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 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 61 | /* An element can hold this many children without extra memory |
| 62 | allocations. */ |
| 63 | #define STATIC_CHILDREN 4 |
| 64 | |
| 65 | /* For best performance, chose a value so that 80-90% of all nodes |
| 66 | have no more than the given number of children. Set this to zero |
| 67 | to minimize the size of the element structure itself (this only |
| 68 | helps if you have lots of leaf nodes with attributes). */ |
| 69 | |
| 70 | /* Also note that pymalloc always allocates blocks in multiples of |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 71 | eight bytes. For the current C version of ElementTree, this means |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 72 | that the number of children should be an even number, at least on |
| 73 | 32-bit platforms. */ |
| 74 | |
| 75 | /* -------------------------------------------------------------------- */ |
| 76 | |
| 77 | #if 0 |
| 78 | static int memory = 0; |
| 79 | #define ALLOC(size, comment)\ |
| 80 | do { memory += size; printf("%8d - %s\n", memory, comment); } while (0) |
| 81 | #define RELEASE(size, comment)\ |
| 82 | do { memory -= size; printf("%8d - %s\n", memory, comment); } while (0) |
| 83 | #else |
| 84 | #define ALLOC(size, comment) |
| 85 | #define RELEASE(size, comment) |
| 86 | #endif |
| 87 | |
| 88 | /* compiler tweaks */ |
| 89 | #if defined(_MSC_VER) |
| 90 | #define LOCAL(type) static __inline type __fastcall |
| 91 | #else |
| 92 | #define LOCAL(type) static type |
| 93 | #endif |
| 94 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 95 | /* macros used to store 'join' flags in string object pointers. note |
| 96 | that all use of text and tail as object pointers must be wrapped in |
| 97 | JOIN_OBJ. see comments in the ElementObject definition for more |
| 98 | info. */ |
| 99 | #define JOIN_GET(p) ((Py_uintptr_t) (p) & 1) |
| 100 | #define JOIN_SET(p, flag) ((void*) ((Py_uintptr_t) (JOIN_OBJ(p)) | (flag))) |
| 101 | #define JOIN_OBJ(p) ((PyObject*) ((Py_uintptr_t) (p) & ~1)) |
| 102 | |
| 103 | /* glue functions (see the init function for details) */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 104 | static PyObject* elementtree_parseerror_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 105 | static PyObject* elementtree_deepcopy_obj; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 106 | static PyObject* elementtree_iter_obj; |
| 107 | static PyObject* elementtree_itertext_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 108 | static PyObject* elementpath_obj; |
| 109 | |
| 110 | /* helpers */ |
| 111 | |
| 112 | LOCAL(PyObject*) |
| 113 | deepcopy(PyObject* object, PyObject* memo) |
| 114 | { |
| 115 | /* do a deep copy of the given object */ |
| 116 | |
| 117 | PyObject* args; |
| 118 | PyObject* result; |
| 119 | |
| 120 | if (!elementtree_deepcopy_obj) { |
| 121 | PyErr_SetString( |
| 122 | PyExc_RuntimeError, |
| 123 | "deepcopy helper not found" |
| 124 | ); |
| 125 | return NULL; |
| 126 | } |
| 127 | |
| 128 | args = PyTuple_New(2); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 129 | if (!args) |
| 130 | return NULL; |
| 131 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 132 | Py_INCREF(object); PyTuple_SET_ITEM(args, 0, (PyObject*) object); |
| 133 | Py_INCREF(memo); PyTuple_SET_ITEM(args, 1, (PyObject*) memo); |
| 134 | |
| 135 | result = PyObject_CallObject(elementtree_deepcopy_obj, args); |
| 136 | |
| 137 | Py_DECREF(args); |
| 138 | |
| 139 | return result; |
| 140 | } |
| 141 | |
| 142 | LOCAL(PyObject*) |
| 143 | list_join(PyObject* list) |
| 144 | { |
| 145 | /* join list elements (destroying the list in the process) */ |
| 146 | |
| 147 | PyObject* joiner; |
| 148 | PyObject* function; |
| 149 | PyObject* args; |
| 150 | PyObject* result; |
| 151 | |
| 152 | switch (PyList_GET_SIZE(list)) { |
| 153 | case 0: |
| 154 | Py_DECREF(list); |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 155 | return PyBytes_FromString(""); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 156 | case 1: |
| 157 | result = PyList_GET_ITEM(list, 0); |
| 158 | Py_INCREF(result); |
| 159 | Py_DECREF(list); |
| 160 | return result; |
| 161 | } |
| 162 | |
| 163 | /* two or more elements: slice out a suitable separator from the |
| 164 | first member, and use that to join the entire list */ |
| 165 | |
| 166 | joiner = PySequence_GetSlice(PyList_GET_ITEM(list, 0), 0, 0); |
| 167 | if (!joiner) |
| 168 | return NULL; |
| 169 | |
| 170 | function = PyObject_GetAttrString(joiner, "join"); |
| 171 | if (!function) { |
| 172 | Py_DECREF(joiner); |
| 173 | return NULL; |
| 174 | } |
| 175 | |
| 176 | args = PyTuple_New(1); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 177 | if (!args) |
| 178 | return NULL; |
| 179 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 180 | PyTuple_SET_ITEM(args, 0, list); |
| 181 | |
| 182 | result = PyObject_CallObject(function, args); |
| 183 | |
| 184 | Py_DECREF(args); /* also removes list */ |
| 185 | Py_DECREF(function); |
| 186 | Py_DECREF(joiner); |
| 187 | |
| 188 | return result; |
| 189 | } |
| 190 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 191 | /* Is the given object an empty dictionary? |
| 192 | */ |
| 193 | static int |
| 194 | is_empty_dict(PyObject *obj) |
| 195 | { |
| 196 | return PyDict_CheckExact(obj) && PyDict_Size(obj) == 0; |
| 197 | } |
| 198 | |
| 199 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 200 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 201 | /* the Element type */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 202 | |
| 203 | typedef struct { |
| 204 | |
| 205 | /* attributes (a dictionary object), or None if no attributes */ |
| 206 | PyObject* attrib; |
| 207 | |
| 208 | /* child elements */ |
| 209 | int length; /* actual number of items */ |
| 210 | int allocated; /* allocated items */ |
| 211 | |
| 212 | /* this either points to _children or to a malloced buffer */ |
| 213 | PyObject* *children; |
| 214 | |
| 215 | PyObject* _children[STATIC_CHILDREN]; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 216 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 217 | } ElementObjectExtra; |
| 218 | |
| 219 | typedef struct { |
| 220 | PyObject_HEAD |
| 221 | |
| 222 | /* element tag (a string). */ |
| 223 | PyObject* tag; |
| 224 | |
| 225 | /* text before first child. note that this is a tagged pointer; |
| 226 | use JOIN_OBJ to get the object pointer. the join flag is used |
| 227 | to distinguish lists created by the tree builder from lists |
| 228 | assigned to the attribute by application code; the former |
| 229 | should be joined before being returned to the user, the latter |
| 230 | should be left intact. */ |
| 231 | PyObject* text; |
| 232 | |
| 233 | /* text after this element, in parent. note that this is a tagged |
| 234 | pointer; use JOIN_OBJ to get the object pointer. */ |
| 235 | PyObject* tail; |
| 236 | |
| 237 | ElementObjectExtra* extra; |
| 238 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 239 | PyObject *weakreflist; /* For tp_weaklistoffset */ |
| 240 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 241 | } ElementObject; |
| 242 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 243 | static PyTypeObject Element_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 244 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 245 | #define Element_CheckExact(op) (Py_TYPE(op) == &Element_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 246 | |
| 247 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 248 | /* Element constructors and destructor */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 249 | |
| 250 | LOCAL(int) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 251 | create_extra(ElementObject* self, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 252 | { |
| 253 | self->extra = PyObject_Malloc(sizeof(ElementObjectExtra)); |
| 254 | if (!self->extra) |
| 255 | return -1; |
| 256 | |
| 257 | if (!attrib) |
| 258 | attrib = Py_None; |
| 259 | |
| 260 | Py_INCREF(attrib); |
| 261 | self->extra->attrib = attrib; |
| 262 | |
| 263 | self->extra->length = 0; |
| 264 | self->extra->allocated = STATIC_CHILDREN; |
| 265 | self->extra->children = self->extra->_children; |
| 266 | |
| 267 | return 0; |
| 268 | } |
| 269 | |
| 270 | LOCAL(void) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 271 | dealloc_extra(ElementObject* self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 272 | { |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 273 | ElementObjectExtra *myextra; |
| 274 | int i; |
| 275 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 276 | if (!self->extra) |
| 277 | return; |
| 278 | |
| 279 | /* Avoid DECREFs calling into this code again (cycles, etc.) |
| 280 | */ |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 281 | myextra = self->extra; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 282 | self->extra = NULL; |
| 283 | |
| 284 | Py_DECREF(myextra->attrib); |
| 285 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 286 | for (i = 0; i < myextra->length; i++) |
| 287 | Py_DECREF(myextra->children[i]); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 288 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 289 | if (myextra->children != myextra->_children) |
| 290 | PyObject_Free(myextra->children); |
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 | PyObject_Free(myextra); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 293 | } |
| 294 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 295 | /* Convenience internal function to create new Element objects with the given |
| 296 | * tag and attributes. |
| 297 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 298 | LOCAL(PyObject*) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 299 | create_new_element(PyObject* tag, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 300 | { |
| 301 | ElementObject* self; |
| 302 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 303 | self = PyObject_GC_New(ElementObject, &Element_Type); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 304 | if (self == NULL) |
| 305 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 306 | self->extra = NULL; |
| 307 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 308 | if (attrib != Py_None && !is_empty_dict(attrib)) { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 309 | if (create_extra(self, attrib) < 0) { |
Thomas Wouters | 477c8d5 | 2006-05-27 19:21:47 +0000 | [diff] [blame] | 310 | PyObject_Del(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 311 | return NULL; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 312 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 313 | } |
| 314 | |
| 315 | Py_INCREF(tag); |
| 316 | self->tag = tag; |
| 317 | |
| 318 | Py_INCREF(Py_None); |
| 319 | self->text = Py_None; |
| 320 | |
| 321 | Py_INCREF(Py_None); |
| 322 | self->tail = Py_None; |
| 323 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 324 | self->weakreflist = NULL; |
| 325 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 326 | ALLOC(sizeof(ElementObject), "create element"); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 327 | PyObject_GC_Track(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 328 | return (PyObject*) self; |
| 329 | } |
| 330 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 331 | static PyObject * |
| 332 | element_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 333 | { |
| 334 | ElementObject *e = (ElementObject *)type->tp_alloc(type, 0); |
| 335 | if (e != NULL) { |
| 336 | Py_INCREF(Py_None); |
| 337 | e->tag = Py_None; |
| 338 | |
| 339 | Py_INCREF(Py_None); |
| 340 | e->text = Py_None; |
| 341 | |
| 342 | Py_INCREF(Py_None); |
| 343 | e->tail = Py_None; |
| 344 | |
| 345 | e->extra = NULL; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 346 | e->weakreflist = NULL; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 347 | } |
| 348 | return (PyObject *)e; |
| 349 | } |
| 350 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 351 | /* Helper function for extracting the attrib dictionary from a keywords dict. |
| 352 | * This is required by some constructors/functions in this module that can |
| 353 | * either accept attrib as a keyword argument or all attributes splashed |
| 354 | * directly into *kwds. |
| 355 | * If there is no 'attrib' keyword, return an empty dict. |
| 356 | */ |
| 357 | static PyObject* |
| 358 | get_attrib_from_keywords(PyObject *kwds) |
| 359 | { |
| 360 | PyObject *attrib_str = PyUnicode_FromString("attrib"); |
| 361 | PyObject *attrib = PyDict_GetItem(kwds, attrib_str); |
| 362 | |
| 363 | if (attrib) { |
| 364 | /* If attrib was found in kwds, copy its value and remove it from |
| 365 | * kwds |
| 366 | */ |
| 367 | if (!PyDict_Check(attrib)) { |
| 368 | Py_DECREF(attrib_str); |
| 369 | PyErr_Format(PyExc_TypeError, "attrib must be dict, not %.100s", |
| 370 | Py_TYPE(attrib)->tp_name); |
| 371 | return NULL; |
| 372 | } |
| 373 | attrib = PyDict_Copy(attrib); |
| 374 | PyDict_DelItem(kwds, attrib_str); |
| 375 | } else { |
| 376 | attrib = PyDict_New(); |
| 377 | } |
| 378 | |
| 379 | Py_DECREF(attrib_str); |
| 380 | |
| 381 | if (attrib) |
| 382 | PyDict_Update(attrib, kwds); |
| 383 | return attrib; |
| 384 | } |
| 385 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 386 | static int |
| 387 | element_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 388 | { |
| 389 | PyObject *tag; |
| 390 | PyObject *tmp; |
| 391 | PyObject *attrib = NULL; |
| 392 | ElementObject *self_elem; |
| 393 | |
| 394 | if (!PyArg_ParseTuple(args, "O|O!:Element", &tag, &PyDict_Type, &attrib)) |
| 395 | return -1; |
| 396 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 397 | if (attrib) { |
| 398 | /* attrib passed as positional arg */ |
| 399 | attrib = PyDict_Copy(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 400 | if (!attrib) |
| 401 | return -1; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 402 | if (kwds) { |
| 403 | if (PyDict_Update(attrib, kwds) < 0) { |
| 404 | return -1; |
| 405 | } |
| 406 | } |
| 407 | } else if (kwds) { |
| 408 | /* have keywords args */ |
| 409 | attrib = get_attrib_from_keywords(kwds); |
| 410 | if (!attrib) |
| 411 | return -1; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 412 | } else { |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 413 | /* no attrib arg, no kwds, so no attributes */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 414 | Py_INCREF(Py_None); |
| 415 | attrib = Py_None; |
| 416 | } |
| 417 | |
| 418 | self_elem = (ElementObject *)self; |
| 419 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 420 | if (attrib != Py_None && !is_empty_dict(attrib)) { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 421 | if (create_extra(self_elem, attrib) < 0) { |
| 422 | PyObject_Del(self_elem); |
| 423 | return -1; |
| 424 | } |
| 425 | } |
| 426 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 427 | /* 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] | 428 | Py_DECREF(attrib); |
| 429 | |
| 430 | /* Replace the objects already pointed to by tag, text and tail. */ |
| 431 | tmp = self_elem->tag; |
| 432 | self_elem->tag = tag; |
| 433 | Py_INCREF(tag); |
| 434 | Py_DECREF(tmp); |
| 435 | |
| 436 | tmp = self_elem->text; |
| 437 | self_elem->text = Py_None; |
| 438 | Py_INCREF(Py_None); |
| 439 | Py_DECREF(JOIN_OBJ(tmp)); |
| 440 | |
| 441 | tmp = self_elem->tail; |
| 442 | self_elem->tail = Py_None; |
| 443 | Py_INCREF(Py_None); |
| 444 | Py_DECREF(JOIN_OBJ(tmp)); |
| 445 | |
| 446 | return 0; |
| 447 | } |
| 448 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 449 | LOCAL(int) |
| 450 | element_resize(ElementObject* self, int extra) |
| 451 | { |
| 452 | int size; |
| 453 | PyObject* *children; |
| 454 | |
| 455 | /* make sure self->children can hold the given number of extra |
| 456 | elements. set an exception and return -1 if allocation failed */ |
| 457 | |
| 458 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 459 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 460 | |
| 461 | size = self->extra->length + extra; |
| 462 | |
| 463 | if (size > self->extra->allocated) { |
| 464 | /* use Python 2.4's list growth strategy */ |
| 465 | size = (size >> 3) + (size < 9 ? 3 : 6) + size; |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 466 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer "children" |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 467 | * which needs at least 4 bytes. |
| 468 | * 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] | 469 | * be safe. |
| 470 | */ |
| 471 | size = size ? size : 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 472 | if (self->extra->children != self->extra->_children) { |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 473 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 474 | * "children", which needs at least 4 bytes. Although it's a |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 475 | * false alarm always assume at least one child to be safe. |
| 476 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 477 | children = PyObject_Realloc(self->extra->children, |
| 478 | size * sizeof(PyObject*)); |
| 479 | if (!children) |
| 480 | goto nomemory; |
| 481 | } else { |
| 482 | children = PyObject_Malloc(size * sizeof(PyObject*)); |
| 483 | if (!children) |
| 484 | goto nomemory; |
| 485 | /* copy existing children from static area to malloc buffer */ |
| 486 | memcpy(children, self->extra->children, |
| 487 | self->extra->length * sizeof(PyObject*)); |
| 488 | } |
| 489 | self->extra->children = children; |
| 490 | self->extra->allocated = size; |
| 491 | } |
| 492 | |
| 493 | return 0; |
| 494 | |
| 495 | nomemory: |
| 496 | PyErr_NoMemory(); |
| 497 | return -1; |
| 498 | } |
| 499 | |
| 500 | LOCAL(int) |
| 501 | element_add_subelement(ElementObject* self, PyObject* element) |
| 502 | { |
| 503 | /* add a child element to a parent */ |
| 504 | |
| 505 | if (element_resize(self, 1) < 0) |
| 506 | return -1; |
| 507 | |
| 508 | Py_INCREF(element); |
| 509 | self->extra->children[self->extra->length] = element; |
| 510 | |
| 511 | self->extra->length++; |
| 512 | |
| 513 | return 0; |
| 514 | } |
| 515 | |
| 516 | LOCAL(PyObject*) |
| 517 | element_get_attrib(ElementObject* self) |
| 518 | { |
| 519 | /* return borrowed reference to attrib dictionary */ |
| 520 | /* note: this function assumes that the extra section exists */ |
| 521 | |
| 522 | PyObject* res = self->extra->attrib; |
| 523 | |
| 524 | if (res == Py_None) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 525 | Py_DECREF(res); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 526 | /* create missing dictionary */ |
| 527 | res = PyDict_New(); |
| 528 | if (!res) |
| 529 | return NULL; |
| 530 | self->extra->attrib = res; |
| 531 | } |
| 532 | |
| 533 | return res; |
| 534 | } |
| 535 | |
| 536 | LOCAL(PyObject*) |
| 537 | element_get_text(ElementObject* self) |
| 538 | { |
| 539 | /* return borrowed reference to text attribute */ |
| 540 | |
| 541 | PyObject* res = self->text; |
| 542 | |
| 543 | if (JOIN_GET(res)) { |
| 544 | res = JOIN_OBJ(res); |
| 545 | if (PyList_CheckExact(res)) { |
| 546 | res = list_join(res); |
| 547 | if (!res) |
| 548 | return NULL; |
| 549 | self->text = res; |
| 550 | } |
| 551 | } |
| 552 | |
| 553 | return res; |
| 554 | } |
| 555 | |
| 556 | LOCAL(PyObject*) |
| 557 | element_get_tail(ElementObject* self) |
| 558 | { |
| 559 | /* return borrowed reference to text attribute */ |
| 560 | |
| 561 | PyObject* res = self->tail; |
| 562 | |
| 563 | if (JOIN_GET(res)) { |
| 564 | res = JOIN_OBJ(res); |
| 565 | if (PyList_CheckExact(res)) { |
| 566 | res = list_join(res); |
| 567 | if (!res) |
| 568 | return NULL; |
| 569 | self->tail = res; |
| 570 | } |
| 571 | } |
| 572 | |
| 573 | return res; |
| 574 | } |
| 575 | |
| 576 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 577 | subelement(PyObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 578 | { |
| 579 | PyObject* elem; |
| 580 | |
| 581 | ElementObject* parent; |
| 582 | PyObject* tag; |
| 583 | PyObject* attrib = NULL; |
| 584 | if (!PyArg_ParseTuple(args, "O!O|O!:SubElement", |
| 585 | &Element_Type, &parent, &tag, |
| 586 | &PyDict_Type, &attrib)) |
| 587 | return NULL; |
| 588 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 589 | if (attrib) { |
| 590 | /* attrib passed as positional arg */ |
| 591 | attrib = PyDict_Copy(attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 592 | if (!attrib) |
| 593 | return NULL; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 594 | if (kwds) { |
| 595 | if (PyDict_Update(attrib, kwds) < 0) { |
| 596 | return NULL; |
| 597 | } |
| 598 | } |
| 599 | } else if (kwds) { |
| 600 | /* have keyword args */ |
| 601 | attrib = get_attrib_from_keywords(kwds); |
| 602 | if (!attrib) |
| 603 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 604 | } else { |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 605 | /* no attrib arg, no kwds, so no attribute */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 606 | Py_INCREF(Py_None); |
| 607 | attrib = Py_None; |
| 608 | } |
| 609 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 610 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 611 | |
| 612 | Py_DECREF(attrib); |
| 613 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 614 | if (element_add_subelement(parent, elem) < 0) { |
| 615 | Py_DECREF(elem); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 616 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 617 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 618 | |
| 619 | return elem; |
| 620 | } |
| 621 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 622 | static int |
| 623 | element_gc_traverse(ElementObject *self, visitproc visit, void *arg) |
| 624 | { |
| 625 | Py_VISIT(self->tag); |
| 626 | Py_VISIT(JOIN_OBJ(self->text)); |
| 627 | Py_VISIT(JOIN_OBJ(self->tail)); |
| 628 | |
| 629 | if (self->extra) { |
| 630 | int i; |
| 631 | Py_VISIT(self->extra->attrib); |
| 632 | |
| 633 | for (i = 0; i < self->extra->length; ++i) |
| 634 | Py_VISIT(self->extra->children[i]); |
| 635 | } |
| 636 | return 0; |
| 637 | } |
| 638 | |
| 639 | static int |
| 640 | element_gc_clear(ElementObject *self) |
| 641 | { |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 642 | Py_CLEAR(self->tag); |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 643 | |
| 644 | /* The following is like Py_CLEAR for self->text and self->tail, but |
| 645 | * written explicitily because the real pointers hide behind access |
| 646 | * macros. |
| 647 | */ |
| 648 | if (self->text) { |
| 649 | PyObject *tmp = JOIN_OBJ(self->text); |
| 650 | self->text = NULL; |
| 651 | Py_DECREF(tmp); |
| 652 | } |
| 653 | |
| 654 | if (self->tail) { |
| 655 | PyObject *tmp = JOIN_OBJ(self->tail); |
| 656 | self->tail = NULL; |
| 657 | Py_DECREF(tmp); |
| 658 | } |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 659 | |
| 660 | /* After dropping all references from extra, it's no longer valid anyway, |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 661 | * so fully deallocate it. |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 662 | */ |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 663 | dealloc_extra(self); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 664 | return 0; |
| 665 | } |
| 666 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 667 | static void |
| 668 | element_dealloc(ElementObject* self) |
| 669 | { |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 670 | PyObject_GC_UnTrack(self); |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 671 | |
| 672 | if (self->weakreflist != NULL) |
| 673 | PyObject_ClearWeakRefs((PyObject *) self); |
| 674 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 675 | /* element_gc_clear clears all references and deallocates extra |
| 676 | */ |
| 677 | element_gc_clear(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 678 | |
| 679 | RELEASE(sizeof(ElementObject), "destroy element"); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 680 | Py_TYPE(self)->tp_free((PyObject *)self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 681 | } |
| 682 | |
| 683 | /* -------------------------------------------------------------------- */ |
| 684 | /* methods (in alphabetical order) */ |
| 685 | |
| 686 | static PyObject* |
| 687 | element_append(ElementObject* self, PyObject* args) |
| 688 | { |
| 689 | PyObject* element; |
| 690 | if (!PyArg_ParseTuple(args, "O!:append", &Element_Type, &element)) |
| 691 | return NULL; |
| 692 | |
| 693 | if (element_add_subelement(self, element) < 0) |
| 694 | return NULL; |
| 695 | |
| 696 | Py_RETURN_NONE; |
| 697 | } |
| 698 | |
| 699 | static PyObject* |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 700 | element_clearmethod(ElementObject* self, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 701 | { |
| 702 | if (!PyArg_ParseTuple(args, ":clear")) |
| 703 | return NULL; |
| 704 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 705 | dealloc_extra(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 706 | |
| 707 | Py_INCREF(Py_None); |
| 708 | Py_DECREF(JOIN_OBJ(self->text)); |
| 709 | self->text = Py_None; |
| 710 | |
| 711 | Py_INCREF(Py_None); |
| 712 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 713 | self->tail = Py_None; |
| 714 | |
| 715 | Py_RETURN_NONE; |
| 716 | } |
| 717 | |
| 718 | static PyObject* |
| 719 | element_copy(ElementObject* self, PyObject* args) |
| 720 | { |
| 721 | int i; |
| 722 | ElementObject* element; |
| 723 | |
| 724 | if (!PyArg_ParseTuple(args, ":__copy__")) |
| 725 | return NULL; |
| 726 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 727 | element = (ElementObject*) create_new_element( |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 728 | self->tag, (self->extra) ? self->extra->attrib : Py_None |
| 729 | ); |
| 730 | if (!element) |
| 731 | return NULL; |
| 732 | |
| 733 | Py_DECREF(JOIN_OBJ(element->text)); |
| 734 | element->text = self->text; |
| 735 | Py_INCREF(JOIN_OBJ(element->text)); |
| 736 | |
| 737 | Py_DECREF(JOIN_OBJ(element->tail)); |
| 738 | element->tail = self->tail; |
| 739 | Py_INCREF(JOIN_OBJ(element->tail)); |
| 740 | |
| 741 | if (self->extra) { |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 742 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 743 | if (element_resize(element, self->extra->length) < 0) { |
| 744 | Py_DECREF(element); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 745 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 746 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 747 | |
| 748 | for (i = 0; i < self->extra->length; i++) { |
| 749 | Py_INCREF(self->extra->children[i]); |
| 750 | element->extra->children[i] = self->extra->children[i]; |
| 751 | } |
| 752 | |
| 753 | element->extra->length = self->extra->length; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 754 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 755 | } |
| 756 | |
| 757 | return (PyObject*) element; |
| 758 | } |
| 759 | |
| 760 | static PyObject* |
| 761 | element_deepcopy(ElementObject* self, PyObject* args) |
| 762 | { |
| 763 | int i; |
| 764 | ElementObject* element; |
| 765 | PyObject* tag; |
| 766 | PyObject* attrib; |
| 767 | PyObject* text; |
| 768 | PyObject* tail; |
| 769 | PyObject* id; |
| 770 | |
| 771 | PyObject* memo; |
| 772 | if (!PyArg_ParseTuple(args, "O:__deepcopy__", &memo)) |
| 773 | return NULL; |
| 774 | |
| 775 | tag = deepcopy(self->tag, memo); |
| 776 | if (!tag) |
| 777 | return NULL; |
| 778 | |
| 779 | if (self->extra) { |
| 780 | attrib = deepcopy(self->extra->attrib, memo); |
| 781 | if (!attrib) { |
| 782 | Py_DECREF(tag); |
| 783 | return NULL; |
| 784 | } |
| 785 | } else { |
| 786 | Py_INCREF(Py_None); |
| 787 | attrib = Py_None; |
| 788 | } |
| 789 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 790 | element = (ElementObject*) create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 791 | |
| 792 | Py_DECREF(tag); |
| 793 | Py_DECREF(attrib); |
| 794 | |
| 795 | if (!element) |
| 796 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 797 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 798 | text = deepcopy(JOIN_OBJ(self->text), memo); |
| 799 | if (!text) |
| 800 | goto error; |
| 801 | Py_DECREF(element->text); |
| 802 | element->text = JOIN_SET(text, JOIN_GET(self->text)); |
| 803 | |
| 804 | tail = deepcopy(JOIN_OBJ(self->tail), memo); |
| 805 | if (!tail) |
| 806 | goto error; |
| 807 | Py_DECREF(element->tail); |
| 808 | element->tail = JOIN_SET(tail, JOIN_GET(self->tail)); |
| 809 | |
| 810 | if (self->extra) { |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 811 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 812 | if (element_resize(element, self->extra->length) < 0) |
| 813 | goto error; |
| 814 | |
| 815 | for (i = 0; i < self->extra->length; i++) { |
| 816 | PyObject* child = deepcopy(self->extra->children[i], memo); |
| 817 | if (!child) { |
| 818 | element->extra->length = i; |
| 819 | goto error; |
| 820 | } |
| 821 | element->extra->children[i] = child; |
| 822 | } |
| 823 | |
| 824 | element->extra->length = self->extra->length; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 825 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 826 | } |
| 827 | |
| 828 | /* add object to memo dictionary (so deepcopy won't visit it again) */ |
Christian Heimes | 217cfd1 | 2007-12-02 14:31:20 +0000 | [diff] [blame] | 829 | id = PyLong_FromLong((Py_uintptr_t) self); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 830 | if (!id) |
| 831 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 832 | |
| 833 | i = PyDict_SetItem(memo, id, (PyObject*) element); |
| 834 | |
| 835 | Py_DECREF(id); |
| 836 | |
| 837 | if (i < 0) |
| 838 | goto error; |
| 839 | |
| 840 | return (PyObject*) element; |
| 841 | |
| 842 | error: |
| 843 | Py_DECREF(element); |
| 844 | return NULL; |
| 845 | } |
| 846 | |
| 847 | LOCAL(int) |
| 848 | checkpath(PyObject* tag) |
| 849 | { |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 850 | Py_ssize_t i; |
| 851 | int check = 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 852 | |
| 853 | /* check if a tag contains an xpath character */ |
| 854 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 855 | #define PATHCHAR(ch) \ |
| 856 | (ch == '/' || ch == '*' || ch == '[' || ch == '@' || ch == '.') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 857 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 858 | if (PyUnicode_Check(tag)) { |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 859 | const Py_ssize_t len = PyUnicode_GET_LENGTH(tag); |
| 860 | void *data = PyUnicode_DATA(tag); |
| 861 | unsigned int kind = PyUnicode_KIND(tag); |
| 862 | for (i = 0; i < len; i++) { |
| 863 | Py_UCS4 ch = PyUnicode_READ(kind, data, i); |
| 864 | if (ch == '{') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 865 | check = 0; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 866 | else if (ch == '}') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 867 | check = 1; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 868 | else if (check && PATHCHAR(ch)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 869 | return 1; |
| 870 | } |
| 871 | return 0; |
| 872 | } |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 873 | if (PyBytes_Check(tag)) { |
| 874 | char *p = PyBytes_AS_STRING(tag); |
| 875 | for (i = 0; i < PyBytes_GET_SIZE(tag); i++) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 876 | if (p[i] == '{') |
| 877 | check = 0; |
| 878 | else if (p[i] == '}') |
| 879 | check = 1; |
| 880 | else if (check && PATHCHAR(p[i])) |
| 881 | return 1; |
| 882 | } |
| 883 | return 0; |
| 884 | } |
| 885 | |
| 886 | return 1; /* unknown type; might be path expression */ |
| 887 | } |
| 888 | |
| 889 | static PyObject* |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 890 | element_extend(ElementObject* self, PyObject* args) |
| 891 | { |
| 892 | PyObject* seq; |
| 893 | Py_ssize_t i, seqlen = 0; |
| 894 | |
| 895 | PyObject* seq_in; |
| 896 | if (!PyArg_ParseTuple(args, "O:extend", &seq_in)) |
| 897 | return NULL; |
| 898 | |
| 899 | seq = PySequence_Fast(seq_in, ""); |
| 900 | if (!seq) { |
| 901 | PyErr_Format( |
| 902 | PyExc_TypeError, |
| 903 | "expected sequence, not \"%.200s\"", Py_TYPE(seq_in)->tp_name |
| 904 | ); |
| 905 | return NULL; |
| 906 | } |
| 907 | |
| 908 | seqlen = PySequence_Size(seq); |
| 909 | for (i = 0; i < seqlen; i++) { |
| 910 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 911 | if (!PyObject_IsInstance(element, (PyObject *)&Element_Type)) { |
| 912 | Py_DECREF(seq); |
| 913 | PyErr_Format( |
| 914 | PyExc_TypeError, |
| 915 | "expected an Element, not \"%.200s\"", |
| 916 | Py_TYPE(element)->tp_name); |
| 917 | return NULL; |
| 918 | } |
| 919 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 920 | if (element_add_subelement(self, element) < 0) { |
| 921 | Py_DECREF(seq); |
| 922 | return NULL; |
| 923 | } |
| 924 | } |
| 925 | |
| 926 | Py_DECREF(seq); |
| 927 | |
| 928 | Py_RETURN_NONE; |
| 929 | } |
| 930 | |
| 931 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 932 | element_find(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 933 | { |
| 934 | int i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 935 | PyObject* tag; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 936 | PyObject* namespaces = Py_None; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 937 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 938 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 939 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:find", kwlist, |
| 940 | &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 941 | return NULL; |
| 942 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 943 | if (checkpath(tag) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 944 | _Py_IDENTIFIER(find); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 945 | return _PyObject_CallMethodId( |
| 946 | elementpath_obj, &PyId_find, "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 947 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 948 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 949 | |
| 950 | if (!self->extra) |
| 951 | Py_RETURN_NONE; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 952 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 953 | for (i = 0; i < self->extra->length; i++) { |
| 954 | PyObject* item = self->extra->children[i]; |
| 955 | if (Element_CheckExact(item) && |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 956 | PyObject_RichCompareBool(((ElementObject*)item)->tag, tag, Py_EQ) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 957 | Py_INCREF(item); |
| 958 | return item; |
| 959 | } |
| 960 | } |
| 961 | |
| 962 | Py_RETURN_NONE; |
| 963 | } |
| 964 | |
| 965 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 966 | element_findtext(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 967 | { |
| 968 | int i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 969 | PyObject* tag; |
| 970 | PyObject* default_value = Py_None; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 971 | PyObject* namespaces = Py_None; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 972 | _Py_IDENTIFIER(findtext); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 973 | static char *kwlist[] = {"path", "default", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 974 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 975 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|OO:findtext", kwlist, |
| 976 | &tag, &default_value, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 977 | return NULL; |
| 978 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 979 | if (checkpath(tag) || namespaces != Py_None) |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 980 | return _PyObject_CallMethodId( |
| 981 | elementpath_obj, &PyId_findtext, "OOOO", self, tag, default_value, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 982 | ); |
| 983 | |
| 984 | if (!self->extra) { |
| 985 | Py_INCREF(default_value); |
| 986 | return default_value; |
| 987 | } |
| 988 | |
| 989 | for (i = 0; i < self->extra->length; i++) { |
| 990 | ElementObject* item = (ElementObject*) self->extra->children[i]; |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 991 | if (Element_CheckExact(item) && (PyObject_RichCompareBool(item->tag, tag, Py_EQ) == 1)) { |
| 992 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 993 | PyObject* text = element_get_text(item); |
| 994 | if (text == Py_None) |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 995 | return PyBytes_FromString(""); |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 996 | Py_XINCREF(text); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 997 | return text; |
| 998 | } |
| 999 | } |
| 1000 | |
| 1001 | Py_INCREF(default_value); |
| 1002 | return default_value; |
| 1003 | } |
| 1004 | |
| 1005 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1006 | element_findall(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1007 | { |
| 1008 | int i; |
| 1009 | PyObject* out; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1010 | PyObject* tag; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1011 | PyObject* namespaces = Py_None; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1012 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1013 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1014 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:findall", kwlist, |
| 1015 | &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1016 | return NULL; |
| 1017 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1018 | if (checkpath(tag) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1019 | _Py_IDENTIFIER(findall); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1020 | return _PyObject_CallMethodId( |
| 1021 | elementpath_obj, &PyId_findall, "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1022 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1023 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1024 | |
| 1025 | out = PyList_New(0); |
| 1026 | if (!out) |
| 1027 | return NULL; |
| 1028 | |
| 1029 | if (!self->extra) |
| 1030 | return out; |
| 1031 | |
| 1032 | for (i = 0; i < self->extra->length; i++) { |
| 1033 | PyObject* item = self->extra->children[i]; |
| 1034 | if (Element_CheckExact(item) && |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1035 | PyObject_RichCompareBool(((ElementObject*)item)->tag, tag, Py_EQ) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1036 | if (PyList_Append(out, item) < 0) { |
| 1037 | Py_DECREF(out); |
| 1038 | return NULL; |
| 1039 | } |
| 1040 | } |
| 1041 | } |
| 1042 | |
| 1043 | return out; |
| 1044 | } |
| 1045 | |
| 1046 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1047 | element_iterfind(ElementObject *self, PyObject *args, PyObject *kwds) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1048 | { |
| 1049 | PyObject* tag; |
| 1050 | PyObject* namespaces = Py_None; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1051 | _Py_IDENTIFIER(iterfind); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1052 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1053 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1054 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:iterfind", kwlist, |
| 1055 | &tag, &namespaces)) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1056 | return NULL; |
| 1057 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1058 | return _PyObject_CallMethodId( |
| 1059 | elementpath_obj, &PyId_iterfind, "OOO", self, tag, namespaces |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1060 | ); |
| 1061 | } |
| 1062 | |
| 1063 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1064 | element_get(ElementObject* self, PyObject* args) |
| 1065 | { |
| 1066 | PyObject* value; |
| 1067 | |
| 1068 | PyObject* key; |
| 1069 | PyObject* default_value = Py_None; |
| 1070 | if (!PyArg_ParseTuple(args, "O|O:get", &key, &default_value)) |
| 1071 | return NULL; |
| 1072 | |
| 1073 | if (!self->extra || self->extra->attrib == Py_None) |
| 1074 | value = default_value; |
| 1075 | else { |
| 1076 | value = PyDict_GetItem(self->extra->attrib, key); |
| 1077 | if (!value) |
| 1078 | value = default_value; |
| 1079 | } |
| 1080 | |
| 1081 | Py_INCREF(value); |
| 1082 | return value; |
| 1083 | } |
| 1084 | |
| 1085 | static PyObject* |
| 1086 | element_getchildren(ElementObject* self, PyObject* args) |
| 1087 | { |
| 1088 | int i; |
| 1089 | PyObject* list; |
| 1090 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1091 | /* FIXME: report as deprecated? */ |
| 1092 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1093 | if (!PyArg_ParseTuple(args, ":getchildren")) |
| 1094 | return NULL; |
| 1095 | |
| 1096 | if (!self->extra) |
| 1097 | return PyList_New(0); |
| 1098 | |
| 1099 | list = PyList_New(self->extra->length); |
| 1100 | if (!list) |
| 1101 | return NULL; |
| 1102 | |
| 1103 | for (i = 0; i < self->extra->length; i++) { |
| 1104 | PyObject* item = self->extra->children[i]; |
| 1105 | Py_INCREF(item); |
| 1106 | PyList_SET_ITEM(list, i, item); |
| 1107 | } |
| 1108 | |
| 1109 | return list; |
| 1110 | } |
| 1111 | |
| 1112 | static PyObject* |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1113 | element_iter(ElementObject* self, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1114 | { |
| 1115 | PyObject* result; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1116 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1117 | PyObject* tag = Py_None; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1118 | if (!PyArg_ParseTuple(args, "|O:iter", &tag)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1119 | return NULL; |
| 1120 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1121 | if (!elementtree_iter_obj) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1122 | PyErr_SetString( |
| 1123 | PyExc_RuntimeError, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1124 | "iter helper not found" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1125 | ); |
| 1126 | return NULL; |
| 1127 | } |
| 1128 | |
| 1129 | args = PyTuple_New(2); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1130 | if (!args) |
| 1131 | return NULL; |
Neal Norwitz | 02876df | 2006-02-07 06:58:52 +0000 | [diff] [blame] | 1132 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1133 | Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self); |
| 1134 | Py_INCREF(tag); PyTuple_SET_ITEM(args, 1, (PyObject*) tag); |
| 1135 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1136 | result = PyObject_CallObject(elementtree_iter_obj, args); |
| 1137 | |
| 1138 | Py_DECREF(args); |
| 1139 | |
| 1140 | return result; |
| 1141 | } |
| 1142 | |
| 1143 | |
| 1144 | static PyObject* |
| 1145 | element_itertext(ElementObject* self, PyObject* args) |
| 1146 | { |
| 1147 | PyObject* result; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1148 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1149 | if (!PyArg_ParseTuple(args, ":itertext")) |
| 1150 | return NULL; |
| 1151 | |
| 1152 | if (!elementtree_itertext_obj) { |
| 1153 | PyErr_SetString( |
| 1154 | PyExc_RuntimeError, |
| 1155 | "itertext helper not found" |
| 1156 | ); |
| 1157 | return NULL; |
| 1158 | } |
| 1159 | |
| 1160 | args = PyTuple_New(1); |
| 1161 | if (!args) |
| 1162 | return NULL; |
| 1163 | |
| 1164 | Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self); |
| 1165 | |
| 1166 | result = PyObject_CallObject(elementtree_itertext_obj, args); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1167 | |
| 1168 | Py_DECREF(args); |
| 1169 | |
| 1170 | return result; |
| 1171 | } |
| 1172 | |
| 1173 | static PyObject* |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1174 | element_getitem(PyObject* self_, Py_ssize_t index) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1175 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1176 | ElementObject* self = (ElementObject*) self_; |
| 1177 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1178 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1179 | PyErr_SetString( |
| 1180 | PyExc_IndexError, |
| 1181 | "child index out of range" |
| 1182 | ); |
| 1183 | return NULL; |
| 1184 | } |
| 1185 | |
| 1186 | Py_INCREF(self->extra->children[index]); |
| 1187 | return self->extra->children[index]; |
| 1188 | } |
| 1189 | |
| 1190 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1191 | element_insert(ElementObject* self, PyObject* args) |
| 1192 | { |
| 1193 | int i; |
| 1194 | |
| 1195 | int index; |
| 1196 | PyObject* element; |
| 1197 | if (!PyArg_ParseTuple(args, "iO!:insert", &index, |
| 1198 | &Element_Type, &element)) |
| 1199 | return NULL; |
| 1200 | |
| 1201 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1202 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1203 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1204 | if (index < 0) { |
| 1205 | index += self->extra->length; |
| 1206 | if (index < 0) |
| 1207 | index = 0; |
| 1208 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1209 | if (index > self->extra->length) |
| 1210 | index = self->extra->length; |
| 1211 | |
| 1212 | if (element_resize(self, 1) < 0) |
| 1213 | return NULL; |
| 1214 | |
| 1215 | for (i = self->extra->length; i > index; i--) |
| 1216 | self->extra->children[i] = self->extra->children[i-1]; |
| 1217 | |
| 1218 | Py_INCREF(element); |
| 1219 | self->extra->children[index] = element; |
| 1220 | |
| 1221 | self->extra->length++; |
| 1222 | |
| 1223 | Py_RETURN_NONE; |
| 1224 | } |
| 1225 | |
| 1226 | static PyObject* |
| 1227 | element_items(ElementObject* self, PyObject* args) |
| 1228 | { |
| 1229 | if (!PyArg_ParseTuple(args, ":items")) |
| 1230 | return NULL; |
| 1231 | |
| 1232 | if (!self->extra || self->extra->attrib == Py_None) |
| 1233 | return PyList_New(0); |
| 1234 | |
| 1235 | return PyDict_Items(self->extra->attrib); |
| 1236 | } |
| 1237 | |
| 1238 | static PyObject* |
| 1239 | element_keys(ElementObject* self, PyObject* args) |
| 1240 | { |
| 1241 | if (!PyArg_ParseTuple(args, ":keys")) |
| 1242 | return NULL; |
| 1243 | |
| 1244 | if (!self->extra || self->extra->attrib == Py_None) |
| 1245 | return PyList_New(0); |
| 1246 | |
| 1247 | return PyDict_Keys(self->extra->attrib); |
| 1248 | } |
| 1249 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1250 | static Py_ssize_t |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1251 | element_length(ElementObject* self) |
| 1252 | { |
| 1253 | if (!self->extra) |
| 1254 | return 0; |
| 1255 | |
| 1256 | return self->extra->length; |
| 1257 | } |
| 1258 | |
| 1259 | static PyObject* |
| 1260 | element_makeelement(PyObject* self, PyObject* args, PyObject* kw) |
| 1261 | { |
| 1262 | PyObject* elem; |
| 1263 | |
| 1264 | PyObject* tag; |
| 1265 | PyObject* attrib; |
| 1266 | if (!PyArg_ParseTuple(args, "OO:makeelement", &tag, &attrib)) |
| 1267 | return NULL; |
| 1268 | |
| 1269 | attrib = PyDict_Copy(attrib); |
| 1270 | if (!attrib) |
| 1271 | return NULL; |
| 1272 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1273 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1274 | |
| 1275 | Py_DECREF(attrib); |
| 1276 | |
| 1277 | return elem; |
| 1278 | } |
| 1279 | |
| 1280 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1281 | element_remove(ElementObject* self, PyObject* args) |
| 1282 | { |
| 1283 | int i; |
| 1284 | |
| 1285 | PyObject* element; |
| 1286 | if (!PyArg_ParseTuple(args, "O!:remove", &Element_Type, &element)) |
| 1287 | return NULL; |
| 1288 | |
| 1289 | if (!self->extra) { |
| 1290 | /* element has no children, so raise exception */ |
| 1291 | PyErr_SetString( |
| 1292 | PyExc_ValueError, |
| 1293 | "list.remove(x): x not in list" |
| 1294 | ); |
| 1295 | return NULL; |
| 1296 | } |
| 1297 | |
| 1298 | for (i = 0; i < self->extra->length; i++) { |
| 1299 | if (self->extra->children[i] == element) |
| 1300 | break; |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1301 | if (PyObject_RichCompareBool(self->extra->children[i], element, Py_EQ) == 1) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1302 | break; |
| 1303 | } |
| 1304 | |
| 1305 | if (i == self->extra->length) { |
| 1306 | /* element is not in children, so raise exception */ |
| 1307 | PyErr_SetString( |
| 1308 | PyExc_ValueError, |
| 1309 | "list.remove(x): x not in list" |
| 1310 | ); |
| 1311 | return NULL; |
| 1312 | } |
| 1313 | |
| 1314 | Py_DECREF(self->extra->children[i]); |
| 1315 | |
| 1316 | self->extra->length--; |
| 1317 | |
| 1318 | for (; i < self->extra->length; i++) |
| 1319 | self->extra->children[i] = self->extra->children[i+1]; |
| 1320 | |
| 1321 | Py_RETURN_NONE; |
| 1322 | } |
| 1323 | |
| 1324 | static PyObject* |
| 1325 | element_repr(ElementObject* self) |
| 1326 | { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1327 | if (self->tag) |
| 1328 | return PyUnicode_FromFormat("<Element %R at %p>", self->tag, self); |
| 1329 | else |
| 1330 | return PyUnicode_FromFormat("<Element at %p>", self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1331 | } |
| 1332 | |
| 1333 | static PyObject* |
| 1334 | element_set(ElementObject* self, PyObject* args) |
| 1335 | { |
| 1336 | PyObject* attrib; |
| 1337 | |
| 1338 | PyObject* key; |
| 1339 | PyObject* value; |
| 1340 | if (!PyArg_ParseTuple(args, "OO:set", &key, &value)) |
| 1341 | return NULL; |
| 1342 | |
| 1343 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1344 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1345 | |
| 1346 | attrib = element_get_attrib(self); |
| 1347 | if (!attrib) |
| 1348 | return NULL; |
| 1349 | |
| 1350 | if (PyDict_SetItem(attrib, key, value) < 0) |
| 1351 | return NULL; |
| 1352 | |
| 1353 | Py_RETURN_NONE; |
| 1354 | } |
| 1355 | |
| 1356 | static int |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1357 | element_setitem(PyObject* self_, Py_ssize_t index, PyObject* item) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1358 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1359 | ElementObject* self = (ElementObject*) self_; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1360 | int i; |
| 1361 | PyObject* old; |
| 1362 | |
| 1363 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1364 | PyErr_SetString( |
| 1365 | PyExc_IndexError, |
| 1366 | "child assignment index out of range"); |
| 1367 | return -1; |
| 1368 | } |
| 1369 | |
| 1370 | old = self->extra->children[index]; |
| 1371 | |
| 1372 | if (item) { |
| 1373 | Py_INCREF(item); |
| 1374 | self->extra->children[index] = item; |
| 1375 | } else { |
| 1376 | self->extra->length--; |
| 1377 | for (i = index; i < self->extra->length; i++) |
| 1378 | self->extra->children[i] = self->extra->children[i+1]; |
| 1379 | } |
| 1380 | |
| 1381 | Py_DECREF(old); |
| 1382 | |
| 1383 | return 0; |
| 1384 | } |
| 1385 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1386 | static PyObject* |
| 1387 | element_subscr(PyObject* self_, PyObject* item) |
| 1388 | { |
| 1389 | ElementObject* self = (ElementObject*) self_; |
| 1390 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1391 | if (PyIndex_Check(item)) { |
| 1392 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1393 | |
| 1394 | if (i == -1 && PyErr_Occurred()) { |
| 1395 | return NULL; |
| 1396 | } |
| 1397 | if (i < 0 && self->extra) |
| 1398 | i += self->extra->length; |
| 1399 | return element_getitem(self_, i); |
| 1400 | } |
| 1401 | else if (PySlice_Check(item)) { |
| 1402 | Py_ssize_t start, stop, step, slicelen, cur, i; |
| 1403 | PyObject* list; |
| 1404 | |
| 1405 | if (!self->extra) |
| 1406 | return PyList_New(0); |
| 1407 | |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 1408 | if (PySlice_GetIndicesEx(item, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1409 | self->extra->length, |
| 1410 | &start, &stop, &step, &slicelen) < 0) { |
| 1411 | return NULL; |
| 1412 | } |
| 1413 | |
| 1414 | if (slicelen <= 0) |
| 1415 | return PyList_New(0); |
| 1416 | else { |
| 1417 | list = PyList_New(slicelen); |
| 1418 | if (!list) |
| 1419 | return NULL; |
| 1420 | |
| 1421 | for (cur = start, i = 0; i < slicelen; |
| 1422 | cur += step, i++) { |
| 1423 | PyObject* item = self->extra->children[cur]; |
| 1424 | Py_INCREF(item); |
| 1425 | PyList_SET_ITEM(list, i, item); |
| 1426 | } |
| 1427 | |
| 1428 | return list; |
| 1429 | } |
| 1430 | } |
| 1431 | else { |
| 1432 | PyErr_SetString(PyExc_TypeError, |
| 1433 | "element indices must be integers"); |
| 1434 | return NULL; |
| 1435 | } |
| 1436 | } |
| 1437 | |
| 1438 | static int |
| 1439 | element_ass_subscr(PyObject* self_, PyObject* item, PyObject* value) |
| 1440 | { |
| 1441 | ElementObject* self = (ElementObject*) self_; |
| 1442 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1443 | if (PyIndex_Check(item)) { |
| 1444 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1445 | |
| 1446 | if (i == -1 && PyErr_Occurred()) { |
| 1447 | return -1; |
| 1448 | } |
| 1449 | if (i < 0 && self->extra) |
| 1450 | i += self->extra->length; |
| 1451 | return element_setitem(self_, i, value); |
| 1452 | } |
| 1453 | else if (PySlice_Check(item)) { |
| 1454 | Py_ssize_t start, stop, step, slicelen, newlen, cur, i; |
| 1455 | |
| 1456 | PyObject* recycle = NULL; |
| 1457 | PyObject* seq = NULL; |
| 1458 | |
| 1459 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1460 | create_extra(self, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1461 | |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 1462 | if (PySlice_GetIndicesEx(item, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1463 | self->extra->length, |
| 1464 | &start, &stop, &step, &slicelen) < 0) { |
| 1465 | return -1; |
| 1466 | } |
| 1467 | |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1468 | if (value == NULL) { |
| 1469 | /* Delete slice */ |
| 1470 | size_t cur; |
| 1471 | Py_ssize_t i; |
| 1472 | |
| 1473 | if (slicelen <= 0) |
| 1474 | return 0; |
| 1475 | |
| 1476 | /* Since we're deleting, the direction of the range doesn't matter, |
| 1477 | * so for simplicity make it always ascending. |
| 1478 | */ |
| 1479 | if (step < 0) { |
| 1480 | stop = start + 1; |
| 1481 | start = stop + step * (slicelen - 1) - 1; |
| 1482 | step = -step; |
| 1483 | } |
| 1484 | |
| 1485 | assert((size_t)slicelen <= PY_SIZE_MAX / sizeof(PyObject *)); |
| 1486 | |
| 1487 | /* recycle is a list that will contain all the children |
| 1488 | * scheduled for removal. |
| 1489 | */ |
| 1490 | if (!(recycle = PyList_New(slicelen))) { |
| 1491 | PyErr_NoMemory(); |
| 1492 | return -1; |
| 1493 | } |
| 1494 | |
| 1495 | /* This loop walks over all the children that have to be deleted, |
| 1496 | * with cur pointing at them. num_moved is the amount of children |
| 1497 | * until the next deleted child that have to be "shifted down" to |
| 1498 | * occupy the deleted's places. |
| 1499 | * Note that in the ith iteration, shifting is done i+i places down |
| 1500 | * because i children were already removed. |
| 1501 | */ |
| 1502 | for (cur = start, i = 0; cur < (size_t)stop; cur += step, ++i) { |
| 1503 | /* Compute how many children have to be moved, clipping at the |
| 1504 | * list end. |
| 1505 | */ |
| 1506 | Py_ssize_t num_moved = step - 1; |
| 1507 | if (cur + step >= (size_t)self->extra->length) { |
| 1508 | num_moved = self->extra->length - cur - 1; |
| 1509 | } |
| 1510 | |
| 1511 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1512 | |
| 1513 | memmove( |
| 1514 | self->extra->children + cur - i, |
| 1515 | self->extra->children + cur + 1, |
| 1516 | num_moved * sizeof(PyObject *)); |
| 1517 | } |
| 1518 | |
| 1519 | /* Leftover "tail" after the last removed child */ |
| 1520 | cur = start + (size_t)slicelen * step; |
| 1521 | if (cur < (size_t)self->extra->length) { |
| 1522 | memmove( |
| 1523 | self->extra->children + cur - slicelen, |
| 1524 | self->extra->children + cur, |
| 1525 | (self->extra->length - cur) * sizeof(PyObject *)); |
| 1526 | } |
| 1527 | |
| 1528 | self->extra->length -= slicelen; |
| 1529 | |
| 1530 | /* Discard the recycle list with all the deleted sub-elements */ |
| 1531 | Py_XDECREF(recycle); |
| 1532 | return 0; |
| 1533 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1534 | else { |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1535 | /* A new slice is actually being assigned */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1536 | seq = PySequence_Fast(value, ""); |
| 1537 | if (!seq) { |
| 1538 | PyErr_Format( |
| 1539 | PyExc_TypeError, |
| 1540 | "expected sequence, not \"%.200s\"", Py_TYPE(value)->tp_name |
| 1541 | ); |
| 1542 | return -1; |
| 1543 | } |
| 1544 | newlen = PySequence_Size(seq); |
| 1545 | } |
| 1546 | |
| 1547 | if (step != 1 && newlen != slicelen) |
| 1548 | { |
| 1549 | PyErr_Format(PyExc_ValueError, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1550 | "attempt to assign sequence of size %zd " |
| 1551 | "to extended slice of size %zd", |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1552 | newlen, slicelen |
| 1553 | ); |
| 1554 | return -1; |
| 1555 | } |
| 1556 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1557 | /* Resize before creating the recycle bin, to prevent refleaks. */ |
| 1558 | if (newlen > slicelen) { |
| 1559 | if (element_resize(self, newlen - slicelen) < 0) { |
| 1560 | if (seq) { |
| 1561 | Py_DECREF(seq); |
| 1562 | } |
| 1563 | return -1; |
| 1564 | } |
| 1565 | } |
| 1566 | |
| 1567 | if (slicelen > 0) { |
| 1568 | /* to avoid recursive calls to this method (via decref), move |
| 1569 | old items to the recycle bin here, and get rid of them when |
| 1570 | we're done modifying the element */ |
| 1571 | recycle = PyList_New(slicelen); |
| 1572 | if (!recycle) { |
| 1573 | if (seq) { |
| 1574 | Py_DECREF(seq); |
| 1575 | } |
| 1576 | return -1; |
| 1577 | } |
| 1578 | for (cur = start, i = 0; i < slicelen; |
| 1579 | cur += step, i++) |
| 1580 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1581 | } |
| 1582 | |
| 1583 | if (newlen < slicelen) { |
| 1584 | /* delete slice */ |
| 1585 | for (i = stop; i < self->extra->length; i++) |
| 1586 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1587 | } else if (newlen > slicelen) { |
| 1588 | /* insert slice */ |
| 1589 | for (i = self->extra->length-1; i >= stop; i--) |
| 1590 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1591 | } |
| 1592 | |
| 1593 | /* replace the slice */ |
| 1594 | for (cur = start, i = 0; i < newlen; |
| 1595 | cur += step, i++) { |
| 1596 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
| 1597 | Py_INCREF(element); |
| 1598 | self->extra->children[cur] = element; |
| 1599 | } |
| 1600 | |
| 1601 | self->extra->length += newlen - slicelen; |
| 1602 | |
| 1603 | if (seq) { |
| 1604 | Py_DECREF(seq); |
| 1605 | } |
| 1606 | |
| 1607 | /* discard the recycle bin, and everything in it */ |
| 1608 | Py_XDECREF(recycle); |
| 1609 | |
| 1610 | return 0; |
| 1611 | } |
| 1612 | else { |
| 1613 | PyErr_SetString(PyExc_TypeError, |
| 1614 | "element indices must be integers"); |
| 1615 | return -1; |
| 1616 | } |
| 1617 | } |
| 1618 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1619 | static PyMethodDef element_methods[] = { |
| 1620 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1621 | {"clear", (PyCFunction) element_clearmethod, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1622 | |
| 1623 | {"get", (PyCFunction) element_get, METH_VARARGS}, |
| 1624 | {"set", (PyCFunction) element_set, METH_VARARGS}, |
| 1625 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1626 | {"find", (PyCFunction) element_find, METH_VARARGS | METH_KEYWORDS}, |
| 1627 | {"findtext", (PyCFunction) element_findtext, METH_VARARGS | METH_KEYWORDS}, |
| 1628 | {"findall", (PyCFunction) element_findall, METH_VARARGS | METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1629 | |
| 1630 | {"append", (PyCFunction) element_append, METH_VARARGS}, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1631 | {"extend", (PyCFunction) element_extend, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1632 | {"insert", (PyCFunction) element_insert, METH_VARARGS}, |
| 1633 | {"remove", (PyCFunction) element_remove, METH_VARARGS}, |
| 1634 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1635 | {"iter", (PyCFunction) element_iter, METH_VARARGS}, |
| 1636 | {"itertext", (PyCFunction) element_itertext, METH_VARARGS}, |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1637 | {"iterfind", (PyCFunction) element_iterfind, METH_VARARGS | METH_KEYWORDS}, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1638 | |
| 1639 | {"getiterator", (PyCFunction) element_iter, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1640 | {"getchildren", (PyCFunction) element_getchildren, METH_VARARGS}, |
| 1641 | |
| 1642 | {"items", (PyCFunction) element_items, METH_VARARGS}, |
| 1643 | {"keys", (PyCFunction) element_keys, METH_VARARGS}, |
| 1644 | |
| 1645 | {"makeelement", (PyCFunction) element_makeelement, METH_VARARGS}, |
| 1646 | |
| 1647 | {"__copy__", (PyCFunction) element_copy, METH_VARARGS}, |
| 1648 | {"__deepcopy__", (PyCFunction) element_deepcopy, METH_VARARGS}, |
| 1649 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1650 | {NULL, NULL} |
| 1651 | }; |
| 1652 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1653 | static PyObject* |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1654 | element_getattro(ElementObject* self, PyObject* nameobj) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1655 | { |
| 1656 | PyObject* res; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1657 | char *name = ""; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1658 | |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1659 | if (PyUnicode_Check(nameobj)) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1660 | name = _PyUnicode_AsString(nameobj); |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1661 | |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 1662 | if (name == NULL) |
| 1663 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1664 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1665 | /* handle common attributes first */ |
| 1666 | if (strcmp(name, "tag") == 0) { |
| 1667 | res = self->tag; |
| 1668 | Py_INCREF(res); |
| 1669 | return res; |
| 1670 | } else if (strcmp(name, "text") == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1671 | res = element_get_text(self); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1672 | Py_INCREF(res); |
| 1673 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1674 | } |
| 1675 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1676 | /* methods */ |
| 1677 | res = PyObject_GenericGetAttr((PyObject*) self, nameobj); |
| 1678 | if (res) |
| 1679 | return res; |
| 1680 | |
| 1681 | /* less common attributes */ |
| 1682 | if (strcmp(name, "tail") == 0) { |
| 1683 | PyErr_Clear(); |
| 1684 | res = element_get_tail(self); |
| 1685 | } else if (strcmp(name, "attrib") == 0) { |
| 1686 | PyErr_Clear(); |
| 1687 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1688 | create_extra(self, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1689 | res = element_get_attrib(self); |
| 1690 | } |
| 1691 | |
| 1692 | if (!res) |
| 1693 | return NULL; |
| 1694 | |
| 1695 | Py_INCREF(res); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1696 | return res; |
| 1697 | } |
| 1698 | |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1699 | static PyObject* |
| 1700 | element_setattro(ElementObject* self, PyObject* nameobj, PyObject* value) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1701 | { |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1702 | char *name = ""; |
| 1703 | if (PyUnicode_Check(nameobj)) |
| 1704 | name = _PyUnicode_AsString(nameobj); |
| 1705 | |
| 1706 | if (name == NULL) |
| 1707 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1708 | |
| 1709 | if (strcmp(name, "tag") == 0) { |
| 1710 | Py_DECREF(self->tag); |
| 1711 | self->tag = value; |
| 1712 | Py_INCREF(self->tag); |
| 1713 | } else if (strcmp(name, "text") == 0) { |
| 1714 | Py_DECREF(JOIN_OBJ(self->text)); |
| 1715 | self->text = value; |
| 1716 | Py_INCREF(self->text); |
| 1717 | } else if (strcmp(name, "tail") == 0) { |
| 1718 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 1719 | self->tail = value; |
| 1720 | Py_INCREF(self->tail); |
| 1721 | } else if (strcmp(name, "attrib") == 0) { |
| 1722 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1723 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1724 | Py_DECREF(self->extra->attrib); |
| 1725 | self->extra->attrib = value; |
| 1726 | Py_INCREF(self->extra->attrib); |
| 1727 | } else { |
| 1728 | PyErr_SetString(PyExc_AttributeError, name); |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1729 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1730 | } |
| 1731 | |
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 | |
| 1735 | static PySequenceMethods element_as_sequence = { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1736 | (lenfunc) element_length, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1737 | 0, /* sq_concat */ |
| 1738 | 0, /* sq_repeat */ |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1739 | element_getitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1740 | 0, |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1741 | element_setitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1742 | 0, |
| 1743 | }; |
| 1744 | |
| 1745 | static PyMappingMethods element_as_mapping = { |
| 1746 | (lenfunc) element_length, |
| 1747 | (binaryfunc) element_subscr, |
| 1748 | (objobjargproc) element_ass_subscr, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1749 | }; |
| 1750 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 1751 | static PyTypeObject Element_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 1752 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1753 | "Element", sizeof(ElementObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1754 | /* methods */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1755 | (destructor)element_dealloc, /* tp_dealloc */ |
| 1756 | 0, /* tp_print */ |
| 1757 | 0, /* tp_getattr */ |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1758 | 0, /* tp_setattr */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1759 | 0, /* tp_reserved */ |
| 1760 | (reprfunc)element_repr, /* tp_repr */ |
| 1761 | 0, /* tp_as_number */ |
| 1762 | &element_as_sequence, /* tp_as_sequence */ |
| 1763 | &element_as_mapping, /* tp_as_mapping */ |
| 1764 | 0, /* tp_hash */ |
| 1765 | 0, /* tp_call */ |
| 1766 | 0, /* tp_str */ |
| 1767 | (getattrofunc)element_getattro, /* tp_getattro */ |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1768 | (setattrofunc)element_setattro, /* tp_setattro */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1769 | 0, /* tp_as_buffer */ |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1770 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 1771 | /* tp_flags */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1772 | 0, /* tp_doc */ |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1773 | (traverseproc)element_gc_traverse, /* tp_traverse */ |
| 1774 | (inquiry)element_gc_clear, /* tp_clear */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1775 | 0, /* tp_richcompare */ |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 1776 | offsetof(ElementObject, weakreflist), /* tp_weaklistoffset */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1777 | 0, /* tp_iter */ |
| 1778 | 0, /* tp_iternext */ |
| 1779 | element_methods, /* tp_methods */ |
| 1780 | 0, /* tp_members */ |
| 1781 | 0, /* tp_getset */ |
| 1782 | 0, /* tp_base */ |
| 1783 | 0, /* tp_dict */ |
| 1784 | 0, /* tp_descr_get */ |
| 1785 | 0, /* tp_descr_set */ |
| 1786 | 0, /* tp_dictoffset */ |
| 1787 | (initproc)element_init, /* tp_init */ |
| 1788 | PyType_GenericAlloc, /* tp_alloc */ |
| 1789 | element_new, /* tp_new */ |
| 1790 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1791 | }; |
| 1792 | |
| 1793 | /* ==================================================================== */ |
| 1794 | /* the tree builder type */ |
| 1795 | |
| 1796 | typedef struct { |
| 1797 | PyObject_HEAD |
| 1798 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1799 | PyObject *root; /* root node (first created node) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1800 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1801 | ElementObject *this; /* current node */ |
| 1802 | ElementObject *last; /* most recently 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 | PyObject *data; /* data collector (string or list), or NULL */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1805 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1806 | PyObject *stack; /* element stack */ |
| 1807 | Py_ssize_t index; /* current stack size (0 means empty) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1808 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1809 | PyObject *element_factory; |
| 1810 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1811 | /* element tracing */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1812 | PyObject *events; /* list of events, or NULL if not collecting */ |
| 1813 | PyObject *start_event_obj; /* event objects (NULL to ignore) */ |
| 1814 | PyObject *end_event_obj; |
| 1815 | PyObject *start_ns_event_obj; |
| 1816 | PyObject *end_ns_event_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1817 | } TreeBuilderObject; |
| 1818 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 1819 | static PyTypeObject TreeBuilder_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1820 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 1821 | #define TreeBuilder_CheckExact(op) (Py_TYPE(op) == &TreeBuilder_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1822 | |
| 1823 | /* -------------------------------------------------------------------- */ |
| 1824 | /* constructor and destructor */ |
| 1825 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1826 | static PyObject * |
| 1827 | treebuilder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1828 | { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1829 | TreeBuilderObject *t = (TreeBuilderObject *)type->tp_alloc(type, 0); |
| 1830 | if (t != NULL) { |
| 1831 | t->root = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1832 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1833 | Py_INCREF(Py_None); |
| 1834 | t->this = (ElementObject *)Py_None; |
| 1835 | Py_INCREF(Py_None); |
| 1836 | t->last = (ElementObject *)Py_None; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1837 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1838 | t->data = NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1839 | t->element_factory = NULL; |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1840 | t->stack = PyList_New(20); |
| 1841 | if (!t->stack) { |
| 1842 | Py_DECREF(t->this); |
| 1843 | Py_DECREF(t->last); |
| 1844 | return NULL; |
| 1845 | } |
| 1846 | t->index = 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1847 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1848 | t->events = NULL; |
| 1849 | t->start_event_obj = t->end_event_obj = NULL; |
| 1850 | t->start_ns_event_obj = t->end_ns_event_obj = NULL; |
| 1851 | } |
| 1852 | return (PyObject *)t; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1853 | } |
| 1854 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1855 | static int |
| 1856 | treebuilder_init(PyObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1857 | { |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1858 | static char *kwlist[] = {"element_factory", NULL}; |
| 1859 | PyObject *element_factory = NULL; |
| 1860 | TreeBuilderObject *self_tb = (TreeBuilderObject *)self; |
| 1861 | |
| 1862 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O:TreeBuilder", kwlist, |
| 1863 | &element_factory)) { |
| 1864 | return -1; |
| 1865 | } |
| 1866 | |
| 1867 | if (element_factory) { |
| 1868 | Py_INCREF(element_factory); |
| 1869 | Py_XDECREF(self_tb->element_factory); |
| 1870 | self_tb->element_factory = element_factory; |
| 1871 | } |
| 1872 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1873 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1874 | } |
| 1875 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1876 | static int |
| 1877 | treebuilder_gc_traverse(TreeBuilderObject *self, visitproc visit, void *arg) |
| 1878 | { |
| 1879 | Py_VISIT(self->root); |
| 1880 | Py_VISIT(self->this); |
| 1881 | Py_VISIT(self->last); |
| 1882 | Py_VISIT(self->data); |
| 1883 | Py_VISIT(self->stack); |
| 1884 | Py_VISIT(self->element_factory); |
| 1885 | return 0; |
| 1886 | } |
| 1887 | |
| 1888 | static int |
| 1889 | treebuilder_gc_clear(TreeBuilderObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1890 | { |
| 1891 | Py_XDECREF(self->end_ns_event_obj); |
| 1892 | Py_XDECREF(self->start_ns_event_obj); |
| 1893 | Py_XDECREF(self->end_event_obj); |
| 1894 | Py_XDECREF(self->start_event_obj); |
| 1895 | Py_XDECREF(self->events); |
| 1896 | Py_DECREF(self->stack); |
| 1897 | Py_XDECREF(self->data); |
| 1898 | Py_DECREF(self->last); |
| 1899 | Py_DECREF(self->this); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1900 | Py_CLEAR(self->element_factory); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1901 | Py_XDECREF(self->root); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1902 | return 0; |
| 1903 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1904 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1905 | static void |
| 1906 | treebuilder_dealloc(TreeBuilderObject *self) |
| 1907 | { |
| 1908 | PyObject_GC_UnTrack(self); |
| 1909 | treebuilder_gc_clear(self); |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 1910 | Py_TYPE(self)->tp_free((PyObject *)self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1911 | } |
| 1912 | |
| 1913 | /* -------------------------------------------------------------------- */ |
| 1914 | /* handlers */ |
| 1915 | |
| 1916 | LOCAL(PyObject*) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1917 | treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag, |
| 1918 | PyObject* attrib) |
| 1919 | { |
| 1920 | PyObject* node; |
| 1921 | PyObject* this; |
| 1922 | |
| 1923 | if (self->data) { |
| 1924 | if (self->this == self->last) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1925 | Py_DECREF(JOIN_OBJ(self->last->text)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1926 | self->last->text = JOIN_SET( |
| 1927 | self->data, PyList_CheckExact(self->data) |
| 1928 | ); |
| 1929 | } else { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1930 | Py_DECREF(JOIN_OBJ(self->last->tail)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1931 | self->last->tail = JOIN_SET( |
| 1932 | self->data, PyList_CheckExact(self->data) |
| 1933 | ); |
| 1934 | } |
| 1935 | self->data = NULL; |
| 1936 | } |
| 1937 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1938 | if (self->element_factory) { |
| 1939 | node = PyObject_CallFunction(self->element_factory, "OO", tag, attrib); |
| 1940 | } else { |
| 1941 | node = create_new_element(tag, attrib); |
| 1942 | } |
| 1943 | if (!node) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1944 | return NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 1945 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1946 | |
| 1947 | this = (PyObject*) self->this; |
| 1948 | |
| 1949 | if (this != Py_None) { |
| 1950 | if (element_add_subelement((ElementObject*) this, node) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1951 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1952 | } else { |
| 1953 | if (self->root) { |
| 1954 | PyErr_SetString( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1955 | elementtree_parseerror_obj, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1956 | "multiple elements on top level" |
| 1957 | ); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1958 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1959 | } |
| 1960 | Py_INCREF(node); |
| 1961 | self->root = node; |
| 1962 | } |
| 1963 | |
| 1964 | if (self->index < PyList_GET_SIZE(self->stack)) { |
| 1965 | if (PyList_SetItem(self->stack, self->index, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1966 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1967 | Py_INCREF(this); |
| 1968 | } else { |
| 1969 | if (PyList_Append(self->stack, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1970 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1971 | } |
| 1972 | self->index++; |
| 1973 | |
| 1974 | Py_DECREF(this); |
| 1975 | Py_INCREF(node); |
| 1976 | self->this = (ElementObject*) node; |
| 1977 | |
| 1978 | Py_DECREF(self->last); |
| 1979 | Py_INCREF(node); |
| 1980 | self->last = (ElementObject*) node; |
| 1981 | |
| 1982 | if (self->start_event_obj) { |
| 1983 | PyObject* res; |
| 1984 | PyObject* action = self->start_event_obj; |
| 1985 | res = PyTuple_New(2); |
| 1986 | if (res) { |
| 1987 | Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action); |
| 1988 | Py_INCREF(node); PyTuple_SET_ITEM(res, 1, (PyObject*) node); |
| 1989 | PyList_Append(self->events, res); |
| 1990 | Py_DECREF(res); |
| 1991 | } else |
| 1992 | PyErr_Clear(); /* FIXME: propagate error */ |
| 1993 | } |
| 1994 | |
| 1995 | return node; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1996 | |
| 1997 | error: |
| 1998 | Py_DECREF(node); |
| 1999 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2000 | } |
| 2001 | |
| 2002 | LOCAL(PyObject*) |
| 2003 | treebuilder_handle_data(TreeBuilderObject* self, PyObject* data) |
| 2004 | { |
| 2005 | if (!self->data) { |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 2006 | if (self->last == (ElementObject*) Py_None) { |
| 2007 | /* ignore calls to data before the first call to start */ |
| 2008 | Py_RETURN_NONE; |
| 2009 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2010 | /* store the first item as is */ |
| 2011 | Py_INCREF(data); self->data = data; |
| 2012 | } else { |
| 2013 | /* more than one item; use a list to collect items */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2014 | if (PyBytes_CheckExact(self->data) && Py_REFCNT(self->data) == 1 && |
| 2015 | PyBytes_CheckExact(data) && PyBytes_GET_SIZE(data) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2016 | /* expat often generates single character data sections; handle |
| 2017 | the most common case by resizing the existing string... */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2018 | Py_ssize_t size = PyBytes_GET_SIZE(self->data); |
| 2019 | if (_PyBytes_Resize(&self->data, size + 1) < 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2020 | return NULL; |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2021 | PyBytes_AS_STRING(self->data)[size] = PyBytes_AS_STRING(data)[0]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2022 | } else if (PyList_CheckExact(self->data)) { |
| 2023 | if (PyList_Append(self->data, data) < 0) |
| 2024 | return NULL; |
| 2025 | } else { |
| 2026 | PyObject* list = PyList_New(2); |
| 2027 | if (!list) |
| 2028 | return NULL; |
| 2029 | PyList_SET_ITEM(list, 0, self->data); |
| 2030 | Py_INCREF(data); PyList_SET_ITEM(list, 1, data); |
| 2031 | self->data = list; |
| 2032 | } |
| 2033 | } |
| 2034 | |
| 2035 | Py_RETURN_NONE; |
| 2036 | } |
| 2037 | |
| 2038 | LOCAL(PyObject*) |
| 2039 | treebuilder_handle_end(TreeBuilderObject* self, PyObject* tag) |
| 2040 | { |
| 2041 | PyObject* item; |
| 2042 | |
| 2043 | if (self->data) { |
| 2044 | if (self->this == self->last) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2045 | Py_DECREF(JOIN_OBJ(self->last->text)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2046 | self->last->text = JOIN_SET( |
| 2047 | self->data, PyList_CheckExact(self->data) |
| 2048 | ); |
| 2049 | } else { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2050 | Py_DECREF(JOIN_OBJ(self->last->tail)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2051 | self->last->tail = JOIN_SET( |
| 2052 | self->data, PyList_CheckExact(self->data) |
| 2053 | ); |
| 2054 | } |
| 2055 | self->data = NULL; |
| 2056 | } |
| 2057 | |
| 2058 | if (self->index == 0) { |
| 2059 | PyErr_SetString( |
| 2060 | PyExc_IndexError, |
| 2061 | "pop from empty stack" |
| 2062 | ); |
| 2063 | return NULL; |
| 2064 | } |
| 2065 | |
| 2066 | self->index--; |
| 2067 | |
| 2068 | item = PyList_GET_ITEM(self->stack, self->index); |
| 2069 | Py_INCREF(item); |
| 2070 | |
| 2071 | Py_DECREF(self->last); |
| 2072 | |
| 2073 | self->last = (ElementObject*) self->this; |
| 2074 | self->this = (ElementObject*) item; |
| 2075 | |
| 2076 | if (self->end_event_obj) { |
| 2077 | PyObject* res; |
| 2078 | PyObject* action = self->end_event_obj; |
| 2079 | PyObject* node = (PyObject*) self->last; |
| 2080 | res = PyTuple_New(2); |
| 2081 | if (res) { |
| 2082 | Py_INCREF(action); PyTuple_SET_ITEM(res, 0, (PyObject*) action); |
| 2083 | Py_INCREF(node); PyTuple_SET_ITEM(res, 1, (PyObject*) node); |
| 2084 | PyList_Append(self->events, res); |
| 2085 | Py_DECREF(res); |
| 2086 | } else |
| 2087 | PyErr_Clear(); /* FIXME: propagate error */ |
| 2088 | } |
| 2089 | |
| 2090 | Py_INCREF(self->last); |
| 2091 | return (PyObject*) self->last; |
| 2092 | } |
| 2093 | |
| 2094 | LOCAL(void) |
| 2095 | treebuilder_handle_namespace(TreeBuilderObject* self, int start, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2096 | PyObject *prefix, PyObject *uri) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2097 | { |
| 2098 | PyObject* res; |
| 2099 | PyObject* action; |
| 2100 | PyObject* parcel; |
| 2101 | |
| 2102 | if (!self->events) |
| 2103 | return; |
| 2104 | |
| 2105 | if (start) { |
| 2106 | if (!self->start_ns_event_obj) |
| 2107 | return; |
| 2108 | action = self->start_ns_event_obj; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2109 | parcel = Py_BuildValue("OO", prefix, uri); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2110 | if (!parcel) |
| 2111 | return; |
| 2112 | Py_INCREF(action); |
| 2113 | } else { |
| 2114 | if (!self->end_ns_event_obj) |
| 2115 | return; |
| 2116 | action = self->end_ns_event_obj; |
| 2117 | Py_INCREF(action); |
| 2118 | parcel = Py_None; |
| 2119 | Py_INCREF(parcel); |
| 2120 | } |
| 2121 | |
| 2122 | res = PyTuple_New(2); |
| 2123 | |
| 2124 | if (res) { |
| 2125 | PyTuple_SET_ITEM(res, 0, action); |
| 2126 | PyTuple_SET_ITEM(res, 1, parcel); |
| 2127 | PyList_Append(self->events, res); |
| 2128 | Py_DECREF(res); |
| 2129 | } else |
| 2130 | PyErr_Clear(); /* FIXME: propagate error */ |
| 2131 | } |
| 2132 | |
| 2133 | /* -------------------------------------------------------------------- */ |
| 2134 | /* methods (in alphabetical order) */ |
| 2135 | |
| 2136 | static PyObject* |
| 2137 | treebuilder_data(TreeBuilderObject* self, PyObject* args) |
| 2138 | { |
| 2139 | PyObject* data; |
| 2140 | if (!PyArg_ParseTuple(args, "O:data", &data)) |
| 2141 | return NULL; |
| 2142 | |
| 2143 | return treebuilder_handle_data(self, data); |
| 2144 | } |
| 2145 | |
| 2146 | static PyObject* |
| 2147 | treebuilder_end(TreeBuilderObject* self, PyObject* args) |
| 2148 | { |
| 2149 | PyObject* tag; |
| 2150 | if (!PyArg_ParseTuple(args, "O:end", &tag)) |
| 2151 | return NULL; |
| 2152 | |
| 2153 | return treebuilder_handle_end(self, tag); |
| 2154 | } |
| 2155 | |
| 2156 | LOCAL(PyObject*) |
| 2157 | treebuilder_done(TreeBuilderObject* self) |
| 2158 | { |
| 2159 | PyObject* res; |
| 2160 | |
| 2161 | /* FIXME: check stack size? */ |
| 2162 | |
| 2163 | if (self->root) |
| 2164 | res = self->root; |
| 2165 | else |
| 2166 | res = Py_None; |
| 2167 | |
| 2168 | Py_INCREF(res); |
| 2169 | return res; |
| 2170 | } |
| 2171 | |
| 2172 | static PyObject* |
| 2173 | treebuilder_close(TreeBuilderObject* self, PyObject* args) |
| 2174 | { |
| 2175 | if (!PyArg_ParseTuple(args, ":close")) |
| 2176 | return NULL; |
| 2177 | |
| 2178 | return treebuilder_done(self); |
| 2179 | } |
| 2180 | |
| 2181 | static PyObject* |
| 2182 | treebuilder_start(TreeBuilderObject* self, PyObject* args) |
| 2183 | { |
| 2184 | PyObject* tag; |
| 2185 | PyObject* attrib = Py_None; |
| 2186 | if (!PyArg_ParseTuple(args, "O|O:start", &tag, &attrib)) |
| 2187 | return NULL; |
| 2188 | |
| 2189 | return treebuilder_handle_start(self, tag, attrib); |
| 2190 | } |
| 2191 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2192 | static PyMethodDef treebuilder_methods[] = { |
| 2193 | {"data", (PyCFunction) treebuilder_data, METH_VARARGS}, |
| 2194 | {"start", (PyCFunction) treebuilder_start, METH_VARARGS}, |
| 2195 | {"end", (PyCFunction) treebuilder_end, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2196 | {"close", (PyCFunction) treebuilder_close, METH_VARARGS}, |
| 2197 | {NULL, NULL} |
| 2198 | }; |
| 2199 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 2200 | static PyTypeObject TreeBuilder_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 2201 | PyVarObject_HEAD_INIT(NULL, 0) |
| 2202 | "TreeBuilder", sizeof(TreeBuilderObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2203 | /* methods */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2204 | (destructor)treebuilder_dealloc, /* tp_dealloc */ |
| 2205 | 0, /* tp_print */ |
| 2206 | 0, /* tp_getattr */ |
| 2207 | 0, /* tp_setattr */ |
| 2208 | 0, /* tp_reserved */ |
| 2209 | 0, /* tp_repr */ |
| 2210 | 0, /* tp_as_number */ |
| 2211 | 0, /* tp_as_sequence */ |
| 2212 | 0, /* tp_as_mapping */ |
| 2213 | 0, /* tp_hash */ |
| 2214 | 0, /* tp_call */ |
| 2215 | 0, /* tp_str */ |
| 2216 | 0, /* tp_getattro */ |
| 2217 | 0, /* tp_setattro */ |
| 2218 | 0, /* tp_as_buffer */ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2219 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 2220 | /* tp_flags */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2221 | 0, /* tp_doc */ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2222 | (traverseproc)treebuilder_gc_traverse, /* tp_traverse */ |
| 2223 | (inquiry)treebuilder_gc_clear, /* tp_clear */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2224 | 0, /* tp_richcompare */ |
| 2225 | 0, /* tp_weaklistoffset */ |
| 2226 | 0, /* tp_iter */ |
| 2227 | 0, /* tp_iternext */ |
| 2228 | treebuilder_methods, /* tp_methods */ |
| 2229 | 0, /* tp_members */ |
| 2230 | 0, /* tp_getset */ |
| 2231 | 0, /* tp_base */ |
| 2232 | 0, /* tp_dict */ |
| 2233 | 0, /* tp_descr_get */ |
| 2234 | 0, /* tp_descr_set */ |
| 2235 | 0, /* tp_dictoffset */ |
| 2236 | (initproc)treebuilder_init, /* tp_init */ |
| 2237 | PyType_GenericAlloc, /* tp_alloc */ |
| 2238 | treebuilder_new, /* tp_new */ |
| 2239 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2240 | }; |
| 2241 | |
| 2242 | /* ==================================================================== */ |
| 2243 | /* the expat interface */ |
| 2244 | |
| 2245 | #if defined(USE_EXPAT) |
| 2246 | |
| 2247 | #include "expat.h" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2248 | #include "pyexpat.h" |
Eli Bendersky | 20d4174 | 2012-06-01 09:48:37 +0300 | [diff] [blame] | 2249 | static struct PyExpat_CAPI *expat_capi; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2250 | #define EXPAT(func) (expat_capi->func) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2251 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2252 | static XML_Memory_Handling_Suite ExpatMemoryHandler = { |
| 2253 | PyObject_Malloc, PyObject_Realloc, PyObject_Free}; |
| 2254 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2255 | typedef struct { |
| 2256 | PyObject_HEAD |
| 2257 | |
| 2258 | XML_Parser parser; |
| 2259 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2260 | PyObject *target; |
| 2261 | PyObject *entity; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2262 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2263 | PyObject *names; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2264 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2265 | PyObject *handle_start; |
| 2266 | PyObject *handle_data; |
| 2267 | PyObject *handle_end; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2268 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2269 | PyObject *handle_comment; |
| 2270 | PyObject *handle_pi; |
| 2271 | PyObject *handle_doctype; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2272 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2273 | PyObject *handle_close; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2274 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2275 | } XMLParserObject; |
| 2276 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 2277 | static PyTypeObject XMLParser_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2278 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2279 | #define XMLParser_CheckExact(op) (Py_TYPE(op) == &XMLParser_Type) |
| 2280 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2281 | /* helpers */ |
| 2282 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2283 | LOCAL(PyObject*) |
| 2284 | makeuniversal(XMLParserObject* self, const char* string) |
| 2285 | { |
| 2286 | /* convert a UTF-8 tag/attribute name from the expat parser |
| 2287 | to a universal name string */ |
| 2288 | |
| 2289 | int size = strlen(string); |
| 2290 | PyObject* key; |
| 2291 | PyObject* value; |
| 2292 | |
| 2293 | /* look the 'raw' name up in the names dictionary */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2294 | key = PyBytes_FromStringAndSize(string, size); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2295 | if (!key) |
| 2296 | return NULL; |
| 2297 | |
| 2298 | value = PyDict_GetItem(self->names, key); |
| 2299 | |
| 2300 | if (value) { |
| 2301 | Py_INCREF(value); |
| 2302 | } else { |
| 2303 | /* new name. convert to universal name, and decode as |
| 2304 | necessary */ |
| 2305 | |
| 2306 | PyObject* tag; |
| 2307 | char* p; |
| 2308 | int i; |
| 2309 | |
| 2310 | /* look for namespace separator */ |
| 2311 | for (i = 0; i < size; i++) |
| 2312 | if (string[i] == '}') |
| 2313 | break; |
| 2314 | if (i != size) { |
| 2315 | /* convert to universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2316 | tag = PyBytes_FromStringAndSize(NULL, size+1); |
| 2317 | p = PyBytes_AS_STRING(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2318 | p[0] = '{'; |
| 2319 | memcpy(p+1, string, size); |
| 2320 | size++; |
| 2321 | } else { |
| 2322 | /* plain name; use key as tag */ |
| 2323 | Py_INCREF(key); |
| 2324 | tag = key; |
| 2325 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2326 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2327 | /* decode universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2328 | p = PyBytes_AS_STRING(tag); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2329 | value = PyUnicode_DecodeUTF8(p, size, "strict"); |
| 2330 | Py_DECREF(tag); |
| 2331 | if (!value) { |
| 2332 | Py_DECREF(key); |
| 2333 | return NULL; |
| 2334 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2335 | |
| 2336 | /* add to names dictionary */ |
| 2337 | if (PyDict_SetItem(self->names, key, value) < 0) { |
| 2338 | Py_DECREF(key); |
| 2339 | Py_DECREF(value); |
| 2340 | return NULL; |
| 2341 | } |
| 2342 | } |
| 2343 | |
| 2344 | Py_DECREF(key); |
| 2345 | return value; |
| 2346 | } |
| 2347 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2348 | /* Set the ParseError exception with the given parameters. |
| 2349 | * If message is not NULL, it's used as the error string. Otherwise, the |
| 2350 | * message string is the default for the given error_code. |
| 2351 | */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2352 | static void |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2353 | 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] | 2354 | { |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2355 | PyObject *errmsg, *error, *position, *code; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2356 | |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2357 | errmsg = PyUnicode_FromFormat("%s: line %d, column %d", |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2358 | message ? message : EXPAT(ErrorString)(error_code), |
| 2359 | line, column); |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2360 | if (errmsg == NULL) |
| 2361 | return; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2362 | |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2363 | error = PyObject_CallFunction(elementtree_parseerror_obj, "O", errmsg); |
| 2364 | Py_DECREF(errmsg); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2365 | if (!error) |
| 2366 | return; |
| 2367 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2368 | /* Add code and position attributes */ |
| 2369 | code = PyLong_FromLong((long)error_code); |
| 2370 | if (!code) { |
| 2371 | Py_DECREF(error); |
| 2372 | return; |
| 2373 | } |
| 2374 | if (PyObject_SetAttrString(error, "code", code) == -1) { |
| 2375 | Py_DECREF(error); |
| 2376 | Py_DECREF(code); |
| 2377 | return; |
| 2378 | } |
| 2379 | Py_DECREF(code); |
| 2380 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2381 | position = Py_BuildValue("(ii)", line, column); |
| 2382 | if (!position) { |
| 2383 | Py_DECREF(error); |
| 2384 | return; |
| 2385 | } |
| 2386 | if (PyObject_SetAttrString(error, "position", position) == -1) { |
| 2387 | Py_DECREF(error); |
| 2388 | Py_DECREF(position); |
| 2389 | return; |
| 2390 | } |
| 2391 | Py_DECREF(position); |
| 2392 | |
| 2393 | PyErr_SetObject(elementtree_parseerror_obj, error); |
| 2394 | Py_DECREF(error); |
| 2395 | } |
| 2396 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2397 | /* -------------------------------------------------------------------- */ |
| 2398 | /* handlers */ |
| 2399 | |
| 2400 | static void |
| 2401 | expat_default_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2402 | int data_len) |
| 2403 | { |
| 2404 | PyObject* key; |
| 2405 | PyObject* value; |
| 2406 | PyObject* res; |
| 2407 | |
| 2408 | if (data_len < 2 || data_in[0] != '&') |
| 2409 | return; |
| 2410 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2411 | key = PyUnicode_DecodeUTF8(data_in + 1, data_len - 2, "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2412 | if (!key) |
| 2413 | return; |
| 2414 | |
| 2415 | value = PyDict_GetItem(self->entity, key); |
| 2416 | |
| 2417 | if (value) { |
| 2418 | if (TreeBuilder_CheckExact(self->target)) |
| 2419 | res = treebuilder_handle_data( |
| 2420 | (TreeBuilderObject*) self->target, value |
| 2421 | ); |
| 2422 | else if (self->handle_data) |
| 2423 | res = PyObject_CallFunction(self->handle_data, "O", value); |
| 2424 | else |
| 2425 | res = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2426 | Py_XDECREF(res); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2427 | } else if (!PyErr_Occurred()) { |
| 2428 | /* Report the first error, not the last */ |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 2429 | char message[128] = "undefined entity "; |
| 2430 | strncat(message, data_in, data_len < 100?data_len:100); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2431 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2432 | XML_ERROR_UNDEFINED_ENTITY, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2433 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2434 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 2435 | message |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2436 | ); |
| 2437 | } |
| 2438 | |
| 2439 | Py_DECREF(key); |
| 2440 | } |
| 2441 | |
| 2442 | static void |
| 2443 | expat_start_handler(XMLParserObject* self, const XML_Char* tag_in, |
| 2444 | const XML_Char **attrib_in) |
| 2445 | { |
| 2446 | PyObject* res; |
| 2447 | PyObject* tag; |
| 2448 | PyObject* attrib; |
| 2449 | int ok; |
| 2450 | |
| 2451 | /* tag name */ |
| 2452 | tag = makeuniversal(self, tag_in); |
| 2453 | if (!tag) |
| 2454 | return; /* parser will look for errors */ |
| 2455 | |
| 2456 | /* attributes */ |
| 2457 | if (attrib_in[0]) { |
| 2458 | attrib = PyDict_New(); |
| 2459 | if (!attrib) |
| 2460 | return; |
| 2461 | while (attrib_in[0] && attrib_in[1]) { |
| 2462 | PyObject* key = makeuniversal(self, attrib_in[0]); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2463 | PyObject* value = PyUnicode_DecodeUTF8(attrib_in[1], strlen(attrib_in[1]), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2464 | if (!key || !value) { |
| 2465 | Py_XDECREF(value); |
| 2466 | Py_XDECREF(key); |
| 2467 | Py_DECREF(attrib); |
| 2468 | return; |
| 2469 | } |
| 2470 | ok = PyDict_SetItem(attrib, key, value); |
| 2471 | Py_DECREF(value); |
| 2472 | Py_DECREF(key); |
| 2473 | if (ok < 0) { |
| 2474 | Py_DECREF(attrib); |
| 2475 | return; |
| 2476 | } |
| 2477 | attrib_in += 2; |
| 2478 | } |
| 2479 | } else { |
| 2480 | Py_INCREF(Py_None); |
| 2481 | attrib = Py_None; |
| 2482 | } |
| 2483 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2484 | /* If we get None, pass an empty dictionary on */ |
| 2485 | if (attrib == Py_None) { |
| 2486 | Py_DECREF(attrib); |
| 2487 | attrib = PyDict_New(); |
| 2488 | if (!attrib) |
| 2489 | return; |
| 2490 | } |
| 2491 | |
| 2492 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2493 | /* shortcut */ |
| 2494 | res = treebuilder_handle_start((TreeBuilderObject*) self->target, |
| 2495 | tag, attrib); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2496 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2497 | else if (self->handle_start) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2498 | res = PyObject_CallFunction(self->handle_start, "OO", tag, attrib); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2499 | } else |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2500 | res = NULL; |
| 2501 | |
| 2502 | Py_DECREF(tag); |
| 2503 | Py_DECREF(attrib); |
| 2504 | |
| 2505 | Py_XDECREF(res); |
| 2506 | } |
| 2507 | |
| 2508 | static void |
| 2509 | expat_data_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2510 | int data_len) |
| 2511 | { |
| 2512 | PyObject* data; |
| 2513 | PyObject* res; |
| 2514 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2515 | data = PyUnicode_DecodeUTF8(data_in, data_len, "strict"); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2516 | if (!data) |
| 2517 | return; /* parser will look for errors */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2518 | |
| 2519 | if (TreeBuilder_CheckExact(self->target)) |
| 2520 | /* shortcut */ |
| 2521 | res = treebuilder_handle_data((TreeBuilderObject*) self->target, data); |
| 2522 | else if (self->handle_data) |
| 2523 | res = PyObject_CallFunction(self->handle_data, "O", data); |
| 2524 | else |
| 2525 | res = NULL; |
| 2526 | |
| 2527 | Py_DECREF(data); |
| 2528 | |
| 2529 | Py_XDECREF(res); |
| 2530 | } |
| 2531 | |
| 2532 | static void |
| 2533 | expat_end_handler(XMLParserObject* self, const XML_Char* tag_in) |
| 2534 | { |
| 2535 | PyObject* tag; |
| 2536 | PyObject* res = NULL; |
| 2537 | |
| 2538 | if (TreeBuilder_CheckExact(self->target)) |
| 2539 | /* shortcut */ |
| 2540 | /* the standard tree builder doesn't look at the end tag */ |
| 2541 | res = treebuilder_handle_end( |
| 2542 | (TreeBuilderObject*) self->target, Py_None |
| 2543 | ); |
| 2544 | else if (self->handle_end) { |
| 2545 | tag = makeuniversal(self, tag_in); |
| 2546 | if (tag) { |
| 2547 | res = PyObject_CallFunction(self->handle_end, "O", tag); |
| 2548 | Py_DECREF(tag); |
| 2549 | } |
| 2550 | } |
| 2551 | |
| 2552 | Py_XDECREF(res); |
| 2553 | } |
| 2554 | |
| 2555 | static void |
| 2556 | expat_start_ns_handler(XMLParserObject* self, const XML_Char* prefix, |
| 2557 | const XML_Char *uri) |
| 2558 | { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2559 | PyObject* sprefix = NULL; |
| 2560 | PyObject* suri = NULL; |
| 2561 | |
| 2562 | suri = PyUnicode_DecodeUTF8(uri, strlen(uri), "strict"); |
| 2563 | if (!suri) |
| 2564 | return; |
| 2565 | |
| 2566 | if (prefix) |
| 2567 | sprefix = PyUnicode_DecodeUTF8(prefix, strlen(prefix), "strict"); |
| 2568 | else |
| 2569 | sprefix = PyUnicode_FromString(""); |
| 2570 | if (!sprefix) { |
| 2571 | Py_DECREF(suri); |
| 2572 | return; |
| 2573 | } |
| 2574 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2575 | treebuilder_handle_namespace( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2576 | (TreeBuilderObject*) self->target, 1, sprefix, suri |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2577 | ); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2578 | |
| 2579 | Py_DECREF(sprefix); |
| 2580 | Py_DECREF(suri); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2581 | } |
| 2582 | |
| 2583 | static void |
| 2584 | expat_end_ns_handler(XMLParserObject* self, const XML_Char* prefix_in) |
| 2585 | { |
| 2586 | treebuilder_handle_namespace( |
| 2587 | (TreeBuilderObject*) self->target, 0, NULL, NULL |
| 2588 | ); |
| 2589 | } |
| 2590 | |
| 2591 | static void |
| 2592 | expat_comment_handler(XMLParserObject* self, const XML_Char* comment_in) |
| 2593 | { |
| 2594 | PyObject* comment; |
| 2595 | PyObject* res; |
| 2596 | |
| 2597 | if (self->handle_comment) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2598 | comment = PyUnicode_DecodeUTF8(comment_in, strlen(comment_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2599 | if (comment) { |
| 2600 | res = PyObject_CallFunction(self->handle_comment, "O", comment); |
| 2601 | Py_XDECREF(res); |
| 2602 | Py_DECREF(comment); |
| 2603 | } |
| 2604 | } |
| 2605 | } |
| 2606 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2607 | static void |
| 2608 | expat_start_doctype_handler(XMLParserObject *self, |
| 2609 | const XML_Char *doctype_name, |
| 2610 | const XML_Char *sysid, |
| 2611 | const XML_Char *pubid, |
| 2612 | int has_internal_subset) |
| 2613 | { |
| 2614 | PyObject *self_pyobj = (PyObject *)self; |
| 2615 | PyObject *doctype_name_obj, *sysid_obj, *pubid_obj; |
| 2616 | PyObject *parser_doctype = NULL; |
| 2617 | PyObject *res = NULL; |
| 2618 | |
| 2619 | doctype_name_obj = makeuniversal(self, doctype_name); |
| 2620 | if (!doctype_name_obj) |
| 2621 | return; |
| 2622 | |
| 2623 | if (sysid) { |
| 2624 | sysid_obj = makeuniversal(self, sysid); |
| 2625 | if (!sysid_obj) { |
| 2626 | Py_DECREF(doctype_name_obj); |
| 2627 | return; |
| 2628 | } |
| 2629 | } else { |
| 2630 | Py_INCREF(Py_None); |
| 2631 | sysid_obj = Py_None; |
| 2632 | } |
| 2633 | |
| 2634 | if (pubid) { |
| 2635 | pubid_obj = makeuniversal(self, pubid); |
| 2636 | if (!pubid_obj) { |
| 2637 | Py_DECREF(doctype_name_obj); |
| 2638 | Py_DECREF(sysid_obj); |
| 2639 | return; |
| 2640 | } |
| 2641 | } else { |
| 2642 | Py_INCREF(Py_None); |
| 2643 | pubid_obj = Py_None; |
| 2644 | } |
| 2645 | |
| 2646 | /* If the target has a handler for doctype, call it. */ |
| 2647 | if (self->handle_doctype) { |
| 2648 | res = PyObject_CallFunction(self->handle_doctype, "OOO", |
| 2649 | doctype_name_obj, pubid_obj, sysid_obj); |
| 2650 | Py_CLEAR(res); |
| 2651 | } |
| 2652 | |
| 2653 | /* Now see if the parser itself has a doctype method. If yes and it's |
| 2654 | * a subclass, call it but warn about deprecation. If it's not a subclass |
| 2655 | * (i.e. vanilla XMLParser), do nothing. |
| 2656 | */ |
| 2657 | parser_doctype = PyObject_GetAttrString(self_pyobj, "doctype"); |
| 2658 | if (parser_doctype) { |
| 2659 | if (!XMLParser_CheckExact(self_pyobj)) { |
| 2660 | if (PyErr_WarnEx(PyExc_DeprecationWarning, |
| 2661 | "This method of XMLParser is deprecated. Define" |
| 2662 | " doctype() method on the TreeBuilder target.", |
| 2663 | 1) < 0) { |
| 2664 | goto clear; |
| 2665 | } |
| 2666 | res = PyObject_CallFunction(parser_doctype, "OOO", |
| 2667 | doctype_name_obj, pubid_obj, sysid_obj); |
| 2668 | Py_CLEAR(res); |
| 2669 | } |
| 2670 | } |
| 2671 | |
| 2672 | clear: |
| 2673 | Py_XDECREF(parser_doctype); |
| 2674 | Py_DECREF(doctype_name_obj); |
| 2675 | Py_DECREF(pubid_obj); |
| 2676 | Py_DECREF(sysid_obj); |
| 2677 | } |
| 2678 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2679 | static void |
| 2680 | expat_pi_handler(XMLParserObject* self, const XML_Char* target_in, |
| 2681 | const XML_Char* data_in) |
| 2682 | { |
| 2683 | PyObject* target; |
| 2684 | PyObject* data; |
| 2685 | PyObject* res; |
| 2686 | |
| 2687 | if (self->handle_pi) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2688 | target = PyUnicode_DecodeUTF8(target_in, strlen(target_in), "strict"); |
| 2689 | data = PyUnicode_DecodeUTF8(data_in, strlen(data_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2690 | if (target && data) { |
| 2691 | res = PyObject_CallFunction(self->handle_pi, "OO", target, data); |
| 2692 | Py_XDECREF(res); |
| 2693 | Py_DECREF(data); |
| 2694 | Py_DECREF(target); |
| 2695 | } else { |
| 2696 | Py_XDECREF(data); |
| 2697 | Py_XDECREF(target); |
| 2698 | } |
| 2699 | } |
| 2700 | } |
| 2701 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2702 | static int |
| 2703 | expat_unknown_encoding_handler(XMLParserObject *self, const XML_Char *name, |
| 2704 | XML_Encoding *info) |
| 2705 | { |
| 2706 | PyObject* u; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2707 | unsigned char s[256]; |
| 2708 | int i; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2709 | void *data; |
| 2710 | unsigned int kind; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2711 | |
| 2712 | memset(info, 0, sizeof(XML_Encoding)); |
| 2713 | |
| 2714 | for (i = 0; i < 256; i++) |
| 2715 | s[i] = i; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2716 | |
Fredrik Lundh | c338999 | 2005-12-25 11:40:19 +0000 | [diff] [blame] | 2717 | u = PyUnicode_Decode((char*) s, 256, name, "replace"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2718 | if (!u) |
| 2719 | return XML_STATUS_ERROR; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2720 | if (PyUnicode_READY(u)) |
| 2721 | return XML_STATUS_ERROR; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2722 | |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2723 | if (PyUnicode_GET_LENGTH(u) != 256) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2724 | Py_DECREF(u); |
| 2725 | return XML_STATUS_ERROR; |
| 2726 | } |
| 2727 | |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2728 | kind = PyUnicode_KIND(u); |
| 2729 | data = PyUnicode_DATA(u); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2730 | for (i = 0; i < 256; i++) { |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 2731 | Py_UCS4 ch = PyUnicode_READ(kind, data, i); |
| 2732 | if (ch != Py_UNICODE_REPLACEMENT_CHARACTER) |
| 2733 | info->map[i] = ch; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2734 | else |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2735 | info->map[i] = -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2736 | } |
| 2737 | |
| 2738 | Py_DECREF(u); |
| 2739 | |
| 2740 | return XML_STATUS_OK; |
| 2741 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2742 | |
| 2743 | /* -------------------------------------------------------------------- */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2744 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2745 | static PyObject * |
| 2746 | xmlparser_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2747 | { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2748 | XMLParserObject *self = (XMLParserObject *)type->tp_alloc(type, 0); |
| 2749 | if (self) { |
| 2750 | self->parser = NULL; |
| 2751 | self->target = self->entity = self->names = NULL; |
| 2752 | self->handle_start = self->handle_data = self->handle_end = NULL; |
| 2753 | self->handle_comment = self->handle_pi = self->handle_close = NULL; |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2754 | self->handle_doctype = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2755 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2756 | return (PyObject *)self; |
| 2757 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2758 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2759 | static int |
| 2760 | xmlparser_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 2761 | { |
| 2762 | XMLParserObject *self_xp = (XMLParserObject *)self; |
| 2763 | PyObject *target = NULL, *html = NULL; |
| 2764 | char *encoding = NULL; |
| 2765 | static char *kwlist[] = {"html", "target", "encoding"}; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2766 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2767 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|OOz:XMLParser", kwlist, |
| 2768 | &html, &target, &encoding)) { |
| 2769 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2770 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2771 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2772 | self_xp->entity = PyDict_New(); |
| 2773 | if (!self_xp->entity) |
| 2774 | return -1; |
| 2775 | |
| 2776 | self_xp->names = PyDict_New(); |
| 2777 | if (!self_xp->names) { |
| 2778 | Py_XDECREF(self_xp->entity); |
| 2779 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2780 | } |
| 2781 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2782 | self_xp->parser = EXPAT(ParserCreate_MM)(encoding, &ExpatMemoryHandler, "}"); |
| 2783 | if (!self_xp->parser) { |
| 2784 | Py_XDECREF(self_xp->entity); |
| 2785 | Py_XDECREF(self_xp->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2786 | PyErr_NoMemory(); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2787 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2788 | } |
| 2789 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2790 | if (target) { |
| 2791 | Py_INCREF(target); |
| 2792 | } else { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2793 | target = treebuilder_new(&TreeBuilder_Type, NULL, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2794 | if (!target) { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2795 | Py_XDECREF(self_xp->entity); |
| 2796 | Py_XDECREF(self_xp->names); |
| 2797 | EXPAT(ParserFree)(self_xp->parser); |
| 2798 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2799 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2800 | } |
| 2801 | self_xp->target = target; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2802 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2803 | self_xp->handle_start = PyObject_GetAttrString(target, "start"); |
| 2804 | self_xp->handle_data = PyObject_GetAttrString(target, "data"); |
| 2805 | self_xp->handle_end = PyObject_GetAttrString(target, "end"); |
| 2806 | self_xp->handle_comment = PyObject_GetAttrString(target, "comment"); |
| 2807 | self_xp->handle_pi = PyObject_GetAttrString(target, "pi"); |
| 2808 | self_xp->handle_close = PyObject_GetAttrString(target, "close"); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2809 | self_xp->handle_doctype = PyObject_GetAttrString(target, "doctype"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2810 | |
| 2811 | PyErr_Clear(); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2812 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2813 | /* configure parser */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2814 | EXPAT(SetUserData)(self_xp->parser, self_xp); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2815 | EXPAT(SetElementHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2816 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2817 | (XML_StartElementHandler) expat_start_handler, |
| 2818 | (XML_EndElementHandler) expat_end_handler |
| 2819 | ); |
| 2820 | EXPAT(SetDefaultHandlerExpand)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2821 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2822 | (XML_DefaultHandler) expat_default_handler |
| 2823 | ); |
| 2824 | EXPAT(SetCharacterDataHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2825 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2826 | (XML_CharacterDataHandler) expat_data_handler |
| 2827 | ); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2828 | if (self_xp->handle_comment) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2829 | EXPAT(SetCommentHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2830 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2831 | (XML_CommentHandler) expat_comment_handler |
| 2832 | ); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2833 | if (self_xp->handle_pi) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2834 | EXPAT(SetProcessingInstructionHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2835 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2836 | (XML_ProcessingInstructionHandler) expat_pi_handler |
| 2837 | ); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2838 | EXPAT(SetStartDoctypeDeclHandler)( |
| 2839 | self_xp->parser, |
| 2840 | (XML_StartDoctypeDeclHandler) expat_start_doctype_handler |
| 2841 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2842 | EXPAT(SetUnknownEncodingHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2843 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2844 | (XML_UnknownEncodingHandler) expat_unknown_encoding_handler, NULL |
| 2845 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2846 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2847 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2848 | } |
| 2849 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2850 | static int |
| 2851 | xmlparser_gc_traverse(XMLParserObject *self, visitproc visit, void *arg) |
| 2852 | { |
| 2853 | Py_VISIT(self->handle_close); |
| 2854 | Py_VISIT(self->handle_pi); |
| 2855 | Py_VISIT(self->handle_comment); |
| 2856 | Py_VISIT(self->handle_end); |
| 2857 | Py_VISIT(self->handle_data); |
| 2858 | Py_VISIT(self->handle_start); |
| 2859 | |
| 2860 | Py_VISIT(self->target); |
| 2861 | Py_VISIT(self->entity); |
| 2862 | Py_VISIT(self->names); |
| 2863 | |
| 2864 | return 0; |
| 2865 | } |
| 2866 | |
| 2867 | static int |
| 2868 | xmlparser_gc_clear(XMLParserObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2869 | { |
| 2870 | EXPAT(ParserFree)(self->parser); |
| 2871 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2872 | Py_XDECREF(self->handle_close); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2873 | Py_XDECREF(self->handle_pi); |
| 2874 | Py_XDECREF(self->handle_comment); |
| 2875 | Py_XDECREF(self->handle_end); |
| 2876 | Py_XDECREF(self->handle_data); |
| 2877 | Py_XDECREF(self->handle_start); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2878 | Py_XDECREF(self->handle_doctype); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2879 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2880 | Py_XDECREF(self->target); |
| 2881 | Py_XDECREF(self->entity); |
| 2882 | Py_XDECREF(self->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2883 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2884 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2885 | } |
| 2886 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2887 | static void |
| 2888 | xmlparser_dealloc(XMLParserObject* self) |
| 2889 | { |
| 2890 | PyObject_GC_UnTrack(self); |
| 2891 | xmlparser_gc_clear(self); |
| 2892 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 2893 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2894 | |
| 2895 | LOCAL(PyObject*) |
| 2896 | expat_parse(XMLParserObject* self, char* data, int data_len, int final) |
| 2897 | { |
| 2898 | int ok; |
| 2899 | |
| 2900 | ok = EXPAT(Parse)(self->parser, data, data_len, final); |
| 2901 | |
| 2902 | if (PyErr_Occurred()) |
| 2903 | return NULL; |
| 2904 | |
| 2905 | if (!ok) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2906 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2907 | EXPAT(GetErrorCode)(self->parser), |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2908 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2909 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 2910 | NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2911 | ); |
| 2912 | return NULL; |
| 2913 | } |
| 2914 | |
| 2915 | Py_RETURN_NONE; |
| 2916 | } |
| 2917 | |
| 2918 | static PyObject* |
| 2919 | xmlparser_close(XMLParserObject* self, PyObject* args) |
| 2920 | { |
| 2921 | /* end feeding data to parser */ |
| 2922 | |
| 2923 | PyObject* res; |
| 2924 | if (!PyArg_ParseTuple(args, ":close")) |
| 2925 | return NULL; |
| 2926 | |
| 2927 | res = expat_parse(self, "", 0, 1); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2928 | if (!res) |
| 2929 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2930 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2931 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2932 | Py_DECREF(res); |
| 2933 | return treebuilder_done((TreeBuilderObject*) self->target); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2934 | } if (self->handle_close) { |
| 2935 | Py_DECREF(res); |
| 2936 | return PyObject_CallFunction(self->handle_close, ""); |
| 2937 | } else |
| 2938 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2939 | } |
| 2940 | |
| 2941 | static PyObject* |
| 2942 | xmlparser_feed(XMLParserObject* self, PyObject* args) |
| 2943 | { |
| 2944 | /* feed data to parser */ |
| 2945 | |
| 2946 | char* data; |
| 2947 | int data_len; |
| 2948 | if (!PyArg_ParseTuple(args, "s#:feed", &data, &data_len)) |
| 2949 | return NULL; |
| 2950 | |
| 2951 | return expat_parse(self, data, data_len, 0); |
| 2952 | } |
| 2953 | |
| 2954 | static PyObject* |
| 2955 | xmlparser_parse(XMLParserObject* self, PyObject* args) |
| 2956 | { |
| 2957 | /* (internal) parse until end of input stream */ |
| 2958 | |
| 2959 | PyObject* reader; |
| 2960 | PyObject* buffer; |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2961 | PyObject* temp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2962 | PyObject* res; |
| 2963 | |
| 2964 | PyObject* fileobj; |
| 2965 | if (!PyArg_ParseTuple(args, "O:_parse", &fileobj)) |
| 2966 | return NULL; |
| 2967 | |
| 2968 | reader = PyObject_GetAttrString(fileobj, "read"); |
| 2969 | if (!reader) |
| 2970 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2971 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2972 | /* read from open file object */ |
| 2973 | for (;;) { |
| 2974 | |
| 2975 | buffer = PyObject_CallFunction(reader, "i", 64*1024); |
| 2976 | |
| 2977 | if (!buffer) { |
| 2978 | /* read failed (e.g. due to KeyboardInterrupt) */ |
| 2979 | Py_DECREF(reader); |
| 2980 | return NULL; |
| 2981 | } |
| 2982 | |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2983 | if (PyUnicode_CheckExact(buffer)) { |
| 2984 | /* A unicode object is encoded into bytes using UTF-8 */ |
| 2985 | if (PyUnicode_GET_SIZE(buffer) == 0) { |
| 2986 | Py_DECREF(buffer); |
| 2987 | break; |
| 2988 | } |
| 2989 | temp = PyUnicode_AsEncodedString(buffer, "utf-8", "surrogatepass"); |
| 2990 | if (!temp) { |
| 2991 | /* Propagate exception from PyUnicode_AsEncodedString */ |
| 2992 | Py_DECREF(buffer); |
| 2993 | Py_DECREF(reader); |
| 2994 | return NULL; |
| 2995 | } |
| 2996 | |
| 2997 | /* Here we no longer need the original buffer since it contains |
| 2998 | * unicode. Make it point to the encoded bytes object. |
| 2999 | */ |
| 3000 | Py_DECREF(buffer); |
| 3001 | buffer = temp; |
| 3002 | } |
| 3003 | else if (!PyBytes_CheckExact(buffer) || PyBytes_GET_SIZE(buffer) == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3004 | Py_DECREF(buffer); |
| 3005 | break; |
| 3006 | } |
| 3007 | |
| 3008 | res = expat_parse( |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 3009 | self, PyBytes_AS_STRING(buffer), PyBytes_GET_SIZE(buffer), 0 |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3010 | ); |
| 3011 | |
| 3012 | Py_DECREF(buffer); |
| 3013 | |
| 3014 | if (!res) { |
| 3015 | Py_DECREF(reader); |
| 3016 | return NULL; |
| 3017 | } |
| 3018 | Py_DECREF(res); |
| 3019 | |
| 3020 | } |
| 3021 | |
| 3022 | Py_DECREF(reader); |
| 3023 | |
| 3024 | res = expat_parse(self, "", 0, 1); |
| 3025 | |
| 3026 | if (res && TreeBuilder_CheckExact(self->target)) { |
| 3027 | Py_DECREF(res); |
| 3028 | return treebuilder_done((TreeBuilderObject*) self->target); |
| 3029 | } |
| 3030 | |
| 3031 | return res; |
| 3032 | } |
| 3033 | |
| 3034 | static PyObject* |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3035 | xmlparser_doctype(XMLParserObject *self, PyObject *args) |
| 3036 | { |
| 3037 | Py_RETURN_NONE; |
| 3038 | } |
| 3039 | |
| 3040 | static PyObject* |
| 3041 | xmlparser_setevents(XMLParserObject *self, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3042 | { |
| 3043 | /* activate element event reporting */ |
| 3044 | |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 3045 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3046 | TreeBuilderObject* target; |
| 3047 | |
| 3048 | PyObject* events; /* event collector */ |
| 3049 | PyObject* event_set = Py_None; |
| 3050 | if (!PyArg_ParseTuple(args, "O!|O:_setevents", &PyList_Type, &events, |
| 3051 | &event_set)) |
| 3052 | return NULL; |
| 3053 | |
| 3054 | if (!TreeBuilder_CheckExact(self->target)) { |
| 3055 | PyErr_SetString( |
| 3056 | PyExc_TypeError, |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 3057 | "event handling only supported for ElementTree.TreeBuilder " |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3058 | "targets" |
| 3059 | ); |
| 3060 | return NULL; |
| 3061 | } |
| 3062 | |
| 3063 | target = (TreeBuilderObject*) self->target; |
| 3064 | |
| 3065 | Py_INCREF(events); |
| 3066 | Py_XDECREF(target->events); |
| 3067 | target->events = events; |
| 3068 | |
| 3069 | /* clear out existing events */ |
| 3070 | Py_XDECREF(target->start_event_obj); target->start_event_obj = NULL; |
| 3071 | Py_XDECREF(target->end_event_obj); target->end_event_obj = NULL; |
| 3072 | Py_XDECREF(target->start_ns_event_obj); target->start_ns_event_obj = NULL; |
| 3073 | Py_XDECREF(target->end_ns_event_obj); target->end_ns_event_obj = NULL; |
| 3074 | |
| 3075 | if (event_set == Py_None) { |
| 3076 | /* default is "end" only */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3077 | target->end_event_obj = PyUnicode_FromString("end"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3078 | Py_RETURN_NONE; |
| 3079 | } |
| 3080 | |
| 3081 | if (!PyTuple_Check(event_set)) /* FIXME: handle arbitrary sequences */ |
| 3082 | goto error; |
| 3083 | |
| 3084 | for (i = 0; i < PyTuple_GET_SIZE(event_set); i++) { |
| 3085 | PyObject* item = PyTuple_GET_ITEM(event_set, i); |
| 3086 | char* event; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3087 | if (PyUnicode_Check(item)) { |
| 3088 | event = _PyUnicode_AsString(item); |
Victor Stinner | 0477bf3 | 2010-03-22 12:11:44 +0000 | [diff] [blame] | 3089 | if (event == NULL) |
| 3090 | goto error; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3091 | } else if (PyBytes_Check(item)) |
| 3092 | event = PyBytes_AS_STRING(item); |
| 3093 | else { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3094 | goto error; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3095 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3096 | if (strcmp(event, "start") == 0) { |
| 3097 | Py_INCREF(item); |
| 3098 | target->start_event_obj = item; |
| 3099 | } else if (strcmp(event, "end") == 0) { |
| 3100 | Py_INCREF(item); |
| 3101 | Py_XDECREF(target->end_event_obj); |
| 3102 | target->end_event_obj = item; |
| 3103 | } else if (strcmp(event, "start-ns") == 0) { |
| 3104 | Py_INCREF(item); |
| 3105 | Py_XDECREF(target->start_ns_event_obj); |
| 3106 | target->start_ns_event_obj = item; |
| 3107 | EXPAT(SetNamespaceDeclHandler)( |
| 3108 | self->parser, |
| 3109 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3110 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3111 | ); |
| 3112 | } else if (strcmp(event, "end-ns") == 0) { |
| 3113 | Py_INCREF(item); |
| 3114 | Py_XDECREF(target->end_ns_event_obj); |
| 3115 | target->end_ns_event_obj = item; |
| 3116 | EXPAT(SetNamespaceDeclHandler)( |
| 3117 | self->parser, |
| 3118 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3119 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3120 | ); |
| 3121 | } else { |
| 3122 | PyErr_Format( |
| 3123 | PyExc_ValueError, |
| 3124 | "unknown event '%s'", event |
| 3125 | ); |
| 3126 | return NULL; |
| 3127 | } |
| 3128 | } |
| 3129 | |
| 3130 | Py_RETURN_NONE; |
| 3131 | |
| 3132 | error: |
| 3133 | PyErr_SetString( |
| 3134 | PyExc_TypeError, |
| 3135 | "invalid event tuple" |
| 3136 | ); |
| 3137 | return NULL; |
| 3138 | } |
| 3139 | |
| 3140 | static PyMethodDef xmlparser_methods[] = { |
| 3141 | {"feed", (PyCFunction) xmlparser_feed, METH_VARARGS}, |
| 3142 | {"close", (PyCFunction) xmlparser_close, METH_VARARGS}, |
| 3143 | {"_parse", (PyCFunction) xmlparser_parse, METH_VARARGS}, |
| 3144 | {"_setevents", (PyCFunction) xmlparser_setevents, METH_VARARGS}, |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3145 | {"doctype", (PyCFunction) xmlparser_doctype, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3146 | {NULL, NULL} |
| 3147 | }; |
| 3148 | |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3149 | static PyObject* |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3150 | xmlparser_getattro(XMLParserObject* self, PyObject* nameobj) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3151 | { |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3152 | if (PyUnicode_Check(nameobj)) { |
| 3153 | PyObject* res; |
| 3154 | if (PyUnicode_CompareWithASCIIString(nameobj, "entity") == 0) |
| 3155 | res = self->entity; |
| 3156 | else if (PyUnicode_CompareWithASCIIString(nameobj, "target") == 0) |
| 3157 | res = self->target; |
| 3158 | else if (PyUnicode_CompareWithASCIIString(nameobj, "version") == 0) { |
| 3159 | return PyUnicode_FromFormat( |
| 3160 | "Expat %d.%d.%d", XML_MAJOR_VERSION, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3161 | XML_MINOR_VERSION, XML_MICRO_VERSION); |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3162 | } |
| 3163 | else |
| 3164 | goto generic; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3165 | |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3166 | Py_INCREF(res); |
| 3167 | return res; |
| 3168 | } |
| 3169 | generic: |
| 3170 | return PyObject_GenericGetAttr((PyObject*) self, nameobj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3171 | } |
| 3172 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 3173 | static PyTypeObject XMLParser_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 3174 | PyVarObject_HEAD_INIT(NULL, 0) |
| 3175 | "XMLParser", sizeof(XMLParserObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3176 | /* methods */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3177 | (destructor)xmlparser_dealloc, /* tp_dealloc */ |
| 3178 | 0, /* tp_print */ |
| 3179 | 0, /* tp_getattr */ |
| 3180 | 0, /* tp_setattr */ |
| 3181 | 0, /* tp_reserved */ |
| 3182 | 0, /* tp_repr */ |
| 3183 | 0, /* tp_as_number */ |
| 3184 | 0, /* tp_as_sequence */ |
| 3185 | 0, /* tp_as_mapping */ |
| 3186 | 0, /* tp_hash */ |
| 3187 | 0, /* tp_call */ |
| 3188 | 0, /* tp_str */ |
| 3189 | (getattrofunc)xmlparser_getattro, /* tp_getattro */ |
| 3190 | 0, /* tp_setattro */ |
| 3191 | 0, /* tp_as_buffer */ |
| 3192 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 3193 | /* tp_flags */ |
| 3194 | 0, /* tp_doc */ |
| 3195 | (traverseproc)xmlparser_gc_traverse, /* tp_traverse */ |
| 3196 | (inquiry)xmlparser_gc_clear, /* tp_clear */ |
| 3197 | 0, /* tp_richcompare */ |
| 3198 | 0, /* tp_weaklistoffset */ |
| 3199 | 0, /* tp_iter */ |
| 3200 | 0, /* tp_iternext */ |
| 3201 | xmlparser_methods, /* tp_methods */ |
| 3202 | 0, /* tp_members */ |
| 3203 | 0, /* tp_getset */ |
| 3204 | 0, /* tp_base */ |
| 3205 | 0, /* tp_dict */ |
| 3206 | 0, /* tp_descr_get */ |
| 3207 | 0, /* tp_descr_set */ |
| 3208 | 0, /* tp_dictoffset */ |
| 3209 | (initproc)xmlparser_init, /* tp_init */ |
| 3210 | PyType_GenericAlloc, /* tp_alloc */ |
| 3211 | xmlparser_new, /* tp_new */ |
| 3212 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3213 | }; |
| 3214 | |
| 3215 | #endif |
| 3216 | |
| 3217 | /* ==================================================================== */ |
| 3218 | /* python module interface */ |
| 3219 | |
| 3220 | static PyMethodDef _functions[] = { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3221 | {"SubElement", (PyCFunction) subelement, METH_VARARGS|METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3222 | {NULL, NULL} |
| 3223 | }; |
| 3224 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3225 | |
| 3226 | static struct PyModuleDef _elementtreemodule = { |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3227 | PyModuleDef_HEAD_INIT, |
| 3228 | "_elementtree", |
| 3229 | NULL, |
| 3230 | -1, |
| 3231 | _functions, |
| 3232 | NULL, |
| 3233 | NULL, |
| 3234 | NULL, |
| 3235 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3236 | }; |
| 3237 | |
Neal Norwitz | f6657e6 | 2006-12-28 04:47:50 +0000 | [diff] [blame] | 3238 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3239 | PyInit__elementtree(void) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3240 | { |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 3241 | PyObject *m, *g, *temp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3242 | char* bootstrap; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3243 | |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3244 | /* Initialize object types */ |
| 3245 | if (PyType_Ready(&TreeBuilder_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3246 | return NULL; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3247 | if (PyType_Ready(&Element_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3248 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3249 | #if defined(USE_EXPAT) |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3250 | if (PyType_Ready(&XMLParser_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3251 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3252 | #endif |
| 3253 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3254 | m = PyModule_Create(&_elementtreemodule); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 3255 | if (!m) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3256 | return NULL; |
| 3257 | |
| 3258 | /* The code below requires that the module gets already added |
| 3259 | to sys.modules. */ |
| 3260 | PyDict_SetItemString(PyImport_GetModuleDict(), |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3261 | _elementtreemodule.m_name, |
| 3262 | m); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3263 | |
| 3264 | /* python glue code */ |
| 3265 | |
| 3266 | g = PyDict_New(); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 3267 | if (!g) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3268 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3269 | |
| 3270 | PyDict_SetItemString(g, "__builtins__", PyEval_GetBuiltins()); |
| 3271 | |
| 3272 | bootstrap = ( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3273 | "def iter(node, tag=None):\n" /* helper */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3274 | " if tag == '*':\n" |
| 3275 | " tag = None\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3276 | " if tag is None or node.tag == tag:\n" |
| 3277 | " yield node\n" |
| 3278 | " for node in node:\n" |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3279 | " for node in iter(node, tag):\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3280 | " yield node\n" |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3281 | |
| 3282 | "def itertext(node):\n" /* helper */ |
| 3283 | " if node.text:\n" |
| 3284 | " yield node.text\n" |
| 3285 | " for e in node:\n" |
| 3286 | " for s in e.itertext():\n" |
| 3287 | " yield s\n" |
| 3288 | " if e.tail:\n" |
| 3289 | " yield e.tail\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3290 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3291 | ); |
| 3292 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3293 | if (!PyRun_String(bootstrap, Py_file_input, g, NULL)) |
| 3294 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3295 | |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 3296 | if (!(temp = PyImport_ImportModule("copy"))) |
| 3297 | return NULL; |
| 3298 | elementtree_deepcopy_obj = PyObject_GetAttrString(temp, "deepcopy"); |
| 3299 | Py_XDECREF(temp); |
| 3300 | |
| 3301 | if (!(elementpath_obj = PyImport_ImportModule("xml.etree.ElementPath"))) |
| 3302 | return NULL; |
| 3303 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3304 | elementtree_iter_obj = PyDict_GetItemString(g, "iter"); |
| 3305 | elementtree_itertext_obj = PyDict_GetItemString(g, "itertext"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3306 | |
Eli Bendersky | 20d4174 | 2012-06-01 09:48:37 +0300 | [diff] [blame] | 3307 | /* link against pyexpat */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3308 | expat_capi = PyCapsule_Import(PyExpat_CAPSULE_NAME, 0); |
| 3309 | if (expat_capi) { |
| 3310 | /* check that it's usable */ |
| 3311 | if (strcmp(expat_capi->magic, PyExpat_CAPI_MAGIC) != 0 || |
| 3312 | expat_capi->size < sizeof(struct PyExpat_CAPI) || |
| 3313 | expat_capi->MAJOR_VERSION != XML_MAJOR_VERSION || |
| 3314 | expat_capi->MINOR_VERSION != XML_MINOR_VERSION || |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3315 | expat_capi->MICRO_VERSION != XML_MICRO_VERSION) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3316 | expat_capi = NULL; |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3317 | } |
| 3318 | } |
| 3319 | if (!expat_capi) { |
| 3320 | PyErr_SetString( |
| 3321 | PyExc_RuntimeError, "cannot load dispatch table from pyexpat" |
| 3322 | ); |
| 3323 | return NULL; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3324 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3325 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3326 | elementtree_parseerror_obj = PyErr_NewException( |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 3327 | "xml.etree.ElementTree.ParseError", PyExc_SyntaxError, NULL |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3328 | ); |
| 3329 | Py_INCREF(elementtree_parseerror_obj); |
| 3330 | PyModule_AddObject(m, "ParseError", elementtree_parseerror_obj); |
| 3331 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 3332 | Py_INCREF((PyObject *)&Element_Type); |
| 3333 | PyModule_AddObject(m, "Element", (PyObject *)&Element_Type); |
| 3334 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 3335 | Py_INCREF((PyObject *)&TreeBuilder_Type); |
| 3336 | PyModule_AddObject(m, "TreeBuilder", (PyObject *)&TreeBuilder_Type); |
| 3337 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3338 | #if defined(USE_EXPAT) |
| 3339 | Py_INCREF((PyObject *)&XMLParser_Type); |
| 3340 | PyModule_AddObject(m, "XMLParser", (PyObject *)&XMLParser_Type); |
| 3341 | #endif |
| 3342 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3343 | return m; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3344 | } |