blob: 9f1cf0a1f7f754a378f209b99102e382d37e4b85 [file] [log] [blame]
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001/* PyByteArray (bytearray) implementation */
2
3#define PY_SSIZE_T_CLEAN
4#include "Python.h"
5#include "structmember.h"
6#include "bytes_methods.h"
7
Antoine Pitroufc8d6f42010-01-17 12:38:54 +00008char _PyByteArray_empty_string[] = "";
Christian Heimes2c9c7a52008-05-26 13:42:13 +00009
10void
11PyByteArray_Fini(void)
12{
Christian Heimes2c9c7a52008-05-26 13:42:13 +000013}
14
15int
16PyByteArray_Init(void)
17{
Christian Heimes2c9c7a52008-05-26 13:42:13 +000018 return 1;
19}
20
21/* end nullbytes support */
22
23/* Helpers */
24
25static int
26_getbytevalue(PyObject* arg, int *value)
27{
28 long face_value;
29
30 if (PyLong_Check(arg)) {
31 face_value = PyLong_AsLong(arg);
Georg Brandl9a54d7c2008-07-16 23:15:30 +000032 } else {
33 PyObject *index = PyNumber_Index(arg);
34 if (index == NULL) {
35 PyErr_Format(PyExc_TypeError, "an integer is required");
Mark Dickinson10de93a2010-07-09 19:25:48 +000036 *value = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +000037 return 0;
38 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +000039 face_value = PyLong_AsLong(index);
40 Py_DECREF(index);
41 }
42
43 if (face_value < 0 || face_value >= 256) {
44 /* this includes the OverflowError in case the long is too large */
45 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
Mark Dickinson10de93a2010-07-09 19:25:48 +000046 *value = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +000047 return 0;
48 }
49
50 *value = face_value;
51 return 1;
52}
53
54static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +000055bytearray_getbuffer(PyByteArrayObject *obj, Py_buffer *view, int flags)
Christian Heimes2c9c7a52008-05-26 13:42:13 +000056{
57 int ret;
58 void *ptr;
59 if (view == NULL) {
60 obj->ob_exports++;
61 return 0;
62 }
Antoine Pitroufc8d6f42010-01-17 12:38:54 +000063 ptr = (void *) PyByteArray_AS_STRING(obj);
Martin v. Löwis423be952008-08-13 15:53:07 +000064 ret = PyBuffer_FillInfo(view, (PyObject*)obj, ptr, Py_SIZE(obj), 0, flags);
Christian Heimes2c9c7a52008-05-26 13:42:13 +000065 if (ret >= 0) {
66 obj->ob_exports++;
67 }
68 return ret;
69}
70
71static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +000072bytearray_releasebuffer(PyByteArrayObject *obj, Py_buffer *view)
Christian Heimes2c9c7a52008-05-26 13:42:13 +000073{
74 obj->ob_exports--;
75}
76
77static Py_ssize_t
78_getbuffer(PyObject *obj, Py_buffer *view)
79{
80 PyBufferProcs *buffer = Py_TYPE(obj)->tp_as_buffer;
81
82 if (buffer == NULL || buffer->bf_getbuffer == NULL)
83 {
84 PyErr_Format(PyExc_TypeError,
85 "Type %.100s doesn't support the buffer API",
86 Py_TYPE(obj)->tp_name);
87 return -1;
88 }
89
90 if (buffer->bf_getbuffer(obj, view, PyBUF_SIMPLE) < 0)
91 return -1;
92 return view->len;
93}
94
Antoine Pitrou5504e892008-12-06 21:27:53 +000095static int
96_canresize(PyByteArrayObject *self)
97{
98 if (self->ob_exports > 0) {
99 PyErr_SetString(PyExc_BufferError,
100 "Existing exports of data: object cannot be re-sized");
101 return 0;
102 }
103 return 1;
104}
105
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000106/* Direct API functions */
107
108PyObject *
109PyByteArray_FromObject(PyObject *input)
110{
111 return PyObject_CallFunctionObjArgs((PyObject *)&PyByteArray_Type,
112 input, NULL);
113}
114
115PyObject *
116PyByteArray_FromStringAndSize(const char *bytes, Py_ssize_t size)
117{
118 PyByteArrayObject *new;
119 Py_ssize_t alloc;
120
121 if (size < 0) {
122 PyErr_SetString(PyExc_SystemError,
123 "Negative size passed to PyByteArray_FromStringAndSize");
124 return NULL;
125 }
126
Neal Norwitz3ce5d922008-08-24 07:08:55 +0000127 /* Prevent buffer overflow when setting alloc to size+1. */
128 if (size == PY_SSIZE_T_MAX) {
129 return PyErr_NoMemory();
130 }
131
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000132 new = PyObject_New(PyByteArrayObject, &PyByteArray_Type);
133 if (new == NULL)
134 return NULL;
135
136 if (size == 0) {
137 new->ob_bytes = NULL;
138 alloc = 0;
139 }
140 else {
141 alloc = size + 1;
Antoine Pitrou39aba4f2011-11-12 21:15:28 +0100142 new->ob_bytes = PyObject_Malloc(alloc);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000143 if (new->ob_bytes == NULL) {
144 Py_DECREF(new);
145 return PyErr_NoMemory();
146 }
Antoine Pitroufc8d6f42010-01-17 12:38:54 +0000147 if (bytes != NULL && size > 0)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000148 memcpy(new->ob_bytes, bytes, size);
149 new->ob_bytes[size] = '\0'; /* Trailing null byte */
150 }
151 Py_SIZE(new) = size;
152 new->ob_alloc = alloc;
153 new->ob_exports = 0;
154
155 return (PyObject *)new;
156}
157
158Py_ssize_t
159PyByteArray_Size(PyObject *self)
160{
161 assert(self != NULL);
162 assert(PyByteArray_Check(self));
163
164 return PyByteArray_GET_SIZE(self);
165}
166
167char *
168PyByteArray_AsString(PyObject *self)
169{
170 assert(self != NULL);
171 assert(PyByteArray_Check(self));
172
173 return PyByteArray_AS_STRING(self);
174}
175
176int
177PyByteArray_Resize(PyObject *self, Py_ssize_t size)
178{
179 void *sval;
180 Py_ssize_t alloc = ((PyByteArrayObject *)self)->ob_alloc;
181
182 assert(self != NULL);
183 assert(PyByteArray_Check(self));
184 assert(size >= 0);
185
Antoine Pitrou5504e892008-12-06 21:27:53 +0000186 if (size == Py_SIZE(self)) {
187 return 0;
188 }
189 if (!_canresize((PyByteArrayObject *)self)) {
190 return -1;
191 }
192
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000193 if (size < alloc / 2) {
194 /* Major downsize; resize down to exact size */
195 alloc = size + 1;
196 }
197 else if (size < alloc) {
198 /* Within allocated size; quick exit */
199 Py_SIZE(self) = size;
200 ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null */
201 return 0;
202 }
203 else if (size <= alloc * 1.125) {
204 /* Moderate upsize; overallocate similar to list_resize() */
205 alloc = size + (size >> 3) + (size < 9 ? 3 : 6);
206 }
207 else {
208 /* Major upsize; resize up to exact size */
209 alloc = size + 1;
210 }
211
Antoine Pitrou39aba4f2011-11-12 21:15:28 +0100212 sval = PyObject_Realloc(((PyByteArrayObject *)self)->ob_bytes, alloc);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000213 if (sval == NULL) {
214 PyErr_NoMemory();
215 return -1;
216 }
217
218 ((PyByteArrayObject *)self)->ob_bytes = sval;
219 Py_SIZE(self) = size;
220 ((PyByteArrayObject *)self)->ob_alloc = alloc;
221 ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null byte */
222
223 return 0;
224}
225
226PyObject *
227PyByteArray_Concat(PyObject *a, PyObject *b)
228{
229 Py_ssize_t size;
230 Py_buffer va, vb;
231 PyByteArrayObject *result = NULL;
232
233 va.len = -1;
234 vb.len = -1;
235 if (_getbuffer(a, &va) < 0 ||
236 _getbuffer(b, &vb) < 0) {
237 PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s",
238 Py_TYPE(a)->tp_name, Py_TYPE(b)->tp_name);
239 goto done;
240 }
241
242 size = va.len + vb.len;
243 if (size < 0) {
Benjamin Petersone0124bd2009-03-09 21:04:33 +0000244 PyErr_NoMemory();
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000245 goto done;
246 }
247
248 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, size);
249 if (result != NULL) {
250 memcpy(result->ob_bytes, va.buf, va.len);
251 memcpy(result->ob_bytes + va.len, vb.buf, vb.len);
252 }
253
254 done:
255 if (va.len != -1)
Martin v. Löwis423be952008-08-13 15:53:07 +0000256 PyBuffer_Release(&va);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000257 if (vb.len != -1)
Martin v. Löwis423be952008-08-13 15:53:07 +0000258 PyBuffer_Release(&vb);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000259 return (PyObject *)result;
260}
261
262/* Functions stuffed into the type object */
263
264static Py_ssize_t
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000265bytearray_length(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000266{
267 return Py_SIZE(self);
268}
269
270static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000271bytearray_iconcat(PyByteArrayObject *self, PyObject *other)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000272{
273 Py_ssize_t mysize;
274 Py_ssize_t size;
275 Py_buffer vo;
276
277 if (_getbuffer(other, &vo) < 0) {
278 PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s",
279 Py_TYPE(other)->tp_name, Py_TYPE(self)->tp_name);
280 return NULL;
281 }
282
283 mysize = Py_SIZE(self);
284 size = mysize + vo.len;
285 if (size < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +0000286 PyBuffer_Release(&vo);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000287 return PyErr_NoMemory();
288 }
289 if (size < self->ob_alloc) {
290 Py_SIZE(self) = size;
291 self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */
292 }
293 else if (PyByteArray_Resize((PyObject *)self, size) < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +0000294 PyBuffer_Release(&vo);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000295 return NULL;
296 }
297 memcpy(self->ob_bytes + mysize, vo.buf, vo.len);
Martin v. Löwis423be952008-08-13 15:53:07 +0000298 PyBuffer_Release(&vo);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000299 Py_INCREF(self);
300 return (PyObject *)self;
301}
302
303static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000304bytearray_repeat(PyByteArrayObject *self, Py_ssize_t count)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000305{
306 PyByteArrayObject *result;
307 Py_ssize_t mysize;
308 Py_ssize_t size;
309
310 if (count < 0)
311 count = 0;
312 mysize = Py_SIZE(self);
Mark Dickinsoncf940c72010-08-10 18:35:01 +0000313 if (count > 0 && mysize > PY_SSIZE_T_MAX / count)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000314 return PyErr_NoMemory();
Mark Dickinsoncf940c72010-08-10 18:35:01 +0000315 size = mysize * count;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000316 result = (PyByteArrayObject *)PyByteArray_FromStringAndSize(NULL, size);
317 if (result != NULL && size != 0) {
318 if (mysize == 1)
319 memset(result->ob_bytes, self->ob_bytes[0], size);
320 else {
321 Py_ssize_t i;
322 for (i = 0; i < count; i++)
323 memcpy(result->ob_bytes + i*mysize, self->ob_bytes, mysize);
324 }
325 }
326 return (PyObject *)result;
327}
328
329static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000330bytearray_irepeat(PyByteArrayObject *self, Py_ssize_t count)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000331{
332 Py_ssize_t mysize;
333 Py_ssize_t size;
334
335 if (count < 0)
336 count = 0;
337 mysize = Py_SIZE(self);
Mark Dickinsoncf940c72010-08-10 18:35:01 +0000338 if (count > 0 && mysize > PY_SSIZE_T_MAX / count)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000339 return PyErr_NoMemory();
Mark Dickinsoncf940c72010-08-10 18:35:01 +0000340 size = mysize * count;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000341 if (size < self->ob_alloc) {
342 Py_SIZE(self) = size;
343 self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */
344 }
345 else if (PyByteArray_Resize((PyObject *)self, size) < 0)
346 return NULL;
347
348 if (mysize == 1)
349 memset(self->ob_bytes, self->ob_bytes[0], size);
350 else {
351 Py_ssize_t i;
352 for (i = 1; i < count; i++)
353 memcpy(self->ob_bytes + i*mysize, self->ob_bytes, mysize);
354 }
355
356 Py_INCREF(self);
357 return (PyObject *)self;
358}
359
360static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000361bytearray_getitem(PyByteArrayObject *self, Py_ssize_t i)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000362{
363 if (i < 0)
364 i += Py_SIZE(self);
365 if (i < 0 || i >= Py_SIZE(self)) {
366 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
367 return NULL;
368 }
369 return PyLong_FromLong((unsigned char)(self->ob_bytes[i]));
370}
371
372static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000373bytearray_subscript(PyByteArrayObject *self, PyObject *index)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000374{
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000375 if (PyIndex_Check(index)) {
376 Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000377
378 if (i == -1 && PyErr_Occurred())
379 return NULL;
380
381 if (i < 0)
382 i += PyByteArray_GET_SIZE(self);
383
384 if (i < 0 || i >= Py_SIZE(self)) {
385 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
386 return NULL;
387 }
388 return PyLong_FromLong((unsigned char)(self->ob_bytes[i]));
389 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000390 else if (PySlice_Check(index)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000391 Py_ssize_t start, stop, step, slicelength, cur, i;
Martin v. Löwis4d0d4712010-12-03 20:14:31 +0000392 if (PySlice_GetIndicesEx(index,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000393 PyByteArray_GET_SIZE(self),
394 &start, &stop, &step, &slicelength) < 0) {
395 return NULL;
396 }
397
398 if (slicelength <= 0)
399 return PyByteArray_FromStringAndSize("", 0);
400 else if (step == 1) {
401 return PyByteArray_FromStringAndSize(self->ob_bytes + start,
402 slicelength);
403 }
404 else {
405 char *source_buf = PyByteArray_AS_STRING(self);
Alexandre Vassalottie2641f42009-04-03 06:38:02 +0000406 char *result_buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000407 PyObject *result;
408
Alexandre Vassalottie2641f42009-04-03 06:38:02 +0000409 result = PyByteArray_FromStringAndSize(NULL, slicelength);
410 if (result == NULL)
411 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000412
Alexandre Vassalottie2641f42009-04-03 06:38:02 +0000413 result_buf = PyByteArray_AS_STRING(result);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000414 for (cur = start, i = 0; i < slicelength;
415 cur += step, i++) {
416 result_buf[i] = source_buf[cur];
417 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000418 return result;
419 }
420 }
421 else {
422 PyErr_SetString(PyExc_TypeError, "bytearray indices must be integers");
423 return NULL;
424 }
425}
426
427static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000428bytearray_setslice(PyByteArrayObject *self, Py_ssize_t lo, Py_ssize_t hi,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000429 PyObject *values)
430{
431 Py_ssize_t avail, needed;
432 void *bytes;
433 Py_buffer vbytes;
434 int res = 0;
435
436 vbytes.len = -1;
437 if (values == (PyObject *)self) {
438 /* Make a copy and call this function recursively */
439 int err;
440 values = PyByteArray_FromObject(values);
441 if (values == NULL)
442 return -1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000443 err = bytearray_setslice(self, lo, hi, values);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000444 Py_DECREF(values);
445 return err;
446 }
447 if (values == NULL) {
448 /* del b[lo:hi] */
449 bytes = NULL;
450 needed = 0;
451 }
452 else {
453 if (_getbuffer(values, &vbytes) < 0) {
454 PyErr_Format(PyExc_TypeError,
Georg Brandl3dbca812008-07-23 16:10:53 +0000455 "can't set bytearray slice from %.100s",
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000456 Py_TYPE(values)->tp_name);
457 return -1;
458 }
459 needed = vbytes.len;
460 bytes = vbytes.buf;
461 }
462
463 if (lo < 0)
464 lo = 0;
465 if (hi < lo)
466 hi = lo;
467 if (hi > Py_SIZE(self))
468 hi = Py_SIZE(self);
469
470 avail = hi - lo;
471 if (avail < 0)
472 lo = hi = avail = 0;
473
474 if (avail != needed) {
475 if (avail > needed) {
Antoine Pitrou5504e892008-12-06 21:27:53 +0000476 if (!_canresize(self)) {
477 res = -1;
478 goto finish;
479 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000480 /*
481 0 lo hi old_size
482 | |<----avail----->|<-----tomove------>|
483 | |<-needed->|<-----tomove------>|
484 0 lo new_hi new_size
485 */
486 memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi,
487 Py_SIZE(self) - hi);
488 }
489 /* XXX(nnorwitz): need to verify this can't overflow! */
490 if (PyByteArray_Resize((PyObject *)self,
491 Py_SIZE(self) + needed - avail) < 0) {
492 res = -1;
493 goto finish;
494 }
495 if (avail < needed) {
496 /*
497 0 lo hi old_size
498 | |<-avail->|<-----tomove------>|
499 | |<----needed---->|<-----tomove------>|
500 0 lo new_hi new_size
501 */
502 memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi,
503 Py_SIZE(self) - lo - needed);
504 }
505 }
506
507 if (needed > 0)
508 memcpy(self->ob_bytes + lo, bytes, needed);
509
510
511 finish:
512 if (vbytes.len != -1)
Martin v. Löwis423be952008-08-13 15:53:07 +0000513 PyBuffer_Release(&vbytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000514 return res;
515}
516
517static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000518bytearray_setitem(PyByteArrayObject *self, Py_ssize_t i, PyObject *value)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000519{
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000520 int ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000521
522 if (i < 0)
523 i += Py_SIZE(self);
524
525 if (i < 0 || i >= Py_SIZE(self)) {
526 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
527 return -1;
528 }
529
530 if (value == NULL)
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000531 return bytearray_setslice(self, i, i+1, NULL);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000532
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000533 if (!_getbytevalue(value, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000534 return -1;
535
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000536 self->ob_bytes[i] = ival;
537 return 0;
538}
539
540static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000541bytearray_ass_subscript(PyByteArrayObject *self, PyObject *index, PyObject *values)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000542{
543 Py_ssize_t start, stop, step, slicelen, needed;
544 char *bytes;
545
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000546 if (PyIndex_Check(index)) {
547 Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000548
549 if (i == -1 && PyErr_Occurred())
550 return -1;
551
552 if (i < 0)
553 i += PyByteArray_GET_SIZE(self);
554
555 if (i < 0 || i >= Py_SIZE(self)) {
556 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
557 return -1;
558 }
559
560 if (values == NULL) {
561 /* Fall through to slice assignment */
562 start = i;
563 stop = i + 1;
564 step = 1;
565 slicelen = 1;
566 }
567 else {
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000568 int ival;
569 if (!_getbytevalue(values, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000570 return -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000571 self->ob_bytes[i] = (char)ival;
572 return 0;
573 }
574 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000575 else if (PySlice_Check(index)) {
Martin v. Löwis4d0d4712010-12-03 20:14:31 +0000576 if (PySlice_GetIndicesEx(index,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000577 PyByteArray_GET_SIZE(self),
578 &start, &stop, &step, &slicelen) < 0) {
579 return -1;
580 }
581 }
582 else {
583 PyErr_SetString(PyExc_TypeError, "bytearray indices must be integer");
584 return -1;
585 }
586
587 if (values == NULL) {
588 bytes = NULL;
589 needed = 0;
590 }
591 else if (values == (PyObject *)self || !PyByteArray_Check(values)) {
Christian Heimes6d26ade2012-11-03 23:07:59 +0100592 int err;
Ezio Melottic64bcbe2012-11-03 21:19:06 +0200593 if (PyNumber_Check(values) || PyUnicode_Check(values)) {
594 PyErr_SetString(PyExc_TypeError,
595 "can assign only bytes, buffers, or iterables "
596 "of ints in range(0, 256)");
597 return -1;
598 }
Georg Brandlf3fa5682010-12-04 17:09:30 +0000599 /* Make a copy and call this function recursively */
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000600 values = PyByteArray_FromObject(values);
601 if (values == NULL)
602 return -1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000603 err = bytearray_ass_subscript(self, index, values);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000604 Py_DECREF(values);
605 return err;
606 }
607 else {
608 assert(PyByteArray_Check(values));
609 bytes = ((PyByteArrayObject *)values)->ob_bytes;
610 needed = Py_SIZE(values);
611 }
612 /* Make sure b[5:2] = ... inserts before 5, not before 2. */
613 if ((step < 0 && start < stop) ||
614 (step > 0 && start > stop))
615 stop = start;
616 if (step == 1) {
617 if (slicelen != needed) {
Antoine Pitrou5504e892008-12-06 21:27:53 +0000618 if (!_canresize(self))
619 return -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000620 if (slicelen > needed) {
621 /*
622 0 start stop old_size
623 | |<---slicelen--->|<-----tomove------>|
624 | |<-needed->|<-----tomove------>|
625 0 lo new_hi new_size
626 */
627 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
628 Py_SIZE(self) - stop);
629 }
630 if (PyByteArray_Resize((PyObject *)self,
631 Py_SIZE(self) + needed - slicelen) < 0)
632 return -1;
633 if (slicelen < needed) {
634 /*
635 0 lo hi old_size
636 | |<-avail->|<-----tomove------>|
637 | |<----needed---->|<-----tomove------>|
638 0 lo new_hi new_size
639 */
640 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
641 Py_SIZE(self) - start - needed);
642 }
643 }
644
645 if (needed > 0)
646 memcpy(self->ob_bytes + start, bytes, needed);
647
648 return 0;
649 }
650 else {
651 if (needed == 0) {
652 /* Delete slice */
Mark Dickinsonbc099642010-01-29 17:27:24 +0000653 size_t cur;
654 Py_ssize_t i;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000655
Antoine Pitrou5504e892008-12-06 21:27:53 +0000656 if (!_canresize(self))
657 return -1;
Mark Dickinson7e3b9482010-08-06 21:33:18 +0000658
659 if (slicelen == 0)
660 /* Nothing to do here. */
661 return 0;
662
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000663 if (step < 0) {
664 stop = start + 1;
665 start = stop + step * (slicelen - 1) - 1;
666 step = -step;
667 }
668 for (cur = start, i = 0;
669 i < slicelen; cur += step, i++) {
670 Py_ssize_t lim = step - 1;
671
Mark Dickinson66f575b2010-02-14 12:53:32 +0000672 if (cur + step >= (size_t)PyByteArray_GET_SIZE(self))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000673 lim = PyByteArray_GET_SIZE(self) - cur - 1;
674
675 memmove(self->ob_bytes + cur - i,
676 self->ob_bytes + cur + 1, lim);
677 }
678 /* Move the tail of the bytes, in one chunk */
Mark Dickinson7e3b9482010-08-06 21:33:18 +0000679 cur = start + (size_t)slicelen*step;
Mark Dickinson66f575b2010-02-14 12:53:32 +0000680 if (cur < (size_t)PyByteArray_GET_SIZE(self)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000681 memmove(self->ob_bytes + cur - slicelen,
682 self->ob_bytes + cur,
683 PyByteArray_GET_SIZE(self) - cur);
684 }
685 if (PyByteArray_Resize((PyObject *)self,
686 PyByteArray_GET_SIZE(self) - slicelen) < 0)
687 return -1;
688
689 return 0;
690 }
691 else {
692 /* Assign slice */
Mark Dickinson7e3b9482010-08-06 21:33:18 +0000693 Py_ssize_t i;
694 size_t cur;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000695
696 if (needed != slicelen) {
697 PyErr_Format(PyExc_ValueError,
698 "attempt to assign bytes of size %zd "
699 "to extended slice of size %zd",
700 needed, slicelen);
701 return -1;
702 }
703 for (cur = start, i = 0; i < slicelen; cur += step, i++)
704 self->ob_bytes[cur] = bytes[i];
705 return 0;
706 }
707 }
708}
709
710static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000711bytearray_init(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000712{
713 static char *kwlist[] = {"source", "encoding", "errors", 0};
714 PyObject *arg = NULL;
715 const char *encoding = NULL;
716 const char *errors = NULL;
717 Py_ssize_t count;
718 PyObject *it;
719 PyObject *(*iternext)(PyObject *);
720
721 if (Py_SIZE(self) != 0) {
722 /* Empty previous contents (yes, do this first of all!) */
723 if (PyByteArray_Resize((PyObject *)self, 0) < 0)
724 return -1;
725 }
726
727 /* Parse arguments */
Georg Brandl3dbca812008-07-23 16:10:53 +0000728 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|Oss:bytearray", kwlist,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000729 &arg, &encoding, &errors))
730 return -1;
731
732 /* Make a quick exit if no first argument */
733 if (arg == NULL) {
734 if (encoding != NULL || errors != NULL) {
735 PyErr_SetString(PyExc_TypeError,
736 "encoding or errors without sequence argument");
737 return -1;
738 }
739 return 0;
740 }
741
742 if (PyUnicode_Check(arg)) {
743 /* Encode via the codec registry */
744 PyObject *encoded, *new;
745 if (encoding == NULL) {
746 PyErr_SetString(PyExc_TypeError,
747 "string argument without an encoding");
748 return -1;
749 }
Marc-André Lemburgb2750b52008-06-06 12:18:17 +0000750 encoded = PyUnicode_AsEncodedString(arg, encoding, errors);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000751 if (encoded == NULL)
752 return -1;
753 assert(PyBytes_Check(encoded));
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000754 new = bytearray_iconcat(self, encoded);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000755 Py_DECREF(encoded);
756 if (new == NULL)
757 return -1;
758 Py_DECREF(new);
759 return 0;
760 }
761
762 /* If it's not unicode, there can't be encoding or errors */
763 if (encoding != NULL || errors != NULL) {
764 PyErr_SetString(PyExc_TypeError,
765 "encoding or errors without a string argument");
766 return -1;
767 }
768
769 /* Is it an int? */
Benjamin Peterson8380dd52010-04-16 22:51:37 +0000770 count = PyNumber_AsSsize_t(arg, PyExc_OverflowError);
771 if (count == -1 && PyErr_Occurred()) {
772 if (PyErr_ExceptionMatches(PyExc_OverflowError))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000773 return -1;
Benjamin Peterson9c0e94f2010-04-16 23:00:53 +0000774 PyErr_Clear();
Benjamin Peterson8380dd52010-04-16 22:51:37 +0000775 }
776 else if (count < 0) {
777 PyErr_SetString(PyExc_ValueError, "negative count");
778 return -1;
779 }
780 else {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000781 if (count > 0) {
782 if (PyByteArray_Resize((PyObject *)self, count))
783 return -1;
784 memset(self->ob_bytes, 0, count);
785 }
786 return 0;
787 }
788
789 /* Use the buffer API */
790 if (PyObject_CheckBuffer(arg)) {
791 Py_ssize_t size;
792 Py_buffer view;
793 if (PyObject_GetBuffer(arg, &view, PyBUF_FULL_RO) < 0)
794 return -1;
795 size = view.len;
796 if (PyByteArray_Resize((PyObject *)self, size) < 0) goto fail;
797 if (PyBuffer_ToContiguous(self->ob_bytes, &view, size, 'C') < 0)
Stefan Krah7d12d9d2012-07-28 12:25:55 +0200798 goto fail;
Martin v. Löwis423be952008-08-13 15:53:07 +0000799 PyBuffer_Release(&view);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000800 return 0;
801 fail:
Martin v. Löwis423be952008-08-13 15:53:07 +0000802 PyBuffer_Release(&view);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000803 return -1;
804 }
805
806 /* XXX Optimize this if the arguments is a list, tuple */
807
808 /* Get the iterator */
809 it = PyObject_GetIter(arg);
810 if (it == NULL)
811 return -1;
812 iternext = *Py_TYPE(it)->tp_iternext;
813
814 /* Run the iterator to exhaustion */
815 for (;;) {
816 PyObject *item;
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000817 int rc, value;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000818
819 /* Get the next item */
820 item = iternext(it);
821 if (item == NULL) {
822 if (PyErr_Occurred()) {
823 if (!PyErr_ExceptionMatches(PyExc_StopIteration))
824 goto error;
825 PyErr_Clear();
826 }
827 break;
828 }
829
830 /* Interpret it as an int (__index__) */
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000831 rc = _getbytevalue(item, &value);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000832 Py_DECREF(item);
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000833 if (!rc)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000834 goto error;
835
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000836 /* Append the byte */
837 if (Py_SIZE(self) < self->ob_alloc)
838 Py_SIZE(self)++;
839 else if (PyByteArray_Resize((PyObject *)self, Py_SIZE(self)+1) < 0)
840 goto error;
841 self->ob_bytes[Py_SIZE(self)-1] = value;
842 }
843
844 /* Clean up and return success */
845 Py_DECREF(it);
846 return 0;
847
848 error:
849 /* Error handling when it != NULL */
850 Py_DECREF(it);
851 return -1;
852}
853
854/* Mostly copied from string_repr, but without the
855 "smart quote" functionality. */
856static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000857bytearray_repr(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000858{
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000859 const char *quote_prefix = "bytearray(b";
860 const char *quote_postfix = ")";
861 Py_ssize_t length = Py_SIZE(self);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200862 /* 15 == strlen(quote_prefix) + 2 + strlen(quote_postfix) + 1 */
Mark Dickinson66f575b2010-02-14 12:53:32 +0000863 size_t newsize;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000864 PyObject *v;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200865 register Py_ssize_t i;
866 register char c;
867 register char *p;
868 int quote;
869 char *test, *start;
870 char *buffer;
871
872 if (length > (PY_SSIZE_T_MAX - 15) / 4) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000873 PyErr_SetString(PyExc_OverflowError,
874 "bytearray object is too large to make repr");
875 return NULL;
876 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200877
878 newsize = 15 + length * 4;
Antoine Pitrou39aba4f2011-11-12 21:15:28 +0100879 buffer = PyObject_Malloc(newsize);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200880 if (buffer == NULL) {
881 PyErr_NoMemory();
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000882 return NULL;
883 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000884
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200885 /* Figure out which quote to use; single is preferred */
886 quote = '\'';
887 start = PyByteArray_AS_STRING(self);
888 for (test = start; test < start+length; ++test) {
889 if (*test == '"') {
890 quote = '\''; /* back to single */
891 break;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000892 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200893 else if (*test == '\'')
894 quote = '"';
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000895 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200896
897 p = buffer;
898 while (*quote_prefix)
899 *p++ = *quote_prefix++;
900 *p++ = quote;
901
902 for (i = 0; i < length; i++) {
903 /* There's at least enough room for a hex escape
904 and a closing quote. */
905 assert(newsize - (p - buffer) >= 5);
906 c = self->ob_bytes[i];
907 if (c == '\'' || c == '\\')
908 *p++ = '\\', *p++ = c;
909 else if (c == '\t')
910 *p++ = '\\', *p++ = 't';
911 else if (c == '\n')
912 *p++ = '\\', *p++ = 'n';
913 else if (c == '\r')
914 *p++ = '\\', *p++ = 'r';
915 else if (c == 0)
916 *p++ = '\\', *p++ = 'x', *p++ = '0', *p++ = '0';
917 else if (c < ' ' || c >= 0x7f) {
918 *p++ = '\\';
919 *p++ = 'x';
Victor Stinnerf5cff562011-10-14 02:13:11 +0200920 *p++ = Py_hexdigits[(c & 0xf0) >> 4];
921 *p++ = Py_hexdigits[c & 0xf];
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200922 }
923 else
924 *p++ = c;
925 }
926 assert(newsize - (p - buffer) >= 1);
927 *p++ = quote;
928 while (*quote_postfix) {
929 *p++ = *quote_postfix++;
930 }
931
932 v = PyUnicode_DecodeASCII(buffer, p - buffer, NULL);
Antoine Pitrou39aba4f2011-11-12 21:15:28 +0100933 PyObject_Free(buffer);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200934 return v;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000935}
936
937static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000938bytearray_str(PyObject *op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000939{
Alexander Belopolskyf0f45142010-08-11 17:31:17 +0000940 if (Py_BytesWarningFlag) {
941 if (PyErr_WarnEx(PyExc_BytesWarning,
942 "str() on a bytearray instance", 1))
943 return NULL;
944 }
945 return bytearray_repr((PyByteArrayObject*)op);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000946}
947
948static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000949bytearray_richcompare(PyObject *self, PyObject *other, int op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000950{
951 Py_ssize_t self_size, other_size;
952 Py_buffer self_bytes, other_bytes;
953 PyObject *res;
954 Py_ssize_t minsize;
955 int cmp;
956
957 /* Bytes can be compared to anything that supports the (binary)
958 buffer API. Except that a comparison with Unicode is always an
959 error, even if the comparison is for equality. */
960 if (PyObject_IsInstance(self, (PyObject*)&PyUnicode_Type) ||
961 PyObject_IsInstance(other, (PyObject*)&PyUnicode_Type)) {
Barry Warsaw9e9dcd62008-10-17 01:50:37 +0000962 if (Py_BytesWarningFlag && (op == Py_EQ || op == Py_NE)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000963 if (PyErr_WarnEx(PyExc_BytesWarning,
Georg Brandle5d68ac2008-06-04 11:30:26 +0000964 "Comparison between bytearray and string", 1))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000965 return NULL;
966 }
967
Brian Curtindfc80e32011-08-10 20:28:54 -0500968 Py_RETURN_NOTIMPLEMENTED;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000969 }
970
971 self_size = _getbuffer(self, &self_bytes);
972 if (self_size < 0) {
973 PyErr_Clear();
Brian Curtindfc80e32011-08-10 20:28:54 -0500974 Py_RETURN_NOTIMPLEMENTED;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000975 }
976
977 other_size = _getbuffer(other, &other_bytes);
978 if (other_size < 0) {
979 PyErr_Clear();
Martin v. Löwis423be952008-08-13 15:53:07 +0000980 PyBuffer_Release(&self_bytes);
Brian Curtindfc80e32011-08-10 20:28:54 -0500981 Py_RETURN_NOTIMPLEMENTED;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000982 }
983
984 if (self_size != other_size && (op == Py_EQ || op == Py_NE)) {
985 /* Shortcut: if the lengths differ, the objects differ */
986 cmp = (op == Py_NE);
987 }
988 else {
989 minsize = self_size;
990 if (other_size < minsize)
991 minsize = other_size;
992
993 cmp = memcmp(self_bytes.buf, other_bytes.buf, minsize);
994 /* In ISO C, memcmp() guarantees to use unsigned bytes! */
995
996 if (cmp == 0) {
997 if (self_size < other_size)
998 cmp = -1;
999 else if (self_size > other_size)
1000 cmp = 1;
1001 }
1002
1003 switch (op) {
1004 case Py_LT: cmp = cmp < 0; break;
1005 case Py_LE: cmp = cmp <= 0; break;
1006 case Py_EQ: cmp = cmp == 0; break;
1007 case Py_NE: cmp = cmp != 0; break;
1008 case Py_GT: cmp = cmp > 0; break;
1009 case Py_GE: cmp = cmp >= 0; break;
1010 }
1011 }
1012
1013 res = cmp ? Py_True : Py_False;
Martin v. Löwis423be952008-08-13 15:53:07 +00001014 PyBuffer_Release(&self_bytes);
1015 PyBuffer_Release(&other_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001016 Py_INCREF(res);
1017 return res;
1018}
1019
1020static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001021bytearray_dealloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001022{
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001023 if (self->ob_exports > 0) {
1024 PyErr_SetString(PyExc_SystemError,
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00001025 "deallocated bytearray object has exported buffers");
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001026 PyErr_Print();
1027 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001028 if (self->ob_bytes != 0) {
Antoine Pitrou39aba4f2011-11-12 21:15:28 +01001029 PyObject_Free(self->ob_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001030 }
1031 Py_TYPE(self)->tp_free((PyObject *)self);
1032}
1033
1034
1035/* -------------------------------------------------------------------- */
1036/* Methods */
1037
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001038#define FASTSEARCH fastsearch
1039#define STRINGLIB(F) stringlib_##F
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001040#define STRINGLIB_CHAR char
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001041#define STRINGLIB_LEN PyByteArray_GET_SIZE
1042#define STRINGLIB_STR PyByteArray_AS_STRING
1043#define STRINGLIB_NEW PyByteArray_FromStringAndSize
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001044#define STRINGLIB_ISSPACE Py_ISSPACE
1045#define STRINGLIB_ISLINEBREAK(x) ((x == '\n') || (x == '\r'))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001046#define STRINGLIB_CHECK_EXACT PyByteArray_CheckExact
1047#define STRINGLIB_MUTABLE 1
1048
1049#include "stringlib/fastsearch.h"
1050#include "stringlib/count.h"
1051#include "stringlib/find.h"
1052#include "stringlib/partition.h"
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001053#include "stringlib/split.h"
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001054#include "stringlib/ctype.h"
1055#include "stringlib/transmogrify.h"
1056
1057
1058/* The following Py_LOCAL_INLINE and Py_LOCAL functions
1059were copied from the old char* style string object. */
1060
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001061/* helper macro to fixup start/end slice values */
1062#define ADJUST_INDICES(start, end, len) \
1063 if (end > len) \
1064 end = len; \
1065 else if (end < 0) { \
1066 end += len; \
1067 if (end < 0) \
1068 end = 0; \
1069 } \
1070 if (start < 0) { \
1071 start += len; \
1072 if (start < 0) \
1073 start = 0; \
1074 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001075
1076Py_LOCAL_INLINE(Py_ssize_t)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001077bytearray_find_internal(PyByteArrayObject *self, PyObject *args, int dir)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001078{
1079 PyObject *subobj;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001080 char byte;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001081 Py_buffer subbuf;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001082 const char *sub;
1083 Py_ssize_t sub_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001084 Py_ssize_t start=0, end=PY_SSIZE_T_MAX;
1085 Py_ssize_t res;
1086
Antoine Pitrouac65d962011-10-20 23:54:17 +02001087 if (!stringlib_parse_args_finds_byte("find/rfind/index/rindex",
1088 args, &subobj, &byte, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001089 return -2;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001090
1091 if (subobj) {
1092 if (_getbuffer(subobj, &subbuf) < 0)
1093 return -2;
1094
1095 sub = subbuf.buf;
1096 sub_len = subbuf.len;
1097 }
1098 else {
1099 sub = &byte;
1100 sub_len = 1;
1101 }
1102
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001103 if (dir > 0)
1104 res = stringlib_find_slice(
1105 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
Antoine Pitrouac65d962011-10-20 23:54:17 +02001106 sub, sub_len, start, end);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001107 else
1108 res = stringlib_rfind_slice(
1109 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
Antoine Pitrouac65d962011-10-20 23:54:17 +02001110 sub, sub_len, start, end);
1111
1112 if (subobj)
1113 PyBuffer_Release(&subbuf);
1114
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001115 return res;
1116}
1117
1118PyDoc_STRVAR(find__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001119"B.find(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001120\n\
1121Return the lowest index in B where subsection sub is found,\n\
Senthil Kumaran53516a82011-07-27 23:33:54 +08001122such that sub is contained within B[start,end]. Optional\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001123arguments start and end are interpreted as in slice notation.\n\
1124\n\
1125Return -1 on failure.");
1126
1127static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001128bytearray_find(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001129{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001130 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001131 if (result == -2)
1132 return NULL;
1133 return PyLong_FromSsize_t(result);
1134}
1135
1136PyDoc_STRVAR(count__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001137"B.count(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001138\n\
1139Return the number of non-overlapping occurrences of subsection sub in\n\
1140bytes B[start:end]. Optional arguments start and end are interpreted\n\
1141as in slice notation.");
1142
1143static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001144bytearray_count(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001145{
1146 PyObject *sub_obj;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001147 const char *str = PyByteArray_AS_STRING(self), *sub;
1148 Py_ssize_t sub_len;
1149 char byte;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001150 Py_ssize_t start = 0, end = PY_SSIZE_T_MAX;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001151
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001152 Py_buffer vsub;
1153 PyObject *count_obj;
1154
Antoine Pitrouac65d962011-10-20 23:54:17 +02001155 if (!stringlib_parse_args_finds_byte("count", args, &sub_obj, &byte,
1156 &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001157 return NULL;
1158
Antoine Pitrouac65d962011-10-20 23:54:17 +02001159 if (sub_obj) {
1160 if (_getbuffer(sub_obj, &vsub) < 0)
1161 return NULL;
1162
1163 sub = vsub.buf;
1164 sub_len = vsub.len;
1165 }
1166 else {
1167 sub = &byte;
1168 sub_len = 1;
1169 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001170
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001171 ADJUST_INDICES(start, end, PyByteArray_GET_SIZE(self));
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001172
1173 count_obj = PyLong_FromSsize_t(
Antoine Pitrouac65d962011-10-20 23:54:17 +02001174 stringlib_count(str + start, end - start, sub, sub_len, PY_SSIZE_T_MAX)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001175 );
Antoine Pitrouac65d962011-10-20 23:54:17 +02001176
1177 if (sub_obj)
1178 PyBuffer_Release(&vsub);
1179
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001180 return count_obj;
1181}
1182
Eli Bendersky4db28d32011-03-03 18:21:02 +00001183PyDoc_STRVAR(clear__doc__,
1184"B.clear() -> None\n\
1185\n\
1186Remove all items from B.");
1187
Victor Stinner6430fd52011-09-29 04:02:13 +02001188static PyObject *
Eli Bendersky4db28d32011-03-03 18:21:02 +00001189bytearray_clear(PyByteArrayObject *self)
1190{
1191 if (PyByteArray_Resize((PyObject *)self, 0) < 0)
1192 return NULL;
1193 Py_RETURN_NONE;
1194}
1195
1196PyDoc_STRVAR(copy__doc__,
1197"B.copy() -> bytearray\n\
1198\n\
1199Return a copy of B.");
1200
1201static PyObject *
1202bytearray_copy(PyByteArrayObject *self)
1203{
1204 return PyByteArray_FromStringAndSize(PyByteArray_AS_STRING((PyObject *)self),
1205 PyByteArray_GET_SIZE(self));
1206}
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001207
1208PyDoc_STRVAR(index__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001209"B.index(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001210\n\
1211Like B.find() but raise ValueError when the subsection is not found.");
1212
1213static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001214bytearray_index(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001215{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001216 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001217 if (result == -2)
1218 return NULL;
1219 if (result == -1) {
1220 PyErr_SetString(PyExc_ValueError,
1221 "subsection not found");
1222 return NULL;
1223 }
1224 return PyLong_FromSsize_t(result);
1225}
1226
1227
1228PyDoc_STRVAR(rfind__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001229"B.rfind(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001230\n\
1231Return the highest index in B where subsection sub is found,\n\
Senthil Kumaran53516a82011-07-27 23:33:54 +08001232such that sub is contained within B[start,end]. Optional\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001233arguments start and end are interpreted as in slice notation.\n\
1234\n\
1235Return -1 on failure.");
1236
1237static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001238bytearray_rfind(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001239{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001240 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001241 if (result == -2)
1242 return NULL;
1243 return PyLong_FromSsize_t(result);
1244}
1245
1246
1247PyDoc_STRVAR(rindex__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001248"B.rindex(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001249\n\
1250Like B.rfind() but raise ValueError when the subsection is not found.");
1251
1252static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001253bytearray_rindex(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001254{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001255 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001256 if (result == -2)
1257 return NULL;
1258 if (result == -1) {
1259 PyErr_SetString(PyExc_ValueError,
1260 "subsection not found");
1261 return NULL;
1262 }
1263 return PyLong_FromSsize_t(result);
1264}
1265
1266
1267static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001268bytearray_contains(PyObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001269{
1270 Py_ssize_t ival = PyNumber_AsSsize_t(arg, PyExc_ValueError);
1271 if (ival == -1 && PyErr_Occurred()) {
1272 Py_buffer varg;
Antoine Pitrou0010d372010-08-15 17:12:55 +00001273 Py_ssize_t pos;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001274 PyErr_Clear();
1275 if (_getbuffer(arg, &varg) < 0)
1276 return -1;
1277 pos = stringlib_find(PyByteArray_AS_STRING(self), Py_SIZE(self),
1278 varg.buf, varg.len, 0);
Martin v. Löwis423be952008-08-13 15:53:07 +00001279 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001280 return pos >= 0;
1281 }
1282 if (ival < 0 || ival >= 256) {
1283 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
1284 return -1;
1285 }
1286
Antoine Pitrou0010d372010-08-15 17:12:55 +00001287 return memchr(PyByteArray_AS_STRING(self), (int) ival, Py_SIZE(self)) != NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001288}
1289
1290
1291/* Matches the end (direction >= 0) or start (direction < 0) of self
1292 * against substr, using the start and end arguments. Returns
1293 * -1 on error, 0 if not found and 1 if found.
1294 */
1295Py_LOCAL(int)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001296_bytearray_tailmatch(PyByteArrayObject *self, PyObject *substr, Py_ssize_t start,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001297 Py_ssize_t end, int direction)
1298{
1299 Py_ssize_t len = PyByteArray_GET_SIZE(self);
1300 const char* str;
1301 Py_buffer vsubstr;
1302 int rv = 0;
1303
1304 str = PyByteArray_AS_STRING(self);
1305
1306 if (_getbuffer(substr, &vsubstr) < 0)
1307 return -1;
1308
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001309 ADJUST_INDICES(start, end, len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001310
1311 if (direction < 0) {
1312 /* startswith */
1313 if (start+vsubstr.len > len) {
1314 goto done;
1315 }
1316 } else {
1317 /* endswith */
1318 if (end-start < vsubstr.len || start > len) {
1319 goto done;
1320 }
1321
1322 if (end-vsubstr.len > start)
1323 start = end - vsubstr.len;
1324 }
1325 if (end-start >= vsubstr.len)
1326 rv = ! memcmp(str+start, vsubstr.buf, vsubstr.len);
1327
1328done:
Martin v. Löwis423be952008-08-13 15:53:07 +00001329 PyBuffer_Release(&vsubstr);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001330 return rv;
1331}
1332
1333
1334PyDoc_STRVAR(startswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001335"B.startswith(prefix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001336\n\
1337Return True if B starts with the specified prefix, False otherwise.\n\
1338With optional start, test B beginning at that position.\n\
1339With optional end, stop comparing B at that position.\n\
Ezio Melottiba42fd52011-04-26 06:09:45 +03001340prefix can also be a tuple of bytes to try.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001341
1342static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001343bytearray_startswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001344{
1345 Py_ssize_t start = 0;
1346 Py_ssize_t end = PY_SSIZE_T_MAX;
1347 PyObject *subobj;
1348 int result;
1349
Jesus Ceaac451502011-04-20 17:09:23 +02001350 if (!stringlib_parse_args_finds("startswith", args, &subobj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001351 return NULL;
1352 if (PyTuple_Check(subobj)) {
1353 Py_ssize_t i;
1354 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001355 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001356 PyTuple_GET_ITEM(subobj, i),
1357 start, end, -1);
1358 if (result == -1)
1359 return NULL;
1360 else if (result) {
1361 Py_RETURN_TRUE;
1362 }
1363 }
1364 Py_RETURN_FALSE;
1365 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001366 result = _bytearray_tailmatch(self, subobj, start, end, -1);
Ezio Melottiba42fd52011-04-26 06:09:45 +03001367 if (result == -1) {
1368 if (PyErr_ExceptionMatches(PyExc_TypeError))
1369 PyErr_Format(PyExc_TypeError, "startswith first arg must be bytes "
1370 "or a tuple of bytes, not %s", Py_TYPE(subobj)->tp_name);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001371 return NULL;
Ezio Melottiba42fd52011-04-26 06:09:45 +03001372 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001373 else
1374 return PyBool_FromLong(result);
1375}
1376
1377PyDoc_STRVAR(endswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001378"B.endswith(suffix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001379\n\
1380Return True if B ends with the specified suffix, False otherwise.\n\
1381With optional start, test B beginning at that position.\n\
1382With optional end, stop comparing B at that position.\n\
Ezio Melottiba42fd52011-04-26 06:09:45 +03001383suffix can also be a tuple of bytes to try.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001384
1385static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001386bytearray_endswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001387{
1388 Py_ssize_t start = 0;
1389 Py_ssize_t end = PY_SSIZE_T_MAX;
1390 PyObject *subobj;
1391 int result;
1392
Jesus Ceaac451502011-04-20 17:09:23 +02001393 if (!stringlib_parse_args_finds("endswith", args, &subobj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001394 return NULL;
1395 if (PyTuple_Check(subobj)) {
1396 Py_ssize_t i;
1397 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001398 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001399 PyTuple_GET_ITEM(subobj, i),
1400 start, end, +1);
1401 if (result == -1)
1402 return NULL;
1403 else if (result) {
1404 Py_RETURN_TRUE;
1405 }
1406 }
1407 Py_RETURN_FALSE;
1408 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001409 result = _bytearray_tailmatch(self, subobj, start, end, +1);
Ezio Melottiba42fd52011-04-26 06:09:45 +03001410 if (result == -1) {
1411 if (PyErr_ExceptionMatches(PyExc_TypeError))
1412 PyErr_Format(PyExc_TypeError, "endswith first arg must be bytes or "
1413 "a tuple of bytes, not %s", Py_TYPE(subobj)->tp_name);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001414 return NULL;
Ezio Melottiba42fd52011-04-26 06:09:45 +03001415 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001416 else
1417 return PyBool_FromLong(result);
1418}
1419
1420
1421PyDoc_STRVAR(translate__doc__,
1422"B.translate(table[, deletechars]) -> bytearray\n\
1423\n\
1424Return a copy of B, where all characters occurring in the\n\
1425optional argument deletechars are removed, and the remaining\n\
1426characters have been mapped through the given translation\n\
1427table, which must be a bytes object of length 256.");
1428
1429static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001430bytearray_translate(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001431{
1432 register char *input, *output;
1433 register const char *table;
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001434 register Py_ssize_t i, c;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001435 PyObject *input_obj = (PyObject*)self;
1436 const char *output_start;
1437 Py_ssize_t inlen;
Georg Brandlccc47b62008-12-28 11:44:14 +00001438 PyObject *result = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001439 int trans_table[256];
Georg Brandlccc47b62008-12-28 11:44:14 +00001440 PyObject *tableobj = NULL, *delobj = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001441 Py_buffer vtable, vdel;
1442
1443 if (!PyArg_UnpackTuple(args, "translate", 1, 2,
1444 &tableobj, &delobj))
1445 return NULL;
1446
Georg Brandlccc47b62008-12-28 11:44:14 +00001447 if (tableobj == Py_None) {
1448 table = NULL;
1449 tableobj = NULL;
1450 } else if (_getbuffer(tableobj, &vtable) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001451 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001452 } else {
1453 if (vtable.len != 256) {
1454 PyErr_SetString(PyExc_ValueError,
1455 "translation table must be 256 characters long");
Georg Brandl953152f2009-07-22 12:03:59 +00001456 PyBuffer_Release(&vtable);
1457 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001458 }
1459 table = (const char*)vtable.buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001460 }
1461
1462 if (delobj != NULL) {
1463 if (_getbuffer(delobj, &vdel) < 0) {
Georg Brandl953152f2009-07-22 12:03:59 +00001464 if (tableobj != NULL)
1465 PyBuffer_Release(&vtable);
1466 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001467 }
1468 }
1469 else {
1470 vdel.buf = NULL;
1471 vdel.len = 0;
1472 }
1473
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001474 inlen = PyByteArray_GET_SIZE(input_obj);
1475 result = PyByteArray_FromStringAndSize((char *)NULL, inlen);
1476 if (result == NULL)
1477 goto done;
1478 output_start = output = PyByteArray_AsString(result);
1479 input = PyByteArray_AS_STRING(input_obj);
1480
Georg Brandlccc47b62008-12-28 11:44:14 +00001481 if (vdel.len == 0 && table != NULL) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001482 /* If no deletions are required, use faster code */
1483 for (i = inlen; --i >= 0; ) {
1484 c = Py_CHARMASK(*input++);
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001485 *output++ = table[c];
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001486 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001487 goto done;
1488 }
Georg Brandlccc47b62008-12-28 11:44:14 +00001489
1490 if (table == NULL) {
1491 for (i = 0; i < 256; i++)
1492 trans_table[i] = Py_CHARMASK(i);
1493 } else {
1494 for (i = 0; i < 256; i++)
1495 trans_table[i] = Py_CHARMASK(table[i]);
1496 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001497
1498 for (i = 0; i < vdel.len; i++)
1499 trans_table[(int) Py_CHARMASK( ((unsigned char*)vdel.buf)[i] )] = -1;
1500
1501 for (i = inlen; --i >= 0; ) {
1502 c = Py_CHARMASK(*input++);
1503 if (trans_table[c] != -1)
1504 if (Py_CHARMASK(*output++ = (char)trans_table[c]) == c)
1505 continue;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001506 }
1507 /* Fix the size of the resulting string */
1508 if (inlen > 0)
1509 PyByteArray_Resize(result, output - output_start);
1510
1511done:
Georg Brandlccc47b62008-12-28 11:44:14 +00001512 if (tableobj != NULL)
1513 PyBuffer_Release(&vtable);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001514 if (delobj != NULL)
Martin v. Löwis423be952008-08-13 15:53:07 +00001515 PyBuffer_Release(&vdel);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001516 return result;
1517}
1518
1519
Georg Brandlabc38772009-04-12 15:51:51 +00001520static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001521bytearray_maketrans(PyObject *null, PyObject *args)
Georg Brandlabc38772009-04-12 15:51:51 +00001522{
Alexander Belopolskyf0f45142010-08-11 17:31:17 +00001523 return _Py_bytes_maketrans(args);
Georg Brandlabc38772009-04-12 15:51:51 +00001524}
1525
1526
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001527/* find and count characters and substrings */
1528
1529#define findchar(target, target_len, c) \
1530 ((char *)memchr((const void *)(target), c, target_len))
1531
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001532
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001533/* Bytes ops must return a string, create a copy */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001534Py_LOCAL(PyByteArrayObject *)
1535return_self(PyByteArrayObject *self)
1536{
Georg Brandl1e7217d2008-05-30 12:02:38 +00001537 /* always return a new bytearray */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001538 return (PyByteArrayObject *)PyByteArray_FromStringAndSize(
1539 PyByteArray_AS_STRING(self),
1540 PyByteArray_GET_SIZE(self));
1541}
1542
1543Py_LOCAL_INLINE(Py_ssize_t)
1544countchar(const char *target, Py_ssize_t target_len, char c, Py_ssize_t maxcount)
1545{
1546 Py_ssize_t count=0;
1547 const char *start=target;
1548 const char *end=target+target_len;
1549
1550 while ( (start=findchar(start, end-start, c)) != NULL ) {
1551 count++;
1552 if (count >= maxcount)
1553 break;
1554 start += 1;
1555 }
1556 return count;
1557}
1558
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001559
1560/* Algorithms for different cases of string replacement */
1561
1562/* len(self)>=1, from="", len(to)>=1, maxcount>=1 */
1563Py_LOCAL(PyByteArrayObject *)
1564replace_interleave(PyByteArrayObject *self,
1565 const char *to_s, Py_ssize_t to_len,
1566 Py_ssize_t maxcount)
1567{
1568 char *self_s, *result_s;
1569 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001570 Py_ssize_t count, i;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001571 PyByteArrayObject *result;
1572
1573 self_len = PyByteArray_GET_SIZE(self);
1574
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001575 /* 1 at the end plus 1 after every character;
1576 count = min(maxcount, self_len + 1) */
1577 if (maxcount <= self_len)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001578 count = maxcount;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001579 else
1580 /* Can't overflow: self_len + 1 <= maxcount <= PY_SSIZE_T_MAX. */
1581 count = self_len + 1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001582
1583 /* Check for overflow */
1584 /* result_len = count * to_len + self_len; */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001585 assert(count > 0);
1586 if (to_len > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001587 PyErr_SetString(PyExc_OverflowError,
1588 "replace string is too long");
1589 return NULL;
1590 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001591 result_len = count * to_len + self_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001592
1593 if (! (result = (PyByteArrayObject *)
1594 PyByteArray_FromStringAndSize(NULL, result_len)) )
1595 return NULL;
1596
1597 self_s = PyByteArray_AS_STRING(self);
1598 result_s = PyByteArray_AS_STRING(result);
1599
1600 /* TODO: special case single character, which doesn't need memcpy */
1601
1602 /* Lay the first one down (guaranteed this will occur) */
1603 Py_MEMCPY(result_s, to_s, to_len);
1604 result_s += to_len;
1605 count -= 1;
1606
1607 for (i=0; i<count; i++) {
1608 *result_s++ = *self_s++;
1609 Py_MEMCPY(result_s, to_s, to_len);
1610 result_s += to_len;
1611 }
1612
1613 /* Copy the rest of the original string */
1614 Py_MEMCPY(result_s, self_s, self_len-i);
1615
1616 return result;
1617}
1618
1619/* Special case for deleting a single character */
1620/* len(self)>=1, len(from)==1, to="", maxcount>=1 */
1621Py_LOCAL(PyByteArrayObject *)
1622replace_delete_single_character(PyByteArrayObject *self,
1623 char from_c, Py_ssize_t maxcount)
1624{
1625 char *self_s, *result_s;
1626 char *start, *next, *end;
1627 Py_ssize_t self_len, result_len;
1628 Py_ssize_t count;
1629 PyByteArrayObject *result;
1630
1631 self_len = PyByteArray_GET_SIZE(self);
1632 self_s = PyByteArray_AS_STRING(self);
1633
1634 count = countchar(self_s, self_len, from_c, maxcount);
1635 if (count == 0) {
1636 return return_self(self);
1637 }
1638
1639 result_len = self_len - count; /* from_len == 1 */
1640 assert(result_len>=0);
1641
1642 if ( (result = (PyByteArrayObject *)
1643 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1644 return NULL;
1645 result_s = PyByteArray_AS_STRING(result);
1646
1647 start = self_s;
1648 end = self_s + self_len;
1649 while (count-- > 0) {
1650 next = findchar(start, end-start, from_c);
1651 if (next == NULL)
1652 break;
1653 Py_MEMCPY(result_s, start, next-start);
1654 result_s += (next-start);
1655 start = next+1;
1656 }
1657 Py_MEMCPY(result_s, start, end-start);
1658
1659 return result;
1660}
1661
1662/* len(self)>=1, len(from)>=2, to="", maxcount>=1 */
1663
1664Py_LOCAL(PyByteArrayObject *)
1665replace_delete_substring(PyByteArrayObject *self,
1666 const char *from_s, Py_ssize_t from_len,
1667 Py_ssize_t maxcount)
1668{
1669 char *self_s, *result_s;
1670 char *start, *next, *end;
1671 Py_ssize_t self_len, result_len;
1672 Py_ssize_t count, offset;
1673 PyByteArrayObject *result;
1674
1675 self_len = PyByteArray_GET_SIZE(self);
1676 self_s = PyByteArray_AS_STRING(self);
1677
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001678 count = stringlib_count(self_s, self_len,
1679 from_s, from_len,
1680 maxcount);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001681
1682 if (count == 0) {
1683 /* no matches */
1684 return return_self(self);
1685 }
1686
1687 result_len = self_len - (count * from_len);
1688 assert (result_len>=0);
1689
1690 if ( (result = (PyByteArrayObject *)
1691 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL )
1692 return NULL;
1693
1694 result_s = PyByteArray_AS_STRING(result);
1695
1696 start = self_s;
1697 end = self_s + self_len;
1698 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001699 offset = stringlib_find(start, end-start,
1700 from_s, from_len,
1701 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001702 if (offset == -1)
1703 break;
1704 next = start + offset;
1705
1706 Py_MEMCPY(result_s, start, next-start);
1707
1708 result_s += (next-start);
1709 start = next+from_len;
1710 }
1711 Py_MEMCPY(result_s, start, end-start);
1712 return result;
1713}
1714
1715/* len(self)>=1, len(from)==len(to)==1, maxcount>=1 */
1716Py_LOCAL(PyByteArrayObject *)
1717replace_single_character_in_place(PyByteArrayObject *self,
1718 char from_c, char to_c,
1719 Py_ssize_t maxcount)
1720{
Antoine Pitroud1188562010-06-09 16:38:55 +00001721 char *self_s, *result_s, *start, *end, *next;
1722 Py_ssize_t self_len;
1723 PyByteArrayObject *result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001724
Antoine Pitroud1188562010-06-09 16:38:55 +00001725 /* The result string will be the same size */
1726 self_s = PyByteArray_AS_STRING(self);
1727 self_len = PyByteArray_GET_SIZE(self);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001728
Antoine Pitroud1188562010-06-09 16:38:55 +00001729 next = findchar(self_s, self_len, from_c);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001730
Antoine Pitroud1188562010-06-09 16:38:55 +00001731 if (next == NULL) {
1732 /* No matches; return the original bytes */
1733 return return_self(self);
1734 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001735
Antoine Pitroud1188562010-06-09 16:38:55 +00001736 /* Need to make a new bytes */
1737 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1738 if (result == NULL)
1739 return NULL;
1740 result_s = PyByteArray_AS_STRING(result);
1741 Py_MEMCPY(result_s, self_s, self_len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001742
Antoine Pitroud1188562010-06-09 16:38:55 +00001743 /* change everything in-place, starting with this one */
1744 start = result_s + (next-self_s);
1745 *start = to_c;
1746 start++;
1747 end = result_s + self_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001748
Antoine Pitroud1188562010-06-09 16:38:55 +00001749 while (--maxcount > 0) {
1750 next = findchar(start, end-start, from_c);
1751 if (next == NULL)
1752 break;
1753 *next = to_c;
1754 start = next+1;
1755 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001756
Antoine Pitroud1188562010-06-09 16:38:55 +00001757 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001758}
1759
1760/* len(self)>=1, len(from)==len(to)>=2, maxcount>=1 */
1761Py_LOCAL(PyByteArrayObject *)
1762replace_substring_in_place(PyByteArrayObject *self,
1763 const char *from_s, Py_ssize_t from_len,
1764 const char *to_s, Py_ssize_t to_len,
1765 Py_ssize_t maxcount)
1766{
1767 char *result_s, *start, *end;
1768 char *self_s;
1769 Py_ssize_t self_len, offset;
1770 PyByteArrayObject *result;
1771
1772 /* The result bytes will be the same size */
1773
1774 self_s = PyByteArray_AS_STRING(self);
1775 self_len = PyByteArray_GET_SIZE(self);
1776
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001777 offset = stringlib_find(self_s, self_len,
1778 from_s, from_len,
1779 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001780 if (offset == -1) {
1781 /* No matches; return the original bytes */
1782 return return_self(self);
1783 }
1784
1785 /* Need to make a new bytes */
1786 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1787 if (result == NULL)
1788 return NULL;
1789 result_s = PyByteArray_AS_STRING(result);
1790 Py_MEMCPY(result_s, self_s, self_len);
1791
1792 /* change everything in-place, starting with this one */
1793 start = result_s + offset;
1794 Py_MEMCPY(start, to_s, from_len);
1795 start += from_len;
1796 end = result_s + self_len;
1797
1798 while ( --maxcount > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001799 offset = stringlib_find(start, end-start,
1800 from_s, from_len,
1801 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001802 if (offset==-1)
1803 break;
1804 Py_MEMCPY(start+offset, to_s, from_len);
1805 start += offset+from_len;
1806 }
1807
1808 return result;
1809}
1810
1811/* len(self)>=1, len(from)==1, len(to)>=2, maxcount>=1 */
1812Py_LOCAL(PyByteArrayObject *)
1813replace_single_character(PyByteArrayObject *self,
1814 char from_c,
1815 const char *to_s, Py_ssize_t to_len,
1816 Py_ssize_t maxcount)
1817{
1818 char *self_s, *result_s;
1819 char *start, *next, *end;
1820 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001821 Py_ssize_t count;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001822 PyByteArrayObject *result;
1823
1824 self_s = PyByteArray_AS_STRING(self);
1825 self_len = PyByteArray_GET_SIZE(self);
1826
1827 count = countchar(self_s, self_len, from_c, maxcount);
1828 if (count == 0) {
1829 /* no matches, return unchanged */
1830 return return_self(self);
1831 }
1832
1833 /* use the difference between current and new, hence the "-1" */
1834 /* result_len = self_len + count * (to_len-1) */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001835 assert(count > 0);
1836 if (to_len - 1 > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001837 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1838 return NULL;
1839 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001840 result_len = self_len + count * (to_len - 1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001841
1842 if ( (result = (PyByteArrayObject *)
1843 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1844 return NULL;
1845 result_s = PyByteArray_AS_STRING(result);
1846
1847 start = self_s;
1848 end = self_s + self_len;
1849 while (count-- > 0) {
1850 next = findchar(start, end-start, from_c);
1851 if (next == NULL)
1852 break;
1853
1854 if (next == start) {
1855 /* replace with the 'to' */
1856 Py_MEMCPY(result_s, to_s, to_len);
1857 result_s += to_len;
1858 start += 1;
1859 } else {
1860 /* copy the unchanged old then the 'to' */
1861 Py_MEMCPY(result_s, start, next-start);
1862 result_s += (next-start);
1863 Py_MEMCPY(result_s, to_s, to_len);
1864 result_s += to_len;
1865 start = next+1;
1866 }
1867 }
1868 /* Copy the remainder of the remaining bytes */
1869 Py_MEMCPY(result_s, start, end-start);
1870
1871 return result;
1872}
1873
1874/* len(self)>=1, len(from)>=2, len(to)>=2, maxcount>=1 */
1875Py_LOCAL(PyByteArrayObject *)
1876replace_substring(PyByteArrayObject *self,
1877 const char *from_s, Py_ssize_t from_len,
1878 const char *to_s, Py_ssize_t to_len,
1879 Py_ssize_t maxcount)
1880{
1881 char *self_s, *result_s;
1882 char *start, *next, *end;
1883 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001884 Py_ssize_t count, offset;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001885 PyByteArrayObject *result;
1886
1887 self_s = PyByteArray_AS_STRING(self);
1888 self_len = PyByteArray_GET_SIZE(self);
1889
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001890 count = stringlib_count(self_s, self_len,
1891 from_s, from_len,
1892 maxcount);
1893
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001894 if (count == 0) {
1895 /* no matches, return unchanged */
1896 return return_self(self);
1897 }
1898
1899 /* Check for overflow */
1900 /* result_len = self_len + count * (to_len-from_len) */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001901 assert(count > 0);
1902 if (to_len - from_len > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001903 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1904 return NULL;
1905 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001906 result_len = self_len + count * (to_len - from_len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001907
1908 if ( (result = (PyByteArrayObject *)
1909 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1910 return NULL;
1911 result_s = PyByteArray_AS_STRING(result);
1912
1913 start = self_s;
1914 end = self_s + self_len;
1915 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001916 offset = stringlib_find(start, end-start,
1917 from_s, from_len,
1918 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001919 if (offset == -1)
1920 break;
1921 next = start+offset;
1922 if (next == start) {
1923 /* replace with the 'to' */
1924 Py_MEMCPY(result_s, to_s, to_len);
1925 result_s += to_len;
1926 start += from_len;
1927 } else {
1928 /* copy the unchanged old then the 'to' */
1929 Py_MEMCPY(result_s, start, next-start);
1930 result_s += (next-start);
1931 Py_MEMCPY(result_s, to_s, to_len);
1932 result_s += to_len;
1933 start = next+from_len;
1934 }
1935 }
1936 /* Copy the remainder of the remaining bytes */
1937 Py_MEMCPY(result_s, start, end-start);
1938
1939 return result;
1940}
1941
1942
1943Py_LOCAL(PyByteArrayObject *)
1944replace(PyByteArrayObject *self,
1945 const char *from_s, Py_ssize_t from_len,
1946 const char *to_s, Py_ssize_t to_len,
1947 Py_ssize_t maxcount)
1948{
1949 if (maxcount < 0) {
1950 maxcount = PY_SSIZE_T_MAX;
1951 } else if (maxcount == 0 || PyByteArray_GET_SIZE(self) == 0) {
1952 /* nothing to do; return the original bytes */
1953 return return_self(self);
1954 }
1955
1956 if (maxcount == 0 ||
1957 (from_len == 0 && to_len == 0)) {
1958 /* nothing to do; return the original bytes */
1959 return return_self(self);
1960 }
1961
1962 /* Handle zero-length special cases */
1963
1964 if (from_len == 0) {
1965 /* insert the 'to' bytes everywhere. */
1966 /* >>> "Python".replace("", ".") */
1967 /* '.P.y.t.h.o.n.' */
1968 return replace_interleave(self, to_s, to_len, maxcount);
1969 }
1970
1971 /* Except for "".replace("", "A") == "A" there is no way beyond this */
1972 /* point for an empty self bytes to generate a non-empty bytes */
1973 /* Special case so the remaining code always gets a non-empty bytes */
1974 if (PyByteArray_GET_SIZE(self) == 0) {
1975 return return_self(self);
1976 }
1977
1978 if (to_len == 0) {
Georg Brandl17cb8a82008-05-30 08:20:09 +00001979 /* delete all occurrences of 'from' bytes */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001980 if (from_len == 1) {
1981 return replace_delete_single_character(
1982 self, from_s[0], maxcount);
1983 } else {
1984 return replace_delete_substring(self, from_s, from_len, maxcount);
1985 }
1986 }
1987
1988 /* Handle special case where both bytes have the same length */
1989
1990 if (from_len == to_len) {
1991 if (from_len == 1) {
1992 return replace_single_character_in_place(
1993 self,
1994 from_s[0],
1995 to_s[0],
1996 maxcount);
1997 } else {
1998 return replace_substring_in_place(
1999 self, from_s, from_len, to_s, to_len, maxcount);
2000 }
2001 }
2002
2003 /* Otherwise use the more generic algorithms */
2004 if (from_len == 1) {
2005 return replace_single_character(self, from_s[0],
2006 to_s, to_len, maxcount);
2007 } else {
2008 /* len('from')>=2, len('to')>=1 */
2009 return replace_substring(self, from_s, from_len, to_s, to_len, maxcount);
2010 }
2011}
2012
2013
2014PyDoc_STRVAR(replace__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002015"B.replace(old, new[, count]) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002016\n\
2017Return a copy of B with all occurrences of subsection\n\
2018old replaced by new. If the optional argument count is\n\
2019given, only the first count occurrences are replaced.");
2020
2021static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002022bytearray_replace(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002023{
2024 Py_ssize_t count = -1;
2025 PyObject *from, *to, *res;
2026 Py_buffer vfrom, vto;
2027
2028 if (!PyArg_ParseTuple(args, "OO|n:replace", &from, &to, &count))
2029 return NULL;
2030
2031 if (_getbuffer(from, &vfrom) < 0)
2032 return NULL;
2033 if (_getbuffer(to, &vto) < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +00002034 PyBuffer_Release(&vfrom);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002035 return NULL;
2036 }
2037
2038 res = (PyObject *)replace((PyByteArrayObject *) self,
2039 vfrom.buf, vfrom.len,
2040 vto.buf, vto.len, count);
2041
Martin v. Löwis423be952008-08-13 15:53:07 +00002042 PyBuffer_Release(&vfrom);
2043 PyBuffer_Release(&vto);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002044 return res;
2045}
2046
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002047PyDoc_STRVAR(split__doc__,
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002048"B.split(sep=None, maxsplit=-1) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002049\n\
2050Return a list of the sections in B, using sep as the delimiter.\n\
2051If sep is not given, B is split on ASCII whitespace characters\n\
2052(space, tab, return, newline, formfeed, vertical tab).\n\
2053If maxsplit is given, at most maxsplit splits are done.");
2054
2055static PyObject *
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002056bytearray_split(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002057{
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002058 static char *kwlist[] = {"sep", "maxsplit", 0};
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002059 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2060 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002061 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002062 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002063 Py_buffer vsub;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002064
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002065 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|On:split",
2066 kwlist, &subobj, &maxsplit))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002067 return NULL;
2068 if (maxsplit < 0)
2069 maxsplit = PY_SSIZE_T_MAX;
2070
2071 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002072 return stringlib_split_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002073
2074 if (_getbuffer(subobj, &vsub) < 0)
2075 return NULL;
2076 sub = vsub.buf;
2077 n = vsub.len;
2078
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002079 list = stringlib_split(
2080 (PyObject*) self, s, len, sub, n, maxsplit
2081 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002082 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002083 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002084}
2085
2086PyDoc_STRVAR(partition__doc__,
2087"B.partition(sep) -> (head, sep, tail)\n\
2088\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002089Search for the separator sep in B, and return the part before it,\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002090the separator itself, and the part after it. If the separator is not\n\
2091found, returns B and two empty bytearray objects.");
2092
2093static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002094bytearray_partition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002095{
2096 PyObject *bytesep, *result;
2097
2098 bytesep = PyByteArray_FromObject(sep_obj);
2099 if (! bytesep)
2100 return NULL;
2101
2102 result = stringlib_partition(
2103 (PyObject*) self,
2104 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2105 bytesep,
2106 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2107 );
2108
2109 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002110 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002111}
2112
2113PyDoc_STRVAR(rpartition__doc__,
Ezio Melotti5b2b2422010-01-25 11:58:28 +00002114"B.rpartition(sep) -> (head, sep, tail)\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002115\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002116Search for the separator sep in B, starting at the end of B,\n\
2117and return the part before it, the separator itself, and the\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002118part after it. If the separator is not found, returns two empty\n\
2119bytearray objects and B.");
2120
2121static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002122bytearray_rpartition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002123{
2124 PyObject *bytesep, *result;
2125
2126 bytesep = PyByteArray_FromObject(sep_obj);
2127 if (! bytesep)
2128 return NULL;
2129
2130 result = stringlib_rpartition(
2131 (PyObject*) self,
2132 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2133 bytesep,
2134 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2135 );
2136
2137 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002138 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002139}
2140
2141PyDoc_STRVAR(rsplit__doc__,
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002142"B.rsplit(sep=None, maxsplit=-1) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002143\n\
2144Return a list of the sections in B, using sep as the delimiter,\n\
2145starting at the end of B and working to the front.\n\
2146If sep is not given, B is split on ASCII whitespace characters\n\
2147(space, tab, return, newline, formfeed, vertical tab).\n\
2148If maxsplit is given, at most maxsplit splits are done.");
2149
2150static PyObject *
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002151bytearray_rsplit(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002152{
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002153 static char *kwlist[] = {"sep", "maxsplit", 0};
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002154 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2155 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002156 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002157 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002158 Py_buffer vsub;
2159
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002160 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|On:rsplit",
2161 kwlist, &subobj, &maxsplit))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002162 return NULL;
2163 if (maxsplit < 0)
2164 maxsplit = PY_SSIZE_T_MAX;
2165
2166 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002167 return stringlib_rsplit_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002168
2169 if (_getbuffer(subobj, &vsub) < 0)
2170 return NULL;
2171 sub = vsub.buf;
2172 n = vsub.len;
2173
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002174 list = stringlib_rsplit(
2175 (PyObject*) self, s, len, sub, n, maxsplit
2176 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002177 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002178 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002179}
2180
2181PyDoc_STRVAR(reverse__doc__,
2182"B.reverse() -> None\n\
2183\n\
2184Reverse the order of the values in B in place.");
2185static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002186bytearray_reverse(PyByteArrayObject *self, PyObject *unused)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002187{
2188 char swap, *head, *tail;
2189 Py_ssize_t i, j, n = Py_SIZE(self);
2190
2191 j = n / 2;
2192 head = self->ob_bytes;
2193 tail = head + n - 1;
2194 for (i = 0; i < j; i++) {
2195 swap = *head;
2196 *head++ = *tail;
2197 *tail-- = swap;
2198 }
2199
2200 Py_RETURN_NONE;
2201}
2202
2203PyDoc_STRVAR(insert__doc__,
2204"B.insert(index, int) -> None\n\
2205\n\
2206Insert a single item into the bytearray before the given index.");
2207static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002208bytearray_insert(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002209{
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002210 PyObject *value;
2211 int ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002212 Py_ssize_t where, n = Py_SIZE(self);
2213
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002214 if (!PyArg_ParseTuple(args, "nO:insert", &where, &value))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002215 return NULL;
2216
2217 if (n == PY_SSIZE_T_MAX) {
2218 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002219 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002220 return NULL;
2221 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002222 if (!_getbytevalue(value, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002223 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002224 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2225 return NULL;
2226
2227 if (where < 0) {
2228 where += n;
2229 if (where < 0)
2230 where = 0;
2231 }
2232 if (where > n)
2233 where = n;
2234 memmove(self->ob_bytes + where + 1, self->ob_bytes + where, n - where);
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002235 self->ob_bytes[where] = ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002236
2237 Py_RETURN_NONE;
2238}
2239
2240PyDoc_STRVAR(append__doc__,
2241"B.append(int) -> None\n\
2242\n\
2243Append a single item to the end of B.");
2244static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002245bytearray_append(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002246{
2247 int value;
2248 Py_ssize_t n = Py_SIZE(self);
2249
2250 if (! _getbytevalue(arg, &value))
2251 return NULL;
2252 if (n == PY_SSIZE_T_MAX) {
2253 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002254 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002255 return NULL;
2256 }
2257 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2258 return NULL;
2259
2260 self->ob_bytes[n] = value;
2261
2262 Py_RETURN_NONE;
2263}
2264
2265PyDoc_STRVAR(extend__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002266"B.extend(iterable_of_ints) -> None\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002267\n\
2268Append all the elements from the iterator or sequence to the\n\
2269end of B.");
2270static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002271bytearray_extend(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002272{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002273 PyObject *it, *item, *bytearray_obj;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002274 Py_ssize_t buf_size = 0, len = 0;
2275 int value;
2276 char *buf;
2277
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002278 /* bytearray_setslice code only accepts something supporting PEP 3118. */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002279 if (PyObject_CheckBuffer(arg)) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002280 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), arg) == -1)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002281 return NULL;
2282
2283 Py_RETURN_NONE;
2284 }
2285
2286 it = PyObject_GetIter(arg);
2287 if (it == NULL)
2288 return NULL;
2289
Ezio Melotti42da6632011-03-15 05:18:48 +02002290 /* Try to determine the length of the argument. 32 is arbitrary. */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002291 buf_size = _PyObject_LengthHint(arg, 32);
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002292 if (buf_size == -1) {
2293 Py_DECREF(it);
2294 return NULL;
2295 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002296
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002297 bytearray_obj = PyByteArray_FromStringAndSize(NULL, buf_size);
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002298 if (bytearray_obj == NULL) {
2299 Py_DECREF(it);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002300 return NULL;
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002301 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002302 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002303
2304 while ((item = PyIter_Next(it)) != NULL) {
2305 if (! _getbytevalue(item, &value)) {
2306 Py_DECREF(item);
2307 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002308 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002309 return NULL;
2310 }
2311 buf[len++] = value;
2312 Py_DECREF(item);
2313
2314 if (len >= buf_size) {
2315 buf_size = len + (len >> 1) + 1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002316 if (PyByteArray_Resize((PyObject *)bytearray_obj, buf_size) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002317 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002318 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002319 return NULL;
2320 }
2321 /* Recompute the `buf' pointer, since the resizing operation may
2322 have invalidated it. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002323 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002324 }
2325 }
2326 Py_DECREF(it);
2327
2328 /* Resize down to exact size. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002329 if (PyByteArray_Resize((PyObject *)bytearray_obj, len) < 0) {
2330 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002331 return NULL;
2332 }
2333
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002334 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), bytearray_obj) == -1) {
2335 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002336 return NULL;
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002337 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002338 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002339
2340 Py_RETURN_NONE;
2341}
2342
2343PyDoc_STRVAR(pop__doc__,
2344"B.pop([index]) -> int\n\
2345\n\
2346Remove and return a single item from B. If no index\n\
Benjamin Petersondcf97b92008-07-02 17:30:14 +00002347argument is given, will pop the last value.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002348static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002349bytearray_pop(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002350{
2351 int value;
2352 Py_ssize_t where = -1, n = Py_SIZE(self);
2353
2354 if (!PyArg_ParseTuple(args, "|n:pop", &where))
2355 return NULL;
2356
2357 if (n == 0) {
Eli Bendersky1bc4f192011-03-04 04:55:25 +00002358 PyErr_SetString(PyExc_IndexError,
2359 "pop from empty bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002360 return NULL;
2361 }
2362 if (where < 0)
2363 where += Py_SIZE(self);
2364 if (where < 0 || where >= Py_SIZE(self)) {
2365 PyErr_SetString(PyExc_IndexError, "pop index out of range");
2366 return NULL;
2367 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002368 if (!_canresize(self))
2369 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002370
2371 value = self->ob_bytes[where];
2372 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2373 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2374 return NULL;
2375
Mark Dickinson54a3db92009-09-06 10:19:23 +00002376 return PyLong_FromLong((unsigned char)value);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002377}
2378
2379PyDoc_STRVAR(remove__doc__,
2380"B.remove(int) -> None\n\
2381\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002382Remove the first occurrence of a value in B.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002383static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002384bytearray_remove(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002385{
2386 int value;
2387 Py_ssize_t where, n = Py_SIZE(self);
2388
2389 if (! _getbytevalue(arg, &value))
2390 return NULL;
2391
2392 for (where = 0; where < n; where++) {
2393 if (self->ob_bytes[where] == value)
2394 break;
2395 }
2396 if (where == n) {
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002397 PyErr_SetString(PyExc_ValueError, "value not found in bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002398 return NULL;
2399 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002400 if (!_canresize(self))
2401 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002402
2403 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2404 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2405 return NULL;
2406
2407 Py_RETURN_NONE;
2408}
2409
2410/* XXX These two helpers could be optimized if argsize == 1 */
2411
2412static Py_ssize_t
2413lstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2414 void *argptr, Py_ssize_t argsize)
2415{
2416 Py_ssize_t i = 0;
2417 while (i < mysize && memchr(argptr, myptr[i], argsize))
2418 i++;
2419 return i;
2420}
2421
2422static Py_ssize_t
2423rstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2424 void *argptr, Py_ssize_t argsize)
2425{
2426 Py_ssize_t i = mysize - 1;
2427 while (i >= 0 && memchr(argptr, myptr[i], argsize))
2428 i--;
2429 return i + 1;
2430}
2431
2432PyDoc_STRVAR(strip__doc__,
2433"B.strip([bytes]) -> bytearray\n\
2434\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002435Strip leading and trailing bytes contained in the argument\n\
2436and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002437If the argument is omitted, strip ASCII whitespace.");
2438static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002439bytearray_strip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002440{
2441 Py_ssize_t left, right, mysize, argsize;
2442 void *myptr, *argptr;
2443 PyObject *arg = Py_None;
2444 Py_buffer varg;
2445 if (!PyArg_ParseTuple(args, "|O:strip", &arg))
2446 return NULL;
2447 if (arg == Py_None) {
2448 argptr = "\t\n\r\f\v ";
2449 argsize = 6;
2450 }
2451 else {
2452 if (_getbuffer(arg, &varg) < 0)
2453 return NULL;
2454 argptr = varg.buf;
2455 argsize = varg.len;
2456 }
2457 myptr = self->ob_bytes;
2458 mysize = Py_SIZE(self);
2459 left = lstrip_helper(myptr, mysize, argptr, argsize);
2460 if (left == mysize)
2461 right = left;
2462 else
2463 right = rstrip_helper(myptr, mysize, argptr, argsize);
2464 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002465 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002466 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2467}
2468
2469PyDoc_STRVAR(lstrip__doc__,
2470"B.lstrip([bytes]) -> bytearray\n\
2471\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002472Strip leading bytes contained in the argument\n\
2473and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002474If the argument is omitted, strip leading ASCII whitespace.");
2475static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002476bytearray_lstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002477{
2478 Py_ssize_t left, right, mysize, argsize;
2479 void *myptr, *argptr;
2480 PyObject *arg = Py_None;
2481 Py_buffer varg;
2482 if (!PyArg_ParseTuple(args, "|O:lstrip", &arg))
2483 return NULL;
2484 if (arg == Py_None) {
2485 argptr = "\t\n\r\f\v ";
2486 argsize = 6;
2487 }
2488 else {
2489 if (_getbuffer(arg, &varg) < 0)
2490 return NULL;
2491 argptr = varg.buf;
2492 argsize = varg.len;
2493 }
2494 myptr = self->ob_bytes;
2495 mysize = Py_SIZE(self);
2496 left = lstrip_helper(myptr, mysize, argptr, argsize);
2497 right = mysize;
2498 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002499 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002500 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2501}
2502
2503PyDoc_STRVAR(rstrip__doc__,
2504"B.rstrip([bytes]) -> bytearray\n\
2505\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002506Strip trailing bytes contained in the argument\n\
2507and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002508If the argument is omitted, strip trailing ASCII whitespace.");
2509static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002510bytearray_rstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002511{
Brett Cannonb94767f2011-02-22 20:15:44 +00002512 Py_ssize_t right, mysize, argsize;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002513 void *myptr, *argptr;
2514 PyObject *arg = Py_None;
2515 Py_buffer varg;
2516 if (!PyArg_ParseTuple(args, "|O:rstrip", &arg))
2517 return NULL;
2518 if (arg == Py_None) {
2519 argptr = "\t\n\r\f\v ";
2520 argsize = 6;
2521 }
2522 else {
2523 if (_getbuffer(arg, &varg) < 0)
2524 return NULL;
2525 argptr = varg.buf;
2526 argsize = varg.len;
2527 }
2528 myptr = self->ob_bytes;
2529 mysize = Py_SIZE(self);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002530 right = rstrip_helper(myptr, mysize, argptr, argsize);
2531 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002532 PyBuffer_Release(&varg);
Brett Cannonb94767f2011-02-22 20:15:44 +00002533 return PyByteArray_FromStringAndSize(self->ob_bytes, right);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002534}
2535
2536PyDoc_STRVAR(decode_doc,
Victor Stinnerc911bbf2010-11-07 19:04:46 +00002537"B.decode(encoding='utf-8', errors='strict') -> str\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002538\n\
Victor Stinnere14e2122010-11-07 18:41:46 +00002539Decode B using the codec registered for encoding. Default encoding\n\
2540is 'utf-8'. errors may be given to set a different error\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002541handling scheme. Default is 'strict' meaning that encoding errors raise\n\
2542a UnicodeDecodeError. Other possible values are 'ignore' and 'replace'\n\
2543as well as any other name registered with codecs.register_error that is\n\
2544able to handle UnicodeDecodeErrors.");
2545
2546static PyObject *
Benjamin Peterson308d6372009-09-18 21:42:35 +00002547bytearray_decode(PyObject *self, PyObject *args, PyObject *kwargs)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002548{
2549 const char *encoding = NULL;
2550 const char *errors = NULL;
Benjamin Peterson308d6372009-09-18 21:42:35 +00002551 static char *kwlist[] = {"encoding", "errors", 0};
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002552
Benjamin Peterson308d6372009-09-18 21:42:35 +00002553 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ss:decode", kwlist, &encoding, &errors))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002554 return NULL;
2555 if (encoding == NULL)
2556 encoding = PyUnicode_GetDefaultEncoding();
Marc-André Lemburgb2750b52008-06-06 12:18:17 +00002557 return PyUnicode_FromEncodedObject(self, encoding, errors);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002558}
2559
2560PyDoc_STRVAR(alloc_doc,
2561"B.__alloc__() -> int\n\
2562\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002563Return the number of bytes actually allocated.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002564
2565static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002566bytearray_alloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002567{
2568 return PyLong_FromSsize_t(self->ob_alloc);
2569}
2570
2571PyDoc_STRVAR(join_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002572"B.join(iterable_of_bytes) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002573\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002574Concatenate any number of bytes/bytearray objects, with B\n\
2575in between each pair, and return the result as a new bytearray.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002576
2577static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002578bytearray_join(PyByteArrayObject *self, PyObject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002579{
2580 PyObject *seq;
2581 Py_ssize_t mysize = Py_SIZE(self);
2582 Py_ssize_t i;
2583 Py_ssize_t n;
2584 PyObject **items;
2585 Py_ssize_t totalsize = 0;
2586 PyObject *result;
2587 char *dest;
2588
2589 seq = PySequence_Fast(it, "can only join an iterable");
2590 if (seq == NULL)
2591 return NULL;
2592 n = PySequence_Fast_GET_SIZE(seq);
2593 items = PySequence_Fast_ITEMS(seq);
2594
2595 /* Compute the total size, and check that they are all bytes */
2596 /* XXX Shouldn't we use _getbuffer() on these items instead? */
2597 for (i = 0; i < n; i++) {
2598 PyObject *obj = items[i];
2599 if (!PyByteArray_Check(obj) && !PyBytes_Check(obj)) {
2600 PyErr_Format(PyExc_TypeError,
2601 "can only join an iterable of bytes "
2602 "(item %ld has type '%.100s')",
2603 /* XXX %ld isn't right on Win64 */
2604 (long)i, Py_TYPE(obj)->tp_name);
2605 goto error;
2606 }
2607 if (i > 0)
2608 totalsize += mysize;
2609 totalsize += Py_SIZE(obj);
2610 if (totalsize < 0) {
2611 PyErr_NoMemory();
2612 goto error;
2613 }
2614 }
2615
2616 /* Allocate the result, and copy the bytes */
2617 result = PyByteArray_FromStringAndSize(NULL, totalsize);
2618 if (result == NULL)
2619 goto error;
2620 dest = PyByteArray_AS_STRING(result);
2621 for (i = 0; i < n; i++) {
2622 PyObject *obj = items[i];
2623 Py_ssize_t size = Py_SIZE(obj);
2624 char *buf;
2625 if (PyByteArray_Check(obj))
2626 buf = PyByteArray_AS_STRING(obj);
2627 else
2628 buf = PyBytes_AS_STRING(obj);
2629 if (i) {
2630 memcpy(dest, self->ob_bytes, mysize);
2631 dest += mysize;
2632 }
2633 memcpy(dest, buf, size);
2634 dest += size;
2635 }
2636
2637 /* Done */
2638 Py_DECREF(seq);
2639 return result;
2640
2641 /* Error handling */
2642 error:
2643 Py_DECREF(seq);
2644 return NULL;
2645}
2646
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002647PyDoc_STRVAR(splitlines__doc__,
2648"B.splitlines([keepends]) -> list of lines\n\
2649\n\
2650Return a list of the lines in B, breaking at line boundaries.\n\
2651Line breaks are not included in the resulting list unless keepends\n\
2652is given and true.");
2653
2654static PyObject*
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002655bytearray_splitlines(PyObject *self, PyObject *args, PyObject *kwds)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002656{
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002657 static char *kwlist[] = {"keepends", 0};
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002658 int keepends = 0;
2659
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002660 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|i:splitlines",
2661 kwlist, &keepends))
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002662 return NULL;
2663
2664 return stringlib_splitlines(
2665 (PyObject*) self, PyByteArray_AS_STRING(self),
2666 PyByteArray_GET_SIZE(self), keepends
2667 );
2668}
2669
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002670PyDoc_STRVAR(fromhex_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002671"bytearray.fromhex(string) -> bytearray (static method)\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002672\n\
2673Create a bytearray object from a string of hexadecimal numbers.\n\
2674Spaces between two numbers are accepted.\n\
2675Example: bytearray.fromhex('B9 01EF') -> bytearray(b'\\xb9\\x01\\xef').");
2676
2677static int
Victor Stinner6430fd52011-09-29 04:02:13 +02002678hex_digit_to_int(Py_UCS4 c)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002679{
2680 if (c >= 128)
2681 return -1;
Eric Smith6dc46f52009-04-27 20:39:49 +00002682 if (Py_ISDIGIT(c))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002683 return c - '0';
2684 else {
Eric Smith6dc46f52009-04-27 20:39:49 +00002685 if (Py_ISUPPER(c))
2686 c = Py_TOLOWER(c);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002687 if (c >= 'a' && c <= 'f')
2688 return c - 'a' + 10;
2689 }
2690 return -1;
2691}
2692
2693static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002694bytearray_fromhex(PyObject *cls, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002695{
2696 PyObject *newbytes, *hexobj;
2697 char *buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002698 Py_ssize_t hexlen, byteslen, i, j;
2699 int top, bot;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002700 void *data;
2701 unsigned int kind;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002702
2703 if (!PyArg_ParseTuple(args, "U:fromhex", &hexobj))
2704 return NULL;
2705 assert(PyUnicode_Check(hexobj));
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002706 if (PyUnicode_READY(hexobj))
2707 return NULL;
2708 kind = PyUnicode_KIND(hexobj);
2709 data = PyUnicode_DATA(hexobj);
2710 hexlen = PyUnicode_GET_LENGTH(hexobj);
2711
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002712 byteslen = hexlen/2; /* This overestimates if there are spaces */
2713 newbytes = PyByteArray_FromStringAndSize(NULL, byteslen);
2714 if (!newbytes)
2715 return NULL;
2716 buf = PyByteArray_AS_STRING(newbytes);
2717 for (i = j = 0; i < hexlen; i += 2) {
2718 /* skip over spaces in the input */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002719 while (PyUnicode_READ(kind, data, i) == ' ')
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002720 i++;
2721 if (i >= hexlen)
2722 break;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002723 top = hex_digit_to_int(PyUnicode_READ(kind, data, i));
2724 bot = hex_digit_to_int(PyUnicode_READ(kind, data, i+1));
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002725 if (top == -1 || bot == -1) {
2726 PyErr_Format(PyExc_ValueError,
2727 "non-hexadecimal number found in "
2728 "fromhex() arg at position %zd", i);
2729 goto error;
2730 }
2731 buf[j++] = (top << 4) + bot;
2732 }
2733 if (PyByteArray_Resize(newbytes, j) < 0)
2734 goto error;
2735 return newbytes;
2736
2737 error:
2738 Py_DECREF(newbytes);
2739 return NULL;
2740}
2741
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002742
2743static PyObject *
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002744_common_reduce(PyByteArrayObject *self, int proto)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002745{
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002746 PyObject *dict;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02002747 _Py_IDENTIFIER(__dict__);
Martin v. Löwis1ee1b6f2011-10-10 18:11:30 +02002748
Martin v. Löwis1ee1b6f2011-10-10 18:11:30 +02002749 dict = _PyObject_GetAttrId((PyObject *)self, &PyId___dict__);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002750 if (dict == NULL) {
2751 PyErr_Clear();
2752 dict = Py_None;
2753 Py_INCREF(dict);
2754 }
2755
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002756 if (proto < 3) {
2757 /* use str based reduction for backwards compatibility with Python 2.x */
2758 PyObject *latin1;
2759 if (self->ob_bytes)
2760 latin1 = PyUnicode_DecodeLatin1(self->ob_bytes, Py_SIZE(self), NULL);
2761 else
2762 latin1 = PyUnicode_FromString("");
2763 return Py_BuildValue("(O(Ns)N)", Py_TYPE(self), latin1, "latin-1", dict);
2764 }
2765 else {
2766 /* use more efficient byte based reduction */
2767 if (self->ob_bytes) {
2768 return Py_BuildValue("(O(y#)N)", Py_TYPE(self), self->ob_bytes, Py_SIZE(self), dict);
2769 }
2770 else {
2771 return Py_BuildValue("(O()N)", Py_TYPE(self), dict);
2772 }
2773 }
2774}
2775
2776PyDoc_STRVAR(reduce_doc, "Return state information for pickling.");
2777
2778static PyObject *
2779bytearray_reduce(PyByteArrayObject *self)
2780{
2781 return _common_reduce(self, 2);
2782}
2783
2784PyDoc_STRVAR(reduce_ex_doc, "Return state information for pickling.");
2785
2786static PyObject *
2787bytearray_reduce_ex(PyByteArrayObject *self, PyObject *args)
2788{
2789 int proto = 0;
2790
2791 if (!PyArg_ParseTuple(args, "|i:__reduce_ex__", &proto))
2792 return NULL;
2793
2794 return _common_reduce(self, proto);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002795}
2796
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002797PyDoc_STRVAR(sizeof_doc,
2798"B.__sizeof__() -> int\n\
2799 \n\
2800Returns the size of B in memory, in bytes");
2801static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002802bytearray_sizeof(PyByteArrayObject *self)
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002803{
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002804 Py_ssize_t res;
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002805
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002806 res = sizeof(PyByteArrayObject) + self->ob_alloc * sizeof(char);
2807 return PyLong_FromSsize_t(res);
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002808}
2809
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002810static PySequenceMethods bytearray_as_sequence = {
2811 (lenfunc)bytearray_length, /* sq_length */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002812 (binaryfunc)PyByteArray_Concat, /* sq_concat */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002813 (ssizeargfunc)bytearray_repeat, /* sq_repeat */
2814 (ssizeargfunc)bytearray_getitem, /* sq_item */
2815 0, /* sq_slice */
2816 (ssizeobjargproc)bytearray_setitem, /* sq_ass_item */
2817 0, /* sq_ass_slice */
2818 (objobjproc)bytearray_contains, /* sq_contains */
2819 (binaryfunc)bytearray_iconcat, /* sq_inplace_concat */
2820 (ssizeargfunc)bytearray_irepeat, /* sq_inplace_repeat */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002821};
2822
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002823static PyMappingMethods bytearray_as_mapping = {
2824 (lenfunc)bytearray_length,
2825 (binaryfunc)bytearray_subscript,
2826 (objobjargproc)bytearray_ass_subscript,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002827};
2828
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002829static PyBufferProcs bytearray_as_buffer = {
2830 (getbufferproc)bytearray_getbuffer,
2831 (releasebufferproc)bytearray_releasebuffer,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002832};
2833
2834static PyMethodDef
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002835bytearray_methods[] = {
2836 {"__alloc__", (PyCFunction)bytearray_alloc, METH_NOARGS, alloc_doc},
2837 {"__reduce__", (PyCFunction)bytearray_reduce, METH_NOARGS, reduce_doc},
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002838 {"__reduce_ex__", (PyCFunction)bytearray_reduce_ex, METH_VARARGS, reduce_ex_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002839 {"__sizeof__", (PyCFunction)bytearray_sizeof, METH_NOARGS, sizeof_doc},
2840 {"append", (PyCFunction)bytearray_append, METH_O, append__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002841 {"capitalize", (PyCFunction)stringlib_capitalize, METH_NOARGS,
2842 _Py_capitalize__doc__},
2843 {"center", (PyCFunction)stringlib_center, METH_VARARGS, center__doc__},
Eli Bendersky4db28d32011-03-03 18:21:02 +00002844 {"clear", (PyCFunction)bytearray_clear, METH_NOARGS, clear__doc__},
2845 {"copy", (PyCFunction)bytearray_copy, METH_NOARGS, copy__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002846 {"count", (PyCFunction)bytearray_count, METH_VARARGS, count__doc__},
Benjamin Peterson308d6372009-09-18 21:42:35 +00002847 {"decode", (PyCFunction)bytearray_decode, METH_VARARGS | METH_KEYWORDS, decode_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002848 {"endswith", (PyCFunction)bytearray_endswith, METH_VARARGS, endswith__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002849 {"expandtabs", (PyCFunction)stringlib_expandtabs, METH_VARARGS,
2850 expandtabs__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002851 {"extend", (PyCFunction)bytearray_extend, METH_O, extend__doc__},
2852 {"find", (PyCFunction)bytearray_find, METH_VARARGS, find__doc__},
2853 {"fromhex", (PyCFunction)bytearray_fromhex, METH_VARARGS|METH_CLASS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002854 fromhex_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002855 {"index", (PyCFunction)bytearray_index, METH_VARARGS, index__doc__},
2856 {"insert", (PyCFunction)bytearray_insert, METH_VARARGS, insert__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002857 {"isalnum", (PyCFunction)stringlib_isalnum, METH_NOARGS,
2858 _Py_isalnum__doc__},
2859 {"isalpha", (PyCFunction)stringlib_isalpha, METH_NOARGS,
2860 _Py_isalpha__doc__},
2861 {"isdigit", (PyCFunction)stringlib_isdigit, METH_NOARGS,
2862 _Py_isdigit__doc__},
2863 {"islower", (PyCFunction)stringlib_islower, METH_NOARGS,
2864 _Py_islower__doc__},
2865 {"isspace", (PyCFunction)stringlib_isspace, METH_NOARGS,
2866 _Py_isspace__doc__},
2867 {"istitle", (PyCFunction)stringlib_istitle, METH_NOARGS,
2868 _Py_istitle__doc__},
2869 {"isupper", (PyCFunction)stringlib_isupper, METH_NOARGS,
2870 _Py_isupper__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002871 {"join", (PyCFunction)bytearray_join, METH_O, join_doc},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002872 {"ljust", (PyCFunction)stringlib_ljust, METH_VARARGS, ljust__doc__},
2873 {"lower", (PyCFunction)stringlib_lower, METH_NOARGS, _Py_lower__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002874 {"lstrip", (PyCFunction)bytearray_lstrip, METH_VARARGS, lstrip__doc__},
2875 {"maketrans", (PyCFunction)bytearray_maketrans, METH_VARARGS|METH_STATIC,
Georg Brandlabc38772009-04-12 15:51:51 +00002876 _Py_maketrans__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002877 {"partition", (PyCFunction)bytearray_partition, METH_O, partition__doc__},
2878 {"pop", (PyCFunction)bytearray_pop, METH_VARARGS, pop__doc__},
2879 {"remove", (PyCFunction)bytearray_remove, METH_O, remove__doc__},
2880 {"replace", (PyCFunction)bytearray_replace, METH_VARARGS, replace__doc__},
2881 {"reverse", (PyCFunction)bytearray_reverse, METH_NOARGS, reverse__doc__},
2882 {"rfind", (PyCFunction)bytearray_rfind, METH_VARARGS, rfind__doc__},
2883 {"rindex", (PyCFunction)bytearray_rindex, METH_VARARGS, rindex__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002884 {"rjust", (PyCFunction)stringlib_rjust, METH_VARARGS, rjust__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002885 {"rpartition", (PyCFunction)bytearray_rpartition, METH_O, rpartition__doc__},
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002886 {"rsplit", (PyCFunction)bytearray_rsplit, METH_VARARGS | METH_KEYWORDS, rsplit__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002887 {"rstrip", (PyCFunction)bytearray_rstrip, METH_VARARGS, rstrip__doc__},
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002888 {"split", (PyCFunction)bytearray_split, METH_VARARGS | METH_KEYWORDS, split__doc__},
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002889 {"splitlines", (PyCFunction)bytearray_splitlines,
2890 METH_VARARGS | METH_KEYWORDS, splitlines__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002891 {"startswith", (PyCFunction)bytearray_startswith, METH_VARARGS ,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002892 startswith__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002893 {"strip", (PyCFunction)bytearray_strip, METH_VARARGS, strip__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002894 {"swapcase", (PyCFunction)stringlib_swapcase, METH_NOARGS,
2895 _Py_swapcase__doc__},
2896 {"title", (PyCFunction)stringlib_title, METH_NOARGS, _Py_title__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002897 {"translate", (PyCFunction)bytearray_translate, METH_VARARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002898 translate__doc__},
2899 {"upper", (PyCFunction)stringlib_upper, METH_NOARGS, _Py_upper__doc__},
2900 {"zfill", (PyCFunction)stringlib_zfill, METH_VARARGS, zfill__doc__},
2901 {NULL}
2902};
2903
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002904PyDoc_STRVAR(bytearray_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002905"bytearray(iterable_of_ints) -> bytearray\n\
2906bytearray(string, encoding[, errors]) -> bytearray\n\
Victor Stinnerbb2e9c42011-12-17 23:18:07 +01002907bytearray(bytes_or_buffer) -> mutable copy of bytes_or_buffer\n\
2908bytearray(int) -> bytes array of size given by the parameter initialized with null bytes\n\
2909bytearray() -> empty bytes array\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002910\n\
2911Construct an mutable bytearray object from:\n\
2912 - an iterable yielding integers in range(256)\n\
2913 - a text string encoded using the specified encoding\n\
Victor Stinnerbb2e9c42011-12-17 23:18:07 +01002914 - a bytes or a buffer object\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002915 - any object implementing the buffer API.\n\
Victor Stinnerbb2e9c42011-12-17 23:18:07 +01002916 - an integer");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002917
2918
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002919static PyObject *bytearray_iter(PyObject *seq);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002920
2921PyTypeObject PyByteArray_Type = {
2922 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2923 "bytearray",
2924 sizeof(PyByteArrayObject),
2925 0,
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002926 (destructor)bytearray_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002927 0, /* tp_print */
2928 0, /* tp_getattr */
2929 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00002930 0, /* tp_reserved */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002931 (reprfunc)bytearray_repr, /* tp_repr */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002932 0, /* tp_as_number */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002933 &bytearray_as_sequence, /* tp_as_sequence */
2934 &bytearray_as_mapping, /* tp_as_mapping */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002935 0, /* tp_hash */
2936 0, /* tp_call */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002937 bytearray_str, /* tp_str */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002938 PyObject_GenericGetAttr, /* tp_getattro */
2939 0, /* tp_setattro */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002940 &bytearray_as_buffer, /* tp_as_buffer */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002941 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /* tp_flags */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002942 bytearray_doc, /* tp_doc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002943 0, /* tp_traverse */
2944 0, /* tp_clear */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002945 (richcmpfunc)bytearray_richcompare, /* tp_richcompare */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002946 0, /* tp_weaklistoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002947 bytearray_iter, /* tp_iter */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002948 0, /* tp_iternext */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002949 bytearray_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002950 0, /* tp_members */
2951 0, /* tp_getset */
2952 0, /* tp_base */
2953 0, /* tp_dict */
2954 0, /* tp_descr_get */
2955 0, /* tp_descr_set */
2956 0, /* tp_dictoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002957 (initproc)bytearray_init, /* tp_init */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002958 PyType_GenericAlloc, /* tp_alloc */
2959 PyType_GenericNew, /* tp_new */
2960 PyObject_Del, /* tp_free */
2961};
2962
2963/*********************** Bytes Iterator ****************************/
2964
2965typedef struct {
2966 PyObject_HEAD
2967 Py_ssize_t it_index;
2968 PyByteArrayObject *it_seq; /* Set to NULL when iterator is exhausted */
2969} bytesiterobject;
2970
2971static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002972bytearrayiter_dealloc(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002973{
2974 _PyObject_GC_UNTRACK(it);
2975 Py_XDECREF(it->it_seq);
2976 PyObject_GC_Del(it);
2977}
2978
2979static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002980bytearrayiter_traverse(bytesiterobject *it, visitproc visit, void *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002981{
2982 Py_VISIT(it->it_seq);
2983 return 0;
2984}
2985
2986static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002987bytearrayiter_next(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002988{
2989 PyByteArrayObject *seq;
2990 PyObject *item;
2991
2992 assert(it != NULL);
2993 seq = it->it_seq;
2994 if (seq == NULL)
2995 return NULL;
2996 assert(PyByteArray_Check(seq));
2997
2998 if (it->it_index < PyByteArray_GET_SIZE(seq)) {
2999 item = PyLong_FromLong(
3000 (unsigned char)seq->ob_bytes[it->it_index]);
3001 if (item != NULL)
3002 ++it->it_index;
3003 return item;
3004 }
3005
3006 Py_DECREF(seq);
3007 it->it_seq = NULL;
3008 return NULL;
3009}
3010
3011static PyObject *
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003012bytearrayiter_length_hint(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003013{
3014 Py_ssize_t len = 0;
3015 if (it->it_seq)
3016 len = PyByteArray_GET_SIZE(it->it_seq) - it->it_index;
3017 return PyLong_FromSsize_t(len);
3018}
3019
3020PyDoc_STRVAR(length_hint_doc,
3021 "Private method returning an estimate of len(list(it)).");
3022
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003023static PyObject *
3024bytearrayiter_reduce(bytesiterobject *it)
3025{
3026 if (it->it_seq != NULL) {
Antoine Pitroua7013882012-04-05 00:04:20 +02003027 return Py_BuildValue("N(O)n", _PyObject_GetBuiltin("iter"),
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003028 it->it_seq, it->it_index);
3029 } else {
3030 PyObject *u = PyUnicode_FromUnicode(NULL, 0);
3031 if (u == NULL)
3032 return NULL;
Antoine Pitroua7013882012-04-05 00:04:20 +02003033 return Py_BuildValue("N(N)", _PyObject_GetBuiltin("iter"), u);
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003034 }
3035}
3036
3037static PyObject *
3038bytearrayiter_setstate(bytesiterobject *it, PyObject *state)
3039{
3040 Py_ssize_t index = PyLong_AsSsize_t(state);
3041 if (index == -1 && PyErr_Occurred())
3042 return NULL;
3043 if (index < 0)
3044 index = 0;
3045 it->it_index = index;
3046 Py_RETURN_NONE;
3047}
3048
3049PyDoc_STRVAR(setstate_doc, "Set state information for unpickling.");
3050
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003051static PyMethodDef bytearrayiter_methods[] = {
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003052 {"__length_hint__", (PyCFunction)bytearrayiter_length_hint, METH_NOARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003053 length_hint_doc},
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003054 {"__reduce__", (PyCFunction)bytearrayiter_reduce, METH_NOARGS,
3055 reduce_doc},
3056 {"__setstate__", (PyCFunction)bytearrayiter_setstate, METH_O,
3057 setstate_doc},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003058 {NULL, NULL} /* sentinel */
3059};
3060
3061PyTypeObject PyByteArrayIter_Type = {
3062 PyVarObject_HEAD_INIT(&PyType_Type, 0)
3063 "bytearray_iterator", /* tp_name */
3064 sizeof(bytesiterobject), /* tp_basicsize */
3065 0, /* tp_itemsize */
3066 /* methods */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003067 (destructor)bytearrayiter_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003068 0, /* tp_print */
3069 0, /* tp_getattr */
3070 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00003071 0, /* tp_reserved */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003072 0, /* tp_repr */
3073 0, /* tp_as_number */
3074 0, /* tp_as_sequence */
3075 0, /* tp_as_mapping */
3076 0, /* tp_hash */
3077 0, /* tp_call */
3078 0, /* tp_str */
3079 PyObject_GenericGetAttr, /* tp_getattro */
3080 0, /* tp_setattro */
3081 0, /* tp_as_buffer */
3082 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */
3083 0, /* tp_doc */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003084 (traverseproc)bytearrayiter_traverse, /* tp_traverse */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003085 0, /* tp_clear */
3086 0, /* tp_richcompare */
3087 0, /* tp_weaklistoffset */
3088 PyObject_SelfIter, /* tp_iter */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003089 (iternextfunc)bytearrayiter_next, /* tp_iternext */
3090 bytearrayiter_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003091 0,
3092};
3093
3094static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003095bytearray_iter(PyObject *seq)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003096{
3097 bytesiterobject *it;
3098
3099 if (!PyByteArray_Check(seq)) {
3100 PyErr_BadInternalCall();
3101 return NULL;
3102 }
3103 it = PyObject_GC_New(bytesiterobject, &PyByteArrayIter_Type);
3104 if (it == NULL)
3105 return NULL;
3106 it->it_index = 0;
3107 Py_INCREF(seq);
3108 it->it_seq = (PyByteArrayObject *)seq;
3109 _PyObject_GC_TRACK(it);
3110 return (PyObject *)it;
3111}