blob: 5f575805d494030c041bbb6e3cce3d115c7e85b4 [file] [log] [blame]
Christian Heimes44720832008-05-26 13:01:01 +00001/* PyBytes (bytearray) implementation */
2
3#define PY_SSIZE_T_CLEAN
4#include "Python.h"
5#include "structmember.h"
6#include "bytes_methods.h"
7
Antoine Pitroue80a6a42010-01-17 12:26:20 +00008char _PyByteArray_empty_string[] = "";
Christian Heimes44720832008-05-26 13:01:01 +00009
10void
11PyByteArray_Fini(void)
12{
Christian Heimes44720832008-05-26 13:01:01 +000013}
14
15int
16PyByteArray_Init(void)
17{
Christian Heimes44720832008-05-26 13:01:01 +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
Georg Brandl3e483f62008-07-16 22:57:41 +000030 if (PyBytes_CheckExact(arg)) {
Christian Heimes44720832008-05-26 13:01:01 +000031 if (Py_SIZE(arg) != 1) {
32 PyErr_SetString(PyExc_ValueError, "string must be of size 1");
33 return 0;
34 }
Georg Brandl3e483f62008-07-16 22:57:41 +000035 *value = Py_CHARMASK(((PyBytesObject*)arg)->ob_sval[0]);
36 return 1;
37 }
38 else if (PyInt_Check(arg) || PyLong_Check(arg)) {
39 face_value = PyLong_AsLong(arg);
Christian Heimes44720832008-05-26 13:01:01 +000040 }
41 else {
Georg Brandl3e483f62008-07-16 22:57:41 +000042 PyObject *index = PyNumber_Index(arg);
43 if (index == NULL) {
44 PyErr_Format(PyExc_TypeError,
45 "an integer or string of size 1 is required");
46 return 0;
47 }
48 face_value = PyLong_AsLong(index);
49 Py_DECREF(index);
50 }
Georg Brandl3e483f62008-07-16 22:57:41 +000051
52 if (face_value < 0 || face_value >= 256) {
Georg Brandl3238a3e2008-07-16 23:17:46 +000053 /* this includes the OverflowError in case the long is too large */
Georg Brandl3e483f62008-07-16 22:57:41 +000054 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
Christian Heimes44720832008-05-26 13:01:01 +000055 return 0;
56 }
57
58 *value = face_value;
59 return 1;
60}
61
62static Py_ssize_t
Benjamin Petersond6720012009-04-18 15:31:34 +000063bytearray_buffer_getreadbuf(PyByteArrayObject *self, Py_ssize_t index, const void **ptr)
Christian Heimes44720832008-05-26 13:01:01 +000064{
65 if ( index != 0 ) {
66 PyErr_SetString(PyExc_SystemError,
67 "accessing non-existent bytes segment");
68 return -1;
69 }
Antoine Pitroue80a6a42010-01-17 12:26:20 +000070 *ptr = (void *)PyByteArray_AS_STRING(self);
Christian Heimes44720832008-05-26 13:01:01 +000071 return Py_SIZE(self);
72}
73
74static Py_ssize_t
Benjamin Petersond6720012009-04-18 15:31:34 +000075bytearray_buffer_getwritebuf(PyByteArrayObject *self, Py_ssize_t index, const void **ptr)
Christian Heimes44720832008-05-26 13:01:01 +000076{
77 if ( index != 0 ) {
78 PyErr_SetString(PyExc_SystemError,
79 "accessing non-existent bytes segment");
80 return -1;
81 }
Antoine Pitroue80a6a42010-01-17 12:26:20 +000082 *ptr = (void *)PyByteArray_AS_STRING(self);
Christian Heimes44720832008-05-26 13:01:01 +000083 return Py_SIZE(self);
84}
85
86static Py_ssize_t
Benjamin Petersond6720012009-04-18 15:31:34 +000087bytearray_buffer_getsegcount(PyByteArrayObject *self, Py_ssize_t *lenp)
Christian Heimes44720832008-05-26 13:01:01 +000088{
89 if ( lenp )
90 *lenp = Py_SIZE(self);
91 return 1;
92}
93
94static Py_ssize_t
Benjamin Petersond6720012009-04-18 15:31:34 +000095bytearray_buffer_getcharbuf(PyByteArrayObject *self, Py_ssize_t index, const char **ptr)
Christian Heimes44720832008-05-26 13:01:01 +000096{
97 if ( index != 0 ) {
98 PyErr_SetString(PyExc_SystemError,
99 "accessing non-existent bytes segment");
100 return -1;
101 }
Antoine Pitroue80a6a42010-01-17 12:26:20 +0000102 *ptr = PyByteArray_AS_STRING(self);
Christian Heimes44720832008-05-26 13:01:01 +0000103 return Py_SIZE(self);
104}
105
106static int
Benjamin Petersond6720012009-04-18 15:31:34 +0000107bytearray_getbuffer(PyByteArrayObject *obj, Py_buffer *view, int flags)
Christian Heimes44720832008-05-26 13:01:01 +0000108{
Antoine Pitrou619f16e2010-06-09 16:24:00 +0000109 int ret;
110 void *ptr;
111 if (view == NULL) {
112 obj->ob_exports++;
113 return 0;
114 }
115 ptr = (void *) PyByteArray_AS_STRING(obj);
116 ret = PyBuffer_FillInfo(view, (PyObject*)obj, ptr, Py_SIZE(obj), 0, flags);
117 if (ret >= 0) {
118 obj->ob_exports++;
119 }
120 return ret;
Christian Heimes44720832008-05-26 13:01:01 +0000121}
122
123static void
Benjamin Petersond6720012009-04-18 15:31:34 +0000124bytearray_releasebuffer(PyByteArrayObject *obj, Py_buffer *view)
Christian Heimes44720832008-05-26 13:01:01 +0000125{
Antoine Pitrou619f16e2010-06-09 16:24:00 +0000126 obj->ob_exports--;
Christian Heimes44720832008-05-26 13:01:01 +0000127}
128
129static Py_ssize_t
130_getbuffer(PyObject *obj, Py_buffer *view)
131{
132 PyBufferProcs *buffer = Py_TYPE(obj)->tp_as_buffer;
133
134 if (buffer == NULL || buffer->bf_getbuffer == NULL)
135 {
136 PyErr_Format(PyExc_TypeError,
137 "Type %.100s doesn't support the buffer API",
138 Py_TYPE(obj)->tp_name);
139 return -1;
140 }
141
142 if (buffer->bf_getbuffer(obj, view, PyBUF_SIMPLE) < 0)
143 return -1;
144 return view->len;
145}
146
Antoine Pitrouae5bece2008-12-06 21:29:24 +0000147static int
148_canresize(PyByteArrayObject *self)
149{
150 if (self->ob_exports > 0) {
151 PyErr_SetString(PyExc_BufferError,
152 "Existing exports of data: object cannot be re-sized");
153 return 0;
154 }
155 return 1;
156}
157
Christian Heimes44720832008-05-26 13:01:01 +0000158/* Direct API functions */
159
160PyObject *
161PyByteArray_FromObject(PyObject *input)
162{
163 return PyObject_CallFunctionObjArgs((PyObject *)&PyByteArray_Type,
164 input, NULL);
165}
166
167PyObject *
168PyByteArray_FromStringAndSize(const char *bytes, Py_ssize_t size)
169{
170 PyByteArrayObject *new;
171 Py_ssize_t alloc;
172
173 if (size < 0) {
174 PyErr_SetString(PyExc_SystemError,
175 "Negative size passed to PyByteArray_FromStringAndSize");
176 return NULL;
177 }
178
179 new = PyObject_New(PyByteArrayObject, &PyByteArray_Type);
180 if (new == NULL)
181 return NULL;
182
183 if (size == 0) {
184 new->ob_bytes = NULL;
185 alloc = 0;
186 }
187 else {
188 alloc = size + 1;
189 new->ob_bytes = PyMem_Malloc(alloc);
190 if (new->ob_bytes == NULL) {
191 Py_DECREF(new);
192 return PyErr_NoMemory();
193 }
Antoine Pitroue80a6a42010-01-17 12:26:20 +0000194 if (bytes != NULL && size > 0)
Christian Heimes44720832008-05-26 13:01:01 +0000195 memcpy(new->ob_bytes, bytes, size);
196 new->ob_bytes[size] = '\0'; /* Trailing null byte */
197 }
198 Py_SIZE(new) = size;
199 new->ob_alloc = alloc;
200 new->ob_exports = 0;
201
202 return (PyObject *)new;
203}
204
205Py_ssize_t
206PyByteArray_Size(PyObject *self)
207{
208 assert(self != NULL);
209 assert(PyByteArray_Check(self));
210
211 return PyByteArray_GET_SIZE(self);
212}
213
214char *
215PyByteArray_AsString(PyObject *self)
216{
217 assert(self != NULL);
218 assert(PyByteArray_Check(self));
219
220 return PyByteArray_AS_STRING(self);
221}
222
223int
224PyByteArray_Resize(PyObject *self, Py_ssize_t size)
225{
226 void *sval;
227 Py_ssize_t alloc = ((PyByteArrayObject *)self)->ob_alloc;
228
229 assert(self != NULL);
230 assert(PyByteArray_Check(self));
231 assert(size >= 0);
232
Antoine Pitrouae5bece2008-12-06 21:29:24 +0000233 if (size == Py_SIZE(self)) {
234 return 0;
235 }
236 if (!_canresize((PyByteArrayObject *)self)) {
237 return -1;
238 }
239
Christian Heimes44720832008-05-26 13:01:01 +0000240 if (size < alloc / 2) {
241 /* Major downsize; resize down to exact size */
242 alloc = size + 1;
243 }
244 else if (size < alloc) {
245 /* Within allocated size; quick exit */
246 Py_SIZE(self) = size;
247 ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null */
248 return 0;
249 }
250 else if (size <= alloc * 1.125) {
251 /* Moderate upsize; overallocate similar to list_resize() */
252 alloc = size + (size >> 3) + (size < 9 ? 3 : 6);
253 }
254 else {
255 /* Major upsize; resize up to exact size */
256 alloc = size + 1;
257 }
258
Christian Heimes44720832008-05-26 13:01:01 +0000259 sval = PyMem_Realloc(((PyByteArrayObject *)self)->ob_bytes, alloc);
260 if (sval == NULL) {
261 PyErr_NoMemory();
262 return -1;
263 }
264
265 ((PyByteArrayObject *)self)->ob_bytes = sval;
266 Py_SIZE(self) = size;
267 ((PyByteArrayObject *)self)->ob_alloc = alloc;
268 ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null byte */
269
270 return 0;
271}
272
273PyObject *
274PyByteArray_Concat(PyObject *a, PyObject *b)
275{
276 Py_ssize_t size;
277 Py_buffer va, vb;
278 PyByteArrayObject *result = NULL;
279
280 va.len = -1;
281 vb.len = -1;
282 if (_getbuffer(a, &va) < 0 ||
283 _getbuffer(b, &vb) < 0) {
284 PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s",
285 Py_TYPE(a)->tp_name, Py_TYPE(b)->tp_name);
286 goto done;
287 }
288
289 size = va.len + vb.len;
290 if (size < 0) {
Hirokazu Yamamotoa3e6c972009-03-05 09:34:14 +0000291 PyErr_NoMemory();
Christian Heimes44720832008-05-26 13:01:01 +0000292 goto done;
293 }
294
295 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, size);
296 if (result != NULL) {
297 memcpy(result->ob_bytes, va.buf, va.len);
298 memcpy(result->ob_bytes + va.len, vb.buf, vb.len);
299 }
300
301 done:
302 if (va.len != -1)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000303 PyBuffer_Release(&va);
Christian Heimes44720832008-05-26 13:01:01 +0000304 if (vb.len != -1)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000305 PyBuffer_Release(&vb);
Christian Heimes44720832008-05-26 13:01:01 +0000306 return (PyObject *)result;
307}
308
309/* Functions stuffed into the type object */
310
311static Py_ssize_t
Benjamin Petersond6720012009-04-18 15:31:34 +0000312bytearray_length(PyByteArrayObject *self)
Christian Heimes44720832008-05-26 13:01:01 +0000313{
314 return Py_SIZE(self);
315}
316
317static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +0000318bytearray_iconcat(PyByteArrayObject *self, PyObject *other)
Christian Heimes44720832008-05-26 13:01:01 +0000319{
320 Py_ssize_t mysize;
321 Py_ssize_t size;
322 Py_buffer vo;
323
324 if (_getbuffer(other, &vo) < 0) {
325 PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s",
326 Py_TYPE(other)->tp_name, Py_TYPE(self)->tp_name);
327 return NULL;
328 }
329
330 mysize = Py_SIZE(self);
331 size = mysize + vo.len;
332 if (size < 0) {
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000333 PyBuffer_Release(&vo);
Christian Heimes44720832008-05-26 13:01:01 +0000334 return PyErr_NoMemory();
335 }
336 if (size < self->ob_alloc) {
337 Py_SIZE(self) = size;
338 self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */
339 }
340 else if (PyByteArray_Resize((PyObject *)self, size) < 0) {
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000341 PyBuffer_Release(&vo);
Christian Heimes44720832008-05-26 13:01:01 +0000342 return NULL;
343 }
344 memcpy(self->ob_bytes + mysize, vo.buf, vo.len);
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000345 PyBuffer_Release(&vo);
Christian Heimes44720832008-05-26 13:01:01 +0000346 Py_INCREF(self);
347 return (PyObject *)self;
348}
349
350static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +0000351bytearray_repeat(PyByteArrayObject *self, Py_ssize_t count)
Christian Heimes44720832008-05-26 13:01:01 +0000352{
353 PyByteArrayObject *result;
354 Py_ssize_t mysize;
355 Py_ssize_t size;
356
357 if (count < 0)
358 count = 0;
359 mysize = Py_SIZE(self);
360 size = mysize * count;
361 if (count != 0 && size / count != mysize)
362 return PyErr_NoMemory();
363 result = (PyByteArrayObject *)PyByteArray_FromStringAndSize(NULL, size);
364 if (result != NULL && size != 0) {
365 if (mysize == 1)
366 memset(result->ob_bytes, self->ob_bytes[0], size);
367 else {
368 Py_ssize_t i;
369 for (i = 0; i < count; i++)
370 memcpy(result->ob_bytes + i*mysize, self->ob_bytes, mysize);
371 }
372 }
373 return (PyObject *)result;
374}
375
376static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +0000377bytearray_irepeat(PyByteArrayObject *self, Py_ssize_t count)
Christian Heimes44720832008-05-26 13:01:01 +0000378{
379 Py_ssize_t mysize;
380 Py_ssize_t size;
381
382 if (count < 0)
383 count = 0;
384 mysize = Py_SIZE(self);
385 size = mysize * count;
386 if (count != 0 && size / count != mysize)
387 return PyErr_NoMemory();
388 if (size < self->ob_alloc) {
389 Py_SIZE(self) = size;
390 self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */
391 }
392 else if (PyByteArray_Resize((PyObject *)self, size) < 0)
393 return NULL;
394
395 if (mysize == 1)
396 memset(self->ob_bytes, self->ob_bytes[0], size);
397 else {
398 Py_ssize_t i;
399 for (i = 1; i < count; i++)
400 memcpy(self->ob_bytes + i*mysize, self->ob_bytes, mysize);
401 }
402
403 Py_INCREF(self);
404 return (PyObject *)self;
405}
406
407static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +0000408bytearray_getitem(PyByteArrayObject *self, Py_ssize_t i)
Christian Heimes44720832008-05-26 13:01:01 +0000409{
410 if (i < 0)
411 i += Py_SIZE(self);
412 if (i < 0 || i >= Py_SIZE(self)) {
413 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
414 return NULL;
415 }
416 return PyInt_FromLong((unsigned char)(self->ob_bytes[i]));
417}
418
419static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +0000420bytearray_subscript(PyByteArrayObject *self, PyObject *index)
Christian Heimes44720832008-05-26 13:01:01 +0000421{
Georg Brandl3e483f62008-07-16 22:57:41 +0000422 if (PyIndex_Check(index)) {
423 Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError);
Christian Heimes44720832008-05-26 13:01:01 +0000424
425 if (i == -1 && PyErr_Occurred())
426 return NULL;
427
428 if (i < 0)
429 i += PyByteArray_GET_SIZE(self);
430
431 if (i < 0 || i >= Py_SIZE(self)) {
432 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
433 return NULL;
434 }
435 return PyInt_FromLong((unsigned char)(self->ob_bytes[i]));
436 }
Georg Brandl3e483f62008-07-16 22:57:41 +0000437 else if (PySlice_Check(index)) {
Christian Heimes44720832008-05-26 13:01:01 +0000438 Py_ssize_t start, stop, step, slicelength, cur, i;
Georg Brandl3e483f62008-07-16 22:57:41 +0000439 if (PySlice_GetIndicesEx((PySliceObject *)index,
Christian Heimes44720832008-05-26 13:01:01 +0000440 PyByteArray_GET_SIZE(self),
441 &start, &stop, &step, &slicelength) < 0) {
442 return NULL;
443 }
444
445 if (slicelength <= 0)
446 return PyByteArray_FromStringAndSize("", 0);
447 else if (step == 1) {
448 return PyByteArray_FromStringAndSize(self->ob_bytes + start,
449 slicelength);
450 }
451 else {
452 char *source_buf = PyByteArray_AS_STRING(self);
453 char *result_buf = (char *)PyMem_Malloc(slicelength);
454 PyObject *result;
455
456 if (result_buf == NULL)
457 return PyErr_NoMemory();
458
459 for (cur = start, i = 0; i < slicelength;
460 cur += step, i++) {
461 result_buf[i] = source_buf[cur];
462 }
463 result = PyByteArray_FromStringAndSize(result_buf, slicelength);
464 PyMem_Free(result_buf);
465 return result;
466 }
467 }
468 else {
469 PyErr_SetString(PyExc_TypeError, "bytearray indices must be integers");
470 return NULL;
471 }
472}
473
474static int
Benjamin Petersond6720012009-04-18 15:31:34 +0000475bytearray_setslice(PyByteArrayObject *self, Py_ssize_t lo, Py_ssize_t hi,
Christian Heimes44720832008-05-26 13:01:01 +0000476 PyObject *values)
477{
478 Py_ssize_t avail, needed;
479 void *bytes;
480 Py_buffer vbytes;
481 int res = 0;
482
483 vbytes.len = -1;
484 if (values == (PyObject *)self) {
485 /* Make a copy and call this function recursively */
486 int err;
487 values = PyByteArray_FromObject(values);
488 if (values == NULL)
489 return -1;
Benjamin Petersond6720012009-04-18 15:31:34 +0000490 err = bytearray_setslice(self, lo, hi, values);
Christian Heimes44720832008-05-26 13:01:01 +0000491 Py_DECREF(values);
492 return err;
493 }
494 if (values == NULL) {
495 /* del b[lo:hi] */
496 bytes = NULL;
497 needed = 0;
498 }
499 else {
500 if (_getbuffer(values, &vbytes) < 0) {
501 PyErr_Format(PyExc_TypeError,
Neal Norwitzc86b54c2008-07-20 19:35:23 +0000502 "can't set bytearray slice from %.100s",
Christian Heimes44720832008-05-26 13:01:01 +0000503 Py_TYPE(values)->tp_name);
504 return -1;
505 }
506 needed = vbytes.len;
507 bytes = vbytes.buf;
508 }
509
510 if (lo < 0)
511 lo = 0;
512 if (hi < lo)
513 hi = lo;
514 if (hi > Py_SIZE(self))
515 hi = Py_SIZE(self);
516
517 avail = hi - lo;
518 if (avail < 0)
519 lo = hi = avail = 0;
520
521 if (avail != needed) {
522 if (avail > needed) {
Antoine Pitrouae5bece2008-12-06 21:29:24 +0000523 if (!_canresize(self)) {
524 res = -1;
525 goto finish;
526 }
Christian Heimes44720832008-05-26 13:01:01 +0000527 /*
528 0 lo hi old_size
529 | |<----avail----->|<-----tomove------>|
530 | |<-needed->|<-----tomove------>|
531 0 lo new_hi new_size
532 */
533 memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi,
534 Py_SIZE(self) - hi);
535 }
536 /* XXX(nnorwitz): need to verify this can't overflow! */
537 if (PyByteArray_Resize((PyObject *)self,
538 Py_SIZE(self) + needed - avail) < 0) {
539 res = -1;
540 goto finish;
541 }
542 if (avail < needed) {
543 /*
544 0 lo hi old_size
545 | |<-avail->|<-----tomove------>|
546 | |<----needed---->|<-----tomove------>|
547 0 lo new_hi new_size
548 */
549 memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi,
550 Py_SIZE(self) - lo - needed);
551 }
552 }
553
554 if (needed > 0)
555 memcpy(self->ob_bytes + lo, bytes, needed);
556
557
558 finish:
559 if (vbytes.len != -1)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000560 PyBuffer_Release(&vbytes);
Christian Heimes44720832008-05-26 13:01:01 +0000561 return res;
562}
563
564static int
Benjamin Petersond6720012009-04-18 15:31:34 +0000565bytearray_setitem(PyByteArrayObject *self, Py_ssize_t i, PyObject *value)
Christian Heimes44720832008-05-26 13:01:01 +0000566{
567 int ival;
568
569 if (i < 0)
570 i += Py_SIZE(self);
571
572 if (i < 0 || i >= Py_SIZE(self)) {
573 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
574 return -1;
575 }
576
577 if (value == NULL)
Benjamin Petersond6720012009-04-18 15:31:34 +0000578 return bytearray_setslice(self, i, i+1, NULL);
Christian Heimes44720832008-05-26 13:01:01 +0000579
580 if (!_getbytevalue(value, &ival))
581 return -1;
582
583 self->ob_bytes[i] = ival;
584 return 0;
585}
586
587static int
Benjamin Petersond6720012009-04-18 15:31:34 +0000588bytearray_ass_subscript(PyByteArrayObject *self, PyObject *index, PyObject *values)
Christian Heimes44720832008-05-26 13:01:01 +0000589{
590 Py_ssize_t start, stop, step, slicelen, needed;
591 char *bytes;
592
Georg Brandl3e483f62008-07-16 22:57:41 +0000593 if (PyIndex_Check(index)) {
594 Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError);
Christian Heimes44720832008-05-26 13:01:01 +0000595
596 if (i == -1 && PyErr_Occurred())
597 return -1;
598
599 if (i < 0)
600 i += PyByteArray_GET_SIZE(self);
601
602 if (i < 0 || i >= Py_SIZE(self)) {
603 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
604 return -1;
605 }
606
607 if (values == NULL) {
608 /* Fall through to slice assignment */
609 start = i;
610 stop = i + 1;
611 step = 1;
612 slicelen = 1;
613 }
614 else {
Georg Brandl3e483f62008-07-16 22:57:41 +0000615 int ival;
616 if (!_getbytevalue(values, &ival))
Christian Heimes44720832008-05-26 13:01:01 +0000617 return -1;
Christian Heimes44720832008-05-26 13:01:01 +0000618 self->ob_bytes[i] = (char)ival;
619 return 0;
620 }
621 }
Georg Brandl3e483f62008-07-16 22:57:41 +0000622 else if (PySlice_Check(index)) {
623 if (PySlice_GetIndicesEx((PySliceObject *)index,
Christian Heimes44720832008-05-26 13:01:01 +0000624 PyByteArray_GET_SIZE(self),
625 &start, &stop, &step, &slicelen) < 0) {
626 return -1;
627 }
628 }
629 else {
630 PyErr_SetString(PyExc_TypeError, "bytearray indices must be integer");
631 return -1;
632 }
633
634 if (values == NULL) {
635 bytes = NULL;
636 needed = 0;
637 }
638 else if (values == (PyObject *)self || !PyByteArray_Check(values)) {
Christian Heimes146a5fe2012-11-03 23:07:59 +0100639 int err;
Ezio Melotti67dc4a82012-11-03 21:10:45 +0200640 if (PyNumber_Check(values) || PyUnicode_Check(values)) {
641 PyErr_SetString(PyExc_TypeError,
642 "can assign only bytes, buffers, or iterables "
643 "of ints in range(0, 256)");
644 return -1;
645 }
Georg Brandl28dadd92011-02-25 10:50:32 +0000646 /* Make a copy and call this function recursively */
Christian Heimes44720832008-05-26 13:01:01 +0000647 values = PyByteArray_FromObject(values);
648 if (values == NULL)
649 return -1;
Benjamin Petersond6720012009-04-18 15:31:34 +0000650 err = bytearray_ass_subscript(self, index, values);
Christian Heimes44720832008-05-26 13:01:01 +0000651 Py_DECREF(values);
652 return err;
653 }
654 else {
655 assert(PyByteArray_Check(values));
656 bytes = ((PyByteArrayObject *)values)->ob_bytes;
657 needed = Py_SIZE(values);
658 }
659 /* Make sure b[5:2] = ... inserts before 5, not before 2. */
660 if ((step < 0 && start < stop) ||
661 (step > 0 && start > stop))
662 stop = start;
663 if (step == 1) {
664 if (slicelen != needed) {
Antoine Pitrouae5bece2008-12-06 21:29:24 +0000665 if (!_canresize(self))
666 return -1;
Christian Heimes44720832008-05-26 13:01:01 +0000667 if (slicelen > needed) {
668 /*
669 0 start stop old_size
670 | |<---slicelen--->|<-----tomove------>|
671 | |<-needed->|<-----tomove------>|
672 0 lo new_hi new_size
673 */
674 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
675 Py_SIZE(self) - stop);
676 }
677 if (PyByteArray_Resize((PyObject *)self,
678 Py_SIZE(self) + needed - slicelen) < 0)
679 return -1;
680 if (slicelen < needed) {
681 /*
682 0 lo hi old_size
683 | |<-avail->|<-----tomove------>|
684 | |<----needed---->|<-----tomove------>|
685 0 lo new_hi new_size
686 */
687 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
688 Py_SIZE(self) - start - needed);
689 }
690 }
691
692 if (needed > 0)
693 memcpy(self->ob_bytes + start, bytes, needed);
694
695 return 0;
696 }
697 else {
698 if (needed == 0) {
699 /* Delete slice */
Mark Dickinson36ecd672010-01-29 17:11:39 +0000700 size_t cur;
701 Py_ssize_t i;
Christian Heimes44720832008-05-26 13:01:01 +0000702
Antoine Pitrouae5bece2008-12-06 21:29:24 +0000703 if (!_canresize(self))
704 return -1;
Christian Heimes44720832008-05-26 13:01:01 +0000705 if (step < 0) {
706 stop = start + 1;
707 start = stop + step * (slicelen - 1) - 1;
708 step = -step;
709 }
710 for (cur = start, i = 0;
711 i < slicelen; cur += step, i++) {
712 Py_ssize_t lim = step - 1;
713
Mark Dickinson2d7911e2010-02-14 12:31:26 +0000714 if (cur + step >= (size_t)PyByteArray_GET_SIZE(self))
Christian Heimes44720832008-05-26 13:01:01 +0000715 lim = PyByteArray_GET_SIZE(self) - cur - 1;
716
717 memmove(self->ob_bytes + cur - i,
718 self->ob_bytes + cur + 1, lim);
719 }
720 /* Move the tail of the bytes, in one chunk */
721 cur = start + slicelen*step;
Mark Dickinson2d7911e2010-02-14 12:31:26 +0000722 if (cur < (size_t)PyByteArray_GET_SIZE(self)) {
Christian Heimes44720832008-05-26 13:01:01 +0000723 memmove(self->ob_bytes + cur - slicelen,
724 self->ob_bytes + cur,
725 PyByteArray_GET_SIZE(self) - cur);
726 }
727 if (PyByteArray_Resize((PyObject *)self,
728 PyByteArray_GET_SIZE(self) - slicelen) < 0)
729 return -1;
730
731 return 0;
732 }
733 else {
734 /* Assign slice */
735 Py_ssize_t cur, i;
736
737 if (needed != slicelen) {
738 PyErr_Format(PyExc_ValueError,
739 "attempt to assign bytes of size %zd "
740 "to extended slice of size %zd",
741 needed, slicelen);
742 return -1;
743 }
744 for (cur = start, i = 0; i < slicelen; cur += step, i++)
745 self->ob_bytes[cur] = bytes[i];
746 return 0;
747 }
748 }
749}
750
751static int
Benjamin Petersond6720012009-04-18 15:31:34 +0000752bytearray_init(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes44720832008-05-26 13:01:01 +0000753{
754 static char *kwlist[] = {"source", "encoding", "errors", 0};
755 PyObject *arg = NULL;
756 const char *encoding = NULL;
757 const char *errors = NULL;
758 Py_ssize_t count;
759 PyObject *it;
760 PyObject *(*iternext)(PyObject *);
761
762 if (Py_SIZE(self) != 0) {
763 /* Empty previous contents (yes, do this first of all!) */
764 if (PyByteArray_Resize((PyObject *)self, 0) < 0)
765 return -1;
766 }
767
768 /* Parse arguments */
Neal Norwitzc86b54c2008-07-20 19:35:23 +0000769 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|Oss:bytearray", kwlist,
Christian Heimes44720832008-05-26 13:01:01 +0000770 &arg, &encoding, &errors))
771 return -1;
772
773 /* Make a quick exit if no first argument */
774 if (arg == NULL) {
775 if (encoding != NULL || errors != NULL) {
776 PyErr_SetString(PyExc_TypeError,
777 "encoding or errors without sequence argument");
778 return -1;
779 }
780 return 0;
781 }
782
783 if (PyBytes_Check(arg)) {
784 PyObject *new, *encoded;
785 if (encoding != NULL) {
Serhiy Storchakac7797dc2015-05-31 20:21:00 +0300786 encoded = _PyCodec_EncodeText(arg, encoding, errors);
Christian Heimes44720832008-05-26 13:01:01 +0000787 if (encoded == NULL)
788 return -1;
789 assert(PyBytes_Check(encoded));
790 }
791 else {
792 encoded = arg;
793 Py_INCREF(arg);
794 }
Benjamin Petersond6720012009-04-18 15:31:34 +0000795 new = bytearray_iconcat(self, arg);
Christian Heimes44720832008-05-26 13:01:01 +0000796 Py_DECREF(encoded);
797 if (new == NULL)
798 return -1;
799 Py_DECREF(new);
800 return 0;
801 }
802
Benjamin Peterson78821dd2009-01-25 17:15:10 +0000803#ifdef Py_USING_UNICODE
Christian Heimes44720832008-05-26 13:01:01 +0000804 if (PyUnicode_Check(arg)) {
805 /* Encode via the codec registry */
806 PyObject *encoded, *new;
807 if (encoding == NULL) {
808 PyErr_SetString(PyExc_TypeError,
809 "unicode argument without an encoding");
810 return -1;
811 }
Serhiy Storchakac7797dc2015-05-31 20:21:00 +0300812 encoded = _PyCodec_EncodeText(arg, encoding, errors);
Christian Heimes44720832008-05-26 13:01:01 +0000813 if (encoded == NULL)
814 return -1;
815 assert(PyBytes_Check(encoded));
Benjamin Petersond6720012009-04-18 15:31:34 +0000816 new = bytearray_iconcat(self, encoded);
Christian Heimes44720832008-05-26 13:01:01 +0000817 Py_DECREF(encoded);
818 if (new == NULL)
819 return -1;
820 Py_DECREF(new);
821 return 0;
822 }
Benjamin Peterson78821dd2009-01-25 17:15:10 +0000823#endif
Christian Heimes44720832008-05-26 13:01:01 +0000824
825 /* If it's not unicode, there can't be encoding or errors */
826 if (encoding != NULL || errors != NULL) {
827 PyErr_SetString(PyExc_TypeError,
828 "encoding or errors without a string argument");
829 return -1;
830 }
831
832 /* Is it an int? */
Benjamin Peterson821a8ea2010-04-16 22:35:38 +0000833 count = PyNumber_AsSsize_t(arg, PyExc_OverflowError);
834 if (count == -1 && PyErr_Occurred()) {
835 if (PyErr_ExceptionMatches(PyExc_OverflowError))
Christian Heimes44720832008-05-26 13:01:01 +0000836 return -1;
Benjamin Petersonae530c22010-04-16 22:52:44 +0000837 PyErr_Clear();
Benjamin Peterson821a8ea2010-04-16 22:35:38 +0000838 }
839 else if (count < 0) {
840 PyErr_SetString(PyExc_ValueError, "negative count");
841 return -1;
842 }
843 else {
Christian Heimes44720832008-05-26 13:01:01 +0000844 if (count > 0) {
845 if (PyByteArray_Resize((PyObject *)self, count))
846 return -1;
847 memset(self->ob_bytes, 0, count);
848 }
849 return 0;
850 }
851
852 /* Use the buffer API */
853 if (PyObject_CheckBuffer(arg)) {
854 Py_ssize_t size;
855 Py_buffer view;
856 if (PyObject_GetBuffer(arg, &view, PyBUF_FULL_RO) < 0)
857 return -1;
858 size = view.len;
859 if (PyByteArray_Resize((PyObject *)self, size) < 0) goto fail;
860 if (PyBuffer_ToContiguous(self->ob_bytes, &view, size, 'C') < 0)
861 goto fail;
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000862 PyBuffer_Release(&view);
Christian Heimes44720832008-05-26 13:01:01 +0000863 return 0;
864 fail:
Martin v. Löwisf91d46a2008-08-12 14:49:50 +0000865 PyBuffer_Release(&view);
Christian Heimes44720832008-05-26 13:01:01 +0000866 return -1;
867 }
868
869 /* XXX Optimize this if the arguments is a list, tuple */
870
871 /* Get the iterator */
872 it = PyObject_GetIter(arg);
873 if (it == NULL)
874 return -1;
875 iternext = *Py_TYPE(it)->tp_iternext;
876
877 /* Run the iterator to exhaustion */
878 for (;;) {
879 PyObject *item;
Georg Brandl3e758462008-07-16 23:10:05 +0000880 int rc, value;
Christian Heimes44720832008-05-26 13:01:01 +0000881
882 /* Get the next item */
883 item = iternext(it);
884 if (item == NULL) {
885 if (PyErr_Occurred()) {
886 if (!PyErr_ExceptionMatches(PyExc_StopIteration))
887 goto error;
888 PyErr_Clear();
889 }
890 break;
891 }
892
893 /* Interpret it as an int (__index__) */
Georg Brandl3e758462008-07-16 23:10:05 +0000894 rc = _getbytevalue(item, &value);
Christian Heimes44720832008-05-26 13:01:01 +0000895 Py_DECREF(item);
Georg Brandl3e758462008-07-16 23:10:05 +0000896 if (!rc)
Christian Heimes44720832008-05-26 13:01:01 +0000897 goto error;
898
Christian Heimes44720832008-05-26 13:01:01 +0000899 /* Append the byte */
900 if (Py_SIZE(self) < self->ob_alloc)
901 Py_SIZE(self)++;
902 else if (PyByteArray_Resize((PyObject *)self, Py_SIZE(self)+1) < 0)
903 goto error;
904 self->ob_bytes[Py_SIZE(self)-1] = value;
905 }
906
907 /* Clean up and return success */
908 Py_DECREF(it);
909 return 0;
910
911 error:
912 /* Error handling when it != NULL */
913 Py_DECREF(it);
914 return -1;
915}
916
917/* Mostly copied from string_repr, but without the
918 "smart quote" functionality. */
919static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +0000920bytearray_repr(PyByteArrayObject *self)
Christian Heimes44720832008-05-26 13:01:01 +0000921{
922 static const char *hexdigits = "0123456789abcdef";
923 const char *quote_prefix = "bytearray(b";
924 const char *quote_postfix = ")";
925 Py_ssize_t length = Py_SIZE(self);
926 /* 14 == strlen(quote_prefix) + 2 + strlen(quote_postfix) */
Mark Dickinson2d7911e2010-02-14 12:31:26 +0000927 size_t newsize;
Christian Heimes44720832008-05-26 13:01:01 +0000928 PyObject *v;
Mark Dickinson2d7911e2010-02-14 12:31:26 +0000929 if (length > (PY_SSIZE_T_MAX - 14) / 4) {
Christian Heimes44720832008-05-26 13:01:01 +0000930 PyErr_SetString(PyExc_OverflowError,
931 "bytearray object is too large to make repr");
932 return NULL;
933 }
Mark Dickinson2d7911e2010-02-14 12:31:26 +0000934 newsize = 14 + 4 * length;
Benjamin Peterson78821dd2009-01-25 17:15:10 +0000935 v = PyString_FromStringAndSize(NULL, newsize);
Christian Heimes44720832008-05-26 13:01:01 +0000936 if (v == NULL) {
937 return NULL;
938 }
939 else {
940 register Py_ssize_t i;
Benjamin Peterson78821dd2009-01-25 17:15:10 +0000941 register char c;
942 register char *p;
Christian Heimes44720832008-05-26 13:01:01 +0000943 int quote;
944
945 /* Figure out which quote to use; single is preferred */
946 quote = '\'';
947 {
948 char *test, *start;
949 start = PyByteArray_AS_STRING(self);
950 for (test = start; test < start+length; ++test) {
951 if (*test == '"') {
952 quote = '\''; /* back to single */
953 goto decided;
954 }
955 else if (*test == '\'')
956 quote = '"';
957 }
958 decided:
959 ;
960 }
961
Benjamin Peterson78821dd2009-01-25 17:15:10 +0000962 p = PyString_AS_STRING(v);
Christian Heimes44720832008-05-26 13:01:01 +0000963 while (*quote_prefix)
964 *p++ = *quote_prefix++;
965 *p++ = quote;
966
967 for (i = 0; i < length; i++) {
968 /* There's at least enough room for a hex escape
969 and a closing quote. */
Benjamin Peterson78821dd2009-01-25 17:15:10 +0000970 assert(newsize - (p - PyString_AS_STRING(v)) >= 5);
Christian Heimes44720832008-05-26 13:01:01 +0000971 c = self->ob_bytes[i];
972 if (c == '\'' || c == '\\')
973 *p++ = '\\', *p++ = c;
974 else if (c == '\t')
975 *p++ = '\\', *p++ = 't';
976 else if (c == '\n')
977 *p++ = '\\', *p++ = 'n';
978 else if (c == '\r')
979 *p++ = '\\', *p++ = 'r';
980 else if (c == 0)
981 *p++ = '\\', *p++ = 'x', *p++ = '0', *p++ = '0';
982 else if (c < ' ' || c >= 0x7f) {
983 *p++ = '\\';
984 *p++ = 'x';
985 *p++ = hexdigits[(c & 0xf0) >> 4];
986 *p++ = hexdigits[c & 0xf];
987 }
988 else
989 *p++ = c;
990 }
Benjamin Peterson78821dd2009-01-25 17:15:10 +0000991 assert(newsize - (p - PyString_AS_STRING(v)) >= 1);
Christian Heimes44720832008-05-26 13:01:01 +0000992 *p++ = quote;
993 while (*quote_postfix) {
994 *p++ = *quote_postfix++;
995 }
996 *p = '\0';
Kristján Valur Jónssonbe580f22014-04-25 09:51:21 +0000997 /* v is cleared on error */
998 (void)_PyString_Resize(&v, (p - PyString_AS_STRING(v)));
Christian Heimes44720832008-05-26 13:01:01 +0000999 return v;
1000 }
1001}
1002
1003static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001004bytearray_str(PyObject *op)
Christian Heimes44720832008-05-26 13:01:01 +00001005{
1006#if 0
1007 if (Py_BytesWarningFlag) {
1008 if (PyErr_WarnEx(PyExc_BytesWarning,
1009 "str() on a bytearray instance", 1))
1010 return NULL;
1011 }
Benjamin Petersond6720012009-04-18 15:31:34 +00001012 return bytearray_repr((PyByteArrayObject*)op);
Christian Heimes44720832008-05-26 13:01:01 +00001013#endif
1014 return PyBytes_FromStringAndSize(((PyByteArrayObject*)op)->ob_bytes, Py_SIZE(op));
1015}
1016
1017static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001018bytearray_richcompare(PyObject *self, PyObject *other, int op)
Christian Heimes44720832008-05-26 13:01:01 +00001019{
1020 Py_ssize_t self_size, other_size;
1021 Py_buffer self_bytes, other_bytes;
1022 PyObject *res;
1023 Py_ssize_t minsize;
Serhiy Storchaka5127ed72015-05-30 17:45:12 +03001024 int cmp, rc;
Christian Heimes44720832008-05-26 13:01:01 +00001025
1026 /* Bytes can be compared to anything that supports the (binary)
1027 buffer API. Except that a comparison with Unicode is always an
1028 error, even if the comparison is for equality. */
Benjamin Peterson78821dd2009-01-25 17:15:10 +00001029#ifdef Py_USING_UNICODE
Serhiy Storchaka5127ed72015-05-30 17:45:12 +03001030 rc = PyObject_IsInstance(self, (PyObject*)&PyUnicode_Type);
1031 if (!rc)
1032 rc = PyObject_IsInstance(other, (PyObject*)&PyUnicode_Type);
1033 if (rc < 0)
1034 return NULL;
1035 if (rc) {
Christian Heimes44720832008-05-26 13:01:01 +00001036 if (Py_BytesWarningFlag && op == Py_EQ) {
1037 if (PyErr_WarnEx(PyExc_BytesWarning,
Ezio Melottid2342082010-01-14 11:34:10 +00001038 "Comparison between bytearray and string", 1))
Christian Heimes44720832008-05-26 13:01:01 +00001039 return NULL;
1040 }
1041
1042 Py_INCREF(Py_NotImplemented);
1043 return Py_NotImplemented;
1044 }
Benjamin Peterson78821dd2009-01-25 17:15:10 +00001045#endif
Christian Heimes44720832008-05-26 13:01:01 +00001046
1047 self_size = _getbuffer(self, &self_bytes);
1048 if (self_size < 0) {
1049 PyErr_Clear();
1050 Py_INCREF(Py_NotImplemented);
1051 return Py_NotImplemented;
1052 }
1053
1054 other_size = _getbuffer(other, &other_bytes);
1055 if (other_size < 0) {
1056 PyErr_Clear();
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001057 PyBuffer_Release(&self_bytes);
Christian Heimes44720832008-05-26 13:01:01 +00001058 Py_INCREF(Py_NotImplemented);
1059 return Py_NotImplemented;
1060 }
1061
1062 if (self_size != other_size && (op == Py_EQ || op == Py_NE)) {
1063 /* Shortcut: if the lengths differ, the objects differ */
1064 cmp = (op == Py_NE);
1065 }
1066 else {
1067 minsize = self_size;
1068 if (other_size < minsize)
1069 minsize = other_size;
1070
1071 cmp = memcmp(self_bytes.buf, other_bytes.buf, minsize);
1072 /* In ISO C, memcmp() guarantees to use unsigned bytes! */
1073
1074 if (cmp == 0) {
1075 if (self_size < other_size)
1076 cmp = -1;
1077 else if (self_size > other_size)
1078 cmp = 1;
1079 }
1080
1081 switch (op) {
1082 case Py_LT: cmp = cmp < 0; break;
1083 case Py_LE: cmp = cmp <= 0; break;
1084 case Py_EQ: cmp = cmp == 0; break;
1085 case Py_NE: cmp = cmp != 0; break;
1086 case Py_GT: cmp = cmp > 0; break;
1087 case Py_GE: cmp = cmp >= 0; break;
1088 }
1089 }
1090
1091 res = cmp ? Py_True : Py_False;
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001092 PyBuffer_Release(&self_bytes);
1093 PyBuffer_Release(&other_bytes);
Christian Heimes44720832008-05-26 13:01:01 +00001094 Py_INCREF(res);
1095 return res;
1096}
1097
1098static void
Benjamin Petersond6720012009-04-18 15:31:34 +00001099bytearray_dealloc(PyByteArrayObject *self)
Christian Heimes44720832008-05-26 13:01:01 +00001100{
Benjamin Peterson9c1f7b22009-03-08 00:21:17 +00001101 if (self->ob_exports > 0) {
1102 PyErr_SetString(PyExc_SystemError,
Georg Brandl517cfdc2009-04-05 13:16:35 +00001103 "deallocated bytearray object has exported buffers");
Benjamin Peterson9c1f7b22009-03-08 00:21:17 +00001104 PyErr_Print();
1105 }
Christian Heimes44720832008-05-26 13:01:01 +00001106 if (self->ob_bytes != 0) {
1107 PyMem_Free(self->ob_bytes);
1108 }
1109 Py_TYPE(self)->tp_free((PyObject *)self);
1110}
1111
1112
1113/* -------------------------------------------------------------------- */
1114/* Methods */
1115
1116#define STRINGLIB_CHAR char
Christian Heimes44720832008-05-26 13:01:01 +00001117#define STRINGLIB_LEN PyByteArray_GET_SIZE
1118#define STRINGLIB_STR PyByteArray_AS_STRING
1119#define STRINGLIB_NEW PyByteArray_FromStringAndSize
Antoine Pitrou64672132010-01-13 07:55:48 +00001120#define STRINGLIB_ISSPACE Py_ISSPACE
1121#define STRINGLIB_ISLINEBREAK(x) ((x == '\n') || (x == '\r'))
Christian Heimes44720832008-05-26 13:01:01 +00001122#define STRINGLIB_CHECK_EXACT PyByteArray_CheckExact
1123#define STRINGLIB_MUTABLE 1
1124
1125#include "stringlib/fastsearch.h"
1126#include "stringlib/count.h"
1127#include "stringlib/find.h"
1128#include "stringlib/partition.h"
Antoine Pitrou64672132010-01-13 07:55:48 +00001129#include "stringlib/split.h"
Christian Heimes44720832008-05-26 13:01:01 +00001130#include "stringlib/ctype.h"
1131#include "stringlib/transmogrify.h"
1132
1133
1134/* The following Py_LOCAL_INLINE and Py_LOCAL functions
1135were copied from the old char* style string object. */
1136
Antoine Pitrou64672132010-01-13 07:55:48 +00001137/* helper macro to fixup start/end slice values */
1138#define ADJUST_INDICES(start, end, len) \
1139 if (end > len) \
1140 end = len; \
1141 else if (end < 0) { \
1142 end += len; \
1143 if (end < 0) \
1144 end = 0; \
1145 } \
1146 if (start < 0) { \
1147 start += len; \
1148 if (start < 0) \
1149 start = 0; \
1150 }
Christian Heimes44720832008-05-26 13:01:01 +00001151
1152Py_LOCAL_INLINE(Py_ssize_t)
Benjamin Petersond6720012009-04-18 15:31:34 +00001153bytearray_find_internal(PyByteArrayObject *self, PyObject *args, int dir)
Christian Heimes44720832008-05-26 13:01:01 +00001154{
1155 PyObject *subobj;
1156 Py_buffer subbuf;
1157 Py_ssize_t start=0, end=PY_SSIZE_T_MAX;
1158 Py_ssize_t res;
1159
Jesus Cea44e81682011-04-20 16:39:15 +02001160 if (!stringlib_parse_args_finds("find/rfind/index/rindex",
1161 args, &subobj, &start, &end))
Christian Heimes44720832008-05-26 13:01:01 +00001162 return -2;
1163 if (_getbuffer(subobj, &subbuf) < 0)
1164 return -2;
1165 if (dir > 0)
1166 res = stringlib_find_slice(
1167 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
1168 subbuf.buf, subbuf.len, start, end);
1169 else
1170 res = stringlib_rfind_slice(
1171 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
1172 subbuf.buf, subbuf.len, start, end);
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001173 PyBuffer_Release(&subbuf);
Christian Heimes44720832008-05-26 13:01:01 +00001174 return res;
1175}
1176
1177PyDoc_STRVAR(find__doc__,
1178"B.find(sub [,start [,end]]) -> int\n\
1179\n\
1180Return the lowest index in B where subsection sub is found,\n\
Senthil Kumaran5e3a19d2011-07-27 23:36:51 +08001181such that sub is contained within B[start,end]. Optional\n\
Christian Heimes44720832008-05-26 13:01:01 +00001182arguments start and end are interpreted as in slice notation.\n\
1183\n\
1184Return -1 on failure.");
1185
1186static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001187bytearray_find(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001188{
Benjamin Petersond6720012009-04-18 15:31:34 +00001189 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes44720832008-05-26 13:01:01 +00001190 if (result == -2)
1191 return NULL;
1192 return PyInt_FromSsize_t(result);
1193}
1194
1195PyDoc_STRVAR(count__doc__,
1196"B.count(sub [,start [,end]]) -> int\n\
1197\n\
1198Return the number of non-overlapping occurrences of subsection sub in\n\
1199bytes B[start:end]. Optional arguments start and end are interpreted\n\
1200as in slice notation.");
1201
1202static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001203bytearray_count(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001204{
1205 PyObject *sub_obj;
1206 const char *str = PyByteArray_AS_STRING(self);
1207 Py_ssize_t start = 0, end = PY_SSIZE_T_MAX;
1208 Py_buffer vsub;
1209 PyObject *count_obj;
1210
Jesus Cea44e81682011-04-20 16:39:15 +02001211 if (!stringlib_parse_args_finds("count", args, &sub_obj, &start, &end))
Christian Heimes44720832008-05-26 13:01:01 +00001212 return NULL;
1213
1214 if (_getbuffer(sub_obj, &vsub) < 0)
1215 return NULL;
1216
Antoine Pitrou64672132010-01-13 07:55:48 +00001217 ADJUST_INDICES(start, end, PyByteArray_GET_SIZE(self));
Christian Heimes44720832008-05-26 13:01:01 +00001218
1219 count_obj = PyInt_FromSsize_t(
Antoine Pitrou64672132010-01-13 07:55:48 +00001220 stringlib_count(str + start, end - start, vsub.buf, vsub.len, PY_SSIZE_T_MAX)
Christian Heimes44720832008-05-26 13:01:01 +00001221 );
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001222 PyBuffer_Release(&vsub);
Christian Heimes44720832008-05-26 13:01:01 +00001223 return count_obj;
1224}
1225
1226
1227PyDoc_STRVAR(index__doc__,
1228"B.index(sub [,start [,end]]) -> int\n\
1229\n\
1230Like B.find() but raise ValueError when the subsection is not found.");
1231
1232static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001233bytearray_index(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001234{
Benjamin Petersond6720012009-04-18 15:31:34 +00001235 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes44720832008-05-26 13:01:01 +00001236 if (result == -2)
1237 return NULL;
1238 if (result == -1) {
1239 PyErr_SetString(PyExc_ValueError,
1240 "subsection not found");
1241 return NULL;
1242 }
1243 return PyInt_FromSsize_t(result);
1244}
1245
1246
1247PyDoc_STRVAR(rfind__doc__,
1248"B.rfind(sub [,start [,end]]) -> int\n\
1249\n\
1250Return the highest index in B where subsection sub is found,\n\
Senthil Kumaran5e3a19d2011-07-27 23:36:51 +08001251such that sub is contained within B[start,end]. Optional\n\
Christian Heimes44720832008-05-26 13:01:01 +00001252arguments start and end are interpreted as in slice notation.\n\
1253\n\
1254Return -1 on failure.");
1255
1256static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001257bytearray_rfind(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001258{
Benjamin Petersond6720012009-04-18 15:31:34 +00001259 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes44720832008-05-26 13:01:01 +00001260 if (result == -2)
1261 return NULL;
1262 return PyInt_FromSsize_t(result);
1263}
1264
1265
1266PyDoc_STRVAR(rindex__doc__,
1267"B.rindex(sub [,start [,end]]) -> int\n\
1268\n\
1269Like B.rfind() but raise ValueError when the subsection is not found.");
1270
1271static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001272bytearray_rindex(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001273{
Benjamin Petersond6720012009-04-18 15:31:34 +00001274 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes44720832008-05-26 13:01:01 +00001275 if (result == -2)
1276 return NULL;
1277 if (result == -1) {
1278 PyErr_SetString(PyExc_ValueError,
1279 "subsection not found");
1280 return NULL;
1281 }
1282 return PyInt_FromSsize_t(result);
1283}
1284
1285
1286static int
Benjamin Petersond6720012009-04-18 15:31:34 +00001287bytearray_contains(PyObject *self, PyObject *arg)
Christian Heimes44720832008-05-26 13:01:01 +00001288{
1289 Py_ssize_t ival = PyNumber_AsSsize_t(arg, PyExc_ValueError);
1290 if (ival == -1 && PyErr_Occurred()) {
1291 Py_buffer varg;
1292 int pos;
1293 PyErr_Clear();
1294 if (_getbuffer(arg, &varg) < 0)
1295 return -1;
1296 pos = stringlib_find(PyByteArray_AS_STRING(self), Py_SIZE(self),
1297 varg.buf, varg.len, 0);
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001298 PyBuffer_Release(&varg);
Christian Heimes44720832008-05-26 13:01:01 +00001299 return pos >= 0;
1300 }
1301 if (ival < 0 || ival >= 256) {
1302 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
1303 return -1;
1304 }
1305
1306 return memchr(PyByteArray_AS_STRING(self), ival, Py_SIZE(self)) != NULL;
1307}
1308
1309
1310/* Matches the end (direction >= 0) or start (direction < 0) of self
1311 * against substr, using the start and end arguments. Returns
1312 * -1 on error, 0 if not found and 1 if found.
1313 */
1314Py_LOCAL(int)
Benjamin Petersond6720012009-04-18 15:31:34 +00001315_bytearray_tailmatch(PyByteArrayObject *self, PyObject *substr, Py_ssize_t start,
Christian Heimes44720832008-05-26 13:01:01 +00001316 Py_ssize_t end, int direction)
1317{
1318 Py_ssize_t len = PyByteArray_GET_SIZE(self);
1319 const char* str;
1320 Py_buffer vsubstr;
1321 int rv = 0;
1322
1323 str = PyByteArray_AS_STRING(self);
1324
1325 if (_getbuffer(substr, &vsubstr) < 0)
1326 return -1;
1327
Antoine Pitrou64672132010-01-13 07:55:48 +00001328 ADJUST_INDICES(start, end, len);
Christian Heimes44720832008-05-26 13:01:01 +00001329
1330 if (direction < 0) {
1331 /* startswith */
1332 if (start+vsubstr.len > len) {
1333 goto done;
1334 }
1335 } else {
1336 /* endswith */
1337 if (end-start < vsubstr.len || start > len) {
1338 goto done;
1339 }
1340
1341 if (end-vsubstr.len > start)
1342 start = end - vsubstr.len;
1343 }
1344 if (end-start >= vsubstr.len)
1345 rv = ! memcmp(str+start, vsubstr.buf, vsubstr.len);
1346
1347done:
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001348 PyBuffer_Release(&vsubstr);
Christian Heimes44720832008-05-26 13:01:01 +00001349 return rv;
1350}
1351
1352
1353PyDoc_STRVAR(startswith__doc__,
1354"B.startswith(prefix [,start [,end]]) -> bool\n\
1355\n\
1356Return True if B starts with the specified prefix, False otherwise.\n\
1357With optional start, test B beginning at that position.\n\
1358With optional end, stop comparing B at that position.\n\
1359prefix can also be a tuple of strings to try.");
1360
1361static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001362bytearray_startswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001363{
1364 Py_ssize_t start = 0;
1365 Py_ssize_t end = PY_SSIZE_T_MAX;
1366 PyObject *subobj;
1367 int result;
1368
Jesus Cea44e81682011-04-20 16:39:15 +02001369 if (!stringlib_parse_args_finds("startswith", args, &subobj, &start, &end))
Christian Heimes44720832008-05-26 13:01:01 +00001370 return NULL;
1371 if (PyTuple_Check(subobj)) {
1372 Py_ssize_t i;
1373 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Petersond6720012009-04-18 15:31:34 +00001374 result = _bytearray_tailmatch(self,
Christian Heimes44720832008-05-26 13:01:01 +00001375 PyTuple_GET_ITEM(subobj, i),
1376 start, end, -1);
1377 if (result == -1)
1378 return NULL;
1379 else if (result) {
1380 Py_RETURN_TRUE;
1381 }
1382 }
1383 Py_RETURN_FALSE;
1384 }
Benjamin Petersond6720012009-04-18 15:31:34 +00001385 result = _bytearray_tailmatch(self, subobj, start, end, -1);
Christian Heimes44720832008-05-26 13:01:01 +00001386 if (result == -1)
1387 return NULL;
1388 else
1389 return PyBool_FromLong(result);
1390}
1391
1392PyDoc_STRVAR(endswith__doc__,
1393"B.endswith(suffix [,start [,end]]) -> bool\n\
1394\n\
1395Return True if B ends with the specified suffix, False otherwise.\n\
1396With optional start, test B beginning at that position.\n\
1397With optional end, stop comparing B at that position.\n\
1398suffix can also be a tuple of strings to try.");
1399
1400static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001401bytearray_endswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001402{
1403 Py_ssize_t start = 0;
1404 Py_ssize_t end = PY_SSIZE_T_MAX;
1405 PyObject *subobj;
1406 int result;
1407
Jesus Cea44e81682011-04-20 16:39:15 +02001408 if (!stringlib_parse_args_finds("endswith", args, &subobj, &start, &end))
Christian Heimes44720832008-05-26 13:01:01 +00001409 return NULL;
1410 if (PyTuple_Check(subobj)) {
1411 Py_ssize_t i;
1412 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Petersond6720012009-04-18 15:31:34 +00001413 result = _bytearray_tailmatch(self,
Christian Heimes44720832008-05-26 13:01:01 +00001414 PyTuple_GET_ITEM(subobj, i),
1415 start, end, +1);
1416 if (result == -1)
1417 return NULL;
1418 else if (result) {
1419 Py_RETURN_TRUE;
1420 }
1421 }
1422 Py_RETURN_FALSE;
1423 }
Benjamin Petersond6720012009-04-18 15:31:34 +00001424 result = _bytearray_tailmatch(self, subobj, start, end, +1);
Christian Heimes44720832008-05-26 13:01:01 +00001425 if (result == -1)
1426 return NULL;
1427 else
1428 return PyBool_FromLong(result);
1429}
1430
1431
1432PyDoc_STRVAR(translate__doc__,
1433"B.translate(table[, deletechars]) -> bytearray\n\
1434\n\
1435Return a copy of B, where all characters occurring in the\n\
1436optional argument deletechars are removed, and the remaining\n\
1437characters have been mapped through the given translation\n\
1438table, which must be a bytes object of length 256.");
1439
1440static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00001441bytearray_translate(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00001442{
1443 register char *input, *output;
1444 register const char *table;
Benjamin Peterson46cc6d12008-11-19 21:49:09 +00001445 register Py_ssize_t i, c;
Christian Heimes44720832008-05-26 13:01:01 +00001446 PyObject *input_obj = (PyObject*)self;
1447 const char *output_start;
1448 Py_ssize_t inlen;
Georg Brandl6425a2f2008-12-28 11:54:53 +00001449 PyObject *result = NULL;
Christian Heimes44720832008-05-26 13:01:01 +00001450 int trans_table[256];
Georg Brandl6425a2f2008-12-28 11:54:53 +00001451 PyObject *tableobj = NULL, *delobj = NULL;
Christian Heimes44720832008-05-26 13:01:01 +00001452 Py_buffer vtable, vdel;
1453
1454 if (!PyArg_UnpackTuple(args, "translate", 1, 2,
1455 &tableobj, &delobj))
1456 return NULL;
1457
Georg Brandl6425a2f2008-12-28 11:54:53 +00001458 if (tableobj == Py_None) {
1459 table = NULL;
1460 tableobj = NULL;
1461 } else if (_getbuffer(tableobj, &vtable) < 0) {
Christian Heimes44720832008-05-26 13:01:01 +00001462 return NULL;
Georg Brandl6425a2f2008-12-28 11:54:53 +00001463 } else {
1464 if (vtable.len != 256) {
1465 PyErr_SetString(PyExc_ValueError,
1466 "translation table must be 256 characters long");
Georg Brandlec812ca2009-07-22 11:57:15 +00001467 PyBuffer_Release(&vtable);
1468 return NULL;
Georg Brandl6425a2f2008-12-28 11:54:53 +00001469 }
1470 table = (const char*)vtable.buf;
Christian Heimes44720832008-05-26 13:01:01 +00001471 }
1472
1473 if (delobj != NULL) {
1474 if (_getbuffer(delobj, &vdel) < 0) {
Georg Brandlec812ca2009-07-22 11:57:15 +00001475 if (tableobj != NULL)
1476 PyBuffer_Release(&vtable);
1477 return NULL;
Christian Heimes44720832008-05-26 13:01:01 +00001478 }
1479 }
1480 else {
1481 vdel.buf = NULL;
1482 vdel.len = 0;
1483 }
1484
Christian Heimes44720832008-05-26 13:01:01 +00001485 inlen = PyByteArray_GET_SIZE(input_obj);
1486 result = PyByteArray_FromStringAndSize((char *)NULL, inlen);
1487 if (result == NULL)
1488 goto done;
1489 output_start = output = PyByteArray_AsString(result);
1490 input = PyByteArray_AS_STRING(input_obj);
1491
Georg Brandl6425a2f2008-12-28 11:54:53 +00001492 if (vdel.len == 0 && table != NULL) {
Christian Heimes44720832008-05-26 13:01:01 +00001493 /* If no deletions are required, use faster code */
1494 for (i = inlen; --i >= 0; ) {
1495 c = Py_CHARMASK(*input++);
Benjamin Peterson46cc6d12008-11-19 21:49:09 +00001496 *output++ = table[c];
Christian Heimes44720832008-05-26 13:01:01 +00001497 }
Christian Heimes44720832008-05-26 13:01:01 +00001498 goto done;
1499 }
Georg Brandl6425a2f2008-12-28 11:54:53 +00001500
1501 if (table == NULL) {
1502 for (i = 0; i < 256; i++)
1503 trans_table[i] = Py_CHARMASK(i);
1504 } else {
1505 for (i = 0; i < 256; i++)
1506 trans_table[i] = Py_CHARMASK(table[i]);
1507 }
Christian Heimes44720832008-05-26 13:01:01 +00001508
1509 for (i = 0; i < vdel.len; i++)
1510 trans_table[(int) Py_CHARMASK( ((unsigned char*)vdel.buf)[i] )] = -1;
1511
1512 for (i = inlen; --i >= 0; ) {
1513 c = Py_CHARMASK(*input++);
1514 if (trans_table[c] != -1)
1515 if (Py_CHARMASK(*output++ = (char)trans_table[c]) == c)
1516 continue;
Christian Heimes44720832008-05-26 13:01:01 +00001517 }
1518 /* Fix the size of the resulting string */
1519 if (inlen > 0)
1520 PyByteArray_Resize(result, output - output_start);
1521
1522done:
Georg Brandl6425a2f2008-12-28 11:54:53 +00001523 if (tableobj != NULL)
1524 PyBuffer_Release(&vtable);
Christian Heimes44720832008-05-26 13:01:01 +00001525 if (delobj != NULL)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001526 PyBuffer_Release(&vdel);
Christian Heimes44720832008-05-26 13:01:01 +00001527 return result;
1528}
1529
1530
Christian Heimes44720832008-05-26 13:01:01 +00001531/* find and count characters and substrings */
1532
1533#define findchar(target, target_len, c) \
1534 ((char *)memchr((const void *)(target), c, target_len))
1535
Christian Heimes44720832008-05-26 13:01:01 +00001536
Benjamin Peterson46cc6d12008-11-19 21:49:09 +00001537/* Bytes ops must return a string, create a copy */
Christian Heimes44720832008-05-26 13:01:01 +00001538Py_LOCAL(PyByteArrayObject *)
1539return_self(PyByteArrayObject *self)
1540{
Christian Heimes44720832008-05-26 13:01:01 +00001541 return (PyByteArrayObject *)PyByteArray_FromStringAndSize(
1542 PyByteArray_AS_STRING(self),
1543 PyByteArray_GET_SIZE(self));
1544}
1545
1546Py_LOCAL_INLINE(Py_ssize_t)
1547countchar(const char *target, Py_ssize_t target_len, char c, Py_ssize_t maxcount)
1548{
1549 Py_ssize_t count=0;
1550 const char *start=target;
1551 const char *end=target+target_len;
1552
1553 while ( (start=findchar(start, end-start, c)) != NULL ) {
1554 count++;
1555 if (count >= maxcount)
1556 break;
1557 start += 1;
1558 }
1559 return count;
1560}
1561
Christian Heimes44720832008-05-26 13:01:01 +00001562
1563/* Algorithms for different cases of string replacement */
1564
1565/* len(self)>=1, from="", len(to)>=1, maxcount>=1 */
1566Py_LOCAL(PyByteArrayObject *)
1567replace_interleave(PyByteArrayObject *self,
1568 const char *to_s, Py_ssize_t to_len,
1569 Py_ssize_t maxcount)
1570{
1571 char *self_s, *result_s;
1572 Py_ssize_t self_len, result_len;
1573 Py_ssize_t count, i, product;
1574 PyByteArrayObject *result;
1575
1576 self_len = PyByteArray_GET_SIZE(self);
1577
1578 /* 1 at the end plus 1 after every character */
1579 count = self_len+1;
1580 if (maxcount < count)
1581 count = maxcount;
1582
1583 /* Check for overflow */
1584 /* result_len = count * to_len + self_len; */
1585 product = count * to_len;
1586 if (product / to_len != count) {
1587 PyErr_SetString(PyExc_OverflowError,
1588 "replace string is too long");
1589 return NULL;
1590 }
1591 result_len = product + self_len;
1592 if (result_len < 0) {
1593 PyErr_SetString(PyExc_OverflowError,
1594 "replace string is too long");
1595 return NULL;
1596 }
1597
1598 if (! (result = (PyByteArrayObject *)
1599 PyByteArray_FromStringAndSize(NULL, result_len)) )
1600 return NULL;
1601
1602 self_s = PyByteArray_AS_STRING(self);
1603 result_s = PyByteArray_AS_STRING(result);
1604
1605 /* TODO: special case single character, which doesn't need memcpy */
1606
1607 /* Lay the first one down (guaranteed this will occur) */
1608 Py_MEMCPY(result_s, to_s, to_len);
1609 result_s += to_len;
1610 count -= 1;
1611
1612 for (i=0; i<count; i++) {
1613 *result_s++ = *self_s++;
1614 Py_MEMCPY(result_s, to_s, to_len);
1615 result_s += to_len;
1616 }
1617
1618 /* Copy the rest of the original string */
1619 Py_MEMCPY(result_s, self_s, self_len-i);
1620
1621 return result;
1622}
1623
1624/* Special case for deleting a single character */
1625/* len(self)>=1, len(from)==1, to="", maxcount>=1 */
1626Py_LOCAL(PyByteArrayObject *)
1627replace_delete_single_character(PyByteArrayObject *self,
1628 char from_c, Py_ssize_t maxcount)
1629{
1630 char *self_s, *result_s;
1631 char *start, *next, *end;
1632 Py_ssize_t self_len, result_len;
1633 Py_ssize_t count;
1634 PyByteArrayObject *result;
1635
1636 self_len = PyByteArray_GET_SIZE(self);
1637 self_s = PyByteArray_AS_STRING(self);
1638
1639 count = countchar(self_s, self_len, from_c, maxcount);
1640 if (count == 0) {
1641 return return_self(self);
1642 }
1643
1644 result_len = self_len - count; /* from_len == 1 */
1645 assert(result_len>=0);
1646
1647 if ( (result = (PyByteArrayObject *)
1648 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1649 return NULL;
1650 result_s = PyByteArray_AS_STRING(result);
1651
1652 start = self_s;
1653 end = self_s + self_len;
1654 while (count-- > 0) {
1655 next = findchar(start, end-start, from_c);
1656 if (next == NULL)
1657 break;
1658 Py_MEMCPY(result_s, start, next-start);
1659 result_s += (next-start);
1660 start = next+1;
1661 }
1662 Py_MEMCPY(result_s, start, end-start);
1663
1664 return result;
1665}
1666
1667/* len(self)>=1, len(from)>=2, to="", maxcount>=1 */
1668
1669Py_LOCAL(PyByteArrayObject *)
1670replace_delete_substring(PyByteArrayObject *self,
1671 const char *from_s, Py_ssize_t from_len,
1672 Py_ssize_t maxcount)
1673{
1674 char *self_s, *result_s;
1675 char *start, *next, *end;
1676 Py_ssize_t self_len, result_len;
1677 Py_ssize_t count, offset;
1678 PyByteArrayObject *result;
1679
1680 self_len = PyByteArray_GET_SIZE(self);
1681 self_s = PyByteArray_AS_STRING(self);
1682
Antoine Pitrou64672132010-01-13 07:55:48 +00001683 count = stringlib_count(self_s, self_len,
1684 from_s, from_len,
1685 maxcount);
Christian Heimes44720832008-05-26 13:01:01 +00001686
1687 if (count == 0) {
1688 /* no matches */
1689 return return_self(self);
1690 }
1691
1692 result_len = self_len - (count * from_len);
1693 assert (result_len>=0);
1694
1695 if ( (result = (PyByteArrayObject *)
1696 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL )
1697 return NULL;
1698
1699 result_s = PyByteArray_AS_STRING(result);
1700
1701 start = self_s;
1702 end = self_s + self_len;
1703 while (count-- > 0) {
Antoine Pitrou64672132010-01-13 07:55:48 +00001704 offset = stringlib_find(start, end-start,
1705 from_s, from_len,
1706 0);
Christian Heimes44720832008-05-26 13:01:01 +00001707 if (offset == -1)
1708 break;
1709 next = start + offset;
1710
1711 Py_MEMCPY(result_s, start, next-start);
1712
1713 result_s += (next-start);
1714 start = next+from_len;
1715 }
1716 Py_MEMCPY(result_s, start, end-start);
1717 return result;
1718}
1719
1720/* len(self)>=1, len(from)==len(to)==1, maxcount>=1 */
1721Py_LOCAL(PyByteArrayObject *)
1722replace_single_character_in_place(PyByteArrayObject *self,
1723 char from_c, char to_c,
1724 Py_ssize_t maxcount)
1725{
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001726 char *self_s, *result_s, *start, *end, *next;
1727 Py_ssize_t self_len;
1728 PyByteArrayObject *result;
Christian Heimes44720832008-05-26 13:01:01 +00001729
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001730 /* The result string will be the same size */
1731 self_s = PyByteArray_AS_STRING(self);
1732 self_len = PyByteArray_GET_SIZE(self);
Christian Heimes44720832008-05-26 13:01:01 +00001733
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001734 next = findchar(self_s, self_len, from_c);
Christian Heimes44720832008-05-26 13:01:01 +00001735
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001736 if (next == NULL) {
1737 /* No matches; return the original bytes */
1738 return return_self(self);
1739 }
Christian Heimes44720832008-05-26 13:01:01 +00001740
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001741 /* Need to make a new bytes */
1742 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1743 if (result == NULL)
1744 return NULL;
1745 result_s = PyByteArray_AS_STRING(result);
1746 Py_MEMCPY(result_s, self_s, self_len);
Christian Heimes44720832008-05-26 13:01:01 +00001747
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001748 /* change everything in-place, starting with this one */
1749 start = result_s + (next-self_s);
1750 *start = to_c;
1751 start++;
1752 end = result_s + self_len;
Christian Heimes44720832008-05-26 13:01:01 +00001753
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001754 while (--maxcount > 0) {
1755 next = findchar(start, end-start, from_c);
1756 if (next == NULL)
1757 break;
1758 *next = to_c;
1759 start = next+1;
1760 }
Christian Heimes44720832008-05-26 13:01:01 +00001761
Antoine Pitrou619f16e2010-06-09 16:24:00 +00001762 return result;
Christian Heimes44720832008-05-26 13:01:01 +00001763}
1764
1765/* len(self)>=1, len(from)==len(to)>=2, maxcount>=1 */
1766Py_LOCAL(PyByteArrayObject *)
1767replace_substring_in_place(PyByteArrayObject *self,
1768 const char *from_s, Py_ssize_t from_len,
1769 const char *to_s, Py_ssize_t to_len,
1770 Py_ssize_t maxcount)
1771{
1772 char *result_s, *start, *end;
1773 char *self_s;
1774 Py_ssize_t self_len, offset;
1775 PyByteArrayObject *result;
1776
1777 /* The result bytes will be the same size */
1778
1779 self_s = PyByteArray_AS_STRING(self);
1780 self_len = PyByteArray_GET_SIZE(self);
1781
Antoine Pitrou64672132010-01-13 07:55:48 +00001782 offset = stringlib_find(self_s, self_len,
1783 from_s, from_len,
1784 0);
Christian Heimes44720832008-05-26 13:01:01 +00001785 if (offset == -1) {
1786 /* No matches; return the original bytes */
1787 return return_self(self);
1788 }
1789
1790 /* Need to make a new bytes */
1791 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1792 if (result == NULL)
1793 return NULL;
1794 result_s = PyByteArray_AS_STRING(result);
1795 Py_MEMCPY(result_s, self_s, self_len);
1796
1797 /* change everything in-place, starting with this one */
1798 start = result_s + offset;
1799 Py_MEMCPY(start, to_s, from_len);
1800 start += from_len;
1801 end = result_s + self_len;
1802
1803 while ( --maxcount > 0) {
Antoine Pitrou64672132010-01-13 07:55:48 +00001804 offset = stringlib_find(start, end-start,
1805 from_s, from_len,
1806 0);
Christian Heimes44720832008-05-26 13:01:01 +00001807 if (offset==-1)
1808 break;
1809 Py_MEMCPY(start+offset, to_s, from_len);
1810 start += offset+from_len;
1811 }
1812
1813 return result;
1814}
1815
1816/* len(self)>=1, len(from)==1, len(to)>=2, maxcount>=1 */
1817Py_LOCAL(PyByteArrayObject *)
1818replace_single_character(PyByteArrayObject *self,
1819 char from_c,
1820 const char *to_s, Py_ssize_t to_len,
1821 Py_ssize_t maxcount)
1822{
1823 char *self_s, *result_s;
1824 char *start, *next, *end;
1825 Py_ssize_t self_len, result_len;
1826 Py_ssize_t count, product;
1827 PyByteArrayObject *result;
1828
1829 self_s = PyByteArray_AS_STRING(self);
1830 self_len = PyByteArray_GET_SIZE(self);
1831
1832 count = countchar(self_s, self_len, from_c, maxcount);
1833 if (count == 0) {
1834 /* no matches, return unchanged */
1835 return return_self(self);
1836 }
1837
1838 /* use the difference between current and new, hence the "-1" */
1839 /* result_len = self_len + count * (to_len-1) */
1840 product = count * (to_len-1);
1841 if (product / (to_len-1) != count) {
1842 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1843 return NULL;
1844 }
1845 result_len = self_len + product;
1846 if (result_len < 0) {
1847 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1848 return NULL;
1849 }
1850
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) {
1859 next = findchar(start, end-start, from_c);
1860 if (next == NULL)
1861 break;
1862
1863 if (next == start) {
1864 /* replace with the 'to' */
1865 Py_MEMCPY(result_s, to_s, to_len);
1866 result_s += to_len;
1867 start += 1;
1868 } else {
1869 /* copy the unchanged old then the 'to' */
1870 Py_MEMCPY(result_s, start, next-start);
1871 result_s += (next-start);
1872 Py_MEMCPY(result_s, to_s, to_len);
1873 result_s += to_len;
1874 start = next+1;
1875 }
1876 }
1877 /* Copy the remainder of the remaining bytes */
1878 Py_MEMCPY(result_s, start, end-start);
1879
1880 return result;
1881}
1882
1883/* len(self)>=1, len(from)>=2, len(to)>=2, maxcount>=1 */
1884Py_LOCAL(PyByteArrayObject *)
1885replace_substring(PyByteArrayObject *self,
1886 const char *from_s, Py_ssize_t from_len,
1887 const char *to_s, Py_ssize_t to_len,
1888 Py_ssize_t maxcount)
1889{
1890 char *self_s, *result_s;
1891 char *start, *next, *end;
1892 Py_ssize_t self_len, result_len;
1893 Py_ssize_t count, offset, product;
1894 PyByteArrayObject *result;
1895
1896 self_s = PyByteArray_AS_STRING(self);
1897 self_len = PyByteArray_GET_SIZE(self);
1898
Antoine Pitrou64672132010-01-13 07:55:48 +00001899 count = stringlib_count(self_s, self_len,
1900 from_s, from_len,
1901 maxcount);
1902
Christian Heimes44720832008-05-26 13:01:01 +00001903 if (count == 0) {
1904 /* no matches, return unchanged */
1905 return return_self(self);
1906 }
1907
1908 /* Check for overflow */
1909 /* result_len = self_len + count * (to_len-from_len) */
1910 product = count * (to_len-from_len);
1911 if (product / (to_len-from_len) != count) {
1912 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1913 return NULL;
1914 }
1915 result_len = self_len + product;
1916 if (result_len < 0) {
1917 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1918 return NULL;
1919 }
1920
1921 if ( (result = (PyByteArrayObject *)
1922 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1923 return NULL;
1924 result_s = PyByteArray_AS_STRING(result);
1925
1926 start = self_s;
1927 end = self_s + self_len;
1928 while (count-- > 0) {
Antoine Pitrou64672132010-01-13 07:55:48 +00001929 offset = stringlib_find(start, end-start,
1930 from_s, from_len,
1931 0);
Christian Heimes44720832008-05-26 13:01:01 +00001932 if (offset == -1)
1933 break;
1934 next = start+offset;
1935 if (next == start) {
1936 /* replace with the 'to' */
1937 Py_MEMCPY(result_s, to_s, to_len);
1938 result_s += to_len;
1939 start += from_len;
1940 } else {
1941 /* copy the unchanged old then the 'to' */
1942 Py_MEMCPY(result_s, start, next-start);
1943 result_s += (next-start);
1944 Py_MEMCPY(result_s, to_s, to_len);
1945 result_s += to_len;
1946 start = next+from_len;
1947 }
1948 }
1949 /* Copy the remainder of the remaining bytes */
1950 Py_MEMCPY(result_s, start, end-start);
1951
1952 return result;
1953}
1954
1955
1956Py_LOCAL(PyByteArrayObject *)
1957replace(PyByteArrayObject *self,
1958 const char *from_s, Py_ssize_t from_len,
1959 const char *to_s, Py_ssize_t to_len,
1960 Py_ssize_t maxcount)
1961{
1962 if (maxcount < 0) {
1963 maxcount = PY_SSIZE_T_MAX;
1964 } else if (maxcount == 0 || PyByteArray_GET_SIZE(self) == 0) {
1965 /* nothing to do; return the original bytes */
1966 return return_self(self);
1967 }
1968
1969 if (maxcount == 0 ||
1970 (from_len == 0 && to_len == 0)) {
1971 /* nothing to do; return the original bytes */
1972 return return_self(self);
1973 }
1974
1975 /* Handle zero-length special cases */
1976
1977 if (from_len == 0) {
1978 /* insert the 'to' bytes everywhere. */
1979 /* >>> "Python".replace("", ".") */
1980 /* '.P.y.t.h.o.n.' */
1981 return replace_interleave(self, to_s, to_len, maxcount);
1982 }
1983
1984 /* Except for "".replace("", "A") == "A" there is no way beyond this */
1985 /* point for an empty self bytes to generate a non-empty bytes */
1986 /* Special case so the remaining code always gets a non-empty bytes */
1987 if (PyByteArray_GET_SIZE(self) == 0) {
1988 return return_self(self);
1989 }
1990
1991 if (to_len == 0) {
1992 /* delete all occurances of 'from' bytes */
1993 if (from_len == 1) {
1994 return replace_delete_single_character(
1995 self, from_s[0], maxcount);
1996 } else {
1997 return replace_delete_substring(self, from_s, from_len, maxcount);
1998 }
1999 }
2000
2001 /* Handle special case where both bytes have the same length */
2002
2003 if (from_len == to_len) {
2004 if (from_len == 1) {
2005 return replace_single_character_in_place(
2006 self,
2007 from_s[0],
2008 to_s[0],
2009 maxcount);
2010 } else {
2011 return replace_substring_in_place(
2012 self, from_s, from_len, to_s, to_len, maxcount);
2013 }
2014 }
2015
2016 /* Otherwise use the more generic algorithms */
2017 if (from_len == 1) {
2018 return replace_single_character(self, from_s[0],
2019 to_s, to_len, maxcount);
2020 } else {
2021 /* len('from')>=2, len('to')>=1 */
2022 return replace_substring(self, from_s, from_len, to_s, to_len, maxcount);
2023 }
2024}
2025
2026
2027PyDoc_STRVAR(replace__doc__,
2028"B.replace(old, new[, count]) -> bytes\n\
2029\n\
2030Return a copy of B with all occurrences of subsection\n\
2031old replaced by new. If the optional argument count is\n\
2032given, only the first count occurrences are replaced.");
2033
2034static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002035bytearray_replace(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002036{
2037 Py_ssize_t count = -1;
2038 PyObject *from, *to, *res;
2039 Py_buffer vfrom, vto;
2040
2041 if (!PyArg_ParseTuple(args, "OO|n:replace", &from, &to, &count))
2042 return NULL;
2043
2044 if (_getbuffer(from, &vfrom) < 0)
2045 return NULL;
2046 if (_getbuffer(to, &vto) < 0) {
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00002047 PyBuffer_Release(&vfrom);
Christian Heimes44720832008-05-26 13:01:01 +00002048 return NULL;
2049 }
2050
2051 res = (PyObject *)replace((PyByteArrayObject *) self,
2052 vfrom.buf, vfrom.len,
2053 vto.buf, vto.len, count);
2054
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00002055 PyBuffer_Release(&vfrom);
2056 PyBuffer_Release(&vto);
Christian Heimes44720832008-05-26 13:01:01 +00002057 return res;
2058}
2059
Christian Heimes44720832008-05-26 13:01:01 +00002060PyDoc_STRVAR(split__doc__,
2061"B.split([sep[, maxsplit]]) -> list of bytearray\n\
2062\n\
2063Return a list of the sections in B, using sep as the delimiter.\n\
2064If sep is not given, B is split on ASCII whitespace characters\n\
2065(space, tab, return, newline, formfeed, vertical tab).\n\
2066If maxsplit is given, at most maxsplit splits are done.");
2067
2068static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002069bytearray_split(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002070{
Antoine Pitrou64672132010-01-13 07:55:48 +00002071 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2072 Py_ssize_t maxsplit = -1;
Christian Heimes44720832008-05-26 13:01:01 +00002073 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrou64672132010-01-13 07:55:48 +00002074 PyObject *list, *subobj = Py_None;
Christian Heimes44720832008-05-26 13:01:01 +00002075 Py_buffer vsub;
Christian Heimes44720832008-05-26 13:01:01 +00002076
2077 if (!PyArg_ParseTuple(args, "|On:split", &subobj, &maxsplit))
2078 return NULL;
2079 if (maxsplit < 0)
2080 maxsplit = PY_SSIZE_T_MAX;
2081
2082 if (subobj == Py_None)
Antoine Pitrou64672132010-01-13 07:55:48 +00002083 return stringlib_split_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes44720832008-05-26 13:01:01 +00002084
2085 if (_getbuffer(subobj, &vsub) < 0)
2086 return NULL;
2087 sub = vsub.buf;
2088 n = vsub.len;
2089
Antoine Pitrou64672132010-01-13 07:55:48 +00002090 list = stringlib_split(
2091 (PyObject*) self, s, len, sub, n, maxsplit
2092 );
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00002093 PyBuffer_Release(&vsub);
Christian Heimes44720832008-05-26 13:01:01 +00002094 return list;
Christian Heimes44720832008-05-26 13:01:01 +00002095}
2096
2097PyDoc_STRVAR(partition__doc__,
2098"B.partition(sep) -> (head, sep, tail)\n\
2099\n\
2100Searches for the separator sep in B, and returns the part before it,\n\
2101the separator itself, and the part after it. If the separator is not\n\
2102found, returns B and two empty bytearray objects.");
2103
2104static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002105bytearray_partition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes44720832008-05-26 13:01:01 +00002106{
2107 PyObject *bytesep, *result;
2108
2109 bytesep = PyByteArray_FromObject(sep_obj);
2110 if (! bytesep)
2111 return NULL;
2112
2113 result = stringlib_partition(
2114 (PyObject*) self,
2115 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2116 bytesep,
2117 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2118 );
2119
2120 Py_DECREF(bytesep);
Antoine Pitrou64672132010-01-13 07:55:48 +00002121 return result;
Christian Heimes44720832008-05-26 13:01:01 +00002122}
2123
2124PyDoc_STRVAR(rpartition__doc__,
Ezio Melotti1fafaab2010-01-25 11:24:37 +00002125"B.rpartition(sep) -> (head, sep, tail)\n\
Christian Heimes44720832008-05-26 13:01:01 +00002126\n\
2127Searches for the separator sep in B, starting at the end of B,\n\
2128and returns the part before it, the separator itself, and the\n\
2129part after it. If the separator is not found, returns two empty\n\
2130bytearray objects and B.");
2131
2132static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002133bytearray_rpartition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes44720832008-05-26 13:01:01 +00002134{
2135 PyObject *bytesep, *result;
2136
2137 bytesep = PyByteArray_FromObject(sep_obj);
2138 if (! bytesep)
2139 return NULL;
2140
2141 result = stringlib_rpartition(
2142 (PyObject*) self,
2143 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2144 bytesep,
2145 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2146 );
2147
2148 Py_DECREF(bytesep);
Antoine Pitrou64672132010-01-13 07:55:48 +00002149 return result;
Christian Heimes44720832008-05-26 13:01:01 +00002150}
2151
2152PyDoc_STRVAR(rsplit__doc__,
2153"B.rsplit(sep[, maxsplit]) -> list of bytearray\n\
2154\n\
2155Return a list of the sections in B, using sep as the delimiter,\n\
2156starting at the end of B and working to the front.\n\
2157If sep is not given, B is split on ASCII whitespace characters\n\
2158(space, tab, return, newline, formfeed, vertical tab).\n\
2159If maxsplit is given, at most maxsplit splits are done.");
2160
2161static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002162bytearray_rsplit(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002163{
Antoine Pitrou64672132010-01-13 07:55:48 +00002164 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2165 Py_ssize_t maxsplit = -1;
Christian Heimes44720832008-05-26 13:01:01 +00002166 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrou64672132010-01-13 07:55:48 +00002167 PyObject *list, *subobj = Py_None;
Christian Heimes44720832008-05-26 13:01:01 +00002168 Py_buffer vsub;
2169
2170 if (!PyArg_ParseTuple(args, "|On:rsplit", &subobj, &maxsplit))
2171 return NULL;
2172 if (maxsplit < 0)
2173 maxsplit = PY_SSIZE_T_MAX;
2174
2175 if (subobj == Py_None)
Antoine Pitrou64672132010-01-13 07:55:48 +00002176 return stringlib_rsplit_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes44720832008-05-26 13:01:01 +00002177
2178 if (_getbuffer(subobj, &vsub) < 0)
2179 return NULL;
2180 sub = vsub.buf;
2181 n = vsub.len;
2182
Antoine Pitrou64672132010-01-13 07:55:48 +00002183 list = stringlib_rsplit(
2184 (PyObject*) self, s, len, sub, n, maxsplit
2185 );
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00002186 PyBuffer_Release(&vsub);
Christian Heimes44720832008-05-26 13:01:01 +00002187 return list;
Christian Heimes44720832008-05-26 13:01:01 +00002188}
2189
2190PyDoc_STRVAR(reverse__doc__,
2191"B.reverse() -> None\n\
2192\n\
2193Reverse the order of the values in B in place.");
2194static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002195bytearray_reverse(PyByteArrayObject *self, PyObject *unused)
Christian Heimes44720832008-05-26 13:01:01 +00002196{
2197 char swap, *head, *tail;
2198 Py_ssize_t i, j, n = Py_SIZE(self);
2199
2200 j = n / 2;
2201 head = self->ob_bytes;
2202 tail = head + n - 1;
2203 for (i = 0; i < j; i++) {
2204 swap = *head;
2205 *head++ = *tail;
2206 *tail-- = swap;
2207 }
2208
2209 Py_RETURN_NONE;
2210}
2211
2212PyDoc_STRVAR(insert__doc__,
2213"B.insert(index, int) -> None\n\
2214\n\
2215Insert a single item into the bytearray before the given index.");
2216static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002217bytearray_insert(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002218{
Georg Brandl3e483f62008-07-16 22:57:41 +00002219 PyObject *value;
2220 int ival;
Christian Heimes44720832008-05-26 13:01:01 +00002221 Py_ssize_t where, n = Py_SIZE(self);
2222
Georg Brandl3e483f62008-07-16 22:57:41 +00002223 if (!PyArg_ParseTuple(args, "nO:insert", &where, &value))
Christian Heimes44720832008-05-26 13:01:01 +00002224 return NULL;
2225
2226 if (n == PY_SSIZE_T_MAX) {
2227 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson135a7cf2009-09-06 10:32:21 +00002228 "cannot add more objects to bytearray");
Christian Heimes44720832008-05-26 13:01:01 +00002229 return NULL;
2230 }
Georg Brandl3e483f62008-07-16 22:57:41 +00002231 if (!_getbytevalue(value, &ival))
Christian Heimes44720832008-05-26 13:01:01 +00002232 return NULL;
Christian Heimes44720832008-05-26 13:01:01 +00002233 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2234 return NULL;
2235
2236 if (where < 0) {
2237 where += n;
2238 if (where < 0)
2239 where = 0;
2240 }
2241 if (where > n)
2242 where = n;
2243 memmove(self->ob_bytes + where + 1, self->ob_bytes + where, n - where);
Georg Brandl3e483f62008-07-16 22:57:41 +00002244 self->ob_bytes[where] = ival;
Christian Heimes44720832008-05-26 13:01:01 +00002245
2246 Py_RETURN_NONE;
2247}
2248
2249PyDoc_STRVAR(append__doc__,
2250"B.append(int) -> None\n\
2251\n\
2252Append a single item to the end of B.");
2253static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002254bytearray_append(PyByteArrayObject *self, PyObject *arg)
Christian Heimes44720832008-05-26 13:01:01 +00002255{
2256 int value;
2257 Py_ssize_t n = Py_SIZE(self);
2258
2259 if (! _getbytevalue(arg, &value))
2260 return NULL;
2261 if (n == PY_SSIZE_T_MAX) {
2262 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson135a7cf2009-09-06 10:32:21 +00002263 "cannot add more objects to bytearray");
Christian Heimes44720832008-05-26 13:01:01 +00002264 return NULL;
2265 }
2266 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2267 return NULL;
2268
2269 self->ob_bytes[n] = value;
2270
2271 Py_RETURN_NONE;
2272}
2273
2274PyDoc_STRVAR(extend__doc__,
2275"B.extend(iterable int) -> None\n\
2276\n\
2277Append all the elements from the iterator or sequence to the\n\
2278end of B.");
2279static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002280bytearray_extend(PyByteArrayObject *self, PyObject *arg)
Christian Heimes44720832008-05-26 13:01:01 +00002281{
Benjamin Petersond6720012009-04-18 15:31:34 +00002282 PyObject *it, *item, *bytearray_obj;
Christian Heimes44720832008-05-26 13:01:01 +00002283 Py_ssize_t buf_size = 0, len = 0;
2284 int value;
2285 char *buf;
2286
Benjamin Petersond6720012009-04-18 15:31:34 +00002287 /* bytearray_setslice code only accepts something supporting PEP 3118. */
Christian Heimes44720832008-05-26 13:01:01 +00002288 if (PyObject_CheckBuffer(arg)) {
Benjamin Petersond6720012009-04-18 15:31:34 +00002289 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), arg) == -1)
Christian Heimes44720832008-05-26 13:01:01 +00002290 return NULL;
2291
2292 Py_RETURN_NONE;
2293 }
2294
2295 it = PyObject_GetIter(arg);
2296 if (it == NULL)
2297 return NULL;
2298
Ezio Melotti24b07bc2011-03-15 18:55:01 +02002299 /* Try to determine the length of the argument. 32 is arbitrary. */
Christian Heimes44720832008-05-26 13:01:01 +00002300 buf_size = _PyObject_LengthHint(arg, 32);
Georg Brandl517cfdc2009-04-05 13:16:35 +00002301 if (buf_size == -1) {
2302 Py_DECREF(it);
2303 return NULL;
2304 }
Christian Heimes44720832008-05-26 13:01:01 +00002305
Benjamin Petersond6720012009-04-18 15:31:34 +00002306 bytearray_obj = PyByteArray_FromStringAndSize(NULL, buf_size);
Antoine Pitroufe941772012-04-01 16:05:46 +02002307 if (bytearray_obj == NULL) {
2308 Py_DECREF(it);
Christian Heimes44720832008-05-26 13:01:01 +00002309 return NULL;
Antoine Pitroufe941772012-04-01 16:05:46 +02002310 }
Benjamin Petersond6720012009-04-18 15:31:34 +00002311 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes44720832008-05-26 13:01:01 +00002312
2313 while ((item = PyIter_Next(it)) != NULL) {
2314 if (! _getbytevalue(item, &value)) {
2315 Py_DECREF(item);
2316 Py_DECREF(it);
Benjamin Petersond6720012009-04-18 15:31:34 +00002317 Py_DECREF(bytearray_obj);
Christian Heimes44720832008-05-26 13:01:01 +00002318 return NULL;
2319 }
2320 buf[len++] = value;
2321 Py_DECREF(item);
2322
2323 if (len >= buf_size) {
2324 buf_size = len + (len >> 1) + 1;
Benjamin Petersond6720012009-04-18 15:31:34 +00002325 if (PyByteArray_Resize((PyObject *)bytearray_obj, buf_size) < 0) {
Christian Heimes44720832008-05-26 13:01:01 +00002326 Py_DECREF(it);
Benjamin Petersond6720012009-04-18 15:31:34 +00002327 Py_DECREF(bytearray_obj);
Christian Heimes44720832008-05-26 13:01:01 +00002328 return NULL;
2329 }
2330 /* Recompute the `buf' pointer, since the resizing operation may
2331 have invalidated it. */
Benjamin Petersond6720012009-04-18 15:31:34 +00002332 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes44720832008-05-26 13:01:01 +00002333 }
2334 }
2335 Py_DECREF(it);
2336
2337 /* Resize down to exact size. */
Benjamin Petersond6720012009-04-18 15:31:34 +00002338 if (PyByteArray_Resize((PyObject *)bytearray_obj, len) < 0) {
2339 Py_DECREF(bytearray_obj);
Christian Heimes44720832008-05-26 13:01:01 +00002340 return NULL;
2341 }
2342
Antoine Pitroufe941772012-04-01 16:05:46 +02002343 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), bytearray_obj) == -1) {
2344 Py_DECREF(bytearray_obj);
Christian Heimes44720832008-05-26 13:01:01 +00002345 return NULL;
Antoine Pitroufe941772012-04-01 16:05:46 +02002346 }
Benjamin Petersond6720012009-04-18 15:31:34 +00002347 Py_DECREF(bytearray_obj);
Christian Heimes44720832008-05-26 13:01:01 +00002348
2349 Py_RETURN_NONE;
2350}
2351
2352PyDoc_STRVAR(pop__doc__,
2353"B.pop([index]) -> int\n\
2354\n\
2355Remove and return a single item from B. If no index\n\
Andrew M. Kuchlingd8972642008-06-21 13:29:12 +00002356argument is given, will pop the last value.");
Christian Heimes44720832008-05-26 13:01:01 +00002357static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002358bytearray_pop(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002359{
2360 int value;
2361 Py_ssize_t where = -1, n = Py_SIZE(self);
2362
2363 if (!PyArg_ParseTuple(args, "|n:pop", &where))
2364 return NULL;
2365
2366 if (n == 0) {
Eli Bendersky680e6eb2011-03-04 06:14:56 +00002367 PyErr_SetString(PyExc_IndexError,
2368 "pop from empty bytearray");
Christian Heimes44720832008-05-26 13:01:01 +00002369 return NULL;
2370 }
2371 if (where < 0)
2372 where += Py_SIZE(self);
2373 if (where < 0 || where >= Py_SIZE(self)) {
2374 PyErr_SetString(PyExc_IndexError, "pop index out of range");
2375 return NULL;
2376 }
Antoine Pitrouae5bece2008-12-06 21:29:24 +00002377 if (!_canresize(self))
2378 return NULL;
Christian Heimes44720832008-05-26 13:01:01 +00002379
2380 value = self->ob_bytes[where];
2381 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2382 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2383 return NULL;
2384
Mark Dickinsonc8a7c7c2009-09-06 10:03:31 +00002385 return PyInt_FromLong((unsigned char)value);
Christian Heimes44720832008-05-26 13:01:01 +00002386}
2387
2388PyDoc_STRVAR(remove__doc__,
2389"B.remove(int) -> None\n\
2390\n\
2391Remove the first occurance of a value in B.");
2392static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002393bytearray_remove(PyByteArrayObject *self, PyObject *arg)
Christian Heimes44720832008-05-26 13:01:01 +00002394{
2395 int value;
2396 Py_ssize_t where, n = Py_SIZE(self);
2397
2398 if (! _getbytevalue(arg, &value))
2399 return NULL;
2400
2401 for (where = 0; where < n; where++) {
2402 if (self->ob_bytes[where] == value)
2403 break;
2404 }
2405 if (where == n) {
Mark Dickinson135a7cf2009-09-06 10:32:21 +00002406 PyErr_SetString(PyExc_ValueError, "value not found in bytearray");
Christian Heimes44720832008-05-26 13:01:01 +00002407 return NULL;
2408 }
Antoine Pitrouae5bece2008-12-06 21:29:24 +00002409 if (!_canresize(self))
2410 return NULL;
Christian Heimes44720832008-05-26 13:01:01 +00002411
2412 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2413 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2414 return NULL;
2415
2416 Py_RETURN_NONE;
2417}
2418
2419/* XXX These two helpers could be optimized if argsize == 1 */
2420
2421static Py_ssize_t
2422lstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2423 void *argptr, Py_ssize_t argsize)
2424{
2425 Py_ssize_t i = 0;
2426 while (i < mysize && memchr(argptr, myptr[i], argsize))
2427 i++;
2428 return i;
2429}
2430
2431static Py_ssize_t
2432rstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2433 void *argptr, Py_ssize_t argsize)
2434{
2435 Py_ssize_t i = mysize - 1;
2436 while (i >= 0 && memchr(argptr, myptr[i], argsize))
2437 i--;
2438 return i + 1;
2439}
2440
2441PyDoc_STRVAR(strip__doc__,
2442"B.strip([bytes]) -> bytearray\n\
2443\n\
2444Strip leading and trailing bytes contained in the argument.\n\
2445If the argument is omitted, strip ASCII whitespace.");
2446static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002447bytearray_strip(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002448{
2449 Py_ssize_t left, right, mysize, argsize;
2450 void *myptr, *argptr;
2451 PyObject *arg = Py_None;
2452 Py_buffer varg;
2453 if (!PyArg_ParseTuple(args, "|O:strip", &arg))
2454 return NULL;
2455 if (arg == Py_None) {
2456 argptr = "\t\n\r\f\v ";
2457 argsize = 6;
2458 }
2459 else {
2460 if (_getbuffer(arg, &varg) < 0)
2461 return NULL;
2462 argptr = varg.buf;
2463 argsize = varg.len;
2464 }
2465 myptr = self->ob_bytes;
2466 mysize = Py_SIZE(self);
2467 left = lstrip_helper(myptr, mysize, argptr, argsize);
2468 if (left == mysize)
2469 right = left;
2470 else
2471 right = rstrip_helper(myptr, mysize, argptr, argsize);
2472 if (arg != Py_None)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00002473 PyBuffer_Release(&varg);
Christian Heimes44720832008-05-26 13:01:01 +00002474 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2475}
2476
2477PyDoc_STRVAR(lstrip__doc__,
2478"B.lstrip([bytes]) -> bytearray\n\
2479\n\
2480Strip leading bytes contained in the argument.\n\
2481If the argument is omitted, strip leading ASCII whitespace.");
2482static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002483bytearray_lstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002484{
2485 Py_ssize_t left, right, mysize, argsize;
2486 void *myptr, *argptr;
2487 PyObject *arg = Py_None;
2488 Py_buffer varg;
2489 if (!PyArg_ParseTuple(args, "|O:lstrip", &arg))
2490 return NULL;
2491 if (arg == Py_None) {
2492 argptr = "\t\n\r\f\v ";
2493 argsize = 6;
2494 }
2495 else {
2496 if (_getbuffer(arg, &varg) < 0)
2497 return NULL;
2498 argptr = varg.buf;
2499 argsize = varg.len;
2500 }
2501 myptr = self->ob_bytes;
2502 mysize = Py_SIZE(self);
2503 left = lstrip_helper(myptr, mysize, argptr, argsize);
2504 right = mysize;
2505 if (arg != Py_None)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00002506 PyBuffer_Release(&varg);
Christian Heimes44720832008-05-26 13:01:01 +00002507 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2508}
2509
2510PyDoc_STRVAR(rstrip__doc__,
2511"B.rstrip([bytes]) -> bytearray\n\
2512\n\
2513Strip trailing bytes contained in the argument.\n\
2514If the argument is omitted, strip trailing ASCII whitespace.");
2515static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002516bytearray_rstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002517{
2518 Py_ssize_t left, right, mysize, argsize;
2519 void *myptr, *argptr;
2520 PyObject *arg = Py_None;
2521 Py_buffer varg;
2522 if (!PyArg_ParseTuple(args, "|O:rstrip", &arg))
2523 return NULL;
2524 if (arg == Py_None) {
2525 argptr = "\t\n\r\f\v ";
2526 argsize = 6;
2527 }
2528 else {
2529 if (_getbuffer(arg, &varg) < 0)
2530 return NULL;
2531 argptr = varg.buf;
2532 argsize = varg.len;
2533 }
2534 myptr = self->ob_bytes;
2535 mysize = Py_SIZE(self);
2536 left = 0;
2537 right = rstrip_helper(myptr, mysize, argptr, argsize);
2538 if (arg != Py_None)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00002539 PyBuffer_Release(&varg);
Christian Heimes44720832008-05-26 13:01:01 +00002540 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2541}
2542
2543PyDoc_STRVAR(decode_doc,
2544"B.decode([encoding[, errors]]) -> unicode object.\n\
2545\n\
2546Decodes B using the codec registered for encoding. encoding defaults\n\
2547to the default encoding. errors may be given to set a different error\n\
2548handling scheme. Default is 'strict' meaning that encoding errors raise\n\
2549a UnicodeDecodeError. Other possible values are 'ignore' and 'replace'\n\
2550as well as any other name registered with codecs.register_error that is\n\
2551able to handle UnicodeDecodeErrors.");
2552
2553static PyObject *
Benjamin Petersondc782b52009-09-18 21:46:21 +00002554bytearray_decode(PyObject *self, PyObject *args, PyObject *kwargs)
Christian Heimes44720832008-05-26 13:01:01 +00002555{
2556 const char *encoding = NULL;
2557 const char *errors = NULL;
Benjamin Petersondc782b52009-09-18 21:46:21 +00002558 static char *kwlist[] = {"encoding", "errors", 0};
Christian Heimes44720832008-05-26 13:01:01 +00002559
Benjamin Petersondc782b52009-09-18 21:46:21 +00002560 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ss:decode", kwlist, &encoding, &errors))
Christian Heimes44720832008-05-26 13:01:01 +00002561 return NULL;
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002562 if (encoding == NULL) {
2563#ifdef Py_USING_UNICODE
Christian Heimes44720832008-05-26 13:01:01 +00002564 encoding = PyUnicode_GetDefaultEncoding();
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002565#else
2566 PyErr_SetString(PyExc_ValueError, "no encoding specified");
2567 return NULL;
2568#endif
2569 }
Serhiy Storchakac7797dc2015-05-31 20:21:00 +03002570 return _PyCodec_DecodeText(self, encoding, errors);
Christian Heimes44720832008-05-26 13:01:01 +00002571}
2572
2573PyDoc_STRVAR(alloc_doc,
2574"B.__alloc__() -> int\n\
2575\n\
2576Returns the number of bytes actually allocated.");
2577
2578static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002579bytearray_alloc(PyByteArrayObject *self)
Christian Heimes44720832008-05-26 13:01:01 +00002580{
2581 return PyInt_FromSsize_t(self->ob_alloc);
2582}
2583
2584PyDoc_STRVAR(join_doc,
2585"B.join(iterable_of_bytes) -> bytes\n\
2586\n\
2587Concatenates any number of bytearray objects, with B in between each pair.");
2588
2589static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002590bytearray_join(PyByteArrayObject *self, PyObject *it)
Christian Heimes44720832008-05-26 13:01:01 +00002591{
2592 PyObject *seq;
2593 Py_ssize_t mysize = Py_SIZE(self);
2594 Py_ssize_t i;
2595 Py_ssize_t n;
2596 PyObject **items;
2597 Py_ssize_t totalsize = 0;
2598 PyObject *result;
2599 char *dest;
2600
2601 seq = PySequence_Fast(it, "can only join an iterable");
2602 if (seq == NULL)
2603 return NULL;
2604 n = PySequence_Fast_GET_SIZE(seq);
2605 items = PySequence_Fast_ITEMS(seq);
2606
2607 /* Compute the total size, and check that they are all bytes */
2608 /* XXX Shouldn't we use _getbuffer() on these items instead? */
2609 for (i = 0; i < n; i++) {
2610 PyObject *obj = items[i];
2611 if (!PyByteArray_Check(obj) && !PyBytes_Check(obj)) {
2612 PyErr_Format(PyExc_TypeError,
2613 "can only join an iterable of bytes "
2614 "(item %ld has type '%.100s')",
2615 /* XXX %ld isn't right on Win64 */
2616 (long)i, Py_TYPE(obj)->tp_name);
2617 goto error;
2618 }
2619 if (i > 0)
2620 totalsize += mysize;
2621 totalsize += Py_SIZE(obj);
2622 if (totalsize < 0) {
2623 PyErr_NoMemory();
2624 goto error;
2625 }
2626 }
2627
2628 /* Allocate the result, and copy the bytes */
2629 result = PyByteArray_FromStringAndSize(NULL, totalsize);
2630 if (result == NULL)
2631 goto error;
2632 dest = PyByteArray_AS_STRING(result);
2633 for (i = 0; i < n; i++) {
2634 PyObject *obj = items[i];
2635 Py_ssize_t size = Py_SIZE(obj);
2636 char *buf;
2637 if (PyByteArray_Check(obj))
2638 buf = PyByteArray_AS_STRING(obj);
2639 else
2640 buf = PyBytes_AS_STRING(obj);
2641 if (i) {
2642 memcpy(dest, self->ob_bytes, mysize);
2643 dest += mysize;
2644 }
2645 memcpy(dest, buf, size);
2646 dest += size;
2647 }
2648
2649 /* Done */
2650 Py_DECREF(seq);
2651 return result;
2652
2653 /* Error handling */
2654 error:
2655 Py_DECREF(seq);
2656 return NULL;
2657}
2658
Antoine Pitrou64672132010-01-13 07:55:48 +00002659PyDoc_STRVAR(splitlines__doc__,
Raymond Hettingeraad5b022012-06-02 01:42:58 -04002660"B.splitlines(keepends=False) -> list of lines\n\
Antoine Pitrou64672132010-01-13 07:55:48 +00002661\n\
2662Return a list of the lines in B, breaking at line boundaries.\n\
2663Line breaks are not included in the resulting list unless keepends\n\
2664is given and true.");
2665
2666static PyObject*
2667bytearray_splitlines(PyObject *self, PyObject *args)
2668{
2669 int keepends = 0;
2670
2671 if (!PyArg_ParseTuple(args, "|i:splitlines", &keepends))
2672 return NULL;
2673
2674 return stringlib_splitlines(
2675 (PyObject*) self, PyByteArray_AS_STRING(self),
2676 PyByteArray_GET_SIZE(self), keepends
2677 );
2678}
2679
Christian Heimes44720832008-05-26 13:01:01 +00002680PyDoc_STRVAR(fromhex_doc,
2681"bytearray.fromhex(string) -> bytearray\n\
2682\n\
2683Create a bytearray object from a string of hexadecimal numbers.\n\
2684Spaces between two numbers are accepted.\n\
2685Example: bytearray.fromhex('B9 01EF') -> bytearray(b'\\xb9\\x01\\xef').");
2686
2687static int
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002688hex_digit_to_int(char c)
Christian Heimes44720832008-05-26 13:01:01 +00002689{
Eric Smithcac7af62009-04-27 19:04:37 +00002690 if (Py_ISDIGIT(c))
Christian Heimes44720832008-05-26 13:01:01 +00002691 return c - '0';
2692 else {
Eric Smithcac7af62009-04-27 19:04:37 +00002693 if (Py_ISUPPER(c))
2694 c = Py_TOLOWER(c);
Christian Heimes44720832008-05-26 13:01:01 +00002695 if (c >= 'a' && c <= 'f')
2696 return c - 'a' + 10;
2697 }
2698 return -1;
2699}
2700
2701static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002702bytearray_fromhex(PyObject *cls, PyObject *args)
Christian Heimes44720832008-05-26 13:01:01 +00002703{
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002704 PyObject *newbytes;
Christian Heimes44720832008-05-26 13:01:01 +00002705 char *buf;
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002706 char *hex;
Christian Heimes44720832008-05-26 13:01:01 +00002707 Py_ssize_t hexlen, byteslen, i, j;
2708 int top, bot;
2709
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002710 if (!PyArg_ParseTuple(args, "s#:fromhex", &hex, &hexlen))
Christian Heimes44720832008-05-26 13:01:01 +00002711 return NULL;
Christian Heimes44720832008-05-26 13:01:01 +00002712 byteslen = hexlen/2; /* This overestimates if there are spaces */
2713 newbytes = PyByteArray_FromStringAndSize(NULL, byteslen);
2714 if (!newbytes)
2715 return NULL;
2716 buf = PyByteArray_AS_STRING(newbytes);
2717 for (i = j = 0; i < hexlen; i += 2) {
2718 /* skip over spaces in the input */
2719 while (hex[i] == ' ')
2720 i++;
2721 if (i >= hexlen)
2722 break;
2723 top = hex_digit_to_int(hex[i]);
2724 bot = hex_digit_to_int(hex[i+1]);
2725 if (top == -1 || bot == -1) {
2726 PyErr_Format(PyExc_ValueError,
2727 "non-hexadecimal number found in "
2728 "fromhex() arg at position %zd", i);
2729 goto error;
2730 }
2731 buf[j++] = (top << 4) + bot;
2732 }
2733 if (PyByteArray_Resize(newbytes, j) < 0)
2734 goto error;
2735 return newbytes;
2736
2737 error:
2738 Py_DECREF(newbytes);
2739 return NULL;
2740}
2741
2742PyDoc_STRVAR(reduce_doc, "Return state information for pickling.");
2743
2744static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002745bytearray_reduce(PyByteArrayObject *self)
Christian Heimes44720832008-05-26 13:01:01 +00002746{
2747 PyObject *latin1, *dict;
2748 if (self->ob_bytes)
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002749#ifdef Py_USING_UNICODE
Christian Heimes44720832008-05-26 13:01:01 +00002750 latin1 = PyUnicode_DecodeLatin1(self->ob_bytes,
2751 Py_SIZE(self), NULL);
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002752#else
Mark Dickinson9d109742009-10-15 15:18:55 +00002753 latin1 = PyString_FromStringAndSize(self->ob_bytes, Py_SIZE(self));
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002754#endif
Christian Heimes44720832008-05-26 13:01:01 +00002755 else
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002756#ifdef Py_USING_UNICODE
Christian Heimes44720832008-05-26 13:01:01 +00002757 latin1 = PyUnicode_FromString("");
Benjamin Peterson78821dd2009-01-25 17:15:10 +00002758#else
2759 latin1 = PyString_FromString("");
2760#endif
Christian Heimes44720832008-05-26 13:01:01 +00002761
2762 dict = PyObject_GetAttrString((PyObject *)self, "__dict__");
2763 if (dict == NULL) {
2764 PyErr_Clear();
2765 dict = Py_None;
2766 Py_INCREF(dict);
2767 }
2768
2769 return Py_BuildValue("(O(Ns)N)", Py_TYPE(self), latin1, "latin-1", dict);
2770}
2771
Robert Schuppenies9be2ec12008-07-10 15:24:04 +00002772PyDoc_STRVAR(sizeof_doc,
2773"B.__sizeof__() -> int\n\
2774 \n\
2775Returns the size of B in memory, in bytes");
2776static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002777bytearray_sizeof(PyByteArrayObject *self)
Robert Schuppenies9be2ec12008-07-10 15:24:04 +00002778{
Georg Brandl517cfdc2009-04-05 13:16:35 +00002779 Py_ssize_t res;
Robert Schuppenies9be2ec12008-07-10 15:24:04 +00002780
Georg Brandl517cfdc2009-04-05 13:16:35 +00002781 res = sizeof(PyByteArrayObject) + self->ob_alloc * sizeof(char);
2782 return PyInt_FromSsize_t(res);
Robert Schuppenies9be2ec12008-07-10 15:24:04 +00002783}
2784
Benjamin Petersond6720012009-04-18 15:31:34 +00002785static PySequenceMethods bytearray_as_sequence = {
2786 (lenfunc)bytearray_length, /* sq_length */
Christian Heimes44720832008-05-26 13:01:01 +00002787 (binaryfunc)PyByteArray_Concat, /* sq_concat */
Benjamin Petersond6720012009-04-18 15:31:34 +00002788 (ssizeargfunc)bytearray_repeat, /* sq_repeat */
2789 (ssizeargfunc)bytearray_getitem, /* sq_item */
2790 0, /* sq_slice */
2791 (ssizeobjargproc)bytearray_setitem, /* sq_ass_item */
2792 0, /* sq_ass_slice */
2793 (objobjproc)bytearray_contains, /* sq_contains */
2794 (binaryfunc)bytearray_iconcat, /* sq_inplace_concat */
2795 (ssizeargfunc)bytearray_irepeat, /* sq_inplace_repeat */
Christian Heimes44720832008-05-26 13:01:01 +00002796};
2797
Benjamin Petersond6720012009-04-18 15:31:34 +00002798static PyMappingMethods bytearray_as_mapping = {
2799 (lenfunc)bytearray_length,
2800 (binaryfunc)bytearray_subscript,
2801 (objobjargproc)bytearray_ass_subscript,
Christian Heimes44720832008-05-26 13:01:01 +00002802};
2803
Benjamin Petersond6720012009-04-18 15:31:34 +00002804static PyBufferProcs bytearray_as_buffer = {
2805 (readbufferproc)bytearray_buffer_getreadbuf,
2806 (writebufferproc)bytearray_buffer_getwritebuf,
2807 (segcountproc)bytearray_buffer_getsegcount,
2808 (charbufferproc)bytearray_buffer_getcharbuf,
2809 (getbufferproc)bytearray_getbuffer,
2810 (releasebufferproc)bytearray_releasebuffer,
Christian Heimes44720832008-05-26 13:01:01 +00002811};
2812
2813static PyMethodDef
Benjamin Petersond6720012009-04-18 15:31:34 +00002814bytearray_methods[] = {
2815 {"__alloc__", (PyCFunction)bytearray_alloc, METH_NOARGS, alloc_doc},
2816 {"__reduce__", (PyCFunction)bytearray_reduce, METH_NOARGS, reduce_doc},
2817 {"__sizeof__", (PyCFunction)bytearray_sizeof, METH_NOARGS, sizeof_doc},
2818 {"append", (PyCFunction)bytearray_append, METH_O, append__doc__},
Christian Heimes44720832008-05-26 13:01:01 +00002819 {"capitalize", (PyCFunction)stringlib_capitalize, METH_NOARGS,
2820 _Py_capitalize__doc__},
2821 {"center", (PyCFunction)stringlib_center, METH_VARARGS, center__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002822 {"count", (PyCFunction)bytearray_count, METH_VARARGS, count__doc__},
Benjamin Petersondc782b52009-09-18 21:46:21 +00002823 {"decode", (PyCFunction)bytearray_decode, METH_VARARGS | METH_KEYWORDS, decode_doc},
Benjamin Petersond6720012009-04-18 15:31:34 +00002824 {"endswith", (PyCFunction)bytearray_endswith, METH_VARARGS, endswith__doc__},
Christian Heimes44720832008-05-26 13:01:01 +00002825 {"expandtabs", (PyCFunction)stringlib_expandtabs, METH_VARARGS,
2826 expandtabs__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002827 {"extend", (PyCFunction)bytearray_extend, METH_O, extend__doc__},
2828 {"find", (PyCFunction)bytearray_find, METH_VARARGS, find__doc__},
2829 {"fromhex", (PyCFunction)bytearray_fromhex, METH_VARARGS|METH_CLASS,
Christian Heimes44720832008-05-26 13:01:01 +00002830 fromhex_doc},
Benjamin Petersond6720012009-04-18 15:31:34 +00002831 {"index", (PyCFunction)bytearray_index, METH_VARARGS, index__doc__},
2832 {"insert", (PyCFunction)bytearray_insert, METH_VARARGS, insert__doc__},
Christian Heimes44720832008-05-26 13:01:01 +00002833 {"isalnum", (PyCFunction)stringlib_isalnum, METH_NOARGS,
2834 _Py_isalnum__doc__},
2835 {"isalpha", (PyCFunction)stringlib_isalpha, METH_NOARGS,
2836 _Py_isalpha__doc__},
2837 {"isdigit", (PyCFunction)stringlib_isdigit, METH_NOARGS,
2838 _Py_isdigit__doc__},
2839 {"islower", (PyCFunction)stringlib_islower, METH_NOARGS,
2840 _Py_islower__doc__},
2841 {"isspace", (PyCFunction)stringlib_isspace, METH_NOARGS,
2842 _Py_isspace__doc__},
2843 {"istitle", (PyCFunction)stringlib_istitle, METH_NOARGS,
2844 _Py_istitle__doc__},
2845 {"isupper", (PyCFunction)stringlib_isupper, METH_NOARGS,
2846 _Py_isupper__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002847 {"join", (PyCFunction)bytearray_join, METH_O, join_doc},
Christian Heimes44720832008-05-26 13:01:01 +00002848 {"ljust", (PyCFunction)stringlib_ljust, METH_VARARGS, ljust__doc__},
2849 {"lower", (PyCFunction)stringlib_lower, METH_NOARGS, _Py_lower__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002850 {"lstrip", (PyCFunction)bytearray_lstrip, METH_VARARGS, lstrip__doc__},
2851 {"partition", (PyCFunction)bytearray_partition, METH_O, partition__doc__},
2852 {"pop", (PyCFunction)bytearray_pop, METH_VARARGS, pop__doc__},
2853 {"remove", (PyCFunction)bytearray_remove, METH_O, remove__doc__},
2854 {"replace", (PyCFunction)bytearray_replace, METH_VARARGS, replace__doc__},
2855 {"reverse", (PyCFunction)bytearray_reverse, METH_NOARGS, reverse__doc__},
2856 {"rfind", (PyCFunction)bytearray_rfind, METH_VARARGS, rfind__doc__},
2857 {"rindex", (PyCFunction)bytearray_rindex, METH_VARARGS, rindex__doc__},
Christian Heimes44720832008-05-26 13:01:01 +00002858 {"rjust", (PyCFunction)stringlib_rjust, METH_VARARGS, rjust__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002859 {"rpartition", (PyCFunction)bytearray_rpartition, METH_O, rpartition__doc__},
2860 {"rsplit", (PyCFunction)bytearray_rsplit, METH_VARARGS, rsplit__doc__},
2861 {"rstrip", (PyCFunction)bytearray_rstrip, METH_VARARGS, rstrip__doc__},
2862 {"split", (PyCFunction)bytearray_split, METH_VARARGS, split__doc__},
Antoine Pitrou64672132010-01-13 07:55:48 +00002863 {"splitlines", (PyCFunction)bytearray_splitlines, METH_VARARGS,
Christian Heimes44720832008-05-26 13:01:01 +00002864 splitlines__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002865 {"startswith", (PyCFunction)bytearray_startswith, METH_VARARGS ,
Christian Heimes44720832008-05-26 13:01:01 +00002866 startswith__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002867 {"strip", (PyCFunction)bytearray_strip, METH_VARARGS, strip__doc__},
Christian Heimes44720832008-05-26 13:01:01 +00002868 {"swapcase", (PyCFunction)stringlib_swapcase, METH_NOARGS,
2869 _Py_swapcase__doc__},
2870 {"title", (PyCFunction)stringlib_title, METH_NOARGS, _Py_title__doc__},
Benjamin Petersond6720012009-04-18 15:31:34 +00002871 {"translate", (PyCFunction)bytearray_translate, METH_VARARGS,
Christian Heimes44720832008-05-26 13:01:01 +00002872 translate__doc__},
2873 {"upper", (PyCFunction)stringlib_upper, METH_NOARGS, _Py_upper__doc__},
2874 {"zfill", (PyCFunction)stringlib_zfill, METH_VARARGS, zfill__doc__},
2875 {NULL}
2876};
2877
Benjamin Petersond6720012009-04-18 15:31:34 +00002878PyDoc_STRVAR(bytearray_doc,
Christian Heimes44720832008-05-26 13:01:01 +00002879"bytearray(iterable_of_ints) -> bytearray.\n\
2880bytearray(string, encoding[, errors]) -> bytearray.\n\
2881bytearray(bytes_or_bytearray) -> mutable copy of bytes_or_bytearray.\n\
2882bytearray(memory_view) -> bytearray.\n\
2883\n\
2884Construct an mutable bytearray object from:\n\
2885 - an iterable yielding integers in range(256)\n\
2886 - a text string encoded using the specified encoding\n\
2887 - a bytes or a bytearray object\n\
2888 - any object implementing the buffer API.\n\
2889\n\
2890bytearray(int) -> bytearray.\n\
2891\n\
2892Construct a zero-initialized bytearray of the given length.");
2893
2894
Benjamin Petersond6720012009-04-18 15:31:34 +00002895static PyObject *bytearray_iter(PyObject *seq);
Christian Heimes44720832008-05-26 13:01:01 +00002896
2897PyTypeObject PyByteArray_Type = {
2898 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2899 "bytearray",
2900 sizeof(PyByteArrayObject),
2901 0,
Benjamin Petersond6720012009-04-18 15:31:34 +00002902 (destructor)bytearray_dealloc, /* tp_dealloc */
Christian Heimes44720832008-05-26 13:01:01 +00002903 0, /* tp_print */
2904 0, /* tp_getattr */
2905 0, /* tp_setattr */
2906 0, /* tp_compare */
Benjamin Petersond6720012009-04-18 15:31:34 +00002907 (reprfunc)bytearray_repr, /* tp_repr */
Christian Heimes44720832008-05-26 13:01:01 +00002908 0, /* tp_as_number */
Benjamin Petersond6720012009-04-18 15:31:34 +00002909 &bytearray_as_sequence, /* tp_as_sequence */
2910 &bytearray_as_mapping, /* tp_as_mapping */
Christian Heimes44720832008-05-26 13:01:01 +00002911 0, /* tp_hash */
2912 0, /* tp_call */
Benjamin Petersond6720012009-04-18 15:31:34 +00002913 bytearray_str, /* tp_str */
Christian Heimes44720832008-05-26 13:01:01 +00002914 PyObject_GenericGetAttr, /* tp_getattro */
2915 0, /* tp_setattro */
Benjamin Petersond6720012009-04-18 15:31:34 +00002916 &bytearray_as_buffer, /* tp_as_buffer */
Christian Heimes44720832008-05-26 13:01:01 +00002917 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE |
2918 Py_TPFLAGS_HAVE_NEWBUFFER, /* tp_flags */
Benjamin Petersond6720012009-04-18 15:31:34 +00002919 bytearray_doc, /* tp_doc */
Christian Heimes44720832008-05-26 13:01:01 +00002920 0, /* tp_traverse */
2921 0, /* tp_clear */
Benjamin Petersond6720012009-04-18 15:31:34 +00002922 (richcmpfunc)bytearray_richcompare, /* tp_richcompare */
Christian Heimes44720832008-05-26 13:01:01 +00002923 0, /* tp_weaklistoffset */
Benjamin Petersond6720012009-04-18 15:31:34 +00002924 bytearray_iter, /* tp_iter */
Christian Heimes44720832008-05-26 13:01:01 +00002925 0, /* tp_iternext */
Benjamin Petersond6720012009-04-18 15:31:34 +00002926 bytearray_methods, /* tp_methods */
Christian Heimes44720832008-05-26 13:01:01 +00002927 0, /* tp_members */
2928 0, /* tp_getset */
2929 0, /* tp_base */
2930 0, /* tp_dict */
2931 0, /* tp_descr_get */
2932 0, /* tp_descr_set */
2933 0, /* tp_dictoffset */
Benjamin Petersond6720012009-04-18 15:31:34 +00002934 (initproc)bytearray_init, /* tp_init */
Christian Heimes44720832008-05-26 13:01:01 +00002935 PyType_GenericAlloc, /* tp_alloc */
2936 PyType_GenericNew, /* tp_new */
2937 PyObject_Del, /* tp_free */
2938};
2939
2940/*********************** Bytes Iterator ****************************/
2941
2942typedef struct {
2943 PyObject_HEAD
2944 Py_ssize_t it_index;
2945 PyByteArrayObject *it_seq; /* Set to NULL when iterator is exhausted */
2946} bytesiterobject;
2947
2948static void
Benjamin Petersond6720012009-04-18 15:31:34 +00002949bytearrayiter_dealloc(bytesiterobject *it)
Christian Heimes44720832008-05-26 13:01:01 +00002950{
2951 _PyObject_GC_UNTRACK(it);
2952 Py_XDECREF(it->it_seq);
2953 PyObject_GC_Del(it);
2954}
2955
2956static int
Benjamin Petersond6720012009-04-18 15:31:34 +00002957bytearrayiter_traverse(bytesiterobject *it, visitproc visit, void *arg)
Christian Heimes44720832008-05-26 13:01:01 +00002958{
2959 Py_VISIT(it->it_seq);
2960 return 0;
2961}
2962
2963static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002964bytearrayiter_next(bytesiterobject *it)
Christian Heimes44720832008-05-26 13:01:01 +00002965{
2966 PyByteArrayObject *seq;
2967 PyObject *item;
2968
2969 assert(it != NULL);
2970 seq = it->it_seq;
2971 if (seq == NULL)
2972 return NULL;
2973 assert(PyByteArray_Check(seq));
2974
2975 if (it->it_index < PyByteArray_GET_SIZE(seq)) {
2976 item = PyInt_FromLong(
2977 (unsigned char)seq->ob_bytes[it->it_index]);
2978 if (item != NULL)
2979 ++it->it_index;
2980 return item;
2981 }
2982
2983 Py_DECREF(seq);
2984 it->it_seq = NULL;
2985 return NULL;
2986}
2987
2988static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00002989bytesarrayiter_length_hint(bytesiterobject *it)
Christian Heimes44720832008-05-26 13:01:01 +00002990{
2991 Py_ssize_t len = 0;
2992 if (it->it_seq)
2993 len = PyByteArray_GET_SIZE(it->it_seq) - it->it_index;
2994 return PyInt_FromSsize_t(len);
2995}
2996
2997PyDoc_STRVAR(length_hint_doc,
2998 "Private method returning an estimate of len(list(it)).");
2999
Benjamin Petersond6720012009-04-18 15:31:34 +00003000static PyMethodDef bytearrayiter_methods[] = {
3001 {"__length_hint__", (PyCFunction)bytesarrayiter_length_hint, METH_NOARGS,
Christian Heimes44720832008-05-26 13:01:01 +00003002 length_hint_doc},
3003 {NULL, NULL} /* sentinel */
3004};
3005
3006PyTypeObject PyByteArrayIter_Type = {
3007 PyVarObject_HEAD_INIT(&PyType_Type, 0)
3008 "bytearray_iterator", /* tp_name */
3009 sizeof(bytesiterobject), /* tp_basicsize */
3010 0, /* tp_itemsize */
3011 /* methods */
Benjamin Petersond6720012009-04-18 15:31:34 +00003012 (destructor)bytearrayiter_dealloc, /* tp_dealloc */
Christian Heimes44720832008-05-26 13:01:01 +00003013 0, /* tp_print */
3014 0, /* tp_getattr */
3015 0, /* tp_setattr */
3016 0, /* tp_compare */
3017 0, /* tp_repr */
3018 0, /* tp_as_number */
3019 0, /* tp_as_sequence */
3020 0, /* tp_as_mapping */
3021 0, /* tp_hash */
3022 0, /* tp_call */
3023 0, /* tp_str */
3024 PyObject_GenericGetAttr, /* tp_getattro */
3025 0, /* tp_setattro */
3026 0, /* tp_as_buffer */
3027 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */
3028 0, /* tp_doc */
Benjamin Petersond6720012009-04-18 15:31:34 +00003029 (traverseproc)bytearrayiter_traverse, /* tp_traverse */
Christian Heimes44720832008-05-26 13:01:01 +00003030 0, /* tp_clear */
3031 0, /* tp_richcompare */
3032 0, /* tp_weaklistoffset */
3033 PyObject_SelfIter, /* tp_iter */
Benjamin Petersond6720012009-04-18 15:31:34 +00003034 (iternextfunc)bytearrayiter_next, /* tp_iternext */
3035 bytearrayiter_methods, /* tp_methods */
Christian Heimes44720832008-05-26 13:01:01 +00003036 0,
3037};
3038
3039static PyObject *
Benjamin Petersond6720012009-04-18 15:31:34 +00003040bytearray_iter(PyObject *seq)
Christian Heimes44720832008-05-26 13:01:01 +00003041{
3042 bytesiterobject *it;
3043
3044 if (!PyByteArray_Check(seq)) {
3045 PyErr_BadInternalCall();
3046 return NULL;
3047 }
3048 it = PyObject_GC_New(bytesiterobject, &PyByteArrayIter_Type);
3049 if (it == NULL)
3050 return NULL;
3051 it->it_index = 0;
3052 Py_INCREF(seq);
3053 it->it_seq = (PyByteArrayObject *)seq;
3054 _PyObject_GC_TRACK(it);
3055 return (PyObject *)it;
3056}