blob: 3ef9e2e26ff4f12c3d632149d351fca419b8e01c [file] [log] [blame]
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001#include "Python.h"
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00002#include "structmember.h" /* for offsetof() */
3#include "_iomodule.h"
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00004
5typedef struct {
6 PyObject_HEAD
7 char *buf;
8 Py_ssize_t pos;
9 Py_ssize_t string_size;
10 size_t buf_size;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000011 PyObject *dict;
12 PyObject *weakreflist;
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000013} bytesio;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +000014
15#define CHECK_CLOSED(self) \
16 if ((self)->buf == NULL) { \
17 PyErr_SetString(PyExc_ValueError, \
18 "I/O operation on closed file."); \
19 return NULL; \
20 }
21
22/* Internal routine to get a line from the buffer of a BytesIO
23 object. Returns the length between the current position to the
24 next newline character. */
25static Py_ssize_t
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000026get_line(bytesio *self, char **output)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +000027{
28 char *n;
29 const char *str_end;
30 Py_ssize_t len;
31
32 assert(self->buf != NULL);
33
34 /* Move to the end of the line, up to the end of the string, s. */
35 str_end = self->buf + self->string_size;
36 for (n = self->buf + self->pos;
37 n < str_end && *n != '\n';
38 n++);
39
40 /* Skip the newline character */
41 if (n < str_end)
42 n++;
43
44 /* Get the length from the current position to the end of the line. */
45 len = n - (self->buf + self->pos);
46 *output = self->buf + self->pos;
47
48 assert(len >= 0);
49 assert(self->pos < PY_SSIZE_T_MAX - len);
50 self->pos += len;
51
52 return len;
53}
54
55/* Internal routine for changing the size of the buffer of BytesIO objects.
56 The caller should ensure that the 'size' argument is non-negative. Returns
57 0 on success, -1 otherwise. */
58static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +000059resize_buffer(bytesio *self, size_t size)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +000060{
61 /* Here, unsigned types are used to avoid dealing with signed integer
62 overflow, which is undefined in C. */
63 size_t alloc = self->buf_size;
64 char *new_buf = NULL;
65
66 assert(self->buf != NULL);
67
68 /* For simplicity, stay in the range of the signed type. Anyway, Python
69 doesn't allow strings to be longer than this. */
70 if (size > PY_SSIZE_T_MAX)
71 goto overflow;
72
73 if (size < alloc / 2) {
74 /* Major downsize; resize down to exact size. */
75 alloc = size + 1;
76 }
77 else if (size < alloc) {
78 /* Within allocated size; quick exit */
79 return 0;
80 }
81 else if (size <= alloc * 1.125) {
82 /* Moderate upsize; overallocate similar to list_resize() */
83 alloc = size + (size >> 3) + (size < 9 ? 3 : 6);
84 }
85 else {
86 /* Major upsize; resize up to exact size */
87 alloc = size + 1;
88 }
89
90 if (alloc > ((size_t)-1) / sizeof(char))
91 goto overflow;
92 new_buf = (char *)PyMem_Realloc(self->buf, alloc * sizeof(char));
93 if (new_buf == NULL) {
94 PyErr_NoMemory();
95 return -1;
96 }
97 self->buf_size = alloc;
98 self->buf = new_buf;
99
100 return 0;
101
102 overflow:
103 PyErr_SetString(PyExc_OverflowError,
104 "new buffer size too large");
105 return -1;
106}
107
108/* Internal routine for writing a string of bytes to the buffer of a BytesIO
109 object. Returns the number of bytes wrote, or -1 on error. */
110static Py_ssize_t
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000111write_bytes(bytesio *self, const char *bytes, Py_ssize_t len)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000112{
113 assert(self->buf != NULL);
114 assert(self->pos >= 0);
115 assert(len >= 0);
116
Alexandre Vassalotti1bfe9dc82008-05-07 01:44:31 +0000117 if ((size_t)self->pos + len > self->buf_size) {
118 if (resize_buffer(self, (size_t)self->pos + len) < 0)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000119 return -1;
120 }
121
122 if (self->pos > self->string_size) {
123 /* In case of overseek, pad with null bytes the buffer region between
124 the end of stream and the current position.
125
126 0 lo string_size hi
127 | |<---used--->|<----------available----------->|
128 | | <--to pad-->|<---to write---> |
129 0 buf position
130 */
131 memset(self->buf + self->string_size, '\0',
132 (self->pos - self->string_size) * sizeof(char));
133 }
134
135 /* Copy the data to the internal buffer, overwriting some of the existing
136 data if self->pos < self->string_size. */
137 memcpy(self->buf + self->pos, bytes, len);
138 self->pos += len;
139
140 /* Set the new length of the internal string if it has changed. */
141 if (self->string_size < self->pos) {
142 self->string_size = self->pos;
143 }
144
145 return len;
146}
147
148static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000149bytesio_get_closed(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000150{
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000151 if (self->buf == NULL) {
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000152 Py_RETURN_TRUE;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000153 }
154 else {
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000155 Py_RETURN_FALSE;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000156 }
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000157}
158
159/* Generic getter for the writable, readable and seekable properties */
160static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000161return_true(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000162{
163 Py_RETURN_TRUE;
164}
165
166PyDoc_STRVAR(flush_doc,
167"flush() -> None. Does nothing.");
168
169static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000170bytesio_flush(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000171{
Antoine Pitrou6be88762010-05-03 16:48:20 +0000172 CHECK_CLOSED(self);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000173 Py_RETURN_NONE;
174}
175
176PyDoc_STRVAR(getval_doc,
Alexandre Vassalotti10dfc1e2008-05-08 01:34:41 +0000177"getvalue() -> bytes.\n"
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000178"\n"
179"Retrieve the entire contents of the BytesIO object.");
180
181static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000182bytesio_getvalue(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000183{
184 CHECK_CLOSED(self);
Christian Heimes72b710a2008-05-26 13:28:38 +0000185 return PyBytes_FromStringAndSize(self->buf, self->string_size);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000186}
187
188PyDoc_STRVAR(isatty_doc,
189"isatty() -> False.\n"
190"\n"
191"Always returns False since BytesIO objects are not connected\n"
192"to a tty-like device.");
193
194static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000195bytesio_isatty(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000196{
197 CHECK_CLOSED(self);
198 Py_RETURN_FALSE;
199}
200
201PyDoc_STRVAR(tell_doc,
202"tell() -> current file position, an integer\n");
203
204static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000205bytesio_tell(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000206{
207 CHECK_CLOSED(self);
208 return PyLong_FromSsize_t(self->pos);
209}
210
211PyDoc_STRVAR(read_doc,
212"read([size]) -> read at most size bytes, returned as a string.\n"
213"\n"
214"If the size argument is negative, read until EOF is reached.\n"
215"Return an empty string at EOF.");
216
217static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000218bytesio_read(bytesio *self, PyObject *args)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000219{
220 Py_ssize_t size, n;
221 char *output;
222 PyObject *arg = Py_None;
223
224 CHECK_CLOSED(self);
225
226 if (!PyArg_ParseTuple(args, "|O:read", &arg))
227 return NULL;
228
229 if (PyLong_Check(arg)) {
230 size = PyLong_AsSsize_t(arg);
Benjamin Petersona8a93042008-09-30 02:18:09 +0000231 if (size == -1 && PyErr_Occurred())
232 return NULL;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000233 }
234 else if (arg == Py_None) {
235 /* Read until EOF is reached, by default. */
236 size = -1;
237 }
238 else {
239 PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
240 Py_TYPE(arg)->tp_name);
241 return NULL;
242 }
243
244 /* adjust invalid sizes */
245 n = self->string_size - self->pos;
246 if (size < 0 || size > n) {
247 size = n;
248 if (size < 0)
249 size = 0;
250 }
251
252 assert(self->buf != NULL);
253 output = self->buf + self->pos;
254 self->pos += size;
255
Christian Heimes72b710a2008-05-26 13:28:38 +0000256 return PyBytes_FromStringAndSize(output, size);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000257}
258
259
260PyDoc_STRVAR(read1_doc,
261"read1(size) -> read at most size bytes, returned as a string.\n"
262"\n"
263"If the size argument is negative or omitted, read until EOF is reached.\n"
264"Return an empty string at EOF.");
265
266static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000267bytesio_read1(bytesio *self, PyObject *n)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000268{
269 PyObject *arg, *res;
270
271 arg = PyTuple_Pack(1, n);
272 if (arg == NULL)
273 return NULL;
274 res = bytesio_read(self, arg);
275 Py_DECREF(arg);
276 return res;
277}
278
279PyDoc_STRVAR(readline_doc,
280"readline([size]) -> next line from the file, as a string.\n"
281"\n"
282"Retain newline. A non-negative size argument limits the maximum\n"
283"number of bytes to return (an incomplete line may be returned then).\n"
284"Return an empty string at EOF.\n");
285
286static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000287bytesio_readline(bytesio *self, PyObject *args)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000288{
289 Py_ssize_t size, n;
290 char *output;
291 PyObject *arg = Py_None;
292
293 CHECK_CLOSED(self);
294
295 if (!PyArg_ParseTuple(args, "|O:readline", &arg))
296 return NULL;
297
298 if (PyLong_Check(arg)) {
299 size = PyLong_AsSsize_t(arg);
Benjamin Petersona8a93042008-09-30 02:18:09 +0000300 if (size == -1 && PyErr_Occurred())
301 return NULL;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000302 }
303 else if (arg == Py_None) {
304 /* No size limit, by default. */
305 size = -1;
306 }
307 else {
308 PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
309 Py_TYPE(arg)->tp_name);
310 return NULL;
311 }
312
313 n = get_line(self, &output);
314
315 if (size >= 0 && size < n) {
316 size = n - size;
317 n -= size;
318 self->pos -= size;
319 }
320
Christian Heimes72b710a2008-05-26 13:28:38 +0000321 return PyBytes_FromStringAndSize(output, n);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000322}
323
324PyDoc_STRVAR(readlines_doc,
325"readlines([size]) -> list of strings, each a line from the file.\n"
326"\n"
327"Call readline() repeatedly and return a list of the lines so read.\n"
328"The optional size argument, if given, is an approximate bound on the\n"
329"total number of bytes in the lines returned.\n");
330
331static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000332bytesio_readlines(bytesio *self, PyObject *args)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000333{
334 Py_ssize_t maxsize, size, n;
335 PyObject *result, *line;
336 char *output;
337 PyObject *arg = Py_None;
338
339 CHECK_CLOSED(self);
340
341 if (!PyArg_ParseTuple(args, "|O:readlines", &arg))
342 return NULL;
343
344 if (PyLong_Check(arg)) {
345 maxsize = PyLong_AsSsize_t(arg);
Benjamin Petersona8a93042008-09-30 02:18:09 +0000346 if (maxsize == -1 && PyErr_Occurred())
347 return NULL;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000348 }
349 else if (arg == Py_None) {
350 /* No size limit, by default. */
351 maxsize = -1;
352 }
353 else {
354 PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
355 Py_TYPE(arg)->tp_name);
356 return NULL;
357 }
358
359 size = 0;
360 result = PyList_New(0);
361 if (!result)
362 return NULL;
363
364 while ((n = get_line(self, &output)) != 0) {
Christian Heimes72b710a2008-05-26 13:28:38 +0000365 line = PyBytes_FromStringAndSize(output, n);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000366 if (!line)
367 goto on_error;
368 if (PyList_Append(result, line) == -1) {
369 Py_DECREF(line);
370 goto on_error;
371 }
372 Py_DECREF(line);
373 size += n;
374 if (maxsize > 0 && size >= maxsize)
375 break;
376 }
377 return result;
378
379 on_error:
380 Py_DECREF(result);
381 return NULL;
382}
383
384PyDoc_STRVAR(readinto_doc,
Alexandre Vassalotti10dfc1e2008-05-08 01:34:41 +0000385"readinto(bytearray) -> int. Read up to len(b) bytes into b.\n"
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000386"\n"
387"Returns number of bytes read (0 for EOF), or None if the object\n"
388"is set not to block as has no data to read.");
389
390static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000391bytesio_readinto(bytesio *self, PyObject *buffer)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000392{
393 void *raw_buffer;
394 Py_ssize_t len;
395
396 CHECK_CLOSED(self);
397
398 if (PyObject_AsWriteBuffer(buffer, &raw_buffer, &len) == -1)
399 return NULL;
400
401 if (self->pos + len > self->string_size)
402 len = self->string_size - self->pos;
403
404 memcpy(raw_buffer, self->buf + self->pos, len);
405 assert(self->pos + len < PY_SSIZE_T_MAX);
406 assert(len >= 0);
407 self->pos += len;
408
409 return PyLong_FromSsize_t(len);
410}
411
412PyDoc_STRVAR(truncate_doc,
413"truncate([size]) -> int. Truncate the file to at most size bytes.\n"
414"\n"
415"Size defaults to the current file position, as returned by tell().\n"
Antoine Pitrou905a2ff2010-01-31 22:47:27 +0000416"The current file position is unchanged. Returns the new size.\n");
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000417
418static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000419bytesio_truncate(bytesio *self, PyObject *args)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000420{
421 Py_ssize_t size;
422 PyObject *arg = Py_None;
423
424 CHECK_CLOSED(self);
425
426 if (!PyArg_ParseTuple(args, "|O:truncate", &arg))
427 return NULL;
428
429 if (PyLong_Check(arg)) {
430 size = PyLong_AsSsize_t(arg);
Benjamin Petersona8a93042008-09-30 02:18:09 +0000431 if (size == -1 && PyErr_Occurred())
432 return NULL;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000433 }
434 else if (arg == Py_None) {
435 /* Truncate to current position if no argument is passed. */
436 size = self->pos;
437 }
438 else {
439 PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
440 Py_TYPE(arg)->tp_name);
441 return NULL;
442 }
443
444 if (size < 0) {
445 PyErr_Format(PyExc_ValueError,
446 "negative size value %zd", size);
447 return NULL;
448 }
449
450 if (size < self->string_size) {
451 self->string_size = size;
452 if (resize_buffer(self, size) < 0)
453 return NULL;
454 }
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000455
456 return PyLong_FromSsize_t(size);
457}
458
459static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000460bytesio_iternext(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000461{
462 char *next;
463 Py_ssize_t n;
464
465 CHECK_CLOSED(self);
466
467 n = get_line(self, &next);
468
469 if (!next || n == 0)
470 return NULL;
471
Christian Heimes72b710a2008-05-26 13:28:38 +0000472 return PyBytes_FromStringAndSize(next, n);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000473}
474
475PyDoc_STRVAR(seek_doc,
476"seek(pos, whence=0) -> int. Change stream position.\n"
477"\n"
478"Seek to byte offset pos relative to position indicated by whence:\n"
479" 0 Start of stream (the default). pos should be >= 0;\n"
480" 1 Current position - pos may be negative;\n"
481" 2 End of stream - pos usually negative.\n"
482"Returns the new absolute position.");
483
484static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000485bytesio_seek(bytesio *self, PyObject *args)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000486{
487 Py_ssize_t pos;
488 int mode = 0;
489
490 CHECK_CLOSED(self);
491
492 if (!PyArg_ParseTuple(args, "n|i:seek", &pos, &mode))
493 return NULL;
494
495 if (pos < 0 && mode == 0) {
496 PyErr_Format(PyExc_ValueError,
497 "negative seek value %zd", pos);
498 return NULL;
499 }
500
501 /* mode 0: offset relative to beginning of the string.
502 mode 1: offset relative to current position.
503 mode 2: offset relative the end of the string. */
504 if (mode == 1) {
505 if (pos > PY_SSIZE_T_MAX - self->pos) {
506 PyErr_SetString(PyExc_OverflowError,
507 "new position too large");
508 return NULL;
509 }
510 pos += self->pos;
511 }
512 else if (mode == 2) {
513 if (pos > PY_SSIZE_T_MAX - self->string_size) {
514 PyErr_SetString(PyExc_OverflowError,
515 "new position too large");
516 return NULL;
517 }
518 pos += self->string_size;
519 }
520 else if (mode != 0) {
521 PyErr_Format(PyExc_ValueError,
522 "invalid whence (%i, should be 0, 1 or 2)", mode);
523 return NULL;
524 }
525
526 if (pos < 0)
527 pos = 0;
528 self->pos = pos;
529
530 return PyLong_FromSsize_t(self->pos);
531}
532
533PyDoc_STRVAR(write_doc,
Alexandre Vassalotti10dfc1e2008-05-08 01:34:41 +0000534"write(bytes) -> int. Write bytes to file.\n"
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000535"\n"
536"Return the number of bytes written.");
537
538static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000539bytesio_write(bytesio *self, PyObject *obj)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000540{
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000541 Py_ssize_t n = 0;
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000542 Py_buffer buf;
543 PyObject *result = NULL;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000544
545 CHECK_CLOSED(self);
546
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000547 if (PyObject_GetBuffer(obj, &buf, PyBUF_CONTIG_RO) < 0)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000548 return NULL;
549
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000550 if (buf.len != 0)
551 n = write_bytes(self, buf.buf, buf.len);
552 if (n >= 0)
553 result = PyLong_FromSsize_t(n);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000554
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000555 PyBuffer_Release(&buf);
556 return result;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000557}
558
559PyDoc_STRVAR(writelines_doc,
Alexandre Vassalotti7d060892008-05-07 01:47:37 +0000560"writelines(sequence_of_strings) -> None. Write strings to the file.\n"
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000561"\n"
Alexandre Vassalotti7d060892008-05-07 01:47:37 +0000562"Note that newlines are not added. The sequence can be any iterable\n"
563"object producing strings. This is equivalent to calling write() for\n"
564"each string.");
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000565
566static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000567bytesio_writelines(bytesio *self, PyObject *v)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000568{
569 PyObject *it, *item;
570 PyObject *ret;
571
572 CHECK_CLOSED(self);
573
574 it = PyObject_GetIter(v);
575 if (it == NULL)
576 return NULL;
577
578 while ((item = PyIter_Next(it)) != NULL) {
579 ret = bytesio_write(self, item);
580 Py_DECREF(item);
581 if (ret == NULL) {
582 Py_DECREF(it);
583 return NULL;
584 }
585 Py_DECREF(ret);
586 }
587 Py_DECREF(it);
588
589 /* See if PyIter_Next failed */
590 if (PyErr_Occurred())
591 return NULL;
592
593 Py_RETURN_NONE;
594}
595
596PyDoc_STRVAR(close_doc,
597"close() -> None. Disable all I/O operations.");
598
599static PyObject *
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000600bytesio_close(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000601{
602 if (self->buf != NULL) {
603 PyMem_Free(self->buf);
604 self->buf = NULL;
605 }
606 Py_RETURN_NONE;
607}
608
Alexandre Vassalotticf76e1a2009-07-22 03:24:36 +0000609/* Pickling support.
610
611 Note that only pickle protocol 2 and onward are supported since we use
612 extended __reduce__ API of PEP 307 to make BytesIO instances picklable.
613
614 Providing support for protocol < 2 would require the __reduce_ex__ method
615 which is notably long-winded when defined properly.
616
617 For BytesIO, the implementation would similar to one coded for
618 object.__reduce_ex__, but slightly less general. To be more specific, we
619 could call bytesio_getstate directly and avoid checking for the presence of
620 a fallback __reduce__ method. However, we would still need a __newobj__
621 function to use the efficient instance representation of PEP 307.
622 */
623
624static PyObject *
625bytesio_getstate(bytesio *self)
626{
627 PyObject *initvalue = bytesio_getvalue(self);
628 PyObject *dict;
629 PyObject *state;
630
631 if (initvalue == NULL)
632 return NULL;
633 if (self->dict == NULL) {
634 Py_INCREF(Py_None);
635 dict = Py_None;
636 }
637 else {
638 dict = PyDict_Copy(self->dict);
639 if (dict == NULL)
640 return NULL;
641 }
642
643 state = Py_BuildValue("(OnN)", initvalue, self->pos, dict);
644 Py_DECREF(initvalue);
645 return state;
646}
647
648static PyObject *
649bytesio_setstate(bytesio *self, PyObject *state)
650{
651 PyObject *result;
652 PyObject *position_obj;
653 PyObject *dict;
654 Py_ssize_t pos;
655
656 assert(state != NULL);
657
658 /* We allow the state tuple to be longer than 3, because we may need
659 someday to extend the object's state without breaking
660 backward-compatibility. */
661 if (!PyTuple_Check(state) || Py_SIZE(state) < 3) {
662 PyErr_Format(PyExc_TypeError,
663 "%.200s.__setstate__ argument should be 3-tuple, got %.200s",
664 Py_TYPE(self)->tp_name, Py_TYPE(state)->tp_name);
665 return NULL;
666 }
667 /* Reset the object to its default state. This is only needed to handle
668 the case of repeated calls to __setstate__. */
669 self->string_size = 0;
670 self->pos = 0;
671
672 /* Set the value of the internal buffer. If state[0] does not support the
673 buffer protocol, bytesio_write will raise the appropriate TypeError. */
674 result = bytesio_write(self, PyTuple_GET_ITEM(state, 0));
675 if (result == NULL)
676 return NULL;
677 Py_DECREF(result);
678
679 /* Set carefully the position value. Alternatively, we could use the seek
680 method instead of modifying self->pos directly to better protect the
681 object internal state against errneous (or malicious) inputs. */
682 position_obj = PyTuple_GET_ITEM(state, 1);
683 if (!PyLong_Check(position_obj)) {
684 PyErr_Format(PyExc_TypeError,
685 "second item of state must be an integer, not %.200s",
686 Py_TYPE(position_obj)->tp_name);
687 return NULL;
688 }
689 pos = PyLong_AsSsize_t(position_obj);
690 if (pos == -1 && PyErr_Occurred())
691 return NULL;
692 if (pos < 0) {
693 PyErr_SetString(PyExc_ValueError,
694 "position value cannot be negative");
695 return NULL;
696 }
697 self->pos = pos;
698
699 /* Set the dictionary of the instance variables. */
700 dict = PyTuple_GET_ITEM(state, 2);
701 if (dict != Py_None) {
702 if (!PyDict_Check(dict)) {
703 PyErr_Format(PyExc_TypeError,
704 "third item of state should be a dict, got a %.200s",
705 Py_TYPE(dict)->tp_name);
706 return NULL;
707 }
708 if (self->dict) {
709 /* Alternatively, we could replace the internal dictionary
710 completely. However, it seems more practical to just update it. */
711 if (PyDict_Update(self->dict, dict) < 0)
712 return NULL;
713 }
714 else {
715 Py_INCREF(dict);
716 self->dict = dict;
717 }
718 }
719
720 Py_RETURN_NONE;
721}
722
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000723static void
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000724bytesio_dealloc(bytesio *self)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000725{
Alexandre Vassalottifc477042009-07-22 02:24:49 +0000726 _PyObject_GC_UNTRACK(self);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000727 if (self->buf != NULL) {
728 PyMem_Free(self->buf);
729 self->buf = NULL;
730 }
Alexandre Vassalottifc477042009-07-22 02:24:49 +0000731 Py_CLEAR(self->dict);
732 if (self->weakreflist != NULL)
733 PyObject_ClearWeakRefs((PyObject *) self);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000734 Py_TYPE(self)->tp_free(self);
735}
736
737static PyObject *
738bytesio_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
739{
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000740 bytesio *self;
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000741
742 assert(type != NULL && type->tp_alloc != NULL);
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000743 self = (bytesio *)type->tp_alloc(type, 0);
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000744 if (self == NULL)
745 return NULL;
746
Alexandre Vassalotticf76e1a2009-07-22 03:24:36 +0000747 /* tp_alloc initializes all the fields to zero. So we don't have to
748 initialize them here. */
749
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000750 self->buf = (char *)PyMem_Malloc(0);
751 if (self->buf == NULL) {
752 Py_DECREF(self);
753 return PyErr_NoMemory();
754 }
755
756 return (PyObject *)self;
757}
758
759static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000760bytesio_init(bytesio *self, PyObject *args, PyObject *kwds)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000761{
Alexandre Vassalottiba5c7432009-08-04 23:19:13 +0000762 char *kwlist[] = {"initial_bytes", NULL};
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000763 PyObject *initvalue = NULL;
764
Alexandre Vassalottiba5c7432009-08-04 23:19:13 +0000765 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O:BytesIO", kwlist,
766 &initvalue))
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000767 return -1;
768
769 /* In case, __init__ is called multiple times. */
770 self->string_size = 0;
771 self->pos = 0;
772
773 if (initvalue && initvalue != Py_None) {
774 PyObject *res;
775 res = bytesio_write(self, initvalue);
776 if (res == NULL)
777 return -1;
778 Py_DECREF(res);
779 self->pos = 0;
780 }
781
782 return 0;
783}
784
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000785static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000786bytesio_traverse(bytesio *self, visitproc visit, void *arg)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000787{
788 Py_VISIT(self->dict);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000789 return 0;
790}
791
792static int
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000793bytesio_clear(bytesio *self)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000794{
795 Py_CLEAR(self->dict);
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000796 return 0;
797}
798
799
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000800static PyGetSetDef bytesio_getsetlist[] = {
801 {"closed", (getter)bytesio_get_closed, NULL,
802 "True if the file is closed."},
Benjamin Peterson1fea3212009-04-19 03:15:20 +0000803 {NULL}, /* sentinel */
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000804};
805
806static struct PyMethodDef bytesio_methods[] = {
807 {"readable", (PyCFunction)return_true, METH_NOARGS, NULL},
808 {"seekable", (PyCFunction)return_true, METH_NOARGS, NULL},
809 {"writable", (PyCFunction)return_true, METH_NOARGS, NULL},
810 {"close", (PyCFunction)bytesio_close, METH_NOARGS, close_doc},
811 {"flush", (PyCFunction)bytesio_flush, METH_NOARGS, flush_doc},
812 {"isatty", (PyCFunction)bytesio_isatty, METH_NOARGS, isatty_doc},
813 {"tell", (PyCFunction)bytesio_tell, METH_NOARGS, tell_doc},
814 {"write", (PyCFunction)bytesio_write, METH_O, write_doc},
815 {"writelines", (PyCFunction)bytesio_writelines, METH_O, writelines_doc},
816 {"read1", (PyCFunction)bytesio_read1, METH_O, read1_doc},
817 {"readinto", (PyCFunction)bytesio_readinto, METH_O, readinto_doc},
818 {"readline", (PyCFunction)bytesio_readline, METH_VARARGS, readline_doc},
819 {"readlines", (PyCFunction)bytesio_readlines, METH_VARARGS, readlines_doc},
820 {"read", (PyCFunction)bytesio_read, METH_VARARGS, read_doc},
Antoine Pitroud5c3f6c2010-09-02 19:48:07 +0000821 {"getvalue", (PyCFunction)bytesio_getvalue, METH_NOARGS, getval_doc},
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000822 {"seek", (PyCFunction)bytesio_seek, METH_VARARGS, seek_doc},
823 {"truncate", (PyCFunction)bytesio_truncate, METH_VARARGS, truncate_doc},
Alexandre Vassalotticf76e1a2009-07-22 03:24:36 +0000824 {"__getstate__", (PyCFunction)bytesio_getstate, METH_NOARGS, NULL},
825 {"__setstate__", (PyCFunction)bytesio_setstate, METH_O, NULL},
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000826 {NULL, NULL} /* sentinel */
827};
828
829PyDoc_STRVAR(bytesio_doc,
830"BytesIO([buffer]) -> object\n"
831"\n"
832"Create a buffered I/O implementation using an in-memory bytes\n"
833"buffer, ready for reading and writing.");
834
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000835PyTypeObject PyBytesIO_Type = {
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000836 PyVarObject_HEAD_INIT(NULL, 0)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000837 "_io.BytesIO", /*tp_name*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000838 sizeof(bytesio), /*tp_basicsize*/
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000839 0, /*tp_itemsize*/
840 (destructor)bytesio_dealloc, /*tp_dealloc*/
841 0, /*tp_print*/
842 0, /*tp_getattr*/
843 0, /*tp_setattr*/
Mark Dickinsone94c6792009-02-02 20:36:42 +0000844 0, /*tp_reserved*/
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000845 0, /*tp_repr*/
846 0, /*tp_as_number*/
847 0, /*tp_as_sequence*/
848 0, /*tp_as_mapping*/
849 0, /*tp_hash*/
850 0, /*tp_call*/
851 0, /*tp_str*/
852 0, /*tp_getattro*/
853 0, /*tp_setattro*/
854 0, /*tp_as_buffer*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000855 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE |
856 Py_TPFLAGS_HAVE_GC, /*tp_flags*/
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000857 bytesio_doc, /*tp_doc*/
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000858 (traverseproc)bytesio_traverse, /*tp_traverse*/
859 (inquiry)bytesio_clear, /*tp_clear*/
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000860 0, /*tp_richcompare*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000861 offsetof(bytesio, weakreflist), /*tp_weaklistoffset*/
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000862 PyObject_SelfIter, /*tp_iter*/
863 (iternextfunc)bytesio_iternext, /*tp_iternext*/
864 bytesio_methods, /*tp_methods*/
865 0, /*tp_members*/
866 bytesio_getsetlist, /*tp_getset*/
867 0, /*tp_base*/
868 0, /*tp_dict*/
869 0, /*tp_descr_get*/
870 0, /*tp_descr_set*/
Benjamin Peterson680bf1a2009-06-12 02:07:12 +0000871 offsetof(bytesio, dict), /*tp_dictoffset*/
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000872 (initproc)bytesio_init, /*tp_init*/
873 0, /*tp_alloc*/
874 bytesio_new, /*tp_new*/
875};