEAR(self->snapshot);
1422 Py_CLEAR(self->errors);
1423 Py_CLEAR(self->raw);
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001424
1425 Py_CLEAR(self->dict);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001426 return 0;
1427}
1428
1429static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001430textiowrapper_dealloc(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001431{
Antoine Pitrou796564c2013-07-30 19:59:21 +02001432 self->finalizing = 1;
1433 if (_PyIOBase_finalize((PyObject *) self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001434 return;
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001435 self->ok = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001436 _PyObject_GC_UNTRACK(self);
1437 if (self->weakreflist != NULL)
1438 PyObject_ClearWeakRefs((PyObject *)self);
Serhiy Storchakaa7c972e2016-11-03 15:37:01 +02001439 textiowrapper_clear(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001440 Py_TYPE(self)->tp_free((PyObject *)self);
1441}
1442
1443static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001444textiowrapper_traverse(textio *self, visitproc visit, void *arg)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001445{
1446 Py_VISIT(self->buffer);
1447 Py_VISIT(self->encoding);
1448 Py_VISIT(self->encoder);
1449 Py_VISIT(self->decoder);
1450 Py_VISIT(self->readnl);
1451 Py_VISIT(self->decoded_chars);
1452 Py_VISIT(self->pending_bytes);
1453 Py_VISIT(self->snapshot);
1454 Py_VISIT(self->errors);
1455 Py_VISIT(self->raw);
1456
1457 Py_VISIT(self->dict);
1458 return 0;
1459}
1460
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001461static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001462textiowrapper_closed_get(textio *self, void *context);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001463
1464/* This macro takes some shortcuts to make the common case faster. */
1465#define CHECK_CLOSED(self) \
1466 do { \
1467 int r; \
1468 PyObject *_res; \
1469 if (Py_TYPE(self) == &PyTextIOWrapper_Type) { \
1470 if (self->raw != NULL) \
1471 r = _PyFileIO_closed(self->raw); \
1472 else { \
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001473 _res = textiowrapper_closed_get(self, NULL); \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001474 if (_res == NULL) \
1475 return NULL; \
1476 r = PyObject_IsTrue(_res); \
1477 Py_DECREF(_res); \
1478 if (r < 0) \
1479 return NULL; \
1480 } \
1481 if (r > 0) { \
1482 PyErr_SetString(PyExc_ValueError, \
1483 "I/O operation on closed file."); \
1484 return NULL; \
1485 } \
1486 } \
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001487 else if (_PyIOBase_check_closed((PyObject *)self, Py_True) == NULL) \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001488 return NULL; \
1489 } while (0)
1490
1491#define CHECK_INITIALIZED(self) \
1492 if (self->ok <= 0) { \
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001493 PyErr_SetString(PyExc_ValueError, \
1494 "I/O operation on uninitialized object"); \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001495 return NULL; \
1496 }
1497
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001498#define CHECK_ATTACHED(self) \
1499 CHECK_INITIALIZED(self); \
1500 if (self->detached) { \
1501 PyErr_SetString(PyExc_ValueError, \
1502 "underlying buffer has been detached"); \
1503 return NULL; \
1504 }
1505
1506#define CHECK_ATTACHED_INT(self) \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001507 if (self->ok <= 0) { \
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001508 PyErr_SetString(PyExc_ValueError, \
1509 "I/O operation on uninitialized object"); \
1510 return -1; \
1511 } else if (self->detached) { \
1512 PyErr_SetString(PyExc_ValueError, \
1513 "underlying buffer has been detached"); \
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001514 return -1; \
1515 }
1516
1517
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001518/*[clinic input]
1519_io.TextIOWrapper.detach
1520[clinic start generated code]*/
1521
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001522static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001523_io_TextIOWrapper_detach_impl(textio *self)
1524/*[clinic end generated code: output=7ba3715cd032d5f2 input=e5a71fbda9e1d9f9]*/
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001525{
1526 PyObject *buffer, *res;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001527 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001528 res = _PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001529 if (res == NULL)
1530 return NULL;
1531 Py_DECREF(res);
1532 buffer = self->buffer;
1533 self->buffer = NULL;
1534 self->detached = 1;
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001535 return buffer;
1536}
1537
Antoine Pitrou24f36292009-03-28 22:16:42 +00001538/* Flush the internal write buffer. This doesn't explicitly flush the
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001539 underlying buffered object, though. */
1540static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001541_textiowrapper_writeflush(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001542{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001543 if (self->pending_bytes == NULL)
1544 return 0;
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001545
Inada Naokibfba8c32019-05-16 15:03:20 +09001546 PyObject *pending = self->pending_bytes;
1547 PyObject *b;
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001548
Inada Naokibfba8c32019-05-16 15:03:20 +09001549 if (PyBytes_Check(pending)) {
1550 b = pending;
1551 Py_INCREF(b);
1552 }
1553 else if (PyUnicode_Check(pending)) {
1554 assert(PyUnicode_IS_ASCII(pending));
1555 assert(PyUnicode_GET_LENGTH(pending) == self->pending_bytes_count);
1556 b = PyBytes_FromStringAndSize(
1557 PyUnicode_DATA(pending), PyUnicode_GET_LENGTH(pending));
1558 if (b == NULL) {
1559 return -1;
1560 }
1561 }
1562 else {
1563 assert(PyList_Check(pending));
1564 b = PyBytes_FromStringAndSize(NULL, self->pending_bytes_count);
1565 if (b == NULL) {
1566 return -1;
1567 }
1568
1569 char *buf = PyBytes_AsString(b);
1570 Py_ssize_t pos = 0;
1571
1572 for (Py_ssize_t i = 0; i < PyList_GET_SIZE(pending); i++) {
1573 PyObject *obj = PyList_GET_ITEM(pending, i);
1574 char *src;
1575 Py_ssize_t len;
1576 if (PyUnicode_Check(obj)) {
1577 assert(PyUnicode_IS_ASCII(obj));
1578 src = PyUnicode_DATA(obj);
1579 len = PyUnicode_GET_LENGTH(obj);
1580 }
1581 else {
1582 assert(PyBytes_Check(obj));
1583 if (PyBytes_AsStringAndSize(obj, &src, &len) < 0) {
1584 Py_DECREF(b);
1585 return -1;
1586 }
1587 }
1588 memcpy(buf + pos, src, len);
1589 pos += len;
1590 }
1591 assert(pos == self->pending_bytes_count);
1592 }
1593
1594 self->pending_bytes_count = 0;
1595 self->pending_bytes = NULL;
Amaury Forgeot d'Arcccd686a2009-08-29 23:00:38 +00001596 Py_DECREF(pending);
Inada Naokibfba8c32019-05-16 15:03:20 +09001597
1598 PyObject *ret;
Gregory P. Smithb9817b02013-02-01 13:03:39 -08001599 do {
1600 ret = PyObject_CallMethodObjArgs(self->buffer,
1601 _PyIO_str_write, b, NULL);
1602 } while (ret == NULL && _PyIO_trap_eintr());
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001603 Py_DECREF(b);
1604 if (ret == NULL)
1605 return -1;
1606 Py_DECREF(ret);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001607 return 0;
1608}
1609
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001610/*[clinic input]
1611_io.TextIOWrapper.write
1612 text: unicode
1613 /
1614[clinic start generated code]*/
1615
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001616static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001617_io_TextIOWrapper_write_impl(textio *self, PyObject *text)
1618/*[clinic end generated code: output=d2deb0d50771fcec input=fdf19153584a0e44]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001619{
1620 PyObject *ret;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001621 PyObject *b;
1622 Py_ssize_t textlen;
1623 int haslf = 0;
Antoine Pitrouc644e7c2014-05-09 00:24:50 +02001624 int needflush = 0, text_needflush = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001625
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001626 if (PyUnicode_READY(text) == -1)
1627 return NULL;
1628
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001629 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001630 CHECK_CLOSED(self);
1631
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001632 if (self->encoder == NULL)
1633 return _unsupported("not writable");
Benjamin Peterson81971ea2009-05-14 22:01:31 +00001634
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001635 Py_INCREF(text);
1636
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001637 textlen = PyUnicode_GET_LENGTH(text);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001638
1639 if ((self->writetranslate && self->writenl != NULL) || self->line_buffering)
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001640 if (PyUnicode_FindChar(text, '\n', 0, PyUnicode_GET_LENGTH(text), 1) != -1)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001641 haslf = 1;
1642
1643 if (haslf && self->writetranslate && self->writenl != NULL) {
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001644 PyObject *newtext = _PyObject_CallMethodId(
1645 text, &PyId_replace, "ss", "\n", self->writenl);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001646 Py_DECREF(text);
1647 if (newtext == NULL)
1648 return NULL;
1649 text = newtext;
1650 }
1651
Antoine Pitroue96ec682011-07-23 21:46:35 +02001652 if (self->write_through)
Antoine Pitrouc644e7c2014-05-09 00:24:50 +02001653 text_needflush = 1;
1654 if (self->line_buffering &&
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001655 (haslf ||
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001656 PyUnicode_FindChar(text, '\r', 0, PyUnicode_GET_LENGTH(text), 1) != -1))
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001657 needflush = 1;
1658
1659 /* XXX What if we were just reading? */
Antoine Pitroue4501852009-05-14 18:55:55 +00001660 if (self->encodefunc != NULL) {
Inada Naokibfba8c32019-05-16 15:03:20 +09001661 if (PyUnicode_IS_ASCII(text) && is_asciicompat_encoding(self->encodefunc)) {
1662 b = text;
1663 Py_INCREF(b);
1664 }
1665 else {
1666 b = (*self->encodefunc)((PyObject *) self, text);
1667 }
Antoine Pitroue4501852009-05-14 18:55:55 +00001668 self->encoding_start_of_stream = 0;
1669 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001670 else
1671 b = PyObject_CallMethodObjArgs(self->encoder,
1672 _PyIO_str_encode, text, NULL);
Inada Naokibfba8c32019-05-16 15:03:20 +09001673
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001674 Py_DECREF(text);
1675 if (b == NULL)
1676 return NULL;
Inada Naokibfba8c32019-05-16 15:03:20 +09001677 if (b != text && !PyBytes_Check(b)) {
Oren Milmana5b4ea12017-08-25 21:14:54 +03001678 PyErr_Format(PyExc_TypeError,
1679 "encoder should return a bytes object, not '%.200s'",
1680 Py_TYPE(b)->tp_name);
1681 Py_DECREF(b);
1682 return NULL;
1683 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001684
Inada Naokibfba8c32019-05-16 15:03:20 +09001685 Py_ssize_t bytes_len;
1686 if (b == text) {
1687 bytes_len = PyUnicode_GET_LENGTH(b);
1688 }
1689 else {
1690 bytes_len = PyBytes_GET_SIZE(b);
1691 }
1692
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001693 if (self->pending_bytes == NULL) {
Inada Naokibfba8c32019-05-16 15:03:20 +09001694 self->pending_bytes_count = 0;
1695 self->pending_bytes = b;
1696 }
1697 else if (!PyList_CheckExact(self->pending_bytes)) {
1698 PyObject *list = PyList_New(2);
1699 if (list == NULL) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001700 Py_DECREF(b);
1701 return NULL;
1702 }
Inada Naokibfba8c32019-05-16 15:03:20 +09001703 PyList_SET_ITEM(list, 0, self->pending_bytes);
1704 PyList_SET_ITEM(list, 1, b);
1705 self->pending_bytes = list;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001706 }
Inada Naokibfba8c32019-05-16 15:03:20 +09001707 else {
1708 if (PyList_Append(self->pending_bytes, b) < 0) {
1709 Py_DECREF(b);
1710 return NULL;
1711 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001712 Py_DECREF(b);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001713 }
Inada Naokibfba8c32019-05-16 15:03:20 +09001714
1715 self->pending_bytes_count += bytes_len;
Antoine Pitrouc644e7c2014-05-09 00:24:50 +02001716 if (self->pending_bytes_count > self->chunk_size || needflush ||
1717 text_needflush) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001718 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001719 return NULL;
1720 }
Antoine Pitrou24f36292009-03-28 22:16:42 +00001721
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001722 if (needflush) {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001723 ret = _PyObject_CallMethodNoArgs(self->buffer, _PyIO_str_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001724 if (ret == NULL)
1725 return NULL;
1726 Py_DECREF(ret);
1727 }
1728
Zackery Spytz23db9352018-06-29 04:14:58 -06001729 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001730 Py_CLEAR(self->snapshot);
1731
1732 if (self->decoder) {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001733 ret = _PyObject_CallMethodIdNoArgs(self->decoder, &PyId_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001734 if (ret == NULL)
1735 return NULL;
1736 Py_DECREF(ret);
1737 }
1738
1739 return PyLong_FromSsize_t(textlen);
1740}
1741
1742/* Steal a reference to chars and store it in the decoded_char buffer;
1743 */
1744static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001745textiowrapper_set_decoded_chars(textio *self, PyObject *chars)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001746{
Serhiy Storchaka48842712016-04-06 09:45:48 +03001747 Py_XSETREF(self->decoded_chars, chars);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001748 self->decoded_chars_used = 0;
1749}
1750
1751static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001752textiowrapper_get_decoded_chars(textio *self, Py_ssize_t n)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001753{
1754 PyObject *chars;
1755 Py_ssize_t avail;
1756
1757 if (self->decoded_chars == NULL)
1758 return PyUnicode_FromStringAndSize(NULL, 0);
1759
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001760 /* decoded_chars is guaranteed to be "ready". */
1761 avail = (PyUnicode_GET_LENGTH(self->decoded_chars)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001762 - self->decoded_chars_used);
1763
1764 assert(avail >= 0);
1765
1766 if (n < 0 || n > avail)
1767 n = avail;
1768
1769 if (self->decoded_chars_used > 0 || n < avail) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001770 chars = PyUnicode_Substring(self->decoded_chars,
1771 self->decoded_chars_used,
1772 self->decoded_chars_used + n);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001773 if (chars == NULL)
1774 return NULL;
1775 }
1776 else {
1777 chars = self->decoded_chars;
1778 Py_INCREF(chars);
1779 }
1780
1781 self->decoded_chars_used += n;
1782 return chars;
1783}
1784
1785/* Read and decode the next chunk of data from the BufferedReader.
1786 */
1787static int
Antoine Pitroue5324562011-11-19 00:39:01 +01001788textiowrapper_read_chunk(textio *self, Py_ssize_t size_hint)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001789{
1790 PyObject *dec_buffer = NULL;
1791 PyObject *dec_flags = NULL;
1792 PyObject *input_chunk = NULL;
Antoine Pitroub8503892014-04-29 10:14:02 +02001793 Py_buffer input_chunk_buf;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001794 PyObject *decoded_chars, *chunk_size;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001795 Py_ssize_t nbytes, nchars;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001796 int eof;
1797
1798 /* The return value is True unless EOF was reached. The decoded string is
1799 * placed in self._decoded_chars (replacing its previous value). The
1800 * entire input chunk is sent to the decoder, though some of it may remain
1801 * buffered in the decoder, yet to be converted.
1802 */
1803
1804 if (self->decoder == NULL) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001805 _unsupported("not readable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001806 return -1;
1807 }
1808
1809 if (self->telling) {
1810 /* To prepare for tell(), we need to snapshot a point in the file
1811 * where the decoder's input buffer is empty.
1812 */
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001813 PyObject *state = _PyObject_CallMethodNoArgs(self->decoder,
1814 _PyIO_str_getstate);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001815 if (state == NULL)
1816 return -1;
1817 /* Given this, we know there was a valid snapshot point
1818 * len(dec_buffer) bytes ago with decoder state (b'', dec_flags).
1819 */
Oren Milmanba7d7362017-08-29 11:58:27 +03001820 if (!PyTuple_Check(state)) {
1821 PyErr_SetString(PyExc_TypeError,
1822 "illegal decoder state");
1823 Py_DECREF(state);
1824 return -1;
1825 }
1826 if (!PyArg_ParseTuple(state,
1827 "OO;illegal decoder state", &dec_buffer, &dec_flags))
1828 {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001829 Py_DECREF(state);
1830 return -1;
1831 }
Antoine Pitroub8503892014-04-29 10:14:02 +02001832
1833 if (!PyBytes_Check(dec_buffer)) {
1834 PyErr_Format(PyExc_TypeError,
Oren Milmanba7d7362017-08-29 11:58:27 +03001835 "illegal decoder state: the first item should be a "
1836 "bytes object, not '%.200s'",
Antoine Pitroub8503892014-04-29 10:14:02 +02001837 Py_TYPE(dec_buffer)->tp_name);
1838 Py_DECREF(state);
1839 return -1;
1840 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001841 Py_INCREF(dec_buffer);
1842 Py_INCREF(dec_flags);
1843 Py_DECREF(state);
1844 }
1845
1846 /* Read a chunk, decode it, and put the result in self._decoded_chars. */
Antoine Pitroue5324562011-11-19 00:39:01 +01001847 if (size_hint > 0) {
Victor Stinnerf8facac2011-11-22 02:30:47 +01001848 size_hint = (Py_ssize_t)(Py_MAX(self->b2cratio, 1.0) * size_hint);
Antoine Pitroue5324562011-11-19 00:39:01 +01001849 }
1850 chunk_size = PyLong_FromSsize_t(Py_MAX(self->chunk_size, size_hint));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001851 if (chunk_size == NULL)
1852 goto fail;
Antoine Pitroub8503892014-04-29 10:14:02 +02001853
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001854 input_chunk = PyObject_CallMethodObjArgs(self->buffer,
Antoine Pitroue96ec682011-07-23 21:46:35 +02001855 (self->has_read1 ? _PyIO_str_read1: _PyIO_str_read),
1856 chunk_size, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001857 Py_DECREF(chunk_size);
1858 if (input_chunk == NULL)
1859 goto fail;
Antoine Pitroub8503892014-04-29 10:14:02 +02001860
1861 if (PyObject_GetBuffer(input_chunk, &input_chunk_buf, 0) != 0) {
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001862 PyErr_Format(PyExc_TypeError,
Antoine Pitroub8503892014-04-29 10:14:02 +02001863 "underlying %s() should have returned a bytes-like object, "
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001864 "not '%.200s'", (self->has_read1 ? "read1": "read"),
1865 Py_TYPE(input_chunk)->tp_name);
1866 goto fail;
1867 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001868
Antoine Pitroub8503892014-04-29 10:14:02 +02001869 nbytes = input_chunk_buf.len;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001870 eof = (nbytes == 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001871
INADA Naoki507434f2017-12-21 09:59:53 +09001872 decoded_chars = _textiowrapper_decode(self->decoder, input_chunk, eof);
1873 PyBuffer_Release(&input_chunk_buf);
1874 if (decoded_chars == NULL)
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001875 goto fail;
INADA Naoki507434f2017-12-21 09:59:53 +09001876
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001877 textiowrapper_set_decoded_chars(self, decoded_chars);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001878 nchars = PyUnicode_GET_LENGTH(decoded_chars);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00001879 if (nchars > 0)
1880 self->b2cratio = (double) nbytes / nchars;
1881 else
1882 self->b2cratio = 0.0;
1883 if (nchars > 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001884 eof = 0;
1885
1886 if (self->telling) {
1887 /* At the snapshot point, len(dec_buffer) bytes before the read, the
1888 * next input to be decoded is dec_buffer + input_chunk.
1889 */
Antoine Pitroub8503892014-04-29 10:14:02 +02001890 PyObject *next_input = dec_buffer;
1891 PyBytes_Concat(&next_input, input_chunk);
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03001892 dec_buffer = NULL; /* Reference lost to PyBytes_Concat */
Antoine Pitroub8503892014-04-29 10:14:02 +02001893 if (next_input == NULL) {
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001894 goto fail;
1895 }
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03001896 PyObject *snapshot = Py_BuildValue("NN", dec_flags, next_input);
1897 if (snapshot == NULL) {
1898 dec_flags = NULL;
1899 goto fail;
1900 }
1901 Py_XSETREF(self->snapshot, snapshot);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001902 }
1903 Py_DECREF(input_chunk);
1904
1905 return (eof == 0);
1906
1907 fail:
1908 Py_XDECREF(dec_buffer);
1909 Py_XDECREF(dec_flags);
1910 Py_XDECREF(input_chunk);
1911 return -1;
1912}
1913
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001914/*[clinic input]
1915_io.TextIOWrapper.read
Serhiy Storchaka762bf402017-03-30 09:15:31 +03001916 size as n: Py_ssize_t(accept={int, NoneType}) = -1
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001917 /
1918[clinic start generated code]*/
1919
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001920static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03001921_io_TextIOWrapper_read_impl(textio *self, Py_ssize_t n)
Serhiy Storchaka762bf402017-03-30 09:15:31 +03001922/*[clinic end generated code: output=7e651ce6cc6a25a6 input=123eecbfe214aeb8]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001923{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001924 PyObject *result = NULL, *chunks = NULL;
1925
Benjamin Peterson10e76b62014-12-21 20:51:50 -06001926 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001927 CHECK_CLOSED(self);
1928
Antoine Pitrou0d739d72010-09-05 23:01:12 +00001929 if (self->decoder == NULL)
1930 return _unsupported("not readable");
Benjamin Petersona1b49012009-03-31 23:11:32 +00001931
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001932 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001933 return NULL;
1934
1935 if (n < 0) {
1936 /* Read everything */
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02001937 PyObject *bytes = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_read);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001938 PyObject *decoded;
1939 if (bytes == NULL)
1940 goto fail;
Victor Stinnerfd821132011-05-25 22:01:33 +02001941
1942 if (Py_TYPE(self->decoder) == &PyIncrementalNewlineDecoder_Type)
1943 decoded = _PyIncrementalNewlineDecoder_decode(self->decoder,
1944 bytes, 1);
1945 else
1946 decoded = PyObject_CallMethodObjArgs(
1947 self->decoder, _PyIO_str_decode, bytes, Py_True, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001948 Py_DECREF(bytes);
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02001949 if (check_decoded(decoded) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001950 goto fail;
1951
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001952 result = textiowrapper_get_decoded_chars(self, -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001953
1954 if (result == NULL) {
1955 Py_DECREF(decoded);
1956 return NULL;
1957 }
1958
1959 PyUnicode_AppendAndDel(&result, decoded);
1960 if (result == NULL)
1961 goto fail;
1962
Zackery Spytz23db9352018-06-29 04:14:58 -06001963 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001964 Py_CLEAR(self->snapshot);
1965 return result;
1966 }
1967 else {
1968 int res = 1;
1969 Py_ssize_t remaining = n;
1970
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00001971 result = textiowrapper_get_decoded_chars(self, n);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001972 if (result == NULL)
1973 goto fail;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001974 if (PyUnicode_READY(result) == -1)
1975 goto fail;
1976 remaining -= PyUnicode_GET_LENGTH(result);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001977
1978 /* Keep reading chunks until we have n characters to return */
1979 while (remaining > 0) {
Antoine Pitroue5324562011-11-19 00:39:01 +01001980 res = textiowrapper_read_chunk(self, remaining);
Gregory P. Smith51359922012-06-23 23:55:39 -07001981 if (res < 0) {
1982 /* NOTE: PyErr_SetFromErrno() calls PyErr_CheckSignals()
1983 when EINTR occurs so we needn't do it ourselves. */
1984 if (_PyIO_trap_eintr()) {
1985 continue;
1986 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001987 goto fail;
Gregory P. Smith51359922012-06-23 23:55:39 -07001988 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001989 if (res == 0) /* EOF */
1990 break;
1991 if (chunks == NULL) {
1992 chunks = PyList_New(0);
1993 if (chunks == NULL)
1994 goto fail;
1995 }
Antoine Pitroue5324562011-11-19 00:39:01 +01001996 if (PyUnicode_GET_LENGTH(result) > 0 &&
1997 PyList_Append(chunks, result) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001998 goto fail;
1999 Py_DECREF(result);
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002000 result = textiowrapper_get_decoded_chars(self, remaining);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002001 if (result == NULL)
2002 goto fail;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002003 remaining -= PyUnicode_GET_LENGTH(result);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002004 }
2005 if (chunks != NULL) {
2006 if (result != NULL && PyList_Append(chunks, result) < 0)
2007 goto fail;
Serhiy Storchaka48842712016-04-06 09:45:48 +03002008 Py_XSETREF(result, PyUnicode_Join(_PyIO_empty_str, chunks));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002009 if (result == NULL)
2010 goto fail;
2011 Py_CLEAR(chunks);
2012 }
2013 return result;
2014 }
2015 fail:
2016 Py_XDECREF(result);
2017 Py_XDECREF(chunks);
2018 return NULL;
2019}
2020
2021
Victor Stinnerf7b8cb62011-09-29 03:28:17 +02002022/* NOTE: `end` must point to the real end of the Py_UCS4 storage,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002023 that is to the NUL character. Otherwise the function will produce
2024 incorrect results. */
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002025static const char *
2026find_control_char(int kind, const char *s, const char *end, Py_UCS4 ch)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002027{
Antoine Pitrouc28e2e52011-11-13 03:53:42 +01002028 if (kind == PyUnicode_1BYTE_KIND) {
2029 assert(ch < 256);
2030 return (char *) memchr((void *) s, (char) ch, end - s);
2031 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002032 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002033 while (PyUnicode_READ(kind, s, 0) > ch)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002034 s += kind;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002035 if (PyUnicode_READ(kind, s, 0) == ch)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002036 return s;
2037 if (s == end)
2038 return NULL;
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002039 s += kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002040 }
2041}
2042
2043Py_ssize_t
2044_PyIO_find_line_ending(
2045 int translated, int universal, PyObject *readnl,
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002046 int kind, const char *start, const char *end, Py_ssize_t *consumed)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002047{
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002048 Py_ssize_t len = ((char*)end - (char*)start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002049
2050 if (translated) {
2051 /* Newlines are already translated, only search for \n */
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002052 const char *pos = find_control_char(kind, start, end, '\n');
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002053 if (pos != NULL)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002054 return (pos - start)/kind + 1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002055 else {
2056 *consumed = len;
2057 return -1;
2058 }
2059 }
2060 else if (universal) {
2061 /* Universal newline search. Find any of \r, \r\n, \n
2062 * The decoder ensures that \r\n are not split in two pieces
2063 */
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002064 const char *s = start;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002065 for (;;) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002066 Py_UCS4 ch;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002067 /* Fast path for non-control chars. The loop always ends
Victor Stinnerf7b8cb62011-09-29 03:28:17 +02002068 since the Unicode string is NUL-terminated. */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002069 while (PyUnicode_READ(kind, s, 0) > '\r')
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002070 s += kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002071 if (s >= end) {
2072 *consumed = len;
2073 return -1;
2074 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002075 ch = PyUnicode_READ(kind, s, 0);
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002076 s += kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002077 if (ch == '\n')
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002078 return (s - start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002079 if (ch == '\r') {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002080 if (PyUnicode_READ(kind, s, 0) == '\n')
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002081 return (s - start)/kind + 1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002082 else
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002083 return (s - start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002084 }
2085 }
2086 }
2087 else {
2088 /* Non-universal mode. */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002089 Py_ssize_t readnl_len = PyUnicode_GET_LENGTH(readnl);
Victor Stinner706768c2014-08-16 01:03:39 +02002090 Py_UCS1 *nl = PyUnicode_1BYTE_DATA(readnl);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002091 /* Assume that readnl is an ASCII character. */
2092 assert(PyUnicode_KIND(readnl) == PyUnicode_1BYTE_KIND);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002093 if (readnl_len == 1) {
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002094 const char *pos = find_control_char(kind, start, end, nl[0]);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002095 if (pos != NULL)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002096 return (pos - start)/kind + 1;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002097 *consumed = len;
2098 return -1;
2099 }
2100 else {
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002101 const char *s = start;
2102 const char *e = end - (readnl_len - 1)*kind;
2103 const char *pos;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002104 if (e < s)
2105 e = s;
2106 while (s < e) {
2107 Py_ssize_t i;
Serhiy Storchakaef1585e2015-12-25 20:01:53 +02002108 const char *pos = find_control_char(kind, s, end, nl[0]);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002109 if (pos == NULL || pos >= e)
2110 break;
2111 for (i = 1; i < readnl_len; i++) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002112 if (PyUnicode_READ(kind, pos, i) != nl[i])
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002113 break;
2114 }
2115 if (i == readnl_len)
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002116 return (pos - start)/kind + readnl_len;
2117 s = pos + kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002118 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002119 pos = find_control_char(kind, e, end, nl[0]);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002120 if (pos == NULL)
2121 *consumed = len;
2122 else
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002123 *consumed = (pos - start)/kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002124 return -1;
2125 }
2126 }
2127}
2128
2129static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002130_textiowrapper_readline(textio *self, Py_ssize_t limit)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002131{
2132 PyObject *line = NULL, *chunks = NULL, *remaining = NULL;
2133 Py_ssize_t start, endpos, chunked, offset_to_buffer;
2134 int res;
2135
2136 CHECK_CLOSED(self);
2137
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002138 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002139 return NULL;
2140
2141 chunked = 0;
2142
2143 while (1) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002144 char *ptr;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002145 Py_ssize_t line_len;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002146 int kind;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002147 Py_ssize_t consumed = 0;
2148
2149 /* First, get some data if necessary */
2150 res = 1;
2151 while (!self->decoded_chars ||
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002152 !PyUnicode_GET_LENGTH(self->decoded_chars)) {
Antoine Pitroue5324562011-11-19 00:39:01 +01002153 res = textiowrapper_read_chunk(self, 0);
Gregory P. Smith51359922012-06-23 23:55:39 -07002154 if (res < 0) {
2155 /* NOTE: PyErr_SetFromErrno() calls PyErr_CheckSignals()
2156 when EINTR occurs so we needn't do it ourselves. */
2157 if (_PyIO_trap_eintr()) {
2158 continue;
2159 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002160 goto error;
Gregory P. Smith51359922012-06-23 23:55:39 -07002161 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002162 if (res == 0)
2163 break;
2164 }
2165 if (res == 0) {
2166 /* end of file */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002167 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002168 Py_CLEAR(self->snapshot);
2169 start = endpos = offset_to_buffer = 0;
2170 break;
2171 }
2172
2173 if (remaining == NULL) {
2174 line = self->decoded_chars;
2175 start = self->decoded_chars_used;
2176 offset_to_buffer = 0;
2177 Py_INCREF(line);
2178 }
2179 else {
2180 assert(self->decoded_chars_used == 0);
2181 line = PyUnicode_Concat(remaining, self->decoded_chars);
2182 start = 0;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002183 offset_to_buffer = PyUnicode_GET_LENGTH(remaining);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002184 Py_CLEAR(remaining);
2185 if (line == NULL)
2186 goto error;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002187 if (PyUnicode_READY(line) == -1)
2188 goto error;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002189 }
2190
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002191 ptr = PyUnicode_DATA(line);
2192 line_len = PyUnicode_GET_LENGTH(line);
2193 kind = PyUnicode_KIND(line);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002194
2195 endpos = _PyIO_find_line_ending(
2196 self->readtranslate, self->readuniversal, self->readnl,
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002197 kind,
Martin v. Löwisc47adb02011-10-07 20:55:35 +02002198 ptr + kind * start,
2199 ptr + kind * line_len,
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002200 &consumed);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002201 if (endpos >= 0) {
2202 endpos += start;
2203 if (limit >= 0 && (endpos - start) + chunked >= limit)
2204 endpos = start + limit - chunked;
2205 break;
2206 }
2207
2208 /* We can put aside up to `endpos` */
2209 endpos = consumed + start;
2210 if (limit >= 0 && (endpos - start) + chunked >= limit) {
2211 /* Didn't find line ending, but reached length limit */
2212 endpos = start + limit - chunked;
2213 break;
2214 }
2215
2216 if (endpos > start) {
2217 /* No line ending seen yet - put aside current data */
2218 PyObject *s;
2219 if (chunks == NULL) {
2220 chunks = PyList_New(0);
2221 if (chunks == NULL)
2222 goto error;
2223 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002224 s = PyUnicode_Substring(line, start, endpos);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002225 if (s == NULL)
2226 goto error;
2227 if (PyList_Append(chunks, s) < 0) {
2228 Py_DECREF(s);
2229 goto error;
2230 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002231 chunked += PyUnicode_GET_LENGTH(s);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002232 Py_DECREF(s);
2233 }
2234 /* There may be some remaining bytes we'll have to prepend to the
2235 next chunk of data */
2236 if (endpos < line_len) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002237 remaining = PyUnicode_Substring(line, endpos, line_len);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002238 if (remaining == NULL)
2239 goto error;
2240 }
2241 Py_CLEAR(line);
2242 /* We have consumed the buffer */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002243 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002244 }
2245
2246 if (line != NULL) {
2247 /* Our line ends in the current buffer */
2248 self->decoded_chars_used = endpos - offset_to_buffer;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002249 if (start > 0 || endpos < PyUnicode_GET_LENGTH(line)) {
2250 PyObject *s = PyUnicode_Substring(line, start, endpos);
2251 Py_CLEAR(line);
2252 if (s == NULL)
2253 goto error;
2254 line = s;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002255 }
2256 }
2257 if (remaining != NULL) {
2258 if (chunks == NULL) {
2259 chunks = PyList_New(0);
2260 if (chunks == NULL)
2261 goto error;
2262 }
2263 if (PyList_Append(chunks, remaining) < 0)
2264 goto error;
2265 Py_CLEAR(remaining);
2266 }
2267 if (chunks != NULL) {
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002268 if (line != NULL) {
2269 if (PyList_Append(chunks, line) < 0)
2270 goto error;
2271 Py_DECREF(line);
2272 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002273 line = PyUnicode_Join(_PyIO_empty_str, chunks);
2274 if (line == NULL)
2275 goto error;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002276 Py_CLEAR(chunks);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002277 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002278 if (line == NULL) {
2279 Py_INCREF(_PyIO_empty_str);
2280 line = _PyIO_empty_str;
2281 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002282
2283 return line;
2284
2285 error:
2286 Py_XDECREF(chunks);
2287 Py_XDECREF(remaining);
2288 Py_XDECREF(line);
2289 return NULL;
2290}
2291
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002292/*[clinic input]
2293_io.TextIOWrapper.readline
2294 size: Py_ssize_t = -1
2295 /
2296[clinic start generated code]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002297
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002298static PyObject *
2299_io_TextIOWrapper_readline_impl(textio *self, Py_ssize_t size)
2300/*[clinic end generated code: output=344afa98804e8b25 input=56c7172483b36db6]*/
2301{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002302 CHECK_ATTACHED(self);
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002303 return _textiowrapper_readline(self, size);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002304}
2305
2306/* Seek and Tell */
2307
2308typedef struct {
2309 Py_off_t start_pos;
2310 int dec_flags;
2311 int bytes_to_feed;
2312 int chars_to_skip;
2313 char need_eof;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002314} cookie_type;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002315
2316/*
2317 To speed up cookie packing/unpacking, we store the fields in a temporary
2318 string and call _PyLong_FromByteArray() or _PyLong_AsByteArray (resp.).
2319 The following macros define at which offsets in the intermediary byte
2320 string the various CookieStruct fields will be stored.
2321 */
2322
2323#define COOKIE_BUF_LEN (sizeof(Py_off_t) + 3 * sizeof(int) + sizeof(char))
2324
Christian Heimes743e0cd2012-10-17 23:52:17 +02002325#if PY_BIG_ENDIAN
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002326/* We want the least significant byte of start_pos to also be the least
2327 significant byte of the cookie, which means that in big-endian mode we
2328 must copy the fields in reverse order. */
2329
2330# define OFF_START_POS (sizeof(char) + 3 * sizeof(int))
2331# define OFF_DEC_FLAGS (sizeof(char) + 2 * sizeof(int))
2332# define OFF_BYTES_TO_FEED (sizeof(char) + sizeof(int))
2333# define OFF_CHARS_TO_SKIP (sizeof(char))
2334# define OFF_NEED_EOF 0
2335
2336#else
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002337/* Little-endian mode: the least significant byte of start_pos will
2338 naturally end up the least significant byte of the cookie. */
2339
2340# define OFF_START_POS 0
2341# define OFF_DEC_FLAGS (sizeof(Py_off_t))
2342# define OFF_BYTES_TO_FEED (sizeof(Py_off_t) + sizeof(int))
2343# define OFF_CHARS_TO_SKIP (sizeof(Py_off_t) + 2 * sizeof(int))
2344# define OFF_NEED_EOF (sizeof(Py_off_t) + 3 * sizeof(int))
2345
2346#endif
2347
2348static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002349textiowrapper_parse_cookie(cookie_type *cookie, PyObject *cookieObj)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002350{
2351 unsigned char buffer[COOKIE_BUF_LEN];
2352 PyLongObject *cookieLong = (PyLongObject *)PyNumber_Long(cookieObj);
2353 if (cookieLong == NULL)
2354 return -1;
2355
2356 if (_PyLong_AsByteArray(cookieLong, buffer, sizeof(buffer),
Christian Heimes743e0cd2012-10-17 23:52:17 +02002357 PY_LITTLE_ENDIAN, 0) < 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002358 Py_DECREF(cookieLong);
2359 return -1;
2360 }
2361 Py_DECREF(cookieLong);
2362
Antoine Pitrou2db74c22009-03-06 21:49:02 +00002363 memcpy(&cookie->start_pos, buffer + OFF_START_POS, sizeof(cookie->start_pos));
2364 memcpy(&cookie->dec_flags, buffer + OFF_DEC_FLAGS, sizeof(cookie->dec_flags));
2365 memcpy(&cookie->bytes_to_feed, buffer + OFF_BYTES_TO_FEED, sizeof(cookie->bytes_to_feed));
2366 memcpy(&cookie->chars_to_skip, buffer + OFF_CHARS_TO_SKIP, sizeof(cookie->chars_to_skip));
2367 memcpy(&cookie->need_eof, buffer + OFF_NEED_EOF, sizeof(cookie->need_eof));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002368
2369 return 0;
2370}
2371
2372static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002373textiowrapper_build_cookie(cookie_type *cookie)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002374{
2375 unsigned char buffer[COOKIE_BUF_LEN];
2376
Antoine Pitrou2db74c22009-03-06 21:49:02 +00002377 memcpy(buffer + OFF_START_POS, &cookie->start_pos, sizeof(cookie->start_pos));
2378 memcpy(buffer + OFF_DEC_FLAGS, &cookie->dec_flags, sizeof(cookie->dec_flags));
2379 memcpy(buffer + OFF_BYTES_TO_FEED, &cookie->bytes_to_feed, sizeof(cookie->bytes_to_feed));
2380 memcpy(buffer + OFF_CHARS_TO_SKIP, &cookie->chars_to_skip, sizeof(cookie->chars_to_skip));
2381 memcpy(buffer + OFF_NEED_EOF, &cookie->need_eof, sizeof(cookie->need_eof));
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002382
Christian Heimes743e0cd2012-10-17 23:52:17 +02002383 return _PyLong_FromByteArray(buffer, sizeof(buffer),
2384 PY_LITTLE_ENDIAN, 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002385}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002386
2387static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002388_textiowrapper_decoder_setstate(textio *self, cookie_type *cookie)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002389{
2390 PyObject *res;
2391 /* When seeking to the start of the stream, we call decoder.reset()
2392 rather than decoder.getstate().
2393 This is for a few decoders such as utf-16 for which the state value
2394 at start is not (b"", 0) but e.g. (b"", 2) (meaning, in the case of
2395 utf-16, that we are expecting a BOM).
2396 */
2397 if (cookie->start_pos == 0 && cookie->dec_flags == 0)
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002398 res = _PyObject_CallMethodNoArgs(self->decoder, _PyIO_str_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002399 else
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002400 res = _PyObject_CallMethodId(self->decoder, &PyId_setstate,
2401 "((yi))", "", cookie->dec_flags);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002402 if (res == NULL)
2403 return -1;
2404 Py_DECREF(res);
2405 return 0;
2406}
2407
Antoine Pitroue4501852009-05-14 18:55:55 +00002408static int
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002409_textiowrapper_encoder_reset(textio *self, int start_of_stream)
Antoine Pitroue4501852009-05-14 18:55:55 +00002410{
2411 PyObject *res;
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002412 if (start_of_stream) {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002413 res = _PyObject_CallMethodNoArgs(self->encoder, _PyIO_str_reset);
Antoine Pitroue4501852009-05-14 18:55:55 +00002414 self->encoding_start_of_stream = 1;
2415 }
2416 else {
2417 res = PyObject_CallMethodObjArgs(self->encoder, _PyIO_str_setstate,
Serhiy Storchakaba85d692017-03-30 09:09:41 +03002418 _PyLong_Zero, NULL);
Antoine Pitroue4501852009-05-14 18:55:55 +00002419 self->encoding_start_of_stream = 0;
2420 }
2421 if (res == NULL)
2422 return -1;
2423 Py_DECREF(res);
2424 return 0;
2425}
2426
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002427static int
2428_textiowrapper_encoder_setstate(textio *self, cookie_type *cookie)
2429{
2430 /* Same as _textiowrapper_decoder_setstate() above. */
2431 return _textiowrapper_encoder_reset(
2432 self, cookie->start_pos == 0 && cookie->dec_flags == 0);
2433}
2434
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002435/*[clinic input]
2436_io.TextIOWrapper.seek
2437 cookie as cookieObj: object
2438 whence: int = 0
2439 /
2440[clinic start generated code]*/
2441
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002442static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002443_io_TextIOWrapper_seek_impl(textio *self, PyObject *cookieObj, int whence)
2444/*[clinic end generated code: output=0a15679764e2d04d input=0458abeb3d7842be]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002445{
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002446 PyObject *posobj;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002447 cookie_type cookie;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002448 PyObject *res;
2449 int cmp;
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002450 PyObject *snapshot;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002451
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002452 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002453 CHECK_CLOSED(self);
2454
2455 Py_INCREF(cookieObj);
2456
2457 if (!self->seekable) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002458 _unsupported("underlying stream is not seekable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002459 goto fail;
2460 }
2461
ngie-eign848037c2019-03-02 23:28:26 -08002462 switch (whence) {
2463 case SEEK_CUR:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002464 /* seek relative to current position */
Serhiy Storchakaba85d692017-03-30 09:09:41 +03002465 cmp = PyObject_RichCompareBool(cookieObj, _PyLong_Zero, Py_EQ);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002466 if (cmp < 0)
2467 goto fail;
2468
2469 if (cmp == 0) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002470 _unsupported("can't do nonzero cur-relative seeks");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002471 goto fail;
2472 }
2473
2474 /* Seeking to the current position should attempt to
2475 * sync the underlying buffer with the current position.
2476 */
2477 Py_DECREF(cookieObj);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002478 cookieObj = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_tell);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002479 if (cookieObj == NULL)
2480 goto fail;
Inada Naoki8c17d922019-03-04 01:22:39 +09002481 break;
2482
ngie-eign848037c2019-03-02 23:28:26 -08002483 case SEEK_END:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002484 /* seek relative to end of file */
Serhiy Storchakaba85d692017-03-30 09:09:41 +03002485 cmp = PyObject_RichCompareBool(cookieObj, _PyLong_Zero, Py_EQ);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002486 if (cmp < 0)
2487 goto fail;
2488
2489 if (cmp == 0) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002490 _unsupported("can't do nonzero end-relative seeks");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002491 goto fail;
2492 }
2493
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002494 res = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002495 if (res == NULL)
2496 goto fail;
2497 Py_DECREF(res);
2498
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002499 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002500 Py_CLEAR(self->snapshot);
2501 if (self->decoder) {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002502 res = _PyObject_CallMethodIdNoArgs(self->decoder, &PyId_reset);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002503 if (res == NULL)
2504 goto fail;
2505 Py_DECREF(res);
2506 }
2507
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002508 res = _PyObject_CallMethodId(self->buffer, &PyId_seek, "ii", 0, 2);
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002509 Py_CLEAR(cookieObj);
2510 if (res == NULL)
2511 goto fail;
2512 if (self->encoder) {
2513 /* If seek() == 0, we are at the start of stream, otherwise not */
Serhiy Storchakaba85d692017-03-30 09:09:41 +03002514 cmp = PyObject_RichCompareBool(res, _PyLong_Zero, Py_EQ);
Antoine Pitrou85e3ee72015-04-13 20:01:21 +02002515 if (cmp < 0 || _textiowrapper_encoder_reset(self, cmp)) {
2516 Py_DECREF(res);
2517 goto fail;
2518 }
2519 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002520 return res;
Inada Naoki8c17d922019-03-04 01:22:39 +09002521
ngie-eign848037c2019-03-02 23:28:26 -08002522 case SEEK_SET:
2523 break;
Inada Naoki8c17d922019-03-04 01:22:39 +09002524
ngie-eign848037c2019-03-02 23:28:26 -08002525 default:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002526 PyErr_Format(PyExc_ValueError,
ngie-eign848037c2019-03-02 23:28:26 -08002527 "invalid whence (%d, should be %d, %d or %d)", whence,
2528 SEEK_SET, SEEK_CUR, SEEK_END);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002529 goto fail;
2530 }
2531
Serhiy Storchakaba85d692017-03-30 09:09:41 +03002532 cmp = PyObject_RichCompareBool(cookieObj, _PyLong_Zero, Py_LT);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002533 if (cmp < 0)
2534 goto fail;
2535
2536 if (cmp == 1) {
2537 PyErr_Format(PyExc_ValueError,
2538 "negative seek position %R", cookieObj);
2539 goto fail;
2540 }
2541
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002542 res = _PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002543 if (res == NULL)
2544 goto fail;
2545 Py_DECREF(res);
2546
2547 /* The strategy of seek() is to go back to the safe start point
2548 * and replay the effect of read(chars_to_skip) from there.
2549 */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002550 if (textiowrapper_parse_cookie(&cookie, cookieObj) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002551 goto fail;
2552
2553 /* Seek back to the safe start point. */
2554 posobj = PyLong_FromOff_t(cookie.start_pos);
2555 if (posobj == NULL)
2556 goto fail;
2557 res = PyObject_CallMethodObjArgs(self->buffer,
2558 _PyIO_str_seek, posobj, NULL);
2559 Py_DECREF(posobj);
2560 if (res == NULL)
2561 goto fail;
2562 Py_DECREF(res);
2563
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002564 textiowrapper_set_decoded_chars(self, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002565 Py_CLEAR(self->snapshot);
2566
2567 /* Restore the decoder to its state from the safe start point. */
2568 if (self->decoder) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002569 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002570 goto fail;
2571 }
2572
2573 if (cookie.chars_to_skip) {
2574 /* Just like _read_chunk, feed the decoder and save a snapshot. */
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002575 PyObject *input_chunk = _PyObject_CallMethodId(
2576 self->buffer, &PyId_read, "i", cookie.bytes_to_feed);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002577 PyObject *decoded;
2578
2579 if (input_chunk == NULL)
2580 goto fail;
2581
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02002582 if (!PyBytes_Check(input_chunk)) {
2583 PyErr_Format(PyExc_TypeError,
2584 "underlying read() should have returned a bytes "
2585 "object, not '%.200s'",
2586 Py_TYPE(input_chunk)->tp_name);
2587 Py_DECREF(input_chunk);
2588 goto fail;
2589 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002590
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002591 snapshot = Py_BuildValue("iN", cookie.dec_flags, input_chunk);
2592 if (snapshot == NULL) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002593 goto fail;
2594 }
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002595 Py_XSETREF(self->snapshot, snapshot);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002596
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002597 decoded = _PyObject_CallMethodId(self->decoder, &PyId_decode,
2598 "Oi", input_chunk, (int)cookie.need_eof);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002599
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02002600 if (check_decoded(decoded) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002601 goto fail;
2602
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002603 textiowrapper_set_decoded_chars(self, decoded);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002604
2605 /* Skip chars_to_skip of the decoded characters. */
Victor Stinner9e30aa52011-11-21 02:49:52 +01002606 if (PyUnicode_GetLength(self->decoded_chars) < cookie.chars_to_skip) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03002607 PyErr_SetString(PyExc_OSError, "can't restore logical file position");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002608 goto fail;
2609 }
2610 self->decoded_chars_used = cookie.chars_to_skip;
2611 }
2612 else {
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002613 snapshot = Py_BuildValue("iy", cookie.dec_flags, "");
2614 if (snapshot == NULL)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002615 goto fail;
Serhiy Storchakafdb5a502018-06-30 20:57:50 +03002616 Py_XSETREF(self->snapshot, snapshot);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002617 }
2618
Antoine Pitroue4501852009-05-14 18:55:55 +00002619 /* Finally, reset the encoder (merely useful for proper BOM handling) */
2620 if (self->encoder) {
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002621 if (_textiowrapper_encoder_setstate(self, &cookie) < 0)
Antoine Pitroue4501852009-05-14 18:55:55 +00002622 goto fail;
2623 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002624 return cookieObj;
2625 fail:
2626 Py_XDECREF(cookieObj);
2627 return NULL;
2628
2629}
2630
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002631/*[clinic input]
2632_io.TextIOWrapper.tell
2633[clinic start generated code]*/
2634
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002635static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002636_io_TextIOWrapper_tell_impl(textio *self)
2637/*[clinic end generated code: output=4f168c08bf34ad5f input=9a2caf88c24f9ddf]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002638{
2639 PyObject *res;
2640 PyObject *posobj = NULL;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002641 cookie_type cookie = {0,0,0,0,0};
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002642 PyObject *next_input;
2643 Py_ssize_t chars_to_skip, chars_decoded;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002644 Py_ssize_t skip_bytes, skip_back;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002645 PyObject *saved_state = NULL;
2646 char *input, *input_end;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002647 Py_ssize_t dec_buffer_len;
2648 int dec_flags;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002649
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002650 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002651 CHECK_CLOSED(self);
2652
2653 if (!self->seekable) {
Antoine Pitrou0d739d72010-09-05 23:01:12 +00002654 _unsupported("underlying stream is not seekable");
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002655 goto fail;
2656 }
2657 if (!self->telling) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03002658 PyErr_SetString(PyExc_OSError,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002659 "telling position disabled by next() call");
2660 goto fail;
2661 }
2662
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002663 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002664 return NULL;
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002665 res = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002666 if (res == NULL)
2667 goto fail;
2668 Py_DECREF(res);
2669
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002670 posobj = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_tell);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002671 if (posobj == NULL)
2672 goto fail;
2673
2674 if (self->decoder == NULL || self->snapshot == NULL) {
Victor Stinner9e30aa52011-11-21 02:49:52 +01002675 assert (self->decoded_chars == NULL || PyUnicode_GetLength(self->decoded_chars) == 0);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002676 return posobj;
2677 }
2678
2679#if defined(HAVE_LARGEFILE_SUPPORT)
2680 cookie.start_pos = PyLong_AsLongLong(posobj);
2681#else
2682 cookie.start_pos = PyLong_AsLong(posobj);
2683#endif
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002684 Py_DECREF(posobj);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002685 if (PyErr_Occurred())
2686 goto fail;
2687
2688 /* Skip backward to the snapshot point (see _read_chunk). */
Oren Milman13614e32017-08-24 19:51:24 +03002689 assert(PyTuple_Check(self->snapshot));
Serhiy Storchakabb72c472015-04-19 20:38:19 +03002690 if (!PyArg_ParseTuple(self->snapshot, "iO", &cookie.dec_flags, &next_input))
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002691 goto fail;
2692
2693 assert (PyBytes_Check(next_input));
2694
2695 cookie.start_pos -= PyBytes_GET_SIZE(next_input);
2696
2697 /* How many decoded characters have been used up since the snapshot? */
2698 if (self->decoded_chars_used == 0) {
2699 /* We haven't moved from the snapshot point. */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002700 return textiowrapper_build_cookie(&cookie);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002701 }
2702
2703 chars_to_skip = self->decoded_chars_used;
2704
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002705 /* Decoder state will be restored at the end */
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002706 saved_state = _PyObject_CallMethodNoArgs(self->decoder,
2707 _PyIO_str_getstate);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002708 if (saved_state == NULL)
2709 goto fail;
2710
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002711#define DECODER_GETSTATE() do { \
Serhiy Storchaka008d88b2015-05-06 09:53:07 +03002712 PyObject *dec_buffer; \
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002713 PyObject *_state = _PyObject_CallMethodNoArgs(self->decoder, \
2714 _PyIO_str_getstate); \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002715 if (_state == NULL) \
2716 goto fail; \
Oren Milman13614e32017-08-24 19:51:24 +03002717 if (!PyTuple_Check(_state)) { \
2718 PyErr_SetString(PyExc_TypeError, \
2719 "illegal decoder state"); \
2720 Py_DECREF(_state); \
2721 goto fail; \
2722 } \
2723 if (!PyArg_ParseTuple(_state, "Oi;illegal decoder state", \
2724 &dec_buffer, &dec_flags)) \
2725 { \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002726 Py_DECREF(_state); \
2727 goto fail; \
2728 } \
Serhiy Storchaka008d88b2015-05-06 09:53:07 +03002729 if (!PyBytes_Check(dec_buffer)) { \
2730 PyErr_Format(PyExc_TypeError, \
Oren Milmanba7d7362017-08-29 11:58:27 +03002731 "illegal decoder state: the first item should be a " \
2732 "bytes object, not '%.200s'", \
Serhiy Storchaka008d88b2015-05-06 09:53:07 +03002733 Py_TYPE(dec_buffer)->tp_name); \
2734 Py_DECREF(_state); \
2735 goto fail; \
2736 } \
2737 dec_buffer_len = PyBytes_GET_SIZE(dec_buffer); \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002738 Py_DECREF(_state); \
2739 } while (0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002740
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002741#define DECODER_DECODE(start, len, res) do { \
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002742 PyObject *_decoded = _PyObject_CallMethodId( \
2743 self->decoder, &PyId_decode, "y#", start, len); \
Serhiy Storchakad03ce4a2013-02-03 17:07:32 +02002744 if (check_decoded(_decoded) < 0) \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002745 goto fail; \
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002746 res = PyUnicode_GET_LENGTH(_decoded); \
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002747 Py_DECREF(_decoded); \
2748 } while (0)
2749
2750 /* Fast search for an acceptable start point, close to our
2751 current pos */
2752 skip_bytes = (Py_ssize_t) (self->b2cratio * chars_to_skip);
2753 skip_back = 1;
2754 assert(skip_back <= PyBytes_GET_SIZE(next_input));
2755 input = PyBytes_AS_STRING(next_input);
2756 while (skip_bytes > 0) {
2757 /* Decode up to temptative start point */
2758 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
2759 goto fail;
2760 DECODER_DECODE(input, skip_bytes, chars_decoded);
2761 if (chars_decoded <= chars_to_skip) {
2762 DECODER_GETSTATE();
2763 if (dec_buffer_len == 0) {
2764 /* Before pos and no bytes buffered in decoder => OK */
2765 cookie.dec_flags = dec_flags;
2766 chars_to_skip -= chars_decoded;
2767 break;
2768 }
2769 /* Skip back by buffered amount and reset heuristic */
2770 skip_bytes -= dec_buffer_len;
2771 skip_back = 1;
2772 }
2773 else {
2774 /* We're too far ahead, skip back a bit */
2775 skip_bytes -= skip_back;
2776 skip_back *= 2;
2777 }
2778 }
2779 if (skip_bytes <= 0) {
2780 skip_bytes = 0;
2781 if (_textiowrapper_decoder_setstate(self, &cookie) < 0)
2782 goto fail;
2783 }
2784
2785 /* Note our initial start point. */
2786 cookie.start_pos += skip_bytes;
Victor Stinner9a282972013-06-24 23:01:33 +02002787 cookie.chars_to_skip = Py_SAFE_DOWNCAST(chars_to_skip, Py_ssize_t, int);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002788 if (chars_to_skip == 0)
2789 goto finally;
2790
2791 /* We should be close to the desired position. Now feed the decoder one
2792 * byte at a time until we reach the `chars_to_skip` target.
2793 * As we go, note the nearest "safe start point" before the current
2794 * location (a point where the decoder has nothing buffered, so seek()
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002795 * can safely start from there and advance to this location).
2796 */
2797 chars_decoded = 0;
2798 input = PyBytes_AS_STRING(next_input);
2799 input_end = input + PyBytes_GET_SIZE(next_input);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002800 input += skip_bytes;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002801 while (input < input_end) {
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002802 Py_ssize_t n;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002803
Serhiy Storchakaec67d182013-08-20 20:04:47 +03002804 DECODER_DECODE(input, (Py_ssize_t)1, n);
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002805 /* We got n chars for 1 byte */
2806 chars_decoded += n;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002807 cookie.bytes_to_feed += 1;
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002808 DECODER_GETSTATE();
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002809
2810 if (dec_buffer_len == 0 && chars_decoded <= chars_to_skip) {
2811 /* Decoder buffer is empty, so this is a safe start point. */
2812 cookie.start_pos += cookie.bytes_to_feed;
2813 chars_to_skip -= chars_decoded;
2814 cookie.dec_flags = dec_flags;
2815 cookie.bytes_to_feed = 0;
2816 chars_decoded = 0;
2817 }
2818 if (chars_decoded >= chars_to_skip)
2819 break;
2820 input++;
2821 }
2822 if (input == input_end) {
2823 /* We didn't get enough decoded data; signal EOF to get more. */
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02002824 PyObject *decoded = _PyObject_CallMethodId(
2825 self->decoder, &PyId_decode, "yi", "", /* final = */ 1);
Serhiy Storchaka94dc6732013-02-03 17:03:31 +02002826 if (check_decoded(decoded) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002827 goto fail;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002828 chars_decoded += PyUnicode_GET_LENGTH(decoded);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002829 Py_DECREF(decoded);
2830 cookie.need_eof = 1;
2831
2832 if (chars_decoded < chars_to_skip) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03002833 PyErr_SetString(PyExc_OSError,
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002834 "can't reconstruct logical file position");
2835 goto fail;
2836 }
2837 }
2838
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002839finally:
Victor Stinner7e425412016-12-09 00:36:19 +01002840 res = _PyObject_CallMethodIdObjArgs(self->decoder, &PyId_setstate, saved_state, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002841 Py_DECREF(saved_state);
2842 if (res == NULL)
2843 return NULL;
2844 Py_DECREF(res);
2845
2846 /* The returned cookie corresponds to the last safe start point. */
2847 cookie.chars_to_skip = Py_SAFE_DOWNCAST(chars_to_skip, Py_ssize_t, int);
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002848 return textiowrapper_build_cookie(&cookie);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002849
Antoine Pitrou211b81d2011-02-25 20:27:33 +00002850fail:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002851 if (saved_state) {
2852 PyObject *type, *value, *traceback;
2853 PyErr_Fetch(&type, &value, &traceback);
Victor Stinner7e425412016-12-09 00:36:19 +01002854 res = _PyObject_CallMethodIdObjArgs(self->decoder, &PyId_setstate, saved_state, NULL);
Serhiy Storchaka04d09eb2015-03-30 09:58:41 +03002855 _PyErr_ChainExceptions(type, value, traceback);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002856 Py_DECREF(saved_state);
Serhiy Storchaka04d09eb2015-03-30 09:58:41 +03002857 Py_XDECREF(res);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002858 }
2859 return NULL;
2860}
2861
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002862/*[clinic input]
2863_io.TextIOWrapper.truncate
2864 pos: object = None
2865 /
2866[clinic start generated code]*/
2867
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002868static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002869_io_TextIOWrapper_truncate_impl(textio *self, PyObject *pos)
2870/*[clinic end generated code: output=90ec2afb9bb7745f input=56ec8baa65aea377]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002871{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002872 PyObject *res;
2873
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002874 CHECK_ATTACHED(self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002875
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002876 res = _PyObject_CallMethodNoArgs((PyObject *)self, _PyIO_str_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002877 if (res == NULL)
2878 return NULL;
2879 Py_DECREF(res);
2880
Antoine Pitrou905a2ff2010-01-31 22:47:27 +00002881 return PyObject_CallMethodObjArgs(self->buffer, _PyIO_str_truncate, pos, NULL);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002882}
2883
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002884static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00002885textiowrapper_repr(textio *self)
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002886{
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002887 PyObject *nameobj, *modeobj, *res, *s;
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002888 int status;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002889
2890 CHECK_INITIALIZED(self);
2891
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002892 res = PyUnicode_FromString("<_io.TextIOWrapper");
2893 if (res == NULL)
2894 return NULL;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002895
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002896 status = Py_ReprEnter((PyObject *)self);
2897 if (status != 0) {
2898 if (status > 0) {
2899 PyErr_Format(PyExc_RuntimeError,
2900 "reentrant call inside %s.__repr__",
2901 Py_TYPE(self)->tp_name);
2902 }
2903 goto error;
2904 }
Martin v. Löwis767046a2011-10-14 15:35:36 +02002905 nameobj = _PyObject_GetAttrId((PyObject *) self, &PyId_name);
Antoine Pitrou716c4442009-05-23 19:04:03 +00002906 if (nameobj == NULL) {
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002907 if (PyErr_ExceptionMatches(PyExc_Exception))
Antoine Pitrou716c4442009-05-23 19:04:03 +00002908 PyErr_Clear();
2909 else
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002910 goto error;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002911 }
2912 else {
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002913 s = PyUnicode_FromFormat(" name=%R", nameobj);
Antoine Pitrou716c4442009-05-23 19:04:03 +00002914 Py_DECREF(nameobj);
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002915 if (s == NULL)
2916 goto error;
2917 PyUnicode_AppendAndDel(&res, s);
2918 if (res == NULL)
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002919 goto error;
Antoine Pitrou716c4442009-05-23 19:04:03 +00002920 }
Martin v. Löwis767046a2011-10-14 15:35:36 +02002921 modeobj = _PyObject_GetAttrId((PyObject *) self, &PyId_mode);
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002922 if (modeobj == NULL) {
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002923 if (PyErr_ExceptionMatches(PyExc_Exception))
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002924 PyErr_Clear();
2925 else
2926 goto error;
2927 }
2928 else {
2929 s = PyUnicode_FromFormat(" mode=%R", modeobj);
2930 Py_DECREF(modeobj);
2931 if (s == NULL)
2932 goto error;
2933 PyUnicode_AppendAndDel(&res, s);
2934 if (res == NULL)
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002935 goto error;
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002936 }
2937 s = PyUnicode_FromFormat("%U encoding=%R>",
2938 res, self->encoding);
2939 Py_DECREF(res);
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002940 if (status == 0) {
2941 Py_ReprLeave((PyObject *)self);
2942 }
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002943 return s;
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002944
2945 error:
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002946 Py_XDECREF(res);
Serhiy Storchakaa5af6e12017-03-19 19:25:29 +02002947 if (status == 0) {
2948 Py_ReprLeave((PyObject *)self);
2949 }
Antoine Pitroua4815ca2011-01-09 20:38:15 +00002950 return NULL;
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00002951}
2952
2953
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002954/* Inquiries */
2955
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002956/*[clinic input]
2957_io.TextIOWrapper.fileno
2958[clinic start generated code]*/
2959
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002960static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002961_io_TextIOWrapper_fileno_impl(textio *self)
2962/*[clinic end generated code: output=21490a4c3da13e6c input=c488ca83d0069f9b]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002963{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002964 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002965 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_fileno);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002966}
2967
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002968/*[clinic input]
2969_io.TextIOWrapper.seekable
2970[clinic start generated code]*/
2971
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002972static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002973_io_TextIOWrapper_seekable_impl(textio *self)
2974/*[clinic end generated code: output=ab223dbbcffc0f00 input=8b005ca06e1fca13]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002975{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002976 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002977 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_seekable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002978}
2979
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002980/*[clinic input]
2981_io.TextIOWrapper.readable
2982[clinic start generated code]*/
2983
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002984static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002985_io_TextIOWrapper_readable_impl(textio *self)
2986/*[clinic end generated code: output=72ff7ba289a8a91b input=0704ea7e01b0d3eb]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002987{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06002988 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02002989 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_readable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002990}
2991
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002992/*[clinic input]
2993_io.TextIOWrapper.writable
2994[clinic start generated code]*/
2995
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002996static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03002997_io_TextIOWrapper_writable_impl(textio *self)
2998/*[clinic end generated code: output=a728c71790d03200 input=c41740bc9d8636e8]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002999{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003000 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003001 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_writable);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003002}
3003
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003004/*[clinic input]
3005_io.TextIOWrapper.isatty
3006[clinic start generated code]*/
3007
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003008static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003009_io_TextIOWrapper_isatty_impl(textio *self)
3010/*[clinic end generated code: output=12be1a35bace882e input=fb68d9f2c99bbfff]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003011{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003012 CHECK_ATTACHED(self);
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003013 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_isatty);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003014}
3015
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003016/*[clinic input]
3017_io.TextIOWrapper.flush
3018[clinic start generated code]*/
3019
Antoine Pitrou243757e2010-11-05 21:15:39 +00003020static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003021_io_TextIOWrapper_flush_impl(textio *self)
3022/*[clinic end generated code: output=59de9165f9c2e4d2 input=928c60590694ab85]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003023{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003024 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003025 CHECK_CLOSED(self);
3026 self->telling = self->seekable;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003027 if (_textiowrapper_writeflush(self) < 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003028 return NULL;
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003029 return _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_flush);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003030}
3031
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003032/*[clinic input]
3033_io.TextIOWrapper.close
3034[clinic start generated code]*/
3035
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003036static PyObject *
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003037_io_TextIOWrapper_close_impl(textio *self)
3038/*[clinic end generated code: output=056ccf8b4876e4f4 input=9c2114315eae1948]*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003039{
3040 PyObject *res;
Antoine Pitrou6be88762010-05-03 16:48:20 +00003041 int r;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003042 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003043
Antoine Pitrou6be88762010-05-03 16:48:20 +00003044 res = textiowrapper_closed_get(self, NULL);
3045 if (res == NULL)
3046 return NULL;
3047 r = PyObject_IsTrue(res);
3048 Py_DECREF(res);
3049 if (r < 0)
3050 return NULL;
Victor Stinnerf6c57832010-05-19 01:17:01 +00003051
Antoine Pitrou6be88762010-05-03 16:48:20 +00003052 if (r > 0) {
3053 Py_RETURN_NONE; /* stream already closed */
3054 }
3055 else {
Benjamin Peterson68623612012-12-20 11:53:11 -06003056 PyObject *exc = NULL, *val, *tb;
Antoine Pitrou796564c2013-07-30 19:59:21 +02003057 if (self->finalizing) {
Victor Stinner61bdb0d2016-12-09 15:39:28 +01003058 res = _PyObject_CallMethodIdObjArgs(self->buffer,
3059 &PyId__dealloc_warn,
3060 self, NULL);
Antoine Pitroue033e062010-10-29 10:38:18 +00003061 if (res)
3062 Py_DECREF(res);
3063 else
3064 PyErr_Clear();
3065 }
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003066 res = _PyObject_CallMethodIdNoArgs((PyObject *)self, &PyId_flush);
Benjamin Peterson68623612012-12-20 11:53:11 -06003067 if (res == NULL)
3068 PyErr_Fetch(&exc, &val, &tb);
Antoine Pitrou6be88762010-05-03 16:48:20 +00003069 else
3070 Py_DECREF(res);
3071
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003072 res = _PyObject_CallMethodIdNoArgs(self->buffer, &PyId_close);
Benjamin Peterson68623612012-12-20 11:53:11 -06003073 if (exc != NULL) {
Serhiy Storchakae2bd2a72014-10-08 22:31:52 +03003074 _PyErr_ChainExceptions(exc, val, tb);
3075 Py_CLEAR(res);
Benjamin Peterson68623612012-12-20 11:53:11 -06003076 }
3077 return res;
Antoine Pitrou6be88762010-05-03 16:48:20 +00003078 }
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003079}
3080
3081static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003082textiowrapper_iternext(textio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003083{
3084 PyObject *line;
3085
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003086 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003087
3088 self->telling = 0;
3089 if (Py_TYPE(self) == &PyTextIOWrapper_Type) {
3090 /* Skip method call overhead for speed */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003091 line = _textiowrapper_readline(self, -1);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003092 }
3093 else {
Jeroen Demeyer762f93f2019-07-08 10:19:25 +02003094 line = _PyObject_CallMethodNoArgs((PyObject *)self,
3095 _PyIO_str_readline);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003096 if (line && !PyUnicode_Check(line)) {
Serhiy Storchaka55fe1ae2017-04-16 10:46:38 +03003097 PyErr_Format(PyExc_OSError,
Serhiy Storchakab6a9c972016-04-17 09:39:28 +03003098 "readline() should have returned a str object, "
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003099 "not '%.200s'", Py_TYPE(line)->tp_name);
3100 Py_DECREF(line);
3101 return NULL;
3102 }
3103 }
3104
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02003105 if (line == NULL || PyUnicode_READY(line) == -1)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003106 return NULL;
3107
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02003108 if (PyUnicode_GET_LENGTH(line) == 0) {
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003109 /* Reached EOF or would have blocked */
3110 Py_DECREF(line);
3111 Py_CLEAR(self->snapshot);
3112 self->telling = self->seekable;
3113 return NULL;
3114 }
3115
3116 return line;
3117}
3118
3119static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003120textiowrapper_name_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003121{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003122 CHECK_ATTACHED(self);
Martin v. Löwis767046a2011-10-14 15:35:36 +02003123 return _PyObject_GetAttrId(self->buffer, &PyId_name);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003124}
3125
3126static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003127textiowrapper_closed_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003128{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003129 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003130 return PyObject_GetAttr(self->buffer, _PyIO_str_closed);
3131}
3132
3133static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003134textiowrapper_newlines_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003135{
3136 PyObject *res;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003137 CHECK_ATTACHED(self);
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003138 if (self->decoder == NULL ||
3139 _PyObject_LookupAttr(self->decoder, _PyIO_str_newlines, &res) == 0)
3140 {
Serhiy Storchaka4d9aec02018-01-16 18:34:21 +02003141 Py_RETURN_NONE;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003142 }
3143 return res;
3144}
3145
3146static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003147textiowrapper_errors_get(textio *self, void *context)
Benjamin Peterson0926ad12009-06-06 18:02:12 +00003148{
3149 CHECK_INITIALIZED(self);
INADA Naoki507434f2017-12-21 09:59:53 +09003150 Py_INCREF(self->errors);
3151 return self->errors;
Benjamin Peterson0926ad12009-06-06 18:02:12 +00003152}
3153
3154static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003155textiowrapper_chunk_size_get(textio *self, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003156{
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003157 CHECK_ATTACHED(self);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003158 return PyLong_FromSsize_t(self->chunk_size);
3159}
3160
3161static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003162textiowrapper_chunk_size_set(textio *self, PyObject *arg, void *context)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003163{
3164 Py_ssize_t n;
Benjamin Peterson10e76b62014-12-21 20:51:50 -06003165 CHECK_ATTACHED_INT(self);
Zackery Spytz842acaa2018-12-17 07:52:45 -07003166 if (arg == NULL) {
3167 PyErr_SetString(PyExc_AttributeError, "cannot delete attribute");
3168 return -1;
3169 }
Antoine Pitroucb4ae812011-07-13 21:07:49 +02003170 n = PyNumber_AsSsize_t(arg, PyExc_ValueError);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003171 if (n == -1 && PyErr_Occurred())
3172 return -1;
3173 if (n <= 0) {
3174 PyErr_SetString(PyExc_ValueError,
3175 "a strictly positive integer is required");
3176 return -1;
3177 }
3178 self->chunk_size = n;
3179 return 0;
3180}
3181
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003182#include "clinic/textio.c.h"
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003183
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003184static PyMethodDef incrementalnewlinedecoder_methods[] = {
3185 _IO_INCREMENTALNEWLINEDECODER_DECODE_METHODDEF
3186 _IO_INCREMENTALNEWLINEDECODER_GETSTATE_METHODDEF
3187 _IO_INCREMENTALNEWLINEDECODER_SETSTATE_METHODDEF
3188 _IO_INCREMENTALNEWLINEDECODER_RESET_METHODDEF
3189 {NULL}
3190};
3191
3192static PyGetSetDef incrementalnewlinedecoder_getset[] = {
3193 {"newlines", (getter)incrementalnewlinedecoder_newlines_get, NULL, NULL},
3194 {NULL}
3195};
3196
3197PyTypeObject PyIncrementalNewlineDecoder_Type = {
3198 PyVarObject_HEAD_INIT(NULL, 0)
3199 "_io.IncrementalNewlineDecoder", /*tp_name*/
3200 sizeof(nldecoder_object), /*tp_basicsize*/
3201 0, /*tp_itemsize*/
3202 (destructor)incrementalnewlinedecoder_dealloc, /*tp_dealloc*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003203 0, /*tp_vectorcall_offset*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003204 0, /*tp_getattr*/
3205 0, /*tp_setattr*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003206 0, /*tp_as_async*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003207 0, /*tp_repr*/
3208 0, /*tp_as_number*/
3209 0, /*tp_as_sequence*/
3210 0, /*tp_as_mapping*/
3211 0, /*tp_hash */
3212 0, /*tp_call*/
3213 0, /*tp_str*/
3214 0, /*tp_getattro*/
3215 0, /*tp_setattro*/
3216 0, /*tp_as_buffer*/
3217 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
3218 _io_IncrementalNewlineDecoder___init____doc__, /* tp_doc */
3219 0, /* tp_traverse */
3220 0, /* tp_clear */
3221 0, /* tp_richcompare */
3222 0, /*tp_weaklistoffset*/
3223 0, /* tp_iter */
3224 0, /* tp_iternext */
3225 incrementalnewlinedecoder_methods, /* tp_methods */
3226 0, /* tp_members */
3227 incrementalnewlinedecoder_getset, /* tp_getset */
3228 0, /* tp_base */
3229 0, /* tp_dict */
3230 0, /* tp_descr_get */
3231 0, /* tp_descr_set */
3232 0, /* tp_dictoffset */
3233 _io_IncrementalNewlineDecoder___init__, /* tp_init */
3234 0, /* tp_alloc */
3235 PyType_GenericNew, /* tp_new */
3236};
3237
3238
3239static PyMethodDef textiowrapper_methods[] = {
3240 _IO_TEXTIOWRAPPER_DETACH_METHODDEF
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02003241 _IO_TEXTIOWRAPPER_RECONFIGURE_METHODDEF
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003242 _IO_TEXTIOWRAPPER_WRITE_METHODDEF
3243 _IO_TEXTIOWRAPPER_READ_METHODDEF
3244 _IO_TEXTIOWRAPPER_READLINE_METHODDEF
3245 _IO_TEXTIOWRAPPER_FLUSH_METHODDEF
3246 _IO_TEXTIOWRAPPER_CLOSE_METHODDEF
3247
3248 _IO_TEXTIOWRAPPER_FILENO_METHODDEF
3249 _IO_TEXTIOWRAPPER_SEEKABLE_METHODDEF
3250 _IO_TEXTIOWRAPPER_READABLE_METHODDEF
3251 _IO_TEXTIOWRAPPER_WRITABLE_METHODDEF
3252 _IO_TEXTIOWRAPPER_ISATTY_METHODDEF
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003253
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003254 _IO_TEXTIOWRAPPER_SEEK_METHODDEF
3255 _IO_TEXTIOWRAPPER_TELL_METHODDEF
3256 _IO_TEXTIOWRAPPER_TRUNCATE_METHODDEF
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003257 {NULL, NULL}
3258};
3259
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003260static PyMemberDef textiowrapper_members[] = {
3261 {"encoding", T_OBJECT, offsetof(textio, encoding), READONLY},
3262 {"buffer", T_OBJECT, offsetof(textio, buffer), READONLY},
3263 {"line_buffering", T_BOOL, offsetof(textio, line_buffering), READONLY},
Antoine Pitrou3c2817b2017-06-03 12:32:28 +02003264 {"write_through", T_BOOL, offsetof(textio, write_through), READONLY},
Antoine Pitrou796564c2013-07-30 19:59:21 +02003265 {"_finalizing", T_BOOL, offsetof(textio, finalizing), 0},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003266 {NULL}
3267};
3268
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003269static PyGetSetDef textiowrapper_getset[] = {
3270 {"name", (getter)textiowrapper_name_get, NULL, NULL},
3271 {"closed", (getter)textiowrapper_closed_get, NULL, NULL},
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003272/* {"mode", (getter)TextIOWrapper_mode_get, NULL, NULL},
3273*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003274 {"newlines", (getter)textiowrapper_newlines_get, NULL, NULL},
3275 {"errors", (getter)textiowrapper_errors_get, NULL, NULL},
3276 {"_CHUNK_SIZE", (getter)textiowrapper_chunk_size_get,
3277 (setter)textiowrapper_chunk_size_set, NULL},
Benjamin Peterson1fea3212009-04-19 03:15:20 +00003278 {NULL}
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003279};
3280
3281PyTypeObject PyTextIOWrapper_Type = {
3282 PyVarObject_HEAD_INIT(NULL, 0)
3283 "_io.TextIOWrapper", /*tp_name*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003284 sizeof(textio), /*tp_basicsize*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003285 0, /*tp_itemsize*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003286 (destructor)textiowrapper_dealloc, /*tp_dealloc*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003287 0, /*tp_vectorcall_offset*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003288 0, /*tp_getattr*/
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00003289 0, /*tps_etattr*/
Jeroen Demeyer530f5062019-05-31 04:13:39 +02003290 0, /*tp_as_async*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003291 (reprfunc)textiowrapper_repr,/*tp_repr*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003292 0, /*tp_as_number*/
3293 0, /*tp_as_sequence*/
3294 0, /*tp_as_mapping*/
3295 0, /*tp_hash */
3296 0, /*tp_call*/
3297 0, /*tp_str*/
3298 0, /*tp_getattro*/
3299 0, /*tp_setattro*/
3300 0, /*tp_as_buffer*/
3301 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE
Antoine Pitrouada319b2019-05-29 22:12:38 +02003302 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003303 _io_TextIOWrapper___init____doc__, /* tp_doc */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003304 (traverseproc)textiowrapper_traverse, /* tp_traverse */
3305 (inquiry)textiowrapper_clear, /* tp_clear */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003306 0, /* tp_richcompare */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003307 offsetof(textio, weakreflist), /*tp_weaklistoffset*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003308 0, /* tp_iter */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003309 (iternextfunc)textiowrapper_iternext, /* tp_iternext */
3310 textiowrapper_methods, /* tp_methods */
3311 textiowrapper_members, /* tp_members */
3312 textiowrapper_getset, /* tp_getset */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003313 0, /* tp_base */
3314 0, /* tp_dict */
3315 0, /* tp_descr_get */
3316 0, /* tp_descr_set */
Benjamin Peterson680bf1a2009-06-12 02:07:12 +00003317 offsetof(textio, dict), /*tp_dictoffset*/
Serhiy Storchakaf24131f2015-04-16 11:19:43 +03003318 _io_TextIOWrapper___init__, /* tp_init */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003319 0, /* tp_alloc */
3320 PyType_GenericNew, /* tp_new */
Antoine Pitrou796564c2013-07-30 19:59:21 +02003321 0, /* tp_free */
3322 0, /* tp_is_gc */
3323 0, /* tp_bases */
3324 0, /* tp_mro */
3325 0, /* tp_cache */
3326 0, /* tp_subclasses */
3327 0, /* tp_weaklist */
3328 0, /* tp_del */
3329 0, /* tp_version_tag */
3330 0, /* tp_finalize */
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00003331};