Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1 | /* PyByteArray (bytearray) implementation */ |
| 2 | |
| 3 | #define PY_SSIZE_T_CLEAN |
| 4 | #include "Python.h" |
| 5 | #include "structmember.h" |
| 6 | #include "bytes_methods.h" |
| 7 | |
| 8 | static PyByteArrayObject *nullbytes = NULL; |
| 9 | |
| 10 | void |
| 11 | PyByteArray_Fini(void) |
| 12 | { |
| 13 | Py_CLEAR(nullbytes); |
| 14 | } |
| 15 | |
| 16 | int |
| 17 | PyByteArray_Init(void) |
| 18 | { |
| 19 | nullbytes = PyObject_New(PyByteArrayObject, &PyByteArray_Type); |
| 20 | if (nullbytes == NULL) |
| 21 | return 0; |
| 22 | nullbytes->ob_bytes = NULL; |
| 23 | Py_SIZE(nullbytes) = nullbytes->ob_alloc = 0; |
| 24 | nullbytes->ob_exports = 0; |
| 25 | return 1; |
| 26 | } |
| 27 | |
| 28 | /* end nullbytes support */ |
| 29 | |
| 30 | /* Helpers */ |
| 31 | |
| 32 | static int |
| 33 | _getbytevalue(PyObject* arg, int *value) |
| 34 | { |
| 35 | long face_value; |
| 36 | |
| 37 | if (PyLong_Check(arg)) { |
| 38 | face_value = PyLong_AsLong(arg); |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 39 | } else { |
| 40 | PyObject *index = PyNumber_Index(arg); |
| 41 | if (index == NULL) { |
| 42 | PyErr_Format(PyExc_TypeError, "an integer is required"); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 43 | return 0; |
| 44 | } |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 45 | face_value = PyLong_AsLong(index); |
| 46 | Py_DECREF(index); |
| 47 | } |
| 48 | |
| 49 | if (face_value < 0 || face_value >= 256) { |
| 50 | /* this includes the OverflowError in case the long is too large */ |
| 51 | PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)"); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 52 | return 0; |
| 53 | } |
| 54 | |
| 55 | *value = face_value; |
| 56 | return 1; |
| 57 | } |
| 58 | |
| 59 | static int |
| 60 | bytes_getbuffer(PyByteArrayObject *obj, Py_buffer *view, int flags) |
| 61 | { |
| 62 | int ret; |
| 63 | void *ptr; |
| 64 | if (view == NULL) { |
| 65 | obj->ob_exports++; |
| 66 | return 0; |
| 67 | } |
| 68 | if (obj->ob_bytes == NULL) |
| 69 | ptr = ""; |
| 70 | else |
| 71 | ptr = obj->ob_bytes; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 72 | ret = PyBuffer_FillInfo(view, (PyObject*)obj, ptr, Py_SIZE(obj), 0, flags); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 73 | if (ret >= 0) { |
| 74 | obj->ob_exports++; |
| 75 | } |
| 76 | return ret; |
| 77 | } |
| 78 | |
| 79 | static void |
| 80 | bytes_releasebuffer(PyByteArrayObject *obj, Py_buffer *view) |
| 81 | { |
| 82 | obj->ob_exports--; |
| 83 | } |
| 84 | |
| 85 | static Py_ssize_t |
| 86 | _getbuffer(PyObject *obj, Py_buffer *view) |
| 87 | { |
| 88 | PyBufferProcs *buffer = Py_TYPE(obj)->tp_as_buffer; |
| 89 | |
| 90 | if (buffer == NULL || buffer->bf_getbuffer == NULL) |
| 91 | { |
| 92 | PyErr_Format(PyExc_TypeError, |
| 93 | "Type %.100s doesn't support the buffer API", |
| 94 | Py_TYPE(obj)->tp_name); |
| 95 | return -1; |
| 96 | } |
| 97 | |
| 98 | if (buffer->bf_getbuffer(obj, view, PyBUF_SIMPLE) < 0) |
| 99 | return -1; |
| 100 | return view->len; |
| 101 | } |
| 102 | |
| 103 | /* Direct API functions */ |
| 104 | |
| 105 | PyObject * |
| 106 | PyByteArray_FromObject(PyObject *input) |
| 107 | { |
| 108 | return PyObject_CallFunctionObjArgs((PyObject *)&PyByteArray_Type, |
| 109 | input, NULL); |
| 110 | } |
| 111 | |
| 112 | PyObject * |
| 113 | PyByteArray_FromStringAndSize(const char *bytes, Py_ssize_t size) |
| 114 | { |
| 115 | PyByteArrayObject *new; |
| 116 | Py_ssize_t alloc; |
| 117 | |
| 118 | if (size < 0) { |
| 119 | PyErr_SetString(PyExc_SystemError, |
| 120 | "Negative size passed to PyByteArray_FromStringAndSize"); |
| 121 | return NULL; |
| 122 | } |
| 123 | |
Neal Norwitz | 3ce5d92 | 2008-08-24 07:08:55 +0000 | [diff] [blame] | 124 | /* Prevent buffer overflow when setting alloc to size+1. */ |
| 125 | if (size == PY_SSIZE_T_MAX) { |
| 126 | return PyErr_NoMemory(); |
| 127 | } |
| 128 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 129 | new = PyObject_New(PyByteArrayObject, &PyByteArray_Type); |
| 130 | if (new == NULL) |
| 131 | return NULL; |
| 132 | |
| 133 | if (size == 0) { |
| 134 | new->ob_bytes = NULL; |
| 135 | alloc = 0; |
| 136 | } |
| 137 | else { |
| 138 | alloc = size + 1; |
| 139 | new->ob_bytes = PyMem_Malloc(alloc); |
| 140 | if (new->ob_bytes == NULL) { |
| 141 | Py_DECREF(new); |
| 142 | return PyErr_NoMemory(); |
| 143 | } |
| 144 | if (bytes != NULL) |
| 145 | memcpy(new->ob_bytes, bytes, size); |
| 146 | new->ob_bytes[size] = '\0'; /* Trailing null byte */ |
| 147 | } |
| 148 | Py_SIZE(new) = size; |
| 149 | new->ob_alloc = alloc; |
| 150 | new->ob_exports = 0; |
| 151 | |
| 152 | return (PyObject *)new; |
| 153 | } |
| 154 | |
| 155 | Py_ssize_t |
| 156 | PyByteArray_Size(PyObject *self) |
| 157 | { |
| 158 | assert(self != NULL); |
| 159 | assert(PyByteArray_Check(self)); |
| 160 | |
| 161 | return PyByteArray_GET_SIZE(self); |
| 162 | } |
| 163 | |
| 164 | char * |
| 165 | PyByteArray_AsString(PyObject *self) |
| 166 | { |
| 167 | assert(self != NULL); |
| 168 | assert(PyByteArray_Check(self)); |
| 169 | |
| 170 | return PyByteArray_AS_STRING(self); |
| 171 | } |
| 172 | |
| 173 | int |
| 174 | PyByteArray_Resize(PyObject *self, Py_ssize_t size) |
| 175 | { |
| 176 | void *sval; |
| 177 | Py_ssize_t alloc = ((PyByteArrayObject *)self)->ob_alloc; |
| 178 | |
| 179 | assert(self != NULL); |
| 180 | assert(PyByteArray_Check(self)); |
| 181 | assert(size >= 0); |
| 182 | |
| 183 | if (size < alloc / 2) { |
| 184 | /* Major downsize; resize down to exact size */ |
| 185 | alloc = size + 1; |
| 186 | } |
| 187 | else if (size < alloc) { |
| 188 | /* Within allocated size; quick exit */ |
| 189 | Py_SIZE(self) = size; |
| 190 | ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null */ |
| 191 | return 0; |
| 192 | } |
| 193 | else if (size <= alloc * 1.125) { |
| 194 | /* Moderate upsize; overallocate similar to list_resize() */ |
| 195 | alloc = size + (size >> 3) + (size < 9 ? 3 : 6); |
| 196 | } |
| 197 | else { |
| 198 | /* Major upsize; resize up to exact size */ |
| 199 | alloc = size + 1; |
| 200 | } |
| 201 | |
| 202 | if (((PyByteArrayObject *)self)->ob_exports > 0) { |
| 203 | /* |
| 204 | fprintf(stderr, "%d: %s", ((PyByteArrayObject *)self)->ob_exports, |
| 205 | ((PyByteArrayObject *)self)->ob_bytes); |
| 206 | */ |
| 207 | PyErr_SetString(PyExc_BufferError, |
| 208 | "Existing exports of data: object cannot be re-sized"); |
| 209 | return -1; |
| 210 | } |
| 211 | |
| 212 | sval = PyMem_Realloc(((PyByteArrayObject *)self)->ob_bytes, alloc); |
| 213 | if (sval == NULL) { |
| 214 | PyErr_NoMemory(); |
| 215 | return -1; |
| 216 | } |
| 217 | |
| 218 | ((PyByteArrayObject *)self)->ob_bytes = sval; |
| 219 | Py_SIZE(self) = size; |
| 220 | ((PyByteArrayObject *)self)->ob_alloc = alloc; |
| 221 | ((PyByteArrayObject *)self)->ob_bytes[size] = '\0'; /* Trailing null byte */ |
| 222 | |
| 223 | return 0; |
| 224 | } |
| 225 | |
| 226 | PyObject * |
| 227 | PyByteArray_Concat(PyObject *a, PyObject *b) |
| 228 | { |
| 229 | Py_ssize_t size; |
| 230 | Py_buffer va, vb; |
| 231 | PyByteArrayObject *result = NULL; |
| 232 | |
| 233 | va.len = -1; |
| 234 | vb.len = -1; |
| 235 | if (_getbuffer(a, &va) < 0 || |
| 236 | _getbuffer(b, &vb) < 0) { |
| 237 | PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s", |
| 238 | Py_TYPE(a)->tp_name, Py_TYPE(b)->tp_name); |
| 239 | goto done; |
| 240 | } |
| 241 | |
| 242 | size = va.len + vb.len; |
| 243 | if (size < 0) { |
| 244 | return PyErr_NoMemory(); |
| 245 | goto done; |
| 246 | } |
| 247 | |
| 248 | result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, size); |
| 249 | if (result != NULL) { |
| 250 | memcpy(result->ob_bytes, va.buf, va.len); |
| 251 | memcpy(result->ob_bytes + va.len, vb.buf, vb.len); |
| 252 | } |
| 253 | |
| 254 | done: |
| 255 | if (va.len != -1) |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 256 | PyBuffer_Release(&va); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 257 | if (vb.len != -1) |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 258 | PyBuffer_Release(&vb); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 259 | return (PyObject *)result; |
| 260 | } |
| 261 | |
| 262 | /* Functions stuffed into the type object */ |
| 263 | |
| 264 | static Py_ssize_t |
| 265 | bytes_length(PyByteArrayObject *self) |
| 266 | { |
| 267 | return Py_SIZE(self); |
| 268 | } |
| 269 | |
| 270 | static PyObject * |
| 271 | bytes_iconcat(PyByteArrayObject *self, PyObject *other) |
| 272 | { |
| 273 | Py_ssize_t mysize; |
| 274 | Py_ssize_t size; |
| 275 | Py_buffer vo; |
| 276 | |
| 277 | if (_getbuffer(other, &vo) < 0) { |
| 278 | PyErr_Format(PyExc_TypeError, "can't concat %.100s to %.100s", |
| 279 | Py_TYPE(other)->tp_name, Py_TYPE(self)->tp_name); |
| 280 | return NULL; |
| 281 | } |
| 282 | |
| 283 | mysize = Py_SIZE(self); |
| 284 | size = mysize + vo.len; |
| 285 | if (size < 0) { |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 286 | PyBuffer_Release(&vo); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 287 | return PyErr_NoMemory(); |
| 288 | } |
| 289 | if (size < self->ob_alloc) { |
| 290 | Py_SIZE(self) = size; |
| 291 | self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */ |
| 292 | } |
| 293 | else if (PyByteArray_Resize((PyObject *)self, size) < 0) { |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 294 | PyBuffer_Release(&vo); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 295 | return NULL; |
| 296 | } |
| 297 | memcpy(self->ob_bytes + mysize, vo.buf, vo.len); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 298 | PyBuffer_Release(&vo); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 299 | Py_INCREF(self); |
| 300 | return (PyObject *)self; |
| 301 | } |
| 302 | |
| 303 | static PyObject * |
| 304 | bytes_repeat(PyByteArrayObject *self, Py_ssize_t count) |
| 305 | { |
| 306 | PyByteArrayObject *result; |
| 307 | Py_ssize_t mysize; |
| 308 | Py_ssize_t size; |
| 309 | |
| 310 | if (count < 0) |
| 311 | count = 0; |
| 312 | mysize = Py_SIZE(self); |
| 313 | size = mysize * count; |
| 314 | if (count != 0 && size / count != mysize) |
| 315 | return PyErr_NoMemory(); |
| 316 | result = (PyByteArrayObject *)PyByteArray_FromStringAndSize(NULL, size); |
| 317 | if (result != NULL && size != 0) { |
| 318 | if (mysize == 1) |
| 319 | memset(result->ob_bytes, self->ob_bytes[0], size); |
| 320 | else { |
| 321 | Py_ssize_t i; |
| 322 | for (i = 0; i < count; i++) |
| 323 | memcpy(result->ob_bytes + i*mysize, self->ob_bytes, mysize); |
| 324 | } |
| 325 | } |
| 326 | return (PyObject *)result; |
| 327 | } |
| 328 | |
| 329 | static PyObject * |
| 330 | bytes_irepeat(PyByteArrayObject *self, Py_ssize_t count) |
| 331 | { |
| 332 | Py_ssize_t mysize; |
| 333 | Py_ssize_t size; |
| 334 | |
| 335 | if (count < 0) |
| 336 | count = 0; |
| 337 | mysize = Py_SIZE(self); |
| 338 | size = mysize * count; |
| 339 | if (count != 0 && size / count != mysize) |
| 340 | return PyErr_NoMemory(); |
| 341 | if (size < self->ob_alloc) { |
| 342 | Py_SIZE(self) = size; |
| 343 | self->ob_bytes[Py_SIZE(self)] = '\0'; /* Trailing null byte */ |
| 344 | } |
| 345 | else if (PyByteArray_Resize((PyObject *)self, size) < 0) |
| 346 | return NULL; |
| 347 | |
| 348 | if (mysize == 1) |
| 349 | memset(self->ob_bytes, self->ob_bytes[0], size); |
| 350 | else { |
| 351 | Py_ssize_t i; |
| 352 | for (i = 1; i < count; i++) |
| 353 | memcpy(self->ob_bytes + i*mysize, self->ob_bytes, mysize); |
| 354 | } |
| 355 | |
| 356 | Py_INCREF(self); |
| 357 | return (PyObject *)self; |
| 358 | } |
| 359 | |
| 360 | static PyObject * |
| 361 | bytes_getitem(PyByteArrayObject *self, Py_ssize_t i) |
| 362 | { |
| 363 | if (i < 0) |
| 364 | i += Py_SIZE(self); |
| 365 | if (i < 0 || i >= Py_SIZE(self)) { |
| 366 | PyErr_SetString(PyExc_IndexError, "bytearray index out of range"); |
| 367 | return NULL; |
| 368 | } |
| 369 | return PyLong_FromLong((unsigned char)(self->ob_bytes[i])); |
| 370 | } |
| 371 | |
| 372 | static PyObject * |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 373 | bytes_subscript(PyByteArrayObject *self, PyObject *index) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 374 | { |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 375 | if (PyIndex_Check(index)) { |
| 376 | Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 377 | |
| 378 | if (i == -1 && PyErr_Occurred()) |
| 379 | return NULL; |
| 380 | |
| 381 | if (i < 0) |
| 382 | i += PyByteArray_GET_SIZE(self); |
| 383 | |
| 384 | if (i < 0 || i >= Py_SIZE(self)) { |
| 385 | PyErr_SetString(PyExc_IndexError, "bytearray index out of range"); |
| 386 | return NULL; |
| 387 | } |
| 388 | return PyLong_FromLong((unsigned char)(self->ob_bytes[i])); |
| 389 | } |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 390 | else if (PySlice_Check(index)) { |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 391 | Py_ssize_t start, stop, step, slicelength, cur, i; |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 392 | if (PySlice_GetIndicesEx((PySliceObject *)index, |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 393 | PyByteArray_GET_SIZE(self), |
| 394 | &start, &stop, &step, &slicelength) < 0) { |
| 395 | return NULL; |
| 396 | } |
| 397 | |
| 398 | if (slicelength <= 0) |
| 399 | return PyByteArray_FromStringAndSize("", 0); |
| 400 | else if (step == 1) { |
| 401 | return PyByteArray_FromStringAndSize(self->ob_bytes + start, |
| 402 | slicelength); |
| 403 | } |
| 404 | else { |
| 405 | char *source_buf = PyByteArray_AS_STRING(self); |
| 406 | char *result_buf = (char *)PyMem_Malloc(slicelength); |
| 407 | PyObject *result; |
| 408 | |
| 409 | if (result_buf == NULL) |
| 410 | return PyErr_NoMemory(); |
| 411 | |
| 412 | for (cur = start, i = 0; i < slicelength; |
| 413 | cur += step, i++) { |
| 414 | result_buf[i] = source_buf[cur]; |
| 415 | } |
| 416 | result = PyByteArray_FromStringAndSize(result_buf, slicelength); |
| 417 | PyMem_Free(result_buf); |
| 418 | return result; |
| 419 | } |
| 420 | } |
| 421 | else { |
| 422 | PyErr_SetString(PyExc_TypeError, "bytearray indices must be integers"); |
| 423 | return NULL; |
| 424 | } |
| 425 | } |
| 426 | |
| 427 | static int |
| 428 | bytes_setslice(PyByteArrayObject *self, Py_ssize_t lo, Py_ssize_t hi, |
| 429 | PyObject *values) |
| 430 | { |
| 431 | Py_ssize_t avail, needed; |
| 432 | void *bytes; |
| 433 | Py_buffer vbytes; |
| 434 | int res = 0; |
| 435 | |
| 436 | vbytes.len = -1; |
| 437 | if (values == (PyObject *)self) { |
| 438 | /* Make a copy and call this function recursively */ |
| 439 | int err; |
| 440 | values = PyByteArray_FromObject(values); |
| 441 | if (values == NULL) |
| 442 | return -1; |
| 443 | err = bytes_setslice(self, lo, hi, values); |
| 444 | Py_DECREF(values); |
| 445 | return err; |
| 446 | } |
| 447 | if (values == NULL) { |
| 448 | /* del b[lo:hi] */ |
| 449 | bytes = NULL; |
| 450 | needed = 0; |
| 451 | } |
| 452 | else { |
| 453 | if (_getbuffer(values, &vbytes) < 0) { |
| 454 | PyErr_Format(PyExc_TypeError, |
Georg Brandl | 3dbca81 | 2008-07-23 16:10:53 +0000 | [diff] [blame] | 455 | "can't set bytearray slice from %.100s", |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 456 | Py_TYPE(values)->tp_name); |
| 457 | return -1; |
| 458 | } |
| 459 | needed = vbytes.len; |
| 460 | bytes = vbytes.buf; |
| 461 | } |
| 462 | |
| 463 | if (lo < 0) |
| 464 | lo = 0; |
| 465 | if (hi < lo) |
| 466 | hi = lo; |
| 467 | if (hi > Py_SIZE(self)) |
| 468 | hi = Py_SIZE(self); |
| 469 | |
| 470 | avail = hi - lo; |
| 471 | if (avail < 0) |
| 472 | lo = hi = avail = 0; |
| 473 | |
| 474 | if (avail != needed) { |
| 475 | if (avail > needed) { |
| 476 | /* |
| 477 | 0 lo hi old_size |
| 478 | | |<----avail----->|<-----tomove------>| |
| 479 | | |<-needed->|<-----tomove------>| |
| 480 | 0 lo new_hi new_size |
| 481 | */ |
| 482 | memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi, |
| 483 | Py_SIZE(self) - hi); |
| 484 | } |
| 485 | /* XXX(nnorwitz): need to verify this can't overflow! */ |
| 486 | if (PyByteArray_Resize((PyObject *)self, |
| 487 | Py_SIZE(self) + needed - avail) < 0) { |
| 488 | res = -1; |
| 489 | goto finish; |
| 490 | } |
| 491 | if (avail < needed) { |
| 492 | /* |
| 493 | 0 lo hi old_size |
| 494 | | |<-avail->|<-----tomove------>| |
| 495 | | |<----needed---->|<-----tomove------>| |
| 496 | 0 lo new_hi new_size |
| 497 | */ |
| 498 | memmove(self->ob_bytes + lo + needed, self->ob_bytes + hi, |
| 499 | Py_SIZE(self) - lo - needed); |
| 500 | } |
| 501 | } |
| 502 | |
| 503 | if (needed > 0) |
| 504 | memcpy(self->ob_bytes + lo, bytes, needed); |
| 505 | |
| 506 | |
| 507 | finish: |
| 508 | if (vbytes.len != -1) |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 509 | PyBuffer_Release(&vbytes); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 510 | return res; |
| 511 | } |
| 512 | |
| 513 | static int |
| 514 | bytes_setitem(PyByteArrayObject *self, Py_ssize_t i, PyObject *value) |
| 515 | { |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 516 | int ival; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 517 | |
| 518 | if (i < 0) |
| 519 | i += Py_SIZE(self); |
| 520 | |
| 521 | if (i < 0 || i >= Py_SIZE(self)) { |
| 522 | PyErr_SetString(PyExc_IndexError, "bytearray index out of range"); |
| 523 | return -1; |
| 524 | } |
| 525 | |
| 526 | if (value == NULL) |
| 527 | return bytes_setslice(self, i, i+1, NULL); |
| 528 | |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 529 | if (!_getbytevalue(value, &ival)) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 530 | return -1; |
| 531 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 532 | self->ob_bytes[i] = ival; |
| 533 | return 0; |
| 534 | } |
| 535 | |
| 536 | static int |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 537 | bytes_ass_subscript(PyByteArrayObject *self, PyObject *index, PyObject *values) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 538 | { |
| 539 | Py_ssize_t start, stop, step, slicelen, needed; |
| 540 | char *bytes; |
| 541 | |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 542 | if (PyIndex_Check(index)) { |
| 543 | Py_ssize_t i = PyNumber_AsSsize_t(index, PyExc_IndexError); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 544 | |
| 545 | if (i == -1 && PyErr_Occurred()) |
| 546 | return -1; |
| 547 | |
| 548 | if (i < 0) |
| 549 | i += PyByteArray_GET_SIZE(self); |
| 550 | |
| 551 | if (i < 0 || i >= Py_SIZE(self)) { |
| 552 | PyErr_SetString(PyExc_IndexError, "bytearray index out of range"); |
| 553 | return -1; |
| 554 | } |
| 555 | |
| 556 | if (values == NULL) { |
| 557 | /* Fall through to slice assignment */ |
| 558 | start = i; |
| 559 | stop = i + 1; |
| 560 | step = 1; |
| 561 | slicelen = 1; |
| 562 | } |
| 563 | else { |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 564 | int ival; |
| 565 | if (!_getbytevalue(values, &ival)) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 566 | return -1; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 567 | self->ob_bytes[i] = (char)ival; |
| 568 | return 0; |
| 569 | } |
| 570 | } |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 571 | else if (PySlice_Check(index)) { |
| 572 | if (PySlice_GetIndicesEx((PySliceObject *)index, |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 573 | PyByteArray_GET_SIZE(self), |
| 574 | &start, &stop, &step, &slicelen) < 0) { |
| 575 | return -1; |
| 576 | } |
| 577 | } |
| 578 | else { |
| 579 | PyErr_SetString(PyExc_TypeError, "bytearray indices must be integer"); |
| 580 | return -1; |
| 581 | } |
| 582 | |
| 583 | if (values == NULL) { |
| 584 | bytes = NULL; |
| 585 | needed = 0; |
| 586 | } |
| 587 | else if (values == (PyObject *)self || !PyByteArray_Check(values)) { |
| 588 | /* Make a copy an call this function recursively */ |
| 589 | int err; |
| 590 | values = PyByteArray_FromObject(values); |
| 591 | if (values == NULL) |
| 592 | return -1; |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 593 | err = bytes_ass_subscript(self, index, values); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 594 | Py_DECREF(values); |
| 595 | return err; |
| 596 | } |
| 597 | else { |
| 598 | assert(PyByteArray_Check(values)); |
| 599 | bytes = ((PyByteArrayObject *)values)->ob_bytes; |
| 600 | needed = Py_SIZE(values); |
| 601 | } |
| 602 | /* Make sure b[5:2] = ... inserts before 5, not before 2. */ |
| 603 | if ((step < 0 && start < stop) || |
| 604 | (step > 0 && start > stop)) |
| 605 | stop = start; |
| 606 | if (step == 1) { |
| 607 | if (slicelen != needed) { |
| 608 | if (slicelen > needed) { |
| 609 | /* |
| 610 | 0 start stop old_size |
| 611 | | |<---slicelen--->|<-----tomove------>| |
| 612 | | |<-needed->|<-----tomove------>| |
| 613 | 0 lo new_hi new_size |
| 614 | */ |
| 615 | memmove(self->ob_bytes + start + needed, self->ob_bytes + stop, |
| 616 | Py_SIZE(self) - stop); |
| 617 | } |
| 618 | if (PyByteArray_Resize((PyObject *)self, |
| 619 | Py_SIZE(self) + needed - slicelen) < 0) |
| 620 | return -1; |
| 621 | if (slicelen < needed) { |
| 622 | /* |
| 623 | 0 lo hi old_size |
| 624 | | |<-avail->|<-----tomove------>| |
| 625 | | |<----needed---->|<-----tomove------>| |
| 626 | 0 lo new_hi new_size |
| 627 | */ |
| 628 | memmove(self->ob_bytes + start + needed, self->ob_bytes + stop, |
| 629 | Py_SIZE(self) - start - needed); |
| 630 | } |
| 631 | } |
| 632 | |
| 633 | if (needed > 0) |
| 634 | memcpy(self->ob_bytes + start, bytes, needed); |
| 635 | |
| 636 | return 0; |
| 637 | } |
| 638 | else { |
| 639 | if (needed == 0) { |
| 640 | /* Delete slice */ |
| 641 | Py_ssize_t cur, i; |
| 642 | |
| 643 | if (step < 0) { |
| 644 | stop = start + 1; |
| 645 | start = stop + step * (slicelen - 1) - 1; |
| 646 | step = -step; |
| 647 | } |
| 648 | for (cur = start, i = 0; |
| 649 | i < slicelen; cur += step, i++) { |
| 650 | Py_ssize_t lim = step - 1; |
| 651 | |
| 652 | if (cur + step >= PyByteArray_GET_SIZE(self)) |
| 653 | lim = PyByteArray_GET_SIZE(self) - cur - 1; |
| 654 | |
| 655 | memmove(self->ob_bytes + cur - i, |
| 656 | self->ob_bytes + cur + 1, lim); |
| 657 | } |
| 658 | /* Move the tail of the bytes, in one chunk */ |
| 659 | cur = start + slicelen*step; |
| 660 | if (cur < PyByteArray_GET_SIZE(self)) { |
| 661 | memmove(self->ob_bytes + cur - slicelen, |
| 662 | self->ob_bytes + cur, |
| 663 | PyByteArray_GET_SIZE(self) - cur); |
| 664 | } |
| 665 | if (PyByteArray_Resize((PyObject *)self, |
| 666 | PyByteArray_GET_SIZE(self) - slicelen) < 0) |
| 667 | return -1; |
| 668 | |
| 669 | return 0; |
| 670 | } |
| 671 | else { |
| 672 | /* Assign slice */ |
| 673 | Py_ssize_t cur, i; |
| 674 | |
| 675 | if (needed != slicelen) { |
| 676 | PyErr_Format(PyExc_ValueError, |
| 677 | "attempt to assign bytes of size %zd " |
| 678 | "to extended slice of size %zd", |
| 679 | needed, slicelen); |
| 680 | return -1; |
| 681 | } |
| 682 | for (cur = start, i = 0; i < slicelen; cur += step, i++) |
| 683 | self->ob_bytes[cur] = bytes[i]; |
| 684 | return 0; |
| 685 | } |
| 686 | } |
| 687 | } |
| 688 | |
| 689 | static int |
| 690 | bytes_init(PyByteArrayObject *self, PyObject *args, PyObject *kwds) |
| 691 | { |
| 692 | static char *kwlist[] = {"source", "encoding", "errors", 0}; |
| 693 | PyObject *arg = NULL; |
| 694 | const char *encoding = NULL; |
| 695 | const char *errors = NULL; |
| 696 | Py_ssize_t count; |
| 697 | PyObject *it; |
| 698 | PyObject *(*iternext)(PyObject *); |
| 699 | |
| 700 | if (Py_SIZE(self) != 0) { |
| 701 | /* Empty previous contents (yes, do this first of all!) */ |
| 702 | if (PyByteArray_Resize((PyObject *)self, 0) < 0) |
| 703 | return -1; |
| 704 | } |
| 705 | |
| 706 | /* Parse arguments */ |
Georg Brandl | 3dbca81 | 2008-07-23 16:10:53 +0000 | [diff] [blame] | 707 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|Oss:bytearray", kwlist, |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 708 | &arg, &encoding, &errors)) |
| 709 | return -1; |
| 710 | |
| 711 | /* Make a quick exit if no first argument */ |
| 712 | if (arg == NULL) { |
| 713 | if (encoding != NULL || errors != NULL) { |
| 714 | PyErr_SetString(PyExc_TypeError, |
| 715 | "encoding or errors without sequence argument"); |
| 716 | return -1; |
| 717 | } |
| 718 | return 0; |
| 719 | } |
| 720 | |
| 721 | if (PyUnicode_Check(arg)) { |
| 722 | /* Encode via the codec registry */ |
| 723 | PyObject *encoded, *new; |
| 724 | if (encoding == NULL) { |
| 725 | PyErr_SetString(PyExc_TypeError, |
| 726 | "string argument without an encoding"); |
| 727 | return -1; |
| 728 | } |
Marc-André Lemburg | b2750b5 | 2008-06-06 12:18:17 +0000 | [diff] [blame] | 729 | encoded = PyUnicode_AsEncodedString(arg, encoding, errors); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 730 | if (encoded == NULL) |
| 731 | return -1; |
| 732 | assert(PyBytes_Check(encoded)); |
| 733 | new = bytes_iconcat(self, encoded); |
| 734 | Py_DECREF(encoded); |
| 735 | if (new == NULL) |
| 736 | return -1; |
| 737 | Py_DECREF(new); |
| 738 | return 0; |
| 739 | } |
| 740 | |
| 741 | /* If it's not unicode, there can't be encoding or errors */ |
| 742 | if (encoding != NULL || errors != NULL) { |
| 743 | PyErr_SetString(PyExc_TypeError, |
| 744 | "encoding or errors without a string argument"); |
| 745 | return -1; |
| 746 | } |
| 747 | |
| 748 | /* Is it an int? */ |
| 749 | count = PyNumber_AsSsize_t(arg, PyExc_ValueError); |
| 750 | if (count == -1 && PyErr_Occurred()) |
| 751 | PyErr_Clear(); |
| 752 | else { |
| 753 | if (count < 0) { |
| 754 | PyErr_SetString(PyExc_ValueError, "negative count"); |
| 755 | return -1; |
| 756 | } |
| 757 | if (count > 0) { |
| 758 | if (PyByteArray_Resize((PyObject *)self, count)) |
| 759 | return -1; |
| 760 | memset(self->ob_bytes, 0, count); |
| 761 | } |
| 762 | return 0; |
| 763 | } |
| 764 | |
| 765 | /* Use the buffer API */ |
| 766 | if (PyObject_CheckBuffer(arg)) { |
| 767 | Py_ssize_t size; |
| 768 | Py_buffer view; |
| 769 | if (PyObject_GetBuffer(arg, &view, PyBUF_FULL_RO) < 0) |
| 770 | return -1; |
| 771 | size = view.len; |
| 772 | if (PyByteArray_Resize((PyObject *)self, size) < 0) goto fail; |
| 773 | if (PyBuffer_ToContiguous(self->ob_bytes, &view, size, 'C') < 0) |
| 774 | goto fail; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 775 | PyBuffer_Release(&view); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 776 | return 0; |
| 777 | fail: |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 778 | PyBuffer_Release(&view); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 779 | return -1; |
| 780 | } |
| 781 | |
| 782 | /* XXX Optimize this if the arguments is a list, tuple */ |
| 783 | |
| 784 | /* Get the iterator */ |
| 785 | it = PyObject_GetIter(arg); |
| 786 | if (it == NULL) |
| 787 | return -1; |
| 788 | iternext = *Py_TYPE(it)->tp_iternext; |
| 789 | |
| 790 | /* Run the iterator to exhaustion */ |
| 791 | for (;;) { |
| 792 | PyObject *item; |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 793 | int rc, value; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 794 | |
| 795 | /* Get the next item */ |
| 796 | item = iternext(it); |
| 797 | if (item == NULL) { |
| 798 | if (PyErr_Occurred()) { |
| 799 | if (!PyErr_ExceptionMatches(PyExc_StopIteration)) |
| 800 | goto error; |
| 801 | PyErr_Clear(); |
| 802 | } |
| 803 | break; |
| 804 | } |
| 805 | |
| 806 | /* Interpret it as an int (__index__) */ |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 807 | rc = _getbytevalue(item, &value); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 808 | Py_DECREF(item); |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 809 | if (!rc) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 810 | goto error; |
| 811 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 812 | /* Append the byte */ |
| 813 | if (Py_SIZE(self) < self->ob_alloc) |
| 814 | Py_SIZE(self)++; |
| 815 | else if (PyByteArray_Resize((PyObject *)self, Py_SIZE(self)+1) < 0) |
| 816 | goto error; |
| 817 | self->ob_bytes[Py_SIZE(self)-1] = value; |
| 818 | } |
| 819 | |
| 820 | /* Clean up and return success */ |
| 821 | Py_DECREF(it); |
| 822 | return 0; |
| 823 | |
| 824 | error: |
| 825 | /* Error handling when it != NULL */ |
| 826 | Py_DECREF(it); |
| 827 | return -1; |
| 828 | } |
| 829 | |
| 830 | /* Mostly copied from string_repr, but without the |
| 831 | "smart quote" functionality. */ |
| 832 | static PyObject * |
| 833 | bytes_repr(PyByteArrayObject *self) |
| 834 | { |
| 835 | static const char *hexdigits = "0123456789abcdef"; |
| 836 | const char *quote_prefix = "bytearray(b"; |
| 837 | const char *quote_postfix = ")"; |
| 838 | Py_ssize_t length = Py_SIZE(self); |
| 839 | /* 14 == strlen(quote_prefix) + 2 + strlen(quote_postfix) */ |
| 840 | size_t newsize = 14 + 4 * length; |
| 841 | PyObject *v; |
| 842 | if (newsize > PY_SSIZE_T_MAX || newsize / 4 - 3 != length) { |
| 843 | PyErr_SetString(PyExc_OverflowError, |
| 844 | "bytearray object is too large to make repr"); |
| 845 | return NULL; |
| 846 | } |
| 847 | v = PyUnicode_FromUnicode(NULL, newsize); |
| 848 | if (v == NULL) { |
| 849 | return NULL; |
| 850 | } |
| 851 | else { |
| 852 | register Py_ssize_t i; |
| 853 | register Py_UNICODE c; |
| 854 | register Py_UNICODE *p; |
| 855 | int quote; |
| 856 | |
| 857 | /* Figure out which quote to use; single is preferred */ |
| 858 | quote = '\''; |
| 859 | { |
| 860 | char *test, *start; |
| 861 | start = PyByteArray_AS_STRING(self); |
| 862 | for (test = start; test < start+length; ++test) { |
| 863 | if (*test == '"') { |
| 864 | quote = '\''; /* back to single */ |
| 865 | goto decided; |
| 866 | } |
| 867 | else if (*test == '\'') |
| 868 | quote = '"'; |
| 869 | } |
| 870 | decided: |
| 871 | ; |
| 872 | } |
| 873 | |
| 874 | p = PyUnicode_AS_UNICODE(v); |
| 875 | while (*quote_prefix) |
| 876 | *p++ = *quote_prefix++; |
| 877 | *p++ = quote; |
| 878 | |
| 879 | for (i = 0; i < length; i++) { |
| 880 | /* There's at least enough room for a hex escape |
| 881 | and a closing quote. */ |
| 882 | assert(newsize - (p - PyUnicode_AS_UNICODE(v)) >= 5); |
| 883 | c = self->ob_bytes[i]; |
| 884 | if (c == '\'' || c == '\\') |
| 885 | *p++ = '\\', *p++ = c; |
| 886 | else if (c == '\t') |
| 887 | *p++ = '\\', *p++ = 't'; |
| 888 | else if (c == '\n') |
| 889 | *p++ = '\\', *p++ = 'n'; |
| 890 | else if (c == '\r') |
| 891 | *p++ = '\\', *p++ = 'r'; |
| 892 | else if (c == 0) |
| 893 | *p++ = '\\', *p++ = 'x', *p++ = '0', *p++ = '0'; |
| 894 | else if (c < ' ' || c >= 0x7f) { |
| 895 | *p++ = '\\'; |
| 896 | *p++ = 'x'; |
| 897 | *p++ = hexdigits[(c & 0xf0) >> 4]; |
| 898 | *p++ = hexdigits[c & 0xf]; |
| 899 | } |
| 900 | else |
| 901 | *p++ = c; |
| 902 | } |
| 903 | assert(newsize - (p - PyUnicode_AS_UNICODE(v)) >= 1); |
| 904 | *p++ = quote; |
| 905 | while (*quote_postfix) { |
| 906 | *p++ = *quote_postfix++; |
| 907 | } |
| 908 | *p = '\0'; |
| 909 | if (PyUnicode_Resize(&v, (p - PyUnicode_AS_UNICODE(v)))) { |
| 910 | Py_DECREF(v); |
| 911 | return NULL; |
| 912 | } |
| 913 | return v; |
| 914 | } |
| 915 | } |
| 916 | |
| 917 | static PyObject * |
| 918 | bytes_str(PyObject *op) |
| 919 | { |
| 920 | if (Py_BytesWarningFlag) { |
| 921 | if (PyErr_WarnEx(PyExc_BytesWarning, |
| 922 | "str() on a bytearray instance", 1)) |
| 923 | return NULL; |
| 924 | } |
| 925 | return bytes_repr((PyByteArrayObject*)op); |
| 926 | } |
| 927 | |
| 928 | static PyObject * |
| 929 | bytes_richcompare(PyObject *self, PyObject *other, int op) |
| 930 | { |
| 931 | Py_ssize_t self_size, other_size; |
| 932 | Py_buffer self_bytes, other_bytes; |
| 933 | PyObject *res; |
| 934 | Py_ssize_t minsize; |
| 935 | int cmp; |
| 936 | |
| 937 | /* Bytes can be compared to anything that supports the (binary) |
| 938 | buffer API. Except that a comparison with Unicode is always an |
| 939 | error, even if the comparison is for equality. */ |
| 940 | if (PyObject_IsInstance(self, (PyObject*)&PyUnicode_Type) || |
| 941 | PyObject_IsInstance(other, (PyObject*)&PyUnicode_Type)) { |
Barry Warsaw | 9e9dcd6 | 2008-10-17 01:50:37 +0000 | [diff] [blame] | 942 | if (Py_BytesWarningFlag && (op == Py_EQ || op == Py_NE)) { |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 943 | if (PyErr_WarnEx(PyExc_BytesWarning, |
Georg Brandl | e5d68ac | 2008-06-04 11:30:26 +0000 | [diff] [blame] | 944 | "Comparison between bytearray and string", 1)) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 945 | return NULL; |
| 946 | } |
| 947 | |
| 948 | Py_INCREF(Py_NotImplemented); |
| 949 | return Py_NotImplemented; |
| 950 | } |
| 951 | |
| 952 | self_size = _getbuffer(self, &self_bytes); |
| 953 | if (self_size < 0) { |
| 954 | PyErr_Clear(); |
| 955 | Py_INCREF(Py_NotImplemented); |
| 956 | return Py_NotImplemented; |
| 957 | } |
| 958 | |
| 959 | other_size = _getbuffer(other, &other_bytes); |
| 960 | if (other_size < 0) { |
| 961 | PyErr_Clear(); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 962 | PyBuffer_Release(&self_bytes); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 963 | Py_INCREF(Py_NotImplemented); |
| 964 | return Py_NotImplemented; |
| 965 | } |
| 966 | |
| 967 | if (self_size != other_size && (op == Py_EQ || op == Py_NE)) { |
| 968 | /* Shortcut: if the lengths differ, the objects differ */ |
| 969 | cmp = (op == Py_NE); |
| 970 | } |
| 971 | else { |
| 972 | minsize = self_size; |
| 973 | if (other_size < minsize) |
| 974 | minsize = other_size; |
| 975 | |
| 976 | cmp = memcmp(self_bytes.buf, other_bytes.buf, minsize); |
| 977 | /* In ISO C, memcmp() guarantees to use unsigned bytes! */ |
| 978 | |
| 979 | if (cmp == 0) { |
| 980 | if (self_size < other_size) |
| 981 | cmp = -1; |
| 982 | else if (self_size > other_size) |
| 983 | cmp = 1; |
| 984 | } |
| 985 | |
| 986 | switch (op) { |
| 987 | case Py_LT: cmp = cmp < 0; break; |
| 988 | case Py_LE: cmp = cmp <= 0; break; |
| 989 | case Py_EQ: cmp = cmp == 0; break; |
| 990 | case Py_NE: cmp = cmp != 0; break; |
| 991 | case Py_GT: cmp = cmp > 0; break; |
| 992 | case Py_GE: cmp = cmp >= 0; break; |
| 993 | } |
| 994 | } |
| 995 | |
| 996 | res = cmp ? Py_True : Py_False; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 997 | PyBuffer_Release(&self_bytes); |
| 998 | PyBuffer_Release(&other_bytes); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 999 | Py_INCREF(res); |
| 1000 | return res; |
| 1001 | } |
| 1002 | |
| 1003 | static void |
| 1004 | bytes_dealloc(PyByteArrayObject *self) |
| 1005 | { |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 1006 | if (self->ob_exports > 0) { |
| 1007 | PyErr_SetString(PyExc_SystemError, |
| 1008 | "deallocated bytearray object has exported buffers"); |
| 1009 | PyErr_Print(); |
| 1010 | } |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1011 | if (self->ob_bytes != 0) { |
| 1012 | PyMem_Free(self->ob_bytes); |
| 1013 | } |
| 1014 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 1015 | } |
| 1016 | |
| 1017 | |
| 1018 | /* -------------------------------------------------------------------- */ |
| 1019 | /* Methods */ |
| 1020 | |
| 1021 | #define STRINGLIB_CHAR char |
| 1022 | #define STRINGLIB_CMP memcmp |
| 1023 | #define STRINGLIB_LEN PyByteArray_GET_SIZE |
| 1024 | #define STRINGLIB_STR PyByteArray_AS_STRING |
| 1025 | #define STRINGLIB_NEW PyByteArray_FromStringAndSize |
| 1026 | #define STRINGLIB_EMPTY nullbytes |
| 1027 | #define STRINGLIB_CHECK_EXACT PyByteArray_CheckExact |
| 1028 | #define STRINGLIB_MUTABLE 1 |
Benjamin Peterson | a786b02 | 2008-08-25 21:05:21 +0000 | [diff] [blame] | 1029 | #define FROM_BYTEARRAY 1 |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1030 | |
| 1031 | #include "stringlib/fastsearch.h" |
| 1032 | #include "stringlib/count.h" |
| 1033 | #include "stringlib/find.h" |
| 1034 | #include "stringlib/partition.h" |
| 1035 | #include "stringlib/ctype.h" |
| 1036 | #include "stringlib/transmogrify.h" |
| 1037 | |
| 1038 | |
| 1039 | /* The following Py_LOCAL_INLINE and Py_LOCAL functions |
| 1040 | were copied from the old char* style string object. */ |
| 1041 | |
| 1042 | Py_LOCAL_INLINE(void) |
| 1043 | _adjust_indices(Py_ssize_t *start, Py_ssize_t *end, Py_ssize_t len) |
| 1044 | { |
| 1045 | if (*end > len) |
| 1046 | *end = len; |
| 1047 | else if (*end < 0) |
| 1048 | *end += len; |
| 1049 | if (*end < 0) |
| 1050 | *end = 0; |
| 1051 | if (*start < 0) |
| 1052 | *start += len; |
| 1053 | if (*start < 0) |
| 1054 | *start = 0; |
| 1055 | } |
| 1056 | |
| 1057 | |
| 1058 | Py_LOCAL_INLINE(Py_ssize_t) |
| 1059 | bytes_find_internal(PyByteArrayObject *self, PyObject *args, int dir) |
| 1060 | { |
| 1061 | PyObject *subobj; |
| 1062 | Py_buffer subbuf; |
| 1063 | Py_ssize_t start=0, end=PY_SSIZE_T_MAX; |
| 1064 | Py_ssize_t res; |
| 1065 | |
| 1066 | if (!PyArg_ParseTuple(args, "O|O&O&:find/rfind/index/rindex", &subobj, |
| 1067 | _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end)) |
| 1068 | return -2; |
| 1069 | if (_getbuffer(subobj, &subbuf) < 0) |
| 1070 | return -2; |
| 1071 | if (dir > 0) |
| 1072 | res = stringlib_find_slice( |
| 1073 | PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), |
| 1074 | subbuf.buf, subbuf.len, start, end); |
| 1075 | else |
| 1076 | res = stringlib_rfind_slice( |
| 1077 | PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), |
| 1078 | subbuf.buf, subbuf.len, start, end); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 1079 | PyBuffer_Release(&subbuf); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1080 | return res; |
| 1081 | } |
| 1082 | |
| 1083 | PyDoc_STRVAR(find__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1084 | "B.find(sub[, start[, end]]) -> int\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1085 | \n\ |
| 1086 | Return the lowest index in B where subsection sub is found,\n\ |
| 1087 | such that sub is contained within s[start,end]. Optional\n\ |
| 1088 | arguments start and end are interpreted as in slice notation.\n\ |
| 1089 | \n\ |
| 1090 | Return -1 on failure."); |
| 1091 | |
| 1092 | static PyObject * |
| 1093 | bytes_find(PyByteArrayObject *self, PyObject *args) |
| 1094 | { |
| 1095 | Py_ssize_t result = bytes_find_internal(self, args, +1); |
| 1096 | if (result == -2) |
| 1097 | return NULL; |
| 1098 | return PyLong_FromSsize_t(result); |
| 1099 | } |
| 1100 | |
| 1101 | PyDoc_STRVAR(count__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1102 | "B.count(sub[, start[, end]]) -> int\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1103 | \n\ |
| 1104 | Return the number of non-overlapping occurrences of subsection sub in\n\ |
| 1105 | bytes B[start:end]. Optional arguments start and end are interpreted\n\ |
| 1106 | as in slice notation."); |
| 1107 | |
| 1108 | static PyObject * |
| 1109 | bytes_count(PyByteArrayObject *self, PyObject *args) |
| 1110 | { |
| 1111 | PyObject *sub_obj; |
| 1112 | const char *str = PyByteArray_AS_STRING(self); |
| 1113 | Py_ssize_t start = 0, end = PY_SSIZE_T_MAX; |
| 1114 | Py_buffer vsub; |
| 1115 | PyObject *count_obj; |
| 1116 | |
| 1117 | if (!PyArg_ParseTuple(args, "O|O&O&:count", &sub_obj, |
| 1118 | _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end)) |
| 1119 | return NULL; |
| 1120 | |
| 1121 | if (_getbuffer(sub_obj, &vsub) < 0) |
| 1122 | return NULL; |
| 1123 | |
| 1124 | _adjust_indices(&start, &end, PyByteArray_GET_SIZE(self)); |
| 1125 | |
| 1126 | count_obj = PyLong_FromSsize_t( |
| 1127 | stringlib_count(str + start, end - start, vsub.buf, vsub.len) |
| 1128 | ); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 1129 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1130 | return count_obj; |
| 1131 | } |
| 1132 | |
| 1133 | |
| 1134 | PyDoc_STRVAR(index__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1135 | "B.index(sub[, start[, end]]) -> int\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1136 | \n\ |
| 1137 | Like B.find() but raise ValueError when the subsection is not found."); |
| 1138 | |
| 1139 | static PyObject * |
| 1140 | bytes_index(PyByteArrayObject *self, PyObject *args) |
| 1141 | { |
| 1142 | Py_ssize_t result = bytes_find_internal(self, args, +1); |
| 1143 | if (result == -2) |
| 1144 | return NULL; |
| 1145 | if (result == -1) { |
| 1146 | PyErr_SetString(PyExc_ValueError, |
| 1147 | "subsection not found"); |
| 1148 | return NULL; |
| 1149 | } |
| 1150 | return PyLong_FromSsize_t(result); |
| 1151 | } |
| 1152 | |
| 1153 | |
| 1154 | PyDoc_STRVAR(rfind__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1155 | "B.rfind(sub[, start[, end]]) -> int\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1156 | \n\ |
| 1157 | Return the highest index in B where subsection sub is found,\n\ |
| 1158 | such that sub is contained within s[start,end]. Optional\n\ |
| 1159 | arguments start and end are interpreted as in slice notation.\n\ |
| 1160 | \n\ |
| 1161 | Return -1 on failure."); |
| 1162 | |
| 1163 | static PyObject * |
| 1164 | bytes_rfind(PyByteArrayObject *self, PyObject *args) |
| 1165 | { |
| 1166 | Py_ssize_t result = bytes_find_internal(self, args, -1); |
| 1167 | if (result == -2) |
| 1168 | return NULL; |
| 1169 | return PyLong_FromSsize_t(result); |
| 1170 | } |
| 1171 | |
| 1172 | |
| 1173 | PyDoc_STRVAR(rindex__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1174 | "B.rindex(sub[, start[, end]]) -> int\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1175 | \n\ |
| 1176 | Like B.rfind() but raise ValueError when the subsection is not found."); |
| 1177 | |
| 1178 | static PyObject * |
| 1179 | bytes_rindex(PyByteArrayObject *self, PyObject *args) |
| 1180 | { |
| 1181 | Py_ssize_t result = bytes_find_internal(self, args, -1); |
| 1182 | if (result == -2) |
| 1183 | return NULL; |
| 1184 | if (result == -1) { |
| 1185 | PyErr_SetString(PyExc_ValueError, |
| 1186 | "subsection not found"); |
| 1187 | return NULL; |
| 1188 | } |
| 1189 | return PyLong_FromSsize_t(result); |
| 1190 | } |
| 1191 | |
| 1192 | |
| 1193 | static int |
| 1194 | bytes_contains(PyObject *self, PyObject *arg) |
| 1195 | { |
| 1196 | Py_ssize_t ival = PyNumber_AsSsize_t(arg, PyExc_ValueError); |
| 1197 | if (ival == -1 && PyErr_Occurred()) { |
| 1198 | Py_buffer varg; |
| 1199 | int pos; |
| 1200 | PyErr_Clear(); |
| 1201 | if (_getbuffer(arg, &varg) < 0) |
| 1202 | return -1; |
| 1203 | pos = stringlib_find(PyByteArray_AS_STRING(self), Py_SIZE(self), |
| 1204 | varg.buf, varg.len, 0); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 1205 | PyBuffer_Release(&varg); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1206 | return pos >= 0; |
| 1207 | } |
| 1208 | if (ival < 0 || ival >= 256) { |
| 1209 | PyErr_SetString(PyExc_ValueError, "byte must be in range(0, 256)"); |
| 1210 | return -1; |
| 1211 | } |
| 1212 | |
| 1213 | return memchr(PyByteArray_AS_STRING(self), ival, Py_SIZE(self)) != NULL; |
| 1214 | } |
| 1215 | |
| 1216 | |
| 1217 | /* Matches the end (direction >= 0) or start (direction < 0) of self |
| 1218 | * against substr, using the start and end arguments. Returns |
| 1219 | * -1 on error, 0 if not found and 1 if found. |
| 1220 | */ |
| 1221 | Py_LOCAL(int) |
| 1222 | _bytes_tailmatch(PyByteArrayObject *self, PyObject *substr, Py_ssize_t start, |
| 1223 | Py_ssize_t end, int direction) |
| 1224 | { |
| 1225 | Py_ssize_t len = PyByteArray_GET_SIZE(self); |
| 1226 | const char* str; |
| 1227 | Py_buffer vsubstr; |
| 1228 | int rv = 0; |
| 1229 | |
| 1230 | str = PyByteArray_AS_STRING(self); |
| 1231 | |
| 1232 | if (_getbuffer(substr, &vsubstr) < 0) |
| 1233 | return -1; |
| 1234 | |
| 1235 | _adjust_indices(&start, &end, len); |
| 1236 | |
| 1237 | if (direction < 0) { |
| 1238 | /* startswith */ |
| 1239 | if (start+vsubstr.len > len) { |
| 1240 | goto done; |
| 1241 | } |
| 1242 | } else { |
| 1243 | /* endswith */ |
| 1244 | if (end-start < vsubstr.len || start > len) { |
| 1245 | goto done; |
| 1246 | } |
| 1247 | |
| 1248 | if (end-vsubstr.len > start) |
| 1249 | start = end - vsubstr.len; |
| 1250 | } |
| 1251 | if (end-start >= vsubstr.len) |
| 1252 | rv = ! memcmp(str+start, vsubstr.buf, vsubstr.len); |
| 1253 | |
| 1254 | done: |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 1255 | PyBuffer_Release(&vsubstr); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1256 | return rv; |
| 1257 | } |
| 1258 | |
| 1259 | |
| 1260 | PyDoc_STRVAR(startswith__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1261 | "B.startswith(prefix[, start[, end]]) -> bool\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1262 | \n\ |
| 1263 | Return True if B starts with the specified prefix, False otherwise.\n\ |
| 1264 | With optional start, test B beginning at that position.\n\ |
| 1265 | With optional end, stop comparing B at that position.\n\ |
| 1266 | prefix can also be a tuple of strings to try."); |
| 1267 | |
| 1268 | static PyObject * |
| 1269 | bytes_startswith(PyByteArrayObject *self, PyObject *args) |
| 1270 | { |
| 1271 | Py_ssize_t start = 0; |
| 1272 | Py_ssize_t end = PY_SSIZE_T_MAX; |
| 1273 | PyObject *subobj; |
| 1274 | int result; |
| 1275 | |
| 1276 | if (!PyArg_ParseTuple(args, "O|O&O&:startswith", &subobj, |
| 1277 | _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end)) |
| 1278 | return NULL; |
| 1279 | if (PyTuple_Check(subobj)) { |
| 1280 | Py_ssize_t i; |
| 1281 | for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) { |
| 1282 | result = _bytes_tailmatch(self, |
| 1283 | PyTuple_GET_ITEM(subobj, i), |
| 1284 | start, end, -1); |
| 1285 | if (result == -1) |
| 1286 | return NULL; |
| 1287 | else if (result) { |
| 1288 | Py_RETURN_TRUE; |
| 1289 | } |
| 1290 | } |
| 1291 | Py_RETURN_FALSE; |
| 1292 | } |
| 1293 | result = _bytes_tailmatch(self, subobj, start, end, -1); |
| 1294 | if (result == -1) |
| 1295 | return NULL; |
| 1296 | else |
| 1297 | return PyBool_FromLong(result); |
| 1298 | } |
| 1299 | |
| 1300 | PyDoc_STRVAR(endswith__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1301 | "B.endswith(suffix[, start[, end]]) -> bool\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1302 | \n\ |
| 1303 | Return True if B ends with the specified suffix, False otherwise.\n\ |
| 1304 | With optional start, test B beginning at that position.\n\ |
| 1305 | With optional end, stop comparing B at that position.\n\ |
| 1306 | suffix can also be a tuple of strings to try."); |
| 1307 | |
| 1308 | static PyObject * |
| 1309 | bytes_endswith(PyByteArrayObject *self, PyObject *args) |
| 1310 | { |
| 1311 | Py_ssize_t start = 0; |
| 1312 | Py_ssize_t end = PY_SSIZE_T_MAX; |
| 1313 | PyObject *subobj; |
| 1314 | int result; |
| 1315 | |
| 1316 | if (!PyArg_ParseTuple(args, "O|O&O&:endswith", &subobj, |
| 1317 | _PyEval_SliceIndex, &start, _PyEval_SliceIndex, &end)) |
| 1318 | return NULL; |
| 1319 | if (PyTuple_Check(subobj)) { |
| 1320 | Py_ssize_t i; |
| 1321 | for (i = 0; i < PyTuple_GET_SIZE(subobj); i++) { |
| 1322 | result = _bytes_tailmatch(self, |
| 1323 | PyTuple_GET_ITEM(subobj, i), |
| 1324 | start, end, +1); |
| 1325 | if (result == -1) |
| 1326 | return NULL; |
| 1327 | else if (result) { |
| 1328 | Py_RETURN_TRUE; |
| 1329 | } |
| 1330 | } |
| 1331 | Py_RETURN_FALSE; |
| 1332 | } |
| 1333 | result = _bytes_tailmatch(self, subobj, start, end, +1); |
| 1334 | if (result == -1) |
| 1335 | return NULL; |
| 1336 | else |
| 1337 | return PyBool_FromLong(result); |
| 1338 | } |
| 1339 | |
| 1340 | |
| 1341 | PyDoc_STRVAR(translate__doc__, |
| 1342 | "B.translate(table[, deletechars]) -> bytearray\n\ |
| 1343 | \n\ |
| 1344 | Return a copy of B, where all characters occurring in the\n\ |
| 1345 | optional argument deletechars are removed, and the remaining\n\ |
| 1346 | characters have been mapped through the given translation\n\ |
| 1347 | table, which must be a bytes object of length 256."); |
| 1348 | |
| 1349 | static PyObject * |
| 1350 | bytes_translate(PyByteArrayObject *self, PyObject *args) |
| 1351 | { |
| 1352 | register char *input, *output; |
| 1353 | register const char *table; |
Benjamin Peterson | 0f3641c | 2008-11-19 22:05:52 +0000 | [diff] [blame] | 1354 | register Py_ssize_t i, c; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1355 | PyObject *input_obj = (PyObject*)self; |
| 1356 | const char *output_start; |
| 1357 | Py_ssize_t inlen; |
| 1358 | PyObject *result; |
| 1359 | int trans_table[256]; |
| 1360 | PyObject *tableobj, *delobj = NULL; |
| 1361 | Py_buffer vtable, vdel; |
| 1362 | |
| 1363 | if (!PyArg_UnpackTuple(args, "translate", 1, 2, |
| 1364 | &tableobj, &delobj)) |
| 1365 | return NULL; |
| 1366 | |
| 1367 | if (_getbuffer(tableobj, &vtable) < 0) |
| 1368 | return NULL; |
| 1369 | |
| 1370 | if (vtable.len != 256) { |
| 1371 | PyErr_SetString(PyExc_ValueError, |
| 1372 | "translation table must be 256 characters long"); |
| 1373 | result = NULL; |
| 1374 | goto done; |
| 1375 | } |
| 1376 | |
| 1377 | if (delobj != NULL) { |
| 1378 | if (_getbuffer(delobj, &vdel) < 0) { |
| 1379 | result = NULL; |
| 1380 | goto done; |
| 1381 | } |
| 1382 | } |
| 1383 | else { |
| 1384 | vdel.buf = NULL; |
| 1385 | vdel.len = 0; |
| 1386 | } |
| 1387 | |
| 1388 | table = (const char *)vtable.buf; |
| 1389 | inlen = PyByteArray_GET_SIZE(input_obj); |
| 1390 | result = PyByteArray_FromStringAndSize((char *)NULL, inlen); |
| 1391 | if (result == NULL) |
| 1392 | goto done; |
| 1393 | output_start = output = PyByteArray_AsString(result); |
| 1394 | input = PyByteArray_AS_STRING(input_obj); |
| 1395 | |
| 1396 | if (vdel.len == 0) { |
| 1397 | /* If no deletions are required, use faster code */ |
| 1398 | for (i = inlen; --i >= 0; ) { |
| 1399 | c = Py_CHARMASK(*input++); |
Benjamin Peterson | 0f3641c | 2008-11-19 22:05:52 +0000 | [diff] [blame] | 1400 | *output++ = table[c]; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1401 | } |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1402 | goto done; |
| 1403 | } |
| 1404 | |
| 1405 | for (i = 0; i < 256; i++) |
| 1406 | trans_table[i] = Py_CHARMASK(table[i]); |
| 1407 | |
| 1408 | for (i = 0; i < vdel.len; i++) |
| 1409 | trans_table[(int) Py_CHARMASK( ((unsigned char*)vdel.buf)[i] )] = -1; |
| 1410 | |
| 1411 | for (i = inlen; --i >= 0; ) { |
| 1412 | c = Py_CHARMASK(*input++); |
| 1413 | if (trans_table[c] != -1) |
| 1414 | if (Py_CHARMASK(*output++ = (char)trans_table[c]) == c) |
| 1415 | continue; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1416 | } |
| 1417 | /* Fix the size of the resulting string */ |
| 1418 | if (inlen > 0) |
| 1419 | PyByteArray_Resize(result, output - output_start); |
| 1420 | |
| 1421 | done: |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 1422 | PyBuffer_Release(&vtable); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1423 | if (delobj != NULL) |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 1424 | PyBuffer_Release(&vdel); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1425 | return result; |
| 1426 | } |
| 1427 | |
| 1428 | |
| 1429 | #define FORWARD 1 |
| 1430 | #define REVERSE -1 |
| 1431 | |
| 1432 | /* find and count characters and substrings */ |
| 1433 | |
| 1434 | #define findchar(target, target_len, c) \ |
| 1435 | ((char *)memchr((const void *)(target), c, target_len)) |
| 1436 | |
| 1437 | /* Don't call if length < 2 */ |
| 1438 | #define Py_STRING_MATCH(target, offset, pattern, length) \ |
| 1439 | (target[offset] == pattern[0] && \ |
| 1440 | target[offset+length-1] == pattern[length-1] && \ |
| 1441 | !memcmp(target+offset+1, pattern+1, length-2) ) |
| 1442 | |
| 1443 | |
Benjamin Peterson | 0f3641c | 2008-11-19 22:05:52 +0000 | [diff] [blame] | 1444 | /* Bytes ops must return a string, create a copy */ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1445 | Py_LOCAL(PyByteArrayObject *) |
| 1446 | return_self(PyByteArrayObject *self) |
| 1447 | { |
Georg Brandl | 1e7217d | 2008-05-30 12:02:38 +0000 | [diff] [blame] | 1448 | /* always return a new bytearray */ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1449 | return (PyByteArrayObject *)PyByteArray_FromStringAndSize( |
| 1450 | PyByteArray_AS_STRING(self), |
| 1451 | PyByteArray_GET_SIZE(self)); |
| 1452 | } |
| 1453 | |
| 1454 | Py_LOCAL_INLINE(Py_ssize_t) |
| 1455 | countchar(const char *target, Py_ssize_t target_len, char c, Py_ssize_t maxcount) |
| 1456 | { |
| 1457 | Py_ssize_t count=0; |
| 1458 | const char *start=target; |
| 1459 | const char *end=target+target_len; |
| 1460 | |
| 1461 | while ( (start=findchar(start, end-start, c)) != NULL ) { |
| 1462 | count++; |
| 1463 | if (count >= maxcount) |
| 1464 | break; |
| 1465 | start += 1; |
| 1466 | } |
| 1467 | return count; |
| 1468 | } |
| 1469 | |
| 1470 | Py_LOCAL(Py_ssize_t) |
| 1471 | findstring(const char *target, Py_ssize_t target_len, |
| 1472 | const char *pattern, Py_ssize_t pattern_len, |
| 1473 | Py_ssize_t start, |
| 1474 | Py_ssize_t end, |
| 1475 | int direction) |
| 1476 | { |
| 1477 | if (start < 0) { |
| 1478 | start += target_len; |
| 1479 | if (start < 0) |
| 1480 | start = 0; |
| 1481 | } |
| 1482 | if (end > target_len) { |
| 1483 | end = target_len; |
| 1484 | } else if (end < 0) { |
| 1485 | end += target_len; |
| 1486 | if (end < 0) |
| 1487 | end = 0; |
| 1488 | } |
| 1489 | |
| 1490 | /* zero-length substrings always match at the first attempt */ |
| 1491 | if (pattern_len == 0) |
| 1492 | return (direction > 0) ? start : end; |
| 1493 | |
| 1494 | end -= pattern_len; |
| 1495 | |
| 1496 | if (direction < 0) { |
| 1497 | for (; end >= start; end--) |
| 1498 | if (Py_STRING_MATCH(target, end, pattern, pattern_len)) |
| 1499 | return end; |
| 1500 | } else { |
| 1501 | for (; start <= end; start++) |
| 1502 | if (Py_STRING_MATCH(target, start, pattern, pattern_len)) |
| 1503 | return start; |
| 1504 | } |
| 1505 | return -1; |
| 1506 | } |
| 1507 | |
| 1508 | Py_LOCAL_INLINE(Py_ssize_t) |
| 1509 | countstring(const char *target, Py_ssize_t target_len, |
| 1510 | const char *pattern, Py_ssize_t pattern_len, |
| 1511 | Py_ssize_t start, |
| 1512 | Py_ssize_t end, |
| 1513 | int direction, Py_ssize_t maxcount) |
| 1514 | { |
| 1515 | Py_ssize_t count=0; |
| 1516 | |
| 1517 | if (start < 0) { |
| 1518 | start += target_len; |
| 1519 | if (start < 0) |
| 1520 | start = 0; |
| 1521 | } |
| 1522 | if (end > target_len) { |
| 1523 | end = target_len; |
| 1524 | } else if (end < 0) { |
| 1525 | end += target_len; |
| 1526 | if (end < 0) |
| 1527 | end = 0; |
| 1528 | } |
| 1529 | |
| 1530 | /* zero-length substrings match everywhere */ |
| 1531 | if (pattern_len == 0 || maxcount == 0) { |
| 1532 | if (target_len+1 < maxcount) |
| 1533 | return target_len+1; |
| 1534 | return maxcount; |
| 1535 | } |
| 1536 | |
| 1537 | end -= pattern_len; |
| 1538 | if (direction < 0) { |
| 1539 | for (; (end >= start); end--) |
| 1540 | if (Py_STRING_MATCH(target, end, pattern, pattern_len)) { |
| 1541 | count++; |
| 1542 | if (--maxcount <= 0) break; |
| 1543 | end -= pattern_len-1; |
| 1544 | } |
| 1545 | } else { |
| 1546 | for (; (start <= end); start++) |
| 1547 | if (Py_STRING_MATCH(target, start, pattern, pattern_len)) { |
| 1548 | count++; |
| 1549 | if (--maxcount <= 0) |
| 1550 | break; |
| 1551 | start += pattern_len-1; |
| 1552 | } |
| 1553 | } |
| 1554 | return count; |
| 1555 | } |
| 1556 | |
| 1557 | |
| 1558 | /* Algorithms for different cases of string replacement */ |
| 1559 | |
| 1560 | /* len(self)>=1, from="", len(to)>=1, maxcount>=1 */ |
| 1561 | Py_LOCAL(PyByteArrayObject *) |
| 1562 | replace_interleave(PyByteArrayObject *self, |
| 1563 | const char *to_s, Py_ssize_t to_len, |
| 1564 | Py_ssize_t maxcount) |
| 1565 | { |
| 1566 | char *self_s, *result_s; |
| 1567 | Py_ssize_t self_len, result_len; |
| 1568 | Py_ssize_t count, i, product; |
| 1569 | PyByteArrayObject *result; |
| 1570 | |
| 1571 | self_len = PyByteArray_GET_SIZE(self); |
| 1572 | |
| 1573 | /* 1 at the end plus 1 after every character */ |
| 1574 | count = self_len+1; |
| 1575 | if (maxcount < count) |
| 1576 | count = maxcount; |
| 1577 | |
| 1578 | /* Check for overflow */ |
| 1579 | /* result_len = count * to_len + self_len; */ |
| 1580 | product = count * to_len; |
| 1581 | if (product / to_len != count) { |
| 1582 | PyErr_SetString(PyExc_OverflowError, |
| 1583 | "replace string is too long"); |
| 1584 | return NULL; |
| 1585 | } |
| 1586 | result_len = product + self_len; |
| 1587 | if (result_len < 0) { |
| 1588 | PyErr_SetString(PyExc_OverflowError, |
| 1589 | "replace string is too long"); |
| 1590 | return NULL; |
| 1591 | } |
| 1592 | |
| 1593 | if (! (result = (PyByteArrayObject *) |
| 1594 | PyByteArray_FromStringAndSize(NULL, result_len)) ) |
| 1595 | return NULL; |
| 1596 | |
| 1597 | self_s = PyByteArray_AS_STRING(self); |
| 1598 | result_s = PyByteArray_AS_STRING(result); |
| 1599 | |
| 1600 | /* TODO: special case single character, which doesn't need memcpy */ |
| 1601 | |
| 1602 | /* Lay the first one down (guaranteed this will occur) */ |
| 1603 | Py_MEMCPY(result_s, to_s, to_len); |
| 1604 | result_s += to_len; |
| 1605 | count -= 1; |
| 1606 | |
| 1607 | for (i=0; i<count; i++) { |
| 1608 | *result_s++ = *self_s++; |
| 1609 | Py_MEMCPY(result_s, to_s, to_len); |
| 1610 | result_s += to_len; |
| 1611 | } |
| 1612 | |
| 1613 | /* Copy the rest of the original string */ |
| 1614 | Py_MEMCPY(result_s, self_s, self_len-i); |
| 1615 | |
| 1616 | return result; |
| 1617 | } |
| 1618 | |
| 1619 | /* Special case for deleting a single character */ |
| 1620 | /* len(self)>=1, len(from)==1, to="", maxcount>=1 */ |
| 1621 | Py_LOCAL(PyByteArrayObject *) |
| 1622 | replace_delete_single_character(PyByteArrayObject *self, |
| 1623 | char from_c, Py_ssize_t maxcount) |
| 1624 | { |
| 1625 | char *self_s, *result_s; |
| 1626 | char *start, *next, *end; |
| 1627 | Py_ssize_t self_len, result_len; |
| 1628 | Py_ssize_t count; |
| 1629 | PyByteArrayObject *result; |
| 1630 | |
| 1631 | self_len = PyByteArray_GET_SIZE(self); |
| 1632 | self_s = PyByteArray_AS_STRING(self); |
| 1633 | |
| 1634 | count = countchar(self_s, self_len, from_c, maxcount); |
| 1635 | if (count == 0) { |
| 1636 | return return_self(self); |
| 1637 | } |
| 1638 | |
| 1639 | result_len = self_len - count; /* from_len == 1 */ |
| 1640 | assert(result_len>=0); |
| 1641 | |
| 1642 | if ( (result = (PyByteArrayObject *) |
| 1643 | PyByteArray_FromStringAndSize(NULL, result_len)) == NULL) |
| 1644 | return NULL; |
| 1645 | result_s = PyByteArray_AS_STRING(result); |
| 1646 | |
| 1647 | start = self_s; |
| 1648 | end = self_s + self_len; |
| 1649 | while (count-- > 0) { |
| 1650 | next = findchar(start, end-start, from_c); |
| 1651 | if (next == NULL) |
| 1652 | break; |
| 1653 | Py_MEMCPY(result_s, start, next-start); |
| 1654 | result_s += (next-start); |
| 1655 | start = next+1; |
| 1656 | } |
| 1657 | Py_MEMCPY(result_s, start, end-start); |
| 1658 | |
| 1659 | return result; |
| 1660 | } |
| 1661 | |
| 1662 | /* len(self)>=1, len(from)>=2, to="", maxcount>=1 */ |
| 1663 | |
| 1664 | Py_LOCAL(PyByteArrayObject *) |
| 1665 | replace_delete_substring(PyByteArrayObject *self, |
| 1666 | const char *from_s, Py_ssize_t from_len, |
| 1667 | Py_ssize_t maxcount) |
| 1668 | { |
| 1669 | char *self_s, *result_s; |
| 1670 | char *start, *next, *end; |
| 1671 | Py_ssize_t self_len, result_len; |
| 1672 | Py_ssize_t count, offset; |
| 1673 | PyByteArrayObject *result; |
| 1674 | |
| 1675 | self_len = PyByteArray_GET_SIZE(self); |
| 1676 | self_s = PyByteArray_AS_STRING(self); |
| 1677 | |
| 1678 | count = countstring(self_s, self_len, |
| 1679 | from_s, from_len, |
| 1680 | 0, self_len, 1, |
| 1681 | maxcount); |
| 1682 | |
| 1683 | if (count == 0) { |
| 1684 | /* no matches */ |
| 1685 | return return_self(self); |
| 1686 | } |
| 1687 | |
| 1688 | result_len = self_len - (count * from_len); |
| 1689 | assert (result_len>=0); |
| 1690 | |
| 1691 | if ( (result = (PyByteArrayObject *) |
| 1692 | PyByteArray_FromStringAndSize(NULL, result_len)) == NULL ) |
| 1693 | return NULL; |
| 1694 | |
| 1695 | result_s = PyByteArray_AS_STRING(result); |
| 1696 | |
| 1697 | start = self_s; |
| 1698 | end = self_s + self_len; |
| 1699 | while (count-- > 0) { |
| 1700 | offset = findstring(start, end-start, |
| 1701 | from_s, from_len, |
| 1702 | 0, end-start, FORWARD); |
| 1703 | if (offset == -1) |
| 1704 | break; |
| 1705 | next = start + offset; |
| 1706 | |
| 1707 | Py_MEMCPY(result_s, start, next-start); |
| 1708 | |
| 1709 | result_s += (next-start); |
| 1710 | start = next+from_len; |
| 1711 | } |
| 1712 | Py_MEMCPY(result_s, start, end-start); |
| 1713 | return result; |
| 1714 | } |
| 1715 | |
| 1716 | /* len(self)>=1, len(from)==len(to)==1, maxcount>=1 */ |
| 1717 | Py_LOCAL(PyByteArrayObject *) |
| 1718 | replace_single_character_in_place(PyByteArrayObject *self, |
| 1719 | char from_c, char to_c, |
| 1720 | Py_ssize_t maxcount) |
| 1721 | { |
| 1722 | char *self_s, *result_s, *start, *end, *next; |
| 1723 | Py_ssize_t self_len; |
| 1724 | PyByteArrayObject *result; |
| 1725 | |
| 1726 | /* The result string will be the same size */ |
| 1727 | self_s = PyByteArray_AS_STRING(self); |
| 1728 | self_len = PyByteArray_GET_SIZE(self); |
| 1729 | |
| 1730 | next = findchar(self_s, self_len, from_c); |
| 1731 | |
| 1732 | if (next == NULL) { |
| 1733 | /* No matches; return the original bytes */ |
| 1734 | return return_self(self); |
| 1735 | } |
| 1736 | |
| 1737 | /* Need to make a new bytes */ |
| 1738 | result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len); |
| 1739 | if (result == NULL) |
| 1740 | return NULL; |
| 1741 | result_s = PyByteArray_AS_STRING(result); |
| 1742 | Py_MEMCPY(result_s, self_s, self_len); |
| 1743 | |
| 1744 | /* change everything in-place, starting with this one */ |
| 1745 | start = result_s + (next-self_s); |
| 1746 | *start = to_c; |
| 1747 | start++; |
| 1748 | end = result_s + self_len; |
| 1749 | |
| 1750 | while (--maxcount > 0) { |
| 1751 | next = findchar(start, end-start, from_c); |
| 1752 | if (next == NULL) |
| 1753 | break; |
| 1754 | *next = to_c; |
| 1755 | start = next+1; |
| 1756 | } |
| 1757 | |
| 1758 | return result; |
| 1759 | } |
| 1760 | |
| 1761 | /* len(self)>=1, len(from)==len(to)>=2, maxcount>=1 */ |
| 1762 | Py_LOCAL(PyByteArrayObject *) |
| 1763 | replace_substring_in_place(PyByteArrayObject *self, |
| 1764 | const char *from_s, Py_ssize_t from_len, |
| 1765 | const char *to_s, Py_ssize_t to_len, |
| 1766 | Py_ssize_t maxcount) |
| 1767 | { |
| 1768 | char *result_s, *start, *end; |
| 1769 | char *self_s; |
| 1770 | Py_ssize_t self_len, offset; |
| 1771 | PyByteArrayObject *result; |
| 1772 | |
| 1773 | /* The result bytes will be the same size */ |
| 1774 | |
| 1775 | self_s = PyByteArray_AS_STRING(self); |
| 1776 | self_len = PyByteArray_GET_SIZE(self); |
| 1777 | |
| 1778 | offset = findstring(self_s, self_len, |
| 1779 | from_s, from_len, |
| 1780 | 0, self_len, FORWARD); |
| 1781 | if (offset == -1) { |
| 1782 | /* No matches; return the original bytes */ |
| 1783 | return return_self(self); |
| 1784 | } |
| 1785 | |
| 1786 | /* Need to make a new bytes */ |
| 1787 | result = (PyByteArrayObject *) PyByteArray_FromStringAndSize(NULL, self_len); |
| 1788 | if (result == NULL) |
| 1789 | return NULL; |
| 1790 | result_s = PyByteArray_AS_STRING(result); |
| 1791 | Py_MEMCPY(result_s, self_s, self_len); |
| 1792 | |
| 1793 | /* change everything in-place, starting with this one */ |
| 1794 | start = result_s + offset; |
| 1795 | Py_MEMCPY(start, to_s, from_len); |
| 1796 | start += from_len; |
| 1797 | end = result_s + self_len; |
| 1798 | |
| 1799 | while ( --maxcount > 0) { |
| 1800 | offset = findstring(start, end-start, |
| 1801 | from_s, from_len, |
| 1802 | 0, end-start, FORWARD); |
| 1803 | if (offset==-1) |
| 1804 | break; |
| 1805 | Py_MEMCPY(start+offset, to_s, from_len); |
| 1806 | start += offset+from_len; |
| 1807 | } |
| 1808 | |
| 1809 | return result; |
| 1810 | } |
| 1811 | |
| 1812 | /* len(self)>=1, len(from)==1, len(to)>=2, maxcount>=1 */ |
| 1813 | Py_LOCAL(PyByteArrayObject *) |
| 1814 | replace_single_character(PyByteArrayObject *self, |
| 1815 | char from_c, |
| 1816 | const char *to_s, Py_ssize_t to_len, |
| 1817 | Py_ssize_t maxcount) |
| 1818 | { |
| 1819 | char *self_s, *result_s; |
| 1820 | char *start, *next, *end; |
| 1821 | Py_ssize_t self_len, result_len; |
| 1822 | Py_ssize_t count, product; |
| 1823 | PyByteArrayObject *result; |
| 1824 | |
| 1825 | self_s = PyByteArray_AS_STRING(self); |
| 1826 | self_len = PyByteArray_GET_SIZE(self); |
| 1827 | |
| 1828 | count = countchar(self_s, self_len, from_c, maxcount); |
| 1829 | if (count == 0) { |
| 1830 | /* no matches, return unchanged */ |
| 1831 | return return_self(self); |
| 1832 | } |
| 1833 | |
| 1834 | /* use the difference between current and new, hence the "-1" */ |
| 1835 | /* result_len = self_len + count * (to_len-1) */ |
| 1836 | product = count * (to_len-1); |
| 1837 | if (product / (to_len-1) != count) { |
| 1838 | PyErr_SetString(PyExc_OverflowError, "replace bytes is too long"); |
| 1839 | return NULL; |
| 1840 | } |
| 1841 | result_len = self_len + product; |
| 1842 | if (result_len < 0) { |
| 1843 | PyErr_SetString(PyExc_OverflowError, "replace bytes is too long"); |
| 1844 | return NULL; |
| 1845 | } |
| 1846 | |
| 1847 | if ( (result = (PyByteArrayObject *) |
| 1848 | PyByteArray_FromStringAndSize(NULL, result_len)) == NULL) |
| 1849 | return NULL; |
| 1850 | result_s = PyByteArray_AS_STRING(result); |
| 1851 | |
| 1852 | start = self_s; |
| 1853 | end = self_s + self_len; |
| 1854 | while (count-- > 0) { |
| 1855 | next = findchar(start, end-start, from_c); |
| 1856 | if (next == NULL) |
| 1857 | break; |
| 1858 | |
| 1859 | if (next == start) { |
| 1860 | /* replace with the 'to' */ |
| 1861 | Py_MEMCPY(result_s, to_s, to_len); |
| 1862 | result_s += to_len; |
| 1863 | start += 1; |
| 1864 | } else { |
| 1865 | /* copy the unchanged old then the 'to' */ |
| 1866 | Py_MEMCPY(result_s, start, next-start); |
| 1867 | result_s += (next-start); |
| 1868 | Py_MEMCPY(result_s, to_s, to_len); |
| 1869 | result_s += to_len; |
| 1870 | start = next+1; |
| 1871 | } |
| 1872 | } |
| 1873 | /* Copy the remainder of the remaining bytes */ |
| 1874 | Py_MEMCPY(result_s, start, end-start); |
| 1875 | |
| 1876 | return result; |
| 1877 | } |
| 1878 | |
| 1879 | /* len(self)>=1, len(from)>=2, len(to)>=2, maxcount>=1 */ |
| 1880 | Py_LOCAL(PyByteArrayObject *) |
| 1881 | replace_substring(PyByteArrayObject *self, |
| 1882 | const char *from_s, Py_ssize_t from_len, |
| 1883 | const char *to_s, Py_ssize_t to_len, |
| 1884 | Py_ssize_t maxcount) |
| 1885 | { |
| 1886 | char *self_s, *result_s; |
| 1887 | char *start, *next, *end; |
| 1888 | Py_ssize_t self_len, result_len; |
| 1889 | Py_ssize_t count, offset, product; |
| 1890 | PyByteArrayObject *result; |
| 1891 | |
| 1892 | self_s = PyByteArray_AS_STRING(self); |
| 1893 | self_len = PyByteArray_GET_SIZE(self); |
| 1894 | |
| 1895 | count = countstring(self_s, self_len, |
| 1896 | from_s, from_len, |
| 1897 | 0, self_len, FORWARD, maxcount); |
| 1898 | if (count == 0) { |
| 1899 | /* no matches, return unchanged */ |
| 1900 | return return_self(self); |
| 1901 | } |
| 1902 | |
| 1903 | /* Check for overflow */ |
| 1904 | /* result_len = self_len + count * (to_len-from_len) */ |
| 1905 | product = count * (to_len-from_len); |
| 1906 | if (product / (to_len-from_len) != count) { |
| 1907 | PyErr_SetString(PyExc_OverflowError, "replace bytes is too long"); |
| 1908 | return NULL; |
| 1909 | } |
| 1910 | result_len = self_len + product; |
| 1911 | if (result_len < 0) { |
| 1912 | PyErr_SetString(PyExc_OverflowError, "replace bytes is too long"); |
| 1913 | return NULL; |
| 1914 | } |
| 1915 | |
| 1916 | if ( (result = (PyByteArrayObject *) |
| 1917 | PyByteArray_FromStringAndSize(NULL, result_len)) == NULL) |
| 1918 | return NULL; |
| 1919 | result_s = PyByteArray_AS_STRING(result); |
| 1920 | |
| 1921 | start = self_s; |
| 1922 | end = self_s + self_len; |
| 1923 | while (count-- > 0) { |
| 1924 | offset = findstring(start, end-start, |
| 1925 | from_s, from_len, |
| 1926 | 0, end-start, FORWARD); |
| 1927 | if (offset == -1) |
| 1928 | break; |
| 1929 | next = start+offset; |
| 1930 | if (next == start) { |
| 1931 | /* replace with the 'to' */ |
| 1932 | Py_MEMCPY(result_s, to_s, to_len); |
| 1933 | result_s += to_len; |
| 1934 | start += from_len; |
| 1935 | } else { |
| 1936 | /* copy the unchanged old then the 'to' */ |
| 1937 | Py_MEMCPY(result_s, start, next-start); |
| 1938 | result_s += (next-start); |
| 1939 | Py_MEMCPY(result_s, to_s, to_len); |
| 1940 | result_s += to_len; |
| 1941 | start = next+from_len; |
| 1942 | } |
| 1943 | } |
| 1944 | /* Copy the remainder of the remaining bytes */ |
| 1945 | Py_MEMCPY(result_s, start, end-start); |
| 1946 | |
| 1947 | return result; |
| 1948 | } |
| 1949 | |
| 1950 | |
| 1951 | Py_LOCAL(PyByteArrayObject *) |
| 1952 | replace(PyByteArrayObject *self, |
| 1953 | const char *from_s, Py_ssize_t from_len, |
| 1954 | const char *to_s, Py_ssize_t to_len, |
| 1955 | Py_ssize_t maxcount) |
| 1956 | { |
| 1957 | if (maxcount < 0) { |
| 1958 | maxcount = PY_SSIZE_T_MAX; |
| 1959 | } else if (maxcount == 0 || PyByteArray_GET_SIZE(self) == 0) { |
| 1960 | /* nothing to do; return the original bytes */ |
| 1961 | return return_self(self); |
| 1962 | } |
| 1963 | |
| 1964 | if (maxcount == 0 || |
| 1965 | (from_len == 0 && to_len == 0)) { |
| 1966 | /* nothing to do; return the original bytes */ |
| 1967 | return return_self(self); |
| 1968 | } |
| 1969 | |
| 1970 | /* Handle zero-length special cases */ |
| 1971 | |
| 1972 | if (from_len == 0) { |
| 1973 | /* insert the 'to' bytes everywhere. */ |
| 1974 | /* >>> "Python".replace("", ".") */ |
| 1975 | /* '.P.y.t.h.o.n.' */ |
| 1976 | return replace_interleave(self, to_s, to_len, maxcount); |
| 1977 | } |
| 1978 | |
| 1979 | /* Except for "".replace("", "A") == "A" there is no way beyond this */ |
| 1980 | /* point for an empty self bytes to generate a non-empty bytes */ |
| 1981 | /* Special case so the remaining code always gets a non-empty bytes */ |
| 1982 | if (PyByteArray_GET_SIZE(self) == 0) { |
| 1983 | return return_self(self); |
| 1984 | } |
| 1985 | |
| 1986 | if (to_len == 0) { |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 1987 | /* delete all occurrences of 'from' bytes */ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 1988 | if (from_len == 1) { |
| 1989 | return replace_delete_single_character( |
| 1990 | self, from_s[0], maxcount); |
| 1991 | } else { |
| 1992 | return replace_delete_substring(self, from_s, from_len, maxcount); |
| 1993 | } |
| 1994 | } |
| 1995 | |
| 1996 | /* Handle special case where both bytes have the same length */ |
| 1997 | |
| 1998 | if (from_len == to_len) { |
| 1999 | if (from_len == 1) { |
| 2000 | return replace_single_character_in_place( |
| 2001 | self, |
| 2002 | from_s[0], |
| 2003 | to_s[0], |
| 2004 | maxcount); |
| 2005 | } else { |
| 2006 | return replace_substring_in_place( |
| 2007 | self, from_s, from_len, to_s, to_len, maxcount); |
| 2008 | } |
| 2009 | } |
| 2010 | |
| 2011 | /* Otherwise use the more generic algorithms */ |
| 2012 | if (from_len == 1) { |
| 2013 | return replace_single_character(self, from_s[0], |
| 2014 | to_s, to_len, maxcount); |
| 2015 | } else { |
| 2016 | /* len('from')>=2, len('to')>=1 */ |
| 2017 | return replace_substring(self, from_s, from_len, to_s, to_len, maxcount); |
| 2018 | } |
| 2019 | } |
| 2020 | |
| 2021 | |
| 2022 | PyDoc_STRVAR(replace__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2023 | "B.replace(old, new[, count]) -> bytearray\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2024 | \n\ |
| 2025 | Return a copy of B with all occurrences of subsection\n\ |
| 2026 | old replaced by new. If the optional argument count is\n\ |
| 2027 | given, only the first count occurrences are replaced."); |
| 2028 | |
| 2029 | static PyObject * |
| 2030 | bytes_replace(PyByteArrayObject *self, PyObject *args) |
| 2031 | { |
| 2032 | Py_ssize_t count = -1; |
| 2033 | PyObject *from, *to, *res; |
| 2034 | Py_buffer vfrom, vto; |
| 2035 | |
| 2036 | if (!PyArg_ParseTuple(args, "OO|n:replace", &from, &to, &count)) |
| 2037 | return NULL; |
| 2038 | |
| 2039 | if (_getbuffer(from, &vfrom) < 0) |
| 2040 | return NULL; |
| 2041 | if (_getbuffer(to, &vto) < 0) { |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2042 | PyBuffer_Release(&vfrom); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2043 | return NULL; |
| 2044 | } |
| 2045 | |
| 2046 | res = (PyObject *)replace((PyByteArrayObject *) self, |
| 2047 | vfrom.buf, vfrom.len, |
| 2048 | vto.buf, vto.len, count); |
| 2049 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2050 | PyBuffer_Release(&vfrom); |
| 2051 | PyBuffer_Release(&vto); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2052 | return res; |
| 2053 | } |
| 2054 | |
| 2055 | |
| 2056 | /* Overallocate the initial list to reduce the number of reallocs for small |
| 2057 | split sizes. Eg, "A A A A A A A A A A".split() (10 elements) has three |
| 2058 | resizes, to sizes 4, 8, then 16. Most observed string splits are for human |
| 2059 | text (roughly 11 words per line) and field delimited data (usually 1-10 |
| 2060 | fields). For large strings the split algorithms are bandwidth limited |
| 2061 | so increasing the preallocation likely will not improve things.*/ |
| 2062 | |
| 2063 | #define MAX_PREALLOC 12 |
| 2064 | |
| 2065 | /* 5 splits gives 6 elements */ |
| 2066 | #define PREALLOC_SIZE(maxsplit) \ |
| 2067 | (maxsplit >= MAX_PREALLOC ? MAX_PREALLOC : maxsplit+1) |
| 2068 | |
| 2069 | #define SPLIT_APPEND(data, left, right) \ |
| 2070 | str = PyByteArray_FromStringAndSize((data) + (left), \ |
| 2071 | (right) - (left)); \ |
| 2072 | if (str == NULL) \ |
| 2073 | goto onError; \ |
| 2074 | if (PyList_Append(list, str)) { \ |
| 2075 | Py_DECREF(str); \ |
| 2076 | goto onError; \ |
| 2077 | } \ |
| 2078 | else \ |
| 2079 | Py_DECREF(str); |
| 2080 | |
| 2081 | #define SPLIT_ADD(data, left, right) { \ |
| 2082 | str = PyByteArray_FromStringAndSize((data) + (left), \ |
| 2083 | (right) - (left)); \ |
| 2084 | if (str == NULL) \ |
| 2085 | goto onError; \ |
| 2086 | if (count < MAX_PREALLOC) { \ |
| 2087 | PyList_SET_ITEM(list, count, str); \ |
| 2088 | } else { \ |
| 2089 | if (PyList_Append(list, str)) { \ |
| 2090 | Py_DECREF(str); \ |
| 2091 | goto onError; \ |
| 2092 | } \ |
| 2093 | else \ |
| 2094 | Py_DECREF(str); \ |
| 2095 | } \ |
| 2096 | count++; } |
| 2097 | |
| 2098 | /* Always force the list to the expected size. */ |
| 2099 | #define FIX_PREALLOC_SIZE(list) Py_SIZE(list) = count |
| 2100 | |
| 2101 | |
| 2102 | Py_LOCAL_INLINE(PyObject *) |
| 2103 | split_char(const char *s, Py_ssize_t len, char ch, Py_ssize_t maxcount) |
| 2104 | { |
| 2105 | register Py_ssize_t i, j, count = 0; |
| 2106 | PyObject *str; |
| 2107 | PyObject *list = PyList_New(PREALLOC_SIZE(maxcount)); |
| 2108 | |
| 2109 | if (list == NULL) |
| 2110 | return NULL; |
| 2111 | |
| 2112 | i = j = 0; |
| 2113 | while ((j < len) && (maxcount-- > 0)) { |
| 2114 | for(; j < len; j++) { |
| 2115 | /* I found that using memchr makes no difference */ |
| 2116 | if (s[j] == ch) { |
| 2117 | SPLIT_ADD(s, i, j); |
| 2118 | i = j = j + 1; |
| 2119 | break; |
| 2120 | } |
| 2121 | } |
| 2122 | } |
| 2123 | if (i <= len) { |
| 2124 | SPLIT_ADD(s, i, len); |
| 2125 | } |
| 2126 | FIX_PREALLOC_SIZE(list); |
| 2127 | return list; |
| 2128 | |
| 2129 | onError: |
| 2130 | Py_DECREF(list); |
| 2131 | return NULL; |
| 2132 | } |
| 2133 | |
| 2134 | |
| 2135 | Py_LOCAL_INLINE(PyObject *) |
| 2136 | split_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxcount) |
| 2137 | { |
| 2138 | register Py_ssize_t i, j, count = 0; |
| 2139 | PyObject *str; |
| 2140 | PyObject *list = PyList_New(PREALLOC_SIZE(maxcount)); |
| 2141 | |
| 2142 | if (list == NULL) |
| 2143 | return NULL; |
| 2144 | |
| 2145 | for (i = j = 0; i < len; ) { |
| 2146 | /* find a token */ |
| 2147 | while (i < len && ISSPACE(s[i])) |
| 2148 | i++; |
| 2149 | j = i; |
| 2150 | while (i < len && !ISSPACE(s[i])) |
| 2151 | i++; |
| 2152 | if (j < i) { |
| 2153 | if (maxcount-- <= 0) |
| 2154 | break; |
| 2155 | SPLIT_ADD(s, j, i); |
| 2156 | while (i < len && ISSPACE(s[i])) |
| 2157 | i++; |
| 2158 | j = i; |
| 2159 | } |
| 2160 | } |
| 2161 | if (j < len) { |
| 2162 | SPLIT_ADD(s, j, len); |
| 2163 | } |
| 2164 | FIX_PREALLOC_SIZE(list); |
| 2165 | return list; |
| 2166 | |
| 2167 | onError: |
| 2168 | Py_DECREF(list); |
| 2169 | return NULL; |
| 2170 | } |
| 2171 | |
| 2172 | PyDoc_STRVAR(split__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2173 | "B.split([sep[, maxsplit]]) -> list of bytearrays\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2174 | \n\ |
| 2175 | Return a list of the sections in B, using sep as the delimiter.\n\ |
| 2176 | If sep is not given, B is split on ASCII whitespace characters\n\ |
| 2177 | (space, tab, return, newline, formfeed, vertical tab).\n\ |
| 2178 | If maxsplit is given, at most maxsplit splits are done."); |
| 2179 | |
| 2180 | static PyObject * |
| 2181 | bytes_split(PyByteArrayObject *self, PyObject *args) |
| 2182 | { |
| 2183 | Py_ssize_t len = PyByteArray_GET_SIZE(self), n, i, j; |
| 2184 | Py_ssize_t maxsplit = -1, count = 0; |
| 2185 | const char *s = PyByteArray_AS_STRING(self), *sub; |
| 2186 | PyObject *list, *str, *subobj = Py_None; |
| 2187 | Py_buffer vsub; |
| 2188 | #ifdef USE_FAST |
| 2189 | Py_ssize_t pos; |
| 2190 | #endif |
| 2191 | |
| 2192 | if (!PyArg_ParseTuple(args, "|On:split", &subobj, &maxsplit)) |
| 2193 | return NULL; |
| 2194 | if (maxsplit < 0) |
| 2195 | maxsplit = PY_SSIZE_T_MAX; |
| 2196 | |
| 2197 | if (subobj == Py_None) |
| 2198 | return split_whitespace(s, len, maxsplit); |
| 2199 | |
| 2200 | if (_getbuffer(subobj, &vsub) < 0) |
| 2201 | return NULL; |
| 2202 | sub = vsub.buf; |
| 2203 | n = vsub.len; |
| 2204 | |
| 2205 | if (n == 0) { |
| 2206 | PyErr_SetString(PyExc_ValueError, "empty separator"); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2207 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2208 | return NULL; |
| 2209 | } |
Benjamin Peterson | c4fe6f3 | 2008-08-19 18:57:56 +0000 | [diff] [blame] | 2210 | if (n == 1) { |
| 2211 | list = split_char(s, len, sub[0], maxsplit); |
| 2212 | PyBuffer_Release(&vsub); |
| 2213 | return list; |
| 2214 | } |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2215 | |
| 2216 | list = PyList_New(PREALLOC_SIZE(maxsplit)); |
| 2217 | if (list == NULL) { |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2218 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2219 | return NULL; |
| 2220 | } |
| 2221 | |
| 2222 | #ifdef USE_FAST |
| 2223 | i = j = 0; |
| 2224 | while (maxsplit-- > 0) { |
| 2225 | pos = fastsearch(s+i, len-i, sub, n, FAST_SEARCH); |
| 2226 | if (pos < 0) |
| 2227 | break; |
| 2228 | j = i+pos; |
| 2229 | SPLIT_ADD(s, i, j); |
| 2230 | i = j + n; |
| 2231 | } |
| 2232 | #else |
| 2233 | i = j = 0; |
| 2234 | while ((j+n <= len) && (maxsplit-- > 0)) { |
| 2235 | for (; j+n <= len; j++) { |
| 2236 | if (Py_STRING_MATCH(s, j, sub, n)) { |
| 2237 | SPLIT_ADD(s, i, j); |
| 2238 | i = j = j + n; |
| 2239 | break; |
| 2240 | } |
| 2241 | } |
| 2242 | } |
| 2243 | #endif |
| 2244 | SPLIT_ADD(s, i, len); |
| 2245 | FIX_PREALLOC_SIZE(list); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2246 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2247 | return list; |
| 2248 | |
| 2249 | onError: |
| 2250 | Py_DECREF(list); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2251 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2252 | return NULL; |
| 2253 | } |
| 2254 | |
| 2255 | /* stringlib's partition shares nullbytes in some cases. |
| 2256 | undo this, we don't want the nullbytes to be shared. */ |
| 2257 | static PyObject * |
| 2258 | make_nullbytes_unique(PyObject *result) |
| 2259 | { |
| 2260 | if (result != NULL) { |
| 2261 | int i; |
| 2262 | assert(PyTuple_Check(result)); |
| 2263 | assert(PyTuple_GET_SIZE(result) == 3); |
| 2264 | for (i = 0; i < 3; i++) { |
| 2265 | if (PyTuple_GET_ITEM(result, i) == (PyObject *)nullbytes) { |
| 2266 | PyObject *new = PyByteArray_FromStringAndSize(NULL, 0); |
| 2267 | if (new == NULL) { |
| 2268 | Py_DECREF(result); |
| 2269 | result = NULL; |
| 2270 | break; |
| 2271 | } |
| 2272 | Py_DECREF(nullbytes); |
| 2273 | PyTuple_SET_ITEM(result, i, new); |
| 2274 | } |
| 2275 | } |
| 2276 | } |
| 2277 | return result; |
| 2278 | } |
| 2279 | |
| 2280 | PyDoc_STRVAR(partition__doc__, |
| 2281 | "B.partition(sep) -> (head, sep, tail)\n\ |
| 2282 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2283 | Search for the separator sep in B, and return the part before it,\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2284 | the separator itself, and the part after it. If the separator is not\n\ |
| 2285 | found, returns B and two empty bytearray objects."); |
| 2286 | |
| 2287 | static PyObject * |
| 2288 | bytes_partition(PyByteArrayObject *self, PyObject *sep_obj) |
| 2289 | { |
| 2290 | PyObject *bytesep, *result; |
| 2291 | |
| 2292 | bytesep = PyByteArray_FromObject(sep_obj); |
| 2293 | if (! bytesep) |
| 2294 | return NULL; |
| 2295 | |
| 2296 | result = stringlib_partition( |
| 2297 | (PyObject*) self, |
| 2298 | PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), |
| 2299 | bytesep, |
| 2300 | PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep) |
| 2301 | ); |
| 2302 | |
| 2303 | Py_DECREF(bytesep); |
| 2304 | return make_nullbytes_unique(result); |
| 2305 | } |
| 2306 | |
| 2307 | PyDoc_STRVAR(rpartition__doc__, |
| 2308 | "B.rpartition(sep) -> (tail, sep, head)\n\ |
| 2309 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2310 | Search for the separator sep in B, starting at the end of B,\n\ |
| 2311 | and return the part before it, the separator itself, and the\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2312 | part after it. If the separator is not found, returns two empty\n\ |
| 2313 | bytearray objects and B."); |
| 2314 | |
| 2315 | static PyObject * |
| 2316 | bytes_rpartition(PyByteArrayObject *self, PyObject *sep_obj) |
| 2317 | { |
| 2318 | PyObject *bytesep, *result; |
| 2319 | |
| 2320 | bytesep = PyByteArray_FromObject(sep_obj); |
| 2321 | if (! bytesep) |
| 2322 | return NULL; |
| 2323 | |
| 2324 | result = stringlib_rpartition( |
| 2325 | (PyObject*) self, |
| 2326 | PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), |
| 2327 | bytesep, |
| 2328 | PyByteArray_AS_STRING(bytesep), PyByteArray_GET_SIZE(bytesep) |
| 2329 | ); |
| 2330 | |
| 2331 | Py_DECREF(bytesep); |
| 2332 | return make_nullbytes_unique(result); |
| 2333 | } |
| 2334 | |
| 2335 | Py_LOCAL_INLINE(PyObject *) |
| 2336 | rsplit_char(const char *s, Py_ssize_t len, char ch, Py_ssize_t maxcount) |
| 2337 | { |
| 2338 | register Py_ssize_t i, j, count=0; |
| 2339 | PyObject *str; |
| 2340 | PyObject *list = PyList_New(PREALLOC_SIZE(maxcount)); |
| 2341 | |
| 2342 | if (list == NULL) |
| 2343 | return NULL; |
| 2344 | |
| 2345 | i = j = len - 1; |
| 2346 | while ((i >= 0) && (maxcount-- > 0)) { |
| 2347 | for (; i >= 0; i--) { |
| 2348 | if (s[i] == ch) { |
| 2349 | SPLIT_ADD(s, i + 1, j + 1); |
| 2350 | j = i = i - 1; |
| 2351 | break; |
| 2352 | } |
| 2353 | } |
| 2354 | } |
| 2355 | if (j >= -1) { |
| 2356 | SPLIT_ADD(s, 0, j + 1); |
| 2357 | } |
| 2358 | FIX_PREALLOC_SIZE(list); |
| 2359 | if (PyList_Reverse(list) < 0) |
| 2360 | goto onError; |
| 2361 | |
| 2362 | return list; |
| 2363 | |
| 2364 | onError: |
| 2365 | Py_DECREF(list); |
| 2366 | return NULL; |
| 2367 | } |
| 2368 | |
| 2369 | Py_LOCAL_INLINE(PyObject *) |
| 2370 | rsplit_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxcount) |
| 2371 | { |
| 2372 | register Py_ssize_t i, j, count = 0; |
| 2373 | PyObject *str; |
| 2374 | PyObject *list = PyList_New(PREALLOC_SIZE(maxcount)); |
| 2375 | |
| 2376 | if (list == NULL) |
| 2377 | return NULL; |
| 2378 | |
| 2379 | for (i = j = len - 1; i >= 0; ) { |
| 2380 | /* find a token */ |
| 2381 | while (i >= 0 && ISSPACE(s[i])) |
| 2382 | i--; |
| 2383 | j = i; |
| 2384 | while (i >= 0 && !ISSPACE(s[i])) |
| 2385 | i--; |
| 2386 | if (j > i) { |
| 2387 | if (maxcount-- <= 0) |
| 2388 | break; |
| 2389 | SPLIT_ADD(s, i + 1, j + 1); |
| 2390 | while (i >= 0 && ISSPACE(s[i])) |
| 2391 | i--; |
| 2392 | j = i; |
| 2393 | } |
| 2394 | } |
| 2395 | if (j >= 0) { |
| 2396 | SPLIT_ADD(s, 0, j + 1); |
| 2397 | } |
| 2398 | FIX_PREALLOC_SIZE(list); |
| 2399 | if (PyList_Reverse(list) < 0) |
| 2400 | goto onError; |
| 2401 | |
| 2402 | return list; |
| 2403 | |
| 2404 | onError: |
| 2405 | Py_DECREF(list); |
| 2406 | return NULL; |
| 2407 | } |
| 2408 | |
| 2409 | PyDoc_STRVAR(rsplit__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2410 | "B.rsplit(sep[, maxsplit]) -> list of bytearrays\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2411 | \n\ |
| 2412 | Return a list of the sections in B, using sep as the delimiter,\n\ |
| 2413 | starting at the end of B and working to the front.\n\ |
| 2414 | If sep is not given, B is split on ASCII whitespace characters\n\ |
| 2415 | (space, tab, return, newline, formfeed, vertical tab).\n\ |
| 2416 | If maxsplit is given, at most maxsplit splits are done."); |
| 2417 | |
| 2418 | static PyObject * |
| 2419 | bytes_rsplit(PyByteArrayObject *self, PyObject *args) |
| 2420 | { |
| 2421 | Py_ssize_t len = PyByteArray_GET_SIZE(self), n, i, j; |
| 2422 | Py_ssize_t maxsplit = -1, count = 0; |
| 2423 | const char *s = PyByteArray_AS_STRING(self), *sub; |
| 2424 | PyObject *list, *str, *subobj = Py_None; |
| 2425 | Py_buffer vsub; |
| 2426 | |
| 2427 | if (!PyArg_ParseTuple(args, "|On:rsplit", &subobj, &maxsplit)) |
| 2428 | return NULL; |
| 2429 | if (maxsplit < 0) |
| 2430 | maxsplit = PY_SSIZE_T_MAX; |
| 2431 | |
| 2432 | if (subobj == Py_None) |
| 2433 | return rsplit_whitespace(s, len, maxsplit); |
| 2434 | |
| 2435 | if (_getbuffer(subobj, &vsub) < 0) |
| 2436 | return NULL; |
| 2437 | sub = vsub.buf; |
| 2438 | n = vsub.len; |
| 2439 | |
| 2440 | if (n == 0) { |
| 2441 | PyErr_SetString(PyExc_ValueError, "empty separator"); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2442 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2443 | return NULL; |
| 2444 | } |
Benjamin Peterson | c4fe6f3 | 2008-08-19 18:57:56 +0000 | [diff] [blame] | 2445 | else if (n == 1) { |
| 2446 | list = rsplit_char(s, len, sub[0], maxsplit); |
| 2447 | PyBuffer_Release(&vsub); |
| 2448 | return list; |
| 2449 | } |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2450 | |
| 2451 | list = PyList_New(PREALLOC_SIZE(maxsplit)); |
| 2452 | if (list == NULL) { |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2453 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2454 | return NULL; |
| 2455 | } |
| 2456 | |
| 2457 | j = len; |
| 2458 | i = j - n; |
| 2459 | |
| 2460 | while ( (i >= 0) && (maxsplit-- > 0) ) { |
| 2461 | for (; i>=0; i--) { |
| 2462 | if (Py_STRING_MATCH(s, i, sub, n)) { |
| 2463 | SPLIT_ADD(s, i + n, j); |
| 2464 | j = i; |
| 2465 | i -= n; |
| 2466 | break; |
| 2467 | } |
| 2468 | } |
| 2469 | } |
| 2470 | SPLIT_ADD(s, 0, j); |
| 2471 | FIX_PREALLOC_SIZE(list); |
| 2472 | if (PyList_Reverse(list) < 0) |
| 2473 | goto onError; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2474 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2475 | return list; |
| 2476 | |
| 2477 | onError: |
| 2478 | Py_DECREF(list); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2479 | PyBuffer_Release(&vsub); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2480 | return NULL; |
| 2481 | } |
| 2482 | |
| 2483 | PyDoc_STRVAR(reverse__doc__, |
| 2484 | "B.reverse() -> None\n\ |
| 2485 | \n\ |
| 2486 | Reverse the order of the values in B in place."); |
| 2487 | static PyObject * |
| 2488 | bytes_reverse(PyByteArrayObject *self, PyObject *unused) |
| 2489 | { |
| 2490 | char swap, *head, *tail; |
| 2491 | Py_ssize_t i, j, n = Py_SIZE(self); |
| 2492 | |
| 2493 | j = n / 2; |
| 2494 | head = self->ob_bytes; |
| 2495 | tail = head + n - 1; |
| 2496 | for (i = 0; i < j; i++) { |
| 2497 | swap = *head; |
| 2498 | *head++ = *tail; |
| 2499 | *tail-- = swap; |
| 2500 | } |
| 2501 | |
| 2502 | Py_RETURN_NONE; |
| 2503 | } |
| 2504 | |
| 2505 | PyDoc_STRVAR(insert__doc__, |
| 2506 | "B.insert(index, int) -> None\n\ |
| 2507 | \n\ |
| 2508 | Insert a single item into the bytearray before the given index."); |
| 2509 | static PyObject * |
| 2510 | bytes_insert(PyByteArrayObject *self, PyObject *args) |
| 2511 | { |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 2512 | PyObject *value; |
| 2513 | int ival; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2514 | Py_ssize_t where, n = Py_SIZE(self); |
| 2515 | |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 2516 | if (!PyArg_ParseTuple(args, "nO:insert", &where, &value)) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2517 | return NULL; |
| 2518 | |
| 2519 | if (n == PY_SSIZE_T_MAX) { |
| 2520 | PyErr_SetString(PyExc_OverflowError, |
| 2521 | "cannot add more objects to bytes"); |
| 2522 | return NULL; |
| 2523 | } |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 2524 | if (!_getbytevalue(value, &ival)) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2525 | return NULL; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2526 | if (PyByteArray_Resize((PyObject *)self, n + 1) < 0) |
| 2527 | return NULL; |
| 2528 | |
| 2529 | if (where < 0) { |
| 2530 | where += n; |
| 2531 | if (where < 0) |
| 2532 | where = 0; |
| 2533 | } |
| 2534 | if (where > n) |
| 2535 | where = n; |
| 2536 | memmove(self->ob_bytes + where + 1, self->ob_bytes + where, n - where); |
Georg Brandl | 9a54d7c | 2008-07-16 23:15:30 +0000 | [diff] [blame] | 2537 | self->ob_bytes[where] = ival; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2538 | |
| 2539 | Py_RETURN_NONE; |
| 2540 | } |
| 2541 | |
| 2542 | PyDoc_STRVAR(append__doc__, |
| 2543 | "B.append(int) -> None\n\ |
| 2544 | \n\ |
| 2545 | Append a single item to the end of B."); |
| 2546 | static PyObject * |
| 2547 | bytes_append(PyByteArrayObject *self, PyObject *arg) |
| 2548 | { |
| 2549 | int value; |
| 2550 | Py_ssize_t n = Py_SIZE(self); |
| 2551 | |
| 2552 | if (! _getbytevalue(arg, &value)) |
| 2553 | return NULL; |
| 2554 | if (n == PY_SSIZE_T_MAX) { |
| 2555 | PyErr_SetString(PyExc_OverflowError, |
| 2556 | "cannot add more objects to bytes"); |
| 2557 | return NULL; |
| 2558 | } |
| 2559 | if (PyByteArray_Resize((PyObject *)self, n + 1) < 0) |
| 2560 | return NULL; |
| 2561 | |
| 2562 | self->ob_bytes[n] = value; |
| 2563 | |
| 2564 | Py_RETURN_NONE; |
| 2565 | } |
| 2566 | |
| 2567 | PyDoc_STRVAR(extend__doc__, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2568 | "B.extend(iterable_of_ints) -> None\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2569 | \n\ |
| 2570 | Append all the elements from the iterator or sequence to the\n\ |
| 2571 | end of B."); |
| 2572 | static PyObject * |
| 2573 | bytes_extend(PyByteArrayObject *self, PyObject *arg) |
| 2574 | { |
| 2575 | PyObject *it, *item, *bytes_obj; |
| 2576 | Py_ssize_t buf_size = 0, len = 0; |
| 2577 | int value; |
| 2578 | char *buf; |
| 2579 | |
| 2580 | /* bytes_setslice code only accepts something supporting PEP 3118. */ |
| 2581 | if (PyObject_CheckBuffer(arg)) { |
| 2582 | if (bytes_setslice(self, Py_SIZE(self), Py_SIZE(self), arg) == -1) |
| 2583 | return NULL; |
| 2584 | |
| 2585 | Py_RETURN_NONE; |
| 2586 | } |
| 2587 | |
| 2588 | it = PyObject_GetIter(arg); |
| 2589 | if (it == NULL) |
| 2590 | return NULL; |
| 2591 | |
| 2592 | /* Try to determine the length of the argument. 32 is abitrary. */ |
| 2593 | buf_size = _PyObject_LengthHint(arg, 32); |
| 2594 | |
| 2595 | bytes_obj = PyByteArray_FromStringAndSize(NULL, buf_size); |
| 2596 | if (bytes_obj == NULL) |
| 2597 | return NULL; |
| 2598 | buf = PyByteArray_AS_STRING(bytes_obj); |
| 2599 | |
| 2600 | while ((item = PyIter_Next(it)) != NULL) { |
| 2601 | if (! _getbytevalue(item, &value)) { |
| 2602 | Py_DECREF(item); |
| 2603 | Py_DECREF(it); |
| 2604 | Py_DECREF(bytes_obj); |
| 2605 | return NULL; |
| 2606 | } |
| 2607 | buf[len++] = value; |
| 2608 | Py_DECREF(item); |
| 2609 | |
| 2610 | if (len >= buf_size) { |
| 2611 | buf_size = len + (len >> 1) + 1; |
| 2612 | if (PyByteArray_Resize((PyObject *)bytes_obj, buf_size) < 0) { |
| 2613 | Py_DECREF(it); |
| 2614 | Py_DECREF(bytes_obj); |
| 2615 | return NULL; |
| 2616 | } |
| 2617 | /* Recompute the `buf' pointer, since the resizing operation may |
| 2618 | have invalidated it. */ |
| 2619 | buf = PyByteArray_AS_STRING(bytes_obj); |
| 2620 | } |
| 2621 | } |
| 2622 | Py_DECREF(it); |
| 2623 | |
| 2624 | /* Resize down to exact size. */ |
| 2625 | if (PyByteArray_Resize((PyObject *)bytes_obj, len) < 0) { |
| 2626 | Py_DECREF(bytes_obj); |
| 2627 | return NULL; |
| 2628 | } |
| 2629 | |
| 2630 | if (bytes_setslice(self, Py_SIZE(self), Py_SIZE(self), bytes_obj) == -1) |
| 2631 | return NULL; |
| 2632 | Py_DECREF(bytes_obj); |
| 2633 | |
| 2634 | Py_RETURN_NONE; |
| 2635 | } |
| 2636 | |
| 2637 | PyDoc_STRVAR(pop__doc__, |
| 2638 | "B.pop([index]) -> int\n\ |
| 2639 | \n\ |
| 2640 | Remove and return a single item from B. If no index\n\ |
Benjamin Peterson | dcf97b9 | 2008-07-02 17:30:14 +0000 | [diff] [blame] | 2641 | argument is given, will pop the last value."); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2642 | static PyObject * |
| 2643 | bytes_pop(PyByteArrayObject *self, PyObject *args) |
| 2644 | { |
| 2645 | int value; |
| 2646 | Py_ssize_t where = -1, n = Py_SIZE(self); |
| 2647 | |
| 2648 | if (!PyArg_ParseTuple(args, "|n:pop", &where)) |
| 2649 | return NULL; |
| 2650 | |
| 2651 | if (n == 0) { |
| 2652 | PyErr_SetString(PyExc_OverflowError, |
| 2653 | "cannot pop an empty bytes"); |
| 2654 | return NULL; |
| 2655 | } |
| 2656 | if (where < 0) |
| 2657 | where += Py_SIZE(self); |
| 2658 | if (where < 0 || where >= Py_SIZE(self)) { |
| 2659 | PyErr_SetString(PyExc_IndexError, "pop index out of range"); |
| 2660 | return NULL; |
| 2661 | } |
| 2662 | |
| 2663 | value = self->ob_bytes[where]; |
| 2664 | memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where); |
| 2665 | if (PyByteArray_Resize((PyObject *)self, n - 1) < 0) |
| 2666 | return NULL; |
| 2667 | |
| 2668 | return PyLong_FromLong(value); |
| 2669 | } |
| 2670 | |
| 2671 | PyDoc_STRVAR(remove__doc__, |
| 2672 | "B.remove(int) -> None\n\ |
| 2673 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2674 | Remove the first occurrence of a value in B."); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2675 | static PyObject * |
| 2676 | bytes_remove(PyByteArrayObject *self, PyObject *arg) |
| 2677 | { |
| 2678 | int value; |
| 2679 | Py_ssize_t where, n = Py_SIZE(self); |
| 2680 | |
| 2681 | if (! _getbytevalue(arg, &value)) |
| 2682 | return NULL; |
| 2683 | |
| 2684 | for (where = 0; where < n; where++) { |
| 2685 | if (self->ob_bytes[where] == value) |
| 2686 | break; |
| 2687 | } |
| 2688 | if (where == n) { |
| 2689 | PyErr_SetString(PyExc_ValueError, "value not found in bytes"); |
| 2690 | return NULL; |
| 2691 | } |
| 2692 | |
| 2693 | memmove(self->ob_bytes + where, self->ob_bytes + where + 1, n - where); |
| 2694 | if (PyByteArray_Resize((PyObject *)self, n - 1) < 0) |
| 2695 | return NULL; |
| 2696 | |
| 2697 | Py_RETURN_NONE; |
| 2698 | } |
| 2699 | |
| 2700 | /* XXX These two helpers could be optimized if argsize == 1 */ |
| 2701 | |
| 2702 | static Py_ssize_t |
| 2703 | lstrip_helper(unsigned char *myptr, Py_ssize_t mysize, |
| 2704 | void *argptr, Py_ssize_t argsize) |
| 2705 | { |
| 2706 | Py_ssize_t i = 0; |
| 2707 | while (i < mysize && memchr(argptr, myptr[i], argsize)) |
| 2708 | i++; |
| 2709 | return i; |
| 2710 | } |
| 2711 | |
| 2712 | static Py_ssize_t |
| 2713 | rstrip_helper(unsigned char *myptr, Py_ssize_t mysize, |
| 2714 | void *argptr, Py_ssize_t argsize) |
| 2715 | { |
| 2716 | Py_ssize_t i = mysize - 1; |
| 2717 | while (i >= 0 && memchr(argptr, myptr[i], argsize)) |
| 2718 | i--; |
| 2719 | return i + 1; |
| 2720 | } |
| 2721 | |
| 2722 | PyDoc_STRVAR(strip__doc__, |
| 2723 | "B.strip([bytes]) -> bytearray\n\ |
| 2724 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2725 | Strip leading and trailing bytes contained in the argument\n\ |
| 2726 | and return the result as a new bytearray.\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2727 | If the argument is omitted, strip ASCII whitespace."); |
| 2728 | static PyObject * |
| 2729 | bytes_strip(PyByteArrayObject *self, PyObject *args) |
| 2730 | { |
| 2731 | Py_ssize_t left, right, mysize, argsize; |
| 2732 | void *myptr, *argptr; |
| 2733 | PyObject *arg = Py_None; |
| 2734 | Py_buffer varg; |
| 2735 | if (!PyArg_ParseTuple(args, "|O:strip", &arg)) |
| 2736 | return NULL; |
| 2737 | if (arg == Py_None) { |
| 2738 | argptr = "\t\n\r\f\v "; |
| 2739 | argsize = 6; |
| 2740 | } |
| 2741 | else { |
| 2742 | if (_getbuffer(arg, &varg) < 0) |
| 2743 | return NULL; |
| 2744 | argptr = varg.buf; |
| 2745 | argsize = varg.len; |
| 2746 | } |
| 2747 | myptr = self->ob_bytes; |
| 2748 | mysize = Py_SIZE(self); |
| 2749 | left = lstrip_helper(myptr, mysize, argptr, argsize); |
| 2750 | if (left == mysize) |
| 2751 | right = left; |
| 2752 | else |
| 2753 | right = rstrip_helper(myptr, mysize, argptr, argsize); |
| 2754 | if (arg != Py_None) |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2755 | PyBuffer_Release(&varg); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2756 | return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left); |
| 2757 | } |
| 2758 | |
| 2759 | PyDoc_STRVAR(lstrip__doc__, |
| 2760 | "B.lstrip([bytes]) -> bytearray\n\ |
| 2761 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2762 | Strip leading bytes contained in the argument\n\ |
| 2763 | and return the result as a new bytearray.\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2764 | If the argument is omitted, strip leading ASCII whitespace."); |
| 2765 | static PyObject * |
| 2766 | bytes_lstrip(PyByteArrayObject *self, PyObject *args) |
| 2767 | { |
| 2768 | Py_ssize_t left, right, mysize, argsize; |
| 2769 | void *myptr, *argptr; |
| 2770 | PyObject *arg = Py_None; |
| 2771 | Py_buffer varg; |
| 2772 | if (!PyArg_ParseTuple(args, "|O:lstrip", &arg)) |
| 2773 | return NULL; |
| 2774 | if (arg == Py_None) { |
| 2775 | argptr = "\t\n\r\f\v "; |
| 2776 | argsize = 6; |
| 2777 | } |
| 2778 | else { |
| 2779 | if (_getbuffer(arg, &varg) < 0) |
| 2780 | return NULL; |
| 2781 | argptr = varg.buf; |
| 2782 | argsize = varg.len; |
| 2783 | } |
| 2784 | myptr = self->ob_bytes; |
| 2785 | mysize = Py_SIZE(self); |
| 2786 | left = lstrip_helper(myptr, mysize, argptr, argsize); |
| 2787 | right = mysize; |
| 2788 | if (arg != Py_None) |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2789 | PyBuffer_Release(&varg); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2790 | return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left); |
| 2791 | } |
| 2792 | |
| 2793 | PyDoc_STRVAR(rstrip__doc__, |
| 2794 | "B.rstrip([bytes]) -> bytearray\n\ |
| 2795 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2796 | Strip trailing bytes contained in the argument\n\ |
| 2797 | and return the result as a new bytearray.\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2798 | If the argument is omitted, strip trailing ASCII whitespace."); |
| 2799 | static PyObject * |
| 2800 | bytes_rstrip(PyByteArrayObject *self, PyObject *args) |
| 2801 | { |
| 2802 | Py_ssize_t left, right, mysize, argsize; |
| 2803 | void *myptr, *argptr; |
| 2804 | PyObject *arg = Py_None; |
| 2805 | Py_buffer varg; |
| 2806 | if (!PyArg_ParseTuple(args, "|O:rstrip", &arg)) |
| 2807 | return NULL; |
| 2808 | if (arg == Py_None) { |
| 2809 | argptr = "\t\n\r\f\v "; |
| 2810 | argsize = 6; |
| 2811 | } |
| 2812 | else { |
| 2813 | if (_getbuffer(arg, &varg) < 0) |
| 2814 | return NULL; |
| 2815 | argptr = varg.buf; |
| 2816 | argsize = varg.len; |
| 2817 | } |
| 2818 | myptr = self->ob_bytes; |
| 2819 | mysize = Py_SIZE(self); |
| 2820 | left = 0; |
| 2821 | right = rstrip_helper(myptr, mysize, argptr, argsize); |
| 2822 | if (arg != Py_None) |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 2823 | PyBuffer_Release(&varg); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2824 | return PyByteArray_FromStringAndSize(self->ob_bytes + left, right - left); |
| 2825 | } |
| 2826 | |
| 2827 | PyDoc_STRVAR(decode_doc, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2828 | "B.decode([encoding[, errors]]) -> str\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2829 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2830 | Decode B using the codec registered for encoding. encoding defaults\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2831 | to the default encoding. errors may be given to set a different error\n\ |
| 2832 | handling scheme. Default is 'strict' meaning that encoding errors raise\n\ |
| 2833 | a UnicodeDecodeError. Other possible values are 'ignore' and 'replace'\n\ |
| 2834 | as well as any other name registered with codecs.register_error that is\n\ |
| 2835 | able to handle UnicodeDecodeErrors."); |
| 2836 | |
| 2837 | static PyObject * |
| 2838 | bytes_decode(PyObject *self, PyObject *args) |
| 2839 | { |
| 2840 | const char *encoding = NULL; |
| 2841 | const char *errors = NULL; |
| 2842 | |
| 2843 | if (!PyArg_ParseTuple(args, "|ss:decode", &encoding, &errors)) |
| 2844 | return NULL; |
| 2845 | if (encoding == NULL) |
| 2846 | encoding = PyUnicode_GetDefaultEncoding(); |
Marc-André Lemburg | b2750b5 | 2008-06-06 12:18:17 +0000 | [diff] [blame] | 2847 | return PyUnicode_FromEncodedObject(self, encoding, errors); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2848 | } |
| 2849 | |
| 2850 | PyDoc_STRVAR(alloc_doc, |
| 2851 | "B.__alloc__() -> int\n\ |
| 2852 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2853 | Return the number of bytes actually allocated."); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2854 | |
| 2855 | static PyObject * |
| 2856 | bytes_alloc(PyByteArrayObject *self) |
| 2857 | { |
| 2858 | return PyLong_FromSsize_t(self->ob_alloc); |
| 2859 | } |
| 2860 | |
| 2861 | PyDoc_STRVAR(join_doc, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2862 | "B.join(iterable_of_bytes) -> bytearray\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2863 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2864 | Concatenate any number of bytes/bytearray objects, with B\n\ |
| 2865 | in between each pair, and return the result as a new bytearray."); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2866 | |
| 2867 | static PyObject * |
| 2868 | bytes_join(PyByteArrayObject *self, PyObject *it) |
| 2869 | { |
| 2870 | PyObject *seq; |
| 2871 | Py_ssize_t mysize = Py_SIZE(self); |
| 2872 | Py_ssize_t i; |
| 2873 | Py_ssize_t n; |
| 2874 | PyObject **items; |
| 2875 | Py_ssize_t totalsize = 0; |
| 2876 | PyObject *result; |
| 2877 | char *dest; |
| 2878 | |
| 2879 | seq = PySequence_Fast(it, "can only join an iterable"); |
| 2880 | if (seq == NULL) |
| 2881 | return NULL; |
| 2882 | n = PySequence_Fast_GET_SIZE(seq); |
| 2883 | items = PySequence_Fast_ITEMS(seq); |
| 2884 | |
| 2885 | /* Compute the total size, and check that they are all bytes */ |
| 2886 | /* XXX Shouldn't we use _getbuffer() on these items instead? */ |
| 2887 | for (i = 0; i < n; i++) { |
| 2888 | PyObject *obj = items[i]; |
| 2889 | if (!PyByteArray_Check(obj) && !PyBytes_Check(obj)) { |
| 2890 | PyErr_Format(PyExc_TypeError, |
| 2891 | "can only join an iterable of bytes " |
| 2892 | "(item %ld has type '%.100s')", |
| 2893 | /* XXX %ld isn't right on Win64 */ |
| 2894 | (long)i, Py_TYPE(obj)->tp_name); |
| 2895 | goto error; |
| 2896 | } |
| 2897 | if (i > 0) |
| 2898 | totalsize += mysize; |
| 2899 | totalsize += Py_SIZE(obj); |
| 2900 | if (totalsize < 0) { |
| 2901 | PyErr_NoMemory(); |
| 2902 | goto error; |
| 2903 | } |
| 2904 | } |
| 2905 | |
| 2906 | /* Allocate the result, and copy the bytes */ |
| 2907 | result = PyByteArray_FromStringAndSize(NULL, totalsize); |
| 2908 | if (result == NULL) |
| 2909 | goto error; |
| 2910 | dest = PyByteArray_AS_STRING(result); |
| 2911 | for (i = 0; i < n; i++) { |
| 2912 | PyObject *obj = items[i]; |
| 2913 | Py_ssize_t size = Py_SIZE(obj); |
| 2914 | char *buf; |
| 2915 | if (PyByteArray_Check(obj)) |
| 2916 | buf = PyByteArray_AS_STRING(obj); |
| 2917 | else |
| 2918 | buf = PyBytes_AS_STRING(obj); |
| 2919 | if (i) { |
| 2920 | memcpy(dest, self->ob_bytes, mysize); |
| 2921 | dest += mysize; |
| 2922 | } |
| 2923 | memcpy(dest, buf, size); |
| 2924 | dest += size; |
| 2925 | } |
| 2926 | |
| 2927 | /* Done */ |
| 2928 | Py_DECREF(seq); |
| 2929 | return result; |
| 2930 | |
| 2931 | /* Error handling */ |
| 2932 | error: |
| 2933 | Py_DECREF(seq); |
| 2934 | return NULL; |
| 2935 | } |
| 2936 | |
| 2937 | PyDoc_STRVAR(fromhex_doc, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 2938 | "bytearray.fromhex(string) -> bytearray (static method)\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 2939 | \n\ |
| 2940 | Create a bytearray object from a string of hexadecimal numbers.\n\ |
| 2941 | Spaces between two numbers are accepted.\n\ |
| 2942 | Example: bytearray.fromhex('B9 01EF') -> bytearray(b'\\xb9\\x01\\xef')."); |
| 2943 | |
| 2944 | static int |
| 2945 | hex_digit_to_int(Py_UNICODE c) |
| 2946 | { |
| 2947 | if (c >= 128) |
| 2948 | return -1; |
| 2949 | if (ISDIGIT(c)) |
| 2950 | return c - '0'; |
| 2951 | else { |
| 2952 | if (ISUPPER(c)) |
| 2953 | c = TOLOWER(c); |
| 2954 | if (c >= 'a' && c <= 'f') |
| 2955 | return c - 'a' + 10; |
| 2956 | } |
| 2957 | return -1; |
| 2958 | } |
| 2959 | |
| 2960 | static PyObject * |
| 2961 | bytes_fromhex(PyObject *cls, PyObject *args) |
| 2962 | { |
| 2963 | PyObject *newbytes, *hexobj; |
| 2964 | char *buf; |
| 2965 | Py_UNICODE *hex; |
| 2966 | Py_ssize_t hexlen, byteslen, i, j; |
| 2967 | int top, bot; |
| 2968 | |
| 2969 | if (!PyArg_ParseTuple(args, "U:fromhex", &hexobj)) |
| 2970 | return NULL; |
| 2971 | assert(PyUnicode_Check(hexobj)); |
| 2972 | hexlen = PyUnicode_GET_SIZE(hexobj); |
| 2973 | hex = PyUnicode_AS_UNICODE(hexobj); |
| 2974 | byteslen = hexlen/2; /* This overestimates if there are spaces */ |
| 2975 | newbytes = PyByteArray_FromStringAndSize(NULL, byteslen); |
| 2976 | if (!newbytes) |
| 2977 | return NULL; |
| 2978 | buf = PyByteArray_AS_STRING(newbytes); |
| 2979 | for (i = j = 0; i < hexlen; i += 2) { |
| 2980 | /* skip over spaces in the input */ |
| 2981 | while (hex[i] == ' ') |
| 2982 | i++; |
| 2983 | if (i >= hexlen) |
| 2984 | break; |
| 2985 | top = hex_digit_to_int(hex[i]); |
| 2986 | bot = hex_digit_to_int(hex[i+1]); |
| 2987 | if (top == -1 || bot == -1) { |
| 2988 | PyErr_Format(PyExc_ValueError, |
| 2989 | "non-hexadecimal number found in " |
| 2990 | "fromhex() arg at position %zd", i); |
| 2991 | goto error; |
| 2992 | } |
| 2993 | buf[j++] = (top << 4) + bot; |
| 2994 | } |
| 2995 | if (PyByteArray_Resize(newbytes, j) < 0) |
| 2996 | goto error; |
| 2997 | return newbytes; |
| 2998 | |
| 2999 | error: |
| 3000 | Py_DECREF(newbytes); |
| 3001 | return NULL; |
| 3002 | } |
| 3003 | |
| 3004 | PyDoc_STRVAR(reduce_doc, "Return state information for pickling."); |
| 3005 | |
| 3006 | static PyObject * |
| 3007 | bytes_reduce(PyByteArrayObject *self) |
| 3008 | { |
| 3009 | PyObject *latin1, *dict; |
| 3010 | if (self->ob_bytes) |
| 3011 | latin1 = PyUnicode_DecodeLatin1(self->ob_bytes, |
| 3012 | Py_SIZE(self), NULL); |
| 3013 | else |
| 3014 | latin1 = PyUnicode_FromString(""); |
| 3015 | |
| 3016 | dict = PyObject_GetAttrString((PyObject *)self, "__dict__"); |
| 3017 | if (dict == NULL) { |
| 3018 | PyErr_Clear(); |
| 3019 | dict = Py_None; |
| 3020 | Py_INCREF(dict); |
| 3021 | } |
| 3022 | |
| 3023 | return Py_BuildValue("(O(Ns)N)", Py_TYPE(self), latin1, "latin-1", dict); |
| 3024 | } |
| 3025 | |
Robert Schuppenies | fbe94c5 | 2008-07-14 10:13:31 +0000 | [diff] [blame] | 3026 | PyDoc_STRVAR(sizeof_doc, |
| 3027 | "B.__sizeof__() -> int\n\ |
| 3028 | \n\ |
| 3029 | Returns the size of B in memory, in bytes"); |
| 3030 | static PyObject * |
| 3031 | bytes_sizeof(PyByteArrayObject *self) |
| 3032 | { |
| 3033 | Py_ssize_t res; |
| 3034 | |
| 3035 | res = sizeof(PyByteArrayObject) + self->ob_alloc * sizeof(char); |
| 3036 | return PyLong_FromSsize_t(res); |
| 3037 | } |
| 3038 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 3039 | static PySequenceMethods bytes_as_sequence = { |
| 3040 | (lenfunc)bytes_length, /* sq_length */ |
| 3041 | (binaryfunc)PyByteArray_Concat, /* sq_concat */ |
| 3042 | (ssizeargfunc)bytes_repeat, /* sq_repeat */ |
| 3043 | (ssizeargfunc)bytes_getitem, /* sq_item */ |
| 3044 | 0, /* sq_slice */ |
| 3045 | (ssizeobjargproc)bytes_setitem, /* sq_ass_item */ |
| 3046 | 0, /* sq_ass_slice */ |
| 3047 | (objobjproc)bytes_contains, /* sq_contains */ |
| 3048 | (binaryfunc)bytes_iconcat, /* sq_inplace_concat */ |
| 3049 | (ssizeargfunc)bytes_irepeat, /* sq_inplace_repeat */ |
| 3050 | }; |
| 3051 | |
| 3052 | static PyMappingMethods bytes_as_mapping = { |
| 3053 | (lenfunc)bytes_length, |
| 3054 | (binaryfunc)bytes_subscript, |
| 3055 | (objobjargproc)bytes_ass_subscript, |
| 3056 | }; |
| 3057 | |
| 3058 | static PyBufferProcs bytes_as_buffer = { |
| 3059 | (getbufferproc)bytes_getbuffer, |
| 3060 | (releasebufferproc)bytes_releasebuffer, |
| 3061 | }; |
| 3062 | |
| 3063 | static PyMethodDef |
| 3064 | bytes_methods[] = { |
| 3065 | {"__alloc__", (PyCFunction)bytes_alloc, METH_NOARGS, alloc_doc}, |
| 3066 | {"__reduce__", (PyCFunction)bytes_reduce, METH_NOARGS, reduce_doc}, |
Robert Schuppenies | fbe94c5 | 2008-07-14 10:13:31 +0000 | [diff] [blame] | 3067 | {"__sizeof__", (PyCFunction)bytes_sizeof, METH_NOARGS, sizeof_doc}, |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 3068 | {"append", (PyCFunction)bytes_append, METH_O, append__doc__}, |
| 3069 | {"capitalize", (PyCFunction)stringlib_capitalize, METH_NOARGS, |
| 3070 | _Py_capitalize__doc__}, |
| 3071 | {"center", (PyCFunction)stringlib_center, METH_VARARGS, center__doc__}, |
| 3072 | {"count", (PyCFunction)bytes_count, METH_VARARGS, count__doc__}, |
| 3073 | {"decode", (PyCFunction)bytes_decode, METH_VARARGS, decode_doc}, |
| 3074 | {"endswith", (PyCFunction)bytes_endswith, METH_VARARGS, endswith__doc__}, |
| 3075 | {"expandtabs", (PyCFunction)stringlib_expandtabs, METH_VARARGS, |
| 3076 | expandtabs__doc__}, |
| 3077 | {"extend", (PyCFunction)bytes_extend, METH_O, extend__doc__}, |
| 3078 | {"find", (PyCFunction)bytes_find, METH_VARARGS, find__doc__}, |
| 3079 | {"fromhex", (PyCFunction)bytes_fromhex, METH_VARARGS|METH_CLASS, |
| 3080 | fromhex_doc}, |
| 3081 | {"index", (PyCFunction)bytes_index, METH_VARARGS, index__doc__}, |
| 3082 | {"insert", (PyCFunction)bytes_insert, METH_VARARGS, insert__doc__}, |
| 3083 | {"isalnum", (PyCFunction)stringlib_isalnum, METH_NOARGS, |
| 3084 | _Py_isalnum__doc__}, |
| 3085 | {"isalpha", (PyCFunction)stringlib_isalpha, METH_NOARGS, |
| 3086 | _Py_isalpha__doc__}, |
| 3087 | {"isdigit", (PyCFunction)stringlib_isdigit, METH_NOARGS, |
| 3088 | _Py_isdigit__doc__}, |
| 3089 | {"islower", (PyCFunction)stringlib_islower, METH_NOARGS, |
| 3090 | _Py_islower__doc__}, |
| 3091 | {"isspace", (PyCFunction)stringlib_isspace, METH_NOARGS, |
| 3092 | _Py_isspace__doc__}, |
| 3093 | {"istitle", (PyCFunction)stringlib_istitle, METH_NOARGS, |
| 3094 | _Py_istitle__doc__}, |
| 3095 | {"isupper", (PyCFunction)stringlib_isupper, METH_NOARGS, |
| 3096 | _Py_isupper__doc__}, |
| 3097 | {"join", (PyCFunction)bytes_join, METH_O, join_doc}, |
| 3098 | {"ljust", (PyCFunction)stringlib_ljust, METH_VARARGS, ljust__doc__}, |
| 3099 | {"lower", (PyCFunction)stringlib_lower, METH_NOARGS, _Py_lower__doc__}, |
| 3100 | {"lstrip", (PyCFunction)bytes_lstrip, METH_VARARGS, lstrip__doc__}, |
| 3101 | {"partition", (PyCFunction)bytes_partition, METH_O, partition__doc__}, |
| 3102 | {"pop", (PyCFunction)bytes_pop, METH_VARARGS, pop__doc__}, |
| 3103 | {"remove", (PyCFunction)bytes_remove, METH_O, remove__doc__}, |
| 3104 | {"replace", (PyCFunction)bytes_replace, METH_VARARGS, replace__doc__}, |
| 3105 | {"reverse", (PyCFunction)bytes_reverse, METH_NOARGS, reverse__doc__}, |
| 3106 | {"rfind", (PyCFunction)bytes_rfind, METH_VARARGS, rfind__doc__}, |
| 3107 | {"rindex", (PyCFunction)bytes_rindex, METH_VARARGS, rindex__doc__}, |
| 3108 | {"rjust", (PyCFunction)stringlib_rjust, METH_VARARGS, rjust__doc__}, |
| 3109 | {"rpartition", (PyCFunction)bytes_rpartition, METH_O, rpartition__doc__}, |
| 3110 | {"rsplit", (PyCFunction)bytes_rsplit, METH_VARARGS, rsplit__doc__}, |
| 3111 | {"rstrip", (PyCFunction)bytes_rstrip, METH_VARARGS, rstrip__doc__}, |
| 3112 | {"split", (PyCFunction)bytes_split, METH_VARARGS, split__doc__}, |
| 3113 | {"splitlines", (PyCFunction)stringlib_splitlines, METH_VARARGS, |
| 3114 | splitlines__doc__}, |
| 3115 | {"startswith", (PyCFunction)bytes_startswith, METH_VARARGS , |
| 3116 | startswith__doc__}, |
| 3117 | {"strip", (PyCFunction)bytes_strip, METH_VARARGS, strip__doc__}, |
| 3118 | {"swapcase", (PyCFunction)stringlib_swapcase, METH_NOARGS, |
| 3119 | _Py_swapcase__doc__}, |
| 3120 | {"title", (PyCFunction)stringlib_title, METH_NOARGS, _Py_title__doc__}, |
| 3121 | {"translate", (PyCFunction)bytes_translate, METH_VARARGS, |
| 3122 | translate__doc__}, |
| 3123 | {"upper", (PyCFunction)stringlib_upper, METH_NOARGS, _Py_upper__doc__}, |
| 3124 | {"zfill", (PyCFunction)stringlib_zfill, METH_VARARGS, zfill__doc__}, |
| 3125 | {NULL} |
| 3126 | }; |
| 3127 | |
| 3128 | PyDoc_STRVAR(bytes_doc, |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 3129 | "bytearray(iterable_of_ints) -> bytearray\n\ |
| 3130 | bytearray(string, encoding[, errors]) -> bytearray\n\ |
| 3131 | bytearray(bytes_or_bytearray) -> mutable copy of bytes_or_bytearray\n\ |
| 3132 | bytearray(memory_view) -> bytearray\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 3133 | \n\ |
| 3134 | Construct an mutable bytearray object from:\n\ |
| 3135 | - an iterable yielding integers in range(256)\n\ |
| 3136 | - a text string encoded using the specified encoding\n\ |
| 3137 | - a bytes or a bytearray object\n\ |
| 3138 | - any object implementing the buffer API.\n\ |
| 3139 | \n\ |
Georg Brandl | 17cb8a8 | 2008-05-30 08:20:09 +0000 | [diff] [blame] | 3140 | bytearray(int) -> bytearray\n\ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 3141 | \n\ |
| 3142 | Construct a zero-initialized bytearray of the given length."); |
| 3143 | |
| 3144 | |
| 3145 | static PyObject *bytes_iter(PyObject *seq); |
| 3146 | |
| 3147 | PyTypeObject PyByteArray_Type = { |
| 3148 | PyVarObject_HEAD_INIT(&PyType_Type, 0) |
| 3149 | "bytearray", |
| 3150 | sizeof(PyByteArrayObject), |
| 3151 | 0, |
| 3152 | (destructor)bytes_dealloc, /* tp_dealloc */ |
| 3153 | 0, /* tp_print */ |
| 3154 | 0, /* tp_getattr */ |
| 3155 | 0, /* tp_setattr */ |
| 3156 | 0, /* tp_compare */ |
| 3157 | (reprfunc)bytes_repr, /* tp_repr */ |
| 3158 | 0, /* tp_as_number */ |
| 3159 | &bytes_as_sequence, /* tp_as_sequence */ |
| 3160 | &bytes_as_mapping, /* tp_as_mapping */ |
| 3161 | 0, /* tp_hash */ |
| 3162 | 0, /* tp_call */ |
| 3163 | bytes_str, /* tp_str */ |
| 3164 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 3165 | 0, /* tp_setattro */ |
| 3166 | &bytes_as_buffer, /* tp_as_buffer */ |
| 3167 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 3168 | bytes_doc, /* tp_doc */ |
| 3169 | 0, /* tp_traverse */ |
| 3170 | 0, /* tp_clear */ |
| 3171 | (richcmpfunc)bytes_richcompare, /* tp_richcompare */ |
| 3172 | 0, /* tp_weaklistoffset */ |
| 3173 | bytes_iter, /* tp_iter */ |
| 3174 | 0, /* tp_iternext */ |
| 3175 | bytes_methods, /* tp_methods */ |
| 3176 | 0, /* tp_members */ |
| 3177 | 0, /* tp_getset */ |
| 3178 | 0, /* tp_base */ |
| 3179 | 0, /* tp_dict */ |
| 3180 | 0, /* tp_descr_get */ |
| 3181 | 0, /* tp_descr_set */ |
| 3182 | 0, /* tp_dictoffset */ |
| 3183 | (initproc)bytes_init, /* tp_init */ |
| 3184 | PyType_GenericAlloc, /* tp_alloc */ |
| 3185 | PyType_GenericNew, /* tp_new */ |
| 3186 | PyObject_Del, /* tp_free */ |
| 3187 | }; |
| 3188 | |
| 3189 | /*********************** Bytes Iterator ****************************/ |
| 3190 | |
| 3191 | typedef struct { |
| 3192 | PyObject_HEAD |
| 3193 | Py_ssize_t it_index; |
| 3194 | PyByteArrayObject *it_seq; /* Set to NULL when iterator is exhausted */ |
| 3195 | } bytesiterobject; |
| 3196 | |
| 3197 | static void |
| 3198 | bytesiter_dealloc(bytesiterobject *it) |
| 3199 | { |
| 3200 | _PyObject_GC_UNTRACK(it); |
| 3201 | Py_XDECREF(it->it_seq); |
| 3202 | PyObject_GC_Del(it); |
| 3203 | } |
| 3204 | |
| 3205 | static int |
| 3206 | bytesiter_traverse(bytesiterobject *it, visitproc visit, void *arg) |
| 3207 | { |
| 3208 | Py_VISIT(it->it_seq); |
| 3209 | return 0; |
| 3210 | } |
| 3211 | |
| 3212 | static PyObject * |
| 3213 | bytesiter_next(bytesiterobject *it) |
| 3214 | { |
| 3215 | PyByteArrayObject *seq; |
| 3216 | PyObject *item; |
| 3217 | |
| 3218 | assert(it != NULL); |
| 3219 | seq = it->it_seq; |
| 3220 | if (seq == NULL) |
| 3221 | return NULL; |
| 3222 | assert(PyByteArray_Check(seq)); |
| 3223 | |
| 3224 | if (it->it_index < PyByteArray_GET_SIZE(seq)) { |
| 3225 | item = PyLong_FromLong( |
| 3226 | (unsigned char)seq->ob_bytes[it->it_index]); |
| 3227 | if (item != NULL) |
| 3228 | ++it->it_index; |
| 3229 | return item; |
| 3230 | } |
| 3231 | |
| 3232 | Py_DECREF(seq); |
| 3233 | it->it_seq = NULL; |
| 3234 | return NULL; |
| 3235 | } |
| 3236 | |
| 3237 | static PyObject * |
| 3238 | bytesiter_length_hint(bytesiterobject *it) |
| 3239 | { |
| 3240 | Py_ssize_t len = 0; |
| 3241 | if (it->it_seq) |
| 3242 | len = PyByteArray_GET_SIZE(it->it_seq) - it->it_index; |
| 3243 | return PyLong_FromSsize_t(len); |
| 3244 | } |
| 3245 | |
| 3246 | PyDoc_STRVAR(length_hint_doc, |
| 3247 | "Private method returning an estimate of len(list(it))."); |
| 3248 | |
| 3249 | static PyMethodDef bytesiter_methods[] = { |
| 3250 | {"__length_hint__", (PyCFunction)bytesiter_length_hint, METH_NOARGS, |
| 3251 | length_hint_doc}, |
| 3252 | {NULL, NULL} /* sentinel */ |
| 3253 | }; |
| 3254 | |
| 3255 | PyTypeObject PyByteArrayIter_Type = { |
| 3256 | PyVarObject_HEAD_INIT(&PyType_Type, 0) |
| 3257 | "bytearray_iterator", /* tp_name */ |
| 3258 | sizeof(bytesiterobject), /* tp_basicsize */ |
| 3259 | 0, /* tp_itemsize */ |
| 3260 | /* methods */ |
| 3261 | (destructor)bytesiter_dealloc, /* tp_dealloc */ |
| 3262 | 0, /* tp_print */ |
| 3263 | 0, /* tp_getattr */ |
| 3264 | 0, /* tp_setattr */ |
| 3265 | 0, /* tp_compare */ |
| 3266 | 0, /* tp_repr */ |
| 3267 | 0, /* tp_as_number */ |
| 3268 | 0, /* tp_as_sequence */ |
| 3269 | 0, /* tp_as_mapping */ |
| 3270 | 0, /* tp_hash */ |
| 3271 | 0, /* tp_call */ |
| 3272 | 0, /* tp_str */ |
| 3273 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 3274 | 0, /* tp_setattro */ |
| 3275 | 0, /* tp_as_buffer */ |
| 3276 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */ |
| 3277 | 0, /* tp_doc */ |
| 3278 | (traverseproc)bytesiter_traverse, /* tp_traverse */ |
| 3279 | 0, /* tp_clear */ |
| 3280 | 0, /* tp_richcompare */ |
| 3281 | 0, /* tp_weaklistoffset */ |
| 3282 | PyObject_SelfIter, /* tp_iter */ |
| 3283 | (iternextfunc)bytesiter_next, /* tp_iternext */ |
| 3284 | bytesiter_methods, /* tp_methods */ |
| 3285 | 0, |
| 3286 | }; |
| 3287 | |
| 3288 | static PyObject * |
| 3289 | bytes_iter(PyObject *seq) |
| 3290 | { |
| 3291 | bytesiterobject *it; |
| 3292 | |
| 3293 | if (!PyByteArray_Check(seq)) { |
| 3294 | PyErr_BadInternalCall(); |
| 3295 | return NULL; |
| 3296 | } |
| 3297 | it = PyObject_GC_New(bytesiterobject, &PyByteArrayIter_Type); |
| 3298 | if (it == NULL) |
| 3299 | return NULL; |
| 3300 | it->it_index = 0; |
| 3301 | Py_INCREF(seq); |
| 3302 | it->it_seq = (PyByteArrayObject *)seq; |
| 3303 | _PyObject_GC_TRACK(it); |
| 3304 | return (PyObject *)it; |
| 3305 | } |