Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1 | /* |
| 2 | * ElementTree |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3 | * $Id: _elementtree.c 3473 2009-01-11 22:53:55Z fredrik $ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 4 | * |
| 5 | * elementtree accelerator |
| 6 | * |
| 7 | * History: |
| 8 | * 1999-06-20 fl created (as part of sgmlop) |
| 9 | * 2001-05-29 fl effdom edition |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 10 | * 2003-02-27 fl elementtree edition (alpha) |
| 11 | * 2004-06-03 fl updates for elementtree 1.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 12 | * 2005-01-05 fl major optimization effort |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 13 | * 2005-01-11 fl first public release (cElementTree 0.8) |
| 14 | * 2005-01-12 fl split element object into base and extras |
| 15 | * 2005-01-13 fl use tagged pointers for tail/text (cElementTree 0.9) |
| 16 | * 2005-01-17 fl added treebuilder close method |
| 17 | * 2005-01-17 fl fixed crash in getchildren |
| 18 | * 2005-01-18 fl removed observer api, added iterparse (cElementTree 0.9.3) |
| 19 | * 2005-01-23 fl revised iterparse api; added namespace event support (0.9.8) |
| 20 | * 2005-01-26 fl added VERSION module property (cElementTree 1.0) |
| 21 | * 2005-01-28 fl added remove method (1.0.1) |
| 22 | * 2005-03-01 fl added iselement function; fixed makeelement aliasing (1.0.2) |
| 23 | * 2005-03-13 fl export Comment and ProcessingInstruction/PI helpers |
| 24 | * 2005-03-26 fl added Comment and PI support to XMLParser |
| 25 | * 2005-03-27 fl event optimizations; complain about bogus events |
| 26 | * 2005-08-08 fl fixed read error handling in parse |
| 27 | * 2005-08-11 fl added runtime test for copy workaround (1.0.3) |
| 28 | * 2005-12-13 fl added expat_capi support (for xml.etree) (1.0.4) |
| 29 | * 2005-12-16 fl added support for non-standard encodings |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 30 | * 2006-03-08 fl fixed a couple of potential null-refs and leaks |
| 31 | * 2006-03-12 fl merge in 2.5 ssize_t changes |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 32 | * 2007-08-25 fl call custom builder's close method from XMLParser |
| 33 | * 2007-08-31 fl added iter, extend from ET 1.3 |
| 34 | * 2007-09-01 fl fixed ParseError exception, setslice source type, etc |
| 35 | * 2007-09-03 fl fixed handling of negative insert indexes |
| 36 | * 2007-09-04 fl added itertext from ET 1.3 |
| 37 | * 2007-09-06 fl added position attribute to ParseError exception |
| 38 | * 2008-06-06 fl delay error reporting in iterparse (from Hrvoje Niksic) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 39 | * |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 40 | * Copyright (c) 1999-2009 by Secret Labs AB. All rights reserved. |
| 41 | * Copyright (c) 1999-2009 by Fredrik Lundh. |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 42 | * |
| 43 | * info@pythonware.com |
| 44 | * http://www.pythonware.com |
| 45 | */ |
| 46 | |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 47 | /* Licensed to PSF under a Contributor Agreement. */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 48 | /* See http://www.python.org/psf/license for licensing details. */ |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 49 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 50 | #include "Python.h" |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 51 | #include "structmember.h" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 52 | |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 53 | #define VERSION "1.0.6" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 54 | |
| 55 | /* -------------------------------------------------------------------- */ |
| 56 | /* configuration */ |
| 57 | |
| 58 | /* Leave defined to include the expat-based XMLParser type */ |
| 59 | #define USE_EXPAT |
| 60 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 61 | /* Define to do all expat calls via pyexpat's embedded expat library */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 62 | /* #define USE_PYEXPAT_CAPI */ |
| 63 | |
| 64 | /* An element can hold this many children without extra memory |
| 65 | allocations. */ |
| 66 | #define STATIC_CHILDREN 4 |
| 67 | |
| 68 | /* For best performance, chose a value so that 80-90% of all nodes |
| 69 | have no more than the given number of children. Set this to zero |
| 70 | to minimize the size of the element structure itself (this only |
| 71 | helps if you have lots of leaf nodes with attributes). */ |
| 72 | |
| 73 | /* Also note that pymalloc always allocates blocks in multiples of |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 74 | eight bytes. For the current C version of ElementTree, this means |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 75 | that the number of children should be an even number, at least on |
| 76 | 32-bit platforms. */ |
| 77 | |
| 78 | /* -------------------------------------------------------------------- */ |
| 79 | |
| 80 | #if 0 |
| 81 | static int memory = 0; |
| 82 | #define ALLOC(size, comment)\ |
| 83 | do { memory += size; printf("%8d - %s\n", memory, comment); } while (0) |
| 84 | #define RELEASE(size, comment)\ |
| 85 | do { memory -= size; printf("%8d - %s\n", memory, comment); } while (0) |
| 86 | #else |
| 87 | #define ALLOC(size, comment) |
| 88 | #define RELEASE(size, comment) |
| 89 | #endif |
| 90 | |
| 91 | /* compiler tweaks */ |
| 92 | #if defined(_MSC_VER) |
| 93 | #define LOCAL(type) static __inline type __fastcall |
| 94 | #else |
| 95 | #define LOCAL(type) static type |
| 96 | #endif |
| 97 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 98 | /* macros used to store 'join' flags in string object pointers. note |
| 99 | that all use of text and tail as object pointers must be wrapped in |
| 100 | JOIN_OBJ. see comments in the ElementObject definition for more |
| 101 | info. */ |
| 102 | #define JOIN_GET(p) ((Py_uintptr_t) (p) & 1) |
| 103 | #define JOIN_SET(p, flag) ((void*) ((Py_uintptr_t) (JOIN_OBJ(p)) | (flag))) |
| 104 | #define JOIN_OBJ(p) ((PyObject*) ((Py_uintptr_t) (p) & ~1)) |
| 105 | |
| 106 | /* glue functions (see the init function for details) */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 107 | static PyObject* elementtree_parseerror_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 108 | static PyObject* elementtree_deepcopy_obj; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 109 | static PyObject* elementtree_iter_obj; |
| 110 | static PyObject* elementtree_itertext_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 111 | static PyObject* elementpath_obj; |
| 112 | |
| 113 | /* helpers */ |
| 114 | |
| 115 | LOCAL(PyObject*) |
| 116 | deepcopy(PyObject* object, PyObject* memo) |
| 117 | { |
| 118 | /* do a deep copy of the given object */ |
| 119 | |
| 120 | PyObject* args; |
| 121 | PyObject* result; |
| 122 | |
| 123 | if (!elementtree_deepcopy_obj) { |
| 124 | PyErr_SetString( |
| 125 | PyExc_RuntimeError, |
| 126 | "deepcopy helper not found" |
| 127 | ); |
| 128 | return NULL; |
| 129 | } |
| 130 | |
| 131 | args = PyTuple_New(2); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 132 | if (!args) |
| 133 | return NULL; |
| 134 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 135 | Py_INCREF(object); PyTuple_SET_ITEM(args, 0, (PyObject*) object); |
| 136 | Py_INCREF(memo); PyTuple_SET_ITEM(args, 1, (PyObject*) memo); |
| 137 | |
| 138 | result = PyObject_CallObject(elementtree_deepcopy_obj, args); |
| 139 | |
| 140 | Py_DECREF(args); |
| 141 | |
| 142 | return result; |
| 143 | } |
| 144 | |
| 145 | LOCAL(PyObject*) |
| 146 | list_join(PyObject* list) |
| 147 | { |
| 148 | /* join list elements (destroying the list in the process) */ |
| 149 | |
| 150 | PyObject* joiner; |
| 151 | PyObject* function; |
| 152 | PyObject* args; |
| 153 | PyObject* result; |
| 154 | |
| 155 | switch (PyList_GET_SIZE(list)) { |
| 156 | case 0: |
| 157 | Py_DECREF(list); |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 158 | return PyBytes_FromString(""); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 159 | case 1: |
| 160 | result = PyList_GET_ITEM(list, 0); |
| 161 | Py_INCREF(result); |
| 162 | Py_DECREF(list); |
| 163 | return result; |
| 164 | } |
| 165 | |
| 166 | /* two or more elements: slice out a suitable separator from the |
| 167 | first member, and use that to join the entire list */ |
| 168 | |
| 169 | joiner = PySequence_GetSlice(PyList_GET_ITEM(list, 0), 0, 0); |
| 170 | if (!joiner) |
| 171 | return NULL; |
| 172 | |
| 173 | function = PyObject_GetAttrString(joiner, "join"); |
| 174 | if (!function) { |
| 175 | Py_DECREF(joiner); |
| 176 | return NULL; |
| 177 | } |
| 178 | |
| 179 | args = PyTuple_New(1); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 180 | if (!args) |
| 181 | return NULL; |
| 182 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 183 | PyTuple_SET_ITEM(args, 0, list); |
| 184 | |
| 185 | result = PyObject_CallObject(function, args); |
| 186 | |
| 187 | Py_DECREF(args); /* also removes list */ |
| 188 | Py_DECREF(function); |
| 189 | Py_DECREF(joiner); |
| 190 | |
| 191 | return result; |
| 192 | } |
| 193 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 194 | /* Is the given object an empty dictionary? |
| 195 | */ |
| 196 | static int |
| 197 | is_empty_dict(PyObject *obj) |
| 198 | { |
| 199 | return PyDict_CheckExact(obj) && PyDict_Size(obj) == 0; |
| 200 | } |
| 201 | |
| 202 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 203 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 204 | /* the Element type */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 205 | |
| 206 | typedef struct { |
| 207 | |
| 208 | /* attributes (a dictionary object), or None if no attributes */ |
| 209 | PyObject* attrib; |
| 210 | |
| 211 | /* child elements */ |
| 212 | int length; /* actual number of items */ |
| 213 | int allocated; /* allocated items */ |
| 214 | |
| 215 | /* this either points to _children or to a malloced buffer */ |
| 216 | PyObject* *children; |
| 217 | |
| 218 | PyObject* _children[STATIC_CHILDREN]; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 219 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 220 | } ElementObjectExtra; |
| 221 | |
| 222 | typedef struct { |
| 223 | PyObject_HEAD |
| 224 | |
| 225 | /* element tag (a string). */ |
| 226 | PyObject* tag; |
| 227 | |
| 228 | /* text before first child. note that this is a tagged pointer; |
| 229 | use JOIN_OBJ to get the object pointer. the join flag is used |
| 230 | to distinguish lists created by the tree builder from lists |
| 231 | assigned to the attribute by application code; the former |
| 232 | should be joined before being returned to the user, the latter |
| 233 | should be left intact. */ |
| 234 | PyObject* text; |
| 235 | |
| 236 | /* text after this element, in parent. note that this is a tagged |
| 237 | pointer; use JOIN_OBJ to get the object pointer. */ |
| 238 | PyObject* tail; |
| 239 | |
| 240 | ElementObjectExtra* extra; |
| 241 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 242 | PyObject *weakreflist; /* For tp_weaklistoffset */ |
| 243 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 244 | } ElementObject; |
| 245 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 246 | static PyTypeObject Element_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 247 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 248 | #define Element_CheckExact(op) (Py_TYPE(op) == &Element_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 249 | |
| 250 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 251 | /* Element constructors and destructor */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 252 | |
| 253 | LOCAL(int) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 254 | create_extra(ElementObject* self, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 255 | { |
| 256 | self->extra = PyObject_Malloc(sizeof(ElementObjectExtra)); |
| 257 | if (!self->extra) |
| 258 | return -1; |
| 259 | |
| 260 | if (!attrib) |
| 261 | attrib = Py_None; |
| 262 | |
| 263 | Py_INCREF(attrib); |
| 264 | self->extra->attrib = attrib; |
| 265 | |
| 266 | self->extra->length = 0; |
| 267 | self->extra->allocated = STATIC_CHILDREN; |
| 268 | self->extra->children = self->extra->_children; |
| 269 | |
| 270 | return 0; |
| 271 | } |
| 272 | |
| 273 | LOCAL(void) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 274 | dealloc_extra(ElementObject* self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 275 | { |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 276 | ElementObjectExtra *myextra; |
| 277 | int i; |
| 278 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 279 | if (!self->extra) |
| 280 | return; |
| 281 | |
| 282 | /* Avoid DECREFs calling into this code again (cycles, etc.) |
| 283 | */ |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 284 | myextra = self->extra; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 285 | self->extra = NULL; |
| 286 | |
| 287 | Py_DECREF(myextra->attrib); |
| 288 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 289 | for (i = 0; i < myextra->length; i++) |
| 290 | Py_DECREF(myextra->children[i]); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 291 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 292 | if (myextra->children != myextra->_children) |
| 293 | PyObject_Free(myextra->children); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 294 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 295 | PyObject_Free(myextra); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 296 | } |
| 297 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 298 | /* Convenience internal function to create new Element objects with the given |
| 299 | * tag and attributes. |
| 300 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 301 | LOCAL(PyObject*) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 302 | create_new_element(PyObject* tag, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 303 | { |
| 304 | ElementObject* self; |
| 305 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 306 | self = PyObject_GC_New(ElementObject, &Element_Type); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 307 | if (self == NULL) |
| 308 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 309 | self->extra = NULL; |
| 310 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 311 | if (attrib != Py_None && !is_empty_dict(attrib)) { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 312 | if (create_extra(self, attrib) < 0) { |
Thomas Wouters | 477c8d5 | 2006-05-27 19:21:47 +0000 | [diff] [blame] | 313 | PyObject_Del(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 314 | return NULL; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 315 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 316 | } |
| 317 | |
| 318 | Py_INCREF(tag); |
| 319 | self->tag = tag; |
| 320 | |
| 321 | Py_INCREF(Py_None); |
| 322 | self->text = Py_None; |
| 323 | |
| 324 | Py_INCREF(Py_None); |
| 325 | self->tail = Py_None; |
| 326 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 327 | self->weakreflist = NULL; |
| 328 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 329 | ALLOC(sizeof(ElementObject), "create element"); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 330 | PyObject_GC_Track(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 331 | return (PyObject*) self; |
| 332 | } |
| 333 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 334 | static PyObject * |
| 335 | element_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 336 | { |
| 337 | ElementObject *e = (ElementObject *)type->tp_alloc(type, 0); |
| 338 | if (e != NULL) { |
| 339 | Py_INCREF(Py_None); |
| 340 | e->tag = Py_None; |
| 341 | |
| 342 | Py_INCREF(Py_None); |
| 343 | e->text = Py_None; |
| 344 | |
| 345 | Py_INCREF(Py_None); |
| 346 | e->tail = Py_None; |
| 347 | |
| 348 | e->extra = NULL; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 349 | e->weakreflist = NULL; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 350 | } |
| 351 | return (PyObject *)e; |
| 352 | } |
| 353 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 354 | /* Helper function for extracting the attrib dictionary from a keywords dict. |
| 355 | * This is required by some constructors/functions in this module that can |
| 356 | * either accept attrib as a keyword argument or all attributes splashed |
| 357 | * directly into *kwds. |
| 358 | * If there is no 'attrib' keyword, return an empty dict. |
| 359 | */ |
| 360 | static PyObject* |
| 361 | get_attrib_from_keywords(PyObject *kwds) |
| 362 | { |
| 363 | PyObject *attrib_str = PyUnicode_FromString("attrib"); |
| 364 | PyObject *attrib = PyDict_GetItem(kwds, attrib_str); |
| 365 | |
| 366 | if (attrib) { |
| 367 | /* If attrib was found in kwds, copy its value and remove it from |
| 368 | * kwds |
| 369 | */ |
| 370 | if (!PyDict_Check(attrib)) { |
| 371 | Py_DECREF(attrib_str); |
| 372 | PyErr_Format(PyExc_TypeError, "attrib must be dict, not %.100s", |
| 373 | Py_TYPE(attrib)->tp_name); |
| 374 | return NULL; |
| 375 | } |
| 376 | attrib = PyDict_Copy(attrib); |
| 377 | PyDict_DelItem(kwds, attrib_str); |
| 378 | } else { |
| 379 | attrib = PyDict_New(); |
| 380 | } |
| 381 | |
| 382 | Py_DECREF(attrib_str); |
| 383 | |
| 384 | if (attrib) |
| 385 | PyDict_Update(attrib, kwds); |
| 386 | return attrib; |
| 387 | } |
| 388 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 389 | static int |
| 390 | element_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 391 | { |
| 392 | PyObject *tag; |
| 393 | PyObject *tmp; |
| 394 | PyObject *attrib = NULL; |
| 395 | ElementObject *self_elem; |
| 396 | |
| 397 | if (!PyArg_ParseTuple(args, "O|O!:Element", &tag, &PyDict_Type, &attrib)) |
| 398 | return -1; |
| 399 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 400 | if (attrib) { |
| 401 | /* attrib passed as positional arg */ |
| 402 | attrib = PyDict_Copy(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 403 | if (!attrib) |
| 404 | return -1; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 405 | if (kwds) { |
| 406 | if (PyDict_Update(attrib, kwds) < 0) { |
| 407 | return -1; |
| 408 | } |
| 409 | } |
| 410 | } else if (kwds) { |
| 411 | /* have keywords args */ |
| 412 | attrib = get_attrib_from_keywords(kwds); |
| 413 | if (!attrib) |
| 414 | return -1; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 415 | } else { |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 416 | /* no attrib arg, no kwds, so no attributes */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 417 | Py_INCREF(Py_None); |
| 418 | attrib = Py_None; |
| 419 | } |
| 420 | |
| 421 | self_elem = (ElementObject *)self; |
| 422 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 423 | if (attrib != Py_None && !is_empty_dict(attrib)) { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 424 | if (create_extra(self_elem, attrib) < 0) { |
| 425 | PyObject_Del(self_elem); |
| 426 | return -1; |
| 427 | } |
| 428 | } |
| 429 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 430 | /* We own a reference to attrib here and it's no longer needed. */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 431 | Py_DECREF(attrib); |
| 432 | |
| 433 | /* Replace the objects already pointed to by tag, text and tail. */ |
| 434 | tmp = self_elem->tag; |
| 435 | self_elem->tag = tag; |
| 436 | Py_INCREF(tag); |
| 437 | Py_DECREF(tmp); |
| 438 | |
| 439 | tmp = self_elem->text; |
| 440 | self_elem->text = Py_None; |
| 441 | Py_INCREF(Py_None); |
| 442 | Py_DECREF(JOIN_OBJ(tmp)); |
| 443 | |
| 444 | tmp = self_elem->tail; |
| 445 | self_elem->tail = Py_None; |
| 446 | Py_INCREF(Py_None); |
| 447 | Py_DECREF(JOIN_OBJ(tmp)); |
| 448 | |
| 449 | return 0; |
| 450 | } |
| 451 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 452 | LOCAL(int) |
| 453 | element_resize(ElementObject* self, int extra) |
| 454 | { |
| 455 | int size; |
| 456 | PyObject* *children; |
| 457 | |
| 458 | /* make sure self->children can hold the given number of extra |
| 459 | elements. set an exception and return -1 if allocation failed */ |
| 460 | |
| 461 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 462 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 463 | |
| 464 | size = self->extra->length + extra; |
| 465 | |
| 466 | if (size > self->extra->allocated) { |
| 467 | /* use Python 2.4's list growth strategy */ |
| 468 | size = (size >> 3) + (size < 9 ? 3 : 6) + size; |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 469 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer "children" |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 470 | * which needs at least 4 bytes. |
| 471 | * Although it's a false alarm always assume at least one child to |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 472 | * be safe. |
| 473 | */ |
| 474 | size = size ? size : 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 475 | if (self->extra->children != self->extra->_children) { |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 476 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 477 | * "children", which needs at least 4 bytes. Although it's a |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 478 | * false alarm always assume at least one child to be safe. |
| 479 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 480 | children = PyObject_Realloc(self->extra->children, |
| 481 | size * sizeof(PyObject*)); |
| 482 | if (!children) |
| 483 | goto nomemory; |
| 484 | } else { |
| 485 | children = PyObject_Malloc(size * sizeof(PyObject*)); |
| 486 | if (!children) |
| 487 | goto nomemory; |
| 488 | /* copy existing children from static area to malloc buffer */ |
| 489 | memcpy(children, self->extra->children, |
| 490 | self->extra->length * sizeof(PyObject*)); |
| 491 | } |
| 492 | self->extra->children = children; |
| 493 | self->extra->allocated = size; |
| 494 | } |
| 495 | |
| 496 | return 0; |
| 497 | |
| 498 | nomemory: |
| 499 | PyErr_NoMemory(); |
| 500 | return -1; |
| 501 | } |
| 502 | |
| 503 | LOCAL(int) |
| 504 | element_add_subelement(ElementObject* self, PyObject* element) |
| 505 | { |
| 506 | /* add a child element to a parent */ |
| 507 | |
| 508 | if (element_resize(self, 1) < 0) |
| 509 | return -1; |
| 510 | |
| 511 | Py_INCREF(element); |
| 512 | self->extra->children[self->extra->length] = element; |
| 513 | |
| 514 | self->extra->length++; |
| 515 | |
| 516 | return 0; |
| 517 | } |
| 518 | |
| 519 | LOCAL(PyObject*) |
| 520 | element_get_attrib(ElementObject* self) |
| 521 | { |
| 522 | /* return borrowed reference to attrib dictionary */ |
| 523 | /* note: this function assumes that the extra section exists */ |
| 524 | |
| 525 | PyObject* res = self->extra->attrib; |
| 526 | |
| 527 | if (res == Py_None) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 528 | Py_DECREF(res); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 529 | /* create missing dictionary */ |
| 530 | res = PyDict_New(); |
| 531 | if (!res) |
| 532 | return NULL; |
| 533 | self->extra->attrib = res; |
| 534 | } |
| 535 | |
| 536 | return res; |
| 537 | } |
| 538 | |
| 539 | LOCAL(PyObject*) |
| 540 | element_get_text(ElementObject* self) |
| 541 | { |
| 542 | /* return borrowed reference to text attribute */ |
| 543 | |
| 544 | PyObject* res = self->text; |
| 545 | |
| 546 | if (JOIN_GET(res)) { |
| 547 | res = JOIN_OBJ(res); |
| 548 | if (PyList_CheckExact(res)) { |
| 549 | res = list_join(res); |
| 550 | if (!res) |
| 551 | return NULL; |
| 552 | self->text = res; |
| 553 | } |
| 554 | } |
| 555 | |
| 556 | return res; |
| 557 | } |
| 558 | |
| 559 | LOCAL(PyObject*) |
| 560 | element_get_tail(ElementObject* self) |
| 561 | { |
| 562 | /* return borrowed reference to text attribute */ |
| 563 | |
| 564 | PyObject* res = self->tail; |
| 565 | |
| 566 | if (JOIN_GET(res)) { |
| 567 | res = JOIN_OBJ(res); |
| 568 | if (PyList_CheckExact(res)) { |
| 569 | res = list_join(res); |
| 570 | if (!res) |
| 571 | return NULL; |
| 572 | self->tail = res; |
| 573 | } |
| 574 | } |
| 575 | |
| 576 | return res; |
| 577 | } |
| 578 | |
| 579 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 580 | subelement(PyObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 581 | { |
| 582 | PyObject* elem; |
| 583 | |
| 584 | ElementObject* parent; |
| 585 | PyObject* tag; |
| 586 | PyObject* attrib = NULL; |
| 587 | if (!PyArg_ParseTuple(args, "O!O|O!:SubElement", |
| 588 | &Element_Type, &parent, &tag, |
| 589 | &PyDict_Type, &attrib)) |
| 590 | return NULL; |
| 591 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 592 | if (attrib) { |
| 593 | /* attrib passed as positional arg */ |
| 594 | attrib = PyDict_Copy(attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 595 | if (!attrib) |
| 596 | return NULL; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 597 | if (kwds) { |
| 598 | if (PyDict_Update(attrib, kwds) < 0) { |
| 599 | return NULL; |
| 600 | } |
| 601 | } |
| 602 | } else if (kwds) { |
| 603 | /* have keyword args */ |
| 604 | attrib = get_attrib_from_keywords(kwds); |
| 605 | if (!attrib) |
| 606 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 607 | } else { |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 608 | /* no attrib arg, no kwds, so no attribute */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 609 | Py_INCREF(Py_None); |
| 610 | attrib = Py_None; |
| 611 | } |
| 612 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 613 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 614 | |
| 615 | Py_DECREF(attrib); |
| 616 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 617 | if (element_add_subelement(parent, elem) < 0) { |
| 618 | Py_DECREF(elem); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 619 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 620 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 621 | |
| 622 | return elem; |
| 623 | } |
| 624 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 625 | static int |
| 626 | element_gc_traverse(ElementObject *self, visitproc visit, void *arg) |
| 627 | { |
| 628 | Py_VISIT(self->tag); |
| 629 | Py_VISIT(JOIN_OBJ(self->text)); |
| 630 | Py_VISIT(JOIN_OBJ(self->tail)); |
| 631 | |
| 632 | if (self->extra) { |
| 633 | int i; |
| 634 | Py_VISIT(self->extra->attrib); |
| 635 | |
| 636 | for (i = 0; i < self->extra->length; ++i) |
| 637 | Py_VISIT(self->extra->children[i]); |
| 638 | } |
| 639 | return 0; |
| 640 | } |
| 641 | |
| 642 | static int |
| 643 | element_gc_clear(ElementObject *self) |
| 644 | { |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 645 | Py_CLEAR(self->tag); |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 646 | |
| 647 | /* The following is like Py_CLEAR for self->text and self->tail, but |
| 648 | * written explicitily because the real pointers hide behind access |
| 649 | * macros. |
| 650 | */ |
| 651 | if (self->text) { |
| 652 | PyObject *tmp = JOIN_OBJ(self->text); |
| 653 | self->text = NULL; |
| 654 | Py_DECREF(tmp); |
| 655 | } |
| 656 | |
| 657 | if (self->tail) { |
| 658 | PyObject *tmp = JOIN_OBJ(self->tail); |
| 659 | self->tail = NULL; |
| 660 | Py_DECREF(tmp); |
| 661 | } |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 662 | |
| 663 | /* After dropping all references from extra, it's no longer valid anyway, |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 664 | * so fully deallocate it. |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 665 | */ |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 666 | dealloc_extra(self); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 667 | return 0; |
| 668 | } |
| 669 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 670 | static void |
| 671 | element_dealloc(ElementObject* self) |
| 672 | { |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 673 | PyObject_GC_UnTrack(self); |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 674 | |
| 675 | if (self->weakreflist != NULL) |
| 676 | PyObject_ClearWeakRefs((PyObject *) self); |
| 677 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 678 | /* element_gc_clear clears all references and deallocates extra |
| 679 | */ |
| 680 | element_gc_clear(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 681 | |
| 682 | RELEASE(sizeof(ElementObject), "destroy element"); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 683 | Py_TYPE(self)->tp_free((PyObject *)self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 684 | } |
| 685 | |
| 686 | /* -------------------------------------------------------------------- */ |
| 687 | /* methods (in alphabetical order) */ |
| 688 | |
| 689 | static PyObject* |
| 690 | element_append(ElementObject* self, PyObject* args) |
| 691 | { |
| 692 | PyObject* element; |
| 693 | if (!PyArg_ParseTuple(args, "O!:append", &Element_Type, &element)) |
| 694 | return NULL; |
| 695 | |
| 696 | if (element_add_subelement(self, element) < 0) |
| 697 | return NULL; |
| 698 | |
| 699 | Py_RETURN_NONE; |
| 700 | } |
| 701 | |
| 702 | static PyObject* |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 703 | element_clearmethod(ElementObject* self, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 704 | { |
| 705 | if (!PyArg_ParseTuple(args, ":clear")) |
| 706 | return NULL; |
| 707 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 708 | dealloc_extra(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 709 | |
| 710 | Py_INCREF(Py_None); |
| 711 | Py_DECREF(JOIN_OBJ(self->text)); |
| 712 | self->text = Py_None; |
| 713 | |
| 714 | Py_INCREF(Py_None); |
| 715 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 716 | self->tail = Py_None; |
| 717 | |
| 718 | Py_RETURN_NONE; |
| 719 | } |
| 720 | |
| 721 | static PyObject* |
| 722 | element_copy(ElementObject* self, PyObject* args) |
| 723 | { |
| 724 | int i; |
| 725 | ElementObject* element; |
| 726 | |
| 727 | if (!PyArg_ParseTuple(args, ":__copy__")) |
| 728 | return NULL; |
| 729 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 730 | element = (ElementObject*) create_new_element( |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 731 | self->tag, (self->extra) ? self->extra->attrib : Py_None |
| 732 | ); |
| 733 | if (!element) |
| 734 | return NULL; |
| 735 | |
| 736 | Py_DECREF(JOIN_OBJ(element->text)); |
| 737 | element->text = self->text; |
| 738 | Py_INCREF(JOIN_OBJ(element->text)); |
| 739 | |
| 740 | Py_DECREF(JOIN_OBJ(element->tail)); |
| 741 | element->tail = self->tail; |
| 742 | Py_INCREF(JOIN_OBJ(element->tail)); |
| 743 | |
| 744 | if (self->extra) { |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 745 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 746 | if (element_resize(element, self->extra->length) < 0) { |
| 747 | Py_DECREF(element); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 748 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 749 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 750 | |
| 751 | for (i = 0; i < self->extra->length; i++) { |
| 752 | Py_INCREF(self->extra->children[i]); |
| 753 | element->extra->children[i] = self->extra->children[i]; |
| 754 | } |
| 755 | |
| 756 | element->extra->length = self->extra->length; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 757 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 758 | } |
| 759 | |
| 760 | return (PyObject*) element; |
| 761 | } |
| 762 | |
| 763 | static PyObject* |
| 764 | element_deepcopy(ElementObject* self, PyObject* args) |
| 765 | { |
| 766 | int i; |
| 767 | ElementObject* element; |
| 768 | PyObject* tag; |
| 769 | PyObject* attrib; |
| 770 | PyObject* text; |
| 771 | PyObject* tail; |
| 772 | PyObject* id; |
| 773 | |
| 774 | PyObject* memo; |
| 775 | if (!PyArg_ParseTuple(args, "O:__deepcopy__", &memo)) |
| 776 | return NULL; |
| 777 | |
| 778 | tag = deepcopy(self->tag, memo); |
| 779 | if (!tag) |
| 780 | return NULL; |
| 781 | |
| 782 | if (self->extra) { |
| 783 | attrib = deepcopy(self->extra->attrib, memo); |
| 784 | if (!attrib) { |
| 785 | Py_DECREF(tag); |
| 786 | return NULL; |
| 787 | } |
| 788 | } else { |
| 789 | Py_INCREF(Py_None); |
| 790 | attrib = Py_None; |
| 791 | } |
| 792 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 793 | element = (ElementObject*) create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 794 | |
| 795 | Py_DECREF(tag); |
| 796 | Py_DECREF(attrib); |
| 797 | |
| 798 | if (!element) |
| 799 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 800 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 801 | text = deepcopy(JOIN_OBJ(self->text), memo); |
| 802 | if (!text) |
| 803 | goto error; |
| 804 | Py_DECREF(element->text); |
| 805 | element->text = JOIN_SET(text, JOIN_GET(self->text)); |
| 806 | |
| 807 | tail = deepcopy(JOIN_OBJ(self->tail), memo); |
| 808 | if (!tail) |
| 809 | goto error; |
| 810 | Py_DECREF(element->tail); |
| 811 | element->tail = JOIN_SET(tail, JOIN_GET(self->tail)); |
| 812 | |
| 813 | if (self->extra) { |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 814 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 815 | if (element_resize(element, self->extra->length) < 0) |
| 816 | goto error; |
| 817 | |
| 818 | for (i = 0; i < self->extra->length; i++) { |
| 819 | PyObject* child = deepcopy(self->extra->children[i], memo); |
| 820 | if (!child) { |
| 821 | element->extra->length = i; |
| 822 | goto error; |
| 823 | } |
| 824 | element->extra->children[i] = child; |
| 825 | } |
| 826 | |
| 827 | element->extra->length = self->extra->length; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 828 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 829 | } |
| 830 | |
| 831 | /* add object to memo dictionary (so deepcopy won't visit it again) */ |
Christian Heimes | 217cfd1 | 2007-12-02 14:31:20 +0000 | [diff] [blame] | 832 | id = PyLong_FromLong((Py_uintptr_t) self); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 833 | if (!id) |
| 834 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 835 | |
| 836 | i = PyDict_SetItem(memo, id, (PyObject*) element); |
| 837 | |
| 838 | Py_DECREF(id); |
| 839 | |
| 840 | if (i < 0) |
| 841 | goto error; |
| 842 | |
| 843 | return (PyObject*) element; |
| 844 | |
| 845 | error: |
| 846 | Py_DECREF(element); |
| 847 | return NULL; |
| 848 | } |
| 849 | |
| 850 | LOCAL(int) |
| 851 | checkpath(PyObject* tag) |
| 852 | { |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 853 | Py_ssize_t i; |
| 854 | int check = 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 855 | |
| 856 | /* check if a tag contains an xpath character */ |
| 857 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 858 | #define PATHCHAR(ch) \ |
| 859 | (ch == '/' || ch == '*' || ch == '[' || ch == '@' || ch == '.') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 860 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 861 | if (PyUnicode_Check(tag)) { |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 862 | const Py_ssize_t len = PyUnicode_GET_LENGTH(tag); |
| 863 | void *data = PyUnicode_DATA(tag); |
| 864 | unsigned int kind = PyUnicode_KIND(tag); |
| 865 | for (i = 0; i < len; i++) { |
| 866 | Py_UCS4 ch = PyUnicode_READ(kind, data, i); |
| 867 | if (ch == '{') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 868 | check = 0; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 869 | else if (ch == '}') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 870 | check = 1; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 871 | else if (check && PATHCHAR(ch)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 872 | return 1; |
| 873 | } |
| 874 | return 0; |
| 875 | } |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 876 | if (PyBytes_Check(tag)) { |
| 877 | char *p = PyBytes_AS_STRING(tag); |
| 878 | for (i = 0; i < PyBytes_GET_SIZE(tag); i++) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 879 | if (p[i] == '{') |
| 880 | check = 0; |
| 881 | else if (p[i] == '}') |
| 882 | check = 1; |
| 883 | else if (check && PATHCHAR(p[i])) |
| 884 | return 1; |
| 885 | } |
| 886 | return 0; |
| 887 | } |
| 888 | |
| 889 | return 1; /* unknown type; might be path expression */ |
| 890 | } |
| 891 | |
| 892 | static PyObject* |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 893 | element_extend(ElementObject* self, PyObject* args) |
| 894 | { |
| 895 | PyObject* seq; |
| 896 | Py_ssize_t i, seqlen = 0; |
| 897 | |
| 898 | PyObject* seq_in; |
| 899 | if (!PyArg_ParseTuple(args, "O:extend", &seq_in)) |
| 900 | return NULL; |
| 901 | |
| 902 | seq = PySequence_Fast(seq_in, ""); |
| 903 | if (!seq) { |
| 904 | PyErr_Format( |
| 905 | PyExc_TypeError, |
| 906 | "expected sequence, not \"%.200s\"", Py_TYPE(seq_in)->tp_name |
| 907 | ); |
| 908 | return NULL; |
| 909 | } |
| 910 | |
| 911 | seqlen = PySequence_Size(seq); |
| 912 | for (i = 0; i < seqlen; i++) { |
| 913 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 914 | if (!PyObject_IsInstance(element, (PyObject *)&Element_Type)) { |
| 915 | Py_DECREF(seq); |
| 916 | PyErr_Format( |
| 917 | PyExc_TypeError, |
| 918 | "expected an Element, not \"%.200s\"", |
| 919 | Py_TYPE(element)->tp_name); |
| 920 | return NULL; |
| 921 | } |
| 922 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 923 | if (element_add_subelement(self, element) < 0) { |
| 924 | Py_DECREF(seq); |
| 925 | return NULL; |
| 926 | } |
| 927 | } |
| 928 | |
| 929 | Py_DECREF(seq); |
| 930 | |
| 931 | Py_RETURN_NONE; |
| 932 | } |
| 933 | |
| 934 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 935 | element_find(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 936 | { |
| 937 | int i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 938 | PyObject* tag; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 939 | PyObject* namespaces = Py_None; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 940 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 941 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 942 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:find", kwlist, |
| 943 | &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 944 | return NULL; |
| 945 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 946 | if (checkpath(tag) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 947 | _Py_IDENTIFIER(find); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 948 | return _PyObject_CallMethodId( |
| 949 | elementpath_obj, &PyId_find, "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 950 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 951 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 952 | |
| 953 | if (!self->extra) |
| 954 | Py_RETURN_NONE; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 955 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 956 | for (i = 0; i < self->extra->length; i++) { |
| 957 | PyObject* item = self->extra->children[i]; |
| 958 | if (Element_CheckExact(item) && |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 959 | PyObject_RichCompareBool(((ElementObject*)item)->tag, tag, Py_EQ) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 960 | Py_INCREF(item); |
| 961 | return item; |
| 962 | } |
| 963 | } |
| 964 | |
| 965 | Py_RETURN_NONE; |
| 966 | } |
| 967 | |
| 968 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 969 | element_findtext(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 970 | { |
| 971 | int i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 972 | PyObject* tag; |
| 973 | PyObject* default_value = Py_None; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 974 | PyObject* namespaces = Py_None; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 975 | _Py_IDENTIFIER(findtext); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 976 | static char *kwlist[] = {"path", "default", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 977 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 978 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|OO:findtext", kwlist, |
| 979 | &tag, &default_value, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 980 | return NULL; |
| 981 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 982 | if (checkpath(tag) || namespaces != Py_None) |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 983 | return _PyObject_CallMethodId( |
| 984 | elementpath_obj, &PyId_findtext, "OOOO", self, tag, default_value, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 985 | ); |
| 986 | |
| 987 | if (!self->extra) { |
| 988 | Py_INCREF(default_value); |
| 989 | return default_value; |
| 990 | } |
| 991 | |
| 992 | for (i = 0; i < self->extra->length; i++) { |
| 993 | ElementObject* item = (ElementObject*) self->extra->children[i]; |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 994 | if (Element_CheckExact(item) && (PyObject_RichCompareBool(item->tag, tag, Py_EQ) == 1)) { |
| 995 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 996 | PyObject* text = element_get_text(item); |
| 997 | if (text == Py_None) |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 998 | return PyBytes_FromString(""); |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 999 | Py_XINCREF(text); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1000 | return text; |
| 1001 | } |
| 1002 | } |
| 1003 | |
| 1004 | Py_INCREF(default_value); |
| 1005 | return default_value; |
| 1006 | } |
| 1007 | |
| 1008 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1009 | element_findall(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1010 | { |
| 1011 | int i; |
| 1012 | PyObject* out; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1013 | PyObject* tag; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1014 | PyObject* namespaces = Py_None; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1015 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1016 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1017 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:findall", kwlist, |
| 1018 | &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1019 | return NULL; |
| 1020 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1021 | if (checkpath(tag) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1022 | _Py_IDENTIFIER(findall); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1023 | return _PyObject_CallMethodId( |
| 1024 | elementpath_obj, &PyId_findall, "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1025 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1026 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1027 | |
| 1028 | out = PyList_New(0); |
| 1029 | if (!out) |
| 1030 | return NULL; |
| 1031 | |
| 1032 | if (!self->extra) |
| 1033 | return out; |
| 1034 | |
| 1035 | for (i = 0; i < self->extra->length; i++) { |
| 1036 | PyObject* item = self->extra->children[i]; |
| 1037 | if (Element_CheckExact(item) && |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1038 | PyObject_RichCompareBool(((ElementObject*)item)->tag, tag, Py_EQ) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1039 | if (PyList_Append(out, item) < 0) { |
| 1040 | Py_DECREF(out); |
| 1041 | return NULL; |
| 1042 | } |
| 1043 | } |
| 1044 | } |
| 1045 | |
| 1046 | return out; |
| 1047 | } |
| 1048 | |
| 1049 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1050 | element_iterfind(ElementObject *self, PyObject *args, PyObject *kwds) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1051 | { |
| 1052 | PyObject* tag; |
| 1053 | PyObject* namespaces = Py_None; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1054 | _Py_IDENTIFIER(iterfind); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1055 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1056 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1057 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:iterfind", kwlist, |
| 1058 | &tag, &namespaces)) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1059 | return NULL; |
| 1060 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1061 | return _PyObject_CallMethodId( |
| 1062 | elementpath_obj, &PyId_iterfind, "OOO", self, tag, namespaces |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1063 | ); |
| 1064 | } |
| 1065 | |
| 1066 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1067 | element_get(ElementObject* self, PyObject* args) |
| 1068 | { |
| 1069 | PyObject* value; |
| 1070 | |
| 1071 | PyObject* key; |
| 1072 | PyObject* default_value = Py_None; |
| 1073 | if (!PyArg_ParseTuple(args, "O|O:get", &key, &default_value)) |
| 1074 | return NULL; |
| 1075 | |
| 1076 | if (!self->extra || self->extra->attrib == Py_None) |
| 1077 | value = default_value; |
| 1078 | else { |
| 1079 | value = PyDict_GetItem(self->extra->attrib, key); |
| 1080 | if (!value) |
| 1081 | value = default_value; |
| 1082 | } |
| 1083 | |
| 1084 | Py_INCREF(value); |
| 1085 | return value; |
| 1086 | } |
| 1087 | |
| 1088 | static PyObject* |
| 1089 | element_getchildren(ElementObject* self, PyObject* args) |
| 1090 | { |
| 1091 | int i; |
| 1092 | PyObject* list; |
| 1093 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1094 | /* FIXME: report as deprecated? */ |
| 1095 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1096 | if (!PyArg_ParseTuple(args, ":getchildren")) |
| 1097 | return NULL; |
| 1098 | |
| 1099 | if (!self->extra) |
| 1100 | return PyList_New(0); |
| 1101 | |
| 1102 | list = PyList_New(self->extra->length); |
| 1103 | if (!list) |
| 1104 | return NULL; |
| 1105 | |
| 1106 | for (i = 0; i < self->extra->length; i++) { |
| 1107 | PyObject* item = self->extra->children[i]; |
| 1108 | Py_INCREF(item); |
| 1109 | PyList_SET_ITEM(list, i, item); |
| 1110 | } |
| 1111 | |
| 1112 | return list; |
| 1113 | } |
| 1114 | |
| 1115 | static PyObject* |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1116 | element_iter(ElementObject* self, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1117 | { |
| 1118 | PyObject* result; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1119 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1120 | PyObject* tag = Py_None; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1121 | if (!PyArg_ParseTuple(args, "|O:iter", &tag)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1122 | return NULL; |
| 1123 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1124 | if (!elementtree_iter_obj) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1125 | PyErr_SetString( |
| 1126 | PyExc_RuntimeError, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1127 | "iter helper not found" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1128 | ); |
| 1129 | return NULL; |
| 1130 | } |
| 1131 | |
| 1132 | args = PyTuple_New(2); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1133 | if (!args) |
| 1134 | return NULL; |
Neal Norwitz | 02876df | 2006-02-07 06:58:52 +0000 | [diff] [blame] | 1135 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1136 | Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self); |
| 1137 | Py_INCREF(tag); PyTuple_SET_ITEM(args, 1, (PyObject*) tag); |
| 1138 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1139 | result = PyObject_CallObject(elementtree_iter_obj, args); |
| 1140 | |
| 1141 | Py_DECREF(args); |
| 1142 | |
| 1143 | return result; |
| 1144 | } |
| 1145 | |
| 1146 | |
| 1147 | static PyObject* |
| 1148 | element_itertext(ElementObject* self, PyObject* args) |
| 1149 | { |
| 1150 | PyObject* result; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1151 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1152 | if (!PyArg_ParseTuple(args, ":itertext")) |
| 1153 | return NULL; |
| 1154 | |
| 1155 | if (!elementtree_itertext_obj) { |
| 1156 | PyErr_SetString( |
| 1157 | PyExc_RuntimeError, |
| 1158 | "itertext helper not found" |
| 1159 | ); |
| 1160 | return NULL; |
| 1161 | } |
| 1162 | |
| 1163 | args = PyTuple_New(1); |
| 1164 | if (!args) |
| 1165 | return NULL; |
| 1166 | |
| 1167 | Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self); |
| 1168 | |
| 1169 | result = PyObject_CallObject(elementtree_itertext_obj, args); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1170 | |
| 1171 | Py_DECREF(args); |
| 1172 | |
| 1173 | return result; |
| 1174 | } |
| 1175 | |
| 1176 | static PyObject* |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1177 | element_getitem(PyObject* self_, Py_ssize_t index) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1178 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1179 | ElementObject* self = (ElementObject*) self_; |
| 1180 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1181 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1182 | PyErr_SetString( |
| 1183 | PyExc_IndexError, |
| 1184 | "child index out of range" |
| 1185 | ); |
| 1186 | return NULL; |
| 1187 | } |
| 1188 | |
| 1189 | Py_INCREF(self->extra->children[index]); |
| 1190 | return self->extra->children[index]; |
| 1191 | } |
| 1192 | |
| 1193 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1194 | element_insert(ElementObject* self, PyObject* args) |
| 1195 | { |
| 1196 | int i; |
| 1197 | |
| 1198 | int index; |
| 1199 | PyObject* element; |
| 1200 | if (!PyArg_ParseTuple(args, "iO!:insert", &index, |
| 1201 | &Element_Type, &element)) |
| 1202 | return NULL; |
| 1203 | |
| 1204 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1205 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1206 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1207 | if (index < 0) { |
| 1208 | index += self->extra->length; |
| 1209 | if (index < 0) |
| 1210 | index = 0; |
| 1211 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1212 | if (index > self->extra->length) |
| 1213 | index = self->extra->length; |
| 1214 | |
| 1215 | if (element_resize(self, 1) < 0) |
| 1216 | return NULL; |
| 1217 | |
| 1218 | for (i = self->extra->length; i > index; i--) |
| 1219 | self->extra->children[i] = self->extra->children[i-1]; |
| 1220 | |
| 1221 | Py_INCREF(element); |
| 1222 | self->extra->children[index] = element; |
| 1223 | |
| 1224 | self->extra->length++; |
| 1225 | |
| 1226 | Py_RETURN_NONE; |
| 1227 | } |
| 1228 | |
| 1229 | static PyObject* |
| 1230 | element_items(ElementObject* self, PyObject* args) |
| 1231 | { |
| 1232 | if (!PyArg_ParseTuple(args, ":items")) |
| 1233 | return NULL; |
| 1234 | |
| 1235 | if (!self->extra || self->extra->attrib == Py_None) |
| 1236 | return PyList_New(0); |
| 1237 | |
| 1238 | return PyDict_Items(self->extra->attrib); |
| 1239 | } |
| 1240 | |
| 1241 | static PyObject* |
| 1242 | element_keys(ElementObject* self, PyObject* args) |
| 1243 | { |
| 1244 | if (!PyArg_ParseTuple(args, ":keys")) |
| 1245 | return NULL; |
| 1246 | |
| 1247 | if (!self->extra || self->extra->attrib == Py_None) |
| 1248 | return PyList_New(0); |
| 1249 | |
| 1250 | return PyDict_Keys(self->extra->attrib); |
| 1251 | } |
| 1252 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1253 | static Py_ssize_t |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1254 | element_length(ElementObject* self) |
| 1255 | { |
| 1256 | if (!self->extra) |
| 1257 | return 0; |
| 1258 | |
| 1259 | return self->extra->length; |
| 1260 | } |
| 1261 | |
| 1262 | static PyObject* |
| 1263 | element_makeelement(PyObject* self, PyObject* args, PyObject* kw) |
| 1264 | { |
| 1265 | PyObject* elem; |
| 1266 | |
| 1267 | PyObject* tag; |
| 1268 | PyObject* attrib; |
| 1269 | if (!PyArg_ParseTuple(args, "OO:makeelement", &tag, &attrib)) |
| 1270 | return NULL; |
| 1271 | |
| 1272 | attrib = PyDict_Copy(attrib); |
| 1273 | if (!attrib) |
| 1274 | return NULL; |
| 1275 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1276 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1277 | |
| 1278 | Py_DECREF(attrib); |
| 1279 | |
| 1280 | return elem; |
| 1281 | } |
| 1282 | |
| 1283 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1284 | element_remove(ElementObject* self, PyObject* args) |
| 1285 | { |
| 1286 | int i; |
| 1287 | |
| 1288 | PyObject* element; |
| 1289 | if (!PyArg_ParseTuple(args, "O!:remove", &Element_Type, &element)) |
| 1290 | return NULL; |
| 1291 | |
| 1292 | if (!self->extra) { |
| 1293 | /* element has no children, so raise exception */ |
| 1294 | PyErr_SetString( |
| 1295 | PyExc_ValueError, |
| 1296 | "list.remove(x): x not in list" |
| 1297 | ); |
| 1298 | return NULL; |
| 1299 | } |
| 1300 | |
| 1301 | for (i = 0; i < self->extra->length; i++) { |
| 1302 | if (self->extra->children[i] == element) |
| 1303 | break; |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1304 | if (PyObject_RichCompareBool(self->extra->children[i], element, Py_EQ) == 1) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1305 | break; |
| 1306 | } |
| 1307 | |
| 1308 | if (i == self->extra->length) { |
| 1309 | /* element is not in children, so raise exception */ |
| 1310 | PyErr_SetString( |
| 1311 | PyExc_ValueError, |
| 1312 | "list.remove(x): x not in list" |
| 1313 | ); |
| 1314 | return NULL; |
| 1315 | } |
| 1316 | |
| 1317 | Py_DECREF(self->extra->children[i]); |
| 1318 | |
| 1319 | self->extra->length--; |
| 1320 | |
| 1321 | for (; i < self->extra->length; i++) |
| 1322 | self->extra->children[i] = self->extra->children[i+1]; |
| 1323 | |
| 1324 | Py_RETURN_NONE; |
| 1325 | } |
| 1326 | |
| 1327 | static PyObject* |
| 1328 | element_repr(ElementObject* self) |
| 1329 | { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1330 | if (self->tag) |
| 1331 | return PyUnicode_FromFormat("<Element %R at %p>", self->tag, self); |
| 1332 | else |
| 1333 | return PyUnicode_FromFormat("<Element at %p>", self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1334 | } |
| 1335 | |
| 1336 | static PyObject* |
| 1337 | element_set(ElementObject* self, PyObject* args) |
| 1338 | { |
| 1339 | PyObject* attrib; |
| 1340 | |
| 1341 | PyObject* key; |
| 1342 | PyObject* value; |
| 1343 | if (!PyArg_ParseTuple(args, "OO:set", &key, &value)) |
| 1344 | return NULL; |
| 1345 | |
| 1346 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1347 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1348 | |
| 1349 | attrib = element_get_attrib(self); |
| 1350 | if (!attrib) |
| 1351 | return NULL; |
| 1352 | |
| 1353 | if (PyDict_SetItem(attrib, key, value) < 0) |
| 1354 | return NULL; |
| 1355 | |
| 1356 | Py_RETURN_NONE; |
| 1357 | } |
| 1358 | |
| 1359 | static int |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1360 | element_setitem(PyObject* self_, Py_ssize_t index, PyObject* item) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1361 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1362 | ElementObject* self = (ElementObject*) self_; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1363 | int i; |
| 1364 | PyObject* old; |
| 1365 | |
| 1366 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1367 | PyErr_SetString( |
| 1368 | PyExc_IndexError, |
| 1369 | "child assignment index out of range"); |
| 1370 | return -1; |
| 1371 | } |
| 1372 | |
| 1373 | old = self->extra->children[index]; |
| 1374 | |
| 1375 | if (item) { |
| 1376 | Py_INCREF(item); |
| 1377 | self->extra->children[index] = item; |
| 1378 | } else { |
| 1379 | self->extra->length--; |
| 1380 | for (i = index; i < self->extra->length; i++) |
| 1381 | self->extra->children[i] = self->extra->children[i+1]; |
| 1382 | } |
| 1383 | |
| 1384 | Py_DECREF(old); |
| 1385 | |
| 1386 | return 0; |
| 1387 | } |
| 1388 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1389 | static PyObject* |
| 1390 | element_subscr(PyObject* self_, PyObject* item) |
| 1391 | { |
| 1392 | ElementObject* self = (ElementObject*) self_; |
| 1393 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1394 | if (PyIndex_Check(item)) { |
| 1395 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1396 | |
| 1397 | if (i == -1 && PyErr_Occurred()) { |
| 1398 | return NULL; |
| 1399 | } |
| 1400 | if (i < 0 && self->extra) |
| 1401 | i += self->extra->length; |
| 1402 | return element_getitem(self_, i); |
| 1403 | } |
| 1404 | else if (PySlice_Check(item)) { |
| 1405 | Py_ssize_t start, stop, step, slicelen, cur, i; |
| 1406 | PyObject* list; |
| 1407 | |
| 1408 | if (!self->extra) |
| 1409 | return PyList_New(0); |
| 1410 | |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 1411 | if (PySlice_GetIndicesEx(item, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1412 | self->extra->length, |
| 1413 | &start, &stop, &step, &slicelen) < 0) { |
| 1414 | return NULL; |
| 1415 | } |
| 1416 | |
| 1417 | if (slicelen <= 0) |
| 1418 | return PyList_New(0); |
| 1419 | else { |
| 1420 | list = PyList_New(slicelen); |
| 1421 | if (!list) |
| 1422 | return NULL; |
| 1423 | |
| 1424 | for (cur = start, i = 0; i < slicelen; |
| 1425 | cur += step, i++) { |
| 1426 | PyObject* item = self->extra->children[cur]; |
| 1427 | Py_INCREF(item); |
| 1428 | PyList_SET_ITEM(list, i, item); |
| 1429 | } |
| 1430 | |
| 1431 | return list; |
| 1432 | } |
| 1433 | } |
| 1434 | else { |
| 1435 | PyErr_SetString(PyExc_TypeError, |
| 1436 | "element indices must be integers"); |
| 1437 | return NULL; |
| 1438 | } |
| 1439 | } |
| 1440 | |
| 1441 | static int |
| 1442 | element_ass_subscr(PyObject* self_, PyObject* item, PyObject* value) |
| 1443 | { |
| 1444 | ElementObject* self = (ElementObject*) self_; |
| 1445 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1446 | if (PyIndex_Check(item)) { |
| 1447 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1448 | |
| 1449 | if (i == -1 && PyErr_Occurred()) { |
| 1450 | return -1; |
| 1451 | } |
| 1452 | if (i < 0 && self->extra) |
| 1453 | i += self->extra->length; |
| 1454 | return element_setitem(self_, i, value); |
| 1455 | } |
| 1456 | else if (PySlice_Check(item)) { |
| 1457 | Py_ssize_t start, stop, step, slicelen, newlen, cur, i; |
| 1458 | |
| 1459 | PyObject* recycle = NULL; |
| 1460 | PyObject* seq = NULL; |
| 1461 | |
| 1462 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1463 | create_extra(self, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1464 | |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 1465 | if (PySlice_GetIndicesEx(item, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1466 | self->extra->length, |
| 1467 | &start, &stop, &step, &slicelen) < 0) { |
| 1468 | return -1; |
| 1469 | } |
| 1470 | |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1471 | if (value == NULL) { |
| 1472 | /* Delete slice */ |
| 1473 | size_t cur; |
| 1474 | Py_ssize_t i; |
| 1475 | |
| 1476 | if (slicelen <= 0) |
| 1477 | return 0; |
| 1478 | |
| 1479 | /* Since we're deleting, the direction of the range doesn't matter, |
| 1480 | * so for simplicity make it always ascending. |
| 1481 | */ |
| 1482 | if (step < 0) { |
| 1483 | stop = start + 1; |
| 1484 | start = stop + step * (slicelen - 1) - 1; |
| 1485 | step = -step; |
| 1486 | } |
| 1487 | |
| 1488 | assert((size_t)slicelen <= PY_SIZE_MAX / sizeof(PyObject *)); |
| 1489 | |
| 1490 | /* recycle is a list that will contain all the children |
| 1491 | * scheduled for removal. |
| 1492 | */ |
| 1493 | if (!(recycle = PyList_New(slicelen))) { |
| 1494 | PyErr_NoMemory(); |
| 1495 | return -1; |
| 1496 | } |
| 1497 | |
| 1498 | /* This loop walks over all the children that have to be deleted, |
| 1499 | * with cur pointing at them. num_moved is the amount of children |
| 1500 | * until the next deleted child that have to be "shifted down" to |
| 1501 | * occupy the deleted's places. |
| 1502 | * Note that in the ith iteration, shifting is done i+i places down |
| 1503 | * because i children were already removed. |
| 1504 | */ |
| 1505 | for (cur = start, i = 0; cur < (size_t)stop; cur += step, ++i) { |
| 1506 | /* Compute how many children have to be moved, clipping at the |
| 1507 | * list end. |
| 1508 | */ |
| 1509 | Py_ssize_t num_moved = step - 1; |
| 1510 | if (cur + step >= (size_t)self->extra->length) { |
| 1511 | num_moved = self->extra->length - cur - 1; |
| 1512 | } |
| 1513 | |
| 1514 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1515 | |
| 1516 | memmove( |
| 1517 | self->extra->children + cur - i, |
| 1518 | self->extra->children + cur + 1, |
| 1519 | num_moved * sizeof(PyObject *)); |
| 1520 | } |
| 1521 | |
| 1522 | /* Leftover "tail" after the last removed child */ |
| 1523 | cur = start + (size_t)slicelen * step; |
| 1524 | if (cur < (size_t)self->extra->length) { |
| 1525 | memmove( |
| 1526 | self->extra->children + cur - slicelen, |
| 1527 | self->extra->children + cur, |
| 1528 | (self->extra->length - cur) * sizeof(PyObject *)); |
| 1529 | } |
| 1530 | |
| 1531 | self->extra->length -= slicelen; |
| 1532 | |
| 1533 | /* Discard the recycle list with all the deleted sub-elements */ |
| 1534 | Py_XDECREF(recycle); |
| 1535 | return 0; |
| 1536 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1537 | else { |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1538 | /* A new slice is actually being assigned */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1539 | seq = PySequence_Fast(value, ""); |
| 1540 | if (!seq) { |
| 1541 | PyErr_Format( |
| 1542 | PyExc_TypeError, |
| 1543 | "expected sequence, not \"%.200s\"", Py_TYPE(value)->tp_name |
| 1544 | ); |
| 1545 | return -1; |
| 1546 | } |
| 1547 | newlen = PySequence_Size(seq); |
| 1548 | } |
| 1549 | |
| 1550 | if (step != 1 && newlen != slicelen) |
| 1551 | { |
| 1552 | PyErr_Format(PyExc_ValueError, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1553 | "attempt to assign sequence of size %zd " |
| 1554 | "to extended slice of size %zd", |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1555 | newlen, slicelen |
| 1556 | ); |
| 1557 | return -1; |
| 1558 | } |
| 1559 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1560 | /* Resize before creating the recycle bin, to prevent refleaks. */ |
| 1561 | if (newlen > slicelen) { |
| 1562 | if (element_resize(self, newlen - slicelen) < 0) { |
| 1563 | if (seq) { |
| 1564 | Py_DECREF(seq); |
| 1565 | } |
| 1566 | return -1; |
| 1567 | } |
| 1568 | } |
| 1569 | |
| 1570 | if (slicelen > 0) { |
| 1571 | /* to avoid recursive calls to this method (via decref), move |
| 1572 | old items to the recycle bin here, and get rid of them when |
| 1573 | we're done modifying the element */ |
| 1574 | recycle = PyList_New(slicelen); |
| 1575 | if (!recycle) { |
| 1576 | if (seq) { |
| 1577 | Py_DECREF(seq); |
| 1578 | } |
| 1579 | return -1; |
| 1580 | } |
| 1581 | for (cur = start, i = 0; i < slicelen; |
| 1582 | cur += step, i++) |
| 1583 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1584 | } |
| 1585 | |
| 1586 | if (newlen < slicelen) { |
| 1587 | /* delete slice */ |
| 1588 | for (i = stop; i < self->extra->length; i++) |
| 1589 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1590 | } else if (newlen > slicelen) { |
| 1591 | /* insert slice */ |
| 1592 | for (i = self->extra->length-1; i >= stop; i--) |
| 1593 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1594 | } |
| 1595 | |
| 1596 | /* replace the slice */ |
| 1597 | for (cur = start, i = 0; i < newlen; |
| 1598 | cur += step, i++) { |
| 1599 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
| 1600 | Py_INCREF(element); |
| 1601 | self->extra->children[cur] = element; |
| 1602 | } |
| 1603 | |
| 1604 | self->extra->length += newlen - slicelen; |
| 1605 | |
| 1606 | if (seq) { |
| 1607 | Py_DECREF(seq); |
| 1608 | } |
| 1609 | |
| 1610 | /* discard the recycle bin, and everything in it */ |
| 1611 | Py_XDECREF(recycle); |
| 1612 | |
| 1613 | return 0; |
| 1614 | } |
| 1615 | else { |
| 1616 | PyErr_SetString(PyExc_TypeError, |
| 1617 | "element indices must be integers"); |
| 1618 | return -1; |
| 1619 | } |
| 1620 | } |
| 1621 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1622 | static PyMethodDef element_methods[] = { |
| 1623 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1624 | {"clear", (PyCFunction) element_clearmethod, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1625 | |
| 1626 | {"get", (PyCFunction) element_get, METH_VARARGS}, |
| 1627 | {"set", (PyCFunction) element_set, METH_VARARGS}, |
| 1628 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1629 | {"find", (PyCFunction) element_find, METH_VARARGS | METH_KEYWORDS}, |
| 1630 | {"findtext", (PyCFunction) element_findtext, METH_VARARGS | METH_KEYWORDS}, |
| 1631 | {"findall", (PyCFunction) element_findall, METH_VARARGS | METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1632 | |
| 1633 | {"append", (PyCFunction) element_append, METH_VARARGS}, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1634 | {"extend", (PyCFunction) element_extend, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1635 | {"insert", (PyCFunction) element_insert, METH_VARARGS}, |
| 1636 | {"remove", (PyCFunction) element_remove, METH_VARARGS}, |
| 1637 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1638 | {"iter", (PyCFunction) element_iter, METH_VARARGS}, |
| 1639 | {"itertext", (PyCFunction) element_itertext, METH_VARARGS}, |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1640 | {"iterfind", (PyCFunction) element_iterfind, METH_VARARGS | METH_KEYWORDS}, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1641 | |
| 1642 | {"getiterator", (PyCFunction) element_iter, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1643 | {"getchildren", (PyCFunction) element_getchildren, METH_VARARGS}, |
| 1644 | |
| 1645 | {"items", (PyCFunction) element_items, METH_VARARGS}, |
| 1646 | {"keys", (PyCFunction) element_keys, METH_VARARGS}, |
| 1647 | |
| 1648 | {"makeelement", (PyCFunction) element_makeelement, METH_VARARGS}, |
| 1649 | |
| 1650 | {"__copy__", (PyCFunction) element_copy, METH_VARARGS}, |
| 1651 | {"__deepcopy__", (PyCFunction) element_deepcopy, METH_VARARGS}, |
| 1652 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1653 | {NULL, NULL} |
| 1654 | }; |
| 1655 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1656 | static PyObject* |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1657 | element_getattro(ElementObject* self, PyObject* nameobj) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1658 | { |
| 1659 | PyObject* res; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1660 | char *name = ""; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1661 | |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1662 | if (PyUnicode_Check(nameobj)) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1663 | name = _PyUnicode_AsString(nameobj); |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1664 | |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 1665 | if (name == NULL) |
| 1666 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1667 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1668 | /* handle common attributes first */ |
| 1669 | if (strcmp(name, "tag") == 0) { |
| 1670 | res = self->tag; |
| 1671 | Py_INCREF(res); |
| 1672 | return res; |
| 1673 | } else if (strcmp(name, "text") == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1674 | res = element_get_text(self); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1675 | Py_INCREF(res); |
| 1676 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1677 | } |
| 1678 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1679 | /* methods */ |
| 1680 | res = PyObject_GenericGetAttr((PyObject*) self, nameobj); |
| 1681 | if (res) |
| 1682 | return res; |
| 1683 | |
| 1684 | /* less common attributes */ |
| 1685 | if (strcmp(name, "tail") == 0) { |
| 1686 | PyErr_Clear(); |
| 1687 | res = element_get_tail(self); |
| 1688 | } else if (strcmp(name, "attrib") == 0) { |
| 1689 | PyErr_Clear(); |
| 1690 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1691 | create_extra(self, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1692 | res = element_get_attrib(self); |
| 1693 | } |
| 1694 | |
| 1695 | if (!res) |
| 1696 | return NULL; |
| 1697 | |
| 1698 | Py_INCREF(res); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1699 | return res; |
| 1700 | } |
| 1701 | |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1702 | static PyObject* |
| 1703 | element_setattro(ElementObject* self, PyObject* nameobj, PyObject* value) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1704 | { |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1705 | char *name = ""; |
| 1706 | if (PyUnicode_Check(nameobj)) |
| 1707 | name = _PyUnicode_AsString(nameobj); |
| 1708 | |
| 1709 | if (name == NULL) |
| 1710 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1711 | |
| 1712 | if (strcmp(name, "tag") == 0) { |
| 1713 | Py_DECREF(self->tag); |
| 1714 | self->tag = value; |
| 1715 | Py_INCREF(self->tag); |
| 1716 | } else if (strcmp(name, "text") == 0) { |
| 1717 | Py_DECREF(JOIN_OBJ(self->text)); |
| 1718 | self->text = value; |
| 1719 | Py_INCREF(self->text); |
| 1720 | } else if (strcmp(name, "tail") == 0) { |
| 1721 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 1722 | self->tail = value; |
| 1723 | Py_INCREF(self->tail); |
| 1724 | } else if (strcmp(name, "attrib") == 0) { |
| 1725 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1726 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1727 | Py_DECREF(self->extra->attrib); |
| 1728 | self->extra->attrib = value; |
| 1729 | Py_INCREF(self->extra->attrib); |
| 1730 | } else { |
| 1731 | PyErr_SetString(PyExc_AttributeError, name); |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1732 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1733 | } |
| 1734 | |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1735 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1736 | } |
| 1737 | |
| 1738 | static PySequenceMethods element_as_sequence = { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1739 | (lenfunc) element_length, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1740 | 0, /* sq_concat */ |
| 1741 | 0, /* sq_repeat */ |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1742 | element_getitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1743 | 0, |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1744 | element_setitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1745 | 0, |
| 1746 | }; |
| 1747 | |
| 1748 | static PyMappingMethods element_as_mapping = { |
| 1749 | (lenfunc) element_length, |
| 1750 | (binaryfunc) element_subscr, |
| 1751 | (objobjargproc) element_ass_subscr, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1752 | }; |
| 1753 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 1754 | static PyTypeObject Element_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 1755 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1756 | "Element", sizeof(ElementObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1757 | /* methods */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1758 | (destructor)element_dealloc, /* tp_dealloc */ |
| 1759 | 0, /* tp_print */ |
| 1760 | 0, /* tp_getattr */ |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1761 | 0, /* tp_setattr */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1762 | 0, /* tp_reserved */ |
| 1763 | (reprfunc)element_repr, /* tp_repr */ |
| 1764 | 0, /* tp_as_number */ |
| 1765 | &element_as_sequence, /* tp_as_sequence */ |
| 1766 | &element_as_mapping, /* tp_as_mapping */ |
| 1767 | 0, /* tp_hash */ |
| 1768 | 0, /* tp_call */ |
| 1769 | 0, /* tp_str */ |
| 1770 | (getattrofunc)element_getattro, /* tp_getattro */ |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1771 | (setattrofunc)element_setattro, /* tp_setattro */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1772 | 0, /* tp_as_buffer */ |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1773 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 1774 | /* tp_flags */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1775 | 0, /* tp_doc */ |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1776 | (traverseproc)element_gc_traverse, /* tp_traverse */ |
| 1777 | (inquiry)element_gc_clear, /* tp_clear */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1778 | 0, /* tp_richcompare */ |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 1779 | offsetof(ElementObject, weakreflist), /* tp_weaklistoffset */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1780 | 0, /* tp_iter */ |
| 1781 | 0, /* tp_iternext */ |
| 1782 | element_methods, /* tp_methods */ |
| 1783 | 0, /* tp_members */ |
| 1784 | 0, /* tp_getset */ |
| 1785 | 0, /* tp_base */ |
| 1786 | 0, /* tp_dict */ |
| 1787 | 0, /* tp_descr_get */ |
| 1788 | 0, /* tp_descr_set */ |
| 1789 | 0, /* tp_dictoffset */ |
| 1790 | (initproc)element_init, /* tp_init */ |
| 1791 | PyType_GenericAlloc, /* tp_alloc */ |
| 1792 | element_new, /* tp_new */ |
| 1793 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1794 | }; |
| 1795 | |
| 1796 | /* ==================================================================== */ |
| 1797 | /* the tree builder type */ |
| 1798 | |
| 1799 | typedef struct { |
| 1800 | PyObject_HEAD |
| 1801 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1802 | PyObject *root; /* root node (first created node) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1803 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1804 | ElementObject *this; /* current node */ |
| 1805 | ElementObject *last; /* most recently created node */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1806 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1807 | PyObject *data; /* data collector (string or list), or NULL */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1808 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1809 | PyObject *stack; /* element stack */ |
| 1810 | Py_ssize_t index; /* current stack size (0 means empty) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1811 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1812 | PyObject *element_factory; |
| 1813 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1814 | /* element tracing */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1815 | PyObject *events; /* list of events, or NULL if not collecting */ |
| 1816 | PyObject *start_event_obj; /* event objects (NULL to ignore) */ |
| 1817 | PyObject *end_event_obj; |
| 1818 | PyObject *start_ns_event_obj; |
| 1819 | PyObject *end_ns_event_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1820 | } TreeBuilderObject; |
| 1821 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 1822 | static PyTypeObject TreeBuilder_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1823 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 1824 | #define TreeBuilder_CheckExact(op) (Py_TYPE(op) == &TreeBuilder_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1825 | |
| 1826 | /* -------------------------------------------------------------------- */ |
| 1827 | /* constructor and destructor */ |
| 1828 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1829 | static PyObject * |
| 1830 | treebuilder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1831 | { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1832 | TreeBuilderObject *t = (TreeBuilderObject *)type->tp_alloc(type, 0); |
| 1833 | if (t != NULL) { |
| 1834 | t->root = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1835 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1836 | Py_INCREF(Py_None); |
| 1837 | t->this = (ElementObject *)Py_None; |
| 1838 | Py_INCREF(Py_None); |
| 1839 | t->last = (ElementObject *)Py_None; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1840 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1841 | t->data = NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1842 | t->element_factory = NULL; |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1843 | t->stack = PyList_New(20); |
| 1844 | if (!t->stack) { |
| 1845 | Py_DECREF(t->this); |
| 1846 | Py_DECREF(t->last); |
| 1847 | return NULL; |
| 1848 | } |
| 1849 | t->index = 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1850 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1851 | t->events = NULL; |
| 1852 | t->start_event_obj = t->end_event_obj = NULL; |
| 1853 | t->start_ns_event_obj = t->end_ns_event_obj = NULL; |
| 1854 | } |
| 1855 | return (PyObject *)t; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1856 | } |
| 1857 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1858 | static int |
| 1859 | treebuilder_init(PyObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1860 | { |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1861 | static char *kwlist[] = {"element_factory", NULL}; |
| 1862 | PyObject *element_factory = NULL; |
| 1863 | TreeBuilderObject *self_tb = (TreeBuilderObject *)self; |
| 1864 | |
| 1865 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O:TreeBuilder", kwlist, |
| 1866 | &element_factory)) { |
| 1867 | return -1; |
| 1868 | } |
| 1869 | |
| 1870 | if (element_factory) { |
| 1871 | Py_INCREF(element_factory); |
| 1872 | Py_XDECREF(self_tb->element_factory); |
| 1873 | self_tb->element_factory = element_factory; |
| 1874 | } |
| 1875 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1876 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1877 | } |
| 1878 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1879 | static int |
| 1880 | treebuilder_gc_traverse(TreeBuilderObject *self, visitproc visit, void *arg) |
| 1881 | { |
| 1882 | Py_VISIT(self->root); |
| 1883 | Py_VISIT(self->this); |
| 1884 | Py_VISIT(self->last); |
| 1885 | Py_VISIT(self->data); |
| 1886 | Py_VISIT(self->stack); |
| 1887 | Py_VISIT(self->element_factory); |
| 1888 | return 0; |
| 1889 | } |
| 1890 | |
| 1891 | static int |
| 1892 | treebuilder_gc_clear(TreeBuilderObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1893 | { |
| 1894 | Py_XDECREF(self->end_ns_event_obj); |
| 1895 | Py_XDECREF(self->start_ns_event_obj); |
| 1896 | Py_XDECREF(self->end_event_obj); |
| 1897 | Py_XDECREF(self->start_event_obj); |
| 1898 | Py_XDECREF(self->events); |
| 1899 | Py_DECREF(self->stack); |
| 1900 | Py_XDECREF(self->data); |
| 1901 | Py_DECREF(self->last); |
| 1902 | Py_DECREF(self->this); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1903 | Py_CLEAR(self->element_factory); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1904 | Py_XDECREF(self->root); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1905 | return 0; |
| 1906 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1907 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1908 | static void |
| 1909 | treebuilder_dealloc(TreeBuilderObject *self) |
| 1910 | { |
| 1911 | PyObject_GC_UnTrack(self); |
| 1912 | treebuilder_gc_clear(self); |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1913 | Py_TYPE(self)->tp_free((PyObject *)self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1914 | } |
| 1915 | |
| 1916 | /* -------------------------------------------------------------------- */ |
| 1917 | /* handlers */ |
| 1918 | |
| 1919 | LOCAL(PyObject*) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1920 | treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag, |
| 1921 | PyObject* attrib) |
| 1922 | { |
| 1923 | PyObject* node; |
| 1924 | PyObject* this; |
| 1925 | |
| 1926 | if (self->data) { |
| 1927 | if (self->this == self->last) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1928 | Py_DECREF(JOIN_OBJ(self->last->text)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1929 | self->last->text = JOIN_SET( |
| 1930 | self->data, PyList_CheckExact(self->data) |
| 1931 | ); |
| 1932 | } else { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1933 | Py_DECREF(JOIN_OBJ(self->last->tail)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1934 | self->last->tail = JOIN_SET( |
| 1935 | self->data, PyList_CheckExact(self->data) |
| 1936 | ); |
| 1937 | } |
| 1938 | self->data = NULL; |
| 1939 | } |
| 1940 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1941 | if (self->element_factory) { |
| 1942 | node = PyObject_CallFunction(self->element_factory, "OO", tag, attrib); |
| 1943 | } else { |
| 1944 | node = create_new_element(tag, attrib); |
| 1945 | } |
| 1946 | if (!node) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1947 | return NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1948 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1949 | |
| 1950 | this = (PyObject*) self->this; |
| 1951 | |
| 1952 | if (this != Py_None) { |
| 1953 | if (element_add_subelement((ElementObject*) this, node) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1954 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1955 | } else { |
| 1956 | if (self->root) { |
| 1957 | PyErr_SetString( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1958 | elementtree_parseerror_obj, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1959 | "multiple elements on top level" |
| 1960 | ); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1961 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1962 | } |
| 1963 | Py_INCREF(node); |
| 1964 | self->root = node; |
| 1965 | } |
| 1966 | |
| 1967 | if (self->index < PyList_GET_SIZE(self->stack)) { |
| 1968 | if (PyList_SetItem(self->stack, self->index, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1969 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1970 | Py_INCREF(this); |
| 1971 | } else { |
| 1972 | if (PyList_Append(self->stack, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1973 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1974 | } |
| 1975 | self->index++; |
| 1976 | |
| 1977 | Py_DECREF(this); |
| 1978 | Py_INCREF(node); |
| 1979 | self->this = (ElementObject*) node; |
| 1980 | |
| 1981 | Py_DECREF(self->last); |
| 1982 | Py_INCREF(node); |
| 1983 | self->last = (ElementObject*) node; |
| 1984 | |
| 1985 | if (self->start_event_obj) { |
| 1986 | PyObject* res; |
| 1987 | PyObject* action = self->start_event_obj; |
| 1988 | res = PyTuple_New(2); |
| 1989 | if (res) { |
| 1990 | Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action); |
| 1991 | Py_INCREF(node); PyTuple_SET_ITEM(res, 1, (PyObject*) node); |
| 1992 | PyList_Append(self->events, res); |
| 1993 | Py_DECREF(res); |
| 1994 | } else |
| 1995 | PyErr_Clear(); /* FIXME: propagate error */ |
| 1996 | } |
| 1997 | |
| 1998 | return node; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1999 | |
| 2000 | error: |
| 2001 | Py_DECREF(node); |
| 2002 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2003 | } |
| 2004 | |
| 2005 | LOCAL(PyObject*) |
| 2006 | treebuilder_handle_data(TreeBuilderObject* self, PyObject* data) |
| 2007 | { |
| 2008 | if (!self->data) { |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 2009 | if (self->last == (ElementObject*) Py_None) { |
| 2010 | /* ignore calls to data before the first call to start */ |
| 2011 | Py_RETURN_NONE; |
| 2012 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2013 | /* store the first item as is */ |
| 2014 | Py_INCREF(data); self->data = data; |
| 2015 | } else { |
| 2016 | /* more than one item; use a list to collect items */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2017 | if (PyBytes_CheckExact(self->data) && Py_REFCNT(self->data) == 1 && |
| 2018 | PyBytes_CheckExact(data) && PyBytes_GET_SIZE(data) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2019 | /* expat often generates single character data sections; handle |
| 2020 | the most common case by resizing the existing string... */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2021 | Py_ssize_t size = PyBytes_GET_SIZE(self->data); |
| 2022 | if (_PyBytes_Resize(&self->data, size + 1) < 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2023 | return NULL; |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2024 | PyBytes_AS_STRING(self->data)[size] = PyBytes_AS_STRING(data)[0]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2025 | } else if (PyList_CheckExact(self->data)) { |
| 2026 | if (PyList_Append(self->data, data) < 0) |
| 2027 | return NULL; |
| 2028 | } else { |
| 2029 | PyObject* list = PyList_New(2); |
| 2030 | if (!list) |
| 2031 | return NULL; |
| 2032 | PyList_SET_ITEM(list, 0, self->data); |
| 2033 | Py_INCREF(data); PyList_SET_ITEM(list, 1, data); |
| 2034 | self->data = list; |
| 2035 | } |
| 2036 | } |
| 2037 | |
| 2038 | Py_RETURN_NONE; |
| 2039 | } |
| 2040 | |
| 2041 | LOCAL(PyObject*) |
| 2042 | treebuilder_handle_end(TreeBuilderObject* self, PyObject* tag) |
| 2043 | { |
| 2044 | PyObject* item; |
| 2045 | |
| 2046 | if (self->data) { |
| 2047 | if (self->this == self->last) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2048 | Py_DECREF(JOIN_OBJ(self->last->text)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2049 | self->last->text = JOIN_SET( |
| 2050 | self->data, PyList_CheckExact(self->data) |
| 2051 | ); |
| 2052 | } else { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2053 | Py_DECREF(JOIN_OBJ(self->last->tail)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2054 | self->last->tail = JOIN_SET( |
| 2055 | self->data, PyList_CheckExact(self->data) |
| 2056 | ); |
| 2057 | } |
| 2058 | self->data = NULL; |
| 2059 | } |
| 2060 | |
| 2061 | if (self->index == 0) { |
| 2062 | PyErr_SetString( |
| 2063 | PyExc_IndexError, |
| 2064 | "pop from empty stack" |
| 2065 | ); |
| 2066 | return NULL; |
| 2067 | } |
| 2068 | |
| 2069 | self->index--; |
| 2070 | |
| 2071 | item = PyList_GET_ITEM(self->stack, self->index); |
| 2072 | Py_INCREF(item); |
| 2073 | |
| 2074 | Py_DECREF(self->last); |
| 2075 | |
| 2076 | self->last = (ElementObject*) self->this; |
| 2077 | self->this = (ElementObject*) item; |
| 2078 | |
| 2079 | if (self->end_event_obj) { |
| 2080 | PyObject* res; |
| 2081 | PyObject* action = self->end_event_obj; |
| 2082 | PyObject* node = (PyObject*) self->last; |
| 2083 | res = PyTuple_New(2); |
| 2084 | if (res) { |
| 2085 | Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action); |
| 2086 | Py_INCREF(node); PyTuple_SET_ITEM(res, 1, (PyObject*) node); |
| 2087 | PyList_Append(self->events, res); |
| 2088 | Py_DECREF(res); |
| 2089 | } else |
| 2090 | PyErr_Clear(); /* FIXME: propagate error */ |
| 2091 | } |
| 2092 | |
| 2093 | Py_INCREF(self->last); |
| 2094 | return (PyObject*) self->last; |
| 2095 | } |
| 2096 | |
| 2097 | LOCAL(void) |
| 2098 | treebuilder_handle_namespace(TreeBuilderObject* self, int start, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2099 | PyObject *prefix, PyObject *uri) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2100 | { |
| 2101 | PyObject* res; |
| 2102 | PyObject* action; |
| 2103 | PyObject* parcel; |
| 2104 | |
| 2105 | if (!self->events) |
| 2106 | return; |
| 2107 | |
| 2108 | if (start) { |
| 2109 | if (!self->start_ns_event_obj) |
| 2110 | return; |
| 2111 | action = self->start_ns_event_obj; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2112 | parcel = Py_BuildValue("OO", prefix, uri); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2113 | if (!parcel) |
| 2114 | return; |
| 2115 | Py_INCREF(action); |
| 2116 | } else { |
| 2117 | if (!self->end_ns_event_obj) |
| 2118 | return; |
| 2119 | action = self->end_ns_event_obj; |
| 2120 | Py_INCREF(action); |
| 2121 | parcel = Py_None; |
| 2122 | Py_INCREF(parcel); |
| 2123 | } |
| 2124 | |
| 2125 | res = PyTuple_New(2); |
| 2126 | |
| 2127 | if (res) { |
| 2128 | PyTuple_SET_ITEM(res, 0, action); |
| 2129 | PyTuple_SET_ITEM(res, 1, parcel); |
| 2130 | PyList_Append(self->events, res); |
| 2131 | Py_DECREF(res); |
| 2132 | } else |
| 2133 | PyErr_Clear(); /* FIXME: propagate error */ |
| 2134 | } |
| 2135 | |
| 2136 | /* -------------------------------------------------------------------- */ |
| 2137 | /* methods (in alphabetical order) */ |
| 2138 | |
| 2139 | static PyObject* |
| 2140 | treebuilder_data(TreeBuilderObject* self, PyObject* args) |
| 2141 | { |
| 2142 | PyObject* data; |
| 2143 | if (!PyArg_ParseTuple(args, "O:data", &data)) |
| 2144 | return NULL; |
| 2145 | |
| 2146 | return treebuilder_handle_data(self, data); |
| 2147 | } |
| 2148 | |
| 2149 | static PyObject* |
| 2150 | treebuilder_end(TreeBuilderObject* self, PyObject* args) |
| 2151 | { |
| 2152 | PyObject* tag; |
| 2153 | if (!PyArg_ParseTuple(args, "O:end", &tag)) |
| 2154 | return NULL; |
| 2155 | |
| 2156 | return treebuilder_handle_end(self, tag); |
| 2157 | } |
| 2158 | |
| 2159 | LOCAL(PyObject*) |
| 2160 | treebuilder_done(TreeBuilderObject* self) |
| 2161 | { |
| 2162 | PyObject* res; |
| 2163 | |
| 2164 | /* FIXME: check stack size? */ |
| 2165 | |
| 2166 | if (self->root) |
| 2167 | res = self->root; |
| 2168 | else |
| 2169 | res = Py_None; |
| 2170 | |
| 2171 | Py_INCREF(res); |
| 2172 | return res; |
| 2173 | } |
| 2174 | |
| 2175 | static PyObject* |
| 2176 | treebuilder_close(TreeBuilderObject* self, PyObject* args) |
| 2177 | { |
| 2178 | if (!PyArg_ParseTuple(args, ":close")) |
| 2179 | return NULL; |
| 2180 | |
| 2181 | return treebuilder_done(self); |
| 2182 | } |
| 2183 | |
| 2184 | static PyObject* |
| 2185 | treebuilder_start(TreeBuilderObject* self, PyObject* args) |
| 2186 | { |
| 2187 | PyObject* tag; |
| 2188 | PyObject* attrib = Py_None; |
| 2189 | if (!PyArg_ParseTuple(args, "O|O:start", &tag, &attrib)) |
| 2190 | return NULL; |
| 2191 | |
| 2192 | return treebuilder_handle_start(self, tag, attrib); |
| 2193 | } |
| 2194 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2195 | static PyMethodDef treebuilder_methods[] = { |
| 2196 | {"data", (PyCFunction) treebuilder_data, METH_VARARGS}, |
| 2197 | {"start", (PyCFunction) treebuilder_start, METH_VARARGS}, |
| 2198 | {"end", (PyCFunction) treebuilder_end, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2199 | {"close", (PyCFunction) treebuilder_close, METH_VARARGS}, |
| 2200 | {NULL, NULL} |
| 2201 | }; |
| 2202 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 2203 | static PyTypeObject TreeBuilder_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 2204 | PyVarObject_HEAD_INIT(NULL, 0) |
| 2205 | "TreeBuilder", sizeof(TreeBuilderObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2206 | /* methods */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2207 | (destructor)treebuilder_dealloc, /* tp_dealloc */ |
| 2208 | 0, /* tp_print */ |
| 2209 | 0, /* tp_getattr */ |
| 2210 | 0, /* tp_setattr */ |
| 2211 | 0, /* tp_reserved */ |
| 2212 | 0, /* tp_repr */ |
| 2213 | 0, /* tp_as_number */ |
| 2214 | 0, /* tp_as_sequence */ |
| 2215 | 0, /* tp_as_mapping */ |
| 2216 | 0, /* tp_hash */ |
| 2217 | 0, /* tp_call */ |
| 2218 | 0, /* tp_str */ |
| 2219 | 0, /* tp_getattro */ |
| 2220 | 0, /* tp_setattro */ |
| 2221 | 0, /* tp_as_buffer */ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2222 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 2223 | /* tp_flags */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2224 | 0, /* tp_doc */ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2225 | (traverseproc)treebuilder_gc_traverse, /* tp_traverse */ |
| 2226 | (inquiry)treebuilder_gc_clear, /* tp_clear */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2227 | 0, /* tp_richcompare */ |
| 2228 | 0, /* tp_weaklistoffset */ |
| 2229 | 0, /* tp_iter */ |
| 2230 | 0, /* tp_iternext */ |
| 2231 | treebuilder_methods, /* tp_methods */ |
| 2232 | 0, /* tp_members */ |
| 2233 | 0, /* tp_getset */ |
| 2234 | 0, /* tp_base */ |
| 2235 | 0, /* tp_dict */ |
| 2236 | 0, /* tp_descr_get */ |
| 2237 | 0, /* tp_descr_set */ |
| 2238 | 0, /* tp_dictoffset */ |
| 2239 | (initproc)treebuilder_init, /* tp_init */ |
| 2240 | PyType_GenericAlloc, /* tp_alloc */ |
| 2241 | treebuilder_new, /* tp_new */ |
| 2242 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2243 | }; |
| 2244 | |
| 2245 | /* ==================================================================== */ |
| 2246 | /* the expat interface */ |
| 2247 | |
| 2248 | #if defined(USE_EXPAT) |
| 2249 | |
| 2250 | #include "expat.h" |
| 2251 | |
| 2252 | #if defined(USE_PYEXPAT_CAPI) |
| 2253 | #include "pyexpat.h" |
| 2254 | static struct PyExpat_CAPI* expat_capi; |
| 2255 | #define EXPAT(func) (expat_capi->func) |
| 2256 | #else |
| 2257 | #define EXPAT(func) (XML_##func) |
| 2258 | #endif |
| 2259 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2260 | static XML_Memory_Handling_Suite ExpatMemoryHandler = { |
| 2261 | PyObject_Malloc, PyObject_Realloc, PyObject_Free}; |
| 2262 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2263 | typedef struct { |
| 2264 | PyObject_HEAD |
| 2265 | |
| 2266 | XML_Parser parser; |
| 2267 | |
| 2268 | PyObject* target; |
| 2269 | PyObject* entity; |
| 2270 | |
| 2271 | PyObject* names; |
| 2272 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2273 | PyObject* handle_start; |
| 2274 | PyObject* handle_data; |
| 2275 | PyObject* handle_end; |
| 2276 | |
| 2277 | PyObject* handle_comment; |
| 2278 | PyObject* handle_pi; |
| 2279 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2280 | PyObject* handle_close; |
| 2281 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2282 | } XMLParserObject; |
| 2283 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 2284 | static PyTypeObject XMLParser_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2285 | |
| 2286 | /* helpers */ |
| 2287 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2288 | LOCAL(PyObject*) |
| 2289 | makeuniversal(XMLParserObject* self, const char* string) |
| 2290 | { |
| 2291 | /* convert a UTF-8 tag/attribute name from the expat parser |
| 2292 | to a universal name string */ |
| 2293 | |
| 2294 | int size = strlen(string); |
| 2295 | PyObject* key; |
| 2296 | PyObject* value; |
| 2297 | |
| 2298 | /* look the 'raw' name up in the names dictionary */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2299 | key = PyBytes_FromStringAndSize(string, size); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2300 | if (!key) |
| 2301 | return NULL; |
| 2302 | |
| 2303 | value = PyDict_GetItem(self->names, key); |
| 2304 | |
| 2305 | if (value) { |
| 2306 | Py_INCREF(value); |
| 2307 | } else { |
| 2308 | /* new name. convert to universal name, and decode as |
| 2309 | necessary */ |
| 2310 | |
| 2311 | PyObject* tag; |
| 2312 | char* p; |
| 2313 | int i; |
| 2314 | |
| 2315 | /* look for namespace separator */ |
| 2316 | for (i = 0; i < size; i++) |
| 2317 | if (string[i] == '}') |
| 2318 | break; |
| 2319 | if (i != size) { |
| 2320 | /* convert to universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2321 | tag = PyBytes_FromStringAndSize(NULL, size+1); |
| 2322 | p = PyBytes_AS_STRING(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2323 | p[0] = '{'; |
| 2324 | memcpy(p+1, string, size); |
| 2325 | size++; |
| 2326 | } else { |
| 2327 | /* plain name; use key as tag */ |
| 2328 | Py_INCREF(key); |
| 2329 | tag = key; |
| 2330 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2331 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2332 | /* decode universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2333 | p = PyBytes_AS_STRING(tag); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2334 | value = PyUnicode_DecodeUTF8(p, size, "strict"); |
| 2335 | Py_DECREF(tag); |
| 2336 | if (!value) { |
| 2337 | Py_DECREF(key); |
| 2338 | return NULL; |
| 2339 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2340 | |
| 2341 | /* add to names dictionary */ |
| 2342 | if (PyDict_SetItem(self->names, key, value) < 0) { |
| 2343 | Py_DECREF(key); |
| 2344 | Py_DECREF(value); |
| 2345 | return NULL; |
| 2346 | } |
| 2347 | } |
| 2348 | |
| 2349 | Py_DECREF(key); |
| 2350 | return value; |
| 2351 | } |
| 2352 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2353 | /* Set the ParseError exception with the given parameters. |
| 2354 | * If message is not NULL, it's used as the error string. Otherwise, the |
| 2355 | * message string is the default for the given error_code. |
| 2356 | */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2357 | static void |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2358 | expat_set_error(enum XML_Error error_code, int line, int column, char *message) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2359 | { |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2360 | PyObject *errmsg, *error, *position, *code; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2361 | |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2362 | errmsg = PyUnicode_FromFormat("%s: line %d, column %d", |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2363 | message ? message : EXPAT(ErrorString)(error_code), |
| 2364 | line, column); |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2365 | if (errmsg == NULL) |
| 2366 | return; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2367 | |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2368 | error = PyObject_CallFunction(elementtree_parseerror_obj, "O", errmsg); |
| 2369 | Py_DECREF(errmsg); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2370 | if (!error) |
| 2371 | return; |
| 2372 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2373 | /* Add code and position attributes */ |
| 2374 | code = PyLong_FromLong((long)error_code); |
| 2375 | if (!code) { |
| 2376 | Py_DECREF(error); |
| 2377 | return; |
| 2378 | } |
| 2379 | if (PyObject_SetAttrString(error, "code", code) == -1) { |
| 2380 | Py_DECREF(error); |
| 2381 | Py_DECREF(code); |
| 2382 | return; |
| 2383 | } |
| 2384 | Py_DECREF(code); |
| 2385 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2386 | position = Py_BuildValue("(ii)", line, column); |
| 2387 | if (!position) { |
| 2388 | Py_DECREF(error); |
| 2389 | return; |
| 2390 | } |
| 2391 | if (PyObject_SetAttrString(error, "position", position) == -1) { |
| 2392 | Py_DECREF(error); |
| 2393 | Py_DECREF(position); |
| 2394 | return; |
| 2395 | } |
| 2396 | Py_DECREF(position); |
| 2397 | |
| 2398 | PyErr_SetObject(elementtree_parseerror_obj, error); |
| 2399 | Py_DECREF(error); |
| 2400 | } |
| 2401 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2402 | /* -------------------------------------------------------------------- */ |
| 2403 | /* handlers */ |
| 2404 | |
| 2405 | static void |
| 2406 | expat_default_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2407 | int data_len) |
| 2408 | { |
| 2409 | PyObject* key; |
| 2410 | PyObject* value; |
| 2411 | PyObject* res; |
| 2412 | |
| 2413 | if (data_len < 2 || data_in[0] != '&') |
| 2414 | return; |
| 2415 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2416 | key = PyUnicode_DecodeUTF8(data_in + 1, data_len - 2, "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2417 | if (!key) |
| 2418 | return; |
| 2419 | |
| 2420 | value = PyDict_GetItem(self->entity, key); |
| 2421 | |
| 2422 | if (value) { |
| 2423 | if (TreeBuilder_CheckExact(self->target)) |
| 2424 | res = treebuilder_handle_data( |
| 2425 | (TreeBuilderObject*) self->target, value |
| 2426 | ); |
| 2427 | else if (self->handle_data) |
| 2428 | res = PyObject_CallFunction(self->handle_data, "O", value); |
| 2429 | else |
| 2430 | res = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2431 | Py_XDECREF(res); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2432 | } else if (!PyErr_Occurred()) { |
| 2433 | /* Report the first error, not the last */ |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 2434 | char message[128] = "undefined entity "; |
| 2435 | strncat(message, data_in, data_len < 100?data_len:100); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2436 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2437 | XML_ERROR_UNDEFINED_ENTITY, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2438 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2439 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 2440 | message |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2441 | ); |
| 2442 | } |
| 2443 | |
| 2444 | Py_DECREF(key); |
| 2445 | } |
| 2446 | |
| 2447 | static void |
| 2448 | expat_start_handler(XMLParserObject* self, const XML_Char* tag_in, |
| 2449 | const XML_Char **attrib_in) |
| 2450 | { |
| 2451 | PyObject* res; |
| 2452 | PyObject* tag; |
| 2453 | PyObject* attrib; |
| 2454 | int ok; |
| 2455 | |
| 2456 | /* tag name */ |
| 2457 | tag = makeuniversal(self, tag_in); |
| 2458 | if (!tag) |
| 2459 | return; /* parser will look for errors */ |
| 2460 | |
| 2461 | /* attributes */ |
| 2462 | if (attrib_in[0]) { |
| 2463 | attrib = PyDict_New(); |
| 2464 | if (!attrib) |
| 2465 | return; |
| 2466 | while (attrib_in[0] && attrib_in[1]) { |
| 2467 | PyObject* key = makeuniversal(self, attrib_in[0]); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2468 | PyObject* value = PyUnicode_DecodeUTF8(attrib_in[1], strlen(attrib_in[1]), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2469 | if (!key || !value) { |
| 2470 | Py_XDECREF(value); |
| 2471 | Py_XDECREF(key); |
| 2472 | Py_DECREF(attrib); |
| 2473 | return; |
| 2474 | } |
| 2475 | ok = PyDict_SetItem(attrib, key, value); |
| 2476 | Py_DECREF(value); |
| 2477 | Py_DECREF(key); |
| 2478 | if (ok < 0) { |
| 2479 | Py_DECREF(attrib); |
| 2480 | return; |
| 2481 | } |
| 2482 | attrib_in += 2; |
| 2483 | } |
| 2484 | } else { |
| 2485 | Py_INCREF(Py_None); |
| 2486 | attrib = Py_None; |
| 2487 | } |
| 2488 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2489 | /* If we get None, pass an empty dictionary on */ |
| 2490 | if (attrib == Py_None) { |
| 2491 | Py_DECREF(attrib); |
| 2492 | attrib = PyDict_New(); |
| 2493 | if (!attrib) |
| 2494 | return; |
| 2495 | } |
| 2496 | |
| 2497 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2498 | /* shortcut */ |
| 2499 | res = treebuilder_handle_start((TreeBuilderObject*) self->target, |
| 2500 | tag, attrib); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2501 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2502 | else if (self->handle_start) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2503 | res = PyObject_CallFunction(self->handle_start, "OO", tag, attrib); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2504 | } else |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2505 | res = NULL; |
| 2506 | |
| 2507 | Py_DECREF(tag); |
| 2508 | Py_DECREF(attrib); |
| 2509 | |
| 2510 | Py_XDECREF(res); |
| 2511 | } |
| 2512 | |
| 2513 | static void |
| 2514 | expat_data_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2515 | int data_len) |
| 2516 | { |
| 2517 | PyObject* data; |
| 2518 | PyObject* res; |
| 2519 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2520 | data = PyUnicode_DecodeUTF8(data_in, data_len, "strict"); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2521 | if (!data) |
| 2522 | return; /* parser will look for errors */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2523 | |
| 2524 | if (TreeBuilder_CheckExact(self->target)) |
| 2525 | /* shortcut */ |
| 2526 | res = treebuilder_handle_data((TreeBuilderObject*) self->target, data); |
| 2527 | else if (self->handle_data) |
| 2528 | res = PyObject_CallFunction(self->handle_data, "O", data); |
| 2529 | else |
| 2530 | res = NULL; |
| 2531 | |
| 2532 | Py_DECREF(data); |
| 2533 | |
| 2534 | Py_XDECREF(res); |
| 2535 | } |
| 2536 | |
| 2537 | static void |
| 2538 | expat_end_handler(XMLParserObject* self, const XML_Char* tag_in) |
| 2539 | { |
| 2540 | PyObject* tag; |
| 2541 | PyObject* res = NULL; |
| 2542 | |
| 2543 | if (TreeBuilder_CheckExact(self->target)) |
| 2544 | /* shortcut */ |
| 2545 | /* the standard tree builder doesn't look at the end tag */ |
| 2546 | res = treebuilder_handle_end( |
| 2547 | (TreeBuilderObject*) self->target, Py_None |
| 2548 | ); |
| 2549 | else if (self->handle_end) { |
| 2550 | tag = makeuniversal(self, tag_in); |
| 2551 | if (tag) { |
| 2552 | res = PyObject_CallFunction(self->handle_end, "O", tag); |
| 2553 | Py_DECREF(tag); |
| 2554 | } |
| 2555 | } |
| 2556 | |
| 2557 | Py_XDECREF(res); |
| 2558 | } |
| 2559 | |
| 2560 | static void |
| 2561 | expat_start_ns_handler(XMLParserObject* self, const XML_Char* prefix, |
| 2562 | const XML_Char *uri) |
| 2563 | { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2564 | PyObject* sprefix = NULL; |
| 2565 | PyObject* suri = NULL; |
| 2566 | |
| 2567 | suri = PyUnicode_DecodeUTF8(uri, strlen(uri), "strict"); |
| 2568 | if (!suri) |
| 2569 | return; |
| 2570 | |
| 2571 | if (prefix) |
| 2572 | sprefix = PyUnicode_DecodeUTF8(prefix, strlen(prefix), "strict"); |
| 2573 | else |
| 2574 | sprefix = PyUnicode_FromString(""); |
| 2575 | if (!sprefix) { |
| 2576 | Py_DECREF(suri); |
| 2577 | return; |
| 2578 | } |
| 2579 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2580 | treebuilder_handle_namespace( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2581 | (TreeBuilderObject*) self->target, 1, sprefix, suri |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2582 | ); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2583 | |
| 2584 | Py_DECREF(sprefix); |
| 2585 | Py_DECREF(suri); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2586 | } |
| 2587 | |
| 2588 | static void |
| 2589 | expat_end_ns_handler(XMLParserObject* self, const XML_Char* prefix_in) |
| 2590 | { |
| 2591 | treebuilder_handle_namespace( |
| 2592 | (TreeBuilderObject*) self->target, 0, NULL, NULL |
| 2593 | ); |
| 2594 | } |
| 2595 | |
| 2596 | static void |
| 2597 | expat_comment_handler(XMLParserObject* self, const XML_Char* comment_in) |
| 2598 | { |
| 2599 | PyObject* comment; |
| 2600 | PyObject* res; |
| 2601 | |
| 2602 | if (self->handle_comment) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2603 | comment = PyUnicode_DecodeUTF8(comment_in, strlen(comment_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2604 | if (comment) { |
| 2605 | res = PyObject_CallFunction(self->handle_comment, "O", comment); |
| 2606 | Py_XDECREF(res); |
| 2607 | Py_DECREF(comment); |
| 2608 | } |
| 2609 | } |
| 2610 | } |
| 2611 | |
| 2612 | static void |
| 2613 | expat_pi_handler(XMLParserObject* self, const XML_Char* target_in, |
| 2614 | const XML_Char* data_in) |
| 2615 | { |
| 2616 | PyObject* target; |
| 2617 | PyObject* data; |
| 2618 | PyObject* res; |
| 2619 | |
| 2620 | if (self->handle_pi) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2621 | target = PyUnicode_DecodeUTF8(target_in, strlen(target_in), "strict"); |
| 2622 | data = PyUnicode_DecodeUTF8(data_in, strlen(data_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2623 | if (target && data) { |
| 2624 | res = PyObject_CallFunction(self->handle_pi, "OO", target, data); |
| 2625 | Py_XDECREF(res); |
| 2626 | Py_DECREF(data); |
| 2627 | Py_DECREF(target); |
| 2628 | } else { |
| 2629 | Py_XDECREF(data); |
| 2630 | Py_XDECREF(target); |
| 2631 | } |
| 2632 | } |
| 2633 | } |
| 2634 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2635 | static int |
| 2636 | expat_unknown_encoding_handler(XMLParserObject *self, const XML_Char *name, |
| 2637 | XML_Encoding *info) |
| 2638 | { |
| 2639 | PyObject* u; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2640 | unsigned char s[256]; |
| 2641 | int i; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2642 | void *data; |
| 2643 | unsigned int kind; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2644 | |
| 2645 | memset(info, 0, sizeof(XML_Encoding)); |
| 2646 | |
| 2647 | for (i = 0; i < 256; i++) |
| 2648 | s[i] = i; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2649 | |
Fredrik Lundh | c338999 | 2005-12-25 11:40:19 +0000 | [diff] [blame] | 2650 | u = PyUnicode_Decode((char*) s, 256, name, "replace"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2651 | if (!u) |
| 2652 | return XML_STATUS_ERROR; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2653 | if (PyUnicode_READY(u)) |
| 2654 | return XML_STATUS_ERROR; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2655 | |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2656 | if (PyUnicode_GET_LENGTH(u) != 256) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2657 | Py_DECREF(u); |
| 2658 | return XML_STATUS_ERROR; |
| 2659 | } |
| 2660 | |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2661 | kind = PyUnicode_KIND(u); |
| 2662 | data = PyUnicode_DATA(u); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2663 | for (i = 0; i < 256; i++) { |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2664 | Py_UCS4 ch = PyUnicode_READ(kind, data, i); |
| 2665 | if (ch != Py_UNICODE_REPLACEMENT_CHARACTER) |
| 2666 | info->map[i] = ch; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2667 | else |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2668 | info->map[i] = -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2669 | } |
| 2670 | |
| 2671 | Py_DECREF(u); |
| 2672 | |
| 2673 | return XML_STATUS_OK; |
| 2674 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2675 | |
| 2676 | /* -------------------------------------------------------------------- */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2677 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2678 | static PyObject * |
| 2679 | xmlparser_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2680 | { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2681 | XMLParserObject *self = (XMLParserObject *)type->tp_alloc(type, 0); |
| 2682 | if (self) { |
| 2683 | self->parser = NULL; |
| 2684 | self->target = self->entity = self->names = NULL; |
| 2685 | self->handle_start = self->handle_data = self->handle_end = NULL; |
| 2686 | self->handle_comment = self->handle_pi = self->handle_close = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2687 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2688 | return (PyObject *)self; |
| 2689 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2690 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2691 | static int |
| 2692 | xmlparser_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 2693 | { |
| 2694 | XMLParserObject *self_xp = (XMLParserObject *)self; |
| 2695 | PyObject *target = NULL, *html = NULL; |
| 2696 | char *encoding = NULL; |
| 2697 | static char *kwlist[] = {"html", "target", "encoding"}; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2698 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2699 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|OOz:XMLParser", kwlist, |
| 2700 | &html, &target, &encoding)) { |
| 2701 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2702 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2703 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2704 | self_xp->entity = PyDict_New(); |
| 2705 | if (!self_xp->entity) |
| 2706 | return -1; |
| 2707 | |
| 2708 | self_xp->names = PyDict_New(); |
| 2709 | if (!self_xp->names) { |
| 2710 | Py_XDECREF(self_xp->entity); |
| 2711 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2712 | } |
| 2713 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2714 | self_xp->parser = EXPAT(ParserCreate_MM)(encoding, &ExpatMemoryHandler, "}"); |
| 2715 | if (!self_xp->parser) { |
| 2716 | Py_XDECREF(self_xp->entity); |
| 2717 | Py_XDECREF(self_xp->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2718 | PyErr_NoMemory(); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2719 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2720 | } |
| 2721 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2722 | if (target) { |
| 2723 | Py_INCREF(target); |
| 2724 | } else { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2725 | target = treebuilder_new(&TreeBuilder_Type, NULL, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2726 | if (!target) { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2727 | Py_XDECREF(self_xp->entity); |
| 2728 | Py_XDECREF(self_xp->names); |
| 2729 | EXPAT(ParserFree)(self_xp->parser); |
| 2730 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2731 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2732 | } |
| 2733 | self_xp->target = target; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2734 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2735 | self_xp->handle_start = PyObject_GetAttrString(target, "start"); |
| 2736 | self_xp->handle_data = PyObject_GetAttrString(target, "data"); |
| 2737 | self_xp->handle_end = PyObject_GetAttrString(target, "end"); |
| 2738 | self_xp->handle_comment = PyObject_GetAttrString(target, "comment"); |
| 2739 | self_xp->handle_pi = PyObject_GetAttrString(target, "pi"); |
| 2740 | self_xp->handle_close = PyObject_GetAttrString(target, "close"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2741 | |
| 2742 | PyErr_Clear(); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2743 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2744 | /* configure parser */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2745 | EXPAT(SetUserData)(self_xp->parser, self_xp); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2746 | EXPAT(SetElementHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2747 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2748 | (XML_StartElementHandler) expat_start_handler, |
| 2749 | (XML_EndElementHandler) expat_end_handler |
| 2750 | ); |
| 2751 | EXPAT(SetDefaultHandlerExpand)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2752 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2753 | (XML_DefaultHandler) expat_default_handler |
| 2754 | ); |
| 2755 | EXPAT(SetCharacterDataHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2756 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2757 | (XML_CharacterDataHandler) expat_data_handler |
| 2758 | ); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2759 | if (self_xp->handle_comment) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2760 | EXPAT(SetCommentHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2761 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2762 | (XML_CommentHandler) expat_comment_handler |
| 2763 | ); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2764 | if (self_xp->handle_pi) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2765 | EXPAT(SetProcessingInstructionHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2766 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2767 | (XML_ProcessingInstructionHandler) expat_pi_handler |
| 2768 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2769 | EXPAT(SetUnknownEncodingHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2770 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2771 | (XML_UnknownEncodingHandler) expat_unknown_encoding_handler, NULL |
| 2772 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2773 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2774 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2775 | } |
| 2776 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2777 | static int |
| 2778 | xmlparser_gc_traverse(XMLParserObject *self, visitproc visit, void *arg) |
| 2779 | { |
| 2780 | Py_VISIT(self->handle_close); |
| 2781 | Py_VISIT(self->handle_pi); |
| 2782 | Py_VISIT(self->handle_comment); |
| 2783 | Py_VISIT(self->handle_end); |
| 2784 | Py_VISIT(self->handle_data); |
| 2785 | Py_VISIT(self->handle_start); |
| 2786 | |
| 2787 | Py_VISIT(self->target); |
| 2788 | Py_VISIT(self->entity); |
| 2789 | Py_VISIT(self->names); |
| 2790 | |
| 2791 | return 0; |
| 2792 | } |
| 2793 | |
| 2794 | static int |
| 2795 | xmlparser_gc_clear(XMLParserObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2796 | { |
| 2797 | EXPAT(ParserFree)(self->parser); |
| 2798 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2799 | Py_XDECREF(self->handle_close); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2800 | Py_XDECREF(self->handle_pi); |
| 2801 | Py_XDECREF(self->handle_comment); |
| 2802 | Py_XDECREF(self->handle_end); |
| 2803 | Py_XDECREF(self->handle_data); |
| 2804 | Py_XDECREF(self->handle_start); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2805 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2806 | Py_XDECREF(self->target); |
| 2807 | Py_XDECREF(self->entity); |
| 2808 | Py_XDECREF(self->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2809 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2810 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2811 | } |
| 2812 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 2813 | static void |
| 2814 | xmlparser_dealloc(XMLParserObject* self) |
| 2815 | { |
| 2816 | PyObject_GC_UnTrack(self); |
| 2817 | xmlparser_gc_clear(self); |
| 2818 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 2819 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2820 | |
| 2821 | LOCAL(PyObject*) |
| 2822 | expat_parse(XMLParserObject* self, char* data, int data_len, int final) |
| 2823 | { |
| 2824 | int ok; |
| 2825 | |
| 2826 | ok = EXPAT(Parse)(self->parser, data, data_len, final); |
| 2827 | |
| 2828 | if (PyErr_Occurred()) |
| 2829 | return NULL; |
| 2830 | |
| 2831 | if (!ok) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2832 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2833 | EXPAT(GetErrorCode)(self->parser), |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2834 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2835 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 2836 | NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2837 | ); |
| 2838 | return NULL; |
| 2839 | } |
| 2840 | |
| 2841 | Py_RETURN_NONE; |
| 2842 | } |
| 2843 | |
| 2844 | static PyObject* |
| 2845 | xmlparser_close(XMLParserObject* self, PyObject* args) |
| 2846 | { |
| 2847 | /* end feeding data to parser */ |
| 2848 | |
| 2849 | PyObject* res; |
| 2850 | if (!PyArg_ParseTuple(args, ":close")) |
| 2851 | return NULL; |
| 2852 | |
| 2853 | res = expat_parse(self, "", 0, 1); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2854 | if (!res) |
| 2855 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2856 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2857 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2858 | Py_DECREF(res); |
| 2859 | return treebuilder_done((TreeBuilderObject*) self->target); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2860 | } if (self->handle_close) { |
| 2861 | Py_DECREF(res); |
| 2862 | return PyObject_CallFunction(self->handle_close, ""); |
| 2863 | } else |
| 2864 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2865 | } |
| 2866 | |
| 2867 | static PyObject* |
| 2868 | xmlparser_feed(XMLParserObject* self, PyObject* args) |
| 2869 | { |
| 2870 | /* feed data to parser */ |
| 2871 | |
| 2872 | char* data; |
| 2873 | int data_len; |
| 2874 | if (!PyArg_ParseTuple(args, "s#:feed", &data, &data_len)) |
| 2875 | return NULL; |
| 2876 | |
| 2877 | return expat_parse(self, data, data_len, 0); |
| 2878 | } |
| 2879 | |
| 2880 | static PyObject* |
| 2881 | xmlparser_parse(XMLParserObject* self, PyObject* args) |
| 2882 | { |
| 2883 | /* (internal) parse until end of input stream */ |
| 2884 | |
| 2885 | PyObject* reader; |
| 2886 | PyObject* buffer; |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2887 | PyObject* temp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2888 | PyObject* res; |
| 2889 | |
| 2890 | PyObject* fileobj; |
| 2891 | if (!PyArg_ParseTuple(args, "O:_parse", &fileobj)) |
| 2892 | return NULL; |
| 2893 | |
| 2894 | reader = PyObject_GetAttrString(fileobj, "read"); |
| 2895 | if (!reader) |
| 2896 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2897 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2898 | /* read from open file object */ |
| 2899 | for (;;) { |
| 2900 | |
| 2901 | buffer = PyObject_CallFunction(reader, "i", 64*1024); |
| 2902 | |
| 2903 | if (!buffer) { |
| 2904 | /* read failed (e.g. due to KeyboardInterrupt) */ |
| 2905 | Py_DECREF(reader); |
| 2906 | return NULL; |
| 2907 | } |
| 2908 | |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2909 | if (PyUnicode_CheckExact(buffer)) { |
| 2910 | /* A unicode object is encoded into bytes using UTF-8 */ |
| 2911 | if (PyUnicode_GET_SIZE(buffer) == 0) { |
| 2912 | Py_DECREF(buffer); |
| 2913 | break; |
| 2914 | } |
| 2915 | temp = PyUnicode_AsEncodedString(buffer, "utf-8", "surrogatepass"); |
| 2916 | if (!temp) { |
| 2917 | /* Propagate exception from PyUnicode_AsEncodedString */ |
| 2918 | Py_DECREF(buffer); |
| 2919 | Py_DECREF(reader); |
| 2920 | return NULL; |
| 2921 | } |
| 2922 | |
| 2923 | /* Here we no longer need the original buffer since it contains |
| 2924 | * unicode. Make it point to the encoded bytes object. |
| 2925 | */ |
| 2926 | Py_DECREF(buffer); |
| 2927 | buffer = temp; |
| 2928 | } |
| 2929 | else if (!PyBytes_CheckExact(buffer) || PyBytes_GET_SIZE(buffer) == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2930 | Py_DECREF(buffer); |
| 2931 | break; |
| 2932 | } |
| 2933 | |
| 2934 | res = expat_parse( |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2935 | self, PyBytes_AS_STRING(buffer), PyBytes_GET_SIZE(buffer), 0 |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2936 | ); |
| 2937 | |
| 2938 | Py_DECREF(buffer); |
| 2939 | |
| 2940 | if (!res) { |
| 2941 | Py_DECREF(reader); |
| 2942 | return NULL; |
| 2943 | } |
| 2944 | Py_DECREF(res); |
| 2945 | |
| 2946 | } |
| 2947 | |
| 2948 | Py_DECREF(reader); |
| 2949 | |
| 2950 | res = expat_parse(self, "", 0, 1); |
| 2951 | |
| 2952 | if (res && TreeBuilder_CheckExact(self->target)) { |
| 2953 | Py_DECREF(res); |
| 2954 | return treebuilder_done((TreeBuilderObject*) self->target); |
| 2955 | } |
| 2956 | |
| 2957 | return res; |
| 2958 | } |
| 2959 | |
| 2960 | static PyObject* |
| 2961 | xmlparser_setevents(XMLParserObject* self, PyObject* args) |
| 2962 | { |
| 2963 | /* activate element event reporting */ |
| 2964 | |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2965 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2966 | TreeBuilderObject* target; |
| 2967 | |
| 2968 | PyObject* events; /* event collector */ |
| 2969 | PyObject* event_set = Py_None; |
| 2970 | if (!PyArg_ParseTuple(args, "O!|O:_setevents", &PyList_Type, &events, |
| 2971 | &event_set)) |
| 2972 | return NULL; |
| 2973 | |
| 2974 | if (!TreeBuilder_CheckExact(self->target)) { |
| 2975 | PyErr_SetString( |
| 2976 | PyExc_TypeError, |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 2977 | "event handling only supported for ElementTree.TreeBuilder " |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2978 | "targets" |
| 2979 | ); |
| 2980 | return NULL; |
| 2981 | } |
| 2982 | |
| 2983 | target = (TreeBuilderObject*) self->target; |
| 2984 | |
| 2985 | Py_INCREF(events); |
| 2986 | Py_XDECREF(target->events); |
| 2987 | target->events = events; |
| 2988 | |
| 2989 | /* clear out existing events */ |
| 2990 | Py_XDECREF(target->start_event_obj); target->start_event_obj = NULL; |
| 2991 | Py_XDECREF(target->end_event_obj); target->end_event_obj = NULL; |
| 2992 | Py_XDECREF(target->start_ns_event_obj); target->start_ns_event_obj = NULL; |
| 2993 | Py_XDECREF(target->end_ns_event_obj); target->end_ns_event_obj = NULL; |
| 2994 | |
| 2995 | if (event_set == Py_None) { |
| 2996 | /* default is "end" only */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2997 | target->end_event_obj = PyUnicode_FromString("end"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2998 | Py_RETURN_NONE; |
| 2999 | } |
| 3000 | |
| 3001 | if (!PyTuple_Check(event_set)) /* FIXME: handle arbitrary sequences */ |
| 3002 | goto error; |
| 3003 | |
| 3004 | for (i = 0; i < PyTuple_GET_SIZE(event_set); i++) { |
| 3005 | PyObject* item = PyTuple_GET_ITEM(event_set, i); |
| 3006 | char* event; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3007 | if (PyUnicode_Check(item)) { |
| 3008 | event = _PyUnicode_AsString(item); |
Victor Stinner | 0477bf3 | 2010-03-22 12:11:44 +0000 | [diff] [blame] | 3009 | if (event == NULL) |
| 3010 | goto error; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3011 | } else if (PyBytes_Check(item)) |
| 3012 | event = PyBytes_AS_STRING(item); |
| 3013 | else { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3014 | goto error; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3015 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3016 | if (strcmp(event, "start") == 0) { |
| 3017 | Py_INCREF(item); |
| 3018 | target->start_event_obj = item; |
| 3019 | } else if (strcmp(event, "end") == 0) { |
| 3020 | Py_INCREF(item); |
| 3021 | Py_XDECREF(target->end_event_obj); |
| 3022 | target->end_event_obj = item; |
| 3023 | } else if (strcmp(event, "start-ns") == 0) { |
| 3024 | Py_INCREF(item); |
| 3025 | Py_XDECREF(target->start_ns_event_obj); |
| 3026 | target->start_ns_event_obj = item; |
| 3027 | EXPAT(SetNamespaceDeclHandler)( |
| 3028 | self->parser, |
| 3029 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3030 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3031 | ); |
| 3032 | } else if (strcmp(event, "end-ns") == 0) { |
| 3033 | Py_INCREF(item); |
| 3034 | Py_XDECREF(target->end_ns_event_obj); |
| 3035 | target->end_ns_event_obj = item; |
| 3036 | EXPAT(SetNamespaceDeclHandler)( |
| 3037 | self->parser, |
| 3038 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3039 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3040 | ); |
| 3041 | } else { |
| 3042 | PyErr_Format( |
| 3043 | PyExc_ValueError, |
| 3044 | "unknown event '%s'", event |
| 3045 | ); |
| 3046 | return NULL; |
| 3047 | } |
| 3048 | } |
| 3049 | |
| 3050 | Py_RETURN_NONE; |
| 3051 | |
| 3052 | error: |
| 3053 | PyErr_SetString( |
| 3054 | PyExc_TypeError, |
| 3055 | "invalid event tuple" |
| 3056 | ); |
| 3057 | return NULL; |
| 3058 | } |
| 3059 | |
| 3060 | static PyMethodDef xmlparser_methods[] = { |
| 3061 | {"feed", (PyCFunction) xmlparser_feed, METH_VARARGS}, |
| 3062 | {"close", (PyCFunction) xmlparser_close, METH_VARARGS}, |
| 3063 | {"_parse", (PyCFunction) xmlparser_parse, METH_VARARGS}, |
| 3064 | {"_setevents", (PyCFunction) xmlparser_setevents, METH_VARARGS}, |
| 3065 | {NULL, NULL} |
| 3066 | }; |
| 3067 | |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3068 | static PyObject* |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3069 | xmlparser_getattro(XMLParserObject* self, PyObject* nameobj) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3070 | { |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3071 | if (PyUnicode_Check(nameobj)) { |
| 3072 | PyObject* res; |
| 3073 | if (PyUnicode_CompareWithASCIIString(nameobj, "entity") == 0) |
| 3074 | res = self->entity; |
| 3075 | else if (PyUnicode_CompareWithASCIIString(nameobj, "target") == 0) |
| 3076 | res = self->target; |
| 3077 | else if (PyUnicode_CompareWithASCIIString(nameobj, "version") == 0) { |
| 3078 | return PyUnicode_FromFormat( |
| 3079 | "Expat %d.%d.%d", XML_MAJOR_VERSION, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3080 | XML_MINOR_VERSION, XML_MICRO_VERSION); |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3081 | } |
| 3082 | else |
| 3083 | goto generic; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3084 | |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3085 | Py_INCREF(res); |
| 3086 | return res; |
| 3087 | } |
| 3088 | generic: |
| 3089 | return PyObject_GenericGetAttr((PyObject*) self, nameobj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3090 | } |
| 3091 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 3092 | static PyTypeObject XMLParser_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 3093 | PyVarObject_HEAD_INIT(NULL, 0) |
| 3094 | "XMLParser", sizeof(XMLParserObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3095 | /* methods */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 3096 | (destructor)xmlparser_dealloc, /* tp_dealloc */ |
| 3097 | 0, /* tp_print */ |
| 3098 | 0, /* tp_getattr */ |
| 3099 | 0, /* tp_setattr */ |
| 3100 | 0, /* tp_reserved */ |
| 3101 | 0, /* tp_repr */ |
| 3102 | 0, /* tp_as_number */ |
| 3103 | 0, /* tp_as_sequence */ |
| 3104 | 0, /* tp_as_mapping */ |
| 3105 | 0, /* tp_hash */ |
| 3106 | 0, /* tp_call */ |
| 3107 | 0, /* tp_str */ |
| 3108 | (getattrofunc)xmlparser_getattro, /* tp_getattro */ |
| 3109 | 0, /* tp_setattro */ |
| 3110 | 0, /* tp_as_buffer */ |
| 3111 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 3112 | /* tp_flags */ |
| 3113 | 0, /* tp_doc */ |
| 3114 | (traverseproc)xmlparser_gc_traverse, /* tp_traverse */ |
| 3115 | (inquiry)xmlparser_gc_clear, /* tp_clear */ |
| 3116 | 0, /* tp_richcompare */ |
| 3117 | 0, /* tp_weaklistoffset */ |
| 3118 | 0, /* tp_iter */ |
| 3119 | 0, /* tp_iternext */ |
| 3120 | xmlparser_methods, /* tp_methods */ |
| 3121 | 0, /* tp_members */ |
| 3122 | 0, /* tp_getset */ |
| 3123 | 0, /* tp_base */ |
| 3124 | 0, /* tp_dict */ |
| 3125 | 0, /* tp_descr_get */ |
| 3126 | 0, /* tp_descr_set */ |
| 3127 | 0, /* tp_dictoffset */ |
| 3128 | (initproc)xmlparser_init, /* tp_init */ |
| 3129 | PyType_GenericAlloc, /* tp_alloc */ |
| 3130 | xmlparser_new, /* tp_new */ |
| 3131 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3132 | }; |
| 3133 | |
| 3134 | #endif |
| 3135 | |
| 3136 | /* ==================================================================== */ |
| 3137 | /* python module interface */ |
| 3138 | |
| 3139 | static PyMethodDef _functions[] = { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3140 | {"SubElement", (PyCFunction) subelement, METH_VARARGS|METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3141 | {NULL, NULL} |
| 3142 | }; |
| 3143 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3144 | |
| 3145 | static struct PyModuleDef _elementtreemodule = { |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3146 | PyModuleDef_HEAD_INIT, |
| 3147 | "_elementtree", |
| 3148 | NULL, |
| 3149 | -1, |
| 3150 | _functions, |
| 3151 | NULL, |
| 3152 | NULL, |
| 3153 | NULL, |
| 3154 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3155 | }; |
| 3156 | |
Neal Norwitz | f6657e6 | 2006-12-28 04:47:50 +0000 | [diff] [blame] | 3157 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3158 | PyInit__elementtree(void) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3159 | { |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 3160 | PyObject *m, *g, *temp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3161 | char* bootstrap; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3162 | |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3163 | /* Initialize object types */ |
| 3164 | if (PyType_Ready(&TreeBuilder_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3165 | return NULL; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3166 | if (PyType_Ready(&Element_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3167 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3168 | #if defined(USE_EXPAT) |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3169 | if (PyType_Ready(&XMLParser_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3170 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3171 | #endif |
| 3172 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3173 | m = PyModule_Create(&_elementtreemodule); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 3174 | if (!m) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3175 | return NULL; |
| 3176 | |
| 3177 | /* The code below requires that the module gets already added |
| 3178 | to sys.modules. */ |
| 3179 | PyDict_SetItemString(PyImport_GetModuleDict(), |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3180 | _elementtreemodule.m_name, |
| 3181 | m); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3182 | |
| 3183 | /* python glue code */ |
| 3184 | |
| 3185 | g = PyDict_New(); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 3186 | if (!g) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3187 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3188 | |
| 3189 | PyDict_SetItemString(g, "__builtins__", PyEval_GetBuiltins()); |
| 3190 | |
| 3191 | bootstrap = ( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3192 | "def iter(node, tag=None):\n" /* helper */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3193 | " if tag == '*':\n" |
| 3194 | " tag = None\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3195 | " if tag is None or node.tag == tag:\n" |
| 3196 | " yield node\n" |
| 3197 | " for node in node:\n" |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3198 | " for node in iter(node, tag):\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3199 | " yield node\n" |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3200 | |
| 3201 | "def itertext(node):\n" /* helper */ |
| 3202 | " if node.text:\n" |
| 3203 | " yield node.text\n" |
| 3204 | " for e in node:\n" |
| 3205 | " for s in e.itertext():\n" |
| 3206 | " yield s\n" |
| 3207 | " if e.tail:\n" |
| 3208 | " yield e.tail\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3209 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3210 | ); |
| 3211 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3212 | if (!PyRun_String(bootstrap, Py_file_input, g, NULL)) |
| 3213 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3214 | |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 3215 | if (!(temp = PyImport_ImportModule("copy"))) |
| 3216 | return NULL; |
| 3217 | elementtree_deepcopy_obj = PyObject_GetAttrString(temp, "deepcopy"); |
| 3218 | Py_XDECREF(temp); |
| 3219 | |
| 3220 | if (!(elementpath_obj = PyImport_ImportModule("xml.etree.ElementPath"))) |
| 3221 | return NULL; |
| 3222 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3223 | elementtree_iter_obj = PyDict_GetItemString(g, "iter"); |
| 3224 | elementtree_itertext_obj = PyDict_GetItemString(g, "itertext"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3225 | |
| 3226 | #if defined(USE_PYEXPAT_CAPI) |
| 3227 | /* link against pyexpat, if possible */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3228 | expat_capi = PyCapsule_Import(PyExpat_CAPSULE_NAME, 0); |
| 3229 | if (expat_capi) { |
| 3230 | /* check that it's usable */ |
| 3231 | if (strcmp(expat_capi->magic, PyExpat_CAPI_MAGIC) != 0 || |
| 3232 | expat_capi->size < sizeof(struct PyExpat_CAPI) || |
| 3233 | expat_capi->MAJOR_VERSION != XML_MAJOR_VERSION || |
| 3234 | expat_capi->MINOR_VERSION != XML_MINOR_VERSION || |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 3235 | expat_capi->MICRO_VERSION != XML_MICRO_VERSION) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3236 | expat_capi = NULL; |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 3237 | } |
| 3238 | } |
| 3239 | if (!expat_capi) { |
| 3240 | PyErr_SetString( |
| 3241 | PyExc_RuntimeError, "cannot load dispatch table from pyexpat" |
| 3242 | ); |
| 3243 | return NULL; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3244 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3245 | #endif |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3246 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3247 | elementtree_parseerror_obj = PyErr_NewException( |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 3248 | "xml.etree.ElementTree.ParseError", PyExc_SyntaxError, NULL |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3249 | ); |
| 3250 | Py_INCREF(elementtree_parseerror_obj); |
| 3251 | PyModule_AddObject(m, "ParseError", elementtree_parseerror_obj); |
| 3252 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 3253 | Py_INCREF((PyObject *)&Element_Type); |
| 3254 | PyModule_AddObject(m, "Element", (PyObject *)&Element_Type); |
| 3255 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 3256 | Py_INCREF((PyObject *)&TreeBuilder_Type); |
| 3257 | PyModule_AddObject(m, "TreeBuilder", (PyObject *)&TreeBuilder_Type); |
| 3258 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame^] | 3259 | #if defined(USE_EXPAT) |
| 3260 | Py_INCREF((PyObject *)&XMLParser_Type); |
| 3261 | PyModule_AddObject(m, "XMLParser", (PyObject *)&XMLParser_Type); |
| 3262 | #endif |
| 3263 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3264 | return m; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3265 | } |