Modules/_elementtree.c

   1 /*
   2  * ElementTree
   3  * $Id: _elementtree.c 2657 2006-03-12 20:50:32Z 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  * 2006-03-08 fl  fixed a couple of potential null-refs and leaks
  37  * 2006-03-12 fl  merge in 2.5 ssize_t changes
  38  *
  39  * Copyright (c) 1999-2006 by Secret Labs AB.  All rights reserved.
  40  * Copyright (c) 1999-2006 by Fredrik Lundh.
  41  *
  42  * info@pythonware.com
  43  * http://www.pythonware.com
  44  */
  45
  46 /* Licensed to PSF under a Contributor Agreement. */
  47 /* See http://www.python.org/2.4/license for licensing details. */
  48
  49 #include "Python.h"
  50
  51 #define VERSION "1.0.6"
  52
  53 /* -------------------------------------------------------------------- */
  54 /* configuration */
  55
  56 /* Leave defined to include the expat-based XMLParser type */
  57 #define USE_EXPAT
  58
  59 /* Define to to all expat calls via pyexpat's embedded expat library */
  60 /* #define USE_PYEXPAT_CAPI */
  61
  62 /* An element can hold this many children without extra memory
  63    allocations. */
  64 #define STATIC_CHILDREN 4
  65
  66 /* For best performance, chose a value so that 80-90% of all nodes
  67    have no more than the given number of children.  Set this to zero
  68    to minimize the size of the element structure itself (this only
  69    helps if you have lots of leaf nodes with attributes). */
  70
  71 /* Also note that pymalloc always allocates blocks in multiples of
  72    eight bytes.  For the current version of cElementTree, this means
  73    that the number of children should be an even number, at least on
  74    32-bit platforms. */
  75
  76 /* -------------------------------------------------------------------- */
  77
  78 #if 0
  79 static int memory = 0;
  80 #define ALLOC(size, comment)\
  81 do { memory += size; printf("%8d - %s\n", memory, comment); } while (0)
  82 #define RELEASE(size, comment)\
  83 do { memory -= size; printf("%8d - %s\n", memory, comment); } while (0)
  84 #else
  85 #define ALLOC(size, comment)
  86 #define RELEASE(size, comment)
  87 #endif
  88
  89 /* compiler tweaks */
  90 #if defined(_MSC_VER)
  91 #define LOCAL(type) static __inline type __fastcall
  92 #else
  93 #define LOCAL(type) static type
  94 #endif
  95
  96 /* compatibility macros */
  97 #if (PY_VERSION_HEX < 0x02050000)
  98 typedef int Py_ssize_t;
  99 #define lenfunc inquiry
 100 #endif
 101
 102 #if (PY_VERSION_HEX < 0x02040000)
 103 #define PyDict_CheckExact PyDict_Check
 104 #if (PY_VERSION_HEX < 0x02020000)
 105 #define PyList_CheckExact PyList_Check
 106 #define PyString_CheckExact PyString_Check
 107 #if (PY_VERSION_HEX >= 0x01060000)
 108 #define Py_USING_UNICODE /* always enabled for 2.0 and 2.1 */
 109 #endif
 110 #endif
 111 #endif
 112
 113 #if !defined(Py_RETURN_NONE)
 114 #define Py_RETURN_NONE return Py_INCREF(Py_None), Py_None
 115 #endif
 116
 117 /* macros used to store 'join' flags in string object pointers.  note
 118    that all use of text and tail as object pointers must be wrapped in
 119    JOIN_OBJ.  see comments in the ElementObject definition for more
 120    info. */
 121 #define JOIN_GET(p) ((Py_uintptr_t) (p) & 1)
 122 #define JOIN_SET(p, flag) ((void*) ((Py_uintptr_t) (JOIN_OBJ(p)) | (flag)))
 123 #define JOIN_OBJ(p) ((PyObject*) ((Py_uintptr_t) (p) & ~1))
 124
 125 /* glue functions (see the init function for details) */
 126 static PyObject* elementtree_copyelement_obj;
 127 static PyObject* elementtree_deepcopy_obj;
 128 static PyObject* elementtree_getiterator_obj;
 129 static PyObject* elementpath_obj;
 130
 131 /* helpers */
 132
 133 LOCAL(PyObject*)
 134 deepcopy(PyObject* object, PyObject* memo)
 135 {
 136     /* do a deep copy of the given object */
 137
 138     PyObject* args;
 139     PyObject* result;
 140
 141     if (!elementtree_deepcopy_obj) {
 142         PyErr_SetString(
 143             PyExc_RuntimeError,
 144             "deepcopy helper not found"
 145             );
 146         return NULL;
 147     }
 148
 149     args = PyTuple_New(2);
 150     if (!args)
 151         return NULL;
 152
 153     Py_INCREF(object); PyTuple_SET_ITEM(args, 0, (PyObject*) object);
 154     Py_INCREF(memo);   PyTuple_SET_ITEM(args, 1, (PyObject*) memo);
 155
 156     result = PyObject_CallObject(elementtree_deepcopy_obj, args);
 157
 158     Py_DECREF(args);
 159
 160     return result;
 161 }
 162
 163 LOCAL(PyObject*)
 164 list_join(PyObject* list)
 165 {
 166     /* join list elements (destroying the list in the process) */
 167
 168     PyObject* joiner;
 169     PyObject* function;
 170     PyObject* args;
 171     PyObject* result;
 172
 173     switch (PyList_GET_SIZE(list)) {
 174     case 0:
 175         Py_DECREF(list);
 176         return PyString_FromString("");
 177     case 1:
 178         result = PyList_GET_ITEM(list, 0);
 179         Py_INCREF(result);
 180         Py_DECREF(list);
 181         return result;
 182     }
 183
 184     /* two or more elements: slice out a suitable separator from the
 185        first member, and use that to join the entire list */
 186
 187     joiner = PySequence_GetSlice(PyList_GET_ITEM(list, 0), 0, 0);
 188     if (!joiner)
 189         return NULL;
 190
 191     function = PyObject_GetAttrString(joiner, "join");
 192     if (!function) {
 193         Py_DECREF(joiner);
 194         return NULL;
 195     }
 196
 197     args = PyTuple_New(1);
 198     if (!args)
 199         return NULL;
 200
 201     PyTuple_SET_ITEM(args, 0, list);
 202
 203     result = PyObject_CallObject(function, args);
 204
 205     Py_DECREF(args); /* also removes list */
 206     Py_DECREF(function);
 207     Py_DECREF(joiner);
 208
 209     return result;
 210 }
 211
 212 #if (PY_VERSION_HEX < 0x02020000)
 213 LOCAL(int)
 214 PyDict_Update(PyObject* dict, PyObject* other)
 215 {
 216     /* PyDict_Update emulation for 2.1 and earlier */
 217
 218     PyObject* res;
 219
 220     res = PyObject_CallMethod(dict, "update", "O", other);
 221     if (!res)
 222         return -1;
 223
 224     Py_DECREF(res);
 225     return 0;
 226 }
 227 #endif
 228
 229 /* -------------------------------------------------------------------- */
 230 /* the element type */
 231
 232 typedef struct {
 233
 234     /* attributes (a dictionary object), or None if no attributes */
 235     PyObject* attrib;
 236
 237     /* child elements */
 238     int length; /* actual number of items */
 239     int allocated; /* allocated items */
 240
 241     /* this either points to _children or to a malloced buffer */
 242     PyObject* *children;
 243
 244     PyObject* _children[STATIC_CHILDREN];
 245
 246 } ElementObjectExtra;
 247
 248 typedef struct {
 249     PyObject_HEAD
 250
 251     /* element tag (a string). */
 252     PyObject* tag;
 253
 254     /* text before first child.  note that this is a tagged pointer;
 255        use JOIN_OBJ to get the object pointer.  the join flag is used
 256        to distinguish lists created by the tree builder from lists
 257        assigned to the attribute by application code; the former
 258        should be joined before being returned to the user, the latter
 259        should be left intact. */
 260     PyObject* text;
 261
 262     /* text after this element, in parent.  note that this is a tagged
 263        pointer; use JOIN_OBJ to get the object pointer. */
 264     PyObject* tail;
 265
 266     ElementObjectExtra* extra;
 267
 268 } ElementObject;
 269
 270 staticforward PyTypeObject Element_Type;
 271
 272 #define Element_CheckExact(op) ((op)->ob_type == &Element_Type)
 273
 274 /* -------------------------------------------------------------------- */
 275 /* element constructor and destructor */
 276
 277 LOCAL(int)
 278 element_new_extra(ElementObject* self, PyObject* attrib)
 279 {
 280     self->extra = PyObject_Malloc(sizeof(ElementObjectExtra));
 281     if (!self->extra)
 282         return -1;
 283
 284     if (!attrib)
 285         attrib = Py_None;
 286
 287     Py_INCREF(attrib);
 288     self->extra->attrib = attrib;
 289
 290     self->extra->length = 0;
 291     self->extra->allocated = STATIC_CHILDREN;
 292     self->extra->children = self->extra->_children;
 293
 294     return 0;
 295 }
 296
 297 LOCAL(void)
 298 element_dealloc_extra(ElementObject* self)
 299 {
 300     int i;
 301
 302     Py_DECREF(self->extra->attrib);
 303
 304     for (i = 0; i < self->extra->length; i++)
 305         Py_DECREF(self->extra->children[i]);
 306
 307     if (self->extra->children != self->extra->_children)
 308         PyObject_Free(self->extra->children);
 309
 310     PyObject_Free(self->extra);
 311 }
 312
 313 LOCAL(PyObject*)
 314 element_new(PyObject* tag, PyObject* attrib)
 315 {
 316     ElementObject* self;
 317
 318     self = PyObject_New(ElementObject, &Element_Type);
 319     if (self == NULL)
 320         return NULL;
 321
 322     /* use None for empty dictionaries */
 323     if (PyDict_CheckExact(attrib) && !PyDict_Size(attrib))
 324         attrib = Py_None;
 325
 326     self->extra = NULL;
 327
 328     if (attrib != Py_None) {
 329
 330         if (element_new_extra(self, attrib) < 0) {
 331             PyObject_Del(self);
 332             return NULL;
 333         }
 334
 335         self->extra->length = 0;
 336         self->extra->allocated = STATIC_CHILDREN;
 337         self->extra->children = self->extra->_children;
 338
 339     }
 340
 341     Py_INCREF(tag);
 342     self->tag = tag;
 343
 344     Py_INCREF(Py_None);
 345     self->text = Py_None;
 346
 347     Py_INCREF(Py_None);
 348     self->tail = Py_None;
 349
 350     ALLOC(sizeof(ElementObject), "create element");
 351
 352     return (PyObject*) self;
 353 }
 354
 355 LOCAL(int)
 356 element_resize(ElementObject* self, int extra)
 357 {
 358     int size;
 359     PyObject* *children;
 360
 361     /* make sure self->children can hold the given number of extra
 362        elements.  set an exception and return -1 if allocation failed */
 363
 364     if (!self->extra)
 365         element_new_extra(self, NULL);
 366
 367     size = self->extra->length + extra;
 368
 369     if (size > self->extra->allocated) {
 370         /* use Python 2.4's list growth strategy */
 371         size = (size >> 3) + (size < 9 ? 3 : 6) + size;
 372         if (self->extra->children != self->extra->_children) {
 373             children = PyObject_Realloc(self->extra->children,
 374                                         size * sizeof(PyObject*));
 375             if (!children)
 376                 goto nomemory;
 377         } else {
 378             children = PyObject_Malloc(size * sizeof(PyObject*));
 379             if (!children)
 380                 goto nomemory;
 381             /* copy existing children from static area to malloc buffer */
 382             memcpy(children, self->extra->children,
 383                    self->extra->length * sizeof(PyObject*));
 384         }
 385         self->extra->children = children;
 386         self->extra->allocated = size;
 387     }
 388
 389     return 0;
 390
 391   nomemory:
 392     PyErr_NoMemory();
 393     return -1;
 394 }
 395
 396 LOCAL(int)
 397 element_add_subelement(ElementObject* self, PyObject* element)
 398 {
 399     /* add a child element to a parent */
 400
 401     if (element_resize(self, 1) < 0)
 402         return -1;
 403
 404     Py_INCREF(element);
 405     self->extra->children[self->extra->length] = element;
 406
 407     self->extra->length++;
 408
 409     return 0;
 410 }
 411
 412 LOCAL(PyObject*)
 413 element_get_attrib(ElementObject* self)
 414 {
 415     /* return borrowed reference to attrib dictionary */
 416     /* note: this function assumes that the extra section exists */
 417
 418     PyObject* res = self->extra->attrib;
 419
 420     if (res == Py_None) {
 421         /* create missing dictionary */
 422         res = PyDict_New();
 423         if (!res)
 424             return NULL;
 425         self->extra->attrib = res;
 426     }
 427
 428     return res;
 429 }
 430
 431 LOCAL(PyObject*)
 432 element_get_text(ElementObject* self)
 433 {
 434     /* return borrowed reference to text attribute */
 435
 436     PyObject* res = self->text;
 437
 438     if (JOIN_GET(res)) {
 439         res = JOIN_OBJ(res);
 440         if (PyList_CheckExact(res)) {
 441             res = list_join(res);
 442             if (!res)
 443                 return NULL;
 444             self->text = res;
 445         }
 446     }
 447
 448     return res;
 449 }
 450
 451 LOCAL(PyObject*)
 452 element_get_tail(ElementObject* self)
 453 {
 454     /* return borrowed reference to text attribute */
 455
 456     PyObject* res = self->tail;
 457
 458     if (JOIN_GET(res)) {
 459         res = JOIN_OBJ(res);
 460         if (PyList_CheckExact(res)) {
 461             res = list_join(res);
 462             if (!res)
 463                 return NULL;
 464             self->tail = res;
 465         }
 466     }
 467
 468     return res;
 469 }
 470
 471 static PyObject*
 472 element(PyObject* self, PyObject* args, PyObject* kw)
 473 {
 474     PyObject* elem;
 475
 476     PyObject* tag;
 477     PyObject* attrib = NULL;
 478     if (!PyArg_ParseTuple(args, "O|O!:Element", &tag,
 479                           &PyDict_Type, &attrib))
 480         return NULL;
 481
 482     if (attrib || kw) {
 483         attrib = (attrib) ? PyDict_Copy(attrib) : PyDict_New();
 484         if (!attrib)
 485             return NULL;
 486         if (kw)
 487             PyDict_Update(attrib, kw);
 488     } else {
 489         Py_INCREF(Py_None);
 490         attrib = Py_None;
 491     }
 492
 493     elem = element_new(tag, attrib);
 494
 495     Py_DECREF(attrib);
 496
 497     return elem;
 498 }
 499
 500 static PyObject*
 501 subelement(PyObject* self, PyObject* args, PyObject* kw)
 502 {
 503     PyObject* elem;
 504
 505     ElementObject* parent;
 506     PyObject* tag;
 507     PyObject* attrib = NULL;
 508     if (!PyArg_ParseTuple(args, "O!O|O!:SubElement",
 509                           &Element_Type, &parent, &tag,
 510                           &PyDict_Type, &attrib))
 511         return NULL;
 512
 513     if (attrib || kw) {
 514         attrib = (attrib) ? PyDict_Copy(attrib) : PyDict_New();
 515         if (!attrib)
 516             return NULL;
 517         if (kw)
 518             PyDict_Update(attrib, kw);
 519     } else {
 520         Py_INCREF(Py_None);
 521         attrib = Py_None;
 522     }
 523
 524     elem = element_new(tag, attrib);
 525
 526     Py_DECREF(attrib);
 527
 528     if (element_add_subelement(parent, elem) < 0) {
 529         Py_DECREF(elem);
 530         return NULL;
 531     }
 532
 533     return elem;
 534 }
 535
 536 static void
 537 element_dealloc(ElementObject* self)
 538 {
 539     if (self->extra)
 540         element_dealloc_extra(self);
 541
 542     /* discard attributes */
 543     Py_DECREF(self->tag);
 544     Py_DECREF(JOIN_OBJ(self->text));
 545     Py_DECREF(JOIN_OBJ(self->tail));
 546
 547     RELEASE(sizeof(ElementObject), "destroy element");
 548
 549     PyObject_Del(self);
 550 }
 551
 552 /* -------------------------------------------------------------------- */
 553 /* methods (in alphabetical order) */
 554
 555 static PyObject*
 556 element_append(ElementObject* self, PyObject* args)
 557 {
 558     PyObject* element;
 559     if (!PyArg_ParseTuple(args, "O!:append", &Element_Type, &element))
 560         return NULL;
 561
 562     if (element_add_subelement(self, element) < 0)
 563         return NULL;
 564
 565     Py_RETURN_NONE;
 566 }
 567
 568 static PyObject*
 569 element_clear(ElementObject* self, PyObject* args)
 570 {
 571     if (!PyArg_ParseTuple(args, ":clear"))
 572         return NULL;
 573
 574     if (self->extra) {
 575         element_dealloc_extra(self);
 576         self->extra = NULL;
 577     }
 578
 579     Py_INCREF(Py_None);
 580     Py_DECREF(JOIN_OBJ(self->text));
 581     self->text = Py_None;
 582
 583     Py_INCREF(Py_None);
 584     Py_DECREF(JOIN_OBJ(self->tail));
 585     self->tail = Py_None;
 586
 587     Py_RETURN_NONE;
 588 }
 589
 590 static PyObject*
 591 element_copy(ElementObject* self, PyObject* args)
 592 {
 593     int i;
 594     ElementObject* element;
 595
 596     if (!PyArg_ParseTuple(args, ":__copy__"))
 597         return NULL;
 598
 599     element = (ElementObject*) element_new(
 600         self->tag, (self->extra) ? self->extra->attrib : Py_None
 601         );
 602     if (!element)
 603         return NULL;
 604
 605     Py_DECREF(JOIN_OBJ(element->text));
 606     element->text = self->text;
 607     Py_INCREF(JOIN_OBJ(element->text));
 608
 609     Py_DECREF(JOIN_OBJ(element->tail));
 610     element->tail = self->tail;
 611     Py_INCREF(JOIN_OBJ(element->tail));
 612
 613     if (self->extra) {
 614
 615         if (element_resize(element, self->extra->length) < 0) {
 616             Py_DECREF(element);
 617             return NULL;
 618         }
 619
 620         for (i = 0; i < self->extra->length; i++) {
 621             Py_INCREF(self->extra->children[i]);
 622             element->extra->children[i] = self->extra->children[i];
 623         }
 624
 625         element->extra->length = self->extra->length;
 626
 627     }
 628
 629     return (PyObject*) element;
 630 }
 631
 632 static PyObject*
 633 element_deepcopy(ElementObject* self, PyObject* args)
 634 {
 635     int i;
 636     ElementObject* element;
 637     PyObject* tag;
 638     PyObject* attrib;
 639     PyObject* text;
 640     PyObject* tail;
 641     PyObject* id;
 642
 643     PyObject* memo;
 644     if (!PyArg_ParseTuple(args, "O:__deepcopy__", &memo))
 645         return NULL;
 646
 647     tag = deepcopy(self->tag, memo);
 648     if (!tag)
 649         return NULL;
 650
 651     if (self->extra) {
 652         attrib = deepcopy(self->extra->attrib, memo);
 653         if (!attrib) {
 654             Py_DECREF(tag);
 655             return NULL;
 656         }
 657     } else {
 658         Py_INCREF(Py_None);
 659         attrib = Py_None;
 660     }
 661
 662     element = (ElementObject*) element_new(tag, attrib);
 663
 664     Py_DECREF(tag);
 665     Py_DECREF(attrib);
 666
 667     if (!element)
 668         return NULL;
 669
 670     text = deepcopy(JOIN_OBJ(self->text), memo);
 671     if (!text)
 672         goto error;
 673     Py_DECREF(element->text);
 674     element->text = JOIN_SET(text, JOIN_GET(self->text));
 675
 676     tail = deepcopy(JOIN_OBJ(self->tail), memo);
 677     if (!tail)
 678         goto error;
 679     Py_DECREF(element->tail);
 680     element->tail = JOIN_SET(tail, JOIN_GET(self->tail));
 681
 682     if (self->extra) {
 683
 684         if (element_resize(element, self->extra->length) < 0)
 685             goto error;
 686
 687         for (i = 0; i < self->extra->length; i++) {
 688             PyObject* child = deepcopy(self->extra->children[i], memo);
 689             if (!child) {
 690                 element->extra->length = i;
 691                 goto error;
 692             }
 693             element->extra->children[i] = child;
 694         }
 695
 696         element->extra->length = self->extra->length;
 697
 698     }
 699
 700     /* add object to memo dictionary (so deepcopy won't visit it again) */
 701     id = PyInt_FromLong((Py_uintptr_t) self);
 702
 703     i = PyDict_SetItem(memo, id, (PyObject*) element);
 704
 705     Py_DECREF(id);
 706
 707     if (i < 0)
 708         goto error;
 709
 710     return (PyObject*) element;
 711
 712   error:
 713     Py_DECREF(element);
 714     return NULL;
 715 }
 716
 717 LOCAL(int)
 718 checkpath(PyObject* tag)
 719 {
 720     Py_ssize_t i;
 721     int check = 1;
 722
 723     /* check if a tag contains an xpath character */
 724
 725 #define PATHCHAR(ch) (ch == '/' || ch == '*' || ch == '[' || ch == '@')
 726
 727 #if defined(Py_USING_UNICODE)
 728     if (PyUnicode_Check(tag)) {
 729         Py_UNICODE *p = PyUnicode_AS_UNICODE(tag);
 730         for (i = 0; i < PyUnicode_GET_SIZE(tag); i++) {
 731             if (p[i] == '{')
 732                 check = 0;
 733             else if (p[i] == '}')
 734                 check = 1;
 735             else if (check && PATHCHAR(p[i]))
 736                 return 1;
 737         }
 738         return 0;
 739     }
 740 #endif
 741     if (PyString_Check(tag)) {
 742         char *p = PyString_AS_STRING(tag);
 743         for (i = 0; i < PyString_GET_SIZE(tag); i++) {
 744             if (p[i] == '{')
 745                 check = 0;
 746             else if (p[i] == '}')
 747                 check = 1;
 748             else if (check && PATHCHAR(p[i]))
 749                 return 1;
 750         }
 751         return 0;
 752     }
 753
 754     return 1; /* unknown type; might be path expression */
 755 }
 756
 757 static PyObject*
 758 element_find(ElementObject* self, PyObject* args)
 759 {
 760     int i;
 761
 762     PyObject* tag;
 763     if (!PyArg_ParseTuple(args, "O:find", &tag))
 764         return NULL;
 765
 766     if (checkpath(tag))
 767         return PyObject_CallMethod(
 768             elementpath_obj, "find", "OO", self, tag
 769             );
 770
 771     if (!self->extra)
 772         Py_RETURN_NONE;
 773
 774     for (i = 0; i < self->extra->length; i++) {
 775         PyObject* item = self->extra->children[i];
 776         if (Element_CheckExact(item) &&
 777             PyObject_Compare(((ElementObject*)item)->tag, tag) == 0) {
 778             Py_INCREF(item);
 779             return item;
 780         }
 781     }
 782
 783     Py_RETURN_NONE;
 784 }
 785
 786 static PyObject*
 787 element_findtext(ElementObject* self, PyObject* args)
 788 {
 789     int i;
 790
 791     PyObject* tag;
 792     PyObject* default_value = Py_None;
 793     if (!PyArg_ParseTuple(args, "O|O:findtext", &tag, &default_value))
 794         return NULL;
 795
 796     if (checkpath(tag))
 797         return PyObject_CallMethod(
 798             elementpath_obj, "findtext", "OOO", self, tag, default_value
 799             );
 800
 801     if (!self->extra) {
 802         Py_INCREF(default_value);
 803         return default_value;
 804     }
 805
 806     for (i = 0; i < self->extra->length; i++) {
 807         ElementObject* item = (ElementObject*) self->extra->children[i];
 808         if (Element_CheckExact(item) && !PyObject_Compare(item->tag, tag)) {
 809             PyObject* text = element_get_text(item);
 810             if (text == Py_None)
 811                 return PyString_FromString("");
 812             Py_XINCREF(text);
 813             return text;
 814         }
 815     }
 816
 817     Py_INCREF(default_value);
 818     return default_value;
 819 }
 820
 821 static PyObject*
 822 element_findall(ElementObject* self, PyObject* args)
 823 {
 824     int i;
 825     PyObject* out;
 826
 827     PyObject* tag;
 828     if (!PyArg_ParseTuple(args, "O:findall", &tag))
 829         return NULL;
 830
 831     if (checkpath(tag))
 832         return PyObject_CallMethod(
 833             elementpath_obj, "findall", "OO", self, tag
 834             );
 835
 836     out = PyList_New(0);
 837     if (!out)
 838         return NULL;
 839
 840     if (!self->extra)
 841         return out;
 842
 843     for (i = 0; i < self->extra->length; i++) {
 844         PyObject* item = self->extra->children[i];
 845         if (Element_CheckExact(item) &&
 846             PyObject_Compare(((ElementObject*)item)->tag, tag) == 0) {
 847             if (PyList_Append(out, item) < 0) {
 848                 Py_DECREF(out);
 849                 return NULL;
 850             }
 851         }
 852     }
 853
 854     return out;
 855 }
 856
 857 static PyObject*
 858 element_get(ElementObject* self, PyObject* args)
 859 {
 860     PyObject* value;
 861
 862     PyObject* key;
 863     PyObject* default_value = Py_None;
 864     if (!PyArg_ParseTuple(args, "O|O:get", &key, &default_value))
 865         return NULL;
 866
 867     if (!self->extra || self->extra->attrib == Py_None)
 868         value = default_value;
 869     else {
 870         value = PyDict_GetItem(self->extra->attrib, key);
 871         if (!value)
 872             value = default_value;
 873     }
 874
 875     Py_INCREF(value);
 876     return value;
 877 }
 878
 879 static PyObject*
 880 element_getchildren(ElementObject* self, PyObject* args)
 881 {
 882     int i;
 883     PyObject* list;
 884
 885     if (!PyArg_ParseTuple(args, ":getchildren"))
 886         return NULL;
 887
 888     if (!self->extra)
 889         return PyList_New(0);
 890
 891     list = PyList_New(self->extra->length);
 892     if (!list)
 893         return NULL;
 894
 895     for (i = 0; i < self->extra->length; i++) {
 896         PyObject* item = self->extra->children[i];
 897         Py_INCREF(item);
 898         PyList_SET_ITEM(list, i, item);
 899     }
 900
 901     return list;
 902 }
 903
 904 static PyObject*
 905 element_getiterator(ElementObject* self, PyObject* args)
 906 {
 907     PyObject* result;
 908
 909     PyObject* tag = Py_None;
 910     if (!PyArg_ParseTuple(args, "|O:getiterator", &tag))
 911         return NULL;
 912
 913     if (!elementtree_getiterator_obj) {
 914         PyErr_SetString(
 915             PyExc_RuntimeError,
 916             "getiterator helper not found"
 917             );
 918         return NULL;
 919     }
 920
 921     args = PyTuple_New(2);
 922     if (!args)
 923         return NULL;
 924
 925     Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self);
 926     Py_INCREF(tag);  PyTuple_SET_ITEM(args, 1, (PyObject*) tag);
 927
 928     result = PyObject_CallObject(elementtree_getiterator_obj, args);
 929
 930     Py_DECREF(args);
 931
 932     return result;
 933 }
 934
 935 static PyObject*
 936 element_getitem(PyObject* self_, Py_ssize_t index)
 937 {
 938     ElementObject* self = (ElementObject*) self_;
 939
 940     if (!self->extra || index < 0 || index >= self->extra->length) {
 941         PyErr_SetString(
 942             PyExc_IndexError,
 943             "child index out of range"
 944             );
 945         return NULL;
 946     }
 947
 948     Py_INCREF(self->extra->children[index]);
 949     return self->extra->children[index];
 950 }
 951
 952 static PyObject*
 953 element_getslice(PyObject* self_, Py_ssize_t start, Py_ssize_t end)
 954 {
 955     ElementObject* self = (ElementObject*) self_;
 956     Py_ssize_t i;
 957     PyObject* list;
 958
 959     if (!self->extra)
 960         return PyList_New(0);
 961
 962     /* standard clamping */
 963     if (start < 0)
 964         start = 0;
 965     if (end < 0)
 966         end = 0;
 967     if (end > self->extra->length)
 968         end = self->extra->length;
 969     if (start > end)
 970         start = end;
 971
 972     list = PyList_New(end - start);
 973     if (!list)
 974         return NULL;
 975
 976     for (i = start; i < end; i++) {
 977         PyObject* item = self->extra->children[i];
 978         Py_INCREF(item);
 979         PyList_SET_ITEM(list, i - start, item);
 980     }
 981
 982     return list;
 983 }
 984
 985 static PyObject*
 986 element_insert(ElementObject* self, PyObject* args)
 987 {
 988     int i;
 989
 990     int index;
 991     PyObject* element;
 992     if (!PyArg_ParseTuple(args, "iO!:insert", &index,
 993                           &Element_Type, &element))
 994         return NULL;
 995
 996     if (!self->extra)
 997         element_new_extra(self, NULL);
 998
 999     if (index < 0)
1000         index = 0;
1001     if (index > self->extra->length)
1002         index = self->extra->length;
1003
1004     if (element_resize(self, 1) < 0)
1005         return NULL;
1006
1007     for (i = self->extra->length; i > index; i--)
1008         self->extra->children[i] = self->extra->children[i-1];
1009
1010     Py_INCREF(element);
1011     self->extra->children[index] = element;
1012
1013     self->extra->length++;
1014
1015     Py_RETURN_NONE;
1016 }
1017
1018 static PyObject*
1019 element_items(ElementObject* self, PyObject* args)
1020 {
1021     if (!PyArg_ParseTuple(args, ":items"))
1022         return NULL;
1023
1024     if (!self->extra || self->extra->attrib == Py_None)
1025         return PyList_New(0);
1026
1027     return PyDict_Items(self->extra->attrib);
1028 }
1029
1030 static PyObject*
1031 element_keys(ElementObject* self, PyObject* args)
1032 {
1033     if (!PyArg_ParseTuple(args, ":keys"))
1034         return NULL;
1035
1036     if (!self->extra || self->extra->attrib == Py_None)
1037         return PyList_New(0);
1038
1039     return PyDict_Keys(self->extra->attrib);
1040 }
1041
1042 static Py_ssize_t
1043 element_length(ElementObject* self)
1044 {
1045     if (!self->extra)
1046         return 0;
1047
1048     return self->extra->length;
1049 }
1050
1051 static PyObject*
1052 element_makeelement(PyObject* self, PyObject* args, PyObject* kw)
1053 {
1054     PyObject* elem;
1055
1056     PyObject* tag;
1057     PyObject* attrib;
1058     if (!PyArg_ParseTuple(args, "OO:makeelement", &tag, &attrib))
1059         return NULL;
1060
1061     attrib = PyDict_Copy(attrib);
1062     if (!attrib)
1063         return NULL;
1064
1065     elem = element_new(tag, attrib);
1066
1067     Py_DECREF(attrib);
1068
1069     return elem;
1070 }
1071
1072 static PyObject*
1073 element_reduce(ElementObject* self, PyObject* args)
1074 {
1075     if (!PyArg_ParseTuple(args, ":__reduce__"))
1076         return NULL;
1077
1078     /* Hack alert: This method is used to work around a __copy__
1079        problem on certain 2.3 and 2.4 versions.  To save time and
1080        simplify the code, we create the copy in here, and use a dummy
1081        copyelement helper to trick the copy module into doing the
1082        right thing. */
1083
1084     if (!elementtree_copyelement_obj) {
1085         PyErr_SetString(
1086             PyExc_RuntimeError,
1087             "copyelement helper not found"
1088             );
1089         return NULL;
1090     }
1091
1092     return Py_BuildValue(
1093         "O(N)", elementtree_copyelement_obj, element_copy(self, args)
1094         );
1095 }
1096
1097 static PyObject*
1098 element_remove(ElementObject* self, PyObject* args)
1099 {
1100     int i;
1101
1102     PyObject* element;
1103     if (!PyArg_ParseTuple(args, "O!:remove", &Element_Type, &element))
1104         return NULL;
1105
1106     if (!self->extra) {
1107         /* element has no children, so raise exception */
1108         PyErr_SetString(
1109             PyExc_ValueError,
1110             "list.remove(x): x not in list"
1111             );
1112         return NULL;
1113     }
1114
1115     for (i = 0; i < self->extra->length; i++) {
1116         if (self->extra->children[i] == element)
1117             break;
1118         if (PyObject_Compare(self->extra->children[i], element) == 0)
1119             break;
1120     }
1121
1122     if (i == self->extra->length) {
1123         /* element is not in children, so raise exception */
1124         PyErr_SetString(
1125             PyExc_ValueError,
1126             "list.remove(x): x not in list"
1127             );
1128         return NULL;
1129     }
1130
1131     Py_DECREF(self->extra->children[i]);
1132
1133     self->extra->length--;
1134
1135     for (; i < self->extra->length; i++)
1136         self->extra->children[i] = self->extra->children[i+1];
1137
1138     Py_RETURN_NONE;
1139 }
1140
1141 static PyObject*
1142 element_repr(ElementObject* self)
1143 {
1144     PyObject* repr;
1145     char buffer[100];
1146
1147     repr = PyString_FromString("<Element ");
1148
1149     PyString_ConcatAndDel(&repr, PyObject_Repr(self->tag));
1150
1151     sprintf(buffer, " at %p>", self);
1152     PyString_ConcatAndDel(&repr, PyString_FromString(buffer));
1153
1154     return repr;
1155 }
1156
1157 static PyObject*
1158 element_set(ElementObject* self, PyObject* args)
1159 {
1160     PyObject* attrib;
1161
1162     PyObject* key;
1163     PyObject* value;
1164     if (!PyArg_ParseTuple(args, "OO:set", &key, &value))
1165         return NULL;
1166
1167     if (!self->extra)
1168         element_new_extra(self, NULL);
1169
1170     attrib = element_get_attrib(self);
1171     if (!attrib)
1172         return NULL;
1173
1174     if (PyDict_SetItem(attrib, key, value) < 0)
1175         return NULL;
1176
1177     Py_RETURN_NONE;
1178 }
1179
1180 static int
1181 element_setslice(PyObject* self_, Py_ssize_t start, Py_ssize_t end, PyObject* item)
1182 {
1183     ElementObject* self = (ElementObject*) self_;
1184     Py_ssize_t i, new, old;
1185     PyObject* recycle = NULL;
1186
1187     if (!self->extra)
1188         element_new_extra(self, NULL);
1189
1190     /* standard clamping */
1191     if (start < 0)
1192         start = 0;
1193     if (end < 0)
1194         end = 0;
1195     if (end > self->extra->length)
1196         end = self->extra->length;
1197     if (start > end)
1198         start = end;
1199
1200     old = end - start;
1201
1202     if (item == NULL)
1203         new = 0;
1204     else if (PyList_CheckExact(item)) {
1205         new = PyList_GET_SIZE(item);
1206     } else {
1207         /* FIXME: support arbitrary sequences? */
1208         PyErr_Format(
1209             PyExc_TypeError,
1210             "expected list, not \"%.200s\"", item->ob_type->tp_name
1211             );
1212         return -1;
1213     }
1214
1215     if (old > 0) {
1216         /* to avoid recursive calls to this method (via decref), move
1217            old items to the recycle bin here, and get rid of them when
1218            we're done modifying the element */
1219         recycle = PyList_New(old);
1220         for (i = 0; i < old; i++)
1221             PyList_SET_ITEM(recycle, i, self->extra->children[i + start]);
1222     }
1223
1224     if (new < old) {
1225         /* delete slice */
1226         for (i = end; i < self->extra->length; i++)
1227             self->extra->children[i + new - old] = self->extra->children[i];
1228     } else if (new > old) {
1229         /* insert slice */
1230         if (element_resize(self, new - old) < 0)
1231             return -1;
1232         for (i = self->extra->length-1; i >= end; i--)
1233             self->extra->children[i + new - old] = self->extra->children[i];
1234     }
1235
1236     /* replace the slice */
1237     for (i = 0; i < new; i++) {
1238         PyObject* element = PyList_GET_ITEM(item, i);
1239         Py_INCREF(element);
1240         self->extra->children[i + start] = element;
1241     }
1242
1243     self->extra->length += new - old;
1244
1245     /* discard the recycle bin, and everything in it */
1246     Py_XDECREF(recycle);
1247
1248     return 0;
1249 }
1250
1251 static int
1252 element_setitem(PyObject* self_, Py_ssize_t index, PyObject* item)
1253 {
1254     ElementObject* self = (ElementObject*) self_;
1255     int i;
1256     PyObject* old;
1257
1258     if (!self->extra || index < 0 || index >= self->extra->length) {
1259         PyErr_SetString(
1260             PyExc_IndexError,
1261             "child assignment index out of range");
1262         return -1;
1263     }
1264
1265     old = self->extra->children[index];
1266
1267     if (item) {
1268         Py_INCREF(item);
1269         self->extra->children[index] = item;
1270     } else {
1271         self->extra->length--;
1272         for (i = index; i < self->extra->length; i++)
1273             self->extra->children[i] = self->extra->children[i+1];
1274     }
1275
1276     Py_DECREF(old);
1277
1278     return 0;
1279 }
1280
1281 static PyMethodDef element_methods[] = {
1282
1283     {"clear", (PyCFunction) element_clear, METH_VARARGS},
1284
1285     {"get", (PyCFunction) element_get, METH_VARARGS},
1286     {"set", (PyCFunction) element_set, METH_VARARGS},
1287
1288     {"find", (PyCFunction) element_find, METH_VARARGS},
1289     {"findtext", (PyCFunction) element_findtext, METH_VARARGS},
1290     {"findall", (PyCFunction) element_findall, METH_VARARGS},
1291
1292     {"append", (PyCFunction) element_append, METH_VARARGS},
1293     {"insert", (PyCFunction) element_insert, METH_VARARGS},
1294     {"remove", (PyCFunction) element_remove, METH_VARARGS},
1295
1296     {"getiterator", (PyCFunction) element_getiterator, METH_VARARGS},
1297     {"getchildren", (PyCFunction) element_getchildren, METH_VARARGS},
1298
1299     {"items", (PyCFunction) element_items, METH_VARARGS},
1300     {"keys", (PyCFunction) element_keys, METH_VARARGS},
1301
1302     {"makeelement", (PyCFunction) element_makeelement, METH_VARARGS},
1303
1304     {"__copy__", (PyCFunction) element_copy, METH_VARARGS},
1305     {"__deepcopy__", (PyCFunction) element_deepcopy, METH_VARARGS},
1306
1307     /* Some 2.3 and 2.4 versions do not handle the __copy__ method on
1308        C objects correctly, so we have to fake it using a __reduce__-
1309        based hack (see the element_reduce implementation above for
1310        details). */
1311
1312     /* The behaviour has been changed in 2.3.5 and 2.4.1, so we're
1313        using a runtime test to figure out if we need to fake things
1314        or now (see the init code below).  The following entry is
1315        enabled only if the hack is needed. */
1316
1317     {"!__reduce__", (PyCFunction) element_reduce, METH_VARARGS},
1318
1319     {NULL, NULL}
1320 };
1321
1322 static PyObject*
1323 element_getattr(ElementObject* self, char* name)
1324 {
1325     PyObject* res;
1326
1327     res = Py_FindMethod(element_methods, (PyObject*) self, name);
1328     if (res)
1329         return res;
1330
1331     PyErr_Clear();
1332
1333     if (strcmp(name, "tag") == 0)
1334         res = self->tag;
1335     else if (strcmp(name, "text") == 0)
1336         res = element_get_text(self);
1337     else if (strcmp(name, "tail") == 0) {
1338         res = element_get_tail(self);
1339     } else if (strcmp(name, "attrib") == 0) {
1340         if (!self->extra)
1341             element_new_extra(self, NULL);
1342         res = element_get_attrib(self);
1343     } else {
1344         PyErr_SetString(PyExc_AttributeError, name);
1345         return NULL;
1346     }
1347
1348     if (!res)
1349         return NULL;
1350
1351     Py_INCREF(res);
1352     return res;
1353 }
1354
1355 static int
1356 element_setattr(ElementObject* self, const char* name, PyObject* value)
1357 {
1358     if (value == NULL) {
1359         PyErr_SetString(
1360             PyExc_AttributeError,
1361             "can't delete element attributes"
1362             );
1363         return -1;
1364     }
1365
1366     if (strcmp(name, "tag") == 0) {
1367         Py_DECREF(self->tag);
1368         self->tag = value;
1369         Py_INCREF(self->tag);
1370     } else if (strcmp(name, "text") == 0) {
1371         Py_DECREF(JOIN_OBJ(self->text));
1372         self->text = value;
1373         Py_INCREF(self->text);
1374     } else if (strcmp(name, "tail") == 0) {
1375         Py_DECREF(JOIN_OBJ(self->tail));
1376         self->tail = value;
1377         Py_INCREF(self->tail);
1378     } else if (strcmp(name, "attrib") == 0) {
1379         if (!self->extra)
1380             element_new_extra(self, NULL);
1381         Py_DECREF(self->extra->attrib);
1382         self->extra->attrib = value;
1383         Py_INCREF(self->extra->attrib);
1384     } else {
1385         PyErr_SetString(PyExc_AttributeError, name);
1386         return -1;
1387     }
1388
1389     return 0;
1390 }
1391
1392 static PySequenceMethods element_as_sequence = {
1393     (lenfunc) element_length,
1394     0, /* sq_concat */
1395     0, /* sq_repeat */
1396     element_getitem,
1397     element_getslice,
1398     element_setitem,
1399     element_setslice,
1400 };
1401
1402 statichere PyTypeObject Element_Type = {
1403     PyObject_HEAD_INIT(NULL)
1404     0, "Element", sizeof(ElementObject), 0,
1405     /* methods */
1406     (destructor)element_dealloc, /* tp_dealloc */
1407     0, /* tp_print */
1408     (getattrfunc)element_getattr, /* tp_getattr */
1409     (setattrfunc)element_setattr, /* tp_setattr */
1410     0, /* tp_compare */
1411     (reprfunc)element_repr, /* tp_repr */
1412     0, /* tp_as_number */
1413     &element_as_sequence, /* tp_as_sequence */
1414 };
1415
1416 /* ==================================================================== */
1417 /* the tree builder type */
1418
1419 typedef struct {
1420     PyObject_HEAD
1421
1422     PyObject* root; /* root node (first created node) */
1423
1424     ElementObject* this; /* current node */
1425     ElementObject* last; /* most recently created node */
1426
1427     PyObject* data; /* data collector (string or list), or NULL */
1428
1429     PyObject* stack; /* element stack */
1430     Py_ssize_t index; /* current stack size (0=empty) */
1431
1432     /* element tracing */
1433     PyObject* events; /* list of events, or NULL if not collecting */
1434     PyObject* start_event_obj; /* event objects (NULL to ignore) */
1435     PyObject* end_event_obj;
1436     PyObject* start_ns_event_obj;
1437     PyObject* end_ns_event_obj;
1438
1439 } TreeBuilderObject;
1440
1441 staticforward PyTypeObject TreeBuilder_Type;
1442
1443 #define TreeBuilder_CheckExact(op) ((op)->ob_type == &TreeBuilder_Type)
1444
1445 /* -------------------------------------------------------------------- */
1446 /* constructor and destructor */
1447
1448 LOCAL(PyObject*)
1449 treebuilder_new(void)
1450 {
1451     TreeBuilderObject* self;
1452
1453     self = PyObject_New(TreeBuilderObject, &TreeBuilder_Type);
1454     if (self == NULL)
1455         return NULL;
1456
1457     self->root = NULL;
1458
1459     Py_INCREF(Py_None);
1460     self->this = (ElementObject*) Py_None;
1461
1462     Py_INCREF(Py_None);
1463     self->last = (ElementObject*) Py_None;
1464
1465     self->data = NULL;
1466
1467     self->stack = PyList_New(20);
1468     self->index = 0;
1469
1470     self->events = NULL;
1471     self->start_event_obj = self->end_event_obj = NULL;
1472     self->start_ns_event_obj = self->end_ns_event_obj = NULL;
1473
1474     ALLOC(sizeof(TreeBuilderObject), "create treebuilder");
1475
1476     return (PyObject*) self;
1477 }
1478
1479 static PyObject*
1480 treebuilder(PyObject* self_, PyObject* args)
1481 {
1482     if (!PyArg_ParseTuple(args, ":TreeBuilder"))
1483         return NULL;
1484
1485     return treebuilder_new();
1486 }
1487
1488 static void
1489 treebuilder_dealloc(TreeBuilderObject* self)
1490 {
1491     Py_XDECREF(self->end_ns_event_obj);
1492     Py_XDECREF(self->start_ns_event_obj);
1493     Py_XDECREF(self->end_event_obj);
1494     Py_XDECREF(self->start_event_obj);
1495     Py_XDECREF(self->events);
1496     Py_DECREF(self->stack);
1497     Py_XDECREF(self->data);
1498     Py_DECREF(self->last);
1499     Py_DECREF(self->this);
1500     Py_XDECREF(self->root);
1501
1502     RELEASE(sizeof(TreeBuilderObject), "destroy treebuilder");
1503
1504     PyObject_Del(self);
1505 }
1506
1507 /* -------------------------------------------------------------------- */
1508 /* handlers */
1509
1510 LOCAL(PyObject*)
1511 treebuilder_handle_xml(TreeBuilderObject* self, PyObject* encoding,
1512                        PyObject* standalone)
1513 {
1514     Py_RETURN_NONE;
1515 }
1516
1517 LOCAL(PyObject*)
1518 treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag,
1519                          PyObject* attrib)
1520 {
1521     PyObject* node;
1522     PyObject* this;
1523
1524     if (self->data) {
1525         if (self->this == self->last) {
1526             Py_DECREF(JOIN_OBJ(self->last->text));
1527             self->last->text = JOIN_SET(
1528                 self->data, PyList_CheckExact(self->data)
1529                 );
1530         } else {
1531             Py_DECREF(JOIN_OBJ(self->last->tail));
1532             self->last->tail = JOIN_SET(
1533                 self->data, PyList_CheckExact(self->data)
1534                 );
1535         }
1536         self->data = NULL;
1537     }
1538
1539     node = element_new(tag, attrib);
1540     if (!node)
1541         return NULL;
1542
1543     this = (PyObject*) self->this;
1544
1545     if (this != Py_None) {
1546         if (element_add_subelement((ElementObject*) this, node) < 0)
1547             goto error;
1548     } else {
1549         if (self->root) {
1550             PyErr_SetString(
1551                 PyExc_SyntaxError,
1552                 "multiple elements on top level"
1553                 );
1554             goto error;
1555         }
1556         Py_INCREF(node);
1557         self->root = node;
1558     }
1559
1560     if (self->index < PyList_GET_SIZE(self->stack)) {
1561         if (PyList_SetItem(self->stack, self->index, this) < 0)
1562             goto error;
1563         Py_INCREF(this);
1564     } else {
1565         if (PyList_Append(self->stack, this) < 0)
1566             goto error;
1567     }
1568     self->index++;
1569
1570     Py_DECREF(this);
1571     Py_INCREF(node);
1572     self->this = (ElementObject*) node;
1573
1574     Py_DECREF(self->last);
1575     Py_INCREF(node);
1576     self->last = (ElementObject*) node;
1577
1578     if (self->start_event_obj) {
1579         PyObject* res;
1580         PyObject* action = self->start_event_obj;
1581         res = PyTuple_New(2);
1582         if (res) {
1583             Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action);
1584             Py_INCREF(node);   PyTuple_SET_ITEM(res, 1, (PyObject*) node);
1585             PyList_Append(self->events, res);
1586             Py_DECREF(res);
1587         } else
1588             PyErr_Clear(); /* FIXME: propagate error */
1589     }
1590
1591     return node;
1592
1593   error:
1594     Py_DECREF(node);
1595     return NULL;
1596 }
1597
1598 LOCAL(PyObject*)
1599 treebuilder_handle_data(TreeBuilderObject* self, PyObject* data)
1600 {
1601     if (!self->data) {
1602         if (self->last == (ElementObject*) Py_None) {
1603             /* ignore calls to data before the first call to start */
1604             Py_RETURN_NONE;
1605         }
1606         /* store the first item as is */
1607         Py_INCREF(data); self->data = data;
1608     } else {
1609         /* more than one item; use a list to collect items */
1610         if (PyString_CheckExact(self->data) && self->data->ob_refcnt == 1 &&
1611             PyString_CheckExact(data) && PyString_GET_SIZE(data) == 1) {
1612             /* expat often generates single character data sections; handle
1613                the most common case by resizing the existing string... */
1614             Py_ssize_t size = PyString_GET_SIZE(self->data);
1615             if (_PyString_Resize(&self->data, size + 1) < 0)
1616                 return NULL;
1617             PyString_AS_STRING(self->data)[size] = PyString_AS_STRING(data)[0];
1618         } else if (PyList_CheckExact(self->data)) {
1619             if (PyList_Append(self->data, data) < 0)
1620                 return NULL;
1621         } else {
1622             PyObject* list = PyList_New(2);
1623             if (!list)
1624                 return NULL;
1625             PyList_SET_ITEM(list, 0, self->data);
1626             Py_INCREF(data); PyList_SET_ITEM(list, 1, data);
1627             self->data = list;
1628         }
1629     }
1630
1631     Py_RETURN_NONE;
1632 }
1633
1634 LOCAL(PyObject*)
1635 treebuilder_handle_end(TreeBuilderObject* self, PyObject* tag)
1636 {
1637     PyObject* item;
1638
1639     if (self->data) {
1640         if (self->this == self->last) {
1641             Py_DECREF(JOIN_OBJ(self->last->text));
1642             self->last->text = JOIN_SET(
1643                 self->data, PyList_CheckExact(self->data)
1644                 );
1645         } else {
1646             Py_DECREF(JOIN_OBJ(self->last->tail));
1647             self->last->tail = JOIN_SET(
1648                 self->data, PyList_CheckExact(self->data)
1649                 );
1650         }
1651         self->data = NULL;
1652     }
1653
1654     if (self->index == 0) {
1655         PyErr_SetString(
1656             PyExc_IndexError,
1657             "pop from empty stack"
1658             );
1659         return NULL;
1660     }
1661
1662     self->index--;
1663
1664     item = PyList_GET_ITEM(self->stack, self->index);
1665     Py_INCREF(item);
1666
1667     Py_DECREF(self->last);
1668
1669     self->last = (ElementObject*) self->this;
1670     self->this = (ElementObject*) item;
1671
1672     if (self->end_event_obj) {
1673         PyObject* res;
1674         PyObject* action = self->end_event_obj;
1675         PyObject* node = (PyObject*) self->last;
1676         res = PyTuple_New(2);
1677         if (res) {
1678             Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action);
1679             Py_INCREF(node);   PyTuple_SET_ITEM(res, 1, (PyObject*) node);
1680             PyList_Append(self->events, res);
1681             Py_DECREF(res);
1682         } else
1683             PyErr_Clear(); /* FIXME: propagate error */
1684     }
1685
1686     Py_INCREF(self->last);
1687     return (PyObject*) self->last;
1688 }
1689
1690 LOCAL(void)
1691 treebuilder_handle_namespace(TreeBuilderObject* self, int start,
1692                              const char* prefix, const char *uri)
1693 {
1694     PyObject* res;
1695     PyObject* action;
1696     PyObject* parcel;
1697
1698     if (!self->events)
1699         return;
1700
1701     if (start) {
1702         if (!self->start_ns_event_obj)
1703             return;
1704         action = self->start_ns_event_obj;
1705         /* FIXME: prefix and uri use utf-8 encoding! */
1706         parcel = Py_BuildValue("ss", (prefix) ? prefix : "", uri);
1707         if (!parcel)
1708             return;
1709         Py_INCREF(action);
1710     } else {
1711         if (!self->end_ns_event_obj)
1712             return;
1713         action = self->end_ns_event_obj;
1714         Py_INCREF(action);
1715         parcel = Py_None;
1716         Py_INCREF(parcel);
1717     }
1718
1719     res = PyTuple_New(2);
1720
1721     if (res) {
1722         PyTuple_SET_ITEM(res, 0, action);
1723         PyTuple_SET_ITEM(res, 1, parcel);
1724         PyList_Append(self->events, res);
1725         Py_DECREF(res);
1726     } else
1727         PyErr_Clear(); /* FIXME: propagate error */
1728 }
1729
1730 /* -------------------------------------------------------------------- */
1731 /* methods (in alphabetical order) */
1732
1733 static PyObject*
1734 treebuilder_data(TreeBuilderObject* self, PyObject* args)
1735 {
1736     PyObject* data;
1737     if (!PyArg_ParseTuple(args, "O:data", &data))
1738         return NULL;
1739
1740     return treebuilder_handle_data(self, data);
1741 }
1742
1743 static PyObject*
1744 treebuilder_end(TreeBuilderObject* self, PyObject* args)
1745 {
1746     PyObject* tag;
1747     if (!PyArg_ParseTuple(args, "O:end", &tag))
1748         return NULL;
1749
1750     return treebuilder_handle_end(self, tag);
1751 }
1752
1753 LOCAL(PyObject*)
1754 treebuilder_done(TreeBuilderObject* self)
1755 {
1756     PyObject* res;
1757
1758     /* FIXME: check stack size? */
1759
1760     if (self->root)
1761         res = self->root;
1762     else
1763         res = Py_None;
1764
1765     Py_INCREF(res);
1766     return res;
1767 }
1768
1769 static PyObject*
1770 treebuilder_close(TreeBuilderObject* self, PyObject* args)
1771 {
1772     if (!PyArg_ParseTuple(args, ":close"))
1773         return NULL;
1774
1775     return treebuilder_done(self);
1776 }
1777
1778 static PyObject*
1779 treebuilder_start(TreeBuilderObject* self, PyObject* args)
1780 {
1781     PyObject* tag;
1782     PyObject* attrib = Py_None;
1783     if (!PyArg_ParseTuple(args, "O|O:start", &tag, &attrib))
1784         return NULL;
1785
1786     return treebuilder_handle_start(self, tag, attrib);
1787 }
1788
1789 static PyObject*
1790 treebuilder_xml(TreeBuilderObject* self, PyObject* args)
1791 {
1792     PyObject* encoding;
1793     PyObject* standalone;
1794     if (!PyArg_ParseTuple(args, "OO:xml", &encoding, &standalone))
1795         return NULL;
1796
1797     return treebuilder_handle_xml(self, encoding, standalone);
1798 }
1799
1800 static PyMethodDef treebuilder_methods[] = {
1801     {"data", (PyCFunction) treebuilder_data, METH_VARARGS},
1802     {"start", (PyCFunction) treebuilder_start, METH_VARARGS},
1803     {"end", (PyCFunction) treebuilder_end, METH_VARARGS},
1804     {"xml", (PyCFunction) treebuilder_xml, METH_VARARGS},
1805     {"close", (PyCFunction) treebuilder_close, METH_VARARGS},
1806     {NULL, NULL}
1807 };
1808
1809 static PyObject*
1810 treebuilder_getattr(TreeBuilderObject* self, char* name)
1811 {
1812     return Py_FindMethod(treebuilder_methods, (PyObject*) self, name);
1813 }
1814
1815 statichere PyTypeObject TreeBuilder_Type = {
1816     PyObject_HEAD_INIT(NULL)
1817     0, "TreeBuilder", sizeof(TreeBuilderObject), 0,
1818     /* methods */
1819     (destructor)treebuilder_dealloc, /* tp_dealloc */
1820     0, /* tp_print */
1821     (getattrfunc)treebuilder_getattr, /* tp_getattr */
1822 };
1823
1824 /* ==================================================================== */
1825 /* the expat interface */
1826
1827 #if defined(USE_EXPAT)
1828
1829 #include "expat.h"
1830
1831 #if defined(USE_PYEXPAT_CAPI)
1832 #include "pyexpat.h"
1833 static struct PyExpat_CAPI* expat_capi;
1834 #define EXPAT(func) (expat_capi->func)
1835 #else
1836 #define EXPAT(func) (XML_##func)
1837 #endif
1838
1839 typedef struct {
1840     PyObject_HEAD
1841
1842     XML_Parser parser;
1843
1844     PyObject* target;
1845     PyObject* entity;
1846
1847     PyObject* names;
1848
1849     PyObject* handle_xml;
1850     PyObject* handle_start;
1851     PyObject* handle_data;
1852     PyObject* handle_end;
1853
1854     PyObject* handle_comment;
1855     PyObject* handle_pi;
1856
1857 } XMLParserObject;
1858
1859 staticforward PyTypeObject XMLParser_Type;
1860
1861 /* helpers */
1862
1863 #if defined(Py_USING_UNICODE)
1864 LOCAL(int)
1865 checkstring(const char* string, int size)
1866 {
1867     int i;
1868
1869     /* check if an 8-bit string contains UTF-8 characters */
1870     for (i = 0; i < size; i++)
1871         if (string[i] & 0x80)
1872             return 1;
1873
1874     return 0;
1875 }
1876 #endif
1877
1878 LOCAL(PyObject*)
1879 makestring(const char* string, int size)
1880 {
1881     /* convert a UTF-8 string to either a 7-bit ascii string or a
1882        Unicode string */
1883
1884 #if defined(Py_USING_UNICODE)
1885     if (checkstring(string, size))
1886         return PyUnicode_DecodeUTF8(string, size, "strict");
1887 #endif
1888
1889     return PyString_FromStringAndSize(string, size);
1890 }
1891
1892 LOCAL(PyObject*)
1893 makeuniversal(XMLParserObject* self, const char* string)
1894 {
1895     /* convert a UTF-8 tag/attribute name from the expat parser
1896        to a universal name string */
1897
1898     int size = strlen(string);
1899     PyObject* key;
1900     PyObject* value;
1901
1902     /* look the 'raw' name up in the names dictionary */
1903     key = PyString_FromStringAndSize(string, size);
1904     if (!key)
1905         return NULL;
1906
1907     value = PyDict_GetItem(self->names, key);
1908
1909     if (value) {
1910         Py_INCREF(value);
1911     } else {
1912         /* new name.  convert to universal name, and decode as
1913            necessary */
1914
1915         PyObject* tag;
1916         char* p;
1917         int i;
1918
1919         /* look for namespace separator */
1920         for (i = 0; i < size; i++)
1921             if (string[i] == '}')
1922                 break;
1923         if (i != size) {
1924             /* convert to universal name */
1925             tag = PyString_FromStringAndSize(NULL, size+1);
1926             p = PyString_AS_STRING(tag);
1927             p[0] = '{';
1928             memcpy(p+1, string, size);
1929             size++;
1930         } else {
1931             /* plain name; use key as tag */
1932             Py_INCREF(key);
1933             tag = key;
1934         }
1935
1936         /* decode universal name */
1937 #if defined(Py_USING_UNICODE)
1938         /* inline makestring, to avoid duplicating the source string if
1939            it's not an utf-8 string */
1940         p = PyString_AS_STRING(tag);
1941         if (checkstring(p, size)) {
1942             value = PyUnicode_DecodeUTF8(p, size, "strict");
1943             Py_DECREF(tag);
1944             if (!value) {
1945                 Py_DECREF(key);
1946                 return NULL;
1947             }
1948         } else
1949 #endif
1950             value = tag; /* use tag as is */
1951
1952         /* add to names dictionary */
1953         if (PyDict_SetItem(self->names, key, value) < 0) {
1954             Py_DECREF(key);
1955             Py_DECREF(value);
1956             return NULL;
1957         }
1958     }
1959
1960     Py_DECREF(key);
1961     return value;
1962 }
1963
1964 /* -------------------------------------------------------------------- */
1965 /* handlers */
1966
1967 static void
1968 expat_default_handler(XMLParserObject* self, const XML_Char* data_in,
1969                       int data_len)
1970 {
1971     PyObject* key;
1972     PyObject* value;
1973     PyObject* res;
1974
1975     if (data_len < 2 || data_in[0] != '&')
1976         return;
1977
1978     key = makestring(data_in + 1, data_len - 2);
1979     if (!key)
1980         return;
1981
1982     value = PyDict_GetItem(self->entity, key);
1983
1984     if (value) {
1985         if (TreeBuilder_CheckExact(self->target))
1986             res = treebuilder_handle_data(
1987                 (TreeBuilderObject*) self->target, value
1988                 );
1989         else if (self->handle_data)
1990             res = PyObject_CallFunction(self->handle_data, "O", value);
1991         else
1992             res = NULL;
1993         Py_XDECREF(res);
1994     } else {
1995         PyErr_Format(
1996             PyExc_SyntaxError, "undefined entity &%s;: line %ld, column %ld",
1997             PyString_AS_STRING(key),
1998             EXPAT(GetErrorLineNumber)(self->parser),
1999             EXPAT(GetErrorColumnNumber)(self->parser)
2000             );
2001     }
2002
2003     Py_DECREF(key);
2004 }
2005
2006 static void
2007 expat_start_handler(XMLParserObject* self, const XML_Char* tag_in,
2008                     const XML_Char **attrib_in)
2009 {
2010     PyObject* res;
2011     PyObject* tag;
2012     PyObject* attrib;
2013     int ok;
2014
2015     /* tag name */
2016     tag = makeuniversal(self, tag_in);
2017     if (!tag)
2018         return; /* parser will look for errors */
2019
2020     /* attributes */
2021     if (attrib_in[0]) {
2022         attrib = PyDict_New();
2023         if (!attrib)
2024             return;
2025         while (attrib_in[0] && attrib_in[1]) {
2026             PyObject* key = makeuniversal(self, attrib_in[0]);
2027             PyObject* value = makestring(attrib_in[1], strlen(attrib_in[1]));
2028             if (!key || !value) {
2029                 Py_XDECREF(value);
2030                 Py_XDECREF(key);
2031                 Py_DECREF(attrib);
2032                 return;
2033             }
2034             ok = PyDict_SetItem(attrib, key, value);
2035             Py_DECREF(value);
2036             Py_DECREF(key);
2037             if (ok < 0) {
2038                 Py_DECREF(attrib);
2039                 return;
2040             }
2041             attrib_in += 2;
2042         }
2043     } else {
2044         Py_INCREF(Py_None);
2045         attrib = Py_None;
2046     }
2047
2048     if (TreeBuilder_CheckExact(self->target))
2049         /* shortcut */
2050         res = treebuilder_handle_start((TreeBuilderObject*) self->target,
2051                                        tag, attrib);
2052     else if (self->handle_start)
2053         res = PyObject_CallFunction(self->handle_start, "OO", tag, attrib);
2054     else
2055         res = NULL;
2056
2057     Py_DECREF(tag);
2058     Py_DECREF(attrib);
2059
2060     Py_XDECREF(res);
2061 }
2062
2063 static void
2064 expat_data_handler(XMLParserObject* self, const XML_Char* data_in,
2065                    int data_len)
2066 {
2067     PyObject* data;
2068     PyObject* res;
2069
2070     data = makestring(data_in, data_len);
2071     if (!data)
2072         return; /* parser will look for errors */
2073
2074     if (TreeBuilder_CheckExact(self->target))
2075         /* shortcut */
2076         res = treebuilder_handle_data((TreeBuilderObject*) self->target, data);
2077     else if (self->handle_data)
2078         res = PyObject_CallFunction(self->handle_data, "O", data);
2079     else
2080         res = NULL;
2081
2082     Py_DECREF(data);
2083
2084     Py_XDECREF(res);
2085 }
2086
2087 static void
2088 expat_end_handler(XMLParserObject* self, const XML_Char* tag_in)
2089 {
2090     PyObject* tag;
2091     PyObject* res = NULL;
2092
2093     if (TreeBuilder_CheckExact(self->target))
2094         /* shortcut */
2095         /* the standard tree builder doesn't look at the end tag */
2096         res = treebuilder_handle_end(
2097             (TreeBuilderObject*) self->target, Py_None
2098             );
2099     else if (self->handle_end) {
2100         tag = makeuniversal(self, tag_in);
2101         if (tag) {
2102             res = PyObject_CallFunction(self->handle_end, "O", tag);
2103             Py_DECREF(tag);
2104         }
2105     }
2106
2107     Py_XDECREF(res);
2108 }
2109
2110 static void
2111 expat_start_ns_handler(XMLParserObject* self, const XML_Char* prefix,
2112                        const XML_Char *uri)
2113 {
2114     treebuilder_handle_namespace(
2115         (TreeBuilderObject*) self->target, 1, prefix, uri
2116         );
2117 }
2118
2119 static void
2120 expat_end_ns_handler(XMLParserObject* self, const XML_Char* prefix_in)
2121 {
2122     treebuilder_handle_namespace(
2123         (TreeBuilderObject*) self->target, 0, NULL, NULL
2124         );
2125 }
2126
2127 static void
2128 expat_comment_handler(XMLParserObject* self, const XML_Char* comment_in)
2129 {
2130     PyObject* comment;
2131     PyObject* res;
2132
2133     if (self->handle_comment) {
2134         comment = makestring(comment_in, strlen(comment_in));
2135         if (comment) {
2136             res = PyObject_CallFunction(self->handle_comment, "O", comment);
2137             Py_XDECREF(res);
2138             Py_DECREF(comment);
2139         }
2140     }
2141 }
2142
2143 static void
2144 expat_pi_handler(XMLParserObject* self, const XML_Char* target_in,
2145                  const XML_Char* data_in)
2146 {
2147     PyObject* target;
2148     PyObject* data;
2149     PyObject* res;
2150
2151     if (self->handle_pi) {
2152         target = makestring(target_in, strlen(target_in));
2153         data = makestring(data_in, strlen(data_in));
2154         if (target && data) {
2155             res = PyObject_CallFunction(self->handle_pi, "OO", target, data);
2156             Py_XDECREF(res);
2157             Py_DECREF(data);
2158             Py_DECREF(target);
2159         } else {
2160             Py_XDECREF(data);
2161             Py_XDECREF(target);
2162         }
2163     }
2164 }
2165
2166 #if defined(Py_USING_UNICODE)
2167 static int
2168 expat_unknown_encoding_handler(XMLParserObject *self, const XML_Char *name,
2169                                XML_Encoding *info)
2170 {
2171     PyObject* u;
2172     Py_UNICODE* p;
2173     unsigned char s[256];
2174     int i;
2175
2176     memset(info, 0, sizeof(XML_Encoding));
2177
2178     for (i = 0; i < 256; i++)
2179         s[i] = i;
2180
2181     u = PyUnicode_Decode((char*) s, 256, name, "replace");
2182     if (!u)
2183         return XML_STATUS_ERROR;
2184
2185     if (PyUnicode_GET_SIZE(u) != 256) {
2186         Py_DECREF(u);
2187         return XML_STATUS_ERROR;
2188     }
2189
2190     p = PyUnicode_AS_UNICODE(u);
2191
2192     for (i = 0; i < 256; i++) {
2193         if (p[i] != Py_UNICODE_REPLACEMENT_CHARACTER)
2194             info->map[i] = p[i];
2195         else
2196             info->map[i] = -1;
2197     }
2198
2199     Py_DECREF(u);
2200
2201     return XML_STATUS_OK;
2202 }
2203 #endif
2204
2205 /* -------------------------------------------------------------------- */
2206 /* constructor and destructor */
2207
2208 static PyObject*
2209 xmlparser(PyObject* self_, PyObject* args, PyObject* kw)
2210 {
2211     XMLParserObject* self;
2212     /* FIXME: does this need to be static? */
2213     static XML_Memory_Handling_Suite memory_handler;
2214
2215     PyObject* target = NULL;
2216     char* encoding = NULL;
2217     static char* kwlist[] = { "target", "encoding", NULL };
2218     if (!PyArg_ParseTupleAndKeywords(args, kw, "|Oz:XMLParser", kwlist,
2219                                      &target, &encoding))
2220         return NULL;
2221
2222 #if defined(USE_PYEXPAT_CAPI)
2223     if (!expat_capi) {
2224         PyErr_SetString(
2225             PyExc_RuntimeError, "cannot load dispatch table from pyexpat"
2226             );
2227         return NULL;
2228     }
2229 #endif
2230
2231     self = PyObject_New(XMLParserObject, &XMLParser_Type);
2232     if (self == NULL)
2233         return NULL;
2234
2235     self->entity = PyDict_New();
2236     if (!self->entity) {
2237         PyObject_Del(self);
2238         return NULL;
2239     }
2240
2241     self->names = PyDict_New();
2242     if (!self->names) {
2243         PyObject_Del(self->entity);
2244         PyObject_Del(self);
2245         return NULL;
2246     }
2247
2248     memory_handler.malloc_fcn = PyObject_Malloc;
2249     memory_handler.realloc_fcn = PyObject_Realloc;
2250     memory_handler.free_fcn = PyObject_Free;
2251
2252     self->parser = EXPAT(ParserCreate_MM)(encoding, &memory_handler, "}");
2253     if (!self->parser) {
2254         PyObject_Del(self->names);
2255         PyObject_Del(self->entity);
2256         PyObject_Del(self);
2257         PyErr_NoMemory();
2258         return NULL;
2259     }
2260
2261     /* setup target handlers */
2262     if (!target) {
2263         target = treebuilder_new();
2264         if (!target) {
2265             EXPAT(ParserFree)(self->parser);
2266             PyObject_Del(self->names);
2267             PyObject_Del(self->entity);
2268             PyObject_Del(self);
2269             return NULL;
2270         }
2271     } else
2272         Py_INCREF(target);
2273     self->target = target;
2274
2275     self->handle_xml = PyObject_GetAttrString(target, "xml");
2276     self->handle_start = PyObject_GetAttrString(target, "start");
2277     self->handle_data = PyObject_GetAttrString(target, "data");
2278     self->handle_end = PyObject_GetAttrString(target, "end");
2279     self->handle_comment = PyObject_GetAttrString(target, "comment");
2280     self->handle_pi = PyObject_GetAttrString(target, "pi");
2281
2282     PyErr_Clear();
2283
2284     /* configure parser */
2285     EXPAT(SetUserData)(self->parser, self);
2286     EXPAT(SetElementHandler)(
2287         self->parser,
2288         (XML_StartElementHandler) expat_start_handler,
2289         (XML_EndElementHandler) expat_end_handler
2290         );
2291     EXPAT(SetDefaultHandlerExpand)(
2292         self->parser,
2293         (XML_DefaultHandler) expat_default_handler
2294         );
2295     EXPAT(SetCharacterDataHandler)(
2296         self->parser,
2297         (XML_CharacterDataHandler) expat_data_handler
2298         );
2299     if (self->handle_comment)
2300         EXPAT(SetCommentHandler)(
2301             self->parser,
2302             (XML_CommentHandler) expat_comment_handler
2303             );
2304     if (self->handle_pi)
2305         EXPAT(SetProcessingInstructionHandler)(
2306             self->parser,
2307             (XML_ProcessingInstructionHandler) expat_pi_handler
2308             );
2309 #if defined(Py_USING_UNICODE)
2310     EXPAT(SetUnknownEncodingHandler)(
2311         self->parser,
2312         (XML_UnknownEncodingHandler) expat_unknown_encoding_handler, NULL
2313         );
2314 #endif
2315
2316     ALLOC(sizeof(XMLParserObject), "create expatparser");
2317
2318     return (PyObject*) self;
2319 }
2320
2321 static void
2322 xmlparser_dealloc(XMLParserObject* self)
2323 {
2324     EXPAT(ParserFree)(self->parser);
2325
2326     Py_XDECREF(self->handle_pi);
2327     Py_XDECREF(self->handle_comment);
2328     Py_XDECREF(self->handle_end);
2329     Py_XDECREF(self->handle_data);
2330     Py_XDECREF(self->handle_start);
2331     Py_XDECREF(self->handle_xml);
2332
2333     Py_DECREF(self->target);
2334     Py_DECREF(self->entity);
2335     Py_DECREF(self->names);
2336
2337     RELEASE(sizeof(XMLParserObject), "destroy expatparser");
2338
2339     PyObject_Del(self);
2340 }
2341
2342 /* -------------------------------------------------------------------- */
2343 /* methods (in alphabetical order) */
2344
2345 LOCAL(PyObject*)
2346 expat_parse(XMLParserObject* self, char* data, int data_len, int final)
2347 {
2348     int ok;
2349
2350     ok = EXPAT(Parse)(self->parser, data, data_len, final);
2351
2352     if (PyErr_Occurred())
2353         return NULL;
2354
2355     if (!ok) {
2356         PyErr_Format(
2357             PyExc_SyntaxError, "%s: line %ld, column %ld",
2358             EXPAT(ErrorString)(EXPAT(GetErrorCode)(self->parser)),
2359             EXPAT(GetErrorLineNumber)(self->parser),
2360             EXPAT(GetErrorColumnNumber)(self->parser)
2361             );
2362         return NULL;
2363     }
2364
2365     Py_RETURN_NONE;
2366 }
2367
2368 static PyObject*
2369 xmlparser_close(XMLParserObject* self, PyObject* args)
2370 {
2371     /* end feeding data to parser */
2372
2373     PyObject* res;
2374     if (!PyArg_ParseTuple(args, ":close"))
2375         return NULL;
2376
2377     res = expat_parse(self, "", 0, 1);
2378
2379     if (res && TreeBuilder_CheckExact(self->target)) {
2380         Py_DECREF(res);
2381         return treebuilder_done((TreeBuilderObject*) self->target);
2382     }
2383
2384     return res;
2385 }
2386
2387 static PyObject*
2388 xmlparser_feed(XMLParserObject* self, PyObject* args)
2389 {
2390     /* feed data to parser */
2391
2392     char* data;
2393     int data_len;
2394     if (!PyArg_ParseTuple(args, "s#:feed", &data, &data_len))
2395         return NULL;
2396
2397     return expat_parse(self, data, data_len, 0);
2398 }
2399
2400 static PyObject*
2401 xmlparser_parse(XMLParserObject* self, PyObject* args)
2402 {
2403     /* (internal) parse until end of input stream */
2404
2405     PyObject* reader;
2406     PyObject* buffer;
2407     PyObject* res;
2408
2409     PyObject* fileobj;
2410     if (!PyArg_ParseTuple(args, "O:_parse", &fileobj))
2411         return NULL;
2412
2413     reader = PyObject_GetAttrString(fileobj, "read");
2414     if (!reader)
2415         return NULL;
2416
2417     /* read from open file object */
2418     for (;;) {
2419
2420         buffer = PyObject_CallFunction(reader, "i", 64*1024);
2421
2422         if (!buffer) {
2423             /* read failed (e.g. due to KeyboardInterrupt) */
2424             Py_DECREF(reader);
2425             return NULL;
2426         }
2427
2428         if (!PyString_CheckExact(buffer) || PyString_GET_SIZE(buffer) == 0) {
2429             Py_DECREF(buffer);
2430             break;
2431         }
2432
2433         res = expat_parse(
2434             self, PyString_AS_STRING(buffer), PyString_GET_SIZE(buffer), 0
2435             );
2436
2437         Py_DECREF(buffer);
2438
2439         if (!res) {
2440             Py_DECREF(reader);
2441             return NULL;
2442         }
2443         Py_DECREF(res);
2444
2445     }
2446
2447     Py_DECREF(reader);
2448
2449     res = expat_parse(self, "", 0, 1);
2450
2451     if (res && TreeBuilder_CheckExact(self->target)) {
2452         Py_DECREF(res);
2453         return treebuilder_done((TreeBuilderObject*) self->target);
2454     }
2455
2456     return res;
2457 }
2458
2459 static PyObject*
2460 xmlparser_setevents(XMLParserObject* self, PyObject* args)
2461 {
2462     /* activate element event reporting */
2463
2464     Py_ssize_t i;
2465     TreeBuilderObject* target;
2466
2467     PyObject* events; /* event collector */
2468     PyObject* event_set = Py_None;
2469     if (!PyArg_ParseTuple(args, "O!|O:_setevents",  &PyList_Type, &events,
2470                           &event_set))
2471         return NULL;
2472
2473     if (!TreeBuilder_CheckExact(self->target)) {
2474         PyErr_SetString(
2475             PyExc_TypeError,
2476             "event handling only supported for cElementTree.Treebuilder "
2477             "targets"
2478             );
2479         return NULL;
2480     }
2481
2482     target = (TreeBuilderObject*) self->target;
2483
2484     Py_INCREF(events);
2485     Py_XDECREF(target->events);
2486     target->events = events;
2487
2488     /* clear out existing events */
2489     Py_XDECREF(target->start_event_obj); target->start_event_obj = NULL;
2490     Py_XDECREF(target->end_event_obj); target->end_event_obj = NULL;
2491     Py_XDECREF(target->start_ns_event_obj); target->start_ns_event_obj = NULL;
2492     Py_XDECREF(target->end_ns_event_obj); target->end_ns_event_obj = NULL;
2493
2494     if (event_set == Py_None) {
2495         /* default is "end" only */
2496         target->end_event_obj = PyString_FromString("end");
2497         Py_RETURN_NONE;
2498     }
2499
2500     if (!PyTuple_Check(event_set)) /* FIXME: handle arbitrary sequences */
2501         goto error;
2502
2503     for (i = 0; i < PyTuple_GET_SIZE(event_set); i++) {
2504         PyObject* item = PyTuple_GET_ITEM(event_set, i);
2505         char* event;
2506         if (!PyString_Check(item))
2507             goto error;
2508         event = PyString_AS_STRING(item);
2509         if (strcmp(event, "start") == 0) {
2510             Py_INCREF(item);
2511             target->start_event_obj = item;
2512         } else if (strcmp(event, "end") == 0) {
2513             Py_INCREF(item);
2514             Py_XDECREF(target->end_event_obj);
2515             target->end_event_obj = item;
2516         } else if (strcmp(event, "start-ns") == 0) {
2517             Py_INCREF(item);
2518             Py_XDECREF(target->start_ns_event_obj);
2519             target->start_ns_event_obj = item;
2520             EXPAT(SetNamespaceDeclHandler)(
2521                 self->parser,
2522                 (XML_StartNamespaceDeclHandler) expat_start_ns_handler,
2523                 (XML_EndNamespaceDeclHandler) expat_end_ns_handler
2524                 );
2525         } else if (strcmp(event, "end-ns") == 0) {
2526             Py_INCREF(item);
2527             Py_XDECREF(target->end_ns_event_obj);
2528             target->end_ns_event_obj = item;
2529             EXPAT(SetNamespaceDeclHandler)(
2530                 self->parser,
2531                 (XML_StartNamespaceDeclHandler) expat_start_ns_handler,
2532                 (XML_EndNamespaceDeclHandler) expat_end_ns_handler
2533                 );
2534         } else {
2535             PyErr_Format(
2536                 PyExc_ValueError,
2537                 "unknown event '%s'", event
2538                 );
2539             return NULL;
2540         }
2541     }
2542
2543     Py_RETURN_NONE;
2544
2545   error:
2546     PyErr_SetString(
2547         PyExc_TypeError,
2548         "invalid event tuple"
2549         );
2550     return NULL;
2551 }
2552
2553 static PyMethodDef xmlparser_methods[] = {
2554     {"feed", (PyCFunction) xmlparser_feed, METH_VARARGS},
2555     {"close", (PyCFunction) xmlparser_close, METH_VARARGS},
2556     {"_parse", (PyCFunction) xmlparser_parse, METH_VARARGS},
2557     {"_setevents", (PyCFunction) xmlparser_setevents, METH_VARARGS},
2558     {NULL, NULL}
2559 };
2560
2561 static PyObject*
2562 xmlparser_getattr(XMLParserObject* self, char* name)
2563 {
2564     PyObject* res;
2565
2566     res = Py_FindMethod(xmlparser_methods, (PyObject*) self, name);
2567     if (res)
2568         return res;
2569
2570     PyErr_Clear();
2571
2572     if (strcmp(name, "entity") == 0)
2573         res = self->entity;
2574     else if (strcmp(name, "target") == 0)
2575         res = self->target;
2576     else if (strcmp(name, "version") == 0) {
2577         char buffer[100];
2578         sprintf(buffer, "Expat %d.%d.%d", XML_MAJOR_VERSION,
2579                 XML_MINOR_VERSION, XML_MICRO_VERSION);
2580         return PyString_FromString(buffer);
2581     } else {
2582         PyErr_SetString(PyExc_AttributeError, name);
2583         return NULL;
2584     }
2585
2586     Py_INCREF(res);
2587     return res;
2588 }
2589
2590 statichere PyTypeObject XMLParser_Type = {
2591     PyObject_HEAD_INIT(NULL)
2592     0, "XMLParser", sizeof(XMLParserObject), 0,
2593     /* methods */
2594     (destructor)xmlparser_dealloc, /* tp_dealloc */
2595     0, /* tp_print */
2596     (getattrfunc)xmlparser_getattr, /* tp_getattr */
2597 };
2598
2599 #endif
2600
2601 /* ==================================================================== */
2602 /* python module interface */
2603
2604 static PyMethodDef _functions[] = {
2605     {"Element", (PyCFunction) element, METH_VARARGS|METH_KEYWORDS},
2606     {"SubElement", (PyCFunction) subelement, METH_VARARGS|METH_KEYWORDS},
2607     {"TreeBuilder", (PyCFunction) treebuilder, METH_VARARGS},
2608 #if defined(USE_EXPAT)
2609     {"XMLParser", (PyCFunction) xmlparser, METH_VARARGS|METH_KEYWORDS},
2610     {"XMLTreeBuilder", (PyCFunction) xmlparser, METH_VARARGS|METH_KEYWORDS},
2611 #endif
2612     {NULL, NULL}
2613 };
2614
2615 DL_EXPORT(void)
2616 init_elementtree(void)
2617 {
2618     PyObject* m;
2619     PyObject* g;
2620     char* bootstrap;
2621 #if defined(USE_PYEXPAT_CAPI)
2622     struct PyExpat_CAPI* capi;
2623 #endif
2624
2625     /* Patch object type */
2626     Element_Type.ob_type = TreeBuilder_Type.ob_type = &PyType_Type;
2627 #if defined(USE_EXPAT)
2628     XMLParser_Type.ob_type = &PyType_Type;
2629 #endif
2630
2631     m = Py_InitModule("_elementtree", _functions);
2632     if (!m)
2633         return;
2634
2635     /* python glue code */
2636
2637     g = PyDict_New();
2638     if (!g)
2639         return;
2640
2641     PyDict_SetItemString(g, "__builtins__", PyEval_GetBuiltins());
2642
2643     bootstrap = (
2644
2645 #if (PY_VERSION_HEX >= 0x02020000 && PY_VERSION_HEX < 0x02030000)
2646         "from __future__ import generators\n" /* enable yield under 2.2 */
2647 #endif
2648
2649         "from copy import copy, deepcopy\n"
2650
2651         "try:\n"
2652         "  from xml.etree import ElementTree\n"
2653         "except ImportError:\n"
2654         "  import ElementTree\n"
2655         "ET = ElementTree\n"
2656         "del ElementTree\n"
2657
2658         "import _elementtree as cElementTree\n"
2659
2660         "try:\n" /* check if copy works as is */
2661         "  copy(cElementTree.Element('x'))\n"
2662         "except:\n"
2663         "  def copyelement(elem):\n"
2664         "    return elem\n"
2665
2666         "def Comment(text=None):\n" /* public */
2667         "  element = cElementTree.Element(ET.Comment)\n"
2668         "  element.text = text\n"
2669         "  return element\n"
2670         "cElementTree.Comment = Comment\n"
2671
2672         "class ElementTree(ET.ElementTree):\n" /* public */
2673         "  def parse(self, source, parser=None):\n"
2674         "    if not hasattr(source, 'read'):\n"
2675         "      source = open(source, 'rb')\n"
2676         "    if parser is not None:\n"
2677         "      while 1:\n"
2678         "        data = source.read(65536)\n"
2679         "        if not data:\n"
2680         "          break\n"
2681         "        parser.feed(data)\n"
2682         "      self._root = parser.close()\n"
2683         "    else:\n"
2684         "      parser = cElementTree.XMLParser()\n"
2685         "      self._root = parser._parse(source)\n"
2686         "    return self._root\n"
2687         "cElementTree.ElementTree = ElementTree\n"
2688
2689         "def getiterator(node, tag=None):\n" /* helper */
2690         "  if tag == '*':\n"
2691         "    tag = None\n"
2692 #if (PY_VERSION_HEX < 0x02020000)
2693         "  nodes = []\n" /* 2.1 doesn't have yield */
2694         "  if tag is None or node.tag == tag:\n"
2695         "    nodes.append(node)\n"
2696         "  for node in node:\n"
2697         "    nodes.extend(getiterator(node, tag))\n"
2698         "  return nodes\n"
2699 #else
2700         "  if tag is None or node.tag == tag:\n"
2701         "    yield node\n"
2702         "  for node in node:\n"
2703         "    for node in getiterator(node, tag):\n"
2704         "      yield node\n"
2705 #endif
2706
2707         "def parse(source, parser=None):\n" /* public */
2708         "  tree = ElementTree()\n"
2709         "  tree.parse(source, parser)\n"
2710         "  return tree\n"
2711         "cElementTree.parse = parse\n"
2712
2713 #if (PY_VERSION_HEX < 0x02020000)
2714         "if hasattr(ET, 'iterparse'):\n"
2715         "    cElementTree.iterparse = ET.iterparse\n" /* delegate on 2.1 */
2716 #else
2717         "class iterparse(object):\n"
2718         " root = None\n"
2719         " def __init__(self, file, events=None):\n"
2720         "  if not hasattr(file, 'read'):\n"
2721         "    file = open(file, 'rb')\n"
2722         "  self._file = file\n"
2723         "  self._events = events\n"
2724         " def __iter__(self):\n"
2725         "  events = []\n"
2726         "  b = cElementTree.TreeBuilder()\n"
2727         "  p = cElementTree.XMLParser(b)\n"
2728         "  p._setevents(events, self._events)\n"
2729         "  while 1:\n"
2730         "    data = self._file.read(16384)\n"
2731         "    if not data:\n"
2732         "      break\n"
2733         "    p.feed(data)\n"
2734         "    for event in events:\n"
2735         "      yield event\n"
2736         "    del events[:]\n"
2737         "  root = p.close()\n"
2738         "  for event in events:\n"
2739         "    yield event\n"
2740         "  self.root = root\n"
2741         "cElementTree.iterparse = iterparse\n"
2742 #endif
2743
2744         "def PI(target, text=None):\n" /* public */
2745         "  element = cElementTree.Element(ET.ProcessingInstruction)\n"
2746         "  element.text = target\n"
2747         "  if text:\n"
2748         "    element.text = element.text + ' ' + text\n"
2749         "  return element\n"
2750
2751         "  elem = cElementTree.Element(ET.PI)\n"
2752         "  elem.text = text\n"
2753         "  return elem\n"
2754         "cElementTree.PI = cElementTree.ProcessingInstruction = PI\n"
2755
2756         "def XML(text):\n" /* public */
2757         "  parser = cElementTree.XMLParser()\n"
2758         "  parser.feed(text)\n"
2759         "  return parser.close()\n"
2760         "cElementTree.XML = cElementTree.fromstring = XML\n"
2761
2762         "def XMLID(text):\n" /* public */
2763         "  tree = XML(text)\n"
2764         "  ids = {}\n"
2765         "  for elem in tree.getiterator():\n"
2766         "    id = elem.get('id')\n"
2767         "    if id:\n"
2768         "      ids[id] = elem\n"
2769         "  return tree, ids\n"
2770         "cElementTree.XMLID = XMLID\n"
2771
2772         "cElementTree.dump = ET.dump\n"
2773         "cElementTree.ElementPath = ElementPath = ET.ElementPath\n"
2774         "cElementTree.iselement = ET.iselement\n"
2775         "cElementTree.QName = ET.QName\n"
2776         "cElementTree.tostring = ET.tostring\n"
2777         "cElementTree.VERSION = '" VERSION "'\n"
2778         "cElementTree.__version__ = '" VERSION "'\n"
2779         "cElementTree.XMLParserError = SyntaxError\n"
2780
2781        );
2782
2783     PyRun_String(bootstrap, Py_file_input, g, NULL);
2784
2785     elementpath_obj = PyDict_GetItemString(g, "ElementPath");
2786
2787     elementtree_copyelement_obj = PyDict_GetItemString(g, "copyelement");
2788     if (elementtree_copyelement_obj) {
2789         /* reduce hack needed; enable reduce method */
2790         PyMethodDef* mp;
2791         for (mp = element_methods; mp->ml_name; mp++)
2792             if (mp->ml_meth == (PyCFunction) element_reduce) {
2793                 mp->ml_name = "__reduce__";
2794                 break;
2795             }
2796     } else
2797         PyErr_Clear();
2798     elementtree_deepcopy_obj = PyDict_GetItemString(g, "deepcopy");
2799     elementtree_getiterator_obj = PyDict_GetItemString(g, "getiterator");
2800
2801 #if defined(USE_PYEXPAT_CAPI)
2802     /* link against pyexpat, if possible */
2803     capi = PyCObject_Import("pyexpat", "expat_CAPI");
2804     if (capi &&
2805         strcmp(capi->magic, PyExpat_CAPI_MAGIC) == 0 &&
2806         capi->size <= sizeof(*expat_capi) &&
2807         capi->MAJOR_VERSION == XML_MAJOR_VERSION &&
2808         capi->MINOR_VERSION == XML_MINOR_VERSION &&
2809         capi->MICRO_VERSION == XML_MICRO_VERSION)
2810         expat_capi = capi;
2811     else
2812         expat_capi = NULL;
2813 #endif
2814
2815 }