blob: 2bb3a29daaa5b69acb8594ae593906b19cd6fab0 [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)) {
Georg Brandlf3fa5682010-12-04 17:09:30 +0000592 /* Make a copy and call this function recursively */
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000593 int err;
594 values = PyByteArray_FromObject(values);
595 if (values == NULL)
596 return -1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000597 err = bytearray_ass_subscript(self, index, values);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000598 Py_DECREF(values);
599 return err;
600 }
601 else {
602 assert(PyByteArray_Check(values));
603 bytes = ((PyByteArrayObject *)values)->ob_bytes;
604 needed = Py_SIZE(values);
605 }
606 /* Make sure b[5:2] = ... inserts before 5, not before 2. */
607 if ((step < 0 && start < stop) ||
608 (step > 0 && start > stop))
609 stop = start;
610 if (step == 1) {
611 if (slicelen != needed) {
Antoine Pitrou5504e892008-12-06 21:27:53 +0000612 if (!_canresize(self))
613 return -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000614 if (slicelen > needed) {
615 /*
616 0 start stop old_size
617 | |<---slicelen--->|<-----tomove------>|
618 | |<-needed->|<-----tomove------>|
619 0 lo new_hi new_size
620 */
621 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
622 Py_SIZE(self) - stop);
623 }
624 if (PyByteArray_Resize((PyObject *)self,
625 Py_SIZE(self) + needed - slicelen) < 0)
626 return -1;
627 if (slicelen < needed) {
628 /*
629 0 lo hi old_size
630 | |<-avail->|<-----tomove------>|
631 | |<----needed---->|<-----tomove------>|
632 0 lo new_hi new_size
633 */
634 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
635 Py_SIZE(self) - start - needed);
636 }
637 }
638
639 if (needed > 0)
640 memcpy(self->ob_bytes + start, bytes, needed);
641
642 return 0;
643 }
644 else {
645 if (needed == 0) {
646 /* Delete slice */
Mark Dickinsonbc099642010-01-29 17:27:24 +0000647 size_t cur;
648 Py_ssize_t i;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000649
Antoine Pitrou5504e892008-12-06 21:27:53 +0000650 if (!_canresize(self))
651 return -1;
Mark Dickinson7e3b9482010-08-06 21:33:18 +0000652
653 if (slicelen == 0)
654 /* Nothing to do here. */
655 return 0;
656
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000657 if (step < 0) {
658 stop = start + 1;
659 start = stop + step * (slicelen - 1) - 1;
660 step = -step;
661 }
662 for (cur = start, i = 0;
663 i < slicelen; cur += step, i++) {
664 Py_ssize_t lim = step - 1;
665
Mark Dickinson66f575b2010-02-14 12:53:32 +0000666 if (cur + step >= (size_t)PyByteArray_GET_SIZE(self))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000667 lim = PyByteArray_GET_SIZE(self) - cur - 1;
668
669 memmove(self->ob_bytes + cur - i,
670 self->ob_bytes + cur + 1, lim);
671 }
672 /* Move the tail of the bytes, in one chunk */
Mark Dickinson7e3b9482010-08-06 21:33:18 +0000673 cur = start + (size_t)slicelen*step;
Mark Dickinson66f575b2010-02-14 12:53:32 +0000674 if (cur < (size_t)PyByteArray_GET_SIZE(self)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000675 memmove(self->ob_bytes + cur - slicelen,
676 self->ob_bytes + cur,
677 PyByteArray_GET_SIZE(self) - cur);
678 }
679 if (PyByteArray_Resize((PyObject *)self,
680 PyByteArray_GET_SIZE(self) - slicelen) < 0)
681 return -1;
682
683 return 0;
684 }
685 else {
686 /* Assign slice */
Mark Dickinson7e3b9482010-08-06 21:33:18 +0000687 Py_ssize_t i;
688 size_t cur;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000689
690 if (needed != slicelen) {
691 PyErr_Format(PyExc_ValueError,
692 "attempt to assign bytes of size %zd "
693 "to extended slice of size %zd",
694 needed, slicelen);
695 return -1;
696 }
697 for (cur = start, i = 0; i < slicelen; cur += step, i++)
698 self->ob_bytes[cur] = bytes[i];
699 return 0;
700 }
701 }
702}
703
704static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000705bytearray_init(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000706{
707 static char *kwlist[] = {"source", "encoding", "errors", 0};
708 PyObject *arg = NULL;
709 const char *encoding = NULL;
710 const char *errors = NULL;
711 Py_ssize_t count;
712 PyObject *it;
713 PyObject *(*iternext)(PyObject *);
714
715 if (Py_SIZE(self) != 0) {
716 /* Empty previous contents (yes, do this first of all!) */
717 if (PyByteArray_Resize((PyObject *)self, 0) < 0)
718 return -1;
719 }
720
721 /* Parse arguments */
Georg Brandl3dbca812008-07-23 16:10:53 +0000722 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|Oss:bytearray", kwlist,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000723 &arg, &encoding, &errors))
724 return -1;
725
726 /* Make a quick exit if no first argument */
727 if (arg == NULL) {
728 if (encoding != NULL || errors != NULL) {
729 PyErr_SetString(PyExc_TypeError,
730 "encoding or errors without sequence argument");
731 return -1;
732 }
733 return 0;
734 }
735
736 if (PyUnicode_Check(arg)) {
737 /* Encode via the codec registry */
738 PyObject *encoded, *new;
739 if (encoding == NULL) {
740 PyErr_SetString(PyExc_TypeError,
741 "string argument without an encoding");
742 return -1;
743 }
Marc-André Lemburgb2750b52008-06-06 12:18:17 +0000744 encoded = PyUnicode_AsEncodedString(arg, encoding, errors);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000745 if (encoded == NULL)
746 return -1;
747 assert(PyBytes_Check(encoded));
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000748 new = bytearray_iconcat(self, encoded);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000749 Py_DECREF(encoded);
750 if (new == NULL)
751 return -1;
752 Py_DECREF(new);
753 return 0;
754 }
755
756 /* If it's not unicode, there can't be encoding or errors */
757 if (encoding != NULL || errors != NULL) {
758 PyErr_SetString(PyExc_TypeError,
759 "encoding or errors without a string argument");
760 return -1;
761 }
762
763 /* Is it an int? */
Benjamin Peterson8380dd52010-04-16 22:51:37 +0000764 count = PyNumber_AsSsize_t(arg, PyExc_OverflowError);
765 if (count == -1 && PyErr_Occurred()) {
766 if (PyErr_ExceptionMatches(PyExc_OverflowError))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000767 return -1;
Benjamin Peterson9c0e94f2010-04-16 23:00:53 +0000768 PyErr_Clear();
Benjamin Peterson8380dd52010-04-16 22:51:37 +0000769 }
770 else if (count < 0) {
771 PyErr_SetString(PyExc_ValueError, "negative count");
772 return -1;
773 }
774 else {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000775 if (count > 0) {
776 if (PyByteArray_Resize((PyObject *)self, count))
777 return -1;
778 memset(self->ob_bytes, 0, count);
779 }
780 return 0;
781 }
782
783 /* Use the buffer API */
784 if (PyObject_CheckBuffer(arg)) {
785 Py_ssize_t size;
786 Py_buffer view;
787 if (PyObject_GetBuffer(arg, &view, PyBUF_FULL_RO) < 0)
788 return -1;
789 size = view.len;
790 if (PyByteArray_Resize((PyObject *)self, size) < 0) goto fail;
791 if (PyBuffer_ToContiguous(self->ob_bytes, &view, size, 'C') < 0)
Stefan Krah7d12d9d2012-07-28 12:25:55 +0200792 goto fail;
Martin v. Löwis423be952008-08-13 15:53:07 +0000793 PyBuffer_Release(&view);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000794 return 0;
795 fail:
Martin v. Löwis423be952008-08-13 15:53:07 +0000796 PyBuffer_Release(&view);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000797 return -1;
798 }
799
800 /* XXX Optimize this if the arguments is a list, tuple */
801
802 /* Get the iterator */
803 it = PyObject_GetIter(arg);
804 if (it == NULL)
805 return -1;
806 iternext = *Py_TYPE(it)->tp_iternext;
807
808 /* Run the iterator to exhaustion */
809 for (;;) {
810 PyObject *item;
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000811 int rc, value;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000812
813 /* Get the next item */
814 item = iternext(it);
815 if (item == NULL) {
816 if (PyErr_Occurred()) {
817 if (!PyErr_ExceptionMatches(PyExc_StopIteration))
818 goto error;
819 PyErr_Clear();
820 }
821 break;
822 }
823
824 /* Interpret it as an int (__index__) */
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000825 rc = _getbytevalue(item, &value);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000826 Py_DECREF(item);
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000827 if (!rc)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000828 goto error;
829
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000830 /* Append the byte */
831 if (Py_SIZE(self) < self->ob_alloc)
832 Py_SIZE(self)++;
833 else if (PyByteArray_Resize((PyObject *)self, Py_SIZE(self)+1) < 0)
834 goto error;
835 self->ob_bytes[Py_SIZE(self)-1] = value;
836 }
837
838 /* Clean up and return success */
839 Py_DECREF(it);
840 return 0;
841
842 error:
843 /* Error handling when it != NULL */
844 Py_DECREF(it);
845 return -1;
846}
847
848/* Mostly copied from string_repr, but without the
849 "smart quote" functionality. */
850static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000851bytearray_repr(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000852{
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000853 const char *quote_prefix = "bytearray(b";
854 const char *quote_postfix = ")";
855 Py_ssize_t length = Py_SIZE(self);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200856 /* 15 == strlen(quote_prefix) + 2 + strlen(quote_postfix) + 1 */
Mark Dickinson66f575b2010-02-14 12:53:32 +0000857 size_t newsize;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000858 PyObject *v;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200859 register Py_ssize_t i;
860 register char c;
861 register char *p;
862 int quote;
863 char *test, *start;
864 char *buffer;
865
866 if (length > (PY_SSIZE_T_MAX - 15) / 4) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000867 PyErr_SetString(PyExc_OverflowError,
868 "bytearray object is too large to make repr");
869 return NULL;
870 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200871
872 newsize = 15 + length * 4;
Antoine Pitrou39aba4f2011-11-12 21:15:28 +0100873 buffer = PyObject_Malloc(newsize);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200874 if (buffer == NULL) {
875 PyErr_NoMemory();
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000876 return NULL;
877 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000878
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200879 /* Figure out which quote to use; single is preferred */
880 quote = '\'';
881 start = PyByteArray_AS_STRING(self);
882 for (test = start; test < start+length; ++test) {
883 if (*test == '"') {
884 quote = '\''; /* back to single */
885 break;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000886 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200887 else if (*test == '\'')
888 quote = '"';
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000889 }
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200890
891 p = buffer;
892 while (*quote_prefix)
893 *p++ = *quote_prefix++;
894 *p++ = quote;
895
896 for (i = 0; i < length; i++) {
897 /* There's at least enough room for a hex escape
898 and a closing quote. */
899 assert(newsize - (p - buffer) >= 5);
900 c = self->ob_bytes[i];
901 if (c == '\'' || c == '\\')
902 *p++ = '\\', *p++ = c;
903 else if (c == '\t')
904 *p++ = '\\', *p++ = 't';
905 else if (c == '\n')
906 *p++ = '\\', *p++ = 'n';
907 else if (c == '\r')
908 *p++ = '\\', *p++ = 'r';
909 else if (c == 0)
910 *p++ = '\\', *p++ = 'x', *p++ = '0', *p++ = '0';
911 else if (c < ' ' || c >= 0x7f) {
912 *p++ = '\\';
913 *p++ = 'x';
Victor Stinnerf5cff562011-10-14 02:13:11 +0200914 *p++ = Py_hexdigits[(c & 0xf0) >> 4];
915 *p++ = Py_hexdigits[c & 0xf];
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200916 }
917 else
918 *p++ = c;
919 }
920 assert(newsize - (p - buffer) >= 1);
921 *p++ = quote;
922 while (*quote_postfix) {
923 *p++ = *quote_postfix++;
924 }
925
926 v = PyUnicode_DecodeASCII(buffer, p - buffer, NULL);
Antoine Pitrou39aba4f2011-11-12 21:15:28 +0100927 PyObject_Free(buffer);
Martin v. Löwisd63a3b82011-09-28 07:41:54 +0200928 return v;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000929}
930
931static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000932bytearray_str(PyObject *op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000933{
Alexander Belopolskyf0f45142010-08-11 17:31:17 +0000934 if (Py_BytesWarningFlag) {
935 if (PyErr_WarnEx(PyExc_BytesWarning,
936 "str() on a bytearray instance", 1))
937 return NULL;
938 }
939 return bytearray_repr((PyByteArrayObject*)op);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000940}
941
942static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000943bytearray_richcompare(PyObject *self, PyObject *other, int op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000944{
945 Py_ssize_t self_size, other_size;
946 Py_buffer self_bytes, other_bytes;
947 PyObject *res;
948 Py_ssize_t minsize;
949 int cmp;
950
951 /* Bytes can be compared to anything that supports the (binary)
952 buffer API. Except that a comparison with Unicode is always an
953 error, even if the comparison is for equality. */
954 if (PyObject_IsInstance(self, (PyObject*)&PyUnicode_Type) ||
955 PyObject_IsInstance(other, (PyObject*)&PyUnicode_Type)) {
Barry Warsaw9e9dcd62008-10-17 01:50:37 +0000956 if (Py_BytesWarningFlag && (op == Py_EQ || op == Py_NE)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000957 if (PyErr_WarnEx(PyExc_BytesWarning,
Georg Brandle5d68ac2008-06-04 11:30:26 +0000958 "Comparison between bytearray and string", 1))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000959 return NULL;
960 }
961
Brian Curtindfc80e32011-08-10 20:28:54 -0500962 Py_RETURN_NOTIMPLEMENTED;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000963 }
964
965 self_size = _getbuffer(self, &self_bytes);
966 if (self_size < 0) {
967 PyErr_Clear();
Brian Curtindfc80e32011-08-10 20:28:54 -0500968 Py_RETURN_NOTIMPLEMENTED;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000969 }
970
971 other_size = _getbuffer(other, &other_bytes);
972 if (other_size < 0) {
973 PyErr_Clear();
Martin v. Löwis423be952008-08-13 15:53:07 +0000974 PyBuffer_Release(&self_bytes);
Brian Curtindfc80e32011-08-10 20:28:54 -0500975 Py_RETURN_NOTIMPLEMENTED;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000976 }
977
978 if (self_size != other_size && (op == Py_EQ || op == Py_NE)) {
979 /* Shortcut: if the lengths differ, the objects differ */
980 cmp = (op == Py_NE);
981 }
982 else {
983 minsize = self_size;
984 if (other_size < minsize)
985 minsize = other_size;
986
987 cmp = memcmp(self_bytes.buf, other_bytes.buf, minsize);
988 /* In ISO C, memcmp() guarantees to use unsigned bytes! */
989
990 if (cmp == 0) {
991 if (self_size < other_size)
992 cmp = -1;
993 else if (self_size > other_size)
994 cmp = 1;
995 }
996
997 switch (op) {
998 case Py_LT: cmp = cmp < 0; break;
999 case Py_LE: cmp = cmp <= 0; break;
1000 case Py_EQ: cmp = cmp == 0; break;
1001 case Py_NE: cmp = cmp != 0; break;
1002 case Py_GT: cmp = cmp > 0; break;
1003 case Py_GE: cmp = cmp >= 0; break;
1004 }
1005 }
1006
1007 res = cmp ? Py_True : Py_False;
Martin v. Löwis423be952008-08-13 15:53:07 +00001008 PyBuffer_Release(&self_bytes);
1009 PyBuffer_Release(&other_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001010 Py_INCREF(res);
1011 return res;
1012}
1013
1014static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001015bytearray_dealloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001016{
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001017 if (self->ob_exports > 0) {
1018 PyErr_SetString(PyExc_SystemError,
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00001019 "deallocated bytearray object has exported buffers");
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001020 PyErr_Print();
1021 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001022 if (self->ob_bytes != 0) {
Antoine Pitrou39aba4f2011-11-12 21:15:28 +01001023 PyObject_Free(self->ob_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001024 }
1025 Py_TYPE(self)->tp_free((PyObject *)self);
1026}
1027
1028
1029/* -------------------------------------------------------------------- */
1030/* Methods */
1031
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001032#define FASTSEARCH fastsearch
1033#define STRINGLIB(F) stringlib_##F
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001034#define STRINGLIB_CHAR char
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001035#define STRINGLIB_LEN PyByteArray_GET_SIZE
1036#define STRINGLIB_STR PyByteArray_AS_STRING
1037#define STRINGLIB_NEW PyByteArray_FromStringAndSize
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001038#define STRINGLIB_ISSPACE Py_ISSPACE
1039#define STRINGLIB_ISLINEBREAK(x) ((x == '\n') || (x == '\r'))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001040#define STRINGLIB_CHECK_EXACT PyByteArray_CheckExact
1041#define STRINGLIB_MUTABLE 1
1042
1043#include "stringlib/fastsearch.h"
1044#include "stringlib/count.h"
1045#include "stringlib/find.h"
1046#include "stringlib/partition.h"
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001047#include "stringlib/split.h"
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001048#include "stringlib/ctype.h"
1049#include "stringlib/transmogrify.h"
1050
1051
1052/* The following Py_LOCAL_INLINE and Py_LOCAL functions
1053were copied from the old char* style string object. */
1054
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001055/* helper macro to fixup start/end slice values */
1056#define ADJUST_INDICES(start, end, len) \
1057 if (end > len) \
1058 end = len; \
1059 else if (end < 0) { \
1060 end += len; \
1061 if (end < 0) \
1062 end = 0; \
1063 } \
1064 if (start < 0) { \
1065 start += len; \
1066 if (start < 0) \
1067 start = 0; \
1068 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001069
1070Py_LOCAL_INLINE(Py_ssize_t)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001071bytearray_find_internal(PyByteArrayObject *self, PyObject *args, int dir)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001072{
1073 PyObject *subobj;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001074 char byte;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001075 Py_buffer subbuf;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001076 const char *sub;
1077 Py_ssize_t sub_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001078 Py_ssize_t start=0, end=PY_SSIZE_T_MAX;
1079 Py_ssize_t res;
1080
Antoine Pitrouac65d962011-10-20 23:54:17 +02001081 if (!stringlib_parse_args_finds_byte("find/rfind/index/rindex",
1082 args, &subobj, &byte, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001083 return -2;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001084
1085 if (subobj) {
1086 if (_getbuffer(subobj, &subbuf) < 0)
1087 return -2;
1088
1089 sub = subbuf.buf;
1090 sub_len = subbuf.len;
1091 }
1092 else {
1093 sub = &byte;
1094 sub_len = 1;
1095 }
1096
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001097 if (dir > 0)
1098 res = stringlib_find_slice(
1099 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
Antoine Pitrouac65d962011-10-20 23:54:17 +02001100 sub, sub_len, start, end);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001101 else
1102 res = stringlib_rfind_slice(
1103 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
Antoine Pitrouac65d962011-10-20 23:54:17 +02001104 sub, sub_len, start, end);
1105
1106 if (subobj)
1107 PyBuffer_Release(&subbuf);
1108
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001109 return res;
1110}
1111
1112PyDoc_STRVAR(find__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001113"B.find(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001114\n\
1115Return the lowest index in B where subsection sub is found,\n\
Senthil Kumaran53516a82011-07-27 23:33:54 +08001116such that sub is contained within B[start,end]. Optional\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001117arguments start and end are interpreted as in slice notation.\n\
1118\n\
1119Return -1 on failure.");
1120
1121static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001122bytearray_find(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001123{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001124 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001125 if (result == -2)
1126 return NULL;
1127 return PyLong_FromSsize_t(result);
1128}
1129
1130PyDoc_STRVAR(count__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001131"B.count(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001132\n\
1133Return the number of non-overlapping occurrences of subsection sub in\n\
1134bytes B[start:end]. Optional arguments start and end are interpreted\n\
1135as in slice notation.");
1136
1137static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001138bytearray_count(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001139{
1140 PyObject *sub_obj;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001141 const char *str = PyByteArray_AS_STRING(self), *sub;
1142 Py_ssize_t sub_len;
1143 char byte;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001144 Py_ssize_t start = 0, end = PY_SSIZE_T_MAX;
Antoine Pitrouac65d962011-10-20 23:54:17 +02001145
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001146 Py_buffer vsub;
1147 PyObject *count_obj;
1148
Antoine Pitrouac65d962011-10-20 23:54:17 +02001149 if (!stringlib_parse_args_finds_byte("count", args, &sub_obj, &byte,
1150 &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001151 return NULL;
1152
Antoine Pitrouac65d962011-10-20 23:54:17 +02001153 if (sub_obj) {
1154 if (_getbuffer(sub_obj, &vsub) < 0)
1155 return NULL;
1156
1157 sub = vsub.buf;
1158 sub_len = vsub.len;
1159 }
1160 else {
1161 sub = &byte;
1162 sub_len = 1;
1163 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001164
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001165 ADJUST_INDICES(start, end, PyByteArray_GET_SIZE(self));
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001166
1167 count_obj = PyLong_FromSsize_t(
Antoine Pitrouac65d962011-10-20 23:54:17 +02001168 stringlib_count(str + start, end - start, sub, sub_len, PY_SSIZE_T_MAX)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001169 );
Antoine Pitrouac65d962011-10-20 23:54:17 +02001170
1171 if (sub_obj)
1172 PyBuffer_Release(&vsub);
1173
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001174 return count_obj;
1175}
1176
Eli Bendersky4db28d32011-03-03 18:21:02 +00001177PyDoc_STRVAR(clear__doc__,
1178"B.clear() -> None\n\
1179\n\
1180Remove all items from B.");
1181
Victor Stinner6430fd52011-09-29 04:02:13 +02001182static PyObject *
Eli Bendersky4db28d32011-03-03 18:21:02 +00001183bytearray_clear(PyByteArrayObject *self)
1184{
1185 if (PyByteArray_Resize((PyObject *)self, 0) < 0)
1186 return NULL;
1187 Py_RETURN_NONE;
1188}
1189
1190PyDoc_STRVAR(copy__doc__,
1191"B.copy() -> bytearray\n\
1192\n\
1193Return a copy of B.");
1194
1195static PyObject *
1196bytearray_copy(PyByteArrayObject *self)
1197{
1198 return PyByteArray_FromStringAndSize(PyByteArray_AS_STRING((PyObject *)self),
1199 PyByteArray_GET_SIZE(self));
1200}
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001201
1202PyDoc_STRVAR(index__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001203"B.index(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001204\n\
1205Like B.find() but raise ValueError when the subsection is not found.");
1206
1207static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001208bytearray_index(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001209{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001210 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001211 if (result == -2)
1212 return NULL;
1213 if (result == -1) {
1214 PyErr_SetString(PyExc_ValueError,
1215 "subsection not found");
1216 return NULL;
1217 }
1218 return PyLong_FromSsize_t(result);
1219}
1220
1221
1222PyDoc_STRVAR(rfind__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001223"B.rfind(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001224\n\
1225Return the highest index in B where subsection sub is found,\n\
Senthil Kumaran53516a82011-07-27 23:33:54 +08001226such that sub is contained within B[start,end]. Optional\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001227arguments start and end are interpreted as in slice notation.\n\
1228\n\
1229Return -1 on failure.");
1230
1231static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001232bytearray_rfind(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001233{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001234 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001235 if (result == -2)
1236 return NULL;
1237 return PyLong_FromSsize_t(result);
1238}
1239
1240
1241PyDoc_STRVAR(rindex__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001242"B.rindex(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001243\n\
1244Like B.rfind() but raise ValueError when the subsection is not found.");
1245
1246static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001247bytearray_rindex(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001248{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001249 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001250 if (result == -2)
1251 return NULL;
1252 if (result == -1) {
1253 PyErr_SetString(PyExc_ValueError,
1254 "subsection not found");
1255 return NULL;
1256 }
1257 return PyLong_FromSsize_t(result);
1258}
1259
1260
1261static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001262bytearray_contains(PyObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001263{
1264 Py_ssize_t ival = PyNumber_AsSsize_t(arg, PyExc_ValueError);
1265 if (ival == -1 && PyErr_Occurred()) {
1266 Py_buffer varg;
Antoine Pitrou0010d372010-08-15 17:12:55 +00001267 Py_ssize_t pos;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001268 PyErr_Clear();
1269 if (_getbuffer(arg, &varg) < 0)
1270 return -1;
1271 pos = stringlib_find(PyByteArray_AS_STRING(self), Py_SIZE(self),
1272 varg.buf, varg.len, 0);
Martin v. Löwis423be952008-08-13 15:53:07 +00001273 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001274 return pos >= 0;
1275 }
1276 if (ival < 0 || ival >= 256) {
1277 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
1278 return -1;
1279 }
1280
Antoine Pitrou0010d372010-08-15 17:12:55 +00001281 return memchr(PyByteArray_AS_STRING(self), (int) ival, Py_SIZE(self)) != NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001282}
1283
1284
1285/* Matches the end (direction >= 0) or start (direction < 0) of self
1286 * against substr, using the start and end arguments. Returns
1287 * -1 on error, 0 if not found and 1 if found.
1288 */
1289Py_LOCAL(int)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001290_bytearray_tailmatch(PyByteArrayObject *self, PyObject *substr, Py_ssize_t start,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001291 Py_ssize_t end, int direction)
1292{
1293 Py_ssize_t len = PyByteArray_GET_SIZE(self);
1294 const char* str;
1295 Py_buffer vsubstr;
1296 int rv = 0;
1297
1298 str = PyByteArray_AS_STRING(self);
1299
1300 if (_getbuffer(substr, &vsubstr) < 0)
1301 return -1;
1302
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001303 ADJUST_INDICES(start, end, len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001304
1305 if (direction < 0) {
1306 /* startswith */
1307 if (start+vsubstr.len > len) {
1308 goto done;
1309 }
1310 } else {
1311 /* endswith */
1312 if (end-start < vsubstr.len || start > len) {
1313 goto done;
1314 }
1315
1316 if (end-vsubstr.len > start)
1317 start = end - vsubstr.len;
1318 }
1319 if (end-start >= vsubstr.len)
1320 rv = ! memcmp(str+start, vsubstr.buf, vsubstr.len);
1321
1322done:
Martin v. Löwis423be952008-08-13 15:53:07 +00001323 PyBuffer_Release(&vsubstr);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001324 return rv;
1325}
1326
1327
1328PyDoc_STRVAR(startswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001329"B.startswith(prefix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001330\n\
1331Return True if B starts with the specified prefix, False otherwise.\n\
1332With optional start, test B beginning at that position.\n\
1333With optional end, stop comparing B at that position.\n\
Ezio Melottiba42fd52011-04-26 06:09:45 +03001334prefix can also be a tuple of bytes to try.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001335
1336static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001337bytearray_startswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001338{
1339 Py_ssize_t start = 0;
1340 Py_ssize_t end = PY_SSIZE_T_MAX;
1341 PyObject *subobj;
1342 int result;
1343
Jesus Ceaac451502011-04-20 17:09:23 +02001344 if (!stringlib_parse_args_finds("startswith", args, &subobj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001345 return NULL;
1346 if (PyTuple_Check(subobj)) {
1347 Py_ssize_t i;
1348 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001349 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001350 PyTuple_GET_ITEM(subobj, i),
1351 start, end, -1);
1352 if (result == -1)
1353 return NULL;
1354 else if (result) {
1355 Py_RETURN_TRUE;
1356 }
1357 }
1358 Py_RETURN_FALSE;
1359 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001360 result = _bytearray_tailmatch(self, subobj, start, end, -1);
Ezio Melottiba42fd52011-04-26 06:09:45 +03001361 if (result == -1) {
1362 if (PyErr_ExceptionMatches(PyExc_TypeError))
1363 PyErr_Format(PyExc_TypeError, "startswith first arg must be bytes "
1364 "or a tuple of bytes, not %s", Py_TYPE(subobj)->tp_name);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001365 return NULL;
Ezio Melottiba42fd52011-04-26 06:09:45 +03001366 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001367 else
1368 return PyBool_FromLong(result);
1369}
1370
1371PyDoc_STRVAR(endswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001372"B.endswith(suffix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001373\n\
1374Return True if B ends with the specified suffix, False otherwise.\n\
1375With optional start, test B beginning at that position.\n\
1376With optional end, stop comparing B at that position.\n\
Ezio Melottiba42fd52011-04-26 06:09:45 +03001377suffix can also be a tuple of bytes to try.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001378
1379static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001380bytearray_endswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001381{
1382 Py_ssize_t start = 0;
1383 Py_ssize_t end = PY_SSIZE_T_MAX;
1384 PyObject *subobj;
1385 int result;
1386
Jesus Ceaac451502011-04-20 17:09:23 +02001387 if (!stringlib_parse_args_finds("endswith", args, &subobj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001388 return NULL;
1389 if (PyTuple_Check(subobj)) {
1390 Py_ssize_t i;
1391 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001392 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001393 PyTuple_GET_ITEM(subobj, i),
1394 start, end, +1);
1395 if (result == -1)
1396 return NULL;
1397 else if (result) {
1398 Py_RETURN_TRUE;
1399 }
1400 }
1401 Py_RETURN_FALSE;
1402 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001403 result = _bytearray_tailmatch(self, subobj, start, end, +1);
Ezio Melottiba42fd52011-04-26 06:09:45 +03001404 if (result == -1) {
1405 if (PyErr_ExceptionMatches(PyExc_TypeError))
1406 PyErr_Format(PyExc_TypeError, "endswith first arg must be bytes or "
1407 "a tuple of bytes, not %s", Py_TYPE(subobj)->tp_name);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001408 return NULL;
Ezio Melottiba42fd52011-04-26 06:09:45 +03001409 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001410 else
1411 return PyBool_FromLong(result);
1412}
1413
1414
1415PyDoc_STRVAR(translate__doc__,
1416"B.translate(table[, deletechars]) -> bytearray\n\
1417\n\
1418Return a copy of B, where all characters occurring in the\n\
1419optional argument deletechars are removed, and the remaining\n\
1420characters have been mapped through the given translation\n\
1421table, which must be a bytes object of length 256.");
1422
1423static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001424bytearray_translate(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001425{
1426 register char *input, *output;
1427 register const char *table;
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001428 register Py_ssize_t i, c;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001429 PyObject *input_obj = (PyObject*)self;
1430 const char *output_start;
1431 Py_ssize_t inlen;
Georg Brandlccc47b62008-12-28 11:44:14 +00001432 PyObject *result = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001433 int trans_table[256];
Georg Brandlccc47b62008-12-28 11:44:14 +00001434 PyObject *tableobj = NULL, *delobj = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001435 Py_buffer vtable, vdel;
1436
1437 if (!PyArg_UnpackTuple(args, "translate", 1, 2,
1438 &tableobj, &delobj))
1439 return NULL;
1440
Georg Brandlccc47b62008-12-28 11:44:14 +00001441 if (tableobj == Py_None) {
1442 table = NULL;
1443 tableobj = NULL;
1444 } else if (_getbuffer(tableobj, &vtable) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001445 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001446 } else {
1447 if (vtable.len != 256) {
1448 PyErr_SetString(PyExc_ValueError,
1449 "translation table must be 256 characters long");
Georg Brandl953152f2009-07-22 12:03:59 +00001450 PyBuffer_Release(&vtable);
1451 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001452 }
1453 table = (const char*)vtable.buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001454 }
1455
1456 if (delobj != NULL) {
1457 if (_getbuffer(delobj, &vdel) < 0) {
Georg Brandl953152f2009-07-22 12:03:59 +00001458 if (tableobj != NULL)
1459 PyBuffer_Release(&vtable);
1460 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001461 }
1462 }
1463 else {
1464 vdel.buf = NULL;
1465 vdel.len = 0;
1466 }
1467
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001468 inlen = PyByteArray_GET_SIZE(input_obj);
1469 result = PyByteArray_FromStringAndSize((char *)NULL, inlen);
1470 if (result == NULL)
1471 goto done;
1472 output_start = output = PyByteArray_AsString(result);
1473 input = PyByteArray_AS_STRING(input_obj);
1474
Georg Brandlccc47b62008-12-28 11:44:14 +00001475 if (vdel.len == 0 && table != NULL) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001476 /* If no deletions are required, use faster code */
1477 for (i = inlen; --i >= 0; ) {
1478 c = Py_CHARMASK(*input++);
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001479 *output++ = table[c];
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001480 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001481 goto done;
1482 }
Georg Brandlccc47b62008-12-28 11:44:14 +00001483
1484 if (table == NULL) {
1485 for (i = 0; i < 256; i++)
1486 trans_table[i] = Py_CHARMASK(i);
1487 } else {
1488 for (i = 0; i < 256; i++)
1489 trans_table[i] = Py_CHARMASK(table[i]);
1490 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001491
1492 for (i = 0; i < vdel.len; i++)
1493 trans_table[(int) Py_CHARMASK( ((unsigned char*)vdel.buf)[i] )] = -1;
1494
1495 for (i = inlen; --i >= 0; ) {
1496 c = Py_CHARMASK(*input++);
1497 if (trans_table[c] != -1)
1498 if (Py_CHARMASK(*output++ = (char)trans_table[c]) == c)
1499 continue;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001500 }
1501 /* Fix the size of the resulting string */
1502 if (inlen > 0)
1503 PyByteArray_Resize(result, output - output_start);
1504
1505done:
Georg Brandlccc47b62008-12-28 11:44:14 +00001506 if (tableobj != NULL)
1507 PyBuffer_Release(&vtable);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001508 if (delobj != NULL)
Martin v. Löwis423be952008-08-13 15:53:07 +00001509 PyBuffer_Release(&vdel);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001510 return result;
1511}
1512
1513
Georg Brandlabc38772009-04-12 15:51:51 +00001514static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001515bytearray_maketrans(PyObject *null, PyObject *args)
Georg Brandlabc38772009-04-12 15:51:51 +00001516{
Alexander Belopolskyf0f45142010-08-11 17:31:17 +00001517 return _Py_bytes_maketrans(args);
Georg Brandlabc38772009-04-12 15:51:51 +00001518}
1519
1520
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001521/* find and count characters and substrings */
1522
1523#define findchar(target, target_len, c) \
1524 ((char *)memchr((const void *)(target), c, target_len))
1525
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001526
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001527/* Bytes ops must return a string, create a copy */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001528Py_LOCAL(PyByteArrayObject *)
1529return_self(PyByteArrayObject *self)
1530{
Georg Brandl1e7217d2008-05-30 12:02:38 +00001531 /* always return a new bytearray */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001532 return (PyByteArrayObject *)PyByteArray_FromStringAndSize(
1533 PyByteArray_AS_STRING(self),
1534 PyByteArray_GET_SIZE(self));
1535}
1536
1537Py_LOCAL_INLINE(Py_ssize_t)
1538countchar(const char *target, Py_ssize_t target_len, char c, Py_ssize_t maxcount)
1539{
1540 Py_ssize_t count=0;
1541 const char *start=target;
1542 const char *end=target+target_len;
1543
1544 while ( (start=findchar(start, end-start, c)) != NULL ) {
1545 count++;
1546 if (count >= maxcount)
1547 break;
1548 start += 1;
1549 }
1550 return count;
1551}
1552
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001553
1554/* Algorithms for different cases of string replacement */
1555
1556/* len(self)>=1, from="", len(to)>=1, maxcount>=1 */
1557Py_LOCAL(PyByteArrayObject *)
1558replace_interleave(PyByteArrayObject *self,
1559 const char *to_s, Py_ssize_t to_len,
1560 Py_ssize_t maxcount)
1561{
1562 char *self_s, *result_s;
1563 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001564 Py_ssize_t count, i;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001565 PyByteArrayObject *result;
1566
1567 self_len = PyByteArray_GET_SIZE(self);
1568
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001569 /* 1 at the end plus 1 after every character;
1570 count = min(maxcount, self_len + 1) */
1571 if (maxcount <= self_len)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001572 count = maxcount;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001573 else
1574 /* Can't overflow: self_len + 1 <= maxcount <= PY_SSIZE_T_MAX. */
1575 count = self_len + 1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001576
1577 /* Check for overflow */
1578 /* result_len = count * to_len + self_len; */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001579 assert(count > 0);
1580 if (to_len > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001581 PyErr_SetString(PyExc_OverflowError,
1582 "replace string is too long");
1583 return NULL;
1584 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001585 result_len = count * to_len + self_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001586
1587 if (! (result = (PyByteArrayObject *)
1588 PyByteArray_FromStringAndSize(NULL, result_len)) )
1589 return NULL;
1590
1591 self_s = PyByteArray_AS_STRING(self);
1592 result_s = PyByteArray_AS_STRING(result);
1593
1594 /* TODO: special case single character, which doesn't need memcpy */
1595
1596 /* Lay the first one down (guaranteed this will occur) */
1597 Py_MEMCPY(result_s, to_s, to_len);
1598 result_s += to_len;
1599 count -= 1;
1600
1601 for (i=0; i<count; i++) {
1602 *result_s++ = *self_s++;
1603 Py_MEMCPY(result_s, to_s, to_len);
1604 result_s += to_len;
1605 }
1606
1607 /* Copy the rest of the original string */
1608 Py_MEMCPY(result_s, self_s, self_len-i);
1609
1610 return result;
1611}
1612
1613/* Special case for deleting a single character */
1614/* len(self)>=1, len(from)==1, to="", maxcount>=1 */
1615Py_LOCAL(PyByteArrayObject *)
1616replace_delete_single_character(PyByteArrayObject *self,
1617 char from_c, Py_ssize_t maxcount)
1618{
1619 char *self_s, *result_s;
1620 char *start, *next, *end;
1621 Py_ssize_t self_len, result_len;
1622 Py_ssize_t count;
1623 PyByteArrayObject *result;
1624
1625 self_len = PyByteArray_GET_SIZE(self);
1626 self_s = PyByteArray_AS_STRING(self);
1627
1628 count = countchar(self_s, self_len, from_c, maxcount);
1629 if (count == 0) {
1630 return return_self(self);
1631 }
1632
1633 result_len = self_len - count; /* from_len == 1 */
1634 assert(result_len>=0);
1635
1636 if ( (result = (PyByteArrayObject *)
1637 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1638 return NULL;
1639 result_s = PyByteArray_AS_STRING(result);
1640
1641 start = self_s;
1642 end = self_s + self_len;
1643 while (count-- > 0) {
1644 next = findchar(start, end-start, from_c);
1645 if (next == NULL)
1646 break;
1647 Py_MEMCPY(result_s, start, next-start);
1648 result_s += (next-start);
1649 start = next+1;
1650 }
1651 Py_MEMCPY(result_s, start, end-start);
1652
1653 return result;
1654}
1655
1656/* len(self)>=1, len(from)>=2, to="", maxcount>=1 */
1657
1658Py_LOCAL(PyByteArrayObject *)
1659replace_delete_substring(PyByteArrayObject *self,
1660 const char *from_s, Py_ssize_t from_len,
1661 Py_ssize_t maxcount)
1662{
1663 char *self_s, *result_s;
1664 char *start, *next, *end;
1665 Py_ssize_t self_len, result_len;
1666 Py_ssize_t count, offset;
1667 PyByteArrayObject *result;
1668
1669 self_len = PyByteArray_GET_SIZE(self);
1670 self_s = PyByteArray_AS_STRING(self);
1671
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001672 count = stringlib_count(self_s, self_len,
1673 from_s, from_len,
1674 maxcount);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001675
1676 if (count == 0) {
1677 /* no matches */
1678 return return_self(self);
1679 }
1680
1681 result_len = self_len - (count * from_len);
1682 assert (result_len>=0);
1683
1684 if ( (result = (PyByteArrayObject *)
1685 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL )
1686 return NULL;
1687
1688 result_s = PyByteArray_AS_STRING(result);
1689
1690 start = self_s;
1691 end = self_s + self_len;
1692 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001693 offset = stringlib_find(start, end-start,
1694 from_s, from_len,
1695 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001696 if (offset == -1)
1697 break;
1698 next = start + offset;
1699
1700 Py_MEMCPY(result_s, start, next-start);
1701
1702 result_s += (next-start);
1703 start = next+from_len;
1704 }
1705 Py_MEMCPY(result_s, start, end-start);
1706 return result;
1707}
1708
1709/* len(self)>=1, len(from)==len(to)==1, maxcount>=1 */
1710Py_LOCAL(PyByteArrayObject *)
1711replace_single_character_in_place(PyByteArrayObject *self,
1712 char from_c, char to_c,
1713 Py_ssize_t maxcount)
1714{
Antoine Pitroud1188562010-06-09 16:38:55 +00001715 char *self_s, *result_s, *start, *end, *next;
1716 Py_ssize_t self_len;
1717 PyByteArrayObject *result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001718
Antoine Pitroud1188562010-06-09 16:38:55 +00001719 /* The result string will be the same size */
1720 self_s = PyByteArray_AS_STRING(self);
1721 self_len = PyByteArray_GET_SIZE(self);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001722
Antoine Pitroud1188562010-06-09 16:38:55 +00001723 next = findchar(self_s, self_len, from_c);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001724
Antoine Pitroud1188562010-06-09 16:38:55 +00001725 if (next == NULL) {
1726 /* No matches; return the original bytes */
1727 return return_self(self);
1728 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001729
Antoine Pitroud1188562010-06-09 16:38:55 +00001730 /* Need to make a new bytes */
1731 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1732 if (result == NULL)
1733 return NULL;
1734 result_s = PyByteArray_AS_STRING(result);
1735 Py_MEMCPY(result_s, self_s, self_len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001736
Antoine Pitroud1188562010-06-09 16:38:55 +00001737 /* change everything in-place, starting with this one */
1738 start = result_s + (next-self_s);
1739 *start = to_c;
1740 start++;
1741 end = result_s + self_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001742
Antoine Pitroud1188562010-06-09 16:38:55 +00001743 while (--maxcount > 0) {
1744 next = findchar(start, end-start, from_c);
1745 if (next == NULL)
1746 break;
1747 *next = to_c;
1748 start = next+1;
1749 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001750
Antoine Pitroud1188562010-06-09 16:38:55 +00001751 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001752}
1753
1754/* len(self)>=1, len(from)==len(to)>=2, maxcount>=1 */
1755Py_LOCAL(PyByteArrayObject *)
1756replace_substring_in_place(PyByteArrayObject *self,
1757 const char *from_s, Py_ssize_t from_len,
1758 const char *to_s, Py_ssize_t to_len,
1759 Py_ssize_t maxcount)
1760{
1761 char *result_s, *start, *end;
1762 char *self_s;
1763 Py_ssize_t self_len, offset;
1764 PyByteArrayObject *result;
1765
1766 /* The result bytes will be the same size */
1767
1768 self_s = PyByteArray_AS_STRING(self);
1769 self_len = PyByteArray_GET_SIZE(self);
1770
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001771 offset = stringlib_find(self_s, self_len,
1772 from_s, from_len,
1773 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001774 if (offset == -1) {
1775 /* No matches; return the original bytes */
1776 return return_self(self);
1777 }
1778
1779 /* Need to make a new bytes */
1780 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1781 if (result == NULL)
1782 return NULL;
1783 result_s = PyByteArray_AS_STRING(result);
1784 Py_MEMCPY(result_s, self_s, self_len);
1785
1786 /* change everything in-place, starting with this one */
1787 start = result_s + offset;
1788 Py_MEMCPY(start, to_s, from_len);
1789 start += from_len;
1790 end = result_s + self_len;
1791
1792 while ( --maxcount > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001793 offset = stringlib_find(start, end-start,
1794 from_s, from_len,
1795 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001796 if (offset==-1)
1797 break;
1798 Py_MEMCPY(start+offset, to_s, from_len);
1799 start += offset+from_len;
1800 }
1801
1802 return result;
1803}
1804
1805/* len(self)>=1, len(from)==1, len(to)>=2, maxcount>=1 */
1806Py_LOCAL(PyByteArrayObject *)
1807replace_single_character(PyByteArrayObject *self,
1808 char from_c,
1809 const char *to_s, Py_ssize_t to_len,
1810 Py_ssize_t maxcount)
1811{
1812 char *self_s, *result_s;
1813 char *start, *next, *end;
1814 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001815 Py_ssize_t count;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001816 PyByteArrayObject *result;
1817
1818 self_s = PyByteArray_AS_STRING(self);
1819 self_len = PyByteArray_GET_SIZE(self);
1820
1821 count = countchar(self_s, self_len, from_c, maxcount);
1822 if (count == 0) {
1823 /* no matches, return unchanged */
1824 return return_self(self);
1825 }
1826
1827 /* use the difference between current and new, hence the "-1" */
1828 /* result_len = self_len + count * (to_len-1) */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001829 assert(count > 0);
1830 if (to_len - 1 > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001831 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1832 return NULL;
1833 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001834 result_len = self_len + count * (to_len - 1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001835
1836 if ( (result = (PyByteArrayObject *)
1837 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1838 return NULL;
1839 result_s = PyByteArray_AS_STRING(result);
1840
1841 start = self_s;
1842 end = self_s + self_len;
1843 while (count-- > 0) {
1844 next = findchar(start, end-start, from_c);
1845 if (next == NULL)
1846 break;
1847
1848 if (next == start) {
1849 /* replace with the 'to' */
1850 Py_MEMCPY(result_s, to_s, to_len);
1851 result_s += to_len;
1852 start += 1;
1853 } else {
1854 /* copy the unchanged old then the 'to' */
1855 Py_MEMCPY(result_s, start, next-start);
1856 result_s += (next-start);
1857 Py_MEMCPY(result_s, to_s, to_len);
1858 result_s += to_len;
1859 start = next+1;
1860 }
1861 }
1862 /* Copy the remainder of the remaining bytes */
1863 Py_MEMCPY(result_s, start, end-start);
1864
1865 return result;
1866}
1867
1868/* len(self)>=1, len(from)>=2, len(to)>=2, maxcount>=1 */
1869Py_LOCAL(PyByteArrayObject *)
1870replace_substring(PyByteArrayObject *self,
1871 const char *from_s, Py_ssize_t from_len,
1872 const char *to_s, Py_ssize_t to_len,
1873 Py_ssize_t maxcount)
1874{
1875 char *self_s, *result_s;
1876 char *start, *next, *end;
1877 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001878 Py_ssize_t count, offset;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001879 PyByteArrayObject *result;
1880
1881 self_s = PyByteArray_AS_STRING(self);
1882 self_len = PyByteArray_GET_SIZE(self);
1883
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001884 count = stringlib_count(self_s, self_len,
1885 from_s, from_len,
1886 maxcount);
1887
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001888 if (count == 0) {
1889 /* no matches, return unchanged */
1890 return return_self(self);
1891 }
1892
1893 /* Check for overflow */
1894 /* result_len = self_len + count * (to_len-from_len) */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001895 assert(count > 0);
1896 if (to_len - from_len > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001897 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1898 return NULL;
1899 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001900 result_len = self_len + count * (to_len - from_len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001901
1902 if ( (result = (PyByteArrayObject *)
1903 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1904 return NULL;
1905 result_s = PyByteArray_AS_STRING(result);
1906
1907 start = self_s;
1908 end = self_s + self_len;
1909 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001910 offset = stringlib_find(start, end-start,
1911 from_s, from_len,
1912 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001913 if (offset == -1)
1914 break;
1915 next = start+offset;
1916 if (next == start) {
1917 /* replace with the 'to' */
1918 Py_MEMCPY(result_s, to_s, to_len);
1919 result_s += to_len;
1920 start += from_len;
1921 } else {
1922 /* copy the unchanged old then the 'to' */
1923 Py_MEMCPY(result_s, start, next-start);
1924 result_s += (next-start);
1925 Py_MEMCPY(result_s, to_s, to_len);
1926 result_s += to_len;
1927 start = next+from_len;
1928 }
1929 }
1930 /* Copy the remainder of the remaining bytes */
1931 Py_MEMCPY(result_s, start, end-start);
1932
1933 return result;
1934}
1935
1936
1937Py_LOCAL(PyByteArrayObject *)
1938replace(PyByteArrayObject *self,
1939 const char *from_s, Py_ssize_t from_len,
1940 const char *to_s, Py_ssize_t to_len,
1941 Py_ssize_t maxcount)
1942{
1943 if (maxcount < 0) {
1944 maxcount = PY_SSIZE_T_MAX;
1945 } else if (maxcount == 0 || PyByteArray_GET_SIZE(self) == 0) {
1946 /* nothing to do; return the original bytes */
1947 return return_self(self);
1948 }
1949
1950 if (maxcount == 0 ||
1951 (from_len == 0 && to_len == 0)) {
1952 /* nothing to do; return the original bytes */
1953 return return_self(self);
1954 }
1955
1956 /* Handle zero-length special cases */
1957
1958 if (from_len == 0) {
1959 /* insert the 'to' bytes everywhere. */
1960 /* >>> "Python".replace("", ".") */
1961 /* '.P.y.t.h.o.n.' */
1962 return replace_interleave(self, to_s, to_len, maxcount);
1963 }
1964
1965 /* Except for "".replace("", "A") == "A" there is no way beyond this */
1966 /* point for an empty self bytes to generate a non-empty bytes */
1967 /* Special case so the remaining code always gets a non-empty bytes */
1968 if (PyByteArray_GET_SIZE(self) == 0) {
1969 return return_self(self);
1970 }
1971
1972 if (to_len == 0) {
Georg Brandl17cb8a82008-05-30 08:20:09 +00001973 /* delete all occurrences of 'from' bytes */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001974 if (from_len == 1) {
1975 return replace_delete_single_character(
1976 self, from_s[0], maxcount);
1977 } else {
1978 return replace_delete_substring(self, from_s, from_len, maxcount);
1979 }
1980 }
1981
1982 /* Handle special case where both bytes have the same length */
1983
1984 if (from_len == to_len) {
1985 if (from_len == 1) {
1986 return replace_single_character_in_place(
1987 self,
1988 from_s[0],
1989 to_s[0],
1990 maxcount);
1991 } else {
1992 return replace_substring_in_place(
1993 self, from_s, from_len, to_s, to_len, maxcount);
1994 }
1995 }
1996
1997 /* Otherwise use the more generic algorithms */
1998 if (from_len == 1) {
1999 return replace_single_character(self, from_s[0],
2000 to_s, to_len, maxcount);
2001 } else {
2002 /* len('from')>=2, len('to')>=1 */
2003 return replace_substring(self, from_s, from_len, to_s, to_len, maxcount);
2004 }
2005}
2006
2007
2008PyDoc_STRVAR(replace__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002009"B.replace(old, new[, count]) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002010\n\
2011Return a copy of B with all occurrences of subsection\n\
2012old replaced by new. If the optional argument count is\n\
2013given, only the first count occurrences are replaced.");
2014
2015static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002016bytearray_replace(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002017{
2018 Py_ssize_t count = -1;
2019 PyObject *from, *to, *res;
2020 Py_buffer vfrom, vto;
2021
2022 if (!PyArg_ParseTuple(args, "OO|n:replace", &from, &to, &count))
2023 return NULL;
2024
2025 if (_getbuffer(from, &vfrom) < 0)
2026 return NULL;
2027 if (_getbuffer(to, &vto) < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +00002028 PyBuffer_Release(&vfrom);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002029 return NULL;
2030 }
2031
2032 res = (PyObject *)replace((PyByteArrayObject *) self,
2033 vfrom.buf, vfrom.len,
2034 vto.buf, vto.len, count);
2035
Martin v. Löwis423be952008-08-13 15:53:07 +00002036 PyBuffer_Release(&vfrom);
2037 PyBuffer_Release(&vto);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002038 return res;
2039}
2040
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002041PyDoc_STRVAR(split__doc__,
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002042"B.split(sep=None, maxsplit=-1) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002043\n\
2044Return a list of the sections in B, using sep as the delimiter.\n\
2045If sep is not given, B is split on ASCII whitespace characters\n\
2046(space, tab, return, newline, formfeed, vertical tab).\n\
2047If maxsplit is given, at most maxsplit splits are done.");
2048
2049static PyObject *
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002050bytearray_split(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002051{
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002052 static char *kwlist[] = {"sep", "maxsplit", 0};
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002053 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2054 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002055 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002056 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002057 Py_buffer vsub;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002058
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002059 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|On:split",
2060 kwlist, &subobj, &maxsplit))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002061 return NULL;
2062 if (maxsplit < 0)
2063 maxsplit = PY_SSIZE_T_MAX;
2064
2065 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002066 return stringlib_split_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002067
2068 if (_getbuffer(subobj, &vsub) < 0)
2069 return NULL;
2070 sub = vsub.buf;
2071 n = vsub.len;
2072
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002073 list = stringlib_split(
2074 (PyObject*) self, s, len, sub, n, maxsplit
2075 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002076 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002077 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002078}
2079
2080PyDoc_STRVAR(partition__doc__,
2081"B.partition(sep) -> (head, sep, tail)\n\
2082\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002083Search for the separator sep in B, and return the part before it,\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002084the separator itself, and the part after it. If the separator is not\n\
2085found, returns B and two empty bytearray objects.");
2086
2087static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002088bytearray_partition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002089{
2090 PyObject *bytesep, *result;
2091
2092 bytesep = PyByteArray_FromObject(sep_obj);
2093 if (! bytesep)
2094 return NULL;
2095
2096 result = stringlib_partition(
2097 (PyObject*) self,
2098 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2099 bytesep,
2100 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2101 );
2102
2103 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002104 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002105}
2106
2107PyDoc_STRVAR(rpartition__doc__,
Ezio Melotti5b2b2422010-01-25 11:58:28 +00002108"B.rpartition(sep) -> (head, sep, tail)\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002109\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002110Search for the separator sep in B, starting at the end of B,\n\
2111and return the part before it, the separator itself, and the\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002112part after it. If the separator is not found, returns two empty\n\
2113bytearray objects and B.");
2114
2115static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002116bytearray_rpartition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002117{
2118 PyObject *bytesep, *result;
2119
2120 bytesep = PyByteArray_FromObject(sep_obj);
2121 if (! bytesep)
2122 return NULL;
2123
2124 result = stringlib_rpartition(
2125 (PyObject*) self,
2126 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2127 bytesep,
2128 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2129 );
2130
2131 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002132 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002133}
2134
2135PyDoc_STRVAR(rsplit__doc__,
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002136"B.rsplit(sep=None, maxsplit=-1) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002137\n\
2138Return a list of the sections in B, using sep as the delimiter,\n\
2139starting at the end of B and working to the front.\n\
2140If sep is not given, B is split on ASCII whitespace characters\n\
2141(space, tab, return, newline, formfeed, vertical tab).\n\
2142If maxsplit is given, at most maxsplit splits are done.");
2143
2144static PyObject *
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002145bytearray_rsplit(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002146{
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002147 static char *kwlist[] = {"sep", "maxsplit", 0};
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002148 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2149 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002150 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002151 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002152 Py_buffer vsub;
2153
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002154 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|On:rsplit",
2155 kwlist, &subobj, &maxsplit))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002156 return NULL;
2157 if (maxsplit < 0)
2158 maxsplit = PY_SSIZE_T_MAX;
2159
2160 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002161 return stringlib_rsplit_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002162
2163 if (_getbuffer(subobj, &vsub) < 0)
2164 return NULL;
2165 sub = vsub.buf;
2166 n = vsub.len;
2167
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002168 list = stringlib_rsplit(
2169 (PyObject*) self, s, len, sub, n, maxsplit
2170 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002171 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002172 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002173}
2174
2175PyDoc_STRVAR(reverse__doc__,
2176"B.reverse() -> None\n\
2177\n\
2178Reverse the order of the values in B in place.");
2179static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002180bytearray_reverse(PyByteArrayObject *self, PyObject *unused)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002181{
2182 char swap, *head, *tail;
2183 Py_ssize_t i, j, n = Py_SIZE(self);
2184
2185 j = n / 2;
2186 head = self->ob_bytes;
2187 tail = head + n - 1;
2188 for (i = 0; i < j; i++) {
2189 swap = *head;
2190 *head++ = *tail;
2191 *tail-- = swap;
2192 }
2193
2194 Py_RETURN_NONE;
2195}
2196
2197PyDoc_STRVAR(insert__doc__,
2198"B.insert(index, int) -> None\n\
2199\n\
2200Insert a single item into the bytearray before the given index.");
2201static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002202bytearray_insert(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002203{
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002204 PyObject *value;
2205 int ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002206 Py_ssize_t where, n = Py_SIZE(self);
2207
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002208 if (!PyArg_ParseTuple(args, "nO:insert", &where, &value))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002209 return NULL;
2210
2211 if (n == PY_SSIZE_T_MAX) {
2212 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002213 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002214 return NULL;
2215 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002216 if (!_getbytevalue(value, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002217 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002218 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2219 return NULL;
2220
2221 if (where < 0) {
2222 where += n;
2223 if (where < 0)
2224 where = 0;
2225 }
2226 if (where > n)
2227 where = n;
2228 memmove(self->ob_bytes + where + 1, self->ob_bytes + where, n - where);
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002229 self->ob_bytes[where] = ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002230
2231 Py_RETURN_NONE;
2232}
2233
2234PyDoc_STRVAR(append__doc__,
2235"B.append(int) -> None\n\
2236\n\
2237Append a single item to the end of B.");
2238static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002239bytearray_append(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002240{
2241 int value;
2242 Py_ssize_t n = Py_SIZE(self);
2243
2244 if (! _getbytevalue(arg, &value))
2245 return NULL;
2246 if (n == PY_SSIZE_T_MAX) {
2247 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002248 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002249 return NULL;
2250 }
2251 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2252 return NULL;
2253
2254 self->ob_bytes[n] = value;
2255
2256 Py_RETURN_NONE;
2257}
2258
2259PyDoc_STRVAR(extend__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002260"B.extend(iterable_of_ints) -> None\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002261\n\
2262Append all the elements from the iterator or sequence to the\n\
2263end of B.");
2264static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002265bytearray_extend(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002266{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002267 PyObject *it, *item, *bytearray_obj;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002268 Py_ssize_t buf_size = 0, len = 0;
2269 int value;
2270 char *buf;
2271
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002272 /* bytearray_setslice code only accepts something supporting PEP 3118. */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002273 if (PyObject_CheckBuffer(arg)) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002274 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), arg) == -1)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002275 return NULL;
2276
2277 Py_RETURN_NONE;
2278 }
2279
2280 it = PyObject_GetIter(arg);
2281 if (it == NULL)
2282 return NULL;
2283
Ezio Melotti42da6632011-03-15 05:18:48 +02002284 /* Try to determine the length of the argument. 32 is arbitrary. */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002285 buf_size = _PyObject_LengthHint(arg, 32);
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002286 if (buf_size == -1) {
2287 Py_DECREF(it);
2288 return NULL;
2289 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002290
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002291 bytearray_obj = PyByteArray_FromStringAndSize(NULL, buf_size);
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002292 if (bytearray_obj == NULL) {
2293 Py_DECREF(it);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002294 return NULL;
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002295 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002296 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002297
2298 while ((item = PyIter_Next(it)) != NULL) {
2299 if (! _getbytevalue(item, &value)) {
2300 Py_DECREF(item);
2301 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002302 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002303 return NULL;
2304 }
2305 buf[len++] = value;
2306 Py_DECREF(item);
2307
2308 if (len >= buf_size) {
2309 buf_size = len + (len >> 1) + 1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002310 if (PyByteArray_Resize((PyObject *)bytearray_obj, buf_size) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002311 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002312 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002313 return NULL;
2314 }
2315 /* Recompute the `buf' pointer, since the resizing operation may
2316 have invalidated it. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002317 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002318 }
2319 }
2320 Py_DECREF(it);
2321
2322 /* Resize down to exact size. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002323 if (PyByteArray_Resize((PyObject *)bytearray_obj, len) < 0) {
2324 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002325 return NULL;
2326 }
2327
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002328 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), bytearray_obj) == -1) {
2329 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002330 return NULL;
Antoine Pitrou58bb82e2012-04-01 16:05:46 +02002331 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002332 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002333
2334 Py_RETURN_NONE;
2335}
2336
2337PyDoc_STRVAR(pop__doc__,
2338"B.pop([index]) -> int\n\
2339\n\
2340Remove and return a single item from B. If no index\n\
Benjamin Petersondcf97b92008-07-02 17:30:14 +00002341argument is given, will pop the last value.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002342static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002343bytearray_pop(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002344{
2345 int value;
2346 Py_ssize_t where = -1, n = Py_SIZE(self);
2347
2348 if (!PyArg_ParseTuple(args, "|n:pop", &where))
2349 return NULL;
2350
2351 if (n == 0) {
Eli Bendersky1bc4f192011-03-04 04:55:25 +00002352 PyErr_SetString(PyExc_IndexError,
2353 "pop from empty bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002354 return NULL;
2355 }
2356 if (where < 0)
2357 where += Py_SIZE(self);
2358 if (where < 0 || where >= Py_SIZE(self)) {
2359 PyErr_SetString(PyExc_IndexError, "pop index out of range");
2360 return NULL;
2361 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002362 if (!_canresize(self))
2363 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002364
2365 value = self->ob_bytes[where];
2366 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2367 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2368 return NULL;
2369
Mark Dickinson54a3db92009-09-06 10:19:23 +00002370 return PyLong_FromLong((unsigned char)value);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002371}
2372
2373PyDoc_STRVAR(remove__doc__,
2374"B.remove(int) -> None\n\
2375\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002376Remove the first occurrence of a value in B.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002377static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002378bytearray_remove(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002379{
2380 int value;
2381 Py_ssize_t where, n = Py_SIZE(self);
2382
2383 if (! _getbytevalue(arg, &value))
2384 return NULL;
2385
2386 for (where = 0; where < n; where++) {
2387 if (self->ob_bytes[where] == value)
2388 break;
2389 }
2390 if (where == n) {
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002391 PyErr_SetString(PyExc_ValueError, "value not found in bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002392 return NULL;
2393 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002394 if (!_canresize(self))
2395 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002396
2397 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2398 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2399 return NULL;
2400
2401 Py_RETURN_NONE;
2402}
2403
2404/* XXX These two helpers could be optimized if argsize == 1 */
2405
2406static Py_ssize_t
2407lstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2408 void *argptr, Py_ssize_t argsize)
2409{
2410 Py_ssize_t i = 0;
2411 while (i < mysize && memchr(argptr, myptr[i], argsize))
2412 i++;
2413 return i;
2414}
2415
2416static Py_ssize_t
2417rstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2418 void *argptr, Py_ssize_t argsize)
2419{
2420 Py_ssize_t i = mysize - 1;
2421 while (i >= 0 && memchr(argptr, myptr[i], argsize))
2422 i--;
2423 return i + 1;
2424}
2425
2426PyDoc_STRVAR(strip__doc__,
2427"B.strip([bytes]) -> bytearray\n\
2428\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002429Strip leading and trailing bytes contained in the argument\n\
2430and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002431If the argument is omitted, strip ASCII whitespace.");
2432static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002433bytearray_strip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002434{
2435 Py_ssize_t left, right, mysize, argsize;
2436 void *myptr, *argptr;
2437 PyObject *arg = Py_None;
2438 Py_buffer varg;
2439 if (!PyArg_ParseTuple(args, "|O:strip", &arg))
2440 return NULL;
2441 if (arg == Py_None) {
2442 argptr = "\t\n\r\f\v ";
2443 argsize = 6;
2444 }
2445 else {
2446 if (_getbuffer(arg, &varg) < 0)
2447 return NULL;
2448 argptr = varg.buf;
2449 argsize = varg.len;
2450 }
2451 myptr = self->ob_bytes;
2452 mysize = Py_SIZE(self);
2453 left = lstrip_helper(myptr, mysize, argptr, argsize);
2454 if (left == mysize)
2455 right = left;
2456 else
2457 right = rstrip_helper(myptr, mysize, argptr, argsize);
2458 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002459 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002460 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2461}
2462
2463PyDoc_STRVAR(lstrip__doc__,
2464"B.lstrip([bytes]) -> bytearray\n\
2465\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002466Strip leading bytes contained in the argument\n\
2467and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002468If the argument is omitted, strip leading ASCII whitespace.");
2469static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002470bytearray_lstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002471{
2472 Py_ssize_t left, right, mysize, argsize;
2473 void *myptr, *argptr;
2474 PyObject *arg = Py_None;
2475 Py_buffer varg;
2476 if (!PyArg_ParseTuple(args, "|O:lstrip", &arg))
2477 return NULL;
2478 if (arg == Py_None) {
2479 argptr = "\t\n\r\f\v ";
2480 argsize = 6;
2481 }
2482 else {
2483 if (_getbuffer(arg, &varg) < 0)
2484 return NULL;
2485 argptr = varg.buf;
2486 argsize = varg.len;
2487 }
2488 myptr = self->ob_bytes;
2489 mysize = Py_SIZE(self);
2490 left = lstrip_helper(myptr, mysize, argptr, argsize);
2491 right = mysize;
2492 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002493 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002494 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2495}
2496
2497PyDoc_STRVAR(rstrip__doc__,
2498"B.rstrip([bytes]) -> bytearray\n\
2499\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002500Strip trailing bytes contained in the argument\n\
2501and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002502If the argument is omitted, strip trailing ASCII whitespace.");
2503static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002504bytearray_rstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002505{
Brett Cannonb94767f2011-02-22 20:15:44 +00002506 Py_ssize_t right, mysize, argsize;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002507 void *myptr, *argptr;
2508 PyObject *arg = Py_None;
2509 Py_buffer varg;
2510 if (!PyArg_ParseTuple(args, "|O:rstrip", &arg))
2511 return NULL;
2512 if (arg == Py_None) {
2513 argptr = "\t\n\r\f\v ";
2514 argsize = 6;
2515 }
2516 else {
2517 if (_getbuffer(arg, &varg) < 0)
2518 return NULL;
2519 argptr = varg.buf;
2520 argsize = varg.len;
2521 }
2522 myptr = self->ob_bytes;
2523 mysize = Py_SIZE(self);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002524 right = rstrip_helper(myptr, mysize, argptr, argsize);
2525 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002526 PyBuffer_Release(&varg);
Brett Cannonb94767f2011-02-22 20:15:44 +00002527 return PyByteArray_FromStringAndSize(self->ob_bytes, right);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002528}
2529
2530PyDoc_STRVAR(decode_doc,
Victor Stinnerc911bbf2010-11-07 19:04:46 +00002531"B.decode(encoding='utf-8', errors='strict') -> str\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002532\n\
Victor Stinnere14e2122010-11-07 18:41:46 +00002533Decode B using the codec registered for encoding. Default encoding\n\
2534is 'utf-8'. errors may be given to set a different error\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002535handling scheme. Default is 'strict' meaning that encoding errors raise\n\
2536a UnicodeDecodeError. Other possible values are 'ignore' and 'replace'\n\
2537as well as any other name registered with codecs.register_error that is\n\
2538able to handle UnicodeDecodeErrors.");
2539
2540static PyObject *
Benjamin Peterson308d6372009-09-18 21:42:35 +00002541bytearray_decode(PyObject *self, PyObject *args, PyObject *kwargs)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002542{
2543 const char *encoding = NULL;
2544 const char *errors = NULL;
Benjamin Peterson308d6372009-09-18 21:42:35 +00002545 static char *kwlist[] = {"encoding", "errors", 0};
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002546
Benjamin Peterson308d6372009-09-18 21:42:35 +00002547 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ss:decode", kwlist, &encoding, &errors))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002548 return NULL;
2549 if (encoding == NULL)
2550 encoding = PyUnicode_GetDefaultEncoding();
Marc-André Lemburgb2750b52008-06-06 12:18:17 +00002551 return PyUnicode_FromEncodedObject(self, encoding, errors);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002552}
2553
2554PyDoc_STRVAR(alloc_doc,
2555"B.__alloc__() -> int\n\
2556\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002557Return the number of bytes actually allocated.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002558
2559static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002560bytearray_alloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002561{
2562 return PyLong_FromSsize_t(self->ob_alloc);
2563}
2564
2565PyDoc_STRVAR(join_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002566"B.join(iterable_of_bytes) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002567\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002568Concatenate any number of bytes/bytearray objects, with B\n\
2569in between each pair, and return the result as a new bytearray.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002570
2571static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002572bytearray_join(PyByteArrayObject *self, PyObject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002573{
2574 PyObject *seq;
2575 Py_ssize_t mysize = Py_SIZE(self);
2576 Py_ssize_t i;
2577 Py_ssize_t n;
2578 PyObject **items;
2579 Py_ssize_t totalsize = 0;
2580 PyObject *result;
2581 char *dest;
2582
2583 seq = PySequence_Fast(it, "can only join an iterable");
2584 if (seq == NULL)
2585 return NULL;
2586 n = PySequence_Fast_GET_SIZE(seq);
2587 items = PySequence_Fast_ITEMS(seq);
2588
2589 /* Compute the total size, and check that they are all bytes */
2590 /* XXX Shouldn't we use _getbuffer() on these items instead? */
2591 for (i = 0; i < n; i++) {
2592 PyObject *obj = items[i];
2593 if (!PyByteArray_Check(obj) && !PyBytes_Check(obj)) {
2594 PyErr_Format(PyExc_TypeError,
2595 "can only join an iterable of bytes "
2596 "(item %ld has type '%.100s')",
2597 /* XXX %ld isn't right on Win64 */
2598 (long)i, Py_TYPE(obj)->tp_name);
2599 goto error;
2600 }
2601 if (i > 0)
2602 totalsize += mysize;
2603 totalsize += Py_SIZE(obj);
2604 if (totalsize < 0) {
2605 PyErr_NoMemory();
2606 goto error;
2607 }
2608 }
2609
2610 /* Allocate the result, and copy the bytes */
2611 result = PyByteArray_FromStringAndSize(NULL, totalsize);
2612 if (result == NULL)
2613 goto error;
2614 dest = PyByteArray_AS_STRING(result);
2615 for (i = 0; i < n; i++) {
2616 PyObject *obj = items[i];
2617 Py_ssize_t size = Py_SIZE(obj);
2618 char *buf;
2619 if (PyByteArray_Check(obj))
2620 buf = PyByteArray_AS_STRING(obj);
2621 else
2622 buf = PyBytes_AS_STRING(obj);
2623 if (i) {
2624 memcpy(dest, self->ob_bytes, mysize);
2625 dest += mysize;
2626 }
2627 memcpy(dest, buf, size);
2628 dest += size;
2629 }
2630
2631 /* Done */
2632 Py_DECREF(seq);
2633 return result;
2634
2635 /* Error handling */
2636 error:
2637 Py_DECREF(seq);
2638 return NULL;
2639}
2640
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002641PyDoc_STRVAR(splitlines__doc__,
2642"B.splitlines([keepends]) -> list of lines\n\
2643\n\
2644Return a list of the lines in B, breaking at line boundaries.\n\
2645Line breaks are not included in the resulting list unless keepends\n\
2646is given and true.");
2647
2648static PyObject*
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002649bytearray_splitlines(PyObject *self, PyObject *args, PyObject *kwds)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002650{
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002651 static char *kwlist[] = {"keepends", 0};
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002652 int keepends = 0;
2653
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002654 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|i:splitlines",
2655 kwlist, &keepends))
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002656 return NULL;
2657
2658 return stringlib_splitlines(
2659 (PyObject*) self, PyByteArray_AS_STRING(self),
2660 PyByteArray_GET_SIZE(self), keepends
2661 );
2662}
2663
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002664PyDoc_STRVAR(fromhex_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002665"bytearray.fromhex(string) -> bytearray (static method)\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002666\n\
2667Create a bytearray object from a string of hexadecimal numbers.\n\
2668Spaces between two numbers are accepted.\n\
2669Example: bytearray.fromhex('B9 01EF') -> bytearray(b'\\xb9\\x01\\xef').");
2670
2671static int
Victor Stinner6430fd52011-09-29 04:02:13 +02002672hex_digit_to_int(Py_UCS4 c)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002673{
2674 if (c >= 128)
2675 return -1;
Eric Smith6dc46f52009-04-27 20:39:49 +00002676 if (Py_ISDIGIT(c))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002677 return c - '0';
2678 else {
Eric Smith6dc46f52009-04-27 20:39:49 +00002679 if (Py_ISUPPER(c))
2680 c = Py_TOLOWER(c);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002681 if (c >= 'a' && c <= 'f')
2682 return c - 'a' + 10;
2683 }
2684 return -1;
2685}
2686
2687static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002688bytearray_fromhex(PyObject *cls, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002689{
2690 PyObject *newbytes, *hexobj;
2691 char *buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002692 Py_ssize_t hexlen, byteslen, i, j;
2693 int top, bot;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002694 void *data;
2695 unsigned int kind;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002696
2697 if (!PyArg_ParseTuple(args, "U:fromhex", &hexobj))
2698 return NULL;
2699 assert(PyUnicode_Check(hexobj));
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002700 if (PyUnicode_READY(hexobj))
2701 return NULL;
2702 kind = PyUnicode_KIND(hexobj);
2703 data = PyUnicode_DATA(hexobj);
2704 hexlen = PyUnicode_GET_LENGTH(hexobj);
2705
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002706 byteslen = hexlen/2; /* This overestimates if there are spaces */
2707 newbytes = PyByteArray_FromStringAndSize(NULL, byteslen);
2708 if (!newbytes)
2709 return NULL;
2710 buf = PyByteArray_AS_STRING(newbytes);
2711 for (i = j = 0; i < hexlen; i += 2) {
2712 /* skip over spaces in the input */
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002713 while (PyUnicode_READ(kind, data, i) == ' ')
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002714 i++;
2715 if (i >= hexlen)
2716 break;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02002717 top = hex_digit_to_int(PyUnicode_READ(kind, data, i));
2718 bot = hex_digit_to_int(PyUnicode_READ(kind, data, i+1));
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002719 if (top == -1 || bot == -1) {
2720 PyErr_Format(PyExc_ValueError,
2721 "non-hexadecimal number found in "
2722 "fromhex() arg at position %zd", i);
2723 goto error;
2724 }
2725 buf[j++] = (top << 4) + bot;
2726 }
2727 if (PyByteArray_Resize(newbytes, j) < 0)
2728 goto error;
2729 return newbytes;
2730
2731 error:
2732 Py_DECREF(newbytes);
2733 return NULL;
2734}
2735
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002736
2737static PyObject *
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002738_common_reduce(PyByteArrayObject *self, int proto)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002739{
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002740 PyObject *dict;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02002741 _Py_IDENTIFIER(__dict__);
Martin v. Löwis1ee1b6f2011-10-10 18:11:30 +02002742
Martin v. Löwis1ee1b6f2011-10-10 18:11:30 +02002743 dict = _PyObject_GetAttrId((PyObject *)self, &PyId___dict__);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002744 if (dict == NULL) {
2745 PyErr_Clear();
2746 dict = Py_None;
2747 Py_INCREF(dict);
2748 }
2749
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002750 if (proto < 3) {
2751 /* use str based reduction for backwards compatibility with Python 2.x */
2752 PyObject *latin1;
2753 if (self->ob_bytes)
2754 latin1 = PyUnicode_DecodeLatin1(self->ob_bytes, Py_SIZE(self), NULL);
2755 else
2756 latin1 = PyUnicode_FromString("");
2757 return Py_BuildValue("(O(Ns)N)", Py_TYPE(self), latin1, "latin-1", dict);
2758 }
2759 else {
2760 /* use more efficient byte based reduction */
2761 if (self->ob_bytes) {
2762 return Py_BuildValue("(O(y#)N)", Py_TYPE(self), self->ob_bytes, Py_SIZE(self), dict);
2763 }
2764 else {
2765 return Py_BuildValue("(O()N)", Py_TYPE(self), dict);
2766 }
2767 }
2768}
2769
2770PyDoc_STRVAR(reduce_doc, "Return state information for pickling.");
2771
2772static PyObject *
2773bytearray_reduce(PyByteArrayObject *self)
2774{
2775 return _common_reduce(self, 2);
2776}
2777
2778PyDoc_STRVAR(reduce_ex_doc, "Return state information for pickling.");
2779
2780static PyObject *
2781bytearray_reduce_ex(PyByteArrayObject *self, PyObject *args)
2782{
2783 int proto = 0;
2784
2785 if (!PyArg_ParseTuple(args, "|i:__reduce_ex__", &proto))
2786 return NULL;
2787
2788 return _common_reduce(self, proto);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002789}
2790
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002791PyDoc_STRVAR(sizeof_doc,
2792"B.__sizeof__() -> int\n\
2793 \n\
2794Returns the size of B in memory, in bytes");
2795static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002796bytearray_sizeof(PyByteArrayObject *self)
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002797{
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002798 Py_ssize_t res;
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002799
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002800 res = sizeof(PyByteArrayObject) + self->ob_alloc * sizeof(char);
2801 return PyLong_FromSsize_t(res);
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002802}
2803
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002804static PySequenceMethods bytearray_as_sequence = {
2805 (lenfunc)bytearray_length, /* sq_length */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002806 (binaryfunc)PyByteArray_Concat, /* sq_concat */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002807 (ssizeargfunc)bytearray_repeat, /* sq_repeat */
2808 (ssizeargfunc)bytearray_getitem, /* sq_item */
2809 0, /* sq_slice */
2810 (ssizeobjargproc)bytearray_setitem, /* sq_ass_item */
2811 0, /* sq_ass_slice */
2812 (objobjproc)bytearray_contains, /* sq_contains */
2813 (binaryfunc)bytearray_iconcat, /* sq_inplace_concat */
2814 (ssizeargfunc)bytearray_irepeat, /* sq_inplace_repeat */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002815};
2816
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002817static PyMappingMethods bytearray_as_mapping = {
2818 (lenfunc)bytearray_length,
2819 (binaryfunc)bytearray_subscript,
2820 (objobjargproc)bytearray_ass_subscript,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002821};
2822
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002823static PyBufferProcs bytearray_as_buffer = {
2824 (getbufferproc)bytearray_getbuffer,
2825 (releasebufferproc)bytearray_releasebuffer,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002826};
2827
2828static PyMethodDef
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002829bytearray_methods[] = {
2830 {"__alloc__", (PyCFunction)bytearray_alloc, METH_NOARGS, alloc_doc},
2831 {"__reduce__", (PyCFunction)bytearray_reduce, METH_NOARGS, reduce_doc},
Antoine Pitroub0e1f8b2011-12-05 20:40:08 +01002832 {"__reduce_ex__", (PyCFunction)bytearray_reduce_ex, METH_VARARGS, reduce_ex_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002833 {"__sizeof__", (PyCFunction)bytearray_sizeof, METH_NOARGS, sizeof_doc},
2834 {"append", (PyCFunction)bytearray_append, METH_O, append__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002835 {"capitalize", (PyCFunction)stringlib_capitalize, METH_NOARGS,
2836 _Py_capitalize__doc__},
2837 {"center", (PyCFunction)stringlib_center, METH_VARARGS, center__doc__},
Eli Bendersky4db28d32011-03-03 18:21:02 +00002838 {"clear", (PyCFunction)bytearray_clear, METH_NOARGS, clear__doc__},
2839 {"copy", (PyCFunction)bytearray_copy, METH_NOARGS, copy__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002840 {"count", (PyCFunction)bytearray_count, METH_VARARGS, count__doc__},
Benjamin Peterson308d6372009-09-18 21:42:35 +00002841 {"decode", (PyCFunction)bytearray_decode, METH_VARARGS | METH_KEYWORDS, decode_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002842 {"endswith", (PyCFunction)bytearray_endswith, METH_VARARGS, endswith__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002843 {"expandtabs", (PyCFunction)stringlib_expandtabs, METH_VARARGS,
2844 expandtabs__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002845 {"extend", (PyCFunction)bytearray_extend, METH_O, extend__doc__},
2846 {"find", (PyCFunction)bytearray_find, METH_VARARGS, find__doc__},
2847 {"fromhex", (PyCFunction)bytearray_fromhex, METH_VARARGS|METH_CLASS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002848 fromhex_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002849 {"index", (PyCFunction)bytearray_index, METH_VARARGS, index__doc__},
2850 {"insert", (PyCFunction)bytearray_insert, METH_VARARGS, insert__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002851 {"isalnum", (PyCFunction)stringlib_isalnum, METH_NOARGS,
2852 _Py_isalnum__doc__},
2853 {"isalpha", (PyCFunction)stringlib_isalpha, METH_NOARGS,
2854 _Py_isalpha__doc__},
2855 {"isdigit", (PyCFunction)stringlib_isdigit, METH_NOARGS,
2856 _Py_isdigit__doc__},
2857 {"islower", (PyCFunction)stringlib_islower, METH_NOARGS,
2858 _Py_islower__doc__},
2859 {"isspace", (PyCFunction)stringlib_isspace, METH_NOARGS,
2860 _Py_isspace__doc__},
2861 {"istitle", (PyCFunction)stringlib_istitle, METH_NOARGS,
2862 _Py_istitle__doc__},
2863 {"isupper", (PyCFunction)stringlib_isupper, METH_NOARGS,
2864 _Py_isupper__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002865 {"join", (PyCFunction)bytearray_join, METH_O, join_doc},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002866 {"ljust", (PyCFunction)stringlib_ljust, METH_VARARGS, ljust__doc__},
2867 {"lower", (PyCFunction)stringlib_lower, METH_NOARGS, _Py_lower__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002868 {"lstrip", (PyCFunction)bytearray_lstrip, METH_VARARGS, lstrip__doc__},
2869 {"maketrans", (PyCFunction)bytearray_maketrans, METH_VARARGS|METH_STATIC,
Georg Brandlabc38772009-04-12 15:51:51 +00002870 _Py_maketrans__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002871 {"partition", (PyCFunction)bytearray_partition, METH_O, partition__doc__},
2872 {"pop", (PyCFunction)bytearray_pop, METH_VARARGS, pop__doc__},
2873 {"remove", (PyCFunction)bytearray_remove, METH_O, remove__doc__},
2874 {"replace", (PyCFunction)bytearray_replace, METH_VARARGS, replace__doc__},
2875 {"reverse", (PyCFunction)bytearray_reverse, METH_NOARGS, reverse__doc__},
2876 {"rfind", (PyCFunction)bytearray_rfind, METH_VARARGS, rfind__doc__},
2877 {"rindex", (PyCFunction)bytearray_rindex, METH_VARARGS, rindex__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002878 {"rjust", (PyCFunction)stringlib_rjust, METH_VARARGS, rjust__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002879 {"rpartition", (PyCFunction)bytearray_rpartition, METH_O, rpartition__doc__},
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002880 {"rsplit", (PyCFunction)bytearray_rsplit, METH_VARARGS | METH_KEYWORDS, rsplit__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002881 {"rstrip", (PyCFunction)bytearray_rstrip, METH_VARARGS, rstrip__doc__},
Ezio Melotticda6b6d2012-02-26 09:39:55 +02002882 {"split", (PyCFunction)bytearray_split, METH_VARARGS | METH_KEYWORDS, split__doc__},
Mark Dickinson0d5f6ad2011-09-24 09:14:39 +01002883 {"splitlines", (PyCFunction)bytearray_splitlines,
2884 METH_VARARGS | METH_KEYWORDS, splitlines__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002885 {"startswith", (PyCFunction)bytearray_startswith, METH_VARARGS ,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002886 startswith__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002887 {"strip", (PyCFunction)bytearray_strip, METH_VARARGS, strip__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002888 {"swapcase", (PyCFunction)stringlib_swapcase, METH_NOARGS,
2889 _Py_swapcase__doc__},
2890 {"title", (PyCFunction)stringlib_title, METH_NOARGS, _Py_title__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002891 {"translate", (PyCFunction)bytearray_translate, METH_VARARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002892 translate__doc__},
2893 {"upper", (PyCFunction)stringlib_upper, METH_NOARGS, _Py_upper__doc__},
2894 {"zfill", (PyCFunction)stringlib_zfill, METH_VARARGS, zfill__doc__},
2895 {NULL}
2896};
2897
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002898PyDoc_STRVAR(bytearray_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002899"bytearray(iterable_of_ints) -> bytearray\n\
2900bytearray(string, encoding[, errors]) -> bytearray\n\
Victor Stinnerbb2e9c42011-12-17 23:18:07 +01002901bytearray(bytes_or_buffer) -> mutable copy of bytes_or_buffer\n\
2902bytearray(int) -> bytes array of size given by the parameter initialized with null bytes\n\
2903bytearray() -> empty bytes array\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002904\n\
2905Construct an mutable bytearray object from:\n\
2906 - an iterable yielding integers in range(256)\n\
2907 - a text string encoded using the specified encoding\n\
Victor Stinnerbb2e9c42011-12-17 23:18:07 +01002908 - a bytes or a buffer object\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002909 - any object implementing the buffer API.\n\
Victor Stinnerbb2e9c42011-12-17 23:18:07 +01002910 - an integer");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002911
2912
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002913static PyObject *bytearray_iter(PyObject *seq);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002914
2915PyTypeObject PyByteArray_Type = {
2916 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2917 "bytearray",
2918 sizeof(PyByteArrayObject),
2919 0,
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002920 (destructor)bytearray_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002921 0, /* tp_print */
2922 0, /* tp_getattr */
2923 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00002924 0, /* tp_reserved */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002925 (reprfunc)bytearray_repr, /* tp_repr */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002926 0, /* tp_as_number */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002927 &bytearray_as_sequence, /* tp_as_sequence */
2928 &bytearray_as_mapping, /* tp_as_mapping */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002929 0, /* tp_hash */
2930 0, /* tp_call */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002931 bytearray_str, /* tp_str */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002932 PyObject_GenericGetAttr, /* tp_getattro */
2933 0, /* tp_setattro */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002934 &bytearray_as_buffer, /* tp_as_buffer */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002935 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /* tp_flags */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002936 bytearray_doc, /* tp_doc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002937 0, /* tp_traverse */
2938 0, /* tp_clear */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002939 (richcmpfunc)bytearray_richcompare, /* tp_richcompare */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002940 0, /* tp_weaklistoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002941 bytearray_iter, /* tp_iter */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002942 0, /* tp_iternext */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002943 bytearray_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002944 0, /* tp_members */
2945 0, /* tp_getset */
2946 0, /* tp_base */
2947 0, /* tp_dict */
2948 0, /* tp_descr_get */
2949 0, /* tp_descr_set */
2950 0, /* tp_dictoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002951 (initproc)bytearray_init, /* tp_init */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002952 PyType_GenericAlloc, /* tp_alloc */
2953 PyType_GenericNew, /* tp_new */
2954 PyObject_Del, /* tp_free */
2955};
2956
2957/*********************** Bytes Iterator ****************************/
2958
2959typedef struct {
2960 PyObject_HEAD
2961 Py_ssize_t it_index;
2962 PyByteArrayObject *it_seq; /* Set to NULL when iterator is exhausted */
2963} bytesiterobject;
2964
2965static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002966bytearrayiter_dealloc(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002967{
2968 _PyObject_GC_UNTRACK(it);
2969 Py_XDECREF(it->it_seq);
2970 PyObject_GC_Del(it);
2971}
2972
2973static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002974bytearrayiter_traverse(bytesiterobject *it, visitproc visit, void *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002975{
2976 Py_VISIT(it->it_seq);
2977 return 0;
2978}
2979
2980static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002981bytearrayiter_next(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002982{
2983 PyByteArrayObject *seq;
2984 PyObject *item;
2985
2986 assert(it != NULL);
2987 seq = it->it_seq;
2988 if (seq == NULL)
2989 return NULL;
2990 assert(PyByteArray_Check(seq));
2991
2992 if (it->it_index < PyByteArray_GET_SIZE(seq)) {
2993 item = PyLong_FromLong(
2994 (unsigned char)seq->ob_bytes[it->it_index]);
2995 if (item != NULL)
2996 ++it->it_index;
2997 return item;
2998 }
2999
3000 Py_DECREF(seq);
3001 it->it_seq = NULL;
3002 return NULL;
3003}
3004
3005static PyObject *
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003006bytearrayiter_length_hint(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003007{
3008 Py_ssize_t len = 0;
3009 if (it->it_seq)
3010 len = PyByteArray_GET_SIZE(it->it_seq) - it->it_index;
3011 return PyLong_FromSsize_t(len);
3012}
3013
3014PyDoc_STRVAR(length_hint_doc,
3015 "Private method returning an estimate of len(list(it)).");
3016
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003017static PyObject *
3018bytearrayiter_reduce(bytesiterobject *it)
3019{
3020 if (it->it_seq != NULL) {
Antoine Pitroua7013882012-04-05 00:04:20 +02003021 return Py_BuildValue("N(O)n", _PyObject_GetBuiltin("iter"),
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003022 it->it_seq, it->it_index);
3023 } else {
3024 PyObject *u = PyUnicode_FromUnicode(NULL, 0);
3025 if (u == NULL)
3026 return NULL;
Antoine Pitroua7013882012-04-05 00:04:20 +02003027 return Py_BuildValue("N(N)", _PyObject_GetBuiltin("iter"), u);
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003028 }
3029}
3030
3031static PyObject *
3032bytearrayiter_setstate(bytesiterobject *it, PyObject *state)
3033{
3034 Py_ssize_t index = PyLong_AsSsize_t(state);
3035 if (index == -1 && PyErr_Occurred())
3036 return NULL;
3037 if (index < 0)
3038 index = 0;
3039 it->it_index = index;
3040 Py_RETURN_NONE;
3041}
3042
3043PyDoc_STRVAR(setstate_doc, "Set state information for unpickling.");
3044
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003045static PyMethodDef bytearrayiter_methods[] = {
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003046 {"__length_hint__", (PyCFunction)bytearrayiter_length_hint, METH_NOARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003047 length_hint_doc},
Kristján Valur Jónsson31668b82012-04-03 10:49:41 +00003048 {"__reduce__", (PyCFunction)bytearrayiter_reduce, METH_NOARGS,
3049 reduce_doc},
3050 {"__setstate__", (PyCFunction)bytearrayiter_setstate, METH_O,
3051 setstate_doc},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003052 {NULL, NULL} /* sentinel */
3053};
3054
3055PyTypeObject PyByteArrayIter_Type = {
3056 PyVarObject_HEAD_INIT(&PyType_Type, 0)
3057 "bytearray_iterator", /* tp_name */
3058 sizeof(bytesiterobject), /* tp_basicsize */
3059 0, /* tp_itemsize */
3060 /* methods */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003061 (destructor)bytearrayiter_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003062 0, /* tp_print */
3063 0, /* tp_getattr */
3064 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00003065 0, /* tp_reserved */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003066 0, /* tp_repr */
3067 0, /* tp_as_number */
3068 0, /* tp_as_sequence */
3069 0, /* tp_as_mapping */
3070 0, /* tp_hash */
3071 0, /* tp_call */
3072 0, /* tp_str */
3073 PyObject_GenericGetAttr, /* tp_getattro */
3074 0, /* tp_setattro */
3075 0, /* tp_as_buffer */
3076 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */
3077 0, /* tp_doc */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003078 (traverseproc)bytearrayiter_traverse, /* tp_traverse */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003079 0, /* tp_clear */
3080 0, /* tp_richcompare */
3081 0, /* tp_weaklistoffset */
3082 PyObject_SelfIter, /* tp_iter */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003083 (iternextfunc)bytearrayiter_next, /* tp_iternext */
3084 bytearrayiter_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003085 0,
3086};
3087
3088static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00003089bytearray_iter(PyObject *seq)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00003090{
3091 bytesiterobject *it;
3092
3093 if (!PyByteArray_Check(seq)) {
3094 PyErr_BadInternalCall();
3095 return NULL;
3096 }
3097 it = PyObject_GC_New(bytesiterobject, &PyByteArrayIter_Type);
3098 if (it == NULL)
3099 return NULL;
3100 it->it_index = 0;
3101 Py_INCREF(seq);
3102 it->it_seq = (PyByteArrayObject *)seq;
3103 _PyObject_GC_TRACK(it);
3104 return (PyObject *)it;
3105}