Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1 | /* |
| 2 | * ElementTree |
| 3 | * $Id: /work/modules/celementtree/cElementTree.c 1128 2005-12-16T21:57:13.668520Z Fredrik $ |
| 4 | * |
| 5 | * elementtree accelerator |
| 6 | * |
| 7 | * History: |
| 8 | * 1999-06-20 fl created (as part of sgmlop) |
| 9 | * 2001-05-29 fl effdom edition |
| 10 | * 2001-06-05 fl backported to unix; fixed bogus free in clear |
| 11 | * 2001-07-10 fl added findall helper |
| 12 | * 2003-02-27 fl elementtree edition (alpha) |
| 13 | * 2004-06-03 fl updates for elementtree 1.2 |
| 14 | * 2005-01-05 fl added universal name cache, Element/SubElement factories |
| 15 | * 2005-01-06 fl moved python helpers into C module; removed 1.5.2 support |
| 16 | * 2005-01-07 fl added 2.1 support; work around broken __copy__ in 2.3 |
| 17 | * 2005-01-08 fl added makeelement method; fixed path support |
| 18 | * 2005-01-10 fl optimized memory usage |
| 19 | * 2005-01-11 fl first public release (cElementTree 0.8) |
| 20 | * 2005-01-12 fl split element object into base and extras |
| 21 | * 2005-01-13 fl use tagged pointers for tail/text (cElementTree 0.9) |
| 22 | * 2005-01-17 fl added treebuilder close method |
| 23 | * 2005-01-17 fl fixed crash in getchildren |
| 24 | * 2005-01-18 fl removed observer api, added iterparse (cElementTree 0.9.3) |
| 25 | * 2005-01-23 fl revised iterparse api; added namespace event support (0.9.8) |
| 26 | * 2005-01-26 fl added VERSION module property (cElementTree 1.0) |
| 27 | * 2005-01-28 fl added remove method (1.0.1) |
| 28 | * 2005-03-01 fl added iselement function; fixed makeelement aliasing (1.0.2) |
| 29 | * 2005-03-13 fl export Comment and ProcessingInstruction/PI helpers |
| 30 | * 2005-03-26 fl added Comment and PI support to XMLParser |
| 31 | * 2005-03-27 fl event optimizations; complain about bogus events |
| 32 | * 2005-08-08 fl fixed read error handling in parse |
| 33 | * 2005-08-11 fl added runtime test for copy workaround (1.0.3) |
| 34 | * 2005-12-13 fl added expat_capi support (for xml.etree) (1.0.4) |
| 35 | * 2005-12-16 fl added support for non-standard encodings |
| 36 | * |
| 37 | * Copyright (c) 1999-2005 by Secret Labs AB. All rights reserved. |
| 38 | * Copyright (c) 1999-2005 by Fredrik Lundh. |
| 39 | * |
| 40 | * info@pythonware.com |
| 41 | * http://www.pythonware.com |
| 42 | */ |
| 43 | |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 44 | /* Licensed to PSF under a Contributor Agreement. */ |
| 45 | /* See http://www.python.org/2.4/license for licensing details. */ |
| 46 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 47 | #include "Python.h" |
| 48 | |
| 49 | #define VERSION "1.0.5" |
| 50 | |
| 51 | /* -------------------------------------------------------------------- */ |
| 52 | /* configuration */ |
| 53 | |
| 54 | /* Leave defined to include the expat-based XMLParser type */ |
| 55 | #define USE_EXPAT |
| 56 | |
| 57 | /* Define to to all expat calls via pyexpat's embedded expat library */ |
| 58 | /* #define USE_PYEXPAT_CAPI */ |
| 59 | |
| 60 | /* An element can hold this many children without extra memory |
| 61 | allocations. */ |
| 62 | #define STATIC_CHILDREN 4 |
| 63 | |
| 64 | /* For best performance, chose a value so that 80-90% of all nodes |
| 65 | have no more than the given number of children. Set this to zero |
| 66 | to minimize the size of the element structure itself (this only |
| 67 | helps if you have lots of leaf nodes with attributes). */ |
| 68 | |
| 69 | /* Also note that pymalloc always allocates blocks in multiples of |
| 70 | eight bytes. For the current version of cElementTree, this means |
| 71 | that the number of children should be an even number, at least on |
| 72 | 32-bit platforms. */ |
| 73 | |
| 74 | /* -------------------------------------------------------------------- */ |
| 75 | |
| 76 | #if 0 |
| 77 | static int memory = 0; |
| 78 | #define ALLOC(size, comment)\ |
| 79 | do { memory += size; printf("%8d - %s\n", memory, comment); } while (0) |
| 80 | #define RELEASE(size, comment)\ |
| 81 | do { memory -= size; printf("%8d - %s\n", memory, comment); } while (0) |
| 82 | #else |
| 83 | #define ALLOC(size, comment) |
| 84 | #define RELEASE(size, comment) |
| 85 | #endif |
| 86 | |
| 87 | /* compiler tweaks */ |
| 88 | #if defined(_MSC_VER) |
| 89 | #define LOCAL(type) static __inline type __fastcall |
| 90 | #else |
| 91 | #define LOCAL(type) static type |
| 92 | #endif |
| 93 | |
| 94 | /* compatibility macros */ |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 95 | #if (PY_VERSION_HEX < 0x02050000) |
| 96 | typedef int Py_ssize_t; |
| 97 | #endif |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 98 | #if (PY_VERSION_HEX < 0x02040000) |
| 99 | #define PyDict_CheckExact PyDict_Check |
| 100 | #if (PY_VERSION_HEX < 0x02020000) |
| 101 | #define PyList_CheckExact PyList_Check |
| 102 | #define PyString_CheckExact PyString_Check |
| 103 | #if (PY_VERSION_HEX >= 0x01060000) |
| 104 | #define Py_USING_UNICODE /* always enabled for 2.0 and 2.1 */ |
| 105 | #endif |
| 106 | #endif |
| 107 | #endif |
| 108 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 109 | #if !defined(Py_RETURN_NONE) |
| 110 | #define Py_RETURN_NONE return Py_INCREF(Py_None), Py_None |
| 111 | #endif |
| 112 | |
| 113 | /* macros used to store 'join' flags in string object pointers. note |
| 114 | that all use of text and tail as object pointers must be wrapped in |
| 115 | JOIN_OBJ. see comments in the ElementObject definition for more |
| 116 | info. */ |
| 117 | #define JOIN_GET(p) ((Py_uintptr_t) (p) & 1) |
| 118 | #define JOIN_SET(p, flag) ((void*) ((Py_uintptr_t) (JOIN_OBJ(p)) | (flag))) |
| 119 | #define JOIN_OBJ(p) ((PyObject*) ((Py_uintptr_t) (p) & ~1)) |
| 120 | |
| 121 | /* glue functions (see the init function for details) */ |
| 122 | static PyObject* elementtree_copyelement_obj; |
| 123 | static PyObject* elementtree_deepcopy_obj; |
| 124 | static PyObject* elementtree_getiterator_obj; |
| 125 | static PyObject* elementpath_obj; |
| 126 | |
| 127 | /* helpers */ |
| 128 | |
| 129 | LOCAL(PyObject*) |
| 130 | deepcopy(PyObject* object, PyObject* memo) |
| 131 | { |
| 132 | /* do a deep copy of the given object */ |
| 133 | |
| 134 | PyObject* args; |
| 135 | PyObject* result; |
| 136 | |
| 137 | if (!elementtree_deepcopy_obj) { |
| 138 | PyErr_SetString( |
| 139 | PyExc_RuntimeError, |
| 140 | "deepcopy helper not found" |
| 141 | ); |
| 142 | return NULL; |
| 143 | } |
| 144 | |
| 145 | args = PyTuple_New(2); |
| 146 | Py_INCREF(object); PyTuple_SET_ITEM(args, 0, (PyObject*) object); |
| 147 | Py_INCREF(memo); PyTuple_SET_ITEM(args, 1, (PyObject*) memo); |
| 148 | |
| 149 | result = PyObject_CallObject(elementtree_deepcopy_obj, args); |
| 150 | |
| 151 | Py_DECREF(args); |
| 152 | |
| 153 | return result; |
| 154 | } |
| 155 | |
| 156 | LOCAL(PyObject*) |
| 157 | list_join(PyObject* list) |
| 158 | { |
| 159 | /* join list elements (destroying the list in the process) */ |
| 160 | |
| 161 | PyObject* joiner; |
| 162 | PyObject* function; |
| 163 | PyObject* args; |
| 164 | PyObject* result; |
| 165 | |
| 166 | switch (PyList_GET_SIZE(list)) { |
| 167 | case 0: |
| 168 | Py_DECREF(list); |
| 169 | return PyString_FromString(""); |
| 170 | case 1: |
| 171 | result = PyList_GET_ITEM(list, 0); |
| 172 | Py_INCREF(result); |
| 173 | Py_DECREF(list); |
| 174 | return result; |
| 175 | } |
| 176 | |
| 177 | /* two or more elements: slice out a suitable separator from the |
| 178 | first member, and use that to join the entire list */ |
| 179 | |
| 180 | joiner = PySequence_GetSlice(PyList_GET_ITEM(list, 0), 0, 0); |
| 181 | if (!joiner) |
| 182 | return NULL; |
| 183 | |
| 184 | function = PyObject_GetAttrString(joiner, "join"); |
| 185 | if (!function) { |
| 186 | Py_DECREF(joiner); |
| 187 | return NULL; |
| 188 | } |
| 189 | |
| 190 | args = PyTuple_New(1); |
| 191 | PyTuple_SET_ITEM(args, 0, list); |
| 192 | |
| 193 | result = PyObject_CallObject(function, args); |
| 194 | |
| 195 | Py_DECREF(args); /* also removes list */ |
| 196 | Py_DECREF(function); |
| 197 | Py_DECREF(joiner); |
| 198 | |
| 199 | return result; |
| 200 | } |
| 201 | |
| 202 | #if (PY_VERSION_HEX < 0x02020000) |
| 203 | LOCAL(int) |
| 204 | PyDict_Update(PyObject* dict, PyObject* other) |
| 205 | { |
| 206 | /* PyDict_Update emulation for 2.1 and earlier */ |
| 207 | |
| 208 | PyObject* res; |
| 209 | |
| 210 | res = PyObject_CallMethod(dict, "update", "O", other); |
| 211 | if (!res) |
| 212 | return -1; |
| 213 | |
| 214 | Py_DECREF(res); |
| 215 | return 0; |
| 216 | } |
| 217 | #endif |
| 218 | |
| 219 | /* -------------------------------------------------------------------- */ |
| 220 | /* the element type */ |
| 221 | |
| 222 | typedef struct { |
| 223 | |
| 224 | /* attributes (a dictionary object), or None if no attributes */ |
| 225 | PyObject* attrib; |
| 226 | |
| 227 | /* child elements */ |
| 228 | int length; /* actual number of items */ |
| 229 | int allocated; /* allocated items */ |
| 230 | |
| 231 | /* this either points to _children or to a malloced buffer */ |
| 232 | PyObject* *children; |
| 233 | |
| 234 | PyObject* _children[STATIC_CHILDREN]; |
| 235 | |
| 236 | } ElementObjectExtra; |
| 237 | |
| 238 | typedef struct { |
| 239 | PyObject_HEAD |
| 240 | |
| 241 | /* element tag (a string). */ |
| 242 | PyObject* tag; |
| 243 | |
| 244 | /* text before first child. note that this is a tagged pointer; |
| 245 | use JOIN_OBJ to get the object pointer. the join flag is used |
| 246 | to distinguish lists created by the tree builder from lists |
| 247 | assigned to the attribute by application code; the former |
| 248 | should be joined before being returned to the user, the latter |
| 249 | should be left intact. */ |
| 250 | PyObject* text; |
| 251 | |
| 252 | /* text after this element, in parent. note that this is a tagged |
| 253 | pointer; use JOIN_OBJ to get the object pointer. */ |
| 254 | PyObject* tail; |
| 255 | |
| 256 | ElementObjectExtra* extra; |
| 257 | |
| 258 | } ElementObject; |
| 259 | |
| 260 | staticforward PyTypeObject Element_Type; |
| 261 | |
| 262 | #define Element_CheckExact(op) ((op)->ob_type == &Element_Type) |
| 263 | |
| 264 | /* -------------------------------------------------------------------- */ |
| 265 | /* element constructor and destructor */ |
| 266 | |
| 267 | LOCAL(int) |
| 268 | element_new_extra(ElementObject* self, PyObject* attrib) |
| 269 | { |
| 270 | self->extra = PyObject_Malloc(sizeof(ElementObjectExtra)); |
| 271 | if (!self->extra) |
| 272 | return -1; |
| 273 | |
| 274 | if (!attrib) |
| 275 | attrib = Py_None; |
| 276 | |
| 277 | Py_INCREF(attrib); |
| 278 | self->extra->attrib = attrib; |
| 279 | |
| 280 | self->extra->length = 0; |
| 281 | self->extra->allocated = STATIC_CHILDREN; |
| 282 | self->extra->children = self->extra->_children; |
| 283 | |
| 284 | return 0; |
| 285 | } |
| 286 | |
| 287 | LOCAL(void) |
| 288 | element_dealloc_extra(ElementObject* self) |
| 289 | { |
| 290 | int i; |
| 291 | |
| 292 | Py_DECREF(self->extra->attrib); |
| 293 | |
| 294 | for (i = 0; i < self->extra->length; i++) |
| 295 | Py_DECREF(self->extra->children[i]); |
| 296 | |
| 297 | if (self->extra->children != self->extra->_children) |
| 298 | PyObject_Free(self->extra->children); |
| 299 | |
| 300 | PyObject_Free(self->extra); |
| 301 | } |
| 302 | |
| 303 | LOCAL(PyObject*) |
| 304 | element_new(PyObject* tag, PyObject* attrib) |
| 305 | { |
| 306 | ElementObject* self; |
| 307 | |
| 308 | self = PyObject_New(ElementObject, &Element_Type); |
| 309 | if (self == NULL) |
| 310 | return NULL; |
| 311 | |
| 312 | /* use None for empty dictionaries */ |
| 313 | if (PyDict_CheckExact(attrib) && !PyDict_Size(attrib)) |
| 314 | attrib = Py_None; |
| 315 | |
| 316 | self->extra = NULL; |
| 317 | |
| 318 | if (attrib != Py_None) { |
| 319 | |
| 320 | if (element_new_extra(self, attrib) < 0) |
| 321 | return NULL; |
| 322 | |
| 323 | self->extra->length = 0; |
| 324 | self->extra->allocated = STATIC_CHILDREN; |
| 325 | self->extra->children = self->extra->_children; |
| 326 | |
| 327 | } |
| 328 | |
| 329 | Py_INCREF(tag); |
| 330 | self->tag = tag; |
| 331 | |
| 332 | Py_INCREF(Py_None); |
| 333 | self->text = Py_None; |
| 334 | |
| 335 | Py_INCREF(Py_None); |
| 336 | self->tail = Py_None; |
| 337 | |
| 338 | ALLOC(sizeof(ElementObject), "create element"); |
| 339 | |
| 340 | return (PyObject*) self; |
| 341 | } |
| 342 | |
| 343 | LOCAL(int) |
| 344 | element_resize(ElementObject* self, int extra) |
| 345 | { |
| 346 | int size; |
| 347 | PyObject* *children; |
| 348 | |
| 349 | /* make sure self->children can hold the given number of extra |
| 350 | elements. set an exception and return -1 if allocation failed */ |
| 351 | |
| 352 | if (!self->extra) |
| 353 | element_new_extra(self, NULL); |
| 354 | |
| 355 | size = self->extra->length + extra; |
| 356 | |
| 357 | if (size > self->extra->allocated) { |
| 358 | /* use Python 2.4's list growth strategy */ |
| 359 | size = (size >> 3) + (size < 9 ? 3 : 6) + size; |
| 360 | if (self->extra->children != self->extra->_children) { |
| 361 | children = PyObject_Realloc(self->extra->children, |
| 362 | size * sizeof(PyObject*)); |
| 363 | if (!children) |
| 364 | goto nomemory; |
| 365 | } else { |
| 366 | children = PyObject_Malloc(size * sizeof(PyObject*)); |
| 367 | if (!children) |
| 368 | goto nomemory; |
| 369 | /* copy existing children from static area to malloc buffer */ |
| 370 | memcpy(children, self->extra->children, |
| 371 | self->extra->length * sizeof(PyObject*)); |
| 372 | } |
| 373 | self->extra->children = children; |
| 374 | self->extra->allocated = size; |
| 375 | } |
| 376 | |
| 377 | return 0; |
| 378 | |
| 379 | nomemory: |
| 380 | PyErr_NoMemory(); |
| 381 | return -1; |
| 382 | } |
| 383 | |
| 384 | LOCAL(int) |
| 385 | element_add_subelement(ElementObject* self, PyObject* element) |
| 386 | { |
| 387 | /* add a child element to a parent */ |
| 388 | |
| 389 | if (element_resize(self, 1) < 0) |
| 390 | return -1; |
| 391 | |
| 392 | Py_INCREF(element); |
| 393 | self->extra->children[self->extra->length] = element; |
| 394 | |
| 395 | self->extra->length++; |
| 396 | |
| 397 | return 0; |
| 398 | } |
| 399 | |
| 400 | LOCAL(PyObject*) |
| 401 | element_get_attrib(ElementObject* self) |
| 402 | { |
| 403 | /* return borrowed reference to attrib dictionary */ |
| 404 | /* note: this function assumes that the extra section exists */ |
| 405 | |
| 406 | PyObject* res = self->extra->attrib; |
| 407 | |
| 408 | if (res == Py_None) { |
| 409 | /* create missing dictionary */ |
| 410 | res = PyDict_New(); |
| 411 | if (!res) |
| 412 | return NULL; |
| 413 | self->extra->attrib = res; |
| 414 | } |
| 415 | |
| 416 | return res; |
| 417 | } |
| 418 | |
| 419 | LOCAL(PyObject*) |
| 420 | element_get_text(ElementObject* self) |
| 421 | { |
| 422 | /* return borrowed reference to text attribute */ |
| 423 | |
| 424 | PyObject* res = self->text; |
| 425 | |
| 426 | if (JOIN_GET(res)) { |
| 427 | res = JOIN_OBJ(res); |
| 428 | if (PyList_CheckExact(res)) { |
| 429 | res = list_join(res); |
| 430 | if (!res) |
| 431 | return NULL; |
| 432 | self->text = res; |
| 433 | } |
| 434 | } |
| 435 | |
| 436 | return res; |
| 437 | } |
| 438 | |
| 439 | LOCAL(PyObject*) |
| 440 | element_get_tail(ElementObject* self) |
| 441 | { |
| 442 | /* return borrowed reference to text attribute */ |
| 443 | |
| 444 | PyObject* res = self->tail; |
| 445 | |
| 446 | if (JOIN_GET(res)) { |
| 447 | res = JOIN_OBJ(res); |
| 448 | if (PyList_CheckExact(res)) { |
| 449 | res = list_join(res); |
| 450 | if (!res) |
| 451 | return NULL; |
| 452 | self->tail = res; |
| 453 | } |
| 454 | } |
| 455 | |
| 456 | return res; |
| 457 | } |
| 458 | |
| 459 | static PyObject* |
| 460 | element(PyObject* self, PyObject* args, PyObject* kw) |
| 461 | { |
| 462 | PyObject* elem; |
| 463 | |
| 464 | PyObject* tag; |
| 465 | PyObject* attrib = NULL; |
| 466 | if (!PyArg_ParseTuple(args, "O|O!:Element", &tag, |
| 467 | &PyDict_Type, &attrib)) |
| 468 | return NULL; |
| 469 | |
| 470 | if (attrib || kw) { |
| 471 | attrib = (attrib) ? PyDict_Copy(attrib) : PyDict_New(); |
| 472 | if (!attrib) |
| 473 | return NULL; |
| 474 | if (kw) |
| 475 | PyDict_Update(attrib, kw); |
| 476 | } else { |
| 477 | Py_INCREF(Py_None); |
| 478 | attrib = Py_None; |
| 479 | } |
| 480 | |
| 481 | elem = element_new(tag, attrib); |
| 482 | |
| 483 | Py_DECREF(attrib); |
| 484 | |
| 485 | return elem; |
| 486 | } |
| 487 | |
| 488 | static PyObject* |
| 489 | subelement(PyObject* self, PyObject* args, PyObject* kw) |
| 490 | { |
| 491 | PyObject* elem; |
| 492 | |
| 493 | ElementObject* parent; |
| 494 | PyObject* tag; |
| 495 | PyObject* attrib = NULL; |
| 496 | if (!PyArg_ParseTuple(args, "O!O|O!:SubElement", |
| 497 | &Element_Type, &parent, &tag, |
| 498 | &PyDict_Type, &attrib)) |
| 499 | return NULL; |
| 500 | |
| 501 | if (attrib || kw) { |
| 502 | attrib = (attrib) ? PyDict_Copy(attrib) : PyDict_New(); |
| 503 | if (!attrib) |
| 504 | return NULL; |
| 505 | if (kw) |
| 506 | PyDict_Update(attrib, kw); |
| 507 | } else { |
| 508 | Py_INCREF(Py_None); |
| 509 | attrib = Py_None; |
| 510 | } |
| 511 | |
| 512 | elem = element_new(tag, attrib); |
| 513 | |
| 514 | Py_DECREF(attrib); |
| 515 | |
| 516 | if (element_add_subelement(parent, elem) < 0) |
| 517 | return NULL; |
| 518 | |
| 519 | return elem; |
| 520 | } |
| 521 | |
| 522 | static void |
| 523 | element_dealloc(ElementObject* self) |
| 524 | { |
| 525 | if (self->extra) |
| 526 | element_dealloc_extra(self); |
| 527 | |
| 528 | /* discard attributes */ |
| 529 | Py_DECREF(self->tag); |
| 530 | Py_DECREF(JOIN_OBJ(self->text)); |
| 531 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 532 | |
| 533 | RELEASE(sizeof(ElementObject), "destroy element"); |
| 534 | |
| 535 | PyObject_Del(self); |
| 536 | } |
| 537 | |
| 538 | /* -------------------------------------------------------------------- */ |
| 539 | /* methods (in alphabetical order) */ |
| 540 | |
| 541 | static PyObject* |
| 542 | element_append(ElementObject* self, PyObject* args) |
| 543 | { |
| 544 | PyObject* element; |
| 545 | if (!PyArg_ParseTuple(args, "O!:append", &Element_Type, &element)) |
| 546 | return NULL; |
| 547 | |
| 548 | if (element_add_subelement(self, element) < 0) |
| 549 | return NULL; |
| 550 | |
| 551 | Py_RETURN_NONE; |
| 552 | } |
| 553 | |
| 554 | static PyObject* |
| 555 | element_clear(ElementObject* self, PyObject* args) |
| 556 | { |
| 557 | if (!PyArg_ParseTuple(args, ":clear")) |
| 558 | return NULL; |
| 559 | |
| 560 | if (self->extra) { |
| 561 | element_dealloc_extra(self); |
| 562 | self->extra = NULL; |
| 563 | } |
| 564 | |
| 565 | Py_INCREF(Py_None); |
| 566 | Py_DECREF(JOIN_OBJ(self->text)); |
| 567 | self->text = Py_None; |
| 568 | |
| 569 | Py_INCREF(Py_None); |
| 570 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 571 | self->tail = Py_None; |
| 572 | |
| 573 | Py_RETURN_NONE; |
| 574 | } |
| 575 | |
| 576 | static PyObject* |
| 577 | element_copy(ElementObject* self, PyObject* args) |
| 578 | { |
| 579 | int i; |
| 580 | ElementObject* element; |
| 581 | |
| 582 | if (!PyArg_ParseTuple(args, ":__copy__")) |
| 583 | return NULL; |
| 584 | |
| 585 | element = (ElementObject*) element_new( |
| 586 | self->tag, (self->extra) ? self->extra->attrib : Py_None |
| 587 | ); |
| 588 | if (!element) |
| 589 | return NULL; |
| 590 | |
| 591 | Py_DECREF(JOIN_OBJ(element->text)); |
| 592 | element->text = self->text; |
| 593 | Py_INCREF(JOIN_OBJ(element->text)); |
| 594 | |
| 595 | Py_DECREF(JOIN_OBJ(element->tail)); |
| 596 | element->tail = self->tail; |
| 597 | Py_INCREF(JOIN_OBJ(element->tail)); |
| 598 | |
| 599 | if (self->extra) { |
| 600 | |
| 601 | if (element_resize(element, self->extra->length) < 0) |
| 602 | return NULL; |
| 603 | |
| 604 | for (i = 0; i < self->extra->length; i++) { |
| 605 | Py_INCREF(self->extra->children[i]); |
| 606 | element->extra->children[i] = self->extra->children[i]; |
| 607 | } |
| 608 | |
| 609 | element->extra->length = self->extra->length; |
| 610 | |
| 611 | } |
| 612 | |
| 613 | return (PyObject*) element; |
| 614 | } |
| 615 | |
| 616 | static PyObject* |
| 617 | element_deepcopy(ElementObject* self, PyObject* args) |
| 618 | { |
| 619 | int i; |
| 620 | ElementObject* element; |
| 621 | PyObject* tag; |
| 622 | PyObject* attrib; |
| 623 | PyObject* text; |
| 624 | PyObject* tail; |
| 625 | PyObject* id; |
| 626 | |
| 627 | PyObject* memo; |
| 628 | if (!PyArg_ParseTuple(args, "O:__deepcopy__", &memo)) |
| 629 | return NULL; |
| 630 | |
| 631 | tag = deepcopy(self->tag, memo); |
| 632 | if (!tag) |
| 633 | return NULL; |
| 634 | |
| 635 | if (self->extra) { |
| 636 | attrib = deepcopy(self->extra->attrib, memo); |
| 637 | if (!attrib) { |
| 638 | Py_DECREF(tag); |
| 639 | return NULL; |
| 640 | } |
| 641 | } else { |
| 642 | Py_INCREF(Py_None); |
| 643 | attrib = Py_None; |
| 644 | } |
| 645 | |
| 646 | element = (ElementObject*) element_new(tag, attrib); |
| 647 | |
| 648 | Py_DECREF(tag); |
| 649 | Py_DECREF(attrib); |
| 650 | |
| 651 | if (!element) |
| 652 | return NULL; |
| 653 | |
| 654 | text = deepcopy(JOIN_OBJ(self->text), memo); |
| 655 | if (!text) |
| 656 | goto error; |
| 657 | Py_DECREF(element->text); |
| 658 | element->text = JOIN_SET(text, JOIN_GET(self->text)); |
| 659 | |
| 660 | tail = deepcopy(JOIN_OBJ(self->tail), memo); |
| 661 | if (!tail) |
| 662 | goto error; |
| 663 | Py_DECREF(element->tail); |
| 664 | element->tail = JOIN_SET(tail, JOIN_GET(self->tail)); |
| 665 | |
| 666 | if (self->extra) { |
| 667 | |
| 668 | if (element_resize(element, self->extra->length) < 0) |
| 669 | goto error; |
| 670 | |
| 671 | for (i = 0; i < self->extra->length; i++) { |
| 672 | PyObject* child = deepcopy(self->extra->children[i], memo); |
| 673 | if (!child) { |
| 674 | element->extra->length = i; |
| 675 | goto error; |
| 676 | } |
| 677 | element->extra->children[i] = child; |
| 678 | } |
| 679 | |
| 680 | element->extra->length = self->extra->length; |
| 681 | |
| 682 | } |
| 683 | |
| 684 | /* add object to memo dictionary (so deepcopy won't visit it again) */ |
| 685 | id = PyInt_FromLong((Py_uintptr_t) self); |
| 686 | |
| 687 | i = PyDict_SetItem(memo, id, (PyObject*) element); |
| 688 | |
| 689 | Py_DECREF(id); |
| 690 | |
| 691 | if (i < 0) |
| 692 | goto error; |
| 693 | |
| 694 | return (PyObject*) element; |
| 695 | |
| 696 | error: |
| 697 | Py_DECREF(element); |
| 698 | return NULL; |
| 699 | } |
| 700 | |
| 701 | LOCAL(int) |
| 702 | checkpath(PyObject* tag) |
| 703 | { |
| 704 | int i, check = 1; |
| 705 | |
| 706 | /* check if a tag contains an xpath character */ |
| 707 | |
| 708 | #define PATHCHAR(ch) (ch == '/' || ch == '*' || ch == '[' || ch == '@') |
| 709 | |
| 710 | #if defined(Py_USING_UNICODE) |
| 711 | if (PyUnicode_Check(tag)) { |
| 712 | Py_UNICODE *p = PyUnicode_AS_UNICODE(tag); |
| 713 | for (i = 0; i < PyUnicode_GET_SIZE(tag); i++) { |
| 714 | if (p[i] == '{') |
| 715 | check = 0; |
| 716 | else if (p[i] == '}') |
| 717 | check = 1; |
| 718 | else if (check && PATHCHAR(p[i])) |
| 719 | return 1; |
| 720 | } |
| 721 | return 0; |
| 722 | } |
| 723 | #endif |
| 724 | if (PyString_Check(tag)) { |
| 725 | char *p = PyString_AS_STRING(tag); |
| 726 | for (i = 0; i < PyString_GET_SIZE(tag); i++) { |
| 727 | if (p[i] == '{') |
| 728 | check = 0; |
| 729 | else if (p[i] == '}') |
| 730 | check = 1; |
| 731 | else if (check && PATHCHAR(p[i])) |
| 732 | return 1; |
| 733 | } |
| 734 | return 0; |
| 735 | } |
| 736 | |
| 737 | return 1; /* unknown type; might be path expression */ |
| 738 | } |
| 739 | |
| 740 | static PyObject* |
| 741 | element_find(ElementObject* self, PyObject* args) |
| 742 | { |
| 743 | int i; |
| 744 | |
| 745 | PyObject* tag; |
| 746 | if (!PyArg_ParseTuple(args, "O:find", &tag)) |
| 747 | return NULL; |
| 748 | |
| 749 | if (checkpath(tag)) |
| 750 | return PyObject_CallMethod( |
| 751 | elementpath_obj, "find", "OO", self, tag |
| 752 | ); |
| 753 | |
| 754 | if (!self->extra) |
| 755 | Py_RETURN_NONE; |
| 756 | |
| 757 | for (i = 0; i < self->extra->length; i++) { |
| 758 | PyObject* item = self->extra->children[i]; |
| 759 | if (Element_CheckExact(item) && |
| 760 | PyObject_Compare(((ElementObject*)item)->tag, tag) == 0) { |
| 761 | Py_INCREF(item); |
| 762 | return item; |
| 763 | } |
| 764 | } |
| 765 | |
| 766 | Py_RETURN_NONE; |
| 767 | } |
| 768 | |
| 769 | static PyObject* |
| 770 | element_findtext(ElementObject* self, PyObject* args) |
| 771 | { |
| 772 | int i; |
| 773 | |
| 774 | PyObject* tag; |
| 775 | PyObject* default_value = Py_None; |
| 776 | if (!PyArg_ParseTuple(args, "O|O:findtext", &tag, &default_value)) |
| 777 | return NULL; |
| 778 | |
| 779 | if (checkpath(tag)) |
| 780 | return PyObject_CallMethod( |
| 781 | elementpath_obj, "findtext", "OOO", self, tag, default_value |
| 782 | ); |
| 783 | |
| 784 | if (!self->extra) { |
| 785 | Py_INCREF(default_value); |
| 786 | return default_value; |
| 787 | } |
| 788 | |
| 789 | for (i = 0; i < self->extra->length; i++) { |
| 790 | ElementObject* item = (ElementObject*) self->extra->children[i]; |
| 791 | if (Element_CheckExact(item) && !PyObject_Compare(item->tag, tag)) { |
| 792 | PyObject* text = element_get_text(item); |
| 793 | if (text == Py_None) |
| 794 | return PyString_FromString(""); |
| 795 | Py_INCREF(text); |
| 796 | return text; |
| 797 | } |
| 798 | } |
| 799 | |
| 800 | Py_INCREF(default_value); |
| 801 | return default_value; |
| 802 | } |
| 803 | |
| 804 | static PyObject* |
| 805 | element_findall(ElementObject* self, PyObject* args) |
| 806 | { |
| 807 | int i; |
| 808 | PyObject* out; |
| 809 | |
| 810 | PyObject* tag; |
| 811 | if (!PyArg_ParseTuple(args, "O:findall", &tag)) |
| 812 | return NULL; |
| 813 | |
| 814 | if (checkpath(tag)) |
| 815 | return PyObject_CallMethod( |
| 816 | elementpath_obj, "findall", "OO", self, tag |
| 817 | ); |
| 818 | |
| 819 | out = PyList_New(0); |
| 820 | if (!out) |
| 821 | return NULL; |
| 822 | |
| 823 | if (!self->extra) |
| 824 | return out; |
| 825 | |
| 826 | for (i = 0; i < self->extra->length; i++) { |
| 827 | PyObject* item = self->extra->children[i]; |
| 828 | if (Element_CheckExact(item) && |
| 829 | PyObject_Compare(((ElementObject*)item)->tag, tag) == 0) { |
| 830 | if (PyList_Append(out, item) < 0) { |
| 831 | Py_DECREF(out); |
| 832 | return NULL; |
| 833 | } |
| 834 | } |
| 835 | } |
| 836 | |
| 837 | return out; |
| 838 | } |
| 839 | |
| 840 | static PyObject* |
| 841 | element_get(ElementObject* self, PyObject* args) |
| 842 | { |
| 843 | PyObject* value; |
| 844 | |
| 845 | PyObject* key; |
| 846 | PyObject* default_value = Py_None; |
| 847 | if (!PyArg_ParseTuple(args, "O|O:get", &key, &default_value)) |
| 848 | return NULL; |
| 849 | |
| 850 | if (!self->extra || self->extra->attrib == Py_None) |
| 851 | value = default_value; |
| 852 | else { |
| 853 | value = PyDict_GetItem(self->extra->attrib, key); |
| 854 | if (!value) |
| 855 | value = default_value; |
| 856 | } |
| 857 | |
| 858 | Py_INCREF(value); |
| 859 | return value; |
| 860 | } |
| 861 | |
| 862 | static PyObject* |
| 863 | element_getchildren(ElementObject* self, PyObject* args) |
| 864 | { |
| 865 | int i; |
| 866 | PyObject* list; |
| 867 | |
| 868 | if (!PyArg_ParseTuple(args, ":getchildren")) |
| 869 | return NULL; |
| 870 | |
| 871 | if (!self->extra) |
| 872 | return PyList_New(0); |
| 873 | |
| 874 | list = PyList_New(self->extra->length); |
| 875 | if (!list) |
| 876 | return NULL; |
| 877 | |
| 878 | for (i = 0; i < self->extra->length; i++) { |
| 879 | PyObject* item = self->extra->children[i]; |
| 880 | Py_INCREF(item); |
| 881 | PyList_SET_ITEM(list, i, item); |
| 882 | } |
| 883 | |
| 884 | return list; |
| 885 | } |
| 886 | |
| 887 | static PyObject* |
| 888 | element_getiterator(ElementObject* self, PyObject* args) |
| 889 | { |
| 890 | PyObject* result; |
| 891 | |
| 892 | PyObject* tag = Py_None; |
| 893 | if (!PyArg_ParseTuple(args, "|O:getiterator", &tag)) |
| 894 | return NULL; |
| 895 | |
| 896 | if (!elementtree_getiterator_obj) { |
| 897 | PyErr_SetString( |
| 898 | PyExc_RuntimeError, |
| 899 | "getiterator helper not found" |
| 900 | ); |
| 901 | return NULL; |
| 902 | } |
| 903 | |
| 904 | args = PyTuple_New(2); |
Neal Norwitz | 02876df | 2006-02-07 06:58:52 +0000 | [diff] [blame] | 905 | if (args == NULL) |
| 906 | return NULL; |
| 907 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 908 | Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self); |
| 909 | Py_INCREF(tag); PyTuple_SET_ITEM(args, 1, (PyObject*) tag); |
| 910 | |
| 911 | result = PyObject_CallObject(elementtree_getiterator_obj, args); |
| 912 | |
| 913 | Py_DECREF(args); |
| 914 | |
| 915 | return result; |
| 916 | } |
| 917 | |
| 918 | static PyObject* |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 919 | element_getitem(PyObject* _self, Py_ssize_t index) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 920 | { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 921 | ElementObject* self = (ElementObject*)_self; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 922 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 923 | PyErr_SetString( |
| 924 | PyExc_IndexError, |
| 925 | "child index out of range" |
| 926 | ); |
| 927 | return NULL; |
| 928 | } |
| 929 | |
| 930 | Py_INCREF(self->extra->children[index]); |
| 931 | return self->extra->children[index]; |
| 932 | } |
| 933 | |
| 934 | static PyObject* |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 935 | element_getslice(PyObject* _self, Py_ssize_t start, Py_ssize_t end) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 936 | { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 937 | ElementObject* self = (ElementObject*)_self; |
| 938 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 939 | PyObject* list; |
| 940 | |
| 941 | if (!self->extra) |
| 942 | return PyList_New(0); |
| 943 | |
| 944 | /* standard clamping */ |
| 945 | if (start < 0) |
| 946 | start = 0; |
| 947 | if (end < 0) |
| 948 | end = 0; |
| 949 | if (end > self->extra->length) |
| 950 | end = self->extra->length; |
| 951 | if (start > end) |
| 952 | start = end; |
| 953 | |
| 954 | list = PyList_New(end - start); |
| 955 | if (!list) |
| 956 | return NULL; |
| 957 | |
| 958 | for (i = start; i < end; i++) { |
| 959 | PyObject* item = self->extra->children[i]; |
| 960 | Py_INCREF(item); |
| 961 | PyList_SET_ITEM(list, i - start, item); |
| 962 | } |
| 963 | |
| 964 | return list; |
| 965 | } |
| 966 | |
| 967 | static PyObject* |
| 968 | element_insert(ElementObject* self, PyObject* args) |
| 969 | { |
| 970 | int i; |
| 971 | |
| 972 | int index; |
| 973 | PyObject* element; |
| 974 | if (!PyArg_ParseTuple(args, "iO!:insert", &index, |
| 975 | &Element_Type, &element)) |
| 976 | return NULL; |
| 977 | |
| 978 | if (!self->extra) |
| 979 | element_new_extra(self, NULL); |
| 980 | |
| 981 | if (index < 0) |
| 982 | index = 0; |
| 983 | if (index > self->extra->length) |
| 984 | index = self->extra->length; |
| 985 | |
| 986 | if (element_resize(self, 1) < 0) |
| 987 | return NULL; |
| 988 | |
| 989 | for (i = self->extra->length; i > index; i--) |
| 990 | self->extra->children[i] = self->extra->children[i-1]; |
| 991 | |
| 992 | Py_INCREF(element); |
| 993 | self->extra->children[index] = element; |
| 994 | |
| 995 | self->extra->length++; |
| 996 | |
| 997 | Py_RETURN_NONE; |
| 998 | } |
| 999 | |
| 1000 | static PyObject* |
| 1001 | element_items(ElementObject* self, PyObject* args) |
| 1002 | { |
| 1003 | if (!PyArg_ParseTuple(args, ":items")) |
| 1004 | return NULL; |
| 1005 | |
| 1006 | if (!self->extra || self->extra->attrib == Py_None) |
| 1007 | return PyList_New(0); |
| 1008 | |
| 1009 | return PyDict_Items(self->extra->attrib); |
| 1010 | } |
| 1011 | |
| 1012 | static PyObject* |
| 1013 | element_keys(ElementObject* self, PyObject* args) |
| 1014 | { |
| 1015 | if (!PyArg_ParseTuple(args, ":keys")) |
| 1016 | return NULL; |
| 1017 | |
| 1018 | if (!self->extra || self->extra->attrib == Py_None) |
| 1019 | return PyList_New(0); |
| 1020 | |
| 1021 | return PyDict_Keys(self->extra->attrib); |
| 1022 | } |
| 1023 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1024 | static Py_ssize_t |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1025 | element_length(ElementObject* self) |
| 1026 | { |
| 1027 | if (!self->extra) |
| 1028 | return 0; |
| 1029 | |
| 1030 | return self->extra->length; |
| 1031 | } |
| 1032 | |
| 1033 | static PyObject* |
| 1034 | element_makeelement(PyObject* self, PyObject* args, PyObject* kw) |
| 1035 | { |
| 1036 | PyObject* elem; |
| 1037 | |
| 1038 | PyObject* tag; |
| 1039 | PyObject* attrib; |
| 1040 | if (!PyArg_ParseTuple(args, "OO:makeelement", &tag, &attrib)) |
| 1041 | return NULL; |
| 1042 | |
| 1043 | attrib = PyDict_Copy(attrib); |
| 1044 | if (!attrib) |
| 1045 | return NULL; |
| 1046 | |
| 1047 | elem = element_new(tag, attrib); |
| 1048 | |
| 1049 | Py_DECREF(attrib); |
| 1050 | |
| 1051 | return elem; |
| 1052 | } |
| 1053 | |
| 1054 | static PyObject* |
| 1055 | element_reduce(ElementObject* self, PyObject* args) |
| 1056 | { |
| 1057 | if (!PyArg_ParseTuple(args, ":__reduce__")) |
| 1058 | return NULL; |
| 1059 | |
| 1060 | /* Hack alert: This method is used to work around a __copy__ |
| 1061 | problem on certain 2.3 and 2.4 versions. To save time and |
| 1062 | simplify the code, we create the copy in here, and use a dummy |
| 1063 | copyelement helper to trick the copy module into doing the |
| 1064 | right thing. */ |
| 1065 | |
| 1066 | if (!elementtree_copyelement_obj) { |
| 1067 | PyErr_SetString( |
| 1068 | PyExc_RuntimeError, |
| 1069 | "copyelement helper not found" |
| 1070 | ); |
| 1071 | return NULL; |
| 1072 | } |
| 1073 | |
| 1074 | return Py_BuildValue( |
| 1075 | "O(N)", elementtree_copyelement_obj, element_copy(self, args) |
| 1076 | ); |
| 1077 | } |
| 1078 | |
| 1079 | static PyObject* |
| 1080 | element_remove(ElementObject* self, PyObject* args) |
| 1081 | { |
| 1082 | int i; |
| 1083 | |
| 1084 | PyObject* element; |
| 1085 | if (!PyArg_ParseTuple(args, "O!:remove", &Element_Type, &element)) |
| 1086 | return NULL; |
| 1087 | |
| 1088 | if (!self->extra) { |
| 1089 | /* element has no children, so raise exception */ |
| 1090 | PyErr_SetString( |
| 1091 | PyExc_ValueError, |
| 1092 | "list.remove(x): x not in list" |
| 1093 | ); |
| 1094 | return NULL; |
| 1095 | } |
| 1096 | |
| 1097 | for (i = 0; i < self->extra->length; i++) { |
| 1098 | if (self->extra->children[i] == element) |
| 1099 | break; |
| 1100 | if (PyObject_Compare(self->extra->children[i], element) == 0) |
| 1101 | break; |
| 1102 | } |
| 1103 | |
| 1104 | if (i == self->extra->length) { |
| 1105 | /* element is not in children, so raise exception */ |
| 1106 | PyErr_SetString( |
| 1107 | PyExc_ValueError, |
| 1108 | "list.remove(x): x not in list" |
| 1109 | ); |
| 1110 | return NULL; |
| 1111 | } |
| 1112 | |
| 1113 | Py_DECREF(self->extra->children[i]); |
| 1114 | |
| 1115 | self->extra->length--; |
| 1116 | |
| 1117 | for (; i < self->extra->length; i++) |
| 1118 | self->extra->children[i] = self->extra->children[i+1]; |
| 1119 | |
| 1120 | Py_RETURN_NONE; |
| 1121 | } |
| 1122 | |
| 1123 | static PyObject* |
| 1124 | element_repr(ElementObject* self) |
| 1125 | { |
| 1126 | PyObject* repr; |
| 1127 | char buffer[100]; |
| 1128 | |
| 1129 | repr = PyString_FromString("<Element "); |
| 1130 | |
| 1131 | PyString_ConcatAndDel(&repr, PyObject_Repr(self->tag)); |
| 1132 | |
| 1133 | sprintf(buffer, " at %p>", self); |
| 1134 | PyString_ConcatAndDel(&repr, PyString_FromString(buffer)); |
| 1135 | |
| 1136 | return repr; |
| 1137 | } |
| 1138 | |
| 1139 | static PyObject* |
| 1140 | element_set(ElementObject* self, PyObject* args) |
| 1141 | { |
| 1142 | PyObject* attrib; |
| 1143 | |
| 1144 | PyObject* key; |
| 1145 | PyObject* value; |
| 1146 | if (!PyArg_ParseTuple(args, "OO:set", &key, &value)) |
| 1147 | return NULL; |
| 1148 | |
| 1149 | if (!self->extra) |
| 1150 | element_new_extra(self, NULL); |
| 1151 | |
| 1152 | attrib = element_get_attrib(self); |
| 1153 | if (!attrib) |
| 1154 | return NULL; |
| 1155 | |
| 1156 | if (PyDict_SetItem(attrib, key, value) < 0) |
| 1157 | return NULL; |
| 1158 | |
| 1159 | Py_RETURN_NONE; |
| 1160 | } |
| 1161 | |
| 1162 | static int |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1163 | element_setslice(PyObject* _self, Py_ssize_t start, Py_ssize_t end, PyObject* item) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1164 | { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1165 | ElementObject* self = (ElementObject*)_self; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1166 | int i, new, old; |
| 1167 | PyObject* recycle = NULL; |
| 1168 | |
| 1169 | if (!self->extra) |
| 1170 | element_new_extra(self, NULL); |
| 1171 | |
| 1172 | /* standard clamping */ |
| 1173 | if (start < 0) |
| 1174 | start = 0; |
| 1175 | if (end < 0) |
| 1176 | end = 0; |
| 1177 | if (end > self->extra->length) |
| 1178 | end = self->extra->length; |
| 1179 | if (start > end) |
| 1180 | start = end; |
| 1181 | |
| 1182 | old = end - start; |
| 1183 | |
| 1184 | if (item == NULL) |
| 1185 | new = 0; |
| 1186 | else if (PyList_CheckExact(item)) { |
| 1187 | new = PyList_GET_SIZE(item); |
| 1188 | } else { |
| 1189 | /* FIXME: support arbitrary sequences? */ |
| 1190 | PyErr_Format( |
| 1191 | PyExc_TypeError, |
| 1192 | "expected list, not \"%.200s\"", item->ob_type->tp_name |
| 1193 | ); |
| 1194 | return -1; |
| 1195 | } |
| 1196 | |
| 1197 | if (old > 0) { |
| 1198 | /* to avoid recursive calls to this method (via decref), move |
| 1199 | old items to the recycle bin here, and get rid of them when |
| 1200 | we're done modifying the element */ |
| 1201 | recycle = PyList_New(old); |
| 1202 | for (i = 0; i < old; i++) |
| 1203 | PyList_SET_ITEM(recycle, i, self->extra->children[i + start]); |
| 1204 | } |
| 1205 | |
| 1206 | if (new < old) { |
| 1207 | /* delete slice */ |
| 1208 | for (i = end; i < self->extra->length; i++) |
| 1209 | self->extra->children[i + new - old] = self->extra->children[i]; |
| 1210 | } else if (new > old) { |
| 1211 | /* insert slice */ |
| 1212 | if (element_resize(self, new - old) < 0) |
| 1213 | return -1; |
| 1214 | for (i = self->extra->length-1; i >= end; i--) |
| 1215 | self->extra->children[i + new - old] = self->extra->children[i]; |
| 1216 | } |
| 1217 | |
| 1218 | /* replace the slice */ |
| 1219 | for (i = 0; i < new; i++) { |
| 1220 | PyObject* element = PyList_GET_ITEM(item, i); |
| 1221 | Py_INCREF(element); |
| 1222 | self->extra->children[i + start] = element; |
| 1223 | } |
| 1224 | |
| 1225 | self->extra->length += new - old; |
| 1226 | |
| 1227 | /* discard the recycle bin, and everything in it */ |
| 1228 | Py_XDECREF(recycle); |
| 1229 | |
| 1230 | return 0; |
| 1231 | } |
| 1232 | |
| 1233 | static int |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1234 | element_setitem(PyObject* _self, Py_ssize_t index, PyObject* item) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1235 | { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1236 | ElementObject* self = (ElementObject*)_self; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1237 | int i; |
| 1238 | PyObject* old; |
| 1239 | |
| 1240 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1241 | PyErr_SetString( |
| 1242 | PyExc_IndexError, |
| 1243 | "child assignment index out of range"); |
| 1244 | return -1; |
| 1245 | } |
| 1246 | |
| 1247 | old = self->extra->children[index]; |
| 1248 | |
| 1249 | if (item) { |
| 1250 | Py_INCREF(item); |
| 1251 | self->extra->children[index] = item; |
| 1252 | } else { |
| 1253 | self->extra->length--; |
| 1254 | for (i = index; i < self->extra->length; i++) |
| 1255 | self->extra->children[i] = self->extra->children[i+1]; |
| 1256 | } |
| 1257 | |
| 1258 | Py_DECREF(old); |
| 1259 | |
| 1260 | return 0; |
| 1261 | } |
| 1262 | |
| 1263 | static PyMethodDef element_methods[] = { |
| 1264 | |
| 1265 | {"clear", (PyCFunction) element_clear, METH_VARARGS}, |
| 1266 | |
| 1267 | {"get", (PyCFunction) element_get, METH_VARARGS}, |
| 1268 | {"set", (PyCFunction) element_set, METH_VARARGS}, |
| 1269 | |
| 1270 | {"find", (PyCFunction) element_find, METH_VARARGS}, |
| 1271 | {"findtext", (PyCFunction) element_findtext, METH_VARARGS}, |
| 1272 | {"findall", (PyCFunction) element_findall, METH_VARARGS}, |
| 1273 | |
| 1274 | {"append", (PyCFunction) element_append, METH_VARARGS}, |
| 1275 | {"insert", (PyCFunction) element_insert, METH_VARARGS}, |
| 1276 | {"remove", (PyCFunction) element_remove, METH_VARARGS}, |
| 1277 | |
| 1278 | {"getiterator", (PyCFunction) element_getiterator, METH_VARARGS}, |
| 1279 | {"getchildren", (PyCFunction) element_getchildren, METH_VARARGS}, |
| 1280 | |
| 1281 | {"items", (PyCFunction) element_items, METH_VARARGS}, |
| 1282 | {"keys", (PyCFunction) element_keys, METH_VARARGS}, |
| 1283 | |
| 1284 | {"makeelement", (PyCFunction) element_makeelement, METH_VARARGS}, |
| 1285 | |
| 1286 | {"__copy__", (PyCFunction) element_copy, METH_VARARGS}, |
| 1287 | {"__deepcopy__", (PyCFunction) element_deepcopy, METH_VARARGS}, |
| 1288 | |
| 1289 | /* Some 2.3 and 2.4 versions do not handle the __copy__ method on |
| 1290 | C objects correctly, so we have to fake it using a __reduce__- |
| 1291 | based hack (see the element_reduce implementation above for |
| 1292 | details). */ |
| 1293 | |
| 1294 | /* The behaviour has been changed in 2.3.5 and 2.4.1, so we're |
| 1295 | using a runtime test to figure out if we need to fake things |
| 1296 | or now (see the init code below). The following entry is |
| 1297 | enabled only if the hack is needed. */ |
| 1298 | |
| 1299 | {"!__reduce__", (PyCFunction) element_reduce, METH_VARARGS}, |
| 1300 | |
| 1301 | {NULL, NULL} |
| 1302 | }; |
| 1303 | |
| 1304 | static PyObject* |
| 1305 | element_getattr(ElementObject* self, char* name) |
| 1306 | { |
| 1307 | PyObject* res; |
| 1308 | |
| 1309 | res = Py_FindMethod(element_methods, (PyObject*) self, name); |
| 1310 | if (res) |
| 1311 | return res; |
| 1312 | |
| 1313 | PyErr_Clear(); |
| 1314 | |
| 1315 | if (strcmp(name, "tag") == 0) |
| 1316 | res = self->tag; |
| 1317 | else if (strcmp(name, "text") == 0) |
| 1318 | res = element_get_text(self); |
| 1319 | else if (strcmp(name, "tail") == 0) { |
| 1320 | res = element_get_tail(self); |
| 1321 | } else if (strcmp(name, "attrib") == 0) { |
| 1322 | if (!self->extra) |
| 1323 | element_new_extra(self, NULL); |
| 1324 | res = element_get_attrib(self); |
| 1325 | } else { |
| 1326 | PyErr_SetString(PyExc_AttributeError, name); |
| 1327 | return NULL; |
| 1328 | } |
| 1329 | |
| 1330 | if (!res) |
| 1331 | return NULL; |
| 1332 | |
| 1333 | Py_INCREF(res); |
| 1334 | return res; |
| 1335 | } |
| 1336 | |
| 1337 | static int |
| 1338 | element_setattr(ElementObject* self, const char* name, PyObject* value) |
| 1339 | { |
| 1340 | if (value == NULL) { |
| 1341 | PyErr_SetString( |
| 1342 | PyExc_AttributeError, |
| 1343 | "can't delete element attributes" |
| 1344 | ); |
| 1345 | return -1; |
| 1346 | } |
| 1347 | |
| 1348 | if (strcmp(name, "tag") == 0) { |
| 1349 | Py_DECREF(self->tag); |
| 1350 | self->tag = value; |
| 1351 | Py_INCREF(self->tag); |
| 1352 | } else if (strcmp(name, "text") == 0) { |
| 1353 | Py_DECREF(JOIN_OBJ(self->text)); |
| 1354 | self->text = value; |
| 1355 | Py_INCREF(self->text); |
| 1356 | } else if (strcmp(name, "tail") == 0) { |
| 1357 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 1358 | self->tail = value; |
| 1359 | Py_INCREF(self->tail); |
| 1360 | } else if (strcmp(name, "attrib") == 0) { |
| 1361 | if (!self->extra) |
| 1362 | element_new_extra(self, NULL); |
| 1363 | Py_DECREF(self->extra->attrib); |
| 1364 | self->extra->attrib = value; |
| 1365 | Py_INCREF(self->extra->attrib); |
| 1366 | } else { |
| 1367 | PyErr_SetString(PyExc_AttributeError, name); |
| 1368 | return -1; |
| 1369 | } |
| 1370 | |
| 1371 | return 0; |
| 1372 | } |
| 1373 | |
| 1374 | static PySequenceMethods element_as_sequence = { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1375 | (lenfunc) element_length, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1376 | 0, /* sq_concat */ |
| 1377 | 0, /* sq_repeat */ |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1378 | element_getitem, |
| 1379 | element_getslice, |
| 1380 | element_setitem, |
| 1381 | element_setslice, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1382 | }; |
| 1383 | |
| 1384 | statichere PyTypeObject Element_Type = { |
| 1385 | PyObject_HEAD_INIT(NULL) |
| 1386 | 0, "Element", sizeof(ElementObject), 0, |
| 1387 | /* methods */ |
| 1388 | (destructor)element_dealloc, /* tp_dealloc */ |
| 1389 | 0, /* tp_print */ |
| 1390 | (getattrfunc)element_getattr, /* tp_getattr */ |
| 1391 | (setattrfunc)element_setattr, /* tp_setattr */ |
| 1392 | 0, /* tp_compare */ |
| 1393 | (reprfunc)element_repr, /* tp_repr */ |
| 1394 | 0, /* tp_as_number */ |
| 1395 | &element_as_sequence, /* tp_as_sequence */ |
| 1396 | }; |
| 1397 | |
| 1398 | /* ==================================================================== */ |
| 1399 | /* the tree builder type */ |
| 1400 | |
| 1401 | typedef struct { |
| 1402 | PyObject_HEAD |
| 1403 | |
| 1404 | PyObject* root; /* root node (first created node) */ |
| 1405 | |
| 1406 | ElementObject* this; /* current node */ |
| 1407 | ElementObject* last; /* most recently created node */ |
| 1408 | |
| 1409 | PyObject* data; /* data collector (string or list), or NULL */ |
| 1410 | |
| 1411 | PyObject* stack; /* element stack */ |
| 1412 | int index; /* current stack size (0=empty) */ |
| 1413 | |
| 1414 | /* element tracing */ |
| 1415 | PyObject* events; /* list of events, or NULL if not collecting */ |
| 1416 | PyObject* start_event_obj; /* event objects (NULL to ignore) */ |
| 1417 | PyObject* end_event_obj; |
| 1418 | PyObject* start_ns_event_obj; |
| 1419 | PyObject* end_ns_event_obj; |
| 1420 | |
| 1421 | } TreeBuilderObject; |
| 1422 | |
| 1423 | staticforward PyTypeObject TreeBuilder_Type; |
| 1424 | |
| 1425 | #define TreeBuilder_CheckExact(op) ((op)->ob_type == &TreeBuilder_Type) |
| 1426 | |
| 1427 | /* -------------------------------------------------------------------- */ |
| 1428 | /* constructor and destructor */ |
| 1429 | |
| 1430 | LOCAL(PyObject*) |
| 1431 | treebuilder_new(void) |
| 1432 | { |
| 1433 | TreeBuilderObject* self; |
| 1434 | |
| 1435 | self = PyObject_New(TreeBuilderObject, &TreeBuilder_Type); |
| 1436 | if (self == NULL) |
| 1437 | return NULL; |
| 1438 | |
| 1439 | self->root = NULL; |
| 1440 | |
| 1441 | Py_INCREF(Py_None); |
| 1442 | self->this = (ElementObject*) Py_None; |
| 1443 | |
| 1444 | Py_INCREF(Py_None); |
| 1445 | self->last = (ElementObject*) Py_None; |
| 1446 | |
| 1447 | self->data = NULL; |
| 1448 | |
| 1449 | self->stack = PyList_New(20); |
| 1450 | self->index = 0; |
| 1451 | |
| 1452 | self->events = NULL; |
| 1453 | self->start_event_obj = self->end_event_obj = NULL; |
| 1454 | self->start_ns_event_obj = self->end_ns_event_obj = NULL; |
| 1455 | |
| 1456 | ALLOC(sizeof(TreeBuilderObject), "create treebuilder"); |
| 1457 | |
| 1458 | return (PyObject*) self; |
| 1459 | } |
| 1460 | |
| 1461 | static PyObject* |
| 1462 | treebuilder(PyObject* _self, PyObject* args) |
| 1463 | { |
| 1464 | if (!PyArg_ParseTuple(args, ":TreeBuilder")) |
| 1465 | return NULL; |
| 1466 | |
| 1467 | return treebuilder_new(); |
| 1468 | } |
| 1469 | |
| 1470 | static void |
| 1471 | treebuilder_dealloc(TreeBuilderObject* self) |
| 1472 | { |
| 1473 | Py_XDECREF(self->end_ns_event_obj); |
| 1474 | Py_XDECREF(self->start_ns_event_obj); |
| 1475 | Py_XDECREF(self->end_event_obj); |
| 1476 | Py_XDECREF(self->start_event_obj); |
| 1477 | Py_XDECREF(self->events); |
| 1478 | Py_DECREF(self->stack); |
| 1479 | Py_XDECREF(self->data); |
| 1480 | Py_DECREF(self->last); |
| 1481 | Py_DECREF(self->this); |
| 1482 | Py_XDECREF(self->root); |
| 1483 | |
| 1484 | RELEASE(sizeof(TreeBuilderObject), "destroy treebuilder"); |
| 1485 | |
| 1486 | PyObject_Del(self); |
| 1487 | } |
| 1488 | |
| 1489 | /* -------------------------------------------------------------------- */ |
| 1490 | /* handlers */ |
| 1491 | |
| 1492 | LOCAL(PyObject*) |
| 1493 | treebuilder_handle_xml(TreeBuilderObject* self, PyObject* encoding, |
| 1494 | PyObject* standalone) |
| 1495 | { |
| 1496 | Py_RETURN_NONE; |
| 1497 | } |
| 1498 | |
| 1499 | LOCAL(PyObject*) |
| 1500 | treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag, |
| 1501 | PyObject* attrib) |
| 1502 | { |
| 1503 | PyObject* node; |
| 1504 | PyObject* this; |
| 1505 | |
| 1506 | if (self->data) { |
| 1507 | if (self->this == self->last) { |
Fredrik Lundh | 0149e3a | 2005-12-18 13:58:25 +0000 | [diff] [blame] | 1508 | Py_DECREF(self->last->text); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1509 | self->last->text = JOIN_SET( |
| 1510 | self->data, PyList_CheckExact(self->data) |
| 1511 | ); |
| 1512 | } else { |
Fredrik Lundh | 0149e3a | 2005-12-18 13:58:25 +0000 | [diff] [blame] | 1513 | Py_DECREF(self->last->tail); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1514 | self->last->tail = JOIN_SET( |
| 1515 | self->data, PyList_CheckExact(self->data) |
| 1516 | ); |
| 1517 | } |
| 1518 | self->data = NULL; |
| 1519 | } |
| 1520 | |
| 1521 | node = element_new(tag, attrib); |
| 1522 | if (!node) |
| 1523 | return NULL; |
| 1524 | |
| 1525 | this = (PyObject*) self->this; |
| 1526 | |
| 1527 | if (this != Py_None) { |
| 1528 | if (element_add_subelement((ElementObject*) this, node) < 0) |
| 1529 | return NULL; |
| 1530 | } else { |
| 1531 | if (self->root) { |
| 1532 | PyErr_SetString( |
| 1533 | PyExc_SyntaxError, |
| 1534 | "multiple elements on top level" |
| 1535 | ); |
| 1536 | return NULL; |
| 1537 | } |
| 1538 | Py_INCREF(node); |
| 1539 | self->root = node; |
| 1540 | } |
| 1541 | |
| 1542 | if (self->index < PyList_GET_SIZE(self->stack)) { |
| 1543 | if (PyList_SetItem(self->stack, self->index, this) < 0) |
| 1544 | return NULL; |
| 1545 | Py_INCREF(this); |
| 1546 | } else { |
| 1547 | if (PyList_Append(self->stack, this) < 0) |
| 1548 | return NULL; |
| 1549 | } |
| 1550 | self->index++; |
| 1551 | |
| 1552 | Py_DECREF(this); |
| 1553 | Py_INCREF(node); |
| 1554 | self->this = (ElementObject*) node; |
| 1555 | |
| 1556 | Py_DECREF(self->last); |
| 1557 | Py_INCREF(node); |
| 1558 | self->last = (ElementObject*) node; |
| 1559 | |
| 1560 | if (self->start_event_obj) { |
| 1561 | PyObject* res; |
| 1562 | PyObject* action = self->start_event_obj; |
| 1563 | res = PyTuple_New(2); |
| 1564 | if (res) { |
| 1565 | Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action); |
| 1566 | Py_INCREF(node); PyTuple_SET_ITEM(res, 1, (PyObject*) node); |
| 1567 | PyList_Append(self->events, res); |
| 1568 | Py_DECREF(res); |
| 1569 | } else |
| 1570 | PyErr_Clear(); /* FIXME: propagate error */ |
| 1571 | } |
| 1572 | |
| 1573 | return node; |
| 1574 | } |
| 1575 | |
| 1576 | LOCAL(PyObject*) |
| 1577 | treebuilder_handle_data(TreeBuilderObject* self, PyObject* data) |
| 1578 | { |
| 1579 | if (!self->data) { |
| 1580 | /* store the first item as is */ |
| 1581 | Py_INCREF(data); self->data = data; |
| 1582 | } else { |
| 1583 | /* more than one item; use a list to collect items */ |
| 1584 | if (PyString_CheckExact(self->data) && self->data->ob_refcnt == 1 && |
| 1585 | PyString_CheckExact(data) && PyString_GET_SIZE(data) == 1) { |
| 1586 | /* expat often generates single character data sections; handle |
| 1587 | the most common case by resizing the existing string... */ |
| 1588 | int size = PyString_GET_SIZE(self->data); |
| 1589 | if (_PyString_Resize(&self->data, size + 1) < 0) |
| 1590 | return NULL; |
| 1591 | PyString_AS_STRING(self->data)[size] = PyString_AS_STRING(data)[0]; |
| 1592 | } else if (PyList_CheckExact(self->data)) { |
| 1593 | if (PyList_Append(self->data, data) < 0) |
| 1594 | return NULL; |
| 1595 | } else { |
| 1596 | PyObject* list = PyList_New(2); |
| 1597 | if (!list) |
| 1598 | return NULL; |
| 1599 | PyList_SET_ITEM(list, 0, self->data); |
| 1600 | Py_INCREF(data); PyList_SET_ITEM(list, 1, data); |
| 1601 | self->data = list; |
| 1602 | } |
| 1603 | } |
| 1604 | |
| 1605 | Py_RETURN_NONE; |
| 1606 | } |
| 1607 | |
| 1608 | LOCAL(PyObject*) |
| 1609 | treebuilder_handle_end(TreeBuilderObject* self, PyObject* tag) |
| 1610 | { |
| 1611 | PyObject* item; |
| 1612 | |
| 1613 | if (self->data) { |
| 1614 | if (self->this == self->last) { |
Fredrik Lundh | 0149e3a | 2005-12-18 13:58:25 +0000 | [diff] [blame] | 1615 | Py_DECREF(self->last->text); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1616 | self->last->text = JOIN_SET( |
| 1617 | self->data, PyList_CheckExact(self->data) |
| 1618 | ); |
| 1619 | } else { |
Fredrik Lundh | 0149e3a | 2005-12-18 13:58:25 +0000 | [diff] [blame] | 1620 | Py_DECREF(self->last->tail); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1621 | self->last->tail = JOIN_SET( |
| 1622 | self->data, PyList_CheckExact(self->data) |
| 1623 | ); |
| 1624 | } |
| 1625 | self->data = NULL; |
| 1626 | } |
| 1627 | |
| 1628 | if (self->index == 0) { |
| 1629 | PyErr_SetString( |
| 1630 | PyExc_IndexError, |
| 1631 | "pop from empty stack" |
| 1632 | ); |
| 1633 | return NULL; |
| 1634 | } |
| 1635 | |
| 1636 | self->index--; |
| 1637 | |
| 1638 | item = PyList_GET_ITEM(self->stack, self->index); |
| 1639 | Py_INCREF(item); |
| 1640 | |
| 1641 | Py_DECREF(self->last); |
| 1642 | |
| 1643 | self->last = (ElementObject*) self->this; |
| 1644 | self->this = (ElementObject*) item; |
| 1645 | |
| 1646 | if (self->end_event_obj) { |
| 1647 | PyObject* res; |
| 1648 | PyObject* action = self->end_event_obj; |
| 1649 | PyObject* node = (PyObject*) self->last; |
| 1650 | res = PyTuple_New(2); |
| 1651 | if (res) { |
| 1652 | Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action); |
| 1653 | Py_INCREF(node); PyTuple_SET_ITEM(res, 1, (PyObject*) node); |
| 1654 | PyList_Append(self->events, res); |
| 1655 | Py_DECREF(res); |
| 1656 | } else |
| 1657 | PyErr_Clear(); /* FIXME: propagate error */ |
| 1658 | } |
| 1659 | |
| 1660 | Py_INCREF(self->last); |
| 1661 | return (PyObject*) self->last; |
| 1662 | } |
| 1663 | |
| 1664 | LOCAL(void) |
| 1665 | treebuilder_handle_namespace(TreeBuilderObject* self, int start, |
| 1666 | const char* prefix, const char *uri) |
| 1667 | { |
| 1668 | PyObject* res; |
| 1669 | PyObject* action; |
| 1670 | PyObject* parcel; |
| 1671 | |
| 1672 | if (!self->events) |
| 1673 | return; |
| 1674 | |
| 1675 | if (start) { |
| 1676 | if (!self->start_ns_event_obj) |
| 1677 | return; |
| 1678 | action = self->start_ns_event_obj; |
| 1679 | /* FIXME: prefix and uri use utf-8 encoding! */ |
| 1680 | parcel = Py_BuildValue("ss", (prefix) ? prefix : "", uri); |
| 1681 | if (!parcel) |
| 1682 | return; |
| 1683 | Py_INCREF(action); |
| 1684 | } else { |
| 1685 | if (!self->end_ns_event_obj) |
| 1686 | return; |
| 1687 | action = self->end_ns_event_obj; |
| 1688 | Py_INCREF(action); |
| 1689 | parcel = Py_None; |
| 1690 | Py_INCREF(parcel); |
| 1691 | } |
| 1692 | |
| 1693 | res = PyTuple_New(2); |
| 1694 | |
| 1695 | if (res) { |
| 1696 | PyTuple_SET_ITEM(res, 0, action); |
| 1697 | PyTuple_SET_ITEM(res, 1, parcel); |
| 1698 | PyList_Append(self->events, res); |
| 1699 | Py_DECREF(res); |
| 1700 | } else |
| 1701 | PyErr_Clear(); /* FIXME: propagate error */ |
| 1702 | } |
| 1703 | |
| 1704 | /* -------------------------------------------------------------------- */ |
| 1705 | /* methods (in alphabetical order) */ |
| 1706 | |
| 1707 | static PyObject* |
| 1708 | treebuilder_data(TreeBuilderObject* self, PyObject* args) |
| 1709 | { |
| 1710 | PyObject* data; |
| 1711 | if (!PyArg_ParseTuple(args, "O:data", &data)) |
| 1712 | return NULL; |
| 1713 | |
| 1714 | return treebuilder_handle_data(self, data); |
| 1715 | } |
| 1716 | |
| 1717 | static PyObject* |
| 1718 | treebuilder_end(TreeBuilderObject* self, PyObject* args) |
| 1719 | { |
| 1720 | PyObject* tag; |
| 1721 | if (!PyArg_ParseTuple(args, "O:end", &tag)) |
| 1722 | return NULL; |
| 1723 | |
| 1724 | return treebuilder_handle_end(self, tag); |
| 1725 | } |
| 1726 | |
| 1727 | LOCAL(PyObject*) |
| 1728 | treebuilder_done(TreeBuilderObject* self) |
| 1729 | { |
| 1730 | PyObject* res; |
| 1731 | |
| 1732 | /* FIXME: check stack size? */ |
| 1733 | |
| 1734 | if (self->root) |
| 1735 | res = self->root; |
| 1736 | else |
| 1737 | res = Py_None; |
| 1738 | |
| 1739 | Py_INCREF(res); |
| 1740 | return res; |
| 1741 | } |
| 1742 | |
| 1743 | static PyObject* |
| 1744 | treebuilder_close(TreeBuilderObject* self, PyObject* args) |
| 1745 | { |
| 1746 | if (!PyArg_ParseTuple(args, ":close")) |
| 1747 | return NULL; |
| 1748 | |
| 1749 | return treebuilder_done(self); |
| 1750 | } |
| 1751 | |
| 1752 | static PyObject* |
| 1753 | treebuilder_start(TreeBuilderObject* self, PyObject* args) |
| 1754 | { |
| 1755 | PyObject* tag; |
| 1756 | PyObject* attrib = Py_None; |
| 1757 | if (!PyArg_ParseTuple(args, "O|O:start", &tag, &attrib)) |
| 1758 | return NULL; |
| 1759 | |
| 1760 | return treebuilder_handle_start(self, tag, attrib); |
| 1761 | } |
| 1762 | |
| 1763 | static PyObject* |
| 1764 | treebuilder_xml(TreeBuilderObject* self, PyObject* args) |
| 1765 | { |
| 1766 | PyObject* encoding; |
| 1767 | PyObject* standalone; |
| 1768 | if (!PyArg_ParseTuple(args, "OO:xml", &encoding, &standalone)) |
| 1769 | return NULL; |
| 1770 | |
| 1771 | return treebuilder_handle_xml(self, encoding, standalone); |
| 1772 | } |
| 1773 | |
| 1774 | static PyMethodDef treebuilder_methods[] = { |
| 1775 | {"data", (PyCFunction) treebuilder_data, METH_VARARGS}, |
| 1776 | {"start", (PyCFunction) treebuilder_start, METH_VARARGS}, |
| 1777 | {"end", (PyCFunction) treebuilder_end, METH_VARARGS}, |
| 1778 | {"xml", (PyCFunction) treebuilder_xml, METH_VARARGS}, |
| 1779 | {"close", (PyCFunction) treebuilder_close, METH_VARARGS}, |
| 1780 | {NULL, NULL} |
| 1781 | }; |
| 1782 | |
| 1783 | static PyObject* |
| 1784 | treebuilder_getattr(TreeBuilderObject* self, char* name) |
| 1785 | { |
| 1786 | return Py_FindMethod(treebuilder_methods, (PyObject*) self, name); |
| 1787 | } |
| 1788 | |
| 1789 | statichere PyTypeObject TreeBuilder_Type = { |
| 1790 | PyObject_HEAD_INIT(NULL) |
| 1791 | 0, "TreeBuilder", sizeof(TreeBuilderObject), 0, |
| 1792 | /* methods */ |
| 1793 | (destructor)treebuilder_dealloc, /* tp_dealloc */ |
| 1794 | 0, /* tp_print */ |
| 1795 | (getattrfunc)treebuilder_getattr, /* tp_getattr */ |
| 1796 | }; |
| 1797 | |
| 1798 | /* ==================================================================== */ |
| 1799 | /* the expat interface */ |
| 1800 | |
| 1801 | #if defined(USE_EXPAT) |
| 1802 | |
| 1803 | #include "expat.h" |
| 1804 | |
| 1805 | #if defined(USE_PYEXPAT_CAPI) |
| 1806 | #include "pyexpat.h" |
| 1807 | static struct PyExpat_CAPI* expat_capi; |
| 1808 | #define EXPAT(func) (expat_capi->func) |
| 1809 | #else |
| 1810 | #define EXPAT(func) (XML_##func) |
| 1811 | #endif |
| 1812 | |
| 1813 | typedef struct { |
| 1814 | PyObject_HEAD |
| 1815 | |
| 1816 | XML_Parser parser; |
| 1817 | |
| 1818 | PyObject* target; |
| 1819 | PyObject* entity; |
| 1820 | |
| 1821 | PyObject* names; |
| 1822 | |
| 1823 | PyObject* handle_xml; |
| 1824 | PyObject* handle_start; |
| 1825 | PyObject* handle_data; |
| 1826 | PyObject* handle_end; |
| 1827 | |
| 1828 | PyObject* handle_comment; |
| 1829 | PyObject* handle_pi; |
| 1830 | |
| 1831 | } XMLParserObject; |
| 1832 | |
| 1833 | staticforward PyTypeObject XMLParser_Type; |
| 1834 | |
| 1835 | /* helpers */ |
| 1836 | |
| 1837 | #if defined(Py_USING_UNICODE) |
| 1838 | LOCAL(int) |
| 1839 | checkstring(const char* string, int size) |
| 1840 | { |
| 1841 | int i; |
| 1842 | |
| 1843 | /* check if an 8-bit string contains UTF-8 characters */ |
| 1844 | for (i = 0; i < size; i++) |
| 1845 | if (string[i] & 0x80) |
| 1846 | return 1; |
| 1847 | |
| 1848 | return 0; |
| 1849 | } |
| 1850 | #endif |
| 1851 | |
| 1852 | LOCAL(PyObject*) |
| 1853 | makestring(const char* string, int size) |
| 1854 | { |
| 1855 | /* convert a UTF-8 string to either a 7-bit ascii string or a |
| 1856 | Unicode string */ |
| 1857 | |
| 1858 | #if defined(Py_USING_UNICODE) |
| 1859 | if (checkstring(string, size)) |
| 1860 | return PyUnicode_DecodeUTF8(string, size, "strict"); |
| 1861 | #endif |
| 1862 | |
| 1863 | return PyString_FromStringAndSize(string, size); |
| 1864 | } |
| 1865 | |
| 1866 | LOCAL(PyObject*) |
| 1867 | makeuniversal(XMLParserObject* self, const char* string) |
| 1868 | { |
| 1869 | /* convert a UTF-8 tag/attribute name from the expat parser |
| 1870 | to a universal name string */ |
| 1871 | |
| 1872 | int size = strlen(string); |
| 1873 | PyObject* key; |
| 1874 | PyObject* value; |
| 1875 | |
| 1876 | /* look the 'raw' name up in the names dictionary */ |
| 1877 | key = PyString_FromStringAndSize(string, size); |
| 1878 | if (!key) |
| 1879 | return NULL; |
| 1880 | |
| 1881 | value = PyDict_GetItem(self->names, key); |
| 1882 | |
| 1883 | if (value) { |
| 1884 | Py_INCREF(value); |
| 1885 | } else { |
| 1886 | /* new name. convert to universal name, and decode as |
| 1887 | necessary */ |
| 1888 | |
| 1889 | PyObject* tag; |
| 1890 | char* p; |
| 1891 | int i; |
| 1892 | |
| 1893 | /* look for namespace separator */ |
| 1894 | for (i = 0; i < size; i++) |
| 1895 | if (string[i] == '}') |
| 1896 | break; |
| 1897 | if (i != size) { |
| 1898 | /* convert to universal name */ |
| 1899 | tag = PyString_FromStringAndSize(NULL, size+1); |
| 1900 | p = PyString_AS_STRING(tag); |
| 1901 | p[0] = '{'; |
| 1902 | memcpy(p+1, string, size); |
| 1903 | size++; |
| 1904 | } else { |
| 1905 | /* plain name; use key as tag */ |
| 1906 | Py_INCREF(key); |
| 1907 | tag = key; |
| 1908 | } |
| 1909 | |
| 1910 | /* decode universal name */ |
| 1911 | #if defined(Py_USING_UNICODE) |
| 1912 | /* inline makestring, to avoid duplicating the source string if |
| 1913 | it's not an utf-8 string */ |
| 1914 | p = PyString_AS_STRING(tag); |
| 1915 | if (checkstring(p, size)) { |
| 1916 | value = PyUnicode_DecodeUTF8(p, size, "strict"); |
| 1917 | Py_DECREF(tag); |
| 1918 | if (!value) { |
| 1919 | Py_DECREF(key); |
| 1920 | return NULL; |
| 1921 | } |
| 1922 | } else |
| 1923 | #endif |
| 1924 | value = tag; /* use tag as is */ |
| 1925 | |
| 1926 | /* add to names dictionary */ |
| 1927 | if (PyDict_SetItem(self->names, key, value) < 0) { |
| 1928 | Py_DECREF(key); |
| 1929 | Py_DECREF(value); |
| 1930 | return NULL; |
| 1931 | } |
| 1932 | } |
| 1933 | |
| 1934 | Py_DECREF(key); |
| 1935 | return value; |
| 1936 | } |
| 1937 | |
| 1938 | /* -------------------------------------------------------------------- */ |
| 1939 | /* handlers */ |
| 1940 | |
| 1941 | static void |
| 1942 | expat_default_handler(XMLParserObject* self, const XML_Char* data_in, |
| 1943 | int data_len) |
| 1944 | { |
| 1945 | PyObject* key; |
| 1946 | PyObject* value; |
| 1947 | PyObject* res; |
| 1948 | |
| 1949 | if (data_len < 2 || data_in[0] != '&') |
| 1950 | return; |
| 1951 | |
| 1952 | key = makestring(data_in + 1, data_len - 2); |
| 1953 | if (!key) |
| 1954 | return; |
| 1955 | |
| 1956 | value = PyDict_GetItem(self->entity, key); |
| 1957 | |
| 1958 | if (value) { |
| 1959 | if (TreeBuilder_CheckExact(self->target)) |
| 1960 | res = treebuilder_handle_data( |
| 1961 | (TreeBuilderObject*) self->target, value |
| 1962 | ); |
| 1963 | else if (self->handle_data) |
| 1964 | res = PyObject_CallFunction(self->handle_data, "O", value); |
| 1965 | else |
| 1966 | res = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1967 | Py_XDECREF(res); |
| 1968 | } else { |
| 1969 | PyErr_Format( |
| 1970 | PyExc_SyntaxError, "undefined entity &%s;: line %d, column %d", |
| 1971 | PyString_AS_STRING(key), |
| 1972 | EXPAT(GetErrorLineNumber)(self->parser), |
| 1973 | EXPAT(GetErrorColumnNumber)(self->parser) |
| 1974 | ); |
| 1975 | } |
| 1976 | |
| 1977 | Py_DECREF(key); |
| 1978 | } |
| 1979 | |
| 1980 | static void |
| 1981 | expat_start_handler(XMLParserObject* self, const XML_Char* tag_in, |
| 1982 | const XML_Char **attrib_in) |
| 1983 | { |
| 1984 | PyObject* res; |
| 1985 | PyObject* tag; |
| 1986 | PyObject* attrib; |
| 1987 | int ok; |
| 1988 | |
| 1989 | /* tag name */ |
| 1990 | tag = makeuniversal(self, tag_in); |
| 1991 | if (!tag) |
| 1992 | return; /* parser will look for errors */ |
| 1993 | |
| 1994 | /* attributes */ |
| 1995 | if (attrib_in[0]) { |
| 1996 | attrib = PyDict_New(); |
| 1997 | if (!attrib) |
| 1998 | return; |
| 1999 | while (attrib_in[0] && attrib_in[1]) { |
| 2000 | PyObject* key = makeuniversal(self, attrib_in[0]); |
| 2001 | PyObject* value = makestring(attrib_in[1], strlen(attrib_in[1])); |
| 2002 | if (!key || !value) { |
| 2003 | Py_XDECREF(value); |
| 2004 | Py_XDECREF(key); |
| 2005 | Py_DECREF(attrib); |
| 2006 | return; |
| 2007 | } |
| 2008 | ok = PyDict_SetItem(attrib, key, value); |
| 2009 | Py_DECREF(value); |
| 2010 | Py_DECREF(key); |
| 2011 | if (ok < 0) { |
| 2012 | Py_DECREF(attrib); |
| 2013 | return; |
| 2014 | } |
| 2015 | attrib_in += 2; |
| 2016 | } |
| 2017 | } else { |
| 2018 | Py_INCREF(Py_None); |
| 2019 | attrib = Py_None; |
| 2020 | } |
| 2021 | |
| 2022 | if (TreeBuilder_CheckExact(self->target)) |
| 2023 | /* shortcut */ |
| 2024 | res = treebuilder_handle_start((TreeBuilderObject*) self->target, |
| 2025 | tag, attrib); |
| 2026 | else if (self->handle_start) |
| 2027 | res = PyObject_CallFunction(self->handle_start, "OO", tag, attrib); |
| 2028 | else |
| 2029 | res = NULL; |
| 2030 | |
| 2031 | Py_DECREF(tag); |
| 2032 | Py_DECREF(attrib); |
| 2033 | |
| 2034 | Py_XDECREF(res); |
| 2035 | } |
| 2036 | |
| 2037 | static void |
| 2038 | expat_data_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2039 | int data_len) |
| 2040 | { |
| 2041 | PyObject* data; |
| 2042 | PyObject* res; |
| 2043 | |
| 2044 | data = makestring(data_in, data_len); |
| 2045 | |
| 2046 | if (TreeBuilder_CheckExact(self->target)) |
| 2047 | /* shortcut */ |
| 2048 | res = treebuilder_handle_data((TreeBuilderObject*) self->target, data); |
| 2049 | else if (self->handle_data) |
| 2050 | res = PyObject_CallFunction(self->handle_data, "O", data); |
| 2051 | else |
| 2052 | res = NULL; |
| 2053 | |
| 2054 | Py_DECREF(data); |
| 2055 | |
| 2056 | Py_XDECREF(res); |
| 2057 | } |
| 2058 | |
| 2059 | static void |
| 2060 | expat_end_handler(XMLParserObject* self, const XML_Char* tag_in) |
| 2061 | { |
| 2062 | PyObject* tag; |
| 2063 | PyObject* res = NULL; |
| 2064 | |
| 2065 | if (TreeBuilder_CheckExact(self->target)) |
| 2066 | /* shortcut */ |
| 2067 | /* the standard tree builder doesn't look at the end tag */ |
| 2068 | res = treebuilder_handle_end( |
| 2069 | (TreeBuilderObject*) self->target, Py_None |
| 2070 | ); |
| 2071 | else if (self->handle_end) { |
| 2072 | tag = makeuniversal(self, tag_in); |
| 2073 | if (tag) { |
| 2074 | res = PyObject_CallFunction(self->handle_end, "O", tag); |
| 2075 | Py_DECREF(tag); |
| 2076 | } |
| 2077 | } |
| 2078 | |
| 2079 | Py_XDECREF(res); |
| 2080 | } |
| 2081 | |
| 2082 | static void |
| 2083 | expat_start_ns_handler(XMLParserObject* self, const XML_Char* prefix, |
| 2084 | const XML_Char *uri) |
| 2085 | { |
| 2086 | treebuilder_handle_namespace( |
| 2087 | (TreeBuilderObject*) self->target, 1, prefix, uri |
| 2088 | ); |
| 2089 | } |
| 2090 | |
| 2091 | static void |
| 2092 | expat_end_ns_handler(XMLParserObject* self, const XML_Char* prefix_in) |
| 2093 | { |
| 2094 | treebuilder_handle_namespace( |
| 2095 | (TreeBuilderObject*) self->target, 0, NULL, NULL |
| 2096 | ); |
| 2097 | } |
| 2098 | |
| 2099 | static void |
| 2100 | expat_comment_handler(XMLParserObject* self, const XML_Char* comment_in) |
| 2101 | { |
| 2102 | PyObject* comment; |
| 2103 | PyObject* res; |
| 2104 | |
| 2105 | if (self->handle_comment) { |
| 2106 | comment = makestring(comment_in, strlen(comment_in)); |
| 2107 | if (comment) { |
| 2108 | res = PyObject_CallFunction(self->handle_comment, "O", comment); |
| 2109 | Py_XDECREF(res); |
| 2110 | Py_DECREF(comment); |
| 2111 | } |
| 2112 | } |
| 2113 | } |
| 2114 | |
| 2115 | static void |
| 2116 | expat_pi_handler(XMLParserObject* self, const XML_Char* target_in, |
| 2117 | const XML_Char* data_in) |
| 2118 | { |
| 2119 | PyObject* target; |
| 2120 | PyObject* data; |
| 2121 | PyObject* res; |
| 2122 | |
| 2123 | if (self->handle_pi) { |
| 2124 | target = makestring(target_in, strlen(target_in)); |
| 2125 | data = makestring(data_in, strlen(data_in)); |
| 2126 | if (target && data) { |
| 2127 | res = PyObject_CallFunction(self->handle_pi, "OO", target, data); |
| 2128 | Py_XDECREF(res); |
| 2129 | Py_DECREF(data); |
| 2130 | Py_DECREF(target); |
| 2131 | } else { |
| 2132 | Py_XDECREF(data); |
| 2133 | Py_XDECREF(target); |
| 2134 | } |
| 2135 | } |
| 2136 | } |
| 2137 | |
| 2138 | #if defined(Py_USING_UNICODE) |
| 2139 | static int |
| 2140 | expat_unknown_encoding_handler(XMLParserObject *self, const XML_Char *name, |
| 2141 | XML_Encoding *info) |
| 2142 | { |
| 2143 | PyObject* u; |
| 2144 | Py_UNICODE* p; |
| 2145 | unsigned char s[256]; |
| 2146 | int i; |
| 2147 | |
| 2148 | memset(info, 0, sizeof(XML_Encoding)); |
| 2149 | |
| 2150 | for (i = 0; i < 256; i++) |
| 2151 | s[i] = i; |
| 2152 | |
Fredrik Lundh | c338999 | 2005-12-25 11:40:19 +0000 | [diff] [blame] | 2153 | u = PyUnicode_Decode((char*) s, 256, name, "replace"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2154 | if (!u) |
| 2155 | return XML_STATUS_ERROR; |
| 2156 | |
| 2157 | if (PyUnicode_GET_SIZE(u) != 256) { |
| 2158 | Py_DECREF(u); |
| 2159 | return XML_STATUS_ERROR; |
| 2160 | } |
| 2161 | |
| 2162 | p = PyUnicode_AS_UNICODE(u); |
| 2163 | |
| 2164 | for (i = 0; i < 256; i++) { |
| 2165 | if (p[i] != Py_UNICODE_REPLACEMENT_CHARACTER) |
| 2166 | info->map[i] = p[i]; |
| 2167 | else |
| 2168 | info->map[i] = -1; |
| 2169 | } |
| 2170 | |
| 2171 | Py_DECREF(u); |
| 2172 | |
| 2173 | return XML_STATUS_OK; |
| 2174 | } |
| 2175 | #endif |
| 2176 | |
| 2177 | /* -------------------------------------------------------------------- */ |
| 2178 | /* constructor and destructor */ |
| 2179 | |
| 2180 | static PyObject* |
| 2181 | xmlparser(PyObject* _self, PyObject* args, PyObject* kw) |
| 2182 | { |
| 2183 | XMLParserObject* self; |
| 2184 | /* FIXME: does this need to be static? */ |
| 2185 | static XML_Memory_Handling_Suite memory_handler; |
| 2186 | |
| 2187 | PyObject* target = NULL; |
| 2188 | char* encoding = NULL; |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame^] | 2189 | static char* kwlist[] = { "target", "encoding", NULL }; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2190 | if (!PyArg_ParseTupleAndKeywords(args, kw, "|Oz:XMLParser", kwlist, |
| 2191 | &target, &encoding)) |
| 2192 | return NULL; |
| 2193 | |
| 2194 | #if defined(USE_PYEXPAT_CAPI) |
| 2195 | if (!expat_capi) { |
| 2196 | PyErr_SetString( |
| 2197 | PyExc_RuntimeError, "cannot load dispatch table from pyexpat" |
| 2198 | ); |
| 2199 | return NULL; |
| 2200 | } |
| 2201 | #endif |
| 2202 | |
| 2203 | self = PyObject_New(XMLParserObject, &XMLParser_Type); |
| 2204 | if (self == NULL) |
| 2205 | return NULL; |
| 2206 | |
| 2207 | self->entity = PyDict_New(); |
| 2208 | if (!self->entity) { |
| 2209 | PyObject_Del(self); |
| 2210 | return NULL; /* FIXME: cleanup on error */ |
| 2211 | } |
| 2212 | |
| 2213 | self->names = PyDict_New(); |
| 2214 | if (!self->names) { |
| 2215 | PyObject_Del(self); |
| 2216 | return NULL; /* FIXME: cleanup on error */ |
| 2217 | } |
| 2218 | |
| 2219 | memory_handler.malloc_fcn = PyObject_Malloc; |
| 2220 | memory_handler.realloc_fcn = PyObject_Realloc; |
| 2221 | memory_handler.free_fcn = PyObject_Free; |
| 2222 | |
| 2223 | self->parser = EXPAT(ParserCreate_MM)(encoding, &memory_handler, "}"); |
| 2224 | if (!self->parser) { |
| 2225 | PyErr_NoMemory(); |
| 2226 | return NULL; /* FIXME: cleanup on error */ |
| 2227 | } |
| 2228 | |
| 2229 | /* setup target handlers */ |
| 2230 | if (!target) { |
| 2231 | target = treebuilder_new(); |
| 2232 | if (!target) { |
| 2233 | PyObject_Del(self); |
| 2234 | return NULL; /* FIXME: cleanup on error */ |
| 2235 | } |
| 2236 | } else |
| 2237 | Py_INCREF(target); |
| 2238 | self->target = target; |
| 2239 | |
| 2240 | self->handle_xml = PyObject_GetAttrString(target, "xml"); |
| 2241 | self->handle_start = PyObject_GetAttrString(target, "start"); |
| 2242 | self->handle_data = PyObject_GetAttrString(target, "data"); |
| 2243 | self->handle_end = PyObject_GetAttrString(target, "end"); |
| 2244 | self->handle_comment = PyObject_GetAttrString(target, "comment"); |
| 2245 | self->handle_pi = PyObject_GetAttrString(target, "pi"); |
| 2246 | |
| 2247 | PyErr_Clear(); |
| 2248 | |
| 2249 | /* configure parser */ |
| 2250 | EXPAT(SetUserData)(self->parser, self); |
| 2251 | EXPAT(SetElementHandler)( |
| 2252 | self->parser, |
| 2253 | (XML_StartElementHandler) expat_start_handler, |
| 2254 | (XML_EndElementHandler) expat_end_handler |
| 2255 | ); |
| 2256 | EXPAT(SetDefaultHandlerExpand)( |
| 2257 | self->parser, |
| 2258 | (XML_DefaultHandler) expat_default_handler |
| 2259 | ); |
| 2260 | EXPAT(SetCharacterDataHandler)( |
| 2261 | self->parser, |
| 2262 | (XML_CharacterDataHandler) expat_data_handler |
| 2263 | ); |
| 2264 | if (self->handle_comment) |
| 2265 | EXPAT(SetCommentHandler)( |
| 2266 | self->parser, |
| 2267 | (XML_CommentHandler) expat_comment_handler |
| 2268 | ); |
| 2269 | if (self->handle_pi) |
| 2270 | EXPAT(SetProcessingInstructionHandler)( |
| 2271 | self->parser, |
| 2272 | (XML_ProcessingInstructionHandler) expat_pi_handler |
| 2273 | ); |
| 2274 | #if defined(Py_USING_UNICODE) |
| 2275 | EXPAT(SetUnknownEncodingHandler)( |
| 2276 | self->parser, |
| 2277 | (XML_UnknownEncodingHandler) expat_unknown_encoding_handler, NULL |
| 2278 | ); |
| 2279 | #endif |
| 2280 | |
| 2281 | ALLOC(sizeof(XMLParserObject), "create expatparser"); |
| 2282 | |
| 2283 | return (PyObject*) self; |
| 2284 | } |
| 2285 | |
| 2286 | static void |
| 2287 | xmlparser_dealloc(XMLParserObject* self) |
| 2288 | { |
| 2289 | EXPAT(ParserFree)(self->parser); |
| 2290 | |
| 2291 | Py_XDECREF(self->handle_pi); |
| 2292 | Py_XDECREF(self->handle_comment); |
| 2293 | Py_XDECREF(self->handle_end); |
| 2294 | Py_XDECREF(self->handle_data); |
| 2295 | Py_XDECREF(self->handle_start); |
| 2296 | Py_XDECREF(self->handle_xml); |
| 2297 | |
| 2298 | Py_DECREF(self->target); |
| 2299 | Py_DECREF(self->entity); |
| 2300 | Py_DECREF(self->names); |
| 2301 | |
| 2302 | RELEASE(sizeof(XMLParserObject), "destroy expatparser"); |
| 2303 | |
| 2304 | PyObject_Del(self); |
| 2305 | } |
| 2306 | |
| 2307 | /* -------------------------------------------------------------------- */ |
| 2308 | /* methods (in alphabetical order) */ |
| 2309 | |
| 2310 | LOCAL(PyObject*) |
| 2311 | expat_parse(XMLParserObject* self, char* data, int data_len, int final) |
| 2312 | { |
| 2313 | int ok; |
| 2314 | |
| 2315 | ok = EXPAT(Parse)(self->parser, data, data_len, final); |
| 2316 | |
| 2317 | if (PyErr_Occurred()) |
| 2318 | return NULL; |
| 2319 | |
| 2320 | if (!ok) { |
| 2321 | PyErr_Format( |
| 2322 | PyExc_SyntaxError, "%s: line %d, column %d", |
| 2323 | EXPAT(ErrorString)(EXPAT(GetErrorCode)(self->parser)), |
| 2324 | EXPAT(GetErrorLineNumber)(self->parser), |
| 2325 | EXPAT(GetErrorColumnNumber)(self->parser) |
| 2326 | ); |
| 2327 | return NULL; |
| 2328 | } |
| 2329 | |
| 2330 | Py_RETURN_NONE; |
| 2331 | } |
| 2332 | |
| 2333 | static PyObject* |
| 2334 | xmlparser_close(XMLParserObject* self, PyObject* args) |
| 2335 | { |
| 2336 | /* end feeding data to parser */ |
| 2337 | |
| 2338 | PyObject* res; |
| 2339 | if (!PyArg_ParseTuple(args, ":close")) |
| 2340 | return NULL; |
| 2341 | |
| 2342 | res = expat_parse(self, "", 0, 1); |
| 2343 | |
| 2344 | if (res && TreeBuilder_CheckExact(self->target)) { |
| 2345 | Py_DECREF(res); |
| 2346 | return treebuilder_done((TreeBuilderObject*) self->target); |
| 2347 | } |
| 2348 | |
| 2349 | return res; |
| 2350 | } |
| 2351 | |
| 2352 | static PyObject* |
| 2353 | xmlparser_feed(XMLParserObject* self, PyObject* args) |
| 2354 | { |
| 2355 | /* feed data to parser */ |
| 2356 | |
| 2357 | char* data; |
| 2358 | int data_len; |
| 2359 | if (!PyArg_ParseTuple(args, "s#:feed", &data, &data_len)) |
| 2360 | return NULL; |
| 2361 | |
| 2362 | return expat_parse(self, data, data_len, 0); |
| 2363 | } |
| 2364 | |
| 2365 | static PyObject* |
| 2366 | xmlparser_parse(XMLParserObject* self, PyObject* args) |
| 2367 | { |
| 2368 | /* (internal) parse until end of input stream */ |
| 2369 | |
| 2370 | PyObject* reader; |
| 2371 | PyObject* buffer; |
| 2372 | PyObject* res; |
| 2373 | |
| 2374 | PyObject* fileobj; |
| 2375 | if (!PyArg_ParseTuple(args, "O:_parse", &fileobj)) |
| 2376 | return NULL; |
| 2377 | |
| 2378 | reader = PyObject_GetAttrString(fileobj, "read"); |
| 2379 | if (!reader) |
| 2380 | return NULL; |
| 2381 | |
| 2382 | /* read from open file object */ |
| 2383 | for (;;) { |
| 2384 | |
| 2385 | buffer = PyObject_CallFunction(reader, "i", 64*1024); |
| 2386 | |
| 2387 | if (!buffer) { |
| 2388 | /* read failed (e.g. due to KeyboardInterrupt) */ |
| 2389 | Py_DECREF(reader); |
| 2390 | return NULL; |
| 2391 | } |
| 2392 | |
| 2393 | if (!PyString_CheckExact(buffer) || PyString_GET_SIZE(buffer) == 0) { |
| 2394 | Py_DECREF(buffer); |
| 2395 | break; |
| 2396 | } |
| 2397 | |
| 2398 | res = expat_parse( |
| 2399 | self, PyString_AS_STRING(buffer), PyString_GET_SIZE(buffer), 0 |
| 2400 | ); |
| 2401 | |
| 2402 | Py_DECREF(buffer); |
| 2403 | |
| 2404 | if (!res) { |
| 2405 | Py_DECREF(reader); |
| 2406 | return NULL; |
| 2407 | } |
| 2408 | Py_DECREF(res); |
| 2409 | |
| 2410 | } |
| 2411 | |
| 2412 | Py_DECREF(reader); |
| 2413 | |
| 2414 | res = expat_parse(self, "", 0, 1); |
| 2415 | |
| 2416 | if (res && TreeBuilder_CheckExact(self->target)) { |
| 2417 | Py_DECREF(res); |
| 2418 | return treebuilder_done((TreeBuilderObject*) self->target); |
| 2419 | } |
| 2420 | |
| 2421 | return res; |
| 2422 | } |
| 2423 | |
| 2424 | static PyObject* |
| 2425 | xmlparser_setevents(XMLParserObject* self, PyObject* args) |
| 2426 | { |
| 2427 | /* activate element event reporting */ |
| 2428 | |
| 2429 | int i; |
| 2430 | TreeBuilderObject* target; |
| 2431 | |
| 2432 | PyObject* events; /* event collector */ |
| 2433 | PyObject* event_set = Py_None; |
| 2434 | if (!PyArg_ParseTuple(args, "O!|O:_setevents", &PyList_Type, &events, |
| 2435 | &event_set)) |
| 2436 | return NULL; |
| 2437 | |
| 2438 | if (!TreeBuilder_CheckExact(self->target)) { |
| 2439 | PyErr_SetString( |
| 2440 | PyExc_TypeError, |
| 2441 | "event handling only supported for cElementTree.Treebuilder " |
| 2442 | "targets" |
| 2443 | ); |
| 2444 | return NULL; |
| 2445 | } |
| 2446 | |
| 2447 | target = (TreeBuilderObject*) self->target; |
| 2448 | |
| 2449 | Py_INCREF(events); |
| 2450 | Py_XDECREF(target->events); |
| 2451 | target->events = events; |
| 2452 | |
| 2453 | /* clear out existing events */ |
| 2454 | Py_XDECREF(target->start_event_obj); target->start_event_obj = NULL; |
| 2455 | Py_XDECREF(target->end_event_obj); target->end_event_obj = NULL; |
| 2456 | Py_XDECREF(target->start_ns_event_obj); target->start_ns_event_obj = NULL; |
| 2457 | Py_XDECREF(target->end_ns_event_obj); target->end_ns_event_obj = NULL; |
| 2458 | |
| 2459 | if (event_set == Py_None) { |
| 2460 | /* default is "end" only */ |
| 2461 | target->end_event_obj = PyString_FromString("end"); |
| 2462 | Py_RETURN_NONE; |
| 2463 | } |
| 2464 | |
| 2465 | if (!PyTuple_Check(event_set)) /* FIXME: handle arbitrary sequences */ |
| 2466 | goto error; |
| 2467 | |
| 2468 | for (i = 0; i < PyTuple_GET_SIZE(event_set); i++) { |
| 2469 | PyObject* item = PyTuple_GET_ITEM(event_set, i); |
| 2470 | char* event; |
| 2471 | if (!PyString_Check(item)) |
| 2472 | goto error; |
| 2473 | event = PyString_AS_STRING(item); |
| 2474 | if (strcmp(event, "start") == 0) { |
| 2475 | Py_INCREF(item); |
| 2476 | target->start_event_obj = item; |
| 2477 | } else if (strcmp(event, "end") == 0) { |
| 2478 | Py_INCREF(item); |
| 2479 | Py_XDECREF(target->end_event_obj); |
| 2480 | target->end_event_obj = item; |
| 2481 | } else if (strcmp(event, "start-ns") == 0) { |
| 2482 | Py_INCREF(item); |
| 2483 | Py_XDECREF(target->start_ns_event_obj); |
| 2484 | target->start_ns_event_obj = item; |
| 2485 | EXPAT(SetNamespaceDeclHandler)( |
| 2486 | self->parser, |
| 2487 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 2488 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 2489 | ); |
| 2490 | } else if (strcmp(event, "end-ns") == 0) { |
| 2491 | Py_INCREF(item); |
| 2492 | Py_XDECREF(target->end_ns_event_obj); |
| 2493 | target->end_ns_event_obj = item; |
| 2494 | EXPAT(SetNamespaceDeclHandler)( |
| 2495 | self->parser, |
| 2496 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 2497 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 2498 | ); |
| 2499 | } else { |
| 2500 | PyErr_Format( |
| 2501 | PyExc_ValueError, |
| 2502 | "unknown event '%s'", event |
| 2503 | ); |
| 2504 | return NULL; |
| 2505 | } |
| 2506 | } |
| 2507 | |
| 2508 | Py_RETURN_NONE; |
| 2509 | |
| 2510 | error: |
| 2511 | PyErr_SetString( |
| 2512 | PyExc_TypeError, |
| 2513 | "invalid event tuple" |
| 2514 | ); |
| 2515 | return NULL; |
| 2516 | } |
| 2517 | |
| 2518 | static PyMethodDef xmlparser_methods[] = { |
| 2519 | {"feed", (PyCFunction) xmlparser_feed, METH_VARARGS}, |
| 2520 | {"close", (PyCFunction) xmlparser_close, METH_VARARGS}, |
| 2521 | {"_parse", (PyCFunction) xmlparser_parse, METH_VARARGS}, |
| 2522 | {"_setevents", (PyCFunction) xmlparser_setevents, METH_VARARGS}, |
| 2523 | {NULL, NULL} |
| 2524 | }; |
| 2525 | |
| 2526 | static PyObject* |
| 2527 | xmlparser_getattr(XMLParserObject* self, char* name) |
| 2528 | { |
| 2529 | PyObject* res; |
| 2530 | |
| 2531 | res = Py_FindMethod(xmlparser_methods, (PyObject*) self, name); |
| 2532 | if (res) |
| 2533 | return res; |
| 2534 | |
| 2535 | PyErr_Clear(); |
| 2536 | |
| 2537 | if (strcmp(name, "entity") == 0) |
| 2538 | res = self->entity; |
| 2539 | else if (strcmp(name, "target") == 0) |
| 2540 | res = self->target; |
| 2541 | else if (strcmp(name, "version") == 0) { |
| 2542 | char buffer[100]; |
| 2543 | sprintf(buffer, "Expat %d.%d.%d", XML_MAJOR_VERSION, |
| 2544 | XML_MINOR_VERSION, XML_MICRO_VERSION); |
| 2545 | return PyString_FromString(buffer); |
| 2546 | } else { |
| 2547 | PyErr_SetString(PyExc_AttributeError, name); |
| 2548 | return NULL; |
| 2549 | } |
| 2550 | |
| 2551 | Py_INCREF(res); |
| 2552 | return res; |
| 2553 | } |
| 2554 | |
| 2555 | statichere PyTypeObject XMLParser_Type = { |
| 2556 | PyObject_HEAD_INIT(NULL) |
| 2557 | 0, "XMLParser", sizeof(XMLParserObject), 0, |
| 2558 | /* methods */ |
| 2559 | (destructor)xmlparser_dealloc, /* tp_dealloc */ |
| 2560 | 0, /* tp_print */ |
| 2561 | (getattrfunc)xmlparser_getattr, /* tp_getattr */ |
| 2562 | }; |
| 2563 | |
| 2564 | #endif |
| 2565 | |
| 2566 | /* ==================================================================== */ |
| 2567 | /* python module interface */ |
| 2568 | |
| 2569 | static PyMethodDef _functions[] = { |
| 2570 | {"Element", (PyCFunction) element, METH_VARARGS|METH_KEYWORDS}, |
| 2571 | {"SubElement", (PyCFunction) subelement, METH_VARARGS|METH_KEYWORDS}, |
| 2572 | {"TreeBuilder", (PyCFunction) treebuilder, METH_VARARGS}, |
| 2573 | #if defined(USE_EXPAT) |
| 2574 | {"XMLParser", (PyCFunction) xmlparser, METH_VARARGS|METH_KEYWORDS}, |
| 2575 | {"XMLTreeBuilder", (PyCFunction) xmlparser, METH_VARARGS|METH_KEYWORDS}, |
| 2576 | #endif |
| 2577 | {NULL, NULL} |
| 2578 | }; |
| 2579 | |
| 2580 | DL_EXPORT(void) |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2581 | init_elementtree(void) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2582 | { |
| 2583 | PyObject* m; |
| 2584 | PyObject* g; |
| 2585 | char* bootstrap; |
| 2586 | #if defined(USE_PYEXPAT_CAPI) |
| 2587 | struct PyExpat_CAPI* capi; |
| 2588 | #endif |
| 2589 | |
| 2590 | /* Patch object type */ |
| 2591 | Element_Type.ob_type = TreeBuilder_Type.ob_type = &PyType_Type; |
| 2592 | #if defined(USE_EXPAT) |
| 2593 | XMLParser_Type.ob_type = &PyType_Type; |
| 2594 | #endif |
| 2595 | |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2596 | m = Py_InitModule("_elementtree", _functions); |
Neal Norwitz | 1ac754f | 2006-01-19 06:09:39 +0000 | [diff] [blame] | 2597 | if (m == NULL) |
| 2598 | return; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2599 | |
| 2600 | /* python glue code */ |
| 2601 | |
| 2602 | g = PyDict_New(); |
Neal Norwitz | 02876df | 2006-02-07 06:58:52 +0000 | [diff] [blame] | 2603 | if (g == NULL) |
| 2604 | return; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2605 | |
| 2606 | PyDict_SetItemString(g, "__builtins__", PyEval_GetBuiltins()); |
| 2607 | |
| 2608 | bootstrap = ( |
| 2609 | |
| 2610 | #if (PY_VERSION_HEX >= 0x02020000 && PY_VERSION_HEX < 0x02030000) |
| 2611 | "from __future__ import generators\n" /* enable yield under 2.2 */ |
| 2612 | #endif |
| 2613 | |
| 2614 | "from copy import copy, deepcopy\n" |
| 2615 | |
| 2616 | "try:\n" |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2617 | " from xml.etree import ElementTree\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2618 | "except ImportError:\n" |
| 2619 | " import ElementTree\n" |
| 2620 | "ET = ElementTree\n" |
| 2621 | "del ElementTree\n" |
| 2622 | |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2623 | "import _elementtree as cElementTree\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2624 | |
| 2625 | "try:\n" /* check if copy works as is */ |
| 2626 | " copy(cElementTree.Element('x'))\n" |
| 2627 | "except:\n" |
| 2628 | " def copyelement(elem):\n" |
| 2629 | " return elem\n" |
| 2630 | |
| 2631 | "def Comment(text=None):\n" /* public */ |
| 2632 | " element = cElementTree.Element(ET.Comment)\n" |
| 2633 | " element.text = text\n" |
| 2634 | " return element\n" |
| 2635 | "cElementTree.Comment = Comment\n" |
| 2636 | |
| 2637 | "class ElementTree(ET.ElementTree):\n" /* public */ |
| 2638 | " def parse(self, source, parser=None):\n" |
| 2639 | " if not hasattr(source, 'read'):\n" |
| 2640 | " source = open(source, 'rb')\n" |
| 2641 | " if parser is not None:\n" |
| 2642 | " while 1:\n" |
| 2643 | " data = source.read(65536)\n" |
| 2644 | " if not data:\n" |
| 2645 | " break\n" |
| 2646 | " parser.feed(data)\n" |
| 2647 | " self._root = parser.close()\n" |
| 2648 | " else:\n" |
| 2649 | " parser = cElementTree.XMLParser()\n" |
| 2650 | " self._root = parser._parse(source)\n" |
| 2651 | " return self._root\n" |
| 2652 | "cElementTree.ElementTree = ElementTree\n" |
| 2653 | |
| 2654 | "def getiterator(node, tag=None):\n" /* helper */ |
| 2655 | " if tag == '*':\n" |
| 2656 | " tag = None\n" |
| 2657 | #if (PY_VERSION_HEX < 0x02020000) |
| 2658 | " nodes = []\n" /* 2.1 doesn't have yield */ |
| 2659 | " if tag is None or node.tag == tag:\n" |
| 2660 | " nodes.append(node)\n" |
| 2661 | " for node in node:\n" |
| 2662 | " nodes.extend(getiterator(node, tag))\n" |
| 2663 | " return nodes\n" |
| 2664 | #else |
| 2665 | " if tag is None or node.tag == tag:\n" |
| 2666 | " yield node\n" |
| 2667 | " for node in node:\n" |
| 2668 | " for node in getiterator(node, tag):\n" |
| 2669 | " yield node\n" |
| 2670 | #endif |
| 2671 | |
| 2672 | "def parse(source, parser=None):\n" /* public */ |
| 2673 | " tree = ElementTree()\n" |
| 2674 | " tree.parse(source, parser)\n" |
| 2675 | " return tree\n" |
| 2676 | "cElementTree.parse = parse\n" |
| 2677 | |
| 2678 | #if (PY_VERSION_HEX < 0x02020000) |
| 2679 | "if hasattr(ET, 'iterparse'):\n" |
| 2680 | " cElementTree.iterparse = ET.iterparse\n" /* delegate on 2.1 */ |
| 2681 | #else |
| 2682 | "class iterparse(object):\n" |
| 2683 | " root = None\n" |
| 2684 | " def __init__(self, file, events=None):\n" |
| 2685 | " if not hasattr(file, 'read'):\n" |
| 2686 | " file = open(file, 'rb')\n" |
| 2687 | " self._file = file\n" |
| 2688 | " self._events = events\n" |
| 2689 | " def __iter__(self):\n" |
| 2690 | " events = []\n" |
| 2691 | " b = cElementTree.TreeBuilder()\n" |
| 2692 | " p = cElementTree.XMLParser(b)\n" |
| 2693 | " p._setevents(events, self._events)\n" |
| 2694 | " while 1:\n" |
| 2695 | " data = self._file.read(16384)\n" |
| 2696 | " if not data:\n" |
| 2697 | " break\n" |
| 2698 | " p.feed(data)\n" |
| 2699 | " for event in events:\n" |
| 2700 | " yield event\n" |
| 2701 | " del events[:]\n" |
| 2702 | " root = p.close()\n" |
| 2703 | " for event in events:\n" |
| 2704 | " yield event\n" |
| 2705 | " self.root = root\n" |
| 2706 | "cElementTree.iterparse = iterparse\n" |
| 2707 | #endif |
| 2708 | |
| 2709 | "def PI(target, text=None):\n" /* public */ |
| 2710 | " element = cElementTree.Element(ET.ProcessingInstruction)\n" |
| 2711 | " element.text = target\n" |
| 2712 | " if text:\n" |
| 2713 | " element.text = element.text + ' ' + text\n" |
| 2714 | " return element\n" |
| 2715 | |
| 2716 | " elem = cElementTree.Element(ET.PI)\n" |
| 2717 | " elem.text = text\n" |
| 2718 | " return elem\n" |
| 2719 | "cElementTree.PI = cElementTree.ProcessingInstruction = PI\n" |
| 2720 | |
| 2721 | "def XML(text):\n" /* public */ |
| 2722 | " parser = cElementTree.XMLParser()\n" |
| 2723 | " parser.feed(text)\n" |
| 2724 | " return parser.close()\n" |
| 2725 | "cElementTree.XML = cElementTree.fromstring = XML\n" |
| 2726 | |
| 2727 | "def XMLID(text):\n" /* public */ |
| 2728 | " tree = XML(text)\n" |
| 2729 | " ids = {}\n" |
| 2730 | " for elem in tree.getiterator():\n" |
| 2731 | " id = elem.get('id')\n" |
| 2732 | " if id:\n" |
| 2733 | " ids[id] = elem\n" |
| 2734 | " return tree, ids\n" |
| 2735 | "cElementTree.XMLID = XMLID\n" |
| 2736 | |
| 2737 | "cElementTree.dump = ET.dump\n" |
| 2738 | "cElementTree.ElementPath = ElementPath = ET.ElementPath\n" |
| 2739 | "cElementTree.iselement = ET.iselement\n" |
| 2740 | "cElementTree.QName = ET.QName\n" |
| 2741 | "cElementTree.tostring = ET.tostring\n" |
| 2742 | "cElementTree.VERSION = '" VERSION "'\n" |
| 2743 | "cElementTree.__version__ = '" VERSION "'\n" |
| 2744 | "cElementTree.XMLParserError = SyntaxError\n" |
| 2745 | |
| 2746 | ); |
| 2747 | |
| 2748 | PyRun_String(bootstrap, Py_file_input, g, NULL); |
| 2749 | |
| 2750 | elementpath_obj = PyDict_GetItemString(g, "ElementPath"); |
| 2751 | |
| 2752 | elementtree_copyelement_obj = PyDict_GetItemString(g, "copyelement"); |
| 2753 | if (elementtree_copyelement_obj) { |
| 2754 | /* reduce hack needed; enable reduce method */ |
| 2755 | PyMethodDef* mp; |
| 2756 | for (mp = element_methods; mp->ml_name; mp++) |
| 2757 | if (mp->ml_meth == (PyCFunction) element_reduce) { |
| 2758 | mp->ml_name = "__reduce__"; |
| 2759 | break; |
| 2760 | } |
| 2761 | } else |
| 2762 | PyErr_Clear(); |
| 2763 | elementtree_deepcopy_obj = PyDict_GetItemString(g, "deepcopy"); |
| 2764 | elementtree_getiterator_obj = PyDict_GetItemString(g, "getiterator"); |
| 2765 | |
| 2766 | #if defined(USE_PYEXPAT_CAPI) |
| 2767 | /* link against pyexpat, if possible */ |
| 2768 | capi = PyCObject_Import("pyexpat", "expat_CAPI"); |
| 2769 | if (capi && |
| 2770 | strcmp(capi->magic, PyExpat_CAPI_MAGIC) == 0 && |
| 2771 | capi->size <= sizeof(*expat_capi) && |
| 2772 | capi->MAJOR_VERSION == XML_MAJOR_VERSION && |
| 2773 | capi->MINOR_VERSION == XML_MINOR_VERSION && |
| 2774 | capi->MICRO_VERSION == XML_MICRO_VERSION) |
| 2775 | expat_capi = capi; |
| 2776 | else |
| 2777 | expat_capi = NULL; |
| 2778 | #endif |
| 2779 | |
| 2780 | } |