blob: 7a7410405ec830d75ed7e20bcce0fe34fb8b6935 [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;
142 new->ob_bytes = PyMem_Malloc(alloc);
143 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
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000212 sval = PyMem_Realloc(((PyByteArrayObject *)self)->ob_bytes, alloc);
213 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)
792 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{
853 static const char *hexdigits = "0123456789abcdef";
854 const char *quote_prefix = "bytearray(b";
855 const char *quote_postfix = ")";
856 Py_ssize_t length = Py_SIZE(self);
857 /* 14 == strlen(quote_prefix) + 2 + strlen(quote_postfix) */
Mark Dickinson66f575b2010-02-14 12:53:32 +0000858 size_t newsize;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000859 PyObject *v;
Mark Dickinson66f575b2010-02-14 12:53:32 +0000860 if (length > (PY_SSIZE_T_MAX - 14) / 4) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000861 PyErr_SetString(PyExc_OverflowError,
862 "bytearray object is too large to make repr");
863 return NULL;
864 }
Mark Dickinson66f575b2010-02-14 12:53:32 +0000865 newsize = 14 + 4 * length;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000866 v = PyUnicode_FromUnicode(NULL, newsize);
867 if (v == NULL) {
868 return NULL;
869 }
870 else {
871 register Py_ssize_t i;
872 register Py_UNICODE c;
873 register Py_UNICODE *p;
874 int quote;
875
876 /* Figure out which quote to use; single is preferred */
877 quote = '\'';
878 {
879 char *test, *start;
880 start = PyByteArray_AS_STRING(self);
881 for (test = start; test < start+length; ++test) {
882 if (*test == '"') {
883 quote = '\''; /* back to single */
884 goto decided;
885 }
886 else if (*test == '\'')
887 quote = '"';
888 }
889 decided:
890 ;
891 }
892
893 p = PyUnicode_AS_UNICODE(v);
894 while (*quote_prefix)
895 *p++ = *quote_prefix++;
896 *p++ = quote;
897
898 for (i = 0; i < length; i++) {
899 /* There's at least enough room for a hex escape
900 and a closing quote. */
901 assert(newsize - (p - PyUnicode_AS_UNICODE(v)) >= 5);
902 c = self->ob_bytes[i];
903 if (c == '\'' || c == '\\')
904 *p++ = '\\', *p++ = c;
905 else if (c == '\t')
906 *p++ = '\\', *p++ = 't';
907 else if (c == '\n')
908 *p++ = '\\', *p++ = 'n';
909 else if (c == '\r')
910 *p++ = '\\', *p++ = 'r';
911 else if (c == 0)
912 *p++ = '\\', *p++ = 'x', *p++ = '0', *p++ = '0';
913 else if (c < ' ' || c >= 0x7f) {
914 *p++ = '\\';
915 *p++ = 'x';
916 *p++ = hexdigits[(c & 0xf0) >> 4];
917 *p++ = hexdigits[c & 0xf];
918 }
919 else
920 *p++ = c;
921 }
922 assert(newsize - (p - PyUnicode_AS_UNICODE(v)) >= 1);
923 *p++ = quote;
924 while (*quote_postfix) {
925 *p++ = *quote_postfix++;
926 }
927 *p = '\0';
928 if (PyUnicode_Resize(&v, (p - PyUnicode_AS_UNICODE(v)))) {
929 Py_DECREF(v);
930 return NULL;
931 }
932 return v;
933 }
934}
935
936static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000937bytearray_str(PyObject *op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000938{
Alexander Belopolskyf0f45142010-08-11 17:31:17 +0000939 if (Py_BytesWarningFlag) {
940 if (PyErr_WarnEx(PyExc_BytesWarning,
941 "str() on a bytearray instance", 1))
942 return NULL;
943 }
944 return bytearray_repr((PyByteArrayObject*)op);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000945}
946
947static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000948bytearray_richcompare(PyObject *self, PyObject *other, int op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000949{
950 Py_ssize_t self_size, other_size;
951 Py_buffer self_bytes, other_bytes;
952 PyObject *res;
953 Py_ssize_t minsize;
954 int cmp;
955
956 /* Bytes can be compared to anything that supports the (binary)
957 buffer API. Except that a comparison with Unicode is always an
958 error, even if the comparison is for equality. */
959 if (PyObject_IsInstance(self, (PyObject*)&PyUnicode_Type) ||
960 PyObject_IsInstance(other, (PyObject*)&PyUnicode_Type)) {
Barry Warsaw9e9dcd62008-10-17 01:50:37 +0000961 if (Py_BytesWarningFlag && (op == Py_EQ || op == Py_NE)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000962 if (PyErr_WarnEx(PyExc_BytesWarning,
Georg Brandle5d68ac2008-06-04 11:30:26 +0000963 "Comparison between bytearray and string", 1))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000964 return NULL;
965 }
966
967 Py_INCREF(Py_NotImplemented);
968 return Py_NotImplemented;
969 }
970
971 self_size = _getbuffer(self, &self_bytes);
972 if (self_size < 0) {
973 PyErr_Clear();
974 Py_INCREF(Py_NotImplemented);
975 return Py_NotImplemented;
976 }
977
978 other_size = _getbuffer(other, &other_bytes);
979 if (other_size < 0) {
980 PyErr_Clear();
Martin v. Löwis423be952008-08-13 15:53:07 +0000981 PyBuffer_Release(&self_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000982 Py_INCREF(Py_NotImplemented);
983 return Py_NotImplemented;
984 }
985
986 if (self_size != other_size && (op == Py_EQ || op == Py_NE)) {
987 /* Shortcut: if the lengths differ, the objects differ */
988 cmp = (op == Py_NE);
989 }
990 else {
991 minsize = self_size;
992 if (other_size < minsize)
993 minsize = other_size;
994
995 cmp = memcmp(self_bytes.buf, other_bytes.buf, minsize);
996 /* In ISO C, memcmp() guarantees to use unsigned bytes! */
997
998 if (cmp == 0) {
999 if (self_size < other_size)
1000 cmp = -1;
1001 else if (self_size > other_size)
1002 cmp = 1;
1003 }
1004
1005 switch (op) {
1006 case Py_LT: cmp = cmp < 0; break;
1007 case Py_LE: cmp = cmp <= 0; break;
1008 case Py_EQ: cmp = cmp == 0; break;
1009 case Py_NE: cmp = cmp != 0; break;
1010 case Py_GT: cmp = cmp > 0; break;
1011 case Py_GE: cmp = cmp >= 0; break;
1012 }
1013 }
1014
1015 res = cmp ? Py_True : Py_False;
Martin v. Löwis423be952008-08-13 15:53:07 +00001016 PyBuffer_Release(&self_bytes);
1017 PyBuffer_Release(&other_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001018 Py_INCREF(res);
1019 return res;
1020}
1021
1022static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001023bytearray_dealloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001024{
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001025 if (self->ob_exports > 0) {
1026 PyErr_SetString(PyExc_SystemError,
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00001027 "deallocated bytearray object has exported buffers");
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001028 PyErr_Print();
1029 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001030 if (self->ob_bytes != 0) {
1031 PyMem_Free(self->ob_bytes);
1032 }
1033 Py_TYPE(self)->tp_free((PyObject *)self);
1034}
1035
1036
1037/* -------------------------------------------------------------------- */
1038/* Methods */
1039
1040#define STRINGLIB_CHAR char
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001041#define STRINGLIB_LEN PyByteArray_GET_SIZE
1042#define STRINGLIB_STR PyByteArray_AS_STRING
1043#define STRINGLIB_NEW PyByteArray_FromStringAndSize
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001044#define STRINGLIB_ISSPACE Py_ISSPACE
1045#define STRINGLIB_ISLINEBREAK(x) ((x == '\n') || (x == '\r'))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001046#define STRINGLIB_CHECK_EXACT PyByteArray_CheckExact
1047#define STRINGLIB_MUTABLE 1
1048
1049#include "stringlib/fastsearch.h"
1050#include "stringlib/count.h"
1051#include "stringlib/find.h"
1052#include "stringlib/partition.h"
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001053#include "stringlib/split.h"
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001054#include "stringlib/ctype.h"
1055#include "stringlib/transmogrify.h"
1056
1057
1058/* The following Py_LOCAL_INLINE and Py_LOCAL functions
1059were copied from the old char* style string object. */
1060
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001061/* helper macro to fixup start/end slice values */
1062#define ADJUST_INDICES(start, end, len) \
1063 if (end > len) \
1064 end = len; \
1065 else if (end < 0) { \
1066 end += len; \
1067 if (end < 0) \
1068 end = 0; \
1069 } \
1070 if (start < 0) { \
1071 start += len; \
1072 if (start < 0) \
1073 start = 0; \
1074 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001075
1076Py_LOCAL_INLINE(Py_ssize_t)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001077bytearray_find_internal(PyByteArrayObject *self, PyObject *args, int dir)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001078{
1079 PyObject *subobj;
1080 Py_buffer subbuf;
1081 Py_ssize_t start=0, end=PY_SSIZE_T_MAX;
1082 Py_ssize_t res;
1083
Jesus Ceaac451502011-04-20 17:09:23 +02001084 if (!stringlib_parse_args_finds("find/rfind/index/rindex",
1085 args, &subobj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001086 return -2;
1087 if (_getbuffer(subobj, &subbuf) < 0)
1088 return -2;
1089 if (dir > 0)
1090 res = stringlib_find_slice(
1091 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
1092 subbuf.buf, subbuf.len, start, end);
1093 else
1094 res = stringlib_rfind_slice(
1095 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
1096 subbuf.buf, subbuf.len, start, end);
Martin v. Löwis423be952008-08-13 15:53:07 +00001097 PyBuffer_Release(&subbuf);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001098 return res;
1099}
1100
1101PyDoc_STRVAR(find__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001102"B.find(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001103\n\
1104Return the lowest index in B where subsection sub is found,\n\
1105such that sub is contained within s[start,end]. Optional\n\
1106arguments start and end are interpreted as in slice notation.\n\
1107\n\
1108Return -1 on failure.");
1109
1110static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001111bytearray_find(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001112{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001113 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001114 if (result == -2)
1115 return NULL;
1116 return PyLong_FromSsize_t(result);
1117}
1118
1119PyDoc_STRVAR(count__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001120"B.count(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001121\n\
1122Return the number of non-overlapping occurrences of subsection sub in\n\
1123bytes B[start:end]. Optional arguments start and end are interpreted\n\
1124as in slice notation.");
1125
1126static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001127bytearray_count(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001128{
1129 PyObject *sub_obj;
1130 const char *str = PyByteArray_AS_STRING(self);
1131 Py_ssize_t start = 0, end = PY_SSIZE_T_MAX;
1132 Py_buffer vsub;
1133 PyObject *count_obj;
1134
Jesus Ceaac451502011-04-20 17:09:23 +02001135 if (!stringlib_parse_args_finds("count", args, &sub_obj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001136 return NULL;
1137
1138 if (_getbuffer(sub_obj, &vsub) < 0)
1139 return NULL;
1140
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001141 ADJUST_INDICES(start, end, PyByteArray_GET_SIZE(self));
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001142
1143 count_obj = PyLong_FromSsize_t(
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001144 stringlib_count(str + start, end - start, vsub.buf, vsub.len, PY_SSIZE_T_MAX)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001145 );
Martin v. Löwis423be952008-08-13 15:53:07 +00001146 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001147 return count_obj;
1148}
1149
1150
1151PyDoc_STRVAR(index__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001152"B.index(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001153\n\
1154Like B.find() but raise ValueError when the subsection is not found.");
1155
1156static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001157bytearray_index(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001158{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001159 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001160 if (result == -2)
1161 return NULL;
1162 if (result == -1) {
1163 PyErr_SetString(PyExc_ValueError,
1164 "subsection not found");
1165 return NULL;
1166 }
1167 return PyLong_FromSsize_t(result);
1168}
1169
1170
1171PyDoc_STRVAR(rfind__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001172"B.rfind(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001173\n\
1174Return the highest index in B where subsection sub is found,\n\
1175such that sub is contained within s[start,end]. Optional\n\
1176arguments start and end are interpreted as in slice notation.\n\
1177\n\
1178Return -1 on failure.");
1179
1180static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001181bytearray_rfind(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001182{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001183 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001184 if (result == -2)
1185 return NULL;
1186 return PyLong_FromSsize_t(result);
1187}
1188
1189
1190PyDoc_STRVAR(rindex__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001191"B.rindex(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001192\n\
1193Like B.rfind() but raise ValueError when the subsection is not found.");
1194
1195static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001196bytearray_rindex(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001197{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001198 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001199 if (result == -2)
1200 return NULL;
1201 if (result == -1) {
1202 PyErr_SetString(PyExc_ValueError,
1203 "subsection not found");
1204 return NULL;
1205 }
1206 return PyLong_FromSsize_t(result);
1207}
1208
1209
1210static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001211bytearray_contains(PyObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001212{
1213 Py_ssize_t ival = PyNumber_AsSsize_t(arg, PyExc_ValueError);
1214 if (ival == -1 && PyErr_Occurred()) {
1215 Py_buffer varg;
Antoine Pitrou0010d372010-08-15 17:12:55 +00001216 Py_ssize_t pos;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001217 PyErr_Clear();
1218 if (_getbuffer(arg, &varg) < 0)
1219 return -1;
1220 pos = stringlib_find(PyByteArray_AS_STRING(self), Py_SIZE(self),
1221 varg.buf, varg.len, 0);
Martin v. Löwis423be952008-08-13 15:53:07 +00001222 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001223 return pos >= 0;
1224 }
1225 if (ival < 0 || ival >= 256) {
1226 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
1227 return -1;
1228 }
1229
Antoine Pitrou0010d372010-08-15 17:12:55 +00001230 return memchr(PyByteArray_AS_STRING(self), (int) ival, Py_SIZE(self)) != NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001231}
1232
1233
1234/* Matches the end (direction >= 0) or start (direction < 0) of self
1235 * against substr, using the start and end arguments. Returns
1236 * -1 on error, 0 if not found and 1 if found.
1237 */
1238Py_LOCAL(int)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001239_bytearray_tailmatch(PyByteArrayObject *self, PyObject *substr, Py_ssize_t start,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001240 Py_ssize_t end, int direction)
1241{
1242 Py_ssize_t len = PyByteArray_GET_SIZE(self);
1243 const char* str;
1244 Py_buffer vsubstr;
1245 int rv = 0;
1246
1247 str = PyByteArray_AS_STRING(self);
1248
1249 if (_getbuffer(substr, &vsubstr) < 0)
1250 return -1;
1251
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001252 ADJUST_INDICES(start, end, len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001253
1254 if (direction < 0) {
1255 /* startswith */
1256 if (start+vsubstr.len > len) {
1257 goto done;
1258 }
1259 } else {
1260 /* endswith */
1261 if (end-start < vsubstr.len || start > len) {
1262 goto done;
1263 }
1264
1265 if (end-vsubstr.len > start)
1266 start = end - vsubstr.len;
1267 }
1268 if (end-start >= vsubstr.len)
1269 rv = ! memcmp(str+start, vsubstr.buf, vsubstr.len);
1270
1271done:
Martin v. Löwis423be952008-08-13 15:53:07 +00001272 PyBuffer_Release(&vsubstr);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001273 return rv;
1274}
1275
1276
1277PyDoc_STRVAR(startswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001278"B.startswith(prefix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001279\n\
1280Return True if B starts with the specified prefix, False otherwise.\n\
1281With optional start, test B beginning at that position.\n\
1282With optional end, stop comparing B at that position.\n\
Ezio Melottiba42fd52011-04-26 06:09:45 +03001283prefix can also be a tuple of bytes to try.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001284
1285static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001286bytearray_startswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001287{
1288 Py_ssize_t start = 0;
1289 Py_ssize_t end = PY_SSIZE_T_MAX;
1290 PyObject *subobj;
1291 int result;
1292
Jesus Ceaac451502011-04-20 17:09:23 +02001293 if (!stringlib_parse_args_finds("startswith", args, &subobj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001294 return NULL;
1295 if (PyTuple_Check(subobj)) {
1296 Py_ssize_t i;
1297 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001298 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001299 PyTuple_GET_ITEM(subobj, i),
1300 start, end, -1);
1301 if (result == -1)
1302 return NULL;
1303 else if (result) {
1304 Py_RETURN_TRUE;
1305 }
1306 }
1307 Py_RETURN_FALSE;
1308 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001309 result = _bytearray_tailmatch(self, subobj, start, end, -1);
Ezio Melottiba42fd52011-04-26 06:09:45 +03001310 if (result == -1) {
1311 if (PyErr_ExceptionMatches(PyExc_TypeError))
1312 PyErr_Format(PyExc_TypeError, "startswith first arg must be bytes "
1313 "or a tuple of bytes, not %s", Py_TYPE(subobj)->tp_name);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001314 return NULL;
Ezio Melottiba42fd52011-04-26 06:09:45 +03001315 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001316 else
1317 return PyBool_FromLong(result);
1318}
1319
1320PyDoc_STRVAR(endswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001321"B.endswith(suffix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001322\n\
1323Return True if B ends with the specified suffix, False otherwise.\n\
1324With optional start, test B beginning at that position.\n\
1325With optional end, stop comparing B at that position.\n\
Ezio Melottiba42fd52011-04-26 06:09:45 +03001326suffix can also be a tuple of bytes to try.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001327
1328static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001329bytearray_endswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001330{
1331 Py_ssize_t start = 0;
1332 Py_ssize_t end = PY_SSIZE_T_MAX;
1333 PyObject *subobj;
1334 int result;
1335
Jesus Ceaac451502011-04-20 17:09:23 +02001336 if (!stringlib_parse_args_finds("endswith", args, &subobj, &start, &end))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001337 return NULL;
1338 if (PyTuple_Check(subobj)) {
1339 Py_ssize_t i;
1340 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001341 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001342 PyTuple_GET_ITEM(subobj, i),
1343 start, end, +1);
1344 if (result == -1)
1345 return NULL;
1346 else if (result) {
1347 Py_RETURN_TRUE;
1348 }
1349 }
1350 Py_RETURN_FALSE;
1351 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001352 result = _bytearray_tailmatch(self, subobj, start, end, +1);
Ezio Melottiba42fd52011-04-26 06:09:45 +03001353 if (result == -1) {
1354 if (PyErr_ExceptionMatches(PyExc_TypeError))
1355 PyErr_Format(PyExc_TypeError, "endswith first arg must be bytes or "
1356 "a tuple of bytes, not %s", Py_TYPE(subobj)->tp_name);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001357 return NULL;
Ezio Melottiba42fd52011-04-26 06:09:45 +03001358 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001359 else
1360 return PyBool_FromLong(result);
1361}
1362
1363
1364PyDoc_STRVAR(translate__doc__,
1365"B.translate(table[, deletechars]) -> bytearray\n\
1366\n\
1367Return a copy of B, where all characters occurring in the\n\
1368optional argument deletechars are removed, and the remaining\n\
1369characters have been mapped through the given translation\n\
1370table, which must be a bytes object of length 256.");
1371
1372static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001373bytearray_translate(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001374{
1375 register char *input, *output;
1376 register const char *table;
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001377 register Py_ssize_t i, c;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001378 PyObject *input_obj = (PyObject*)self;
1379 const char *output_start;
1380 Py_ssize_t inlen;
Georg Brandlccc47b62008-12-28 11:44:14 +00001381 PyObject *result = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001382 int trans_table[256];
Georg Brandlccc47b62008-12-28 11:44:14 +00001383 PyObject *tableobj = NULL, *delobj = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001384 Py_buffer vtable, vdel;
1385
1386 if (!PyArg_UnpackTuple(args, "translate", 1, 2,
1387 &tableobj, &delobj))
1388 return NULL;
1389
Georg Brandlccc47b62008-12-28 11:44:14 +00001390 if (tableobj == Py_None) {
1391 table = NULL;
1392 tableobj = NULL;
1393 } else if (_getbuffer(tableobj, &vtable) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001394 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001395 } else {
1396 if (vtable.len != 256) {
1397 PyErr_SetString(PyExc_ValueError,
1398 "translation table must be 256 characters long");
Georg Brandl953152f2009-07-22 12:03:59 +00001399 PyBuffer_Release(&vtable);
1400 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001401 }
1402 table = (const char*)vtable.buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001403 }
1404
1405 if (delobj != NULL) {
1406 if (_getbuffer(delobj, &vdel) < 0) {
Georg Brandl953152f2009-07-22 12:03:59 +00001407 if (tableobj != NULL)
1408 PyBuffer_Release(&vtable);
1409 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001410 }
1411 }
1412 else {
1413 vdel.buf = NULL;
1414 vdel.len = 0;
1415 }
1416
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001417 inlen = PyByteArray_GET_SIZE(input_obj);
1418 result = PyByteArray_FromStringAndSize((char *)NULL, inlen);
1419 if (result == NULL)
1420 goto done;
1421 output_start = output = PyByteArray_AsString(result);
1422 input = PyByteArray_AS_STRING(input_obj);
1423
Georg Brandlccc47b62008-12-28 11:44:14 +00001424 if (vdel.len == 0 && table != NULL) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001425 /* If no deletions are required, use faster code */
1426 for (i = inlen; --i >= 0; ) {
1427 c = Py_CHARMASK(*input++);
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001428 *output++ = table[c];
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001429 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001430 goto done;
1431 }
Georg Brandlccc47b62008-12-28 11:44:14 +00001432
1433 if (table == NULL) {
1434 for (i = 0; i < 256; i++)
1435 trans_table[i] = Py_CHARMASK(i);
1436 } else {
1437 for (i = 0; i < 256; i++)
1438 trans_table[i] = Py_CHARMASK(table[i]);
1439 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001440
1441 for (i = 0; i < vdel.len; i++)
1442 trans_table[(int) Py_CHARMASK( ((unsigned char*)vdel.buf)[i] )] = -1;
1443
1444 for (i = inlen; --i >= 0; ) {
1445 c = Py_CHARMASK(*input++);
1446 if (trans_table[c] != -1)
1447 if (Py_CHARMASK(*output++ = (char)trans_table[c]) == c)
1448 continue;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001449 }
1450 /* Fix the size of the resulting string */
1451 if (inlen > 0)
1452 PyByteArray_Resize(result, output - output_start);
1453
1454done:
Georg Brandlccc47b62008-12-28 11:44:14 +00001455 if (tableobj != NULL)
1456 PyBuffer_Release(&vtable);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001457 if (delobj != NULL)
Martin v. Löwis423be952008-08-13 15:53:07 +00001458 PyBuffer_Release(&vdel);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001459 return result;
1460}
1461
1462
Georg Brandlabc38772009-04-12 15:51:51 +00001463static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001464bytearray_maketrans(PyObject *null, PyObject *args)
Georg Brandlabc38772009-04-12 15:51:51 +00001465{
Alexander Belopolskyf0f45142010-08-11 17:31:17 +00001466 return _Py_bytes_maketrans(args);
Georg Brandlabc38772009-04-12 15:51:51 +00001467}
1468
1469
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001470/* find and count characters and substrings */
1471
1472#define findchar(target, target_len, c) \
1473 ((char *)memchr((const void *)(target), c, target_len))
1474
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001475
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001476/* Bytes ops must return a string, create a copy */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001477Py_LOCAL(PyByteArrayObject *)
1478return_self(PyByteArrayObject *self)
1479{
Georg Brandl1e7217d2008-05-30 12:02:38 +00001480 /* always return a new bytearray */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001481 return (PyByteArrayObject *)PyByteArray_FromStringAndSize(
1482 PyByteArray_AS_STRING(self),
1483 PyByteArray_GET_SIZE(self));
1484}
1485
1486Py_LOCAL_INLINE(Py_ssize_t)
1487countchar(const char *target, Py_ssize_t target_len, char c, Py_ssize_t maxcount)
1488{
1489 Py_ssize_t count=0;
1490 const char *start=target;
1491 const char *end=target+target_len;
1492
1493 while ( (start=findchar(start, end-start, c)) != NULL ) {
1494 count++;
1495 if (count >= maxcount)
1496 break;
1497 start += 1;
1498 }
1499 return count;
1500}
1501
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001502
1503/* Algorithms for different cases of string replacement */
1504
1505/* len(self)>=1, from="", len(to)>=1, maxcount>=1 */
1506Py_LOCAL(PyByteArrayObject *)
1507replace_interleave(PyByteArrayObject *self,
1508 const char *to_s, Py_ssize_t to_len,
1509 Py_ssize_t maxcount)
1510{
1511 char *self_s, *result_s;
1512 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001513 Py_ssize_t count, i;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001514 PyByteArrayObject *result;
1515
1516 self_len = PyByteArray_GET_SIZE(self);
1517
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001518 /* 1 at the end plus 1 after every character;
1519 count = min(maxcount, self_len + 1) */
1520 if (maxcount <= self_len)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001521 count = maxcount;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001522 else
1523 /* Can't overflow: self_len + 1 <= maxcount <= PY_SSIZE_T_MAX. */
1524 count = self_len + 1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001525
1526 /* Check for overflow */
1527 /* result_len = count * to_len + self_len; */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001528 assert(count > 0);
1529 if (to_len > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001530 PyErr_SetString(PyExc_OverflowError,
1531 "replace string is too long");
1532 return NULL;
1533 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001534 result_len = count * to_len + self_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001535
1536 if (! (result = (PyByteArrayObject *)
1537 PyByteArray_FromStringAndSize(NULL, result_len)) )
1538 return NULL;
1539
1540 self_s = PyByteArray_AS_STRING(self);
1541 result_s = PyByteArray_AS_STRING(result);
1542
1543 /* TODO: special case single character, which doesn't need memcpy */
1544
1545 /* Lay the first one down (guaranteed this will occur) */
1546 Py_MEMCPY(result_s, to_s, to_len);
1547 result_s += to_len;
1548 count -= 1;
1549
1550 for (i=0; i<count; i++) {
1551 *result_s++ = *self_s++;
1552 Py_MEMCPY(result_s, to_s, to_len);
1553 result_s += to_len;
1554 }
1555
1556 /* Copy the rest of the original string */
1557 Py_MEMCPY(result_s, self_s, self_len-i);
1558
1559 return result;
1560}
1561
1562/* Special case for deleting a single character */
1563/* len(self)>=1, len(from)==1, to="", maxcount>=1 */
1564Py_LOCAL(PyByteArrayObject *)
1565replace_delete_single_character(PyByteArrayObject *self,
1566 char from_c, Py_ssize_t maxcount)
1567{
1568 char *self_s, *result_s;
1569 char *start, *next, *end;
1570 Py_ssize_t self_len, result_len;
1571 Py_ssize_t count;
1572 PyByteArrayObject *result;
1573
1574 self_len = PyByteArray_GET_SIZE(self);
1575 self_s = PyByteArray_AS_STRING(self);
1576
1577 count = countchar(self_s, self_len, from_c, maxcount);
1578 if (count == 0) {
1579 return return_self(self);
1580 }
1581
1582 result_len = self_len - count; /* from_len == 1 */
1583 assert(result_len>=0);
1584
1585 if ( (result = (PyByteArrayObject *)
1586 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1587 return NULL;
1588 result_s = PyByteArray_AS_STRING(result);
1589
1590 start = self_s;
1591 end = self_s + self_len;
1592 while (count-- > 0) {
1593 next = findchar(start, end-start, from_c);
1594 if (next == NULL)
1595 break;
1596 Py_MEMCPY(result_s, start, next-start);
1597 result_s += (next-start);
1598 start = next+1;
1599 }
1600 Py_MEMCPY(result_s, start, end-start);
1601
1602 return result;
1603}
1604
1605/* len(self)>=1, len(from)>=2, to="", maxcount>=1 */
1606
1607Py_LOCAL(PyByteArrayObject *)
1608replace_delete_substring(PyByteArrayObject *self,
1609 const char *from_s, Py_ssize_t from_len,
1610 Py_ssize_t maxcount)
1611{
1612 char *self_s, *result_s;
1613 char *start, *next, *end;
1614 Py_ssize_t self_len, result_len;
1615 Py_ssize_t count, offset;
1616 PyByteArrayObject *result;
1617
1618 self_len = PyByteArray_GET_SIZE(self);
1619 self_s = PyByteArray_AS_STRING(self);
1620
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001621 count = stringlib_count(self_s, self_len,
1622 from_s, from_len,
1623 maxcount);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001624
1625 if (count == 0) {
1626 /* no matches */
1627 return return_self(self);
1628 }
1629
1630 result_len = self_len - (count * from_len);
1631 assert (result_len>=0);
1632
1633 if ( (result = (PyByteArrayObject *)
1634 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL )
1635 return NULL;
1636
1637 result_s = PyByteArray_AS_STRING(result);
1638
1639 start = self_s;
1640 end = self_s + self_len;
1641 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001642 offset = stringlib_find(start, end-start,
1643 from_s, from_len,
1644 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001645 if (offset == -1)
1646 break;
1647 next = start + offset;
1648
1649 Py_MEMCPY(result_s, start, next-start);
1650
1651 result_s += (next-start);
1652 start = next+from_len;
1653 }
1654 Py_MEMCPY(result_s, start, end-start);
1655 return result;
1656}
1657
1658/* len(self)>=1, len(from)==len(to)==1, maxcount>=1 */
1659Py_LOCAL(PyByteArrayObject *)
1660replace_single_character_in_place(PyByteArrayObject *self,
1661 char from_c, char to_c,
1662 Py_ssize_t maxcount)
1663{
Antoine Pitroud1188562010-06-09 16:38:55 +00001664 char *self_s, *result_s, *start, *end, *next;
1665 Py_ssize_t self_len;
1666 PyByteArrayObject *result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001667
Antoine Pitroud1188562010-06-09 16:38:55 +00001668 /* The result string will be the same size */
1669 self_s = PyByteArray_AS_STRING(self);
1670 self_len = PyByteArray_GET_SIZE(self);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001671
Antoine Pitroud1188562010-06-09 16:38:55 +00001672 next = findchar(self_s, self_len, from_c);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001673
Antoine Pitroud1188562010-06-09 16:38:55 +00001674 if (next == NULL) {
1675 /* No matches; return the original bytes */
1676 return return_self(self);
1677 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001678
Antoine Pitroud1188562010-06-09 16:38:55 +00001679 /* Need to make a new bytes */
1680 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1681 if (result == NULL)
1682 return NULL;
1683 result_s = PyByteArray_AS_STRING(result);
1684 Py_MEMCPY(result_s, self_s, self_len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001685
Antoine Pitroud1188562010-06-09 16:38:55 +00001686 /* change everything in-place, starting with this one */
1687 start = result_s + (next-self_s);
1688 *start = to_c;
1689 start++;
1690 end = result_s + self_len;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001691
Antoine Pitroud1188562010-06-09 16:38:55 +00001692 while (--maxcount > 0) {
1693 next = findchar(start, end-start, from_c);
1694 if (next == NULL)
1695 break;
1696 *next = to_c;
1697 start = next+1;
1698 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001699
Antoine Pitroud1188562010-06-09 16:38:55 +00001700 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001701}
1702
1703/* len(self)>=1, len(from)==len(to)>=2, maxcount>=1 */
1704Py_LOCAL(PyByteArrayObject *)
1705replace_substring_in_place(PyByteArrayObject *self,
1706 const char *from_s, Py_ssize_t from_len,
1707 const char *to_s, Py_ssize_t to_len,
1708 Py_ssize_t maxcount)
1709{
1710 char *result_s, *start, *end;
1711 char *self_s;
1712 Py_ssize_t self_len, offset;
1713 PyByteArrayObject *result;
1714
1715 /* The result bytes will be the same size */
1716
1717 self_s = PyByteArray_AS_STRING(self);
1718 self_len = PyByteArray_GET_SIZE(self);
1719
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001720 offset = stringlib_find(self_s, self_len,
1721 from_s, from_len,
1722 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001723 if (offset == -1) {
1724 /* No matches; return the original bytes */
1725 return return_self(self);
1726 }
1727
1728 /* Need to make a new bytes */
1729 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1730 if (result == NULL)
1731 return NULL;
1732 result_s = PyByteArray_AS_STRING(result);
1733 Py_MEMCPY(result_s, self_s, self_len);
1734
1735 /* change everything in-place, starting with this one */
1736 start = result_s + offset;
1737 Py_MEMCPY(start, to_s, from_len);
1738 start += from_len;
1739 end = result_s + self_len;
1740
1741 while ( --maxcount > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001742 offset = stringlib_find(start, end-start,
1743 from_s, from_len,
1744 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001745 if (offset==-1)
1746 break;
1747 Py_MEMCPY(start+offset, to_s, from_len);
1748 start += offset+from_len;
1749 }
1750
1751 return result;
1752}
1753
1754/* len(self)>=1, len(from)==1, len(to)>=2, maxcount>=1 */
1755Py_LOCAL(PyByteArrayObject *)
1756replace_single_character(PyByteArrayObject *self,
1757 char from_c,
1758 const char *to_s, Py_ssize_t to_len,
1759 Py_ssize_t maxcount)
1760{
1761 char *self_s, *result_s;
1762 char *start, *next, *end;
1763 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001764 Py_ssize_t count;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001765 PyByteArrayObject *result;
1766
1767 self_s = PyByteArray_AS_STRING(self);
1768 self_len = PyByteArray_GET_SIZE(self);
1769
1770 count = countchar(self_s, self_len, from_c, maxcount);
1771 if (count == 0) {
1772 /* no matches, return unchanged */
1773 return return_self(self);
1774 }
1775
1776 /* use the difference between current and new, hence the "-1" */
1777 /* result_len = self_len + count * (to_len-1) */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001778 assert(count > 0);
1779 if (to_len - 1 > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001780 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1781 return NULL;
1782 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001783 result_len = self_len + count * (to_len - 1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001784
1785 if ( (result = (PyByteArrayObject *)
1786 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1787 return NULL;
1788 result_s = PyByteArray_AS_STRING(result);
1789
1790 start = self_s;
1791 end = self_s + self_len;
1792 while (count-- > 0) {
1793 next = findchar(start, end-start, from_c);
1794 if (next == NULL)
1795 break;
1796
1797 if (next == start) {
1798 /* replace with the 'to' */
1799 Py_MEMCPY(result_s, to_s, to_len);
1800 result_s += to_len;
1801 start += 1;
1802 } else {
1803 /* copy the unchanged old then the 'to' */
1804 Py_MEMCPY(result_s, start, next-start);
1805 result_s += (next-start);
1806 Py_MEMCPY(result_s, to_s, to_len);
1807 result_s += to_len;
1808 start = next+1;
1809 }
1810 }
1811 /* Copy the remainder of the remaining bytes */
1812 Py_MEMCPY(result_s, start, end-start);
1813
1814 return result;
1815}
1816
1817/* len(self)>=1, len(from)>=2, len(to)>=2, maxcount>=1 */
1818Py_LOCAL(PyByteArrayObject *)
1819replace_substring(PyByteArrayObject *self,
1820 const char *from_s, Py_ssize_t from_len,
1821 const char *to_s, Py_ssize_t to_len,
1822 Py_ssize_t maxcount)
1823{
1824 char *self_s, *result_s;
1825 char *start, *next, *end;
1826 Py_ssize_t self_len, result_len;
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001827 Py_ssize_t count, offset;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001828 PyByteArrayObject *result;
1829
1830 self_s = PyByteArray_AS_STRING(self);
1831 self_len = PyByteArray_GET_SIZE(self);
1832
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001833 count = stringlib_count(self_s, self_len,
1834 from_s, from_len,
1835 maxcount);
1836
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001837 if (count == 0) {
1838 /* no matches, return unchanged */
1839 return return_self(self);
1840 }
1841
1842 /* Check for overflow */
1843 /* result_len = self_len + count * (to_len-from_len) */
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001844 assert(count > 0);
1845 if (to_len - from_len > (PY_SSIZE_T_MAX - self_len) / count) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001846 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1847 return NULL;
1848 }
Mark Dickinsoncf940c72010-08-10 18:35:01 +00001849 result_len = self_len + count * (to_len - from_len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001850
1851 if ( (result = (PyByteArrayObject *)
1852 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1853 return NULL;
1854 result_s = PyByteArray_AS_STRING(result);
1855
1856 start = self_s;
1857 end = self_s + self_len;
1858 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001859 offset = stringlib_find(start, end-start,
1860 from_s, from_len,
1861 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001862 if (offset == -1)
1863 break;
1864 next = start+offset;
1865 if (next == start) {
1866 /* replace with the 'to' */
1867 Py_MEMCPY(result_s, to_s, to_len);
1868 result_s += to_len;
1869 start += from_len;
1870 } else {
1871 /* copy the unchanged old then the 'to' */
1872 Py_MEMCPY(result_s, start, next-start);
1873 result_s += (next-start);
1874 Py_MEMCPY(result_s, to_s, to_len);
1875 result_s += to_len;
1876 start = next+from_len;
1877 }
1878 }
1879 /* Copy the remainder of the remaining bytes */
1880 Py_MEMCPY(result_s, start, end-start);
1881
1882 return result;
1883}
1884
1885
1886Py_LOCAL(PyByteArrayObject *)
1887replace(PyByteArrayObject *self,
1888 const char *from_s, Py_ssize_t from_len,
1889 const char *to_s, Py_ssize_t to_len,
1890 Py_ssize_t maxcount)
1891{
1892 if (maxcount < 0) {
1893 maxcount = PY_SSIZE_T_MAX;
1894 } else if (maxcount == 0 || PyByteArray_GET_SIZE(self) == 0) {
1895 /* nothing to do; return the original bytes */
1896 return return_self(self);
1897 }
1898
1899 if (maxcount == 0 ||
1900 (from_len == 0 && to_len == 0)) {
1901 /* nothing to do; return the original bytes */
1902 return return_self(self);
1903 }
1904
1905 /* Handle zero-length special cases */
1906
1907 if (from_len == 0) {
1908 /* insert the 'to' bytes everywhere. */
1909 /* >>> "Python".replace("", ".") */
1910 /* '.P.y.t.h.o.n.' */
1911 return replace_interleave(self, to_s, to_len, maxcount);
1912 }
1913
1914 /* Except for "".replace("", "A") == "A" there is no way beyond this */
1915 /* point for an empty self bytes to generate a non-empty bytes */
1916 /* Special case so the remaining code always gets a non-empty bytes */
1917 if (PyByteArray_GET_SIZE(self) == 0) {
1918 return return_self(self);
1919 }
1920
1921 if (to_len == 0) {
Georg Brandl17cb8a82008-05-30 08:20:09 +00001922 /* delete all occurrences of 'from' bytes */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001923 if (from_len == 1) {
1924 return replace_delete_single_character(
1925 self, from_s[0], maxcount);
1926 } else {
1927 return replace_delete_substring(self, from_s, from_len, maxcount);
1928 }
1929 }
1930
1931 /* Handle special case where both bytes have the same length */
1932
1933 if (from_len == to_len) {
1934 if (from_len == 1) {
1935 return replace_single_character_in_place(
1936 self,
1937 from_s[0],
1938 to_s[0],
1939 maxcount);
1940 } else {
1941 return replace_substring_in_place(
1942 self, from_s, from_len, to_s, to_len, maxcount);
1943 }
1944 }
1945
1946 /* Otherwise use the more generic algorithms */
1947 if (from_len == 1) {
1948 return replace_single_character(self, from_s[0],
1949 to_s, to_len, maxcount);
1950 } else {
1951 /* len('from')>=2, len('to')>=1 */
1952 return replace_substring(self, from_s, from_len, to_s, to_len, maxcount);
1953 }
1954}
1955
1956
1957PyDoc_STRVAR(replace__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001958"B.replace(old, new[, count]) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001959\n\
1960Return a copy of B with all occurrences of subsection\n\
1961old replaced by new. If the optional argument count is\n\
1962given, only the first count occurrences are replaced.");
1963
1964static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001965bytearray_replace(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001966{
1967 Py_ssize_t count = -1;
1968 PyObject *from, *to, *res;
1969 Py_buffer vfrom, vto;
1970
1971 if (!PyArg_ParseTuple(args, "OO|n:replace", &from, &to, &count))
1972 return NULL;
1973
1974 if (_getbuffer(from, &vfrom) < 0)
1975 return NULL;
1976 if (_getbuffer(to, &vto) < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +00001977 PyBuffer_Release(&vfrom);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001978 return NULL;
1979 }
1980
1981 res = (PyObject *)replace((PyByteArrayObject *) self,
1982 vfrom.buf, vfrom.len,
1983 vto.buf, vto.len, count);
1984
Martin v. Löwis423be952008-08-13 15:53:07 +00001985 PyBuffer_Release(&vfrom);
1986 PyBuffer_Release(&vto);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001987 return res;
1988}
1989
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001990PyDoc_STRVAR(split__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001991"B.split([sep[, maxsplit]]) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001992\n\
1993Return a list of the sections in B, using sep as the delimiter.\n\
1994If sep is not given, B is split on ASCII whitespace characters\n\
1995(space, tab, return, newline, formfeed, vertical tab).\n\
1996If maxsplit is given, at most maxsplit splits are done.");
1997
1998static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001999bytearray_split(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002000{
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002001 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2002 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002003 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002004 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002005 Py_buffer vsub;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002006
2007 if (!PyArg_ParseTuple(args, "|On:split", &subobj, &maxsplit))
2008 return NULL;
2009 if (maxsplit < 0)
2010 maxsplit = PY_SSIZE_T_MAX;
2011
2012 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002013 return stringlib_split_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002014
2015 if (_getbuffer(subobj, &vsub) < 0)
2016 return NULL;
2017 sub = vsub.buf;
2018 n = vsub.len;
2019
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002020 list = stringlib_split(
2021 (PyObject*) self, s, len, sub, n, maxsplit
2022 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002023 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002024 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002025}
2026
2027PyDoc_STRVAR(partition__doc__,
2028"B.partition(sep) -> (head, sep, tail)\n\
2029\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002030Search for the separator sep in B, and return the part before it,\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002031the separator itself, and the part after it. If the separator is not\n\
2032found, returns B and two empty bytearray objects.");
2033
2034static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002035bytearray_partition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002036{
2037 PyObject *bytesep, *result;
2038
2039 bytesep = PyByteArray_FromObject(sep_obj);
2040 if (! bytesep)
2041 return NULL;
2042
2043 result = stringlib_partition(
2044 (PyObject*) self,
2045 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2046 bytesep,
2047 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2048 );
2049
2050 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002051 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002052}
2053
2054PyDoc_STRVAR(rpartition__doc__,
Ezio Melotti5b2b2422010-01-25 11:58:28 +00002055"B.rpartition(sep) -> (head, sep, tail)\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002056\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002057Search for the separator sep in B, starting at the end of B,\n\
2058and return the part before it, the separator itself, and the\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002059part after it. If the separator is not found, returns two empty\n\
2060bytearray objects and B.");
2061
2062static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002063bytearray_rpartition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002064{
2065 PyObject *bytesep, *result;
2066
2067 bytesep = PyByteArray_FromObject(sep_obj);
2068 if (! bytesep)
2069 return NULL;
2070
2071 result = stringlib_rpartition(
2072 (PyObject*) self,
2073 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2074 bytesep,
2075 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2076 );
2077
2078 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002079 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002080}
2081
2082PyDoc_STRVAR(rsplit__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002083"B.rsplit(sep[, maxsplit]) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002084\n\
2085Return a list of the sections in B, using sep as the delimiter,\n\
2086starting at the end of B and working to the front.\n\
2087If sep is not given, B is split on ASCII whitespace characters\n\
2088(space, tab, return, newline, formfeed, vertical tab).\n\
2089If maxsplit is given, at most maxsplit splits are done.");
2090
2091static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002092bytearray_rsplit(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002093{
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002094 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2095 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002096 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002097 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002098 Py_buffer vsub;
2099
2100 if (!PyArg_ParseTuple(args, "|On:rsplit", &subobj, &maxsplit))
2101 return NULL;
2102 if (maxsplit < 0)
2103 maxsplit = PY_SSIZE_T_MAX;
2104
2105 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002106 return stringlib_rsplit_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002107
2108 if (_getbuffer(subobj, &vsub) < 0)
2109 return NULL;
2110 sub = vsub.buf;
2111 n = vsub.len;
2112
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002113 list = stringlib_rsplit(
2114 (PyObject*) self, s, len, sub, n, maxsplit
2115 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002116 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002117 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002118}
2119
2120PyDoc_STRVAR(reverse__doc__,
2121"B.reverse() -> None\n\
2122\n\
2123Reverse the order of the values in B in place.");
2124static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002125bytearray_reverse(PyByteArrayObject *self, PyObject *unused)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002126{
2127 char swap, *head, *tail;
2128 Py_ssize_t i, j, n = Py_SIZE(self);
2129
2130 j = n / 2;
2131 head = self->ob_bytes;
2132 tail = head + n - 1;
2133 for (i = 0; i < j; i++) {
2134 swap = *head;
2135 *head++ = *tail;
2136 *tail-- = swap;
2137 }
2138
2139 Py_RETURN_NONE;
2140}
2141
2142PyDoc_STRVAR(insert__doc__,
2143"B.insert(index, int) -> None\n\
2144\n\
2145Insert a single item into the bytearray before the given index.");
2146static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002147bytearray_insert(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002148{
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002149 PyObject *value;
2150 int ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002151 Py_ssize_t where, n = Py_SIZE(self);
2152
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002153 if (!PyArg_ParseTuple(args, "nO:insert", &where, &value))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002154 return NULL;
2155
2156 if (n == PY_SSIZE_T_MAX) {
2157 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002158 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002159 return NULL;
2160 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002161 if (!_getbytevalue(value, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002162 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002163 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2164 return NULL;
2165
2166 if (where < 0) {
2167 where += n;
2168 if (where < 0)
2169 where = 0;
2170 }
2171 if (where > n)
2172 where = n;
2173 memmove(self->ob_bytes + where + 1, self->ob_bytes + where, n - where);
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002174 self->ob_bytes[where] = ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002175
2176 Py_RETURN_NONE;
2177}
2178
2179PyDoc_STRVAR(append__doc__,
2180"B.append(int) -> None\n\
2181\n\
2182Append a single item to the end of B.");
2183static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002184bytearray_append(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002185{
2186 int value;
2187 Py_ssize_t n = Py_SIZE(self);
2188
2189 if (! _getbytevalue(arg, &value))
2190 return NULL;
2191 if (n == PY_SSIZE_T_MAX) {
2192 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002193 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002194 return NULL;
2195 }
2196 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2197 return NULL;
2198
2199 self->ob_bytes[n] = value;
2200
2201 Py_RETURN_NONE;
2202}
2203
2204PyDoc_STRVAR(extend__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002205"B.extend(iterable_of_ints) -> None\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002206\n\
2207Append all the elements from the iterator or sequence to the\n\
2208end of B.");
2209static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002210bytearray_extend(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002211{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002212 PyObject *it, *item, *bytearray_obj;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002213 Py_ssize_t buf_size = 0, len = 0;
2214 int value;
2215 char *buf;
2216
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002217 /* bytearray_setslice code only accepts something supporting PEP 3118. */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002218 if (PyObject_CheckBuffer(arg)) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002219 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), arg) == -1)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002220 return NULL;
2221
2222 Py_RETURN_NONE;
2223 }
2224
2225 it = PyObject_GetIter(arg);
2226 if (it == NULL)
2227 return NULL;
2228
Ezio Melotti42da6632011-03-15 05:18:48 +02002229 /* Try to determine the length of the argument. 32 is arbitrary. */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002230 buf_size = _PyObject_LengthHint(arg, 32);
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002231 if (buf_size == -1) {
2232 Py_DECREF(it);
2233 return NULL;
2234 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002235
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002236 bytearray_obj = PyByteArray_FromStringAndSize(NULL, buf_size);
2237 if (bytearray_obj == NULL)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002238 return NULL;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002239 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002240
2241 while ((item = PyIter_Next(it)) != NULL) {
2242 if (! _getbytevalue(item, &value)) {
2243 Py_DECREF(item);
2244 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002245 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002246 return NULL;
2247 }
2248 buf[len++] = value;
2249 Py_DECREF(item);
2250
2251 if (len >= buf_size) {
2252 buf_size = len + (len >> 1) + 1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002253 if (PyByteArray_Resize((PyObject *)bytearray_obj, buf_size) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002254 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002255 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002256 return NULL;
2257 }
2258 /* Recompute the `buf' pointer, since the resizing operation may
2259 have invalidated it. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002260 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002261 }
2262 }
2263 Py_DECREF(it);
2264
2265 /* Resize down to exact size. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002266 if (PyByteArray_Resize((PyObject *)bytearray_obj, len) < 0) {
2267 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002268 return NULL;
2269 }
2270
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002271 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), bytearray_obj) == -1)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002272 return NULL;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002273 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002274
2275 Py_RETURN_NONE;
2276}
2277
2278PyDoc_STRVAR(pop__doc__,
2279"B.pop([index]) -> int\n\
2280\n\
2281Remove and return a single item from B. If no index\n\
Benjamin Petersondcf97b92008-07-02 17:30:14 +00002282argument is given, will pop the last value.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002283static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002284bytearray_pop(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002285{
2286 int value;
2287 Py_ssize_t where = -1, n = Py_SIZE(self);
2288
2289 if (!PyArg_ParseTuple(args, "|n:pop", &where))
2290 return NULL;
2291
2292 if (n == 0) {
Eli Benderskye0c8635d82011-03-04 05:10:57 +00002293 PyErr_SetString(PyExc_IndexError,
2294 "pop from empty bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002295 return NULL;
2296 }
2297 if (where < 0)
2298 where += Py_SIZE(self);
2299 if (where < 0 || where >= Py_SIZE(self)) {
2300 PyErr_SetString(PyExc_IndexError, "pop index out of range");
2301 return NULL;
2302 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002303 if (!_canresize(self))
2304 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002305
2306 value = self->ob_bytes[where];
2307 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2308 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2309 return NULL;
2310
Mark Dickinson54a3db92009-09-06 10:19:23 +00002311 return PyLong_FromLong((unsigned char)value);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002312}
2313
2314PyDoc_STRVAR(remove__doc__,
2315"B.remove(int) -> None\n\
2316\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002317Remove the first occurrence of a value in B.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002318static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002319bytearray_remove(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002320{
2321 int value;
2322 Py_ssize_t where, n = Py_SIZE(self);
2323
2324 if (! _getbytevalue(arg, &value))
2325 return NULL;
2326
2327 for (where = 0; where < n; where++) {
2328 if (self->ob_bytes[where] == value)
2329 break;
2330 }
2331 if (where == n) {
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002332 PyErr_SetString(PyExc_ValueError, "value not found in bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002333 return NULL;
2334 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002335 if (!_canresize(self))
2336 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002337
2338 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2339 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2340 return NULL;
2341
2342 Py_RETURN_NONE;
2343}
2344
2345/* XXX These two helpers could be optimized if argsize == 1 */
2346
2347static Py_ssize_t
2348lstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2349 void *argptr, Py_ssize_t argsize)
2350{
2351 Py_ssize_t i = 0;
2352 while (i < mysize && memchr(argptr, myptr[i], argsize))
2353 i++;
2354 return i;
2355}
2356
2357static Py_ssize_t
2358rstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2359 void *argptr, Py_ssize_t argsize)
2360{
2361 Py_ssize_t i = mysize - 1;
2362 while (i >= 0 && memchr(argptr, myptr[i], argsize))
2363 i--;
2364 return i + 1;
2365}
2366
2367PyDoc_STRVAR(strip__doc__,
2368"B.strip([bytes]) -> bytearray\n\
2369\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002370Strip leading and trailing bytes contained in the argument\n\
2371and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002372If the argument is omitted, strip ASCII whitespace.");
2373static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002374bytearray_strip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002375{
2376 Py_ssize_t left, right, mysize, argsize;
2377 void *myptr, *argptr;
2378 PyObject *arg = Py_None;
2379 Py_buffer varg;
2380 if (!PyArg_ParseTuple(args, "|O:strip", &arg))
2381 return NULL;
2382 if (arg == Py_None) {
2383 argptr = "\t\n\r\f\v ";
2384 argsize = 6;
2385 }
2386 else {
2387 if (_getbuffer(arg, &varg) < 0)
2388 return NULL;
2389 argptr = varg.buf;
2390 argsize = varg.len;
2391 }
2392 myptr = self->ob_bytes;
2393 mysize = Py_SIZE(self);
2394 left = lstrip_helper(myptr, mysize, argptr, argsize);
2395 if (left == mysize)
2396 right = left;
2397 else
2398 right = rstrip_helper(myptr, mysize, argptr, argsize);
2399 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002400 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002401 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2402}
2403
2404PyDoc_STRVAR(lstrip__doc__,
2405"B.lstrip([bytes]) -> bytearray\n\
2406\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002407Strip leading bytes contained in the argument\n\
2408and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002409If the argument is omitted, strip leading ASCII whitespace.");
2410static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002411bytearray_lstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002412{
2413 Py_ssize_t left, right, mysize, argsize;
2414 void *myptr, *argptr;
2415 PyObject *arg = Py_None;
2416 Py_buffer varg;
2417 if (!PyArg_ParseTuple(args, "|O:lstrip", &arg))
2418 return NULL;
2419 if (arg == Py_None) {
2420 argptr = "\t\n\r\f\v ";
2421 argsize = 6;
2422 }
2423 else {
2424 if (_getbuffer(arg, &varg) < 0)
2425 return NULL;
2426 argptr = varg.buf;
2427 argsize = varg.len;
2428 }
2429 myptr = self->ob_bytes;
2430 mysize = Py_SIZE(self);
2431 left = lstrip_helper(myptr, mysize, argptr, argsize);
2432 right = mysize;
2433 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002434 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002435 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2436}
2437
2438PyDoc_STRVAR(rstrip__doc__,
2439"B.rstrip([bytes]) -> bytearray\n\
2440\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002441Strip trailing bytes contained in the argument\n\
2442and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002443If the argument is omitted, strip trailing ASCII whitespace.");
2444static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002445bytearray_rstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002446{
2447 Py_ssize_t left, right, mysize, argsize;
2448 void *myptr, *argptr;
2449 PyObject *arg = Py_None;
2450 Py_buffer varg;
2451 if (!PyArg_ParseTuple(args, "|O:rstrip", &arg))
2452 return NULL;
2453 if (arg == Py_None) {
2454 argptr = "\t\n\r\f\v ";
2455 argsize = 6;
2456 }
2457 else {
2458 if (_getbuffer(arg, &varg) < 0)
2459 return NULL;
2460 argptr = varg.buf;
2461 argsize = varg.len;
2462 }
2463 myptr = self->ob_bytes;
2464 mysize = Py_SIZE(self);
2465 left = 0;
2466 right = rstrip_helper(myptr, mysize, argptr, argsize);
2467 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002468 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002469 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2470}
2471
2472PyDoc_STRVAR(decode_doc,
Victor Stinnerc911bbf2010-11-07 19:04:46 +00002473"B.decode(encoding='utf-8', errors='strict') -> str\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002474\n\
Victor Stinnere14e2122010-11-07 18:41:46 +00002475Decode B using the codec registered for encoding. Default encoding\n\
2476is 'utf-8'. errors may be given to set a different error\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002477handling scheme. Default is 'strict' meaning that encoding errors raise\n\
2478a UnicodeDecodeError. Other possible values are 'ignore' and 'replace'\n\
2479as well as any other name registered with codecs.register_error that is\n\
2480able to handle UnicodeDecodeErrors.");
2481
2482static PyObject *
Benjamin Peterson308d6372009-09-18 21:42:35 +00002483bytearray_decode(PyObject *self, PyObject *args, PyObject *kwargs)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002484{
2485 const char *encoding = NULL;
2486 const char *errors = NULL;
Benjamin Peterson308d6372009-09-18 21:42:35 +00002487 static char *kwlist[] = {"encoding", "errors", 0};
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002488
Benjamin Peterson308d6372009-09-18 21:42:35 +00002489 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ss:decode", kwlist, &encoding, &errors))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002490 return NULL;
2491 if (encoding == NULL)
2492 encoding = PyUnicode_GetDefaultEncoding();
Marc-André Lemburgb2750b52008-06-06 12:18:17 +00002493 return PyUnicode_FromEncodedObject(self, encoding, errors);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002494}
2495
2496PyDoc_STRVAR(alloc_doc,
2497"B.__alloc__() -> int\n\
2498\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002499Return the number of bytes actually allocated.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002500
2501static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002502bytearray_alloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002503{
2504 return PyLong_FromSsize_t(self->ob_alloc);
2505}
2506
2507PyDoc_STRVAR(join_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002508"B.join(iterable_of_bytes) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002509\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002510Concatenate any number of bytes/bytearray objects, with B\n\
2511in between each pair, and return the result as a new bytearray.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002512
2513static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002514bytearray_join(PyByteArrayObject *self, PyObject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002515{
2516 PyObject *seq;
2517 Py_ssize_t mysize = Py_SIZE(self);
2518 Py_ssize_t i;
2519 Py_ssize_t n;
2520 PyObject **items;
2521 Py_ssize_t totalsize = 0;
2522 PyObject *result;
2523 char *dest;
2524
2525 seq = PySequence_Fast(it, "can only join an iterable");
2526 if (seq == NULL)
2527 return NULL;
2528 n = PySequence_Fast_GET_SIZE(seq);
2529 items = PySequence_Fast_ITEMS(seq);
2530
2531 /* Compute the total size, and check that they are all bytes */
2532 /* XXX Shouldn't we use _getbuffer() on these items instead? */
2533 for (i = 0; i < n; i++) {
2534 PyObject *obj = items[i];
2535 if (!PyByteArray_Check(obj) && !PyBytes_Check(obj)) {
2536 PyErr_Format(PyExc_TypeError,
2537 "can only join an iterable of bytes "
2538 "(item %ld has type '%.100s')",
2539 /* XXX %ld isn't right on Win64 */
2540 (long)i, Py_TYPE(obj)->tp_name);
2541 goto error;
2542 }
2543 if (i > 0)
2544 totalsize += mysize;
2545 totalsize += Py_SIZE(obj);
2546 if (totalsize < 0) {
2547 PyErr_NoMemory();
2548 goto error;
2549 }
2550 }
2551
2552 /* Allocate the result, and copy the bytes */
2553 result = PyByteArray_FromStringAndSize(NULL, totalsize);
2554 if (result == NULL)
2555 goto error;
2556 dest = PyByteArray_AS_STRING(result);
2557 for (i = 0; i < n; i++) {
2558 PyObject *obj = items[i];
2559 Py_ssize_t size = Py_SIZE(obj);
2560 char *buf;
2561 if (PyByteArray_Check(obj))
2562 buf = PyByteArray_AS_STRING(obj);
2563 else
2564 buf = PyBytes_AS_STRING(obj);
2565 if (i) {
2566 memcpy(dest, self->ob_bytes, mysize);
2567 dest += mysize;
2568 }
2569 memcpy(dest, buf, size);
2570 dest += size;
2571 }
2572
2573 /* Done */
2574 Py_DECREF(seq);
2575 return result;
2576
2577 /* Error handling */
2578 error:
2579 Py_DECREF(seq);
2580 return NULL;
2581}
2582
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002583PyDoc_STRVAR(splitlines__doc__,
2584"B.splitlines([keepends]) -> list of lines\n\
2585\n\
2586Return a list of the lines in B, breaking at line boundaries.\n\
2587Line breaks are not included in the resulting list unless keepends\n\
2588is given and true.");
2589
2590static PyObject*
2591bytearray_splitlines(PyObject *self, PyObject *args)
2592{
2593 int keepends = 0;
2594
2595 if (!PyArg_ParseTuple(args, "|i:splitlines", &keepends))
2596 return NULL;
2597
2598 return stringlib_splitlines(
2599 (PyObject*) self, PyByteArray_AS_STRING(self),
2600 PyByteArray_GET_SIZE(self), keepends
2601 );
2602}
2603
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002604PyDoc_STRVAR(fromhex_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002605"bytearray.fromhex(string) -> bytearray (static method)\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002606\n\
2607Create a bytearray object from a string of hexadecimal numbers.\n\
2608Spaces between two numbers are accepted.\n\
2609Example: bytearray.fromhex('B9 01EF') -> bytearray(b'\\xb9\\x01\\xef').");
2610
2611static int
2612hex_digit_to_int(Py_UNICODE c)
2613{
2614 if (c >= 128)
2615 return -1;
Eric Smith6dc46f52009-04-27 20:39:49 +00002616 if (Py_ISDIGIT(c))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002617 return c - '0';
2618 else {
Eric Smith6dc46f52009-04-27 20:39:49 +00002619 if (Py_ISUPPER(c))
2620 c = Py_TOLOWER(c);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002621 if (c >= 'a' && c <= 'f')
2622 return c - 'a' + 10;
2623 }
2624 return -1;
2625}
2626
2627static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002628bytearray_fromhex(PyObject *cls, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002629{
2630 PyObject *newbytes, *hexobj;
2631 char *buf;
2632 Py_UNICODE *hex;
2633 Py_ssize_t hexlen, byteslen, i, j;
2634 int top, bot;
2635
2636 if (!PyArg_ParseTuple(args, "U:fromhex", &hexobj))
2637 return NULL;
2638 assert(PyUnicode_Check(hexobj));
2639 hexlen = PyUnicode_GET_SIZE(hexobj);
2640 hex = PyUnicode_AS_UNICODE(hexobj);
2641 byteslen = hexlen/2; /* This overestimates if there are spaces */
2642 newbytes = PyByteArray_FromStringAndSize(NULL, byteslen);
2643 if (!newbytes)
2644 return NULL;
2645 buf = PyByteArray_AS_STRING(newbytes);
2646 for (i = j = 0; i < hexlen; i += 2) {
2647 /* skip over spaces in the input */
2648 while (hex[i] == ' ')
2649 i++;
2650 if (i >= hexlen)
2651 break;
2652 top = hex_digit_to_int(hex[i]);
2653 bot = hex_digit_to_int(hex[i+1]);
2654 if (top == -1 || bot == -1) {
2655 PyErr_Format(PyExc_ValueError,
2656 "non-hexadecimal number found in "
2657 "fromhex() arg at position %zd", i);
2658 goto error;
2659 }
2660 buf[j++] = (top << 4) + bot;
2661 }
2662 if (PyByteArray_Resize(newbytes, j) < 0)
2663 goto error;
2664 return newbytes;
2665
2666 error:
2667 Py_DECREF(newbytes);
2668 return NULL;
2669}
2670
2671PyDoc_STRVAR(reduce_doc, "Return state information for pickling.");
2672
2673static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002674bytearray_reduce(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002675{
2676 PyObject *latin1, *dict;
2677 if (self->ob_bytes)
2678 latin1 = PyUnicode_DecodeLatin1(self->ob_bytes,
2679 Py_SIZE(self), NULL);
2680 else
2681 latin1 = PyUnicode_FromString("");
2682
2683 dict = PyObject_GetAttrString((PyObject *)self, "__dict__");
2684 if (dict == NULL) {
2685 PyErr_Clear();
2686 dict = Py_None;
2687 Py_INCREF(dict);
2688 }
2689
2690 return Py_BuildValue("(O(Ns)N)", Py_TYPE(self), latin1, "latin-1", dict);
2691}
2692
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002693PyDoc_STRVAR(sizeof_doc,
2694"B.__sizeof__() -> int\n\
2695 \n\
2696Returns the size of B in memory, in bytes");
2697static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002698bytearray_sizeof(PyByteArrayObject *self)
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002699{
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002700 Py_ssize_t res;
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002701
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002702 res = sizeof(PyByteArrayObject) + self->ob_alloc * sizeof(char);
2703 return PyLong_FromSsize_t(res);
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002704}
2705
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002706static PySequenceMethods bytearray_as_sequence = {
2707 (lenfunc)bytearray_length, /* sq_length */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002708 (binaryfunc)PyByteArray_Concat, /* sq_concat */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002709 (ssizeargfunc)bytearray_repeat, /* sq_repeat */
2710 (ssizeargfunc)bytearray_getitem, /* sq_item */
2711 0, /* sq_slice */
2712 (ssizeobjargproc)bytearray_setitem, /* sq_ass_item */
2713 0, /* sq_ass_slice */
2714 (objobjproc)bytearray_contains, /* sq_contains */
2715 (binaryfunc)bytearray_iconcat, /* sq_inplace_concat */
2716 (ssizeargfunc)bytearray_irepeat, /* sq_inplace_repeat */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002717};
2718
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002719static PyMappingMethods bytearray_as_mapping = {
2720 (lenfunc)bytearray_length,
2721 (binaryfunc)bytearray_subscript,
2722 (objobjargproc)bytearray_ass_subscript,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002723};
2724
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002725static PyBufferProcs bytearray_as_buffer = {
2726 (getbufferproc)bytearray_getbuffer,
2727 (releasebufferproc)bytearray_releasebuffer,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002728};
2729
2730static PyMethodDef
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002731bytearray_methods[] = {
2732 {"__alloc__", (PyCFunction)bytearray_alloc, METH_NOARGS, alloc_doc},
2733 {"__reduce__", (PyCFunction)bytearray_reduce, METH_NOARGS, reduce_doc},
2734 {"__sizeof__", (PyCFunction)bytearray_sizeof, METH_NOARGS, sizeof_doc},
2735 {"append", (PyCFunction)bytearray_append, METH_O, append__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002736 {"capitalize", (PyCFunction)stringlib_capitalize, METH_NOARGS,
2737 _Py_capitalize__doc__},
2738 {"center", (PyCFunction)stringlib_center, METH_VARARGS, center__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002739 {"count", (PyCFunction)bytearray_count, METH_VARARGS, count__doc__},
Benjamin Peterson308d6372009-09-18 21:42:35 +00002740 {"decode", (PyCFunction)bytearray_decode, METH_VARARGS | METH_KEYWORDS, decode_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002741 {"endswith", (PyCFunction)bytearray_endswith, METH_VARARGS, endswith__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002742 {"expandtabs", (PyCFunction)stringlib_expandtabs, METH_VARARGS,
2743 expandtabs__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002744 {"extend", (PyCFunction)bytearray_extend, METH_O, extend__doc__},
2745 {"find", (PyCFunction)bytearray_find, METH_VARARGS, find__doc__},
2746 {"fromhex", (PyCFunction)bytearray_fromhex, METH_VARARGS|METH_CLASS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002747 fromhex_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002748 {"index", (PyCFunction)bytearray_index, METH_VARARGS, index__doc__},
2749 {"insert", (PyCFunction)bytearray_insert, METH_VARARGS, insert__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002750 {"isalnum", (PyCFunction)stringlib_isalnum, METH_NOARGS,
2751 _Py_isalnum__doc__},
2752 {"isalpha", (PyCFunction)stringlib_isalpha, METH_NOARGS,
2753 _Py_isalpha__doc__},
2754 {"isdigit", (PyCFunction)stringlib_isdigit, METH_NOARGS,
2755 _Py_isdigit__doc__},
2756 {"islower", (PyCFunction)stringlib_islower, METH_NOARGS,
2757 _Py_islower__doc__},
2758 {"isspace", (PyCFunction)stringlib_isspace, METH_NOARGS,
2759 _Py_isspace__doc__},
2760 {"istitle", (PyCFunction)stringlib_istitle, METH_NOARGS,
2761 _Py_istitle__doc__},
2762 {"isupper", (PyCFunction)stringlib_isupper, METH_NOARGS,
2763 _Py_isupper__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002764 {"join", (PyCFunction)bytearray_join, METH_O, join_doc},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002765 {"ljust", (PyCFunction)stringlib_ljust, METH_VARARGS, ljust__doc__},
2766 {"lower", (PyCFunction)stringlib_lower, METH_NOARGS, _Py_lower__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002767 {"lstrip", (PyCFunction)bytearray_lstrip, METH_VARARGS, lstrip__doc__},
2768 {"maketrans", (PyCFunction)bytearray_maketrans, METH_VARARGS|METH_STATIC,
Georg Brandlabc38772009-04-12 15:51:51 +00002769 _Py_maketrans__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002770 {"partition", (PyCFunction)bytearray_partition, METH_O, partition__doc__},
2771 {"pop", (PyCFunction)bytearray_pop, METH_VARARGS, pop__doc__},
2772 {"remove", (PyCFunction)bytearray_remove, METH_O, remove__doc__},
2773 {"replace", (PyCFunction)bytearray_replace, METH_VARARGS, replace__doc__},
2774 {"reverse", (PyCFunction)bytearray_reverse, METH_NOARGS, reverse__doc__},
2775 {"rfind", (PyCFunction)bytearray_rfind, METH_VARARGS, rfind__doc__},
2776 {"rindex", (PyCFunction)bytearray_rindex, METH_VARARGS, rindex__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002777 {"rjust", (PyCFunction)stringlib_rjust, METH_VARARGS, rjust__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002778 {"rpartition", (PyCFunction)bytearray_rpartition, METH_O, rpartition__doc__},
2779 {"rsplit", (PyCFunction)bytearray_rsplit, METH_VARARGS, rsplit__doc__},
2780 {"rstrip", (PyCFunction)bytearray_rstrip, METH_VARARGS, rstrip__doc__},
2781 {"split", (PyCFunction)bytearray_split, METH_VARARGS, split__doc__},
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002782 {"splitlines", (PyCFunction)bytearray_splitlines, METH_VARARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002783 splitlines__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002784 {"startswith", (PyCFunction)bytearray_startswith, METH_VARARGS ,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002785 startswith__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002786 {"strip", (PyCFunction)bytearray_strip, METH_VARARGS, strip__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002787 {"swapcase", (PyCFunction)stringlib_swapcase, METH_NOARGS,
2788 _Py_swapcase__doc__},
2789 {"title", (PyCFunction)stringlib_title, METH_NOARGS, _Py_title__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002790 {"translate", (PyCFunction)bytearray_translate, METH_VARARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002791 translate__doc__},
2792 {"upper", (PyCFunction)stringlib_upper, METH_NOARGS, _Py_upper__doc__},
2793 {"zfill", (PyCFunction)stringlib_zfill, METH_VARARGS, zfill__doc__},
2794 {NULL}
2795};
2796
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002797PyDoc_STRVAR(bytearray_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002798"bytearray(iterable_of_ints) -> bytearray\n\
2799bytearray(string, encoding[, errors]) -> bytearray\n\
2800bytearray(bytes_or_bytearray) -> mutable copy of bytes_or_bytearray\n\
2801bytearray(memory_view) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002802\n\
2803Construct an mutable bytearray object from:\n\
2804 - an iterable yielding integers in range(256)\n\
2805 - a text string encoded using the specified encoding\n\
2806 - a bytes or a bytearray object\n\
2807 - any object implementing the buffer API.\n\
2808\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002809bytearray(int) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002810\n\
2811Construct a zero-initialized bytearray of the given length.");
2812
2813
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002814static PyObject *bytearray_iter(PyObject *seq);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002815
2816PyTypeObject PyByteArray_Type = {
2817 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2818 "bytearray",
2819 sizeof(PyByteArrayObject),
2820 0,
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002821 (destructor)bytearray_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002822 0, /* tp_print */
2823 0, /* tp_getattr */
2824 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00002825 0, /* tp_reserved */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002826 (reprfunc)bytearray_repr, /* tp_repr */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002827 0, /* tp_as_number */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002828 &bytearray_as_sequence, /* tp_as_sequence */
2829 &bytearray_as_mapping, /* tp_as_mapping */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002830 0, /* tp_hash */
2831 0, /* tp_call */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002832 bytearray_str, /* tp_str */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002833 PyObject_GenericGetAttr, /* tp_getattro */
2834 0, /* tp_setattro */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002835 &bytearray_as_buffer, /* tp_as_buffer */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002836 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /* tp_flags */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002837 bytearray_doc, /* tp_doc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002838 0, /* tp_traverse */
2839 0, /* tp_clear */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002840 (richcmpfunc)bytearray_richcompare, /* tp_richcompare */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002841 0, /* tp_weaklistoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002842 bytearray_iter, /* tp_iter */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002843 0, /* tp_iternext */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002844 bytearray_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002845 0, /* tp_members */
2846 0, /* tp_getset */
2847 0, /* tp_base */
2848 0, /* tp_dict */
2849 0, /* tp_descr_get */
2850 0, /* tp_descr_set */
2851 0, /* tp_dictoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002852 (initproc)bytearray_init, /* tp_init */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002853 PyType_GenericAlloc, /* tp_alloc */
2854 PyType_GenericNew, /* tp_new */
2855 PyObject_Del, /* tp_free */
2856};
2857
2858/*********************** Bytes Iterator ****************************/
2859
2860typedef struct {
2861 PyObject_HEAD
2862 Py_ssize_t it_index;
2863 PyByteArrayObject *it_seq; /* Set to NULL when iterator is exhausted */
2864} bytesiterobject;
2865
2866static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002867bytearrayiter_dealloc(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002868{
2869 _PyObject_GC_UNTRACK(it);
2870 Py_XDECREF(it->it_seq);
2871 PyObject_GC_Del(it);
2872}
2873
2874static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002875bytearrayiter_traverse(bytesiterobject *it, visitproc visit, void *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002876{
2877 Py_VISIT(it->it_seq);
2878 return 0;
2879}
2880
2881static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002882bytearrayiter_next(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002883{
2884 PyByteArrayObject *seq;
2885 PyObject *item;
2886
2887 assert(it != NULL);
2888 seq = it->it_seq;
2889 if (seq == NULL)
2890 return NULL;
2891 assert(PyByteArray_Check(seq));
2892
2893 if (it->it_index < PyByteArray_GET_SIZE(seq)) {
2894 item = PyLong_FromLong(
2895 (unsigned char)seq->ob_bytes[it->it_index]);
2896 if (item != NULL)
2897 ++it->it_index;
2898 return item;
2899 }
2900
2901 Py_DECREF(seq);
2902 it->it_seq = NULL;
2903 return NULL;
2904}
2905
2906static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002907bytesarrayiter_length_hint(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002908{
2909 Py_ssize_t len = 0;
2910 if (it->it_seq)
2911 len = PyByteArray_GET_SIZE(it->it_seq) - it->it_index;
2912 return PyLong_FromSsize_t(len);
2913}
2914
2915PyDoc_STRVAR(length_hint_doc,
2916 "Private method returning an estimate of len(list(it)).");
2917
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002918static PyMethodDef bytearrayiter_methods[] = {
2919 {"__length_hint__", (PyCFunction)bytesarrayiter_length_hint, METH_NOARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002920 length_hint_doc},
2921 {NULL, NULL} /* sentinel */
2922};
2923
2924PyTypeObject PyByteArrayIter_Type = {
2925 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2926 "bytearray_iterator", /* tp_name */
2927 sizeof(bytesiterobject), /* tp_basicsize */
2928 0, /* tp_itemsize */
2929 /* methods */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002930 (destructor)bytearrayiter_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002931 0, /* tp_print */
2932 0, /* tp_getattr */
2933 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00002934 0, /* tp_reserved */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002935 0, /* tp_repr */
2936 0, /* tp_as_number */
2937 0, /* tp_as_sequence */
2938 0, /* tp_as_mapping */
2939 0, /* tp_hash */
2940 0, /* tp_call */
2941 0, /* tp_str */
2942 PyObject_GenericGetAttr, /* tp_getattro */
2943 0, /* tp_setattro */
2944 0, /* tp_as_buffer */
2945 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */
2946 0, /* tp_doc */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002947 (traverseproc)bytearrayiter_traverse, /* tp_traverse */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002948 0, /* tp_clear */
2949 0, /* tp_richcompare */
2950 0, /* tp_weaklistoffset */
2951 PyObject_SelfIter, /* tp_iter */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002952 (iternextfunc)bytearrayiter_next, /* tp_iternext */
2953 bytearrayiter_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002954 0,
2955};
2956
2957static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002958bytearray_iter(PyObject *seq)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002959{
2960 bytesiterobject *it;
2961
2962 if (!PyByteArray_Check(seq)) {
2963 PyErr_BadInternalCall();
2964 return NULL;
2965 }
2966 it = PyObject_GC_New(bytesiterobject, &PyByteArrayIter_Type);
2967 if (it == NULL)
2968 return NULL;
2969 it->it_index = 0;
2970 Py_INCREF(seq);
2971 it->it_seq = (PyByteArrayObject *)seq;
2972 _PyObject_GC_TRACK(it);
2973 return (PyObject *)it;
2974}