blob: 7d525222b917f43a34dc912c83313274dae84364 [file] [log] [blame]
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001/* PyByteArray (bytearray) implementation */
2
3#define PY_SSIZE_T_CLEAN
4#include "Python.h"
5#include "structmember.h"
6#include "bytes_methods.h"
7
Antoine Pitroufc8d6f42010-01-17 12:38:54 +00008char _PyByteArray_empty_string[] = "";
Christian Heimes2c9c7a52008-05-26 13:42:13 +00009
10void
11PyByteArray_Fini(void)
12{
Christian Heimes2c9c7a52008-05-26 13:42:13 +000013}
14
15int
16PyByteArray_Init(void)
17{
Christian Heimes2c9c7a52008-05-26 13:42:13 +000018 return 1;
19}
20
21/* end nullbytes support */
22
23/* Helpers */
24
25static int
26_getbytevalue(PyObject* arg, int *value)
27{
28 long face_value;
29
30 if (PyLong_Check(arg)) {
31 face_value = PyLong_AsLong(arg);
Georg Brandl9a54d7c2008-07-16 23:15:30 +000032 } else {
33 PyObject *index = PyNumber_Index(arg);
34 if (index == NULL) {
35 PyErr_Format(PyExc_TypeError, "an integer is required");
Christian Heimes2c9c7a52008-05-26 13:42:13 +000036 return 0;
37 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +000038 face_value = PyLong_AsLong(index);
39 Py_DECREF(index);
40 }
41
42 if (face_value < 0 || face_value >= 256) {
43 /* this includes the OverflowError in case the long is too large */
44 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
Christian Heimes2c9c7a52008-05-26 13:42:13 +000045 return 0;
46 }
47
48 *value = face_value;
49 return 1;
50}
51
52static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +000053bytearray_getbuffer(PyByteArrayObject *obj, Py_buffer *view, int flags)
Christian Heimes2c9c7a52008-05-26 13:42:13 +000054{
55 int ret;
56 void *ptr;
57 if (view == NULL) {
58 obj->ob_exports++;
59 return 0;
60 }
Antoine Pitroufc8d6f42010-01-17 12:38:54 +000061 ptr = (void *) PyByteArray_AS_STRING(obj);
Martin v. Löwis423be952008-08-13 15:53:07 +000062 ret = PyBuffer_FillInfo(view, (PyObject*)obj, ptr, Py_SIZE(obj), 0, flags);
Christian Heimes2c9c7a52008-05-26 13:42:13 +000063 if (ret >= 0) {
64 obj->ob_exports++;
65 }
66 return ret;
67}
68
69static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +000070bytearray_releasebuffer(PyByteArrayObject *obj, Py_buffer *view)
Christian Heimes2c9c7a52008-05-26 13:42:13 +000071{
72 obj->ob_exports--;
73}
74
75static Py_ssize_t
76_getbuffer(PyObject *obj, Py_buffer *view)
77{
78 PyBufferProcs *buffer = Py_TYPE(obj)->tp_as_buffer;
79
80 if (buffer == NULL || buffer->bf_getbuffer == NULL)
81 {
82 PyErr_Format(PyExc_TypeError,
83 "Type %.100s doesn't support the buffer API",
84 Py_TYPE(obj)->tp_name);
85 return -1;
86 }
87
88 if (buffer->bf_getbuffer(obj, view, PyBUF_SIMPLE) < 0)
89 return -1;
90 return view->len;
91}
92
Antoine Pitrou5504e892008-12-06 21:27:53 +000093static int
94_canresize(PyByteArrayObject *self)
95{
96 if (self->ob_exports > 0) {
97 PyErr_SetString(PyExc_BufferError,
98 "Existing exports of data: object cannot be re-sized");
99 return 0;
100 }
101 return 1;
102}
103
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000104/* Direct API functions */
105
106PyObject *
107PyByteArray_FromObject(PyObject *input)
108{
109 return PyObject_CallFunctionObjArgs((PyObject *)&PyByteArray_Type,
110 input, NULL);
111}
112
113PyObject *
114PyByteArray_FromStringAndSize(const char *bytes, Py_ssize_t size)
115{
116 PyByteArrayObject *new;
117 Py_ssize_t alloc;
118
119 if (size < 0) {
120 PyErr_SetString(PyExc_SystemError,
121 "Negative size passed to PyByteArray_FromStringAndSize");
122 return NULL;
123 }
124
Neal Norwitz3ce5d922008-08-24 07:08:55 +0000125 /* Prevent buffer overflow when setting alloc to size+1. */
126 if (size == PY_SSIZE_T_MAX) {
127 return PyErr_NoMemory();
128 }
129
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000130 new = PyObject_New(PyByteArrayObject, &PyByteArray_Type);
131 if (new == NULL)
132 return NULL;
133
134 if (size == 0) {
135 new->ob_bytes = NULL;
136 alloc = 0;
137 }
138 else {
139 alloc = size + 1;
140 new->ob_bytes = PyMem_Malloc(alloc);
141 if (new->ob_bytes == NULL) {
142 Py_DECREF(new);
143 return PyErr_NoMemory();
144 }
Antoine Pitroufc8d6f42010-01-17 12:38:54 +0000145 if (bytes != NULL && size > 0)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000146 memcpy(new->ob_bytes, bytes, size);
147 new->ob_bytes[size] = '\0'; /* Trailing null byte */
148 }
149 Py_SIZE(new) = size;
150 new->ob_alloc = alloc;
151 new->ob_exports = 0;
152
153 return (PyObject *)new;
154}
155
156Py_ssize_t
157PyByteArray_Size(PyObject *self)
158{
159 assert(self != NULL);
160 assert(PyByteArray_Check(self));
161
162 return PyByteArray_GET_SIZE(self);
163}
164
165char *
166PyByteArray_AsString(PyObject *self)
167{
168 assert(self != NULL);
169 assert(PyByteArray_Check(self));
170
171 return PyByteArray_AS_STRING(self);
172}
173
174int
175PyByteArray_Resize(PyObject *self, Py_ssize_t size)
176{
177 void *sval;
178 Py_ssize_t alloc = ((PyByteArrayObject *)self)->ob_alloc;
179
180 assert(self != NULL);
181 assert(PyByteArray_Check(self));
182 assert(size >= 0);
183
Antoine Pitrou5504e892008-12-06 21:27:53 +0000184 if (size == Py_SIZE(self)) {
185 return 0;
186 }
187 if (!_canresize((PyByteArrayObject *)self)) {
188 return -1;
189 }
190
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000191 if (size < alloc / 2) {
192 /* Major downsize; resize down to exact size */
193 alloc = size + 1;
194 }
195 else if (size < alloc) {
196 /* Within allocated size; quick exit */
197 Py_SIZE(self) = size;
198 ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null */
199 return 0;
200 }
201 else if (size <= alloc * 1.125) {
202 /* Moderate upsize; overallocate similar to list_resize() */
203 alloc = size + (size >> 3) + (size < 9 ? 3 : 6);
204 }
205 else {
206 /* Major upsize; resize up to exact size */
207 alloc = size + 1;
208 }
209
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000210 sval = PyMem_Realloc(((PyByteArrayObject *)self)->ob_bytes, alloc);
211 if (sval == NULL) {
212 PyErr_NoMemory();
213 return -1;
214 }
215
216 ((PyByteArrayObject *)self)->ob_bytes = sval;
217 Py_SIZE(self) = size;
218 ((PyByteArrayObject *)self)->ob_alloc = alloc;
219 ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null byte */
220
221 return 0;
222}
223
224PyObject *
225PyByteArray_Concat(PyObject *a, PyObject *b)
226{
227 Py_ssize_t size;
228 Py_buffer va, vb;
229 PyByteArrayObject *result = NULL;
230
231 va.len = -1;
232 vb.len = -1;
233 if (_getbuffer(a, &va) < 0 ||
234 _getbuffer(b, &vb) < 0) {
235 PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s",
236 Py_TYPE(a)->tp_name, Py_TYPE(b)->tp_name);
237 goto done;
238 }
239
240 size = va.len + vb.len;
241 if (size < 0) {
Benjamin Petersone0124bd2009-03-09 21:04:33 +0000242 PyErr_NoMemory();
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000243 goto done;
244 }
245
246 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, size);
247 if (result != NULL) {
248 memcpy(result->ob_bytes, va.buf, va.len);
249 memcpy(result->ob_bytes + va.len, vb.buf, vb.len);
250 }
251
252 done:
253 if (va.len != -1)
Martin v. Löwis423be952008-08-13 15:53:07 +0000254 PyBuffer_Release(&va);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000255 if (vb.len != -1)
Martin v. Löwis423be952008-08-13 15:53:07 +0000256 PyBuffer_Release(&vb);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000257 return (PyObject *)result;
258}
259
260/* Functions stuffed into the type object */
261
262static Py_ssize_t
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000263bytearray_length(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000264{
265 return Py_SIZE(self);
266}
267
268static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000269bytearray_iconcat(PyByteArrayObject *self, PyObject *other)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000270{
271 Py_ssize_t mysize;
272 Py_ssize_t size;
273 Py_buffer vo;
274
275 if (_getbuffer(other, &vo) < 0) {
276 PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s",
277 Py_TYPE(other)->tp_name, Py_TYPE(self)->tp_name);
278 return NULL;
279 }
280
281 mysize = Py_SIZE(self);
282 size = mysize + vo.len;
283 if (size < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +0000284 PyBuffer_Release(&vo);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000285 return PyErr_NoMemory();
286 }
287 if (size < self->ob_alloc) {
288 Py_SIZE(self) = size;
289 self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */
290 }
291 else if (PyByteArray_Resize((PyObject *)self, size) < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +0000292 PyBuffer_Release(&vo);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000293 return NULL;
294 }
295 memcpy(self->ob_bytes + mysize, vo.buf, vo.len);
Martin v. Löwis423be952008-08-13 15:53:07 +0000296 PyBuffer_Release(&vo);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000297 Py_INCREF(self);
298 return (PyObject *)self;
299}
300
301static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000302bytearray_repeat(PyByteArrayObject *self, Py_ssize_t count)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000303{
304 PyByteArrayObject *result;
305 Py_ssize_t mysize;
306 Py_ssize_t size;
307
308 if (count < 0)
309 count = 0;
310 mysize = Py_SIZE(self);
311 size = mysize * count;
312 if (count != 0 && size / count != mysize)
313 return PyErr_NoMemory();
314 result = (PyByteArrayObject *)PyByteArray_FromStringAndSize(NULL, size);
315 if (result != NULL && size != 0) {
316 if (mysize == 1)
317 memset(result->ob_bytes, self->ob_bytes[0], size);
318 else {
319 Py_ssize_t i;
320 for (i = 0; i < count; i++)
321 memcpy(result->ob_bytes + i*mysize, self->ob_bytes, mysize);
322 }
323 }
324 return (PyObject *)result;
325}
326
327static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000328bytearray_irepeat(PyByteArrayObject *self, Py_ssize_t count)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000329{
330 Py_ssize_t mysize;
331 Py_ssize_t size;
332
333 if (count < 0)
334 count = 0;
335 mysize = Py_SIZE(self);
336 size = mysize * count;
337 if (count != 0 && size / count != mysize)
338 return PyErr_NoMemory();
339 if (size < self->ob_alloc) {
340 Py_SIZE(self) = size;
341 self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */
342 }
343 else if (PyByteArray_Resize((PyObject *)self, size) < 0)
344 return NULL;
345
346 if (mysize == 1)
347 memset(self->ob_bytes, self->ob_bytes[0], size);
348 else {
349 Py_ssize_t i;
350 for (i = 1; i < count; i++)
351 memcpy(self->ob_bytes + i*mysize, self->ob_bytes, mysize);
352 }
353
354 Py_INCREF(self);
355 return (PyObject *)self;
356}
357
358static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000359bytearray_getitem(PyByteArrayObject *self, Py_ssize_t i)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000360{
361 if (i < 0)
362 i += Py_SIZE(self);
363 if (i < 0 || i >= Py_SIZE(self)) {
364 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
365 return NULL;
366 }
367 return PyLong_FromLong((unsigned char)(self->ob_bytes[i]));
368}
369
370static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000371bytearray_subscript(PyByteArrayObject *self, PyObject *index)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000372{
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000373 if (PyIndex_Check(index)) {
374 Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000375
376 if (i == -1 && PyErr_Occurred())
377 return NULL;
378
379 if (i < 0)
380 i += PyByteArray_GET_SIZE(self);
381
382 if (i < 0 || i >= Py_SIZE(self)) {
383 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
384 return NULL;
385 }
386 return PyLong_FromLong((unsigned char)(self->ob_bytes[i]));
387 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000388 else if (PySlice_Check(index)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000389 Py_ssize_t start, stop, step, slicelength, cur, i;
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000390 if (PySlice_GetIndicesEx((PySliceObject *)index,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000391 PyByteArray_GET_SIZE(self),
392 &start, &stop, &step, &slicelength) < 0) {
393 return NULL;
394 }
395
396 if (slicelength <= 0)
397 return PyByteArray_FromStringAndSize("", 0);
398 else if (step == 1) {
399 return PyByteArray_FromStringAndSize(self->ob_bytes + start,
400 slicelength);
401 }
402 else {
403 char *source_buf = PyByteArray_AS_STRING(self);
Alexandre Vassalottie2641f42009-04-03 06:38:02 +0000404 char *result_buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000405 PyObject *result;
406
Alexandre Vassalottie2641f42009-04-03 06:38:02 +0000407 result = PyByteArray_FromStringAndSize(NULL, slicelength);
408 if (result == NULL)
409 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000410
Alexandre Vassalottie2641f42009-04-03 06:38:02 +0000411 result_buf = PyByteArray_AS_STRING(result);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000412 for (cur = start, i = 0; i < slicelength;
413 cur += step, i++) {
414 result_buf[i] = source_buf[cur];
415 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000416 return result;
417 }
418 }
419 else {
420 PyErr_SetString(PyExc_TypeError, "bytearray indices must be integers");
421 return NULL;
422 }
423}
424
425static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000426bytearray_setslice(PyByteArrayObject *self, Py_ssize_t lo, Py_ssize_t hi,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000427 PyObject *values)
428{
429 Py_ssize_t avail, needed;
430 void *bytes;
431 Py_buffer vbytes;
432 int res = 0;
433
434 vbytes.len = -1;
435 if (values == (PyObject *)self) {
436 /* Make a copy and call this function recursively */
437 int err;
438 values = PyByteArray_FromObject(values);
439 if (values == NULL)
440 return -1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000441 err = bytearray_setslice(self, lo, hi, values);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000442 Py_DECREF(values);
443 return err;
444 }
445 if (values == NULL) {
446 /* del b[lo:hi] */
447 bytes = NULL;
448 needed = 0;
449 }
450 else {
451 if (_getbuffer(values, &vbytes) < 0) {
452 PyErr_Format(PyExc_TypeError,
Georg Brandl3dbca812008-07-23 16:10:53 +0000453 "can't set bytearray slice from %.100s",
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000454 Py_TYPE(values)->tp_name);
455 return -1;
456 }
457 needed = vbytes.len;
458 bytes = vbytes.buf;
459 }
460
461 if (lo < 0)
462 lo = 0;
463 if (hi < lo)
464 hi = lo;
465 if (hi > Py_SIZE(self))
466 hi = Py_SIZE(self);
467
468 avail = hi - lo;
469 if (avail < 0)
470 lo = hi = avail = 0;
471
472 if (avail != needed) {
473 if (avail > needed) {
Antoine Pitrou5504e892008-12-06 21:27:53 +0000474 if (!_canresize(self)) {
475 res = -1;
476 goto finish;
477 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000478 /*
479 0 lo hi old_size
480 | |<----avail----->|<-----tomove------>|
481 | |<-needed->|<-----tomove------>|
482 0 lo new_hi new_size
483 */
484 memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi,
485 Py_SIZE(self) - hi);
486 }
487 /* XXX(nnorwitz): need to verify this can't overflow! */
488 if (PyByteArray_Resize((PyObject *)self,
489 Py_SIZE(self) + needed - avail) < 0) {
490 res = -1;
491 goto finish;
492 }
493 if (avail < needed) {
494 /*
495 0 lo hi old_size
496 | |<-avail->|<-----tomove------>|
497 | |<----needed---->|<-----tomove------>|
498 0 lo new_hi new_size
499 */
500 memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi,
501 Py_SIZE(self) - lo - needed);
502 }
503 }
504
505 if (needed > 0)
506 memcpy(self->ob_bytes + lo, bytes, needed);
507
508
509 finish:
510 if (vbytes.len != -1)
Martin v. Löwis423be952008-08-13 15:53:07 +0000511 PyBuffer_Release(&vbytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000512 return res;
513}
514
515static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000516bytearray_setitem(PyByteArrayObject *self, Py_ssize_t i, PyObject *value)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000517{
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000518 int ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000519
520 if (i < 0)
521 i += Py_SIZE(self);
522
523 if (i < 0 || i >= Py_SIZE(self)) {
524 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
525 return -1;
526 }
527
528 if (value == NULL)
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000529 return bytearray_setslice(self, i, i+1, NULL);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000530
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000531 if (!_getbytevalue(value, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000532 return -1;
533
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000534 self->ob_bytes[i] = ival;
535 return 0;
536}
537
538static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000539bytearray_ass_subscript(PyByteArrayObject *self, PyObject *index, PyObject *values)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000540{
541 Py_ssize_t start, stop, step, slicelen, needed;
542 char *bytes;
543
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000544 if (PyIndex_Check(index)) {
545 Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000546
547 if (i == -1 && PyErr_Occurred())
548 return -1;
549
550 if (i < 0)
551 i += PyByteArray_GET_SIZE(self);
552
553 if (i < 0 || i >= Py_SIZE(self)) {
554 PyErr_SetString(PyExc_IndexError, "bytearray index out of range");
555 return -1;
556 }
557
558 if (values == NULL) {
559 /* Fall through to slice assignment */
560 start = i;
561 stop = i + 1;
562 step = 1;
563 slicelen = 1;
564 }
565 else {
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000566 int ival;
567 if (!_getbytevalue(values, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000568 return -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000569 self->ob_bytes[i] = (char)ival;
570 return 0;
571 }
572 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000573 else if (PySlice_Check(index)) {
574 if (PySlice_GetIndicesEx((PySliceObject *)index,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000575 PyByteArray_GET_SIZE(self),
576 &start, &stop, &step, &slicelen) < 0) {
577 return -1;
578 }
579 }
580 else {
581 PyErr_SetString(PyExc_TypeError, "bytearray indices must be integer");
582 return -1;
583 }
584
585 if (values == NULL) {
586 bytes = NULL;
587 needed = 0;
588 }
589 else if (values == (PyObject *)self || !PyByteArray_Check(values)) {
590 /* Make a copy an call this function recursively */
591 int err;
592 values = PyByteArray_FromObject(values);
593 if (values == NULL)
594 return -1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000595 err = bytearray_ass_subscript(self, index, values);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000596 Py_DECREF(values);
597 return err;
598 }
599 else {
600 assert(PyByteArray_Check(values));
601 bytes = ((PyByteArrayObject *)values)->ob_bytes;
602 needed = Py_SIZE(values);
603 }
604 /* Make sure b[5:2] = ... inserts before 5, not before 2. */
605 if ((step < 0 && start < stop) ||
606 (step > 0 && start > stop))
607 stop = start;
608 if (step == 1) {
609 if (slicelen != needed) {
Antoine Pitrou5504e892008-12-06 21:27:53 +0000610 if (!_canresize(self))
611 return -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000612 if (slicelen > needed) {
613 /*
614 0 start stop old_size
615 | |<---slicelen--->|<-----tomove------>|
616 | |<-needed->|<-----tomove------>|
617 0 lo new_hi new_size
618 */
619 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
620 Py_SIZE(self) - stop);
621 }
622 if (PyByteArray_Resize((PyObject *)self,
623 Py_SIZE(self) + needed - slicelen) < 0)
624 return -1;
625 if (slicelen < needed) {
626 /*
627 0 lo hi old_size
628 | |<-avail->|<-----tomove------>|
629 | |<----needed---->|<-----tomove------>|
630 0 lo new_hi new_size
631 */
632 memmove(self->ob_bytes + start + needed, self->ob_bytes + stop,
633 Py_SIZE(self) - start - needed);
634 }
635 }
636
637 if (needed > 0)
638 memcpy(self->ob_bytes + start, bytes, needed);
639
640 return 0;
641 }
642 else {
643 if (needed == 0) {
644 /* Delete slice */
645 Py_ssize_t cur, i;
646
Antoine Pitrou5504e892008-12-06 21:27:53 +0000647 if (!_canresize(self))
648 return -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000649 if (step < 0) {
650 stop = start + 1;
651 start = stop + step * (slicelen - 1) - 1;
652 step = -step;
653 }
654 for (cur = start, i = 0;
655 i < slicelen; cur += step, i++) {
656 Py_ssize_t lim = step - 1;
657
658 if (cur + step >= PyByteArray_GET_SIZE(self))
659 lim = PyByteArray_GET_SIZE(self) - cur - 1;
660
661 memmove(self->ob_bytes + cur - i,
662 self->ob_bytes + cur + 1, lim);
663 }
664 /* Move the tail of the bytes, in one chunk */
665 cur = start + slicelen*step;
666 if (cur < PyByteArray_GET_SIZE(self)) {
667 memmove(self->ob_bytes + cur - slicelen,
668 self->ob_bytes + cur,
669 PyByteArray_GET_SIZE(self) - cur);
670 }
671 if (PyByteArray_Resize((PyObject *)self,
672 PyByteArray_GET_SIZE(self) - slicelen) < 0)
673 return -1;
674
675 return 0;
676 }
677 else {
678 /* Assign slice */
679 Py_ssize_t cur, i;
680
681 if (needed != slicelen) {
682 PyErr_Format(PyExc_ValueError,
683 "attempt to assign bytes of size %zd "
684 "to extended slice of size %zd",
685 needed, slicelen);
686 return -1;
687 }
688 for (cur = start, i = 0; i < slicelen; cur += step, i++)
689 self->ob_bytes[cur] = bytes[i];
690 return 0;
691 }
692 }
693}
694
695static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000696bytearray_init(PyByteArrayObject *self, PyObject *args, PyObject *kwds)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000697{
698 static char *kwlist[] = {"source", "encoding", "errors", 0};
699 PyObject *arg = NULL;
700 const char *encoding = NULL;
701 const char *errors = NULL;
702 Py_ssize_t count;
703 PyObject *it;
704 PyObject *(*iternext)(PyObject *);
705
706 if (Py_SIZE(self) != 0) {
707 /* Empty previous contents (yes, do this first of all!) */
708 if (PyByteArray_Resize((PyObject *)self, 0) < 0)
709 return -1;
710 }
711
712 /* Parse arguments */
Georg Brandl3dbca812008-07-23 16:10:53 +0000713 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|Oss:bytearray", kwlist,
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000714 &arg, &encoding, &errors))
715 return -1;
716
717 /* Make a quick exit if no first argument */
718 if (arg == NULL) {
719 if (encoding != NULL || errors != NULL) {
720 PyErr_SetString(PyExc_TypeError,
721 "encoding or errors without sequence argument");
722 return -1;
723 }
724 return 0;
725 }
726
727 if (PyUnicode_Check(arg)) {
728 /* Encode via the codec registry */
729 PyObject *encoded, *new;
730 if (encoding == NULL) {
731 PyErr_SetString(PyExc_TypeError,
732 "string argument without an encoding");
733 return -1;
734 }
Marc-André Lemburgb2750b52008-06-06 12:18:17 +0000735 encoded = PyUnicode_AsEncodedString(arg, encoding, errors);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000736 if (encoded == NULL)
737 return -1;
738 assert(PyBytes_Check(encoded));
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000739 new = bytearray_iconcat(self, encoded);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000740 Py_DECREF(encoded);
741 if (new == NULL)
742 return -1;
743 Py_DECREF(new);
744 return 0;
745 }
746
747 /* If it's not unicode, there can't be encoding or errors */
748 if (encoding != NULL || errors != NULL) {
749 PyErr_SetString(PyExc_TypeError,
750 "encoding or errors without a string argument");
751 return -1;
752 }
753
754 /* Is it an int? */
755 count = PyNumber_AsSsize_t(arg, PyExc_ValueError);
756 if (count == -1 && PyErr_Occurred())
757 PyErr_Clear();
758 else {
759 if (count < 0) {
760 PyErr_SetString(PyExc_ValueError, "negative count");
761 return -1;
762 }
763 if (count > 0) {
764 if (PyByteArray_Resize((PyObject *)self, count))
765 return -1;
766 memset(self->ob_bytes, 0, count);
767 }
768 return 0;
769 }
770
771 /* Use the buffer API */
772 if (PyObject_CheckBuffer(arg)) {
773 Py_ssize_t size;
774 Py_buffer view;
775 if (PyObject_GetBuffer(arg, &view, PyBUF_FULL_RO) < 0)
776 return -1;
777 size = view.len;
778 if (PyByteArray_Resize((PyObject *)self, size) < 0) goto fail;
779 if (PyBuffer_ToContiguous(self->ob_bytes, &view, size, 'C') < 0)
780 goto fail;
Martin v. Löwis423be952008-08-13 15:53:07 +0000781 PyBuffer_Release(&view);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000782 return 0;
783 fail:
Martin v. Löwis423be952008-08-13 15:53:07 +0000784 PyBuffer_Release(&view);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000785 return -1;
786 }
787
788 /* XXX Optimize this if the arguments is a list, tuple */
789
790 /* Get the iterator */
791 it = PyObject_GetIter(arg);
792 if (it == NULL)
793 return -1;
794 iternext = *Py_TYPE(it)->tp_iternext;
795
796 /* Run the iterator to exhaustion */
797 for (;;) {
798 PyObject *item;
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000799 int rc, value;
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000800
801 /* Get the next item */
802 item = iternext(it);
803 if (item == NULL) {
804 if (PyErr_Occurred()) {
805 if (!PyErr_ExceptionMatches(PyExc_StopIteration))
806 goto error;
807 PyErr_Clear();
808 }
809 break;
810 }
811
812 /* Interpret it as an int (__index__) */
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000813 rc = _getbytevalue(item, &value);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000814 Py_DECREF(item);
Georg Brandl9a54d7c2008-07-16 23:15:30 +0000815 if (!rc)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000816 goto error;
817
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000818 /* Append the byte */
819 if (Py_SIZE(self) < self->ob_alloc)
820 Py_SIZE(self)++;
821 else if (PyByteArray_Resize((PyObject *)self, Py_SIZE(self)+1) < 0)
822 goto error;
823 self->ob_bytes[Py_SIZE(self)-1] = value;
824 }
825
826 /* Clean up and return success */
827 Py_DECREF(it);
828 return 0;
829
830 error:
831 /* Error handling when it != NULL */
832 Py_DECREF(it);
833 return -1;
834}
835
836/* Mostly copied from string_repr, but without the
837 "smart quote" functionality. */
838static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000839bytearray_repr(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000840{
841 static const char *hexdigits = "0123456789abcdef";
842 const char *quote_prefix = "bytearray(b";
843 const char *quote_postfix = ")";
844 Py_ssize_t length = Py_SIZE(self);
845 /* 14 == strlen(quote_prefix) + 2 + strlen(quote_postfix) */
846 size_t newsize = 14 + 4 * length;
847 PyObject *v;
848 if (newsize > PY_SSIZE_T_MAX || newsize / 4 - 3 != length) {
849 PyErr_SetString(PyExc_OverflowError,
850 "bytearray object is too large to make repr");
851 return NULL;
852 }
853 v = PyUnicode_FromUnicode(NULL, newsize);
854 if (v == NULL) {
855 return NULL;
856 }
857 else {
858 register Py_ssize_t i;
859 register Py_UNICODE c;
860 register Py_UNICODE *p;
861 int quote;
862
863 /* Figure out which quote to use; single is preferred */
864 quote = '\'';
865 {
866 char *test, *start;
867 start = PyByteArray_AS_STRING(self);
868 for (test = start; test < start+length; ++test) {
869 if (*test == '"') {
870 quote = '\''; /* back to single */
871 goto decided;
872 }
873 else if (*test == '\'')
874 quote = '"';
875 }
876 decided:
877 ;
878 }
879
880 p = PyUnicode_AS_UNICODE(v);
881 while (*quote_prefix)
882 *p++ = *quote_prefix++;
883 *p++ = quote;
884
885 for (i = 0; i < length; i++) {
886 /* There's at least enough room for a hex escape
887 and a closing quote. */
888 assert(newsize - (p - PyUnicode_AS_UNICODE(v)) >= 5);
889 c = self->ob_bytes[i];
890 if (c == '\'' || c == '\\')
891 *p++ = '\\', *p++ = c;
892 else if (c == '\t')
893 *p++ = '\\', *p++ = 't';
894 else if (c == '\n')
895 *p++ = '\\', *p++ = 'n';
896 else if (c == '\r')
897 *p++ = '\\', *p++ = 'r';
898 else if (c == 0)
899 *p++ = '\\', *p++ = 'x', *p++ = '0', *p++ = '0';
900 else if (c < ' ' || c >= 0x7f) {
901 *p++ = '\\';
902 *p++ = 'x';
903 *p++ = hexdigits[(c & 0xf0) >> 4];
904 *p++ = hexdigits[c & 0xf];
905 }
906 else
907 *p++ = c;
908 }
909 assert(newsize - (p - PyUnicode_AS_UNICODE(v)) >= 1);
910 *p++ = quote;
911 while (*quote_postfix) {
912 *p++ = *quote_postfix++;
913 }
914 *p = '\0';
915 if (PyUnicode_Resize(&v, (p - PyUnicode_AS_UNICODE(v)))) {
916 Py_DECREF(v);
917 return NULL;
918 }
919 return v;
920 }
921}
922
923static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000924bytearray_str(PyObject *op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000925{
926 if (Py_BytesWarningFlag) {
927 if (PyErr_WarnEx(PyExc_BytesWarning,
928 "str() on a bytearray instance", 1))
929 return NULL;
930 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000931 return bytearray_repr((PyByteArrayObject*)op);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000932}
933
934static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +0000935bytearray_richcompare(PyObject *self, PyObject *other, int op)
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000936{
937 Py_ssize_t self_size, other_size;
938 Py_buffer self_bytes, other_bytes;
939 PyObject *res;
940 Py_ssize_t minsize;
941 int cmp;
942
943 /* Bytes can be compared to anything that supports the (binary)
944 buffer API. Except that a comparison with Unicode is always an
945 error, even if the comparison is for equality. */
946 if (PyObject_IsInstance(self, (PyObject*)&PyUnicode_Type) ||
947 PyObject_IsInstance(other, (PyObject*)&PyUnicode_Type)) {
Barry Warsaw9e9dcd62008-10-17 01:50:37 +0000948 if (Py_BytesWarningFlag && (op == Py_EQ || op == Py_NE)) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000949 if (PyErr_WarnEx(PyExc_BytesWarning,
Georg Brandle5d68ac2008-06-04 11:30:26 +0000950 "Comparison between bytearray and string", 1))
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000951 return NULL;
952 }
953
954 Py_INCREF(Py_NotImplemented);
955 return Py_NotImplemented;
956 }
957
958 self_size = _getbuffer(self, &self_bytes);
959 if (self_size < 0) {
960 PyErr_Clear();
961 Py_INCREF(Py_NotImplemented);
962 return Py_NotImplemented;
963 }
964
965 other_size = _getbuffer(other, &other_bytes);
966 if (other_size < 0) {
967 PyErr_Clear();
Martin v. Löwis423be952008-08-13 15:53:07 +0000968 PyBuffer_Release(&self_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +0000969 Py_INCREF(Py_NotImplemented);
970 return Py_NotImplemented;
971 }
972
973 if (self_size != other_size && (op == Py_EQ || op == Py_NE)) {
974 /* Shortcut: if the lengths differ, the objects differ */
975 cmp = (op == Py_NE);
976 }
977 else {
978 minsize = self_size;
979 if (other_size < minsize)
980 minsize = other_size;
981
982 cmp = memcmp(self_bytes.buf, other_bytes.buf, minsize);
983 /* In ISO C, memcmp() guarantees to use unsigned bytes! */
984
985 if (cmp == 0) {
986 if (self_size < other_size)
987 cmp = -1;
988 else if (self_size > other_size)
989 cmp = 1;
990 }
991
992 switch (op) {
993 case Py_LT: cmp = cmp < 0; break;
994 case Py_LE: cmp = cmp <= 0; break;
995 case Py_EQ: cmp = cmp == 0; break;
996 case Py_NE: cmp = cmp != 0; break;
997 case Py_GT: cmp = cmp > 0; break;
998 case Py_GE: cmp = cmp >= 0; break;
999 }
1000 }
1001
1002 res = cmp ? Py_True : Py_False;
Martin v. Löwis423be952008-08-13 15:53:07 +00001003 PyBuffer_Release(&self_bytes);
1004 PyBuffer_Release(&other_bytes);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001005 Py_INCREF(res);
1006 return res;
1007}
1008
1009static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001010bytearray_dealloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001011{
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001012 if (self->ob_exports > 0) {
1013 PyErr_SetString(PyExc_SystemError,
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00001014 "deallocated bytearray object has exported buffers");
Benjamin Petersone0124bd2009-03-09 21:04:33 +00001015 PyErr_Print();
1016 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001017 if (self->ob_bytes != 0) {
1018 PyMem_Free(self->ob_bytes);
1019 }
1020 Py_TYPE(self)->tp_free((PyObject *)self);
1021}
1022
1023
1024/* -------------------------------------------------------------------- */
1025/* Methods */
1026
1027#define STRINGLIB_CHAR char
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001028#define STRINGLIB_LEN PyByteArray_GET_SIZE
1029#define STRINGLIB_STR PyByteArray_AS_STRING
1030#define STRINGLIB_NEW PyByteArray_FromStringAndSize
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001031#define STRINGLIB_ISSPACE Py_ISSPACE
1032#define STRINGLIB_ISLINEBREAK(x) ((x == '\n') || (x == '\r'))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001033#define STRINGLIB_CHECK_EXACT PyByteArray_CheckExact
1034#define STRINGLIB_MUTABLE 1
1035
1036#include "stringlib/fastsearch.h"
1037#include "stringlib/count.h"
1038#include "stringlib/find.h"
1039#include "stringlib/partition.h"
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001040#include "stringlib/split.h"
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001041#include "stringlib/ctype.h"
1042#include "stringlib/transmogrify.h"
1043
1044
1045/* The following Py_LOCAL_INLINE and Py_LOCAL functions
1046were copied from the old char* style string object. */
1047
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001048/* helper macro to fixup start/end slice values */
1049#define ADJUST_INDICES(start, end, len) \
1050 if (end > len) \
1051 end = len; \
1052 else if (end < 0) { \
1053 end += len; \
1054 if (end < 0) \
1055 end = 0; \
1056 } \
1057 if (start < 0) { \
1058 start += len; \
1059 if (start < 0) \
1060 start = 0; \
1061 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001062
1063Py_LOCAL_INLINE(Py_ssize_t)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001064bytearray_find_internal(PyByteArrayObject *self, PyObject *args, int dir)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001065{
1066 PyObject *subobj;
1067 Py_buffer subbuf;
1068 Py_ssize_t start=0, end=PY_SSIZE_T_MAX;
1069 Py_ssize_t res;
1070
1071 if (!PyArg_ParseTuple(args, "O|O&O&:find/rfind/index/rindex", &subobj,
1072 _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end))
1073 return -2;
1074 if (_getbuffer(subobj, &subbuf) < 0)
1075 return -2;
1076 if (dir > 0)
1077 res = stringlib_find_slice(
1078 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
1079 subbuf.buf, subbuf.len, start, end);
1080 else
1081 res = stringlib_rfind_slice(
1082 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
1083 subbuf.buf, subbuf.len, start, end);
Martin v. Löwis423be952008-08-13 15:53:07 +00001084 PyBuffer_Release(&subbuf);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001085 return res;
1086}
1087
1088PyDoc_STRVAR(find__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001089"B.find(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001090\n\
1091Return the lowest index in B where subsection sub is found,\n\
1092such that sub is contained within s[start,end]. Optional\n\
1093arguments start and end are interpreted as in slice notation.\n\
1094\n\
1095Return -1 on failure.");
1096
1097static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001098bytearray_find(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001099{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001100 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001101 if (result == -2)
1102 return NULL;
1103 return PyLong_FromSsize_t(result);
1104}
1105
1106PyDoc_STRVAR(count__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001107"B.count(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001108\n\
1109Return the number of non-overlapping occurrences of subsection sub in\n\
1110bytes B[start:end]. Optional arguments start and end are interpreted\n\
1111as in slice notation.");
1112
1113static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001114bytearray_count(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001115{
1116 PyObject *sub_obj;
1117 const char *str = PyByteArray_AS_STRING(self);
1118 Py_ssize_t start = 0, end = PY_SSIZE_T_MAX;
1119 Py_buffer vsub;
1120 PyObject *count_obj;
1121
1122 if (!PyArg_ParseTuple(args, "O|O&O&:count", &sub_obj,
1123 _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end))
1124 return NULL;
1125
1126 if (_getbuffer(sub_obj, &vsub) < 0)
1127 return NULL;
1128
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001129 ADJUST_INDICES(start, end, PyByteArray_GET_SIZE(self));
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001130
1131 count_obj = PyLong_FromSsize_t(
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001132 stringlib_count(str + start, end - start, vsub.buf, vsub.len, PY_SSIZE_T_MAX)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001133 );
Martin v. Löwis423be952008-08-13 15:53:07 +00001134 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001135 return count_obj;
1136}
1137
1138
1139PyDoc_STRVAR(index__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001140"B.index(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001141\n\
1142Like B.find() but raise ValueError when the subsection is not found.");
1143
1144static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001145bytearray_index(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001146{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001147 Py_ssize_t result = bytearray_find_internal(self, args, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001148 if (result == -2)
1149 return NULL;
1150 if (result == -1) {
1151 PyErr_SetString(PyExc_ValueError,
1152 "subsection not found");
1153 return NULL;
1154 }
1155 return PyLong_FromSsize_t(result);
1156}
1157
1158
1159PyDoc_STRVAR(rfind__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001160"B.rfind(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001161\n\
1162Return the highest index in B where subsection sub is found,\n\
1163such that sub is contained within s[start,end]. Optional\n\
1164arguments start and end are interpreted as in slice notation.\n\
1165\n\
1166Return -1 on failure.");
1167
1168static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001169bytearray_rfind(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001170{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001171 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001172 if (result == -2)
1173 return NULL;
1174 return PyLong_FromSsize_t(result);
1175}
1176
1177
1178PyDoc_STRVAR(rindex__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001179"B.rindex(sub[, start[, end]]) -> int\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001180\n\
1181Like B.rfind() but raise ValueError when the subsection is not found.");
1182
1183static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001184bytearray_rindex(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001185{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001186 Py_ssize_t result = bytearray_find_internal(self, args, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001187 if (result == -2)
1188 return NULL;
1189 if (result == -1) {
1190 PyErr_SetString(PyExc_ValueError,
1191 "subsection not found");
1192 return NULL;
1193 }
1194 return PyLong_FromSsize_t(result);
1195}
1196
1197
1198static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001199bytearray_contains(PyObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001200{
1201 Py_ssize_t ival = PyNumber_AsSsize_t(arg, PyExc_ValueError);
1202 if (ival == -1 && PyErr_Occurred()) {
1203 Py_buffer varg;
1204 int pos;
1205 PyErr_Clear();
1206 if (_getbuffer(arg, &varg) < 0)
1207 return -1;
1208 pos = stringlib_find(PyByteArray_AS_STRING(self), Py_SIZE(self),
1209 varg.buf, varg.len, 0);
Martin v. Löwis423be952008-08-13 15:53:07 +00001210 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001211 return pos >= 0;
1212 }
1213 if (ival < 0 || ival >= 256) {
1214 PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)");
1215 return -1;
1216 }
1217
1218 return memchr(PyByteArray_AS_STRING(self), ival, Py_SIZE(self)) != NULL;
1219}
1220
1221
1222/* Matches the end (direction >= 0) or start (direction < 0) of self
1223 * against substr, using the start and end arguments. Returns
1224 * -1 on error, 0 if not found and 1 if found.
1225 */
1226Py_LOCAL(int)
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001227_bytearray_tailmatch(PyByteArrayObject *self, PyObject *substr, Py_ssize_t start,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001228 Py_ssize_t end, int direction)
1229{
1230 Py_ssize_t len = PyByteArray_GET_SIZE(self);
1231 const char* str;
1232 Py_buffer vsubstr;
1233 int rv = 0;
1234
1235 str = PyByteArray_AS_STRING(self);
1236
1237 if (_getbuffer(substr, &vsubstr) < 0)
1238 return -1;
1239
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001240 ADJUST_INDICES(start, end, len);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001241
1242 if (direction < 0) {
1243 /* startswith */
1244 if (start+vsubstr.len > len) {
1245 goto done;
1246 }
1247 } else {
1248 /* endswith */
1249 if (end-start < vsubstr.len || start > len) {
1250 goto done;
1251 }
1252
1253 if (end-vsubstr.len > start)
1254 start = end - vsubstr.len;
1255 }
1256 if (end-start >= vsubstr.len)
1257 rv = ! memcmp(str+start, vsubstr.buf, vsubstr.len);
1258
1259done:
Martin v. Löwis423be952008-08-13 15:53:07 +00001260 PyBuffer_Release(&vsubstr);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001261 return rv;
1262}
1263
1264
1265PyDoc_STRVAR(startswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001266"B.startswith(prefix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001267\n\
1268Return True if B starts with the specified prefix, False otherwise.\n\
1269With optional start, test B beginning at that position.\n\
1270With optional end, stop comparing B at that position.\n\
1271prefix can also be a tuple of strings to try.");
1272
1273static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001274bytearray_startswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001275{
1276 Py_ssize_t start = 0;
1277 Py_ssize_t end = PY_SSIZE_T_MAX;
1278 PyObject *subobj;
1279 int result;
1280
1281 if (!PyArg_ParseTuple(args, "O|O&O&:startswith", &subobj,
1282 _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end))
1283 return NULL;
1284 if (PyTuple_Check(subobj)) {
1285 Py_ssize_t i;
1286 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001287 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001288 PyTuple_GET_ITEM(subobj, i),
1289 start, end, -1);
1290 if (result == -1)
1291 return NULL;
1292 else if (result) {
1293 Py_RETURN_TRUE;
1294 }
1295 }
1296 Py_RETURN_FALSE;
1297 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001298 result = _bytearray_tailmatch(self, subobj, start, end, -1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001299 if (result == -1)
1300 return NULL;
1301 else
1302 return PyBool_FromLong(result);
1303}
1304
1305PyDoc_STRVAR(endswith__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001306"B.endswith(suffix[, start[, end]]) -> bool\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001307\n\
1308Return True if B ends with the specified suffix, False otherwise.\n\
1309With optional start, test B beginning at that position.\n\
1310With optional end, stop comparing B at that position.\n\
1311suffix can also be a tuple of strings to try.");
1312
1313static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001314bytearray_endswith(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001315{
1316 Py_ssize_t start = 0;
1317 Py_ssize_t end = PY_SSIZE_T_MAX;
1318 PyObject *subobj;
1319 int result;
1320
1321 if (!PyArg_ParseTuple(args, "O|O&O&:endswith", &subobj,
1322 _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end))
1323 return NULL;
1324 if (PyTuple_Check(subobj)) {
1325 Py_ssize_t i;
1326 for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001327 result = _bytearray_tailmatch(self,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001328 PyTuple_GET_ITEM(subobj, i),
1329 start, end, +1);
1330 if (result == -1)
1331 return NULL;
1332 else if (result) {
1333 Py_RETURN_TRUE;
1334 }
1335 }
1336 Py_RETURN_FALSE;
1337 }
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001338 result = _bytearray_tailmatch(self, subobj, start, end, +1);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001339 if (result == -1)
1340 return NULL;
1341 else
1342 return PyBool_FromLong(result);
1343}
1344
1345
1346PyDoc_STRVAR(translate__doc__,
1347"B.translate(table[, deletechars]) -> bytearray\n\
1348\n\
1349Return a copy of B, where all characters occurring in the\n\
1350optional argument deletechars are removed, and the remaining\n\
1351characters have been mapped through the given translation\n\
1352table, which must be a bytes object of length 256.");
1353
1354static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001355bytearray_translate(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001356{
1357 register char *input, *output;
1358 register const char *table;
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001359 register Py_ssize_t i, c;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001360 PyObject *input_obj = (PyObject*)self;
1361 const char *output_start;
1362 Py_ssize_t inlen;
Georg Brandlccc47b62008-12-28 11:44:14 +00001363 PyObject *result = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001364 int trans_table[256];
Georg Brandlccc47b62008-12-28 11:44:14 +00001365 PyObject *tableobj = NULL, *delobj = NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001366 Py_buffer vtable, vdel;
1367
1368 if (!PyArg_UnpackTuple(args, "translate", 1, 2,
1369 &tableobj, &delobj))
1370 return NULL;
1371
Georg Brandlccc47b62008-12-28 11:44:14 +00001372 if (tableobj == Py_None) {
1373 table = NULL;
1374 tableobj = NULL;
1375 } else if (_getbuffer(tableobj, &vtable) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001376 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001377 } else {
1378 if (vtable.len != 256) {
1379 PyErr_SetString(PyExc_ValueError,
1380 "translation table must be 256 characters long");
Georg Brandl953152f2009-07-22 12:03:59 +00001381 PyBuffer_Release(&vtable);
1382 return NULL;
Georg Brandlccc47b62008-12-28 11:44:14 +00001383 }
1384 table = (const char*)vtable.buf;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001385 }
1386
1387 if (delobj != NULL) {
1388 if (_getbuffer(delobj, &vdel) < 0) {
Georg Brandl953152f2009-07-22 12:03:59 +00001389 if (tableobj != NULL)
1390 PyBuffer_Release(&vtable);
1391 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001392 }
1393 }
1394 else {
1395 vdel.buf = NULL;
1396 vdel.len = 0;
1397 }
1398
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001399 inlen = PyByteArray_GET_SIZE(input_obj);
1400 result = PyByteArray_FromStringAndSize((char *)NULL, inlen);
1401 if (result == NULL)
1402 goto done;
1403 output_start = output = PyByteArray_AsString(result);
1404 input = PyByteArray_AS_STRING(input_obj);
1405
Georg Brandlccc47b62008-12-28 11:44:14 +00001406 if (vdel.len == 0 && table != NULL) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001407 /* If no deletions are required, use faster code */
1408 for (i = inlen; --i >= 0; ) {
1409 c = Py_CHARMASK(*input++);
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001410 *output++ = table[c];
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001411 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001412 goto done;
1413 }
Georg Brandlccc47b62008-12-28 11:44:14 +00001414
1415 if (table == NULL) {
1416 for (i = 0; i < 256; i++)
1417 trans_table[i] = Py_CHARMASK(i);
1418 } else {
1419 for (i = 0; i < 256; i++)
1420 trans_table[i] = Py_CHARMASK(table[i]);
1421 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001422
1423 for (i = 0; i < vdel.len; i++)
1424 trans_table[(int) Py_CHARMASK( ((unsigned char*)vdel.buf)[i] )] = -1;
1425
1426 for (i = inlen; --i >= 0; ) {
1427 c = Py_CHARMASK(*input++);
1428 if (trans_table[c] != -1)
1429 if (Py_CHARMASK(*output++ = (char)trans_table[c]) == c)
1430 continue;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001431 }
1432 /* Fix the size of the resulting string */
1433 if (inlen > 0)
1434 PyByteArray_Resize(result, output - output_start);
1435
1436done:
Georg Brandlccc47b62008-12-28 11:44:14 +00001437 if (tableobj != NULL)
1438 PyBuffer_Release(&vtable);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001439 if (delobj != NULL)
Martin v. Löwis423be952008-08-13 15:53:07 +00001440 PyBuffer_Release(&vdel);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001441 return result;
1442}
1443
1444
Georg Brandlabc38772009-04-12 15:51:51 +00001445static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001446bytearray_maketrans(PyObject *null, PyObject *args)
Georg Brandlabc38772009-04-12 15:51:51 +00001447{
1448 return _Py_bytes_maketrans(args);
1449}
1450
1451
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001452/* find and count characters and substrings */
1453
1454#define findchar(target, target_len, c) \
1455 ((char *)memchr((const void *)(target), c, target_len))
1456
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001457
Benjamin Peterson0f3641c2008-11-19 22:05:52 +00001458/* Bytes ops must return a string, create a copy */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001459Py_LOCAL(PyByteArrayObject *)
1460return_self(PyByteArrayObject *self)
1461{
Georg Brandl1e7217d2008-05-30 12:02:38 +00001462 /* always return a new bytearray */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001463 return (PyByteArrayObject *)PyByteArray_FromStringAndSize(
1464 PyByteArray_AS_STRING(self),
1465 PyByteArray_GET_SIZE(self));
1466}
1467
1468Py_LOCAL_INLINE(Py_ssize_t)
1469countchar(const char *target, Py_ssize_t target_len, char c, Py_ssize_t maxcount)
1470{
1471 Py_ssize_t count=0;
1472 const char *start=target;
1473 const char *end=target+target_len;
1474
1475 while ( (start=findchar(start, end-start, c)) != NULL ) {
1476 count++;
1477 if (count >= maxcount)
1478 break;
1479 start += 1;
1480 }
1481 return count;
1482}
1483
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001484
1485/* Algorithms for different cases of string replacement */
1486
1487/* len(self)>=1, from="", len(to)>=1, maxcount>=1 */
1488Py_LOCAL(PyByteArrayObject *)
1489replace_interleave(PyByteArrayObject *self,
1490 const char *to_s, Py_ssize_t to_len,
1491 Py_ssize_t maxcount)
1492{
1493 char *self_s, *result_s;
1494 Py_ssize_t self_len, result_len;
1495 Py_ssize_t count, i, product;
1496 PyByteArrayObject *result;
1497
1498 self_len = PyByteArray_GET_SIZE(self);
1499
1500 /* 1 at the end plus 1 after every character */
1501 count = self_len+1;
1502 if (maxcount < count)
1503 count = maxcount;
1504
1505 /* Check for overflow */
1506 /* result_len = count * to_len + self_len; */
1507 product = count * to_len;
1508 if (product / to_len != count) {
1509 PyErr_SetString(PyExc_OverflowError,
1510 "replace string is too long");
1511 return NULL;
1512 }
1513 result_len = product + self_len;
1514 if (result_len < 0) {
1515 PyErr_SetString(PyExc_OverflowError,
1516 "replace string is too long");
1517 return NULL;
1518 }
1519
1520 if (! (result = (PyByteArrayObject *)
1521 PyByteArray_FromStringAndSize(NULL, result_len)) )
1522 return NULL;
1523
1524 self_s = PyByteArray_AS_STRING(self);
1525 result_s = PyByteArray_AS_STRING(result);
1526
1527 /* TODO: special case single character, which doesn't need memcpy */
1528
1529 /* Lay the first one down (guaranteed this will occur) */
1530 Py_MEMCPY(result_s, to_s, to_len);
1531 result_s += to_len;
1532 count -= 1;
1533
1534 for (i=0; i<count; i++) {
1535 *result_s++ = *self_s++;
1536 Py_MEMCPY(result_s, to_s, to_len);
1537 result_s += to_len;
1538 }
1539
1540 /* Copy the rest of the original string */
1541 Py_MEMCPY(result_s, self_s, self_len-i);
1542
1543 return result;
1544}
1545
1546/* Special case for deleting a single character */
1547/* len(self)>=1, len(from)==1, to="", maxcount>=1 */
1548Py_LOCAL(PyByteArrayObject *)
1549replace_delete_single_character(PyByteArrayObject *self,
1550 char from_c, Py_ssize_t maxcount)
1551{
1552 char *self_s, *result_s;
1553 char *start, *next, *end;
1554 Py_ssize_t self_len, result_len;
1555 Py_ssize_t count;
1556 PyByteArrayObject *result;
1557
1558 self_len = PyByteArray_GET_SIZE(self);
1559 self_s = PyByteArray_AS_STRING(self);
1560
1561 count = countchar(self_s, self_len, from_c, maxcount);
1562 if (count == 0) {
1563 return return_self(self);
1564 }
1565
1566 result_len = self_len - count; /* from_len == 1 */
1567 assert(result_len>=0);
1568
1569 if ( (result = (PyByteArrayObject *)
1570 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1571 return NULL;
1572 result_s = PyByteArray_AS_STRING(result);
1573
1574 start = self_s;
1575 end = self_s + self_len;
1576 while (count-- > 0) {
1577 next = findchar(start, end-start, from_c);
1578 if (next == NULL)
1579 break;
1580 Py_MEMCPY(result_s, start, next-start);
1581 result_s += (next-start);
1582 start = next+1;
1583 }
1584 Py_MEMCPY(result_s, start, end-start);
1585
1586 return result;
1587}
1588
1589/* len(self)>=1, len(from)>=2, to="", maxcount>=1 */
1590
1591Py_LOCAL(PyByteArrayObject *)
1592replace_delete_substring(PyByteArrayObject *self,
1593 const char *from_s, Py_ssize_t from_len,
1594 Py_ssize_t maxcount)
1595{
1596 char *self_s, *result_s;
1597 char *start, *next, *end;
1598 Py_ssize_t self_len, result_len;
1599 Py_ssize_t count, offset;
1600 PyByteArrayObject *result;
1601
1602 self_len = PyByteArray_GET_SIZE(self);
1603 self_s = PyByteArray_AS_STRING(self);
1604
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001605 count = stringlib_count(self_s, self_len,
1606 from_s, from_len,
1607 maxcount);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001608
1609 if (count == 0) {
1610 /* no matches */
1611 return return_self(self);
1612 }
1613
1614 result_len = self_len - (count * from_len);
1615 assert (result_len>=0);
1616
1617 if ( (result = (PyByteArrayObject *)
1618 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL )
1619 return NULL;
1620
1621 result_s = PyByteArray_AS_STRING(result);
1622
1623 start = self_s;
1624 end = self_s + self_len;
1625 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001626 offset = stringlib_find(start, end-start,
1627 from_s, from_len,
1628 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001629 if (offset == -1)
1630 break;
1631 next = start + offset;
1632
1633 Py_MEMCPY(result_s, start, next-start);
1634
1635 result_s += (next-start);
1636 start = next+from_len;
1637 }
1638 Py_MEMCPY(result_s, start, end-start);
1639 return result;
1640}
1641
1642/* len(self)>=1, len(from)==len(to)==1, maxcount>=1 */
1643Py_LOCAL(PyByteArrayObject *)
1644replace_single_character_in_place(PyByteArrayObject *self,
1645 char from_c, char to_c,
1646 Py_ssize_t maxcount)
1647{
1648 char *self_s, *result_s, *start, *end, *next;
1649 Py_ssize_t self_len;
1650 PyByteArrayObject *result;
1651
1652 /* The result string will be the same size */
1653 self_s = PyByteArray_AS_STRING(self);
1654 self_len = PyByteArray_GET_SIZE(self);
1655
1656 next = findchar(self_s, self_len, from_c);
1657
1658 if (next == NULL) {
1659 /* No matches; return the original bytes */
1660 return return_self(self);
1661 }
1662
1663 /* Need to make a new bytes */
1664 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1665 if (result == NULL)
1666 return NULL;
1667 result_s = PyByteArray_AS_STRING(result);
1668 Py_MEMCPY(result_s, self_s, self_len);
1669
1670 /* change everything in-place, starting with this one */
1671 start = result_s + (next-self_s);
1672 *start = to_c;
1673 start++;
1674 end = result_s + self_len;
1675
1676 while (--maxcount > 0) {
1677 next = findchar(start, end-start, from_c);
1678 if (next == NULL)
1679 break;
1680 *next = to_c;
1681 start = next+1;
1682 }
1683
1684 return result;
1685}
1686
1687/* len(self)>=1, len(from)==len(to)>=2, maxcount>=1 */
1688Py_LOCAL(PyByteArrayObject *)
1689replace_substring_in_place(PyByteArrayObject *self,
1690 const char *from_s, Py_ssize_t from_len,
1691 const char *to_s, Py_ssize_t to_len,
1692 Py_ssize_t maxcount)
1693{
1694 char *result_s, *start, *end;
1695 char *self_s;
1696 Py_ssize_t self_len, offset;
1697 PyByteArrayObject *result;
1698
1699 /* The result bytes will be the same size */
1700
1701 self_s = PyByteArray_AS_STRING(self);
1702 self_len = PyByteArray_GET_SIZE(self);
1703
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001704 offset = stringlib_find(self_s, self_len,
1705 from_s, from_len,
1706 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001707 if (offset == -1) {
1708 /* No matches; return the original bytes */
1709 return return_self(self);
1710 }
1711
1712 /* Need to make a new bytes */
1713 result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len);
1714 if (result == NULL)
1715 return NULL;
1716 result_s = PyByteArray_AS_STRING(result);
1717 Py_MEMCPY(result_s, self_s, self_len);
1718
1719 /* change everything in-place, starting with this one */
1720 start = result_s + offset;
1721 Py_MEMCPY(start, to_s, from_len);
1722 start += from_len;
1723 end = result_s + self_len;
1724
1725 while ( --maxcount > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001726 offset = stringlib_find(start, end-start,
1727 from_s, from_len,
1728 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001729 if (offset==-1)
1730 break;
1731 Py_MEMCPY(start+offset, to_s, from_len);
1732 start += offset+from_len;
1733 }
1734
1735 return result;
1736}
1737
1738/* len(self)>=1, len(from)==1, len(to)>=2, maxcount>=1 */
1739Py_LOCAL(PyByteArrayObject *)
1740replace_single_character(PyByteArrayObject *self,
1741 char from_c,
1742 const char *to_s, Py_ssize_t to_len,
1743 Py_ssize_t maxcount)
1744{
1745 char *self_s, *result_s;
1746 char *start, *next, *end;
1747 Py_ssize_t self_len, result_len;
1748 Py_ssize_t count, product;
1749 PyByteArrayObject *result;
1750
1751 self_s = PyByteArray_AS_STRING(self);
1752 self_len = PyByteArray_GET_SIZE(self);
1753
1754 count = countchar(self_s, self_len, from_c, maxcount);
1755 if (count == 0) {
1756 /* no matches, return unchanged */
1757 return return_self(self);
1758 }
1759
1760 /* use the difference between current and new, hence the "-1" */
1761 /* result_len = self_len + count * (to_len-1) */
1762 product = count * (to_len-1);
1763 if (product / (to_len-1) != count) {
1764 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1765 return NULL;
1766 }
1767 result_len = self_len + product;
1768 if (result_len < 0) {
1769 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1770 return NULL;
1771 }
1772
1773 if ( (result = (PyByteArrayObject *)
1774 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1775 return NULL;
1776 result_s = PyByteArray_AS_STRING(result);
1777
1778 start = self_s;
1779 end = self_s + self_len;
1780 while (count-- > 0) {
1781 next = findchar(start, end-start, from_c);
1782 if (next == NULL)
1783 break;
1784
1785 if (next == start) {
1786 /* replace with the 'to' */
1787 Py_MEMCPY(result_s, to_s, to_len);
1788 result_s += to_len;
1789 start += 1;
1790 } else {
1791 /* copy the unchanged old then the 'to' */
1792 Py_MEMCPY(result_s, start, next-start);
1793 result_s += (next-start);
1794 Py_MEMCPY(result_s, to_s, to_len);
1795 result_s += to_len;
1796 start = next+1;
1797 }
1798 }
1799 /* Copy the remainder of the remaining bytes */
1800 Py_MEMCPY(result_s, start, end-start);
1801
1802 return result;
1803}
1804
1805/* len(self)>=1, len(from)>=2, len(to)>=2, maxcount>=1 */
1806Py_LOCAL(PyByteArrayObject *)
1807replace_substring(PyByteArrayObject *self,
1808 const char *from_s, Py_ssize_t from_len,
1809 const char *to_s, Py_ssize_t to_len,
1810 Py_ssize_t maxcount)
1811{
1812 char *self_s, *result_s;
1813 char *start, *next, *end;
1814 Py_ssize_t self_len, result_len;
1815 Py_ssize_t count, offset, product;
1816 PyByteArrayObject *result;
1817
1818 self_s = PyByteArray_AS_STRING(self);
1819 self_len = PyByteArray_GET_SIZE(self);
1820
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001821 count = stringlib_count(self_s, self_len,
1822 from_s, from_len,
1823 maxcount);
1824
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001825 if (count == 0) {
1826 /* no matches, return unchanged */
1827 return return_self(self);
1828 }
1829
1830 /* Check for overflow */
1831 /* result_len = self_len + count * (to_len-from_len) */
1832 product = count * (to_len-from_len);
1833 if (product / (to_len-from_len) != count) {
1834 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1835 return NULL;
1836 }
1837 result_len = self_len + product;
1838 if (result_len < 0) {
1839 PyErr_SetString(PyExc_OverflowError, "replace bytes is too long");
1840 return NULL;
1841 }
1842
1843 if ( (result = (PyByteArrayObject *)
1844 PyByteArray_FromStringAndSize(NULL, result_len)) == NULL)
1845 return NULL;
1846 result_s = PyByteArray_AS_STRING(result);
1847
1848 start = self_s;
1849 end = self_s + self_len;
1850 while (count-- > 0) {
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001851 offset = stringlib_find(start, end-start,
1852 from_s, from_len,
1853 0);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001854 if (offset == -1)
1855 break;
1856 next = start+offset;
1857 if (next == start) {
1858 /* replace with the 'to' */
1859 Py_MEMCPY(result_s, to_s, to_len);
1860 result_s += to_len;
1861 start += from_len;
1862 } else {
1863 /* copy the unchanged old then the 'to' */
1864 Py_MEMCPY(result_s, start, next-start);
1865 result_s += (next-start);
1866 Py_MEMCPY(result_s, to_s, to_len);
1867 result_s += to_len;
1868 start = next+from_len;
1869 }
1870 }
1871 /* Copy the remainder of the remaining bytes */
1872 Py_MEMCPY(result_s, start, end-start);
1873
1874 return result;
1875}
1876
1877
1878Py_LOCAL(PyByteArrayObject *)
1879replace(PyByteArrayObject *self,
1880 const char *from_s, Py_ssize_t from_len,
1881 const char *to_s, Py_ssize_t to_len,
1882 Py_ssize_t maxcount)
1883{
1884 if (maxcount < 0) {
1885 maxcount = PY_SSIZE_T_MAX;
1886 } else if (maxcount == 0 || PyByteArray_GET_SIZE(self) == 0) {
1887 /* nothing to do; return the original bytes */
1888 return return_self(self);
1889 }
1890
1891 if (maxcount == 0 ||
1892 (from_len == 0 && to_len == 0)) {
1893 /* nothing to do; return the original bytes */
1894 return return_self(self);
1895 }
1896
1897 /* Handle zero-length special cases */
1898
1899 if (from_len == 0) {
1900 /* insert the 'to' bytes everywhere. */
1901 /* >>> "Python".replace("", ".") */
1902 /* '.P.y.t.h.o.n.' */
1903 return replace_interleave(self, to_s, to_len, maxcount);
1904 }
1905
1906 /* Except for "".replace("", "A") == "A" there is no way beyond this */
1907 /* point for an empty self bytes to generate a non-empty bytes */
1908 /* Special case so the remaining code always gets a non-empty bytes */
1909 if (PyByteArray_GET_SIZE(self) == 0) {
1910 return return_self(self);
1911 }
1912
1913 if (to_len == 0) {
Georg Brandl17cb8a82008-05-30 08:20:09 +00001914 /* delete all occurrences of 'from' bytes */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001915 if (from_len == 1) {
1916 return replace_delete_single_character(
1917 self, from_s[0], maxcount);
1918 } else {
1919 return replace_delete_substring(self, from_s, from_len, maxcount);
1920 }
1921 }
1922
1923 /* Handle special case where both bytes have the same length */
1924
1925 if (from_len == to_len) {
1926 if (from_len == 1) {
1927 return replace_single_character_in_place(
1928 self,
1929 from_s[0],
1930 to_s[0],
1931 maxcount);
1932 } else {
1933 return replace_substring_in_place(
1934 self, from_s, from_len, to_s, to_len, maxcount);
1935 }
1936 }
1937
1938 /* Otherwise use the more generic algorithms */
1939 if (from_len == 1) {
1940 return replace_single_character(self, from_s[0],
1941 to_s, to_len, maxcount);
1942 } else {
1943 /* len('from')>=2, len('to')>=1 */
1944 return replace_substring(self, from_s, from_len, to_s, to_len, maxcount);
1945 }
1946}
1947
1948
1949PyDoc_STRVAR(replace__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001950"B.replace(old, new[, count]) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001951\n\
1952Return a copy of B with all occurrences of subsection\n\
1953old replaced by new. If the optional argument count is\n\
1954given, only the first count occurrences are replaced.");
1955
1956static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001957bytearray_replace(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001958{
1959 Py_ssize_t count = -1;
1960 PyObject *from, *to, *res;
1961 Py_buffer vfrom, vto;
1962
1963 if (!PyArg_ParseTuple(args, "OO|n:replace", &from, &to, &count))
1964 return NULL;
1965
1966 if (_getbuffer(from, &vfrom) < 0)
1967 return NULL;
1968 if (_getbuffer(to, &vto) < 0) {
Martin v. Löwis423be952008-08-13 15:53:07 +00001969 PyBuffer_Release(&vfrom);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001970 return NULL;
1971 }
1972
1973 res = (PyObject *)replace((PyByteArrayObject *) self,
1974 vfrom.buf, vfrom.len,
1975 vto.buf, vto.len, count);
1976
Martin v. Löwis423be952008-08-13 15:53:07 +00001977 PyBuffer_Release(&vfrom);
1978 PyBuffer_Release(&vto);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001979 return res;
1980}
1981
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001982PyDoc_STRVAR(split__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00001983"B.split([sep[, maxsplit]]) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001984\n\
1985Return a list of the sections in B, using sep as the delimiter.\n\
1986If sep is not given, B is split on ASCII whitespace characters\n\
1987(space, tab, return, newline, formfeed, vertical tab).\n\
1988If maxsplit is given, at most maxsplit splits are done.");
1989
1990static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00001991bytearray_split(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001992{
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001993 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
1994 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001995 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00001996 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001997 Py_buffer vsub;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00001998
1999 if (!PyArg_ParseTuple(args, "|On:split", &subobj, &maxsplit))
2000 return NULL;
2001 if (maxsplit < 0)
2002 maxsplit = PY_SSIZE_T_MAX;
2003
2004 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002005 return stringlib_split_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002006
2007 if (_getbuffer(subobj, &vsub) < 0)
2008 return NULL;
2009 sub = vsub.buf;
2010 n = vsub.len;
2011
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002012 list = stringlib_split(
2013 (PyObject*) self, s, len, sub, n, maxsplit
2014 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002015 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002016 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002017}
2018
2019PyDoc_STRVAR(partition__doc__,
2020"B.partition(sep) -> (head, sep, tail)\n\
2021\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002022Search for the separator sep in B, and return the part before it,\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002023the separator itself, and the part after it. If the separator is not\n\
2024found, returns B and two empty bytearray objects.");
2025
2026static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002027bytearray_partition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002028{
2029 PyObject *bytesep, *result;
2030
2031 bytesep = PyByteArray_FromObject(sep_obj);
2032 if (! bytesep)
2033 return NULL;
2034
2035 result = stringlib_partition(
2036 (PyObject*) self,
2037 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2038 bytesep,
2039 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2040 );
2041
2042 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002043 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002044}
2045
2046PyDoc_STRVAR(rpartition__doc__,
2047"B.rpartition(sep) -> (tail, sep, head)\n\
2048\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002049Search for the separator sep in B, starting at the end of B,\n\
2050and return the part before it, the separator itself, and the\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002051part after it. If the separator is not found, returns two empty\n\
2052bytearray objects and B.");
2053
2054static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002055bytearray_rpartition(PyByteArrayObject *self, PyObject *sep_obj)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002056{
2057 PyObject *bytesep, *result;
2058
2059 bytesep = PyByteArray_FromObject(sep_obj);
2060 if (! bytesep)
2061 return NULL;
2062
2063 result = stringlib_rpartition(
2064 (PyObject*) self,
2065 PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
2066 bytesep,
2067 PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep)
2068 );
2069
2070 Py_DECREF(bytesep);
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002071 return result;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002072}
2073
2074PyDoc_STRVAR(rsplit__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002075"B.rsplit(sep[, maxsplit]) -> list of bytearrays\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002076\n\
2077Return a list of the sections in B, using sep as the delimiter,\n\
2078starting at the end of B and working to the front.\n\
2079If sep is not given, B is split on ASCII whitespace characters\n\
2080(space, tab, return, newline, formfeed, vertical tab).\n\
2081If maxsplit is given, at most maxsplit splits are done.");
2082
2083static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002084bytearray_rsplit(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002085{
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002086 Py_ssize_t len = PyByteArray_GET_SIZE(self), n;
2087 Py_ssize_t maxsplit = -1;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002088 const char *s = PyByteArray_AS_STRING(self), *sub;
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002089 PyObject *list, *subobj = Py_None;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002090 Py_buffer vsub;
2091
2092 if (!PyArg_ParseTuple(args, "|On:rsplit", &subobj, &maxsplit))
2093 return NULL;
2094 if (maxsplit < 0)
2095 maxsplit = PY_SSIZE_T_MAX;
2096
2097 if (subobj == Py_None)
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002098 return stringlib_rsplit_whitespace((PyObject*) self, s, len, maxsplit);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002099
2100 if (_getbuffer(subobj, &vsub) < 0)
2101 return NULL;
2102 sub = vsub.buf;
2103 n = vsub.len;
2104
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002105 list = stringlib_rsplit(
2106 (PyObject*) self, s, len, sub, n, maxsplit
2107 );
Martin v. Löwis423be952008-08-13 15:53:07 +00002108 PyBuffer_Release(&vsub);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002109 return list;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002110}
2111
2112PyDoc_STRVAR(reverse__doc__,
2113"B.reverse() -> None\n\
2114\n\
2115Reverse the order of the values in B in place.");
2116static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002117bytearray_reverse(PyByteArrayObject *self, PyObject *unused)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002118{
2119 char swap, *head, *tail;
2120 Py_ssize_t i, j, n = Py_SIZE(self);
2121
2122 j = n / 2;
2123 head = self->ob_bytes;
2124 tail = head + n - 1;
2125 for (i = 0; i < j; i++) {
2126 swap = *head;
2127 *head++ = *tail;
2128 *tail-- = swap;
2129 }
2130
2131 Py_RETURN_NONE;
2132}
2133
2134PyDoc_STRVAR(insert__doc__,
2135"B.insert(index, int) -> None\n\
2136\n\
2137Insert a single item into the bytearray before the given index.");
2138static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002139bytearray_insert(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002140{
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002141 PyObject *value;
2142 int ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002143 Py_ssize_t where, n = Py_SIZE(self);
2144
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002145 if (!PyArg_ParseTuple(args, "nO:insert", &where, &value))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002146 return NULL;
2147
2148 if (n == PY_SSIZE_T_MAX) {
2149 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002150 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002151 return NULL;
2152 }
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002153 if (!_getbytevalue(value, &ival))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002154 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002155 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2156 return NULL;
2157
2158 if (where < 0) {
2159 where += n;
2160 if (where < 0)
2161 where = 0;
2162 }
2163 if (where > n)
2164 where = n;
2165 memmove(self->ob_bytes + where + 1, self->ob_bytes + where, n - where);
Georg Brandl9a54d7c2008-07-16 23:15:30 +00002166 self->ob_bytes[where] = ival;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002167
2168 Py_RETURN_NONE;
2169}
2170
2171PyDoc_STRVAR(append__doc__,
2172"B.append(int) -> None\n\
2173\n\
2174Append a single item to the end of B.");
2175static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002176bytearray_append(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002177{
2178 int value;
2179 Py_ssize_t n = Py_SIZE(self);
2180
2181 if (! _getbytevalue(arg, &value))
2182 return NULL;
2183 if (n == PY_SSIZE_T_MAX) {
2184 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002185 "cannot add more objects to bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002186 return NULL;
2187 }
2188 if (PyByteArray_Resize((PyObject *)self, n + 1) < 0)
2189 return NULL;
2190
2191 self->ob_bytes[n] = value;
2192
2193 Py_RETURN_NONE;
2194}
2195
2196PyDoc_STRVAR(extend__doc__,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002197"B.extend(iterable_of_ints) -> None\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002198\n\
2199Append all the elements from the iterator or sequence to the\n\
2200end of B.");
2201static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002202bytearray_extend(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002203{
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002204 PyObject *it, *item, *bytearray_obj;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002205 Py_ssize_t buf_size = 0, len = 0;
2206 int value;
2207 char *buf;
2208
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002209 /* bytearray_setslice code only accepts something supporting PEP 3118. */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002210 if (PyObject_CheckBuffer(arg)) {
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002211 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), arg) == -1)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002212 return NULL;
2213
2214 Py_RETURN_NONE;
2215 }
2216
2217 it = PyObject_GetIter(arg);
2218 if (it == NULL)
2219 return NULL;
2220
2221 /* Try to determine the length of the argument. 32 is abitrary. */
2222 buf_size = _PyObject_LengthHint(arg, 32);
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002223 if (buf_size == -1) {
2224 Py_DECREF(it);
2225 return NULL;
2226 }
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002227
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002228 bytearray_obj = PyByteArray_FromStringAndSize(NULL, buf_size);
2229 if (bytearray_obj == NULL)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002230 return NULL;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002231 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002232
2233 while ((item = PyIter_Next(it)) != NULL) {
2234 if (! _getbytevalue(item, &value)) {
2235 Py_DECREF(item);
2236 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002237 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002238 return NULL;
2239 }
2240 buf[len++] = value;
2241 Py_DECREF(item);
2242
2243 if (len >= buf_size) {
2244 buf_size = len + (len >> 1) + 1;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002245 if (PyByteArray_Resize((PyObject *)bytearray_obj, buf_size) < 0) {
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002246 Py_DECREF(it);
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002247 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002248 return NULL;
2249 }
2250 /* Recompute the `buf' pointer, since the resizing operation may
2251 have invalidated it. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002252 buf = PyByteArray_AS_STRING(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002253 }
2254 }
2255 Py_DECREF(it);
2256
2257 /* Resize down to exact size. */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002258 if (PyByteArray_Resize((PyObject *)bytearray_obj, len) < 0) {
2259 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002260 return NULL;
2261 }
2262
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002263 if (bytearray_setslice(self, Py_SIZE(self), Py_SIZE(self), bytearray_obj) == -1)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002264 return NULL;
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002265 Py_DECREF(bytearray_obj);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002266
2267 Py_RETURN_NONE;
2268}
2269
2270PyDoc_STRVAR(pop__doc__,
2271"B.pop([index]) -> int\n\
2272\n\
2273Remove and return a single item from B. If no index\n\
Benjamin Petersondcf97b92008-07-02 17:30:14 +00002274argument is given, will pop the last value.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002275static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002276bytearray_pop(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002277{
2278 int value;
2279 Py_ssize_t where = -1, n = Py_SIZE(self);
2280
2281 if (!PyArg_ParseTuple(args, "|n:pop", &where))
2282 return NULL;
2283
2284 if (n == 0) {
2285 PyErr_SetString(PyExc_OverflowError,
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002286 "cannot pop an empty bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002287 return NULL;
2288 }
2289 if (where < 0)
2290 where += Py_SIZE(self);
2291 if (where < 0 || where >= Py_SIZE(self)) {
2292 PyErr_SetString(PyExc_IndexError, "pop index out of range");
2293 return NULL;
2294 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002295 if (!_canresize(self))
2296 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002297
2298 value = self->ob_bytes[where];
2299 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2300 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2301 return NULL;
2302
Mark Dickinson54a3db92009-09-06 10:19:23 +00002303 return PyLong_FromLong((unsigned char)value);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002304}
2305
2306PyDoc_STRVAR(remove__doc__,
2307"B.remove(int) -> None\n\
2308\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002309Remove the first occurrence of a value in B.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002310static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002311bytearray_remove(PyByteArrayObject *self, PyObject *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002312{
2313 int value;
2314 Py_ssize_t where, n = Py_SIZE(self);
2315
2316 if (! _getbytevalue(arg, &value))
2317 return NULL;
2318
2319 for (where = 0; where < n; where++) {
2320 if (self->ob_bytes[where] == value)
2321 break;
2322 }
2323 if (where == n) {
Mark Dickinson2b6705f2009-09-06 10:34:47 +00002324 PyErr_SetString(PyExc_ValueError, "value not found in bytearray");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002325 return NULL;
2326 }
Antoine Pitrou5504e892008-12-06 21:27:53 +00002327 if (!_canresize(self))
2328 return NULL;
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002329
2330 memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where);
2331 if (PyByteArray_Resize((PyObject *)self, n - 1) < 0)
2332 return NULL;
2333
2334 Py_RETURN_NONE;
2335}
2336
2337/* XXX These two helpers could be optimized if argsize == 1 */
2338
2339static Py_ssize_t
2340lstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2341 void *argptr, Py_ssize_t argsize)
2342{
2343 Py_ssize_t i = 0;
2344 while (i < mysize && memchr(argptr, myptr[i], argsize))
2345 i++;
2346 return i;
2347}
2348
2349static Py_ssize_t
2350rstrip_helper(unsigned char *myptr, Py_ssize_t mysize,
2351 void *argptr, Py_ssize_t argsize)
2352{
2353 Py_ssize_t i = mysize - 1;
2354 while (i >= 0 && memchr(argptr, myptr[i], argsize))
2355 i--;
2356 return i + 1;
2357}
2358
2359PyDoc_STRVAR(strip__doc__,
2360"B.strip([bytes]) -> bytearray\n\
2361\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002362Strip leading and trailing bytes contained in the argument\n\
2363and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002364If the argument is omitted, strip ASCII whitespace.");
2365static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002366bytearray_strip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002367{
2368 Py_ssize_t left, right, mysize, argsize;
2369 void *myptr, *argptr;
2370 PyObject *arg = Py_None;
2371 Py_buffer varg;
2372 if (!PyArg_ParseTuple(args, "|O:strip", &arg))
2373 return NULL;
2374 if (arg == Py_None) {
2375 argptr = "\t\n\r\f\v ";
2376 argsize = 6;
2377 }
2378 else {
2379 if (_getbuffer(arg, &varg) < 0)
2380 return NULL;
2381 argptr = varg.buf;
2382 argsize = varg.len;
2383 }
2384 myptr = self->ob_bytes;
2385 mysize = Py_SIZE(self);
2386 left = lstrip_helper(myptr, mysize, argptr, argsize);
2387 if (left == mysize)
2388 right = left;
2389 else
2390 right = rstrip_helper(myptr, mysize, argptr, argsize);
2391 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002392 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002393 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2394}
2395
2396PyDoc_STRVAR(lstrip__doc__,
2397"B.lstrip([bytes]) -> bytearray\n\
2398\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002399Strip leading bytes contained in the argument\n\
2400and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002401If the argument is omitted, strip leading ASCII whitespace.");
2402static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002403bytearray_lstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002404{
2405 Py_ssize_t left, right, mysize, argsize;
2406 void *myptr, *argptr;
2407 PyObject *arg = Py_None;
2408 Py_buffer varg;
2409 if (!PyArg_ParseTuple(args, "|O:lstrip", &arg))
2410 return NULL;
2411 if (arg == Py_None) {
2412 argptr = "\t\n\r\f\v ";
2413 argsize = 6;
2414 }
2415 else {
2416 if (_getbuffer(arg, &varg) < 0)
2417 return NULL;
2418 argptr = varg.buf;
2419 argsize = varg.len;
2420 }
2421 myptr = self->ob_bytes;
2422 mysize = Py_SIZE(self);
2423 left = lstrip_helper(myptr, mysize, argptr, argsize);
2424 right = mysize;
2425 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002426 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002427 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2428}
2429
2430PyDoc_STRVAR(rstrip__doc__,
2431"B.rstrip([bytes]) -> bytearray\n\
2432\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002433Strip trailing bytes contained in the argument\n\
2434and return the result as a new bytearray.\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002435If the argument is omitted, strip trailing ASCII whitespace.");
2436static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002437bytearray_rstrip(PyByteArrayObject *self, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002438{
2439 Py_ssize_t left, right, mysize, argsize;
2440 void *myptr, *argptr;
2441 PyObject *arg = Py_None;
2442 Py_buffer varg;
2443 if (!PyArg_ParseTuple(args, "|O:rstrip", &arg))
2444 return NULL;
2445 if (arg == Py_None) {
2446 argptr = "\t\n\r\f\v ";
2447 argsize = 6;
2448 }
2449 else {
2450 if (_getbuffer(arg, &varg) < 0)
2451 return NULL;
2452 argptr = varg.buf;
2453 argsize = varg.len;
2454 }
2455 myptr = self->ob_bytes;
2456 mysize = Py_SIZE(self);
2457 left = 0;
2458 right = rstrip_helper(myptr, mysize, argptr, argsize);
2459 if (arg != Py_None)
Martin v. Löwis423be952008-08-13 15:53:07 +00002460 PyBuffer_Release(&varg);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002461 return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left);
2462}
2463
2464PyDoc_STRVAR(decode_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002465"B.decode([encoding[, errors]]) -> str\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002466\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002467Decode B using the codec registered for encoding. encoding defaults\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002468to the default encoding. errors may be given to set a different error\n\
2469handling scheme. Default is 'strict' meaning that encoding errors raise\n\
2470a UnicodeDecodeError. Other possible values are 'ignore' and 'replace'\n\
2471as well as any other name registered with codecs.register_error that is\n\
2472able to handle UnicodeDecodeErrors.");
2473
2474static PyObject *
Benjamin Peterson308d6372009-09-18 21:42:35 +00002475bytearray_decode(PyObject *self, PyObject *args, PyObject *kwargs)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002476{
2477 const char *encoding = NULL;
2478 const char *errors = NULL;
Benjamin Peterson308d6372009-09-18 21:42:35 +00002479 static char *kwlist[] = {"encoding", "errors", 0};
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002480
Benjamin Peterson308d6372009-09-18 21:42:35 +00002481 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ss:decode", kwlist, &encoding, &errors))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002482 return NULL;
2483 if (encoding == NULL)
2484 encoding = PyUnicode_GetDefaultEncoding();
Marc-André Lemburgb2750b52008-06-06 12:18:17 +00002485 return PyUnicode_FromEncodedObject(self, encoding, errors);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002486}
2487
2488PyDoc_STRVAR(alloc_doc,
2489"B.__alloc__() -> int\n\
2490\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002491Return the number of bytes actually allocated.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002492
2493static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002494bytearray_alloc(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002495{
2496 return PyLong_FromSsize_t(self->ob_alloc);
2497}
2498
2499PyDoc_STRVAR(join_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002500"B.join(iterable_of_bytes) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002501\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002502Concatenate any number of bytes/bytearray objects, with B\n\
2503in between each pair, and return the result as a new bytearray.");
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002504
2505static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002506bytearray_join(PyByteArrayObject *self, PyObject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002507{
2508 PyObject *seq;
2509 Py_ssize_t mysize = Py_SIZE(self);
2510 Py_ssize_t i;
2511 Py_ssize_t n;
2512 PyObject **items;
2513 Py_ssize_t totalsize = 0;
2514 PyObject *result;
2515 char *dest;
2516
2517 seq = PySequence_Fast(it, "can only join an iterable");
2518 if (seq == NULL)
2519 return NULL;
2520 n = PySequence_Fast_GET_SIZE(seq);
2521 items = PySequence_Fast_ITEMS(seq);
2522
2523 /* Compute the total size, and check that they are all bytes */
2524 /* XXX Shouldn't we use _getbuffer() on these items instead? */
2525 for (i = 0; i < n; i++) {
2526 PyObject *obj = items[i];
2527 if (!PyByteArray_Check(obj) && !PyBytes_Check(obj)) {
2528 PyErr_Format(PyExc_TypeError,
2529 "can only join an iterable of bytes "
2530 "(item %ld has type '%.100s')",
2531 /* XXX %ld isn't right on Win64 */
2532 (long)i, Py_TYPE(obj)->tp_name);
2533 goto error;
2534 }
2535 if (i > 0)
2536 totalsize += mysize;
2537 totalsize += Py_SIZE(obj);
2538 if (totalsize < 0) {
2539 PyErr_NoMemory();
2540 goto error;
2541 }
2542 }
2543
2544 /* Allocate the result, and copy the bytes */
2545 result = PyByteArray_FromStringAndSize(NULL, totalsize);
2546 if (result == NULL)
2547 goto error;
2548 dest = PyByteArray_AS_STRING(result);
2549 for (i = 0; i < n; i++) {
2550 PyObject *obj = items[i];
2551 Py_ssize_t size = Py_SIZE(obj);
2552 char *buf;
2553 if (PyByteArray_Check(obj))
2554 buf = PyByteArray_AS_STRING(obj);
2555 else
2556 buf = PyBytes_AS_STRING(obj);
2557 if (i) {
2558 memcpy(dest, self->ob_bytes, mysize);
2559 dest += mysize;
2560 }
2561 memcpy(dest, buf, size);
2562 dest += size;
2563 }
2564
2565 /* Done */
2566 Py_DECREF(seq);
2567 return result;
2568
2569 /* Error handling */
2570 error:
2571 Py_DECREF(seq);
2572 return NULL;
2573}
2574
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002575PyDoc_STRVAR(splitlines__doc__,
2576"B.splitlines([keepends]) -> list of lines\n\
2577\n\
2578Return a list of the lines in B, breaking at line boundaries.\n\
2579Line breaks are not included in the resulting list unless keepends\n\
2580is given and true.");
2581
2582static PyObject*
2583bytearray_splitlines(PyObject *self, PyObject *args)
2584{
2585 int keepends = 0;
2586
2587 if (!PyArg_ParseTuple(args, "|i:splitlines", &keepends))
2588 return NULL;
2589
2590 return stringlib_splitlines(
2591 (PyObject*) self, PyByteArray_AS_STRING(self),
2592 PyByteArray_GET_SIZE(self), keepends
2593 );
2594}
2595
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002596PyDoc_STRVAR(fromhex_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002597"bytearray.fromhex(string) -> bytearray (static method)\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002598\n\
2599Create a bytearray object from a string of hexadecimal numbers.\n\
2600Spaces between two numbers are accepted.\n\
2601Example: bytearray.fromhex('B9 01EF') -> bytearray(b'\\xb9\\x01\\xef').");
2602
2603static int
2604hex_digit_to_int(Py_UNICODE c)
2605{
2606 if (c >= 128)
2607 return -1;
Eric Smith6dc46f52009-04-27 20:39:49 +00002608 if (Py_ISDIGIT(c))
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002609 return c - '0';
2610 else {
Eric Smith6dc46f52009-04-27 20:39:49 +00002611 if (Py_ISUPPER(c))
2612 c = Py_TOLOWER(c);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002613 if (c >= 'a' && c <= 'f')
2614 return c - 'a' + 10;
2615 }
2616 return -1;
2617}
2618
2619static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002620bytearray_fromhex(PyObject *cls, PyObject *args)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002621{
2622 PyObject *newbytes, *hexobj;
2623 char *buf;
2624 Py_UNICODE *hex;
2625 Py_ssize_t hexlen, byteslen, i, j;
2626 int top, bot;
2627
2628 if (!PyArg_ParseTuple(args, "U:fromhex", &hexobj))
2629 return NULL;
2630 assert(PyUnicode_Check(hexobj));
2631 hexlen = PyUnicode_GET_SIZE(hexobj);
2632 hex = PyUnicode_AS_UNICODE(hexobj);
2633 byteslen = hexlen/2; /* This overestimates if there are spaces */
2634 newbytes = PyByteArray_FromStringAndSize(NULL, byteslen);
2635 if (!newbytes)
2636 return NULL;
2637 buf = PyByteArray_AS_STRING(newbytes);
2638 for (i = j = 0; i < hexlen; i += 2) {
2639 /* skip over spaces in the input */
2640 while (hex[i] == ' ')
2641 i++;
2642 if (i >= hexlen)
2643 break;
2644 top = hex_digit_to_int(hex[i]);
2645 bot = hex_digit_to_int(hex[i+1]);
2646 if (top == -1 || bot == -1) {
2647 PyErr_Format(PyExc_ValueError,
2648 "non-hexadecimal number found in "
2649 "fromhex() arg at position %zd", i);
2650 goto error;
2651 }
2652 buf[j++] = (top << 4) + bot;
2653 }
2654 if (PyByteArray_Resize(newbytes, j) < 0)
2655 goto error;
2656 return newbytes;
2657
2658 error:
2659 Py_DECREF(newbytes);
2660 return NULL;
2661}
2662
2663PyDoc_STRVAR(reduce_doc, "Return state information for pickling.");
2664
2665static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002666bytearray_reduce(PyByteArrayObject *self)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002667{
2668 PyObject *latin1, *dict;
2669 if (self->ob_bytes)
2670 latin1 = PyUnicode_DecodeLatin1(self->ob_bytes,
2671 Py_SIZE(self), NULL);
2672 else
2673 latin1 = PyUnicode_FromString("");
2674
2675 dict = PyObject_GetAttrString((PyObject *)self, "__dict__");
2676 if (dict == NULL) {
2677 PyErr_Clear();
2678 dict = Py_None;
2679 Py_INCREF(dict);
2680 }
2681
2682 return Py_BuildValue("(O(Ns)N)", Py_TYPE(self), latin1, "latin-1", dict);
2683}
2684
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002685PyDoc_STRVAR(sizeof_doc,
2686"B.__sizeof__() -> int\n\
2687 \n\
2688Returns the size of B in memory, in bytes");
2689static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002690bytearray_sizeof(PyByteArrayObject *self)
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002691{
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002692 Py_ssize_t res;
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002693
Benjamin Petersonef3e4c22009-04-11 19:48:14 +00002694 res = sizeof(PyByteArrayObject) + self->ob_alloc * sizeof(char);
2695 return PyLong_FromSsize_t(res);
Robert Schuppeniesfbe94c52008-07-14 10:13:31 +00002696}
2697
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002698static PySequenceMethods bytearray_as_sequence = {
2699 (lenfunc)bytearray_length, /* sq_length */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002700 (binaryfunc)PyByteArray_Concat, /* sq_concat */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002701 (ssizeargfunc)bytearray_repeat, /* sq_repeat */
2702 (ssizeargfunc)bytearray_getitem, /* sq_item */
2703 0, /* sq_slice */
2704 (ssizeobjargproc)bytearray_setitem, /* sq_ass_item */
2705 0, /* sq_ass_slice */
2706 (objobjproc)bytearray_contains, /* sq_contains */
2707 (binaryfunc)bytearray_iconcat, /* sq_inplace_concat */
2708 (ssizeargfunc)bytearray_irepeat, /* sq_inplace_repeat */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002709};
2710
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002711static PyMappingMethods bytearray_as_mapping = {
2712 (lenfunc)bytearray_length,
2713 (binaryfunc)bytearray_subscript,
2714 (objobjargproc)bytearray_ass_subscript,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002715};
2716
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002717static PyBufferProcs bytearray_as_buffer = {
2718 (getbufferproc)bytearray_getbuffer,
2719 (releasebufferproc)bytearray_releasebuffer,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002720};
2721
2722static PyMethodDef
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002723bytearray_methods[] = {
2724 {"__alloc__", (PyCFunction)bytearray_alloc, METH_NOARGS, alloc_doc},
2725 {"__reduce__", (PyCFunction)bytearray_reduce, METH_NOARGS, reduce_doc},
2726 {"__sizeof__", (PyCFunction)bytearray_sizeof, METH_NOARGS, sizeof_doc},
2727 {"append", (PyCFunction)bytearray_append, METH_O, append__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002728 {"capitalize", (PyCFunction)stringlib_capitalize, METH_NOARGS,
2729 _Py_capitalize__doc__},
2730 {"center", (PyCFunction)stringlib_center, METH_VARARGS, center__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002731 {"count", (PyCFunction)bytearray_count, METH_VARARGS, count__doc__},
Benjamin Peterson308d6372009-09-18 21:42:35 +00002732 {"decode", (PyCFunction)bytearray_decode, METH_VARARGS | METH_KEYWORDS, decode_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002733 {"endswith", (PyCFunction)bytearray_endswith, METH_VARARGS, endswith__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002734 {"expandtabs", (PyCFunction)stringlib_expandtabs, METH_VARARGS,
2735 expandtabs__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002736 {"extend", (PyCFunction)bytearray_extend, METH_O, extend__doc__},
2737 {"find", (PyCFunction)bytearray_find, METH_VARARGS, find__doc__},
2738 {"fromhex", (PyCFunction)bytearray_fromhex, METH_VARARGS|METH_CLASS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002739 fromhex_doc},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002740 {"index", (PyCFunction)bytearray_index, METH_VARARGS, index__doc__},
2741 {"insert", (PyCFunction)bytearray_insert, METH_VARARGS, insert__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002742 {"isalnum", (PyCFunction)stringlib_isalnum, METH_NOARGS,
2743 _Py_isalnum__doc__},
2744 {"isalpha", (PyCFunction)stringlib_isalpha, METH_NOARGS,
2745 _Py_isalpha__doc__},
2746 {"isdigit", (PyCFunction)stringlib_isdigit, METH_NOARGS,
2747 _Py_isdigit__doc__},
2748 {"islower", (PyCFunction)stringlib_islower, METH_NOARGS,
2749 _Py_islower__doc__},
2750 {"isspace", (PyCFunction)stringlib_isspace, METH_NOARGS,
2751 _Py_isspace__doc__},
2752 {"istitle", (PyCFunction)stringlib_istitle, METH_NOARGS,
2753 _Py_istitle__doc__},
2754 {"isupper", (PyCFunction)stringlib_isupper, METH_NOARGS,
2755 _Py_isupper__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002756 {"join", (PyCFunction)bytearray_join, METH_O, join_doc},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002757 {"ljust", (PyCFunction)stringlib_ljust, METH_VARARGS, ljust__doc__},
2758 {"lower", (PyCFunction)stringlib_lower, METH_NOARGS, _Py_lower__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002759 {"lstrip", (PyCFunction)bytearray_lstrip, METH_VARARGS, lstrip__doc__},
2760 {"maketrans", (PyCFunction)bytearray_maketrans, METH_VARARGS|METH_STATIC,
Georg Brandlabc38772009-04-12 15:51:51 +00002761 _Py_maketrans__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002762 {"partition", (PyCFunction)bytearray_partition, METH_O, partition__doc__},
2763 {"pop", (PyCFunction)bytearray_pop, METH_VARARGS, pop__doc__},
2764 {"remove", (PyCFunction)bytearray_remove, METH_O, remove__doc__},
2765 {"replace", (PyCFunction)bytearray_replace, METH_VARARGS, replace__doc__},
2766 {"reverse", (PyCFunction)bytearray_reverse, METH_NOARGS, reverse__doc__},
2767 {"rfind", (PyCFunction)bytearray_rfind, METH_VARARGS, rfind__doc__},
2768 {"rindex", (PyCFunction)bytearray_rindex, METH_VARARGS, rindex__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002769 {"rjust", (PyCFunction)stringlib_rjust, METH_VARARGS, rjust__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002770 {"rpartition", (PyCFunction)bytearray_rpartition, METH_O, rpartition__doc__},
2771 {"rsplit", (PyCFunction)bytearray_rsplit, METH_VARARGS, rsplit__doc__},
2772 {"rstrip", (PyCFunction)bytearray_rstrip, METH_VARARGS, rstrip__doc__},
2773 {"split", (PyCFunction)bytearray_split, METH_VARARGS, split__doc__},
Antoine Pitrouf2c54842010-01-13 08:07:53 +00002774 {"splitlines", (PyCFunction)bytearray_splitlines, METH_VARARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002775 splitlines__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002776 {"startswith", (PyCFunction)bytearray_startswith, METH_VARARGS ,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002777 startswith__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002778 {"strip", (PyCFunction)bytearray_strip, METH_VARARGS, strip__doc__},
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002779 {"swapcase", (PyCFunction)stringlib_swapcase, METH_NOARGS,
2780 _Py_swapcase__doc__},
2781 {"title", (PyCFunction)stringlib_title, METH_NOARGS, _Py_title__doc__},
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002782 {"translate", (PyCFunction)bytearray_translate, METH_VARARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002783 translate__doc__},
2784 {"upper", (PyCFunction)stringlib_upper, METH_NOARGS, _Py_upper__doc__},
2785 {"zfill", (PyCFunction)stringlib_zfill, METH_VARARGS, zfill__doc__},
2786 {NULL}
2787};
2788
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002789PyDoc_STRVAR(bytearray_doc,
Georg Brandl17cb8a82008-05-30 08:20:09 +00002790"bytearray(iterable_of_ints) -> bytearray\n\
2791bytearray(string, encoding[, errors]) -> bytearray\n\
2792bytearray(bytes_or_bytearray) -> mutable copy of bytes_or_bytearray\n\
2793bytearray(memory_view) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002794\n\
2795Construct an mutable bytearray object from:\n\
2796 - an iterable yielding integers in range(256)\n\
2797 - a text string encoded using the specified encoding\n\
2798 - a bytes or a bytearray object\n\
2799 - any object implementing the buffer API.\n\
2800\n\
Georg Brandl17cb8a82008-05-30 08:20:09 +00002801bytearray(int) -> bytearray\n\
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002802\n\
2803Construct a zero-initialized bytearray of the given length.");
2804
2805
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002806static PyObject *bytearray_iter(PyObject *seq);
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002807
2808PyTypeObject PyByteArray_Type = {
2809 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2810 "bytearray",
2811 sizeof(PyByteArrayObject),
2812 0,
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002813 (destructor)bytearray_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002814 0, /* tp_print */
2815 0, /* tp_getattr */
2816 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00002817 0, /* tp_reserved */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002818 (reprfunc)bytearray_repr, /* tp_repr */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002819 0, /* tp_as_number */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002820 &bytearray_as_sequence, /* tp_as_sequence */
2821 &bytearray_as_mapping, /* tp_as_mapping */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002822 0, /* tp_hash */
2823 0, /* tp_call */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002824 bytearray_str, /* tp_str */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002825 PyObject_GenericGetAttr, /* tp_getattro */
2826 0, /* tp_setattro */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002827 &bytearray_as_buffer, /* tp_as_buffer */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002828 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /* tp_flags */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002829 bytearray_doc, /* tp_doc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002830 0, /* tp_traverse */
2831 0, /* tp_clear */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002832 (richcmpfunc)bytearray_richcompare, /* tp_richcompare */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002833 0, /* tp_weaklistoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002834 bytearray_iter, /* tp_iter */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002835 0, /* tp_iternext */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002836 bytearray_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002837 0, /* tp_members */
2838 0, /* tp_getset */
2839 0, /* tp_base */
2840 0, /* tp_dict */
2841 0, /* tp_descr_get */
2842 0, /* tp_descr_set */
2843 0, /* tp_dictoffset */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002844 (initproc)bytearray_init, /* tp_init */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002845 PyType_GenericAlloc, /* tp_alloc */
2846 PyType_GenericNew, /* tp_new */
2847 PyObject_Del, /* tp_free */
2848};
2849
2850/*********************** Bytes Iterator ****************************/
2851
2852typedef struct {
2853 PyObject_HEAD
2854 Py_ssize_t it_index;
2855 PyByteArrayObject *it_seq; /* Set to NULL when iterator is exhausted */
2856} bytesiterobject;
2857
2858static void
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002859bytearrayiter_dealloc(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002860{
2861 _PyObject_GC_UNTRACK(it);
2862 Py_XDECREF(it->it_seq);
2863 PyObject_GC_Del(it);
2864}
2865
2866static int
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002867bytearrayiter_traverse(bytesiterobject *it, visitproc visit, void *arg)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002868{
2869 Py_VISIT(it->it_seq);
2870 return 0;
2871}
2872
2873static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002874bytearrayiter_next(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002875{
2876 PyByteArrayObject *seq;
2877 PyObject *item;
2878
2879 assert(it != NULL);
2880 seq = it->it_seq;
2881 if (seq == NULL)
2882 return NULL;
2883 assert(PyByteArray_Check(seq));
2884
2885 if (it->it_index < PyByteArray_GET_SIZE(seq)) {
2886 item = PyLong_FromLong(
2887 (unsigned char)seq->ob_bytes[it->it_index]);
2888 if (item != NULL)
2889 ++it->it_index;
2890 return item;
2891 }
2892
2893 Py_DECREF(seq);
2894 it->it_seq = NULL;
2895 return NULL;
2896}
2897
2898static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002899bytesarrayiter_length_hint(bytesiterobject *it)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002900{
2901 Py_ssize_t len = 0;
2902 if (it->it_seq)
2903 len = PyByteArray_GET_SIZE(it->it_seq) - it->it_index;
2904 return PyLong_FromSsize_t(len);
2905}
2906
2907PyDoc_STRVAR(length_hint_doc,
2908 "Private method returning an estimate of len(list(it)).");
2909
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002910static PyMethodDef bytearrayiter_methods[] = {
2911 {"__length_hint__", (PyCFunction)bytesarrayiter_length_hint, METH_NOARGS,
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002912 length_hint_doc},
2913 {NULL, NULL} /* sentinel */
2914};
2915
2916PyTypeObject PyByteArrayIter_Type = {
2917 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2918 "bytearray_iterator", /* tp_name */
2919 sizeof(bytesiterobject), /* tp_basicsize */
2920 0, /* tp_itemsize */
2921 /* methods */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002922 (destructor)bytearrayiter_dealloc, /* tp_dealloc */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002923 0, /* tp_print */
2924 0, /* tp_getattr */
2925 0, /* tp_setattr */
Mark Dickinsone94c6792009-02-02 20:36:42 +00002926 0, /* tp_reserved */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002927 0, /* tp_repr */
2928 0, /* tp_as_number */
2929 0, /* tp_as_sequence */
2930 0, /* tp_as_mapping */
2931 0, /* tp_hash */
2932 0, /* tp_call */
2933 0, /* tp_str */
2934 PyObject_GenericGetAttr, /* tp_getattro */
2935 0, /* tp_setattro */
2936 0, /* tp_as_buffer */
2937 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */
2938 0, /* tp_doc */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002939 (traverseproc)bytearrayiter_traverse, /* tp_traverse */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002940 0, /* tp_clear */
2941 0, /* tp_richcompare */
2942 0, /* tp_weaklistoffset */
2943 PyObject_SelfIter, /* tp_iter */
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002944 (iternextfunc)bytearrayiter_next, /* tp_iternext */
2945 bytearrayiter_methods, /* tp_methods */
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002946 0,
2947};
2948
2949static PyObject *
Benjamin Peterson153c70f2009-04-18 15:42:12 +00002950bytearray_iter(PyObject *seq)
Christian Heimes2c9c7a52008-05-26 13:42:13 +00002951{
2952 bytesiterobject *it;
2953
2954 if (!PyByteArray_Check(seq)) {
2955 PyErr_BadInternalCall();
2956 return NULL;
2957 }
2958 it = PyObject_GC_New(bytesiterobject, &PyByteArrayIter_Type);
2959 if (it == NULL)
2960 return NULL;
2961 it->it_index = 0;
2962 Py_INCREF(seq);
2963 it->it_seq = (PyByteArrayObject *)seq;
2964 _PyObject_GC_TRACK(it);
2965 return (PyObject *)it;
2966}