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))) |
Antoine Pitrou | ca8aa4a | 2012-09-20 20:56:47 +0200 | [diff] [blame] | 101 | #define JOIN_OBJ(p) ((PyObject*) ((Py_uintptr_t) (p) & ~(Py_uintptr_t)1)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 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; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 106 | static PyObject* elementpath_obj; |
| 107 | |
| 108 | /* helpers */ |
| 109 | |
| 110 | LOCAL(PyObject*) |
| 111 | deepcopy(PyObject* object, PyObject* memo) |
| 112 | { |
| 113 | /* do a deep copy of the given object */ |
| 114 | |
| 115 | PyObject* args; |
| 116 | PyObject* result; |
| 117 | |
| 118 | if (!elementtree_deepcopy_obj) { |
| 119 | PyErr_SetString( |
| 120 | PyExc_RuntimeError, |
| 121 | "deepcopy helper not found" |
| 122 | ); |
| 123 | return NULL; |
| 124 | } |
| 125 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 126 | args = PyTuple_Pack(2, object, memo); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 127 | if (!args) |
| 128 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 129 | result = PyObject_CallObject(elementtree_deepcopy_obj, args); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 130 | Py_DECREF(args); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 131 | return result; |
| 132 | } |
| 133 | |
| 134 | LOCAL(PyObject*) |
| 135 | list_join(PyObject* list) |
| 136 | { |
| 137 | /* join list elements (destroying the list in the process) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 138 | PyObject* joiner; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 139 | PyObject* result; |
| 140 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 141 | joiner = PyUnicode_FromStringAndSize("", 0); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 142 | if (!joiner) |
| 143 | return NULL; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 144 | result = PyUnicode_Join(joiner, list); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 145 | Py_DECREF(joiner); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 146 | if (result) |
| 147 | Py_DECREF(list); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 148 | return result; |
| 149 | } |
| 150 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 151 | /* Is the given object an empty dictionary? |
| 152 | */ |
| 153 | static int |
| 154 | is_empty_dict(PyObject *obj) |
| 155 | { |
| 156 | return PyDict_CheckExact(obj) && PyDict_Size(obj) == 0; |
| 157 | } |
| 158 | |
| 159 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 160 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 161 | /* the Element type */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 162 | |
| 163 | typedef struct { |
| 164 | |
| 165 | /* attributes (a dictionary object), or None if no attributes */ |
| 166 | PyObject* attrib; |
| 167 | |
| 168 | /* child elements */ |
| 169 | int length; /* actual number of items */ |
| 170 | int allocated; /* allocated items */ |
| 171 | |
| 172 | /* this either points to _children or to a malloced buffer */ |
| 173 | PyObject* *children; |
| 174 | |
| 175 | PyObject* _children[STATIC_CHILDREN]; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 176 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 177 | } ElementObjectExtra; |
| 178 | |
| 179 | typedef struct { |
| 180 | PyObject_HEAD |
| 181 | |
| 182 | /* element tag (a string). */ |
| 183 | PyObject* tag; |
| 184 | |
| 185 | /* text before first child. note that this is a tagged pointer; |
| 186 | use JOIN_OBJ to get the object pointer. the join flag is used |
| 187 | to distinguish lists created by the tree builder from lists |
| 188 | assigned to the attribute by application code; the former |
| 189 | should be joined before being returned to the user, the latter |
| 190 | should be left intact. */ |
| 191 | PyObject* text; |
| 192 | |
| 193 | /* text after this element, in parent. note that this is a tagged |
| 194 | pointer; use JOIN_OBJ to get the object pointer. */ |
| 195 | PyObject* tail; |
| 196 | |
| 197 | ElementObjectExtra* extra; |
| 198 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 199 | PyObject *weakreflist; /* For tp_weaklistoffset */ |
| 200 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 201 | } ElementObject; |
| 202 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 203 | static PyTypeObject Element_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 204 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 205 | #define Element_CheckExact(op) (Py_TYPE(op) == &Element_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 206 | |
| 207 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 208 | /* Element constructors and destructor */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 209 | |
| 210 | LOCAL(int) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 211 | create_extra(ElementObject* self, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 212 | { |
| 213 | self->extra = PyObject_Malloc(sizeof(ElementObjectExtra)); |
| 214 | if (!self->extra) |
| 215 | return -1; |
| 216 | |
| 217 | if (!attrib) |
| 218 | attrib = Py_None; |
| 219 | |
| 220 | Py_INCREF(attrib); |
| 221 | self->extra->attrib = attrib; |
| 222 | |
| 223 | self->extra->length = 0; |
| 224 | self->extra->allocated = STATIC_CHILDREN; |
| 225 | self->extra->children = self->extra->_children; |
| 226 | |
| 227 | return 0; |
| 228 | } |
| 229 | |
| 230 | LOCAL(void) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 231 | dealloc_extra(ElementObject* self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 232 | { |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 233 | ElementObjectExtra *myextra; |
| 234 | int i; |
| 235 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 236 | if (!self->extra) |
| 237 | return; |
| 238 | |
| 239 | /* Avoid DECREFs calling into this code again (cycles, etc.) |
| 240 | */ |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 241 | myextra = self->extra; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 242 | self->extra = NULL; |
| 243 | |
| 244 | Py_DECREF(myextra->attrib); |
| 245 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 246 | for (i = 0; i < myextra->length; i++) |
| 247 | Py_DECREF(myextra->children[i]); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 248 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 249 | if (myextra->children != myextra->_children) |
| 250 | PyObject_Free(myextra->children); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 251 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 252 | PyObject_Free(myextra); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 253 | } |
| 254 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 255 | /* Convenience internal function to create new Element objects with the given |
| 256 | * tag and attributes. |
| 257 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 258 | LOCAL(PyObject*) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 259 | create_new_element(PyObject* tag, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 260 | { |
| 261 | ElementObject* self; |
| 262 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 263 | self = PyObject_GC_New(ElementObject, &Element_Type); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 264 | if (self == NULL) |
| 265 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 266 | self->extra = NULL; |
| 267 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 268 | if (attrib != Py_None && !is_empty_dict(attrib)) { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 269 | if (create_extra(self, attrib) < 0) { |
Thomas Wouters | 477c8d5 | 2006-05-27 19:21:47 +0000 | [diff] [blame] | 270 | PyObject_Del(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 271 | return NULL; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 272 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 273 | } |
| 274 | |
| 275 | Py_INCREF(tag); |
| 276 | self->tag = tag; |
| 277 | |
| 278 | Py_INCREF(Py_None); |
| 279 | self->text = Py_None; |
| 280 | |
| 281 | Py_INCREF(Py_None); |
| 282 | self->tail = Py_None; |
| 283 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 284 | self->weakreflist = NULL; |
| 285 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 286 | ALLOC(sizeof(ElementObject), "create element"); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 287 | PyObject_GC_Track(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 288 | return (PyObject*) self; |
| 289 | } |
| 290 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 291 | static PyObject * |
| 292 | element_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 293 | { |
| 294 | ElementObject *e = (ElementObject *)type->tp_alloc(type, 0); |
| 295 | if (e != NULL) { |
| 296 | Py_INCREF(Py_None); |
| 297 | e->tag = Py_None; |
| 298 | |
| 299 | Py_INCREF(Py_None); |
| 300 | e->text = Py_None; |
| 301 | |
| 302 | Py_INCREF(Py_None); |
| 303 | e->tail = Py_None; |
| 304 | |
| 305 | e->extra = NULL; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 306 | e->weakreflist = NULL; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 307 | } |
| 308 | return (PyObject *)e; |
| 309 | } |
| 310 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 311 | /* Helper function for extracting the attrib dictionary from a keywords dict. |
| 312 | * This is required by some constructors/functions in this module that can |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 313 | * either accept attrib as a keyword argument or all attributes splashed |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 314 | * directly into *kwds. |
Eli Bendersky | d4cb4b7 | 2013-04-22 05:25:25 -0700 | [diff] [blame] | 315 | * |
| 316 | * Return a dictionary with the content of kwds merged into the content of |
| 317 | * attrib. If there is no attrib keyword, return a copy of kwds. |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 318 | */ |
| 319 | static PyObject* |
| 320 | get_attrib_from_keywords(PyObject *kwds) |
| 321 | { |
| 322 | PyObject *attrib_str = PyUnicode_FromString("attrib"); |
| 323 | PyObject *attrib = PyDict_GetItem(kwds, attrib_str); |
| 324 | |
| 325 | if (attrib) { |
| 326 | /* If attrib was found in kwds, copy its value and remove it from |
| 327 | * kwds |
| 328 | */ |
| 329 | if (!PyDict_Check(attrib)) { |
| 330 | Py_DECREF(attrib_str); |
| 331 | PyErr_Format(PyExc_TypeError, "attrib must be dict, not %.100s", |
| 332 | Py_TYPE(attrib)->tp_name); |
| 333 | return NULL; |
| 334 | } |
| 335 | attrib = PyDict_Copy(attrib); |
| 336 | PyDict_DelItem(kwds, attrib_str); |
| 337 | } else { |
| 338 | attrib = PyDict_New(); |
| 339 | } |
| 340 | |
| 341 | Py_DECREF(attrib_str); |
| 342 | |
| 343 | if (attrib) |
| 344 | PyDict_Update(attrib, kwds); |
| 345 | return attrib; |
| 346 | } |
| 347 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 348 | static int |
| 349 | element_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 350 | { |
| 351 | PyObject *tag; |
| 352 | PyObject *tmp; |
| 353 | PyObject *attrib = NULL; |
| 354 | ElementObject *self_elem; |
| 355 | |
| 356 | if (!PyArg_ParseTuple(args, "O|O!:Element", &tag, &PyDict_Type, &attrib)) |
| 357 | return -1; |
| 358 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 359 | if (attrib) { |
| 360 | /* attrib passed as positional arg */ |
| 361 | attrib = PyDict_Copy(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 362 | if (!attrib) |
| 363 | return -1; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 364 | if (kwds) { |
| 365 | if (PyDict_Update(attrib, kwds) < 0) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 366 | Py_DECREF(attrib); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 367 | return -1; |
| 368 | } |
| 369 | } |
| 370 | } else if (kwds) { |
| 371 | /* have keywords args */ |
| 372 | attrib = get_attrib_from_keywords(kwds); |
| 373 | if (!attrib) |
| 374 | return -1; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 375 | } |
| 376 | |
| 377 | self_elem = (ElementObject *)self; |
| 378 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 379 | if (attrib != NULL && !is_empty_dict(attrib)) { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 380 | if (create_extra(self_elem, attrib) < 0) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 381 | Py_DECREF(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 382 | return -1; |
| 383 | } |
| 384 | } |
| 385 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 386 | /* We own a reference to attrib here and it's no longer needed. */ |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 387 | Py_XDECREF(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 388 | |
| 389 | /* Replace the objects already pointed to by tag, text and tail. */ |
| 390 | tmp = self_elem->tag; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 391 | Py_INCREF(tag); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 392 | self_elem->tag = tag; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 393 | Py_DECREF(tmp); |
| 394 | |
| 395 | tmp = self_elem->text; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 396 | Py_INCREF(Py_None); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 397 | self_elem->text = Py_None; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 398 | Py_DECREF(JOIN_OBJ(tmp)); |
| 399 | |
| 400 | tmp = self_elem->tail; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 401 | Py_INCREF(Py_None); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 402 | self_elem->tail = Py_None; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 403 | Py_DECREF(JOIN_OBJ(tmp)); |
| 404 | |
| 405 | return 0; |
| 406 | } |
| 407 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 408 | LOCAL(int) |
| 409 | element_resize(ElementObject* self, int extra) |
| 410 | { |
| 411 | int size; |
| 412 | PyObject* *children; |
| 413 | |
| 414 | /* make sure self->children can hold the given number of extra |
| 415 | elements. set an exception and return -1 if allocation failed */ |
| 416 | |
| 417 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 418 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 419 | |
| 420 | size = self->extra->length + extra; |
| 421 | |
| 422 | if (size > self->extra->allocated) { |
| 423 | /* use Python 2.4's list growth strategy */ |
| 424 | size = (size >> 3) + (size < 9 ? 3 : 6) + size; |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 425 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer "children" |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 426 | * which needs at least 4 bytes. |
| 427 | * 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] | 428 | * be safe. |
| 429 | */ |
| 430 | size = size ? size : 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 431 | if (self->extra->children != self->extra->_children) { |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 432 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 433 | * "children", which needs at least 4 bytes. Although it's a |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 434 | * false alarm always assume at least one child to be safe. |
| 435 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 436 | children = PyObject_Realloc(self->extra->children, |
| 437 | size * sizeof(PyObject*)); |
| 438 | if (!children) |
| 439 | goto nomemory; |
| 440 | } else { |
| 441 | children = PyObject_Malloc(size * sizeof(PyObject*)); |
| 442 | if (!children) |
| 443 | goto nomemory; |
| 444 | /* copy existing children from static area to malloc buffer */ |
| 445 | memcpy(children, self->extra->children, |
| 446 | self->extra->length * sizeof(PyObject*)); |
| 447 | } |
| 448 | self->extra->children = children; |
| 449 | self->extra->allocated = size; |
| 450 | } |
| 451 | |
| 452 | return 0; |
| 453 | |
| 454 | nomemory: |
| 455 | PyErr_NoMemory(); |
| 456 | return -1; |
| 457 | } |
| 458 | |
| 459 | LOCAL(int) |
| 460 | element_add_subelement(ElementObject* self, PyObject* element) |
| 461 | { |
| 462 | /* add a child element to a parent */ |
| 463 | |
| 464 | if (element_resize(self, 1) < 0) |
| 465 | return -1; |
| 466 | |
| 467 | Py_INCREF(element); |
| 468 | self->extra->children[self->extra->length] = element; |
| 469 | |
| 470 | self->extra->length++; |
| 471 | |
| 472 | return 0; |
| 473 | } |
| 474 | |
| 475 | LOCAL(PyObject*) |
| 476 | element_get_attrib(ElementObject* self) |
| 477 | { |
| 478 | /* return borrowed reference to attrib dictionary */ |
| 479 | /* note: this function assumes that the extra section exists */ |
| 480 | |
| 481 | PyObject* res = self->extra->attrib; |
| 482 | |
| 483 | if (res == Py_None) { |
| 484 | /* create missing dictionary */ |
| 485 | res = PyDict_New(); |
| 486 | if (!res) |
| 487 | return NULL; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 488 | Py_DECREF(Py_None); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 489 | self->extra->attrib = res; |
| 490 | } |
| 491 | |
| 492 | return res; |
| 493 | } |
| 494 | |
| 495 | LOCAL(PyObject*) |
| 496 | element_get_text(ElementObject* self) |
| 497 | { |
| 498 | /* return borrowed reference to text attribute */ |
| 499 | |
| 500 | PyObject* res = self->text; |
| 501 | |
| 502 | if (JOIN_GET(res)) { |
| 503 | res = JOIN_OBJ(res); |
| 504 | if (PyList_CheckExact(res)) { |
| 505 | res = list_join(res); |
| 506 | if (!res) |
| 507 | return NULL; |
| 508 | self->text = res; |
| 509 | } |
| 510 | } |
| 511 | |
| 512 | return res; |
| 513 | } |
| 514 | |
| 515 | LOCAL(PyObject*) |
| 516 | element_get_tail(ElementObject* self) |
| 517 | { |
| 518 | /* return borrowed reference to text attribute */ |
| 519 | |
| 520 | PyObject* res = self->tail; |
| 521 | |
| 522 | if (JOIN_GET(res)) { |
| 523 | res = JOIN_OBJ(res); |
| 524 | if (PyList_CheckExact(res)) { |
| 525 | res = list_join(res); |
| 526 | if (!res) |
| 527 | return NULL; |
| 528 | self->tail = res; |
| 529 | } |
| 530 | } |
| 531 | |
| 532 | return res; |
| 533 | } |
| 534 | |
| 535 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 536 | subelement(PyObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 537 | { |
| 538 | PyObject* elem; |
| 539 | |
| 540 | ElementObject* parent; |
| 541 | PyObject* tag; |
| 542 | PyObject* attrib = NULL; |
| 543 | if (!PyArg_ParseTuple(args, "O!O|O!:SubElement", |
| 544 | &Element_Type, &parent, &tag, |
| 545 | &PyDict_Type, &attrib)) |
| 546 | return NULL; |
| 547 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 548 | if (attrib) { |
| 549 | /* attrib passed as positional arg */ |
| 550 | attrib = PyDict_Copy(attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 551 | if (!attrib) |
| 552 | return NULL; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 553 | if (kwds) { |
| 554 | if (PyDict_Update(attrib, kwds) < 0) { |
| 555 | return NULL; |
| 556 | } |
| 557 | } |
| 558 | } else if (kwds) { |
| 559 | /* have keyword args */ |
| 560 | attrib = get_attrib_from_keywords(kwds); |
| 561 | if (!attrib) |
| 562 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 563 | } else { |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 564 | /* no attrib arg, no kwds, so no attribute */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 565 | Py_INCREF(Py_None); |
| 566 | attrib = Py_None; |
| 567 | } |
| 568 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 569 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 570 | |
| 571 | Py_DECREF(attrib); |
| 572 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 573 | if (element_add_subelement(parent, elem) < 0) { |
| 574 | Py_DECREF(elem); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 575 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 576 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 577 | |
| 578 | return elem; |
| 579 | } |
| 580 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 581 | static int |
| 582 | element_gc_traverse(ElementObject *self, visitproc visit, void *arg) |
| 583 | { |
| 584 | Py_VISIT(self->tag); |
| 585 | Py_VISIT(JOIN_OBJ(self->text)); |
| 586 | Py_VISIT(JOIN_OBJ(self->tail)); |
| 587 | |
| 588 | if (self->extra) { |
| 589 | int i; |
| 590 | Py_VISIT(self->extra->attrib); |
| 591 | |
| 592 | for (i = 0; i < self->extra->length; ++i) |
| 593 | Py_VISIT(self->extra->children[i]); |
| 594 | } |
| 595 | return 0; |
| 596 | } |
| 597 | |
| 598 | static int |
| 599 | element_gc_clear(ElementObject *self) |
| 600 | { |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 601 | Py_CLEAR(self->tag); |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 602 | |
| 603 | /* The following is like Py_CLEAR for self->text and self->tail, but |
| 604 | * written explicitily because the real pointers hide behind access |
| 605 | * macros. |
| 606 | */ |
| 607 | if (self->text) { |
| 608 | PyObject *tmp = JOIN_OBJ(self->text); |
| 609 | self->text = NULL; |
| 610 | Py_DECREF(tmp); |
| 611 | } |
| 612 | |
| 613 | if (self->tail) { |
| 614 | PyObject *tmp = JOIN_OBJ(self->tail); |
| 615 | self->tail = NULL; |
| 616 | Py_DECREF(tmp); |
| 617 | } |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 618 | |
| 619 | /* After dropping all references from extra, it's no longer valid anyway, |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 620 | * so fully deallocate it. |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 621 | */ |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 622 | dealloc_extra(self); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 623 | return 0; |
| 624 | } |
| 625 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 626 | static void |
| 627 | element_dealloc(ElementObject* self) |
| 628 | { |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 629 | PyObject_GC_UnTrack(self); |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 630 | |
| 631 | if (self->weakreflist != NULL) |
| 632 | PyObject_ClearWeakRefs((PyObject *) self); |
| 633 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 634 | /* element_gc_clear clears all references and deallocates extra |
| 635 | */ |
| 636 | element_gc_clear(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 637 | |
| 638 | RELEASE(sizeof(ElementObject), "destroy element"); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 639 | Py_TYPE(self)->tp_free((PyObject *)self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 640 | } |
| 641 | |
| 642 | /* -------------------------------------------------------------------- */ |
| 643 | /* methods (in alphabetical order) */ |
| 644 | |
| 645 | static PyObject* |
| 646 | element_append(ElementObject* self, PyObject* args) |
| 647 | { |
| 648 | PyObject* element; |
| 649 | if (!PyArg_ParseTuple(args, "O!:append", &Element_Type, &element)) |
| 650 | return NULL; |
| 651 | |
| 652 | if (element_add_subelement(self, element) < 0) |
| 653 | return NULL; |
| 654 | |
| 655 | Py_RETURN_NONE; |
| 656 | } |
| 657 | |
| 658 | static PyObject* |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 659 | element_clearmethod(ElementObject* self, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 660 | { |
| 661 | if (!PyArg_ParseTuple(args, ":clear")) |
| 662 | return NULL; |
| 663 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 664 | dealloc_extra(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 665 | |
| 666 | Py_INCREF(Py_None); |
| 667 | Py_DECREF(JOIN_OBJ(self->text)); |
| 668 | self->text = Py_None; |
| 669 | |
| 670 | Py_INCREF(Py_None); |
| 671 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 672 | self->tail = Py_None; |
| 673 | |
| 674 | Py_RETURN_NONE; |
| 675 | } |
| 676 | |
| 677 | static PyObject* |
| 678 | element_copy(ElementObject* self, PyObject* args) |
| 679 | { |
| 680 | int i; |
| 681 | ElementObject* element; |
| 682 | |
| 683 | if (!PyArg_ParseTuple(args, ":__copy__")) |
| 684 | return NULL; |
| 685 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 686 | element = (ElementObject*) create_new_element( |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 687 | self->tag, (self->extra) ? self->extra->attrib : Py_None |
| 688 | ); |
| 689 | if (!element) |
| 690 | return NULL; |
| 691 | |
| 692 | Py_DECREF(JOIN_OBJ(element->text)); |
| 693 | element->text = self->text; |
| 694 | Py_INCREF(JOIN_OBJ(element->text)); |
| 695 | |
| 696 | Py_DECREF(JOIN_OBJ(element->tail)); |
| 697 | element->tail = self->tail; |
| 698 | Py_INCREF(JOIN_OBJ(element->tail)); |
| 699 | |
| 700 | if (self->extra) { |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 701 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 702 | if (element_resize(element, self->extra->length) < 0) { |
| 703 | Py_DECREF(element); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 704 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 705 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 706 | |
| 707 | for (i = 0; i < self->extra->length; i++) { |
| 708 | Py_INCREF(self->extra->children[i]); |
| 709 | element->extra->children[i] = self->extra->children[i]; |
| 710 | } |
| 711 | |
| 712 | element->extra->length = self->extra->length; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 713 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 714 | } |
| 715 | |
| 716 | return (PyObject*) element; |
| 717 | } |
| 718 | |
| 719 | static PyObject* |
| 720 | element_deepcopy(ElementObject* self, PyObject* args) |
| 721 | { |
| 722 | int i; |
| 723 | ElementObject* element; |
| 724 | PyObject* tag; |
| 725 | PyObject* attrib; |
| 726 | PyObject* text; |
| 727 | PyObject* tail; |
| 728 | PyObject* id; |
| 729 | |
| 730 | PyObject* memo; |
| 731 | if (!PyArg_ParseTuple(args, "O:__deepcopy__", &memo)) |
| 732 | return NULL; |
| 733 | |
| 734 | tag = deepcopy(self->tag, memo); |
| 735 | if (!tag) |
| 736 | return NULL; |
| 737 | |
| 738 | if (self->extra) { |
| 739 | attrib = deepcopy(self->extra->attrib, memo); |
| 740 | if (!attrib) { |
| 741 | Py_DECREF(tag); |
| 742 | return NULL; |
| 743 | } |
| 744 | } else { |
| 745 | Py_INCREF(Py_None); |
| 746 | attrib = Py_None; |
| 747 | } |
| 748 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 749 | element = (ElementObject*) create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 750 | |
| 751 | Py_DECREF(tag); |
| 752 | Py_DECREF(attrib); |
| 753 | |
| 754 | if (!element) |
| 755 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 756 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 757 | text = deepcopy(JOIN_OBJ(self->text), memo); |
| 758 | if (!text) |
| 759 | goto error; |
| 760 | Py_DECREF(element->text); |
| 761 | element->text = JOIN_SET(text, JOIN_GET(self->text)); |
| 762 | |
| 763 | tail = deepcopy(JOIN_OBJ(self->tail), memo); |
| 764 | if (!tail) |
| 765 | goto error; |
| 766 | Py_DECREF(element->tail); |
| 767 | element->tail = JOIN_SET(tail, JOIN_GET(self->tail)); |
| 768 | |
| 769 | if (self->extra) { |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 770 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 771 | if (element_resize(element, self->extra->length) < 0) |
| 772 | goto error; |
| 773 | |
| 774 | for (i = 0; i < self->extra->length; i++) { |
| 775 | PyObject* child = deepcopy(self->extra->children[i], memo); |
| 776 | if (!child) { |
| 777 | element->extra->length = i; |
| 778 | goto error; |
| 779 | } |
| 780 | element->extra->children[i] = child; |
| 781 | } |
| 782 | |
| 783 | element->extra->length = self->extra->length; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 784 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 785 | } |
| 786 | |
| 787 | /* add object to memo dictionary (so deepcopy won't visit it again) */ |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 788 | id = PyLong_FromSsize_t((Py_uintptr_t) self); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 789 | if (!id) |
| 790 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 791 | |
| 792 | i = PyDict_SetItem(memo, id, (PyObject*) element); |
| 793 | |
| 794 | Py_DECREF(id); |
| 795 | |
| 796 | if (i < 0) |
| 797 | goto error; |
| 798 | |
| 799 | return (PyObject*) element; |
| 800 | |
| 801 | error: |
| 802 | Py_DECREF(element); |
| 803 | return NULL; |
| 804 | } |
| 805 | |
Martin v. Löwis | bce1666 | 2012-06-17 10:41:22 +0200 | [diff] [blame] | 806 | static PyObject* |
| 807 | element_sizeof(PyObject* _self, PyObject* args) |
| 808 | { |
| 809 | ElementObject *self = (ElementObject*)_self; |
| 810 | Py_ssize_t result = sizeof(ElementObject); |
| 811 | if (self->extra) { |
| 812 | result += sizeof(ElementObjectExtra); |
| 813 | if (self->extra->children != self->extra->_children) |
| 814 | result += sizeof(PyObject*) * self->extra->allocated; |
| 815 | } |
| 816 | return PyLong_FromSsize_t(result); |
| 817 | } |
| 818 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 819 | /* dict keys for getstate/setstate. */ |
| 820 | #define PICKLED_TAG "tag" |
| 821 | #define PICKLED_CHILDREN "_children" |
| 822 | #define PICKLED_ATTRIB "attrib" |
| 823 | #define PICKLED_TAIL "tail" |
| 824 | #define PICKLED_TEXT "text" |
| 825 | |
| 826 | /* __getstate__ returns a fabricated instance dict as in the pure-Python |
| 827 | * Element implementation, for interoperability/interchangeability. This |
| 828 | * makes the pure-Python implementation details an API, but (a) there aren't |
| 829 | * any unnecessary structures there; and (b) it buys compatibility with 3.2 |
| 830 | * pickles. See issue #16076. |
| 831 | */ |
| 832 | static PyObject * |
| 833 | element_getstate(ElementObject *self) |
| 834 | { |
| 835 | int i, noattrib; |
| 836 | PyObject *instancedict = NULL, *children; |
| 837 | |
| 838 | /* Build a list of children. */ |
| 839 | children = PyList_New(self->extra ? self->extra->length : 0); |
| 840 | if (!children) |
| 841 | return NULL; |
| 842 | for (i = 0; i < PyList_GET_SIZE(children); i++) { |
| 843 | PyObject *child = self->extra->children[i]; |
| 844 | Py_INCREF(child); |
| 845 | PyList_SET_ITEM(children, i, child); |
| 846 | } |
| 847 | |
| 848 | /* Construct the state object. */ |
| 849 | noattrib = (self->extra == NULL || self->extra->attrib == Py_None); |
| 850 | if (noattrib) |
| 851 | instancedict = Py_BuildValue("{sOsOs{}sOsO}", |
| 852 | PICKLED_TAG, self->tag, |
| 853 | PICKLED_CHILDREN, children, |
| 854 | PICKLED_ATTRIB, |
| 855 | PICKLED_TEXT, self->text, |
| 856 | PICKLED_TAIL, self->tail); |
| 857 | else |
| 858 | instancedict = Py_BuildValue("{sOsOsOsOsO}", |
| 859 | PICKLED_TAG, self->tag, |
| 860 | PICKLED_CHILDREN, children, |
| 861 | PICKLED_ATTRIB, self->extra->attrib, |
| 862 | PICKLED_TEXT, self->text, |
| 863 | PICKLED_TAIL, self->tail); |
Eli Bendersky | b8f6dc8 | 2013-01-12 05:20:16 -0800 | [diff] [blame] | 864 | if (instancedict) { |
| 865 | Py_DECREF(children); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 866 | return instancedict; |
Eli Bendersky | b8f6dc8 | 2013-01-12 05:20:16 -0800 | [diff] [blame] | 867 | } |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 868 | else { |
| 869 | for (i = 0; i < PyList_GET_SIZE(children); i++) |
| 870 | Py_DECREF(PyList_GET_ITEM(children, i)); |
| 871 | Py_DECREF(children); |
| 872 | |
| 873 | return NULL; |
| 874 | } |
| 875 | } |
| 876 | |
| 877 | static PyObject * |
| 878 | element_setstate_from_attributes(ElementObject *self, |
| 879 | PyObject *tag, |
| 880 | PyObject *attrib, |
| 881 | PyObject *text, |
| 882 | PyObject *tail, |
| 883 | PyObject *children) |
| 884 | { |
| 885 | Py_ssize_t i, nchildren; |
| 886 | |
| 887 | if (!tag) { |
| 888 | PyErr_SetString(PyExc_TypeError, "tag may not be NULL"); |
| 889 | return NULL; |
| 890 | } |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 891 | |
| 892 | Py_CLEAR(self->tag); |
| 893 | self->tag = tag; |
| 894 | Py_INCREF(self->tag); |
| 895 | |
| 896 | Py_CLEAR(self->text); |
Eli Bendersky | b8f6dc8 | 2013-01-12 05:20:16 -0800 | [diff] [blame] | 897 | self->text = text ? text : Py_None; |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 898 | Py_INCREF(self->text); |
| 899 | |
| 900 | Py_CLEAR(self->tail); |
Eli Bendersky | b8f6dc8 | 2013-01-12 05:20:16 -0800 | [diff] [blame] | 901 | self->tail = tail ? tail : Py_None; |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 902 | Py_INCREF(self->tail); |
| 903 | |
| 904 | /* Handle ATTRIB and CHILDREN. */ |
| 905 | if (!children && !attrib) |
| 906 | Py_RETURN_NONE; |
| 907 | |
| 908 | /* Compute 'nchildren'. */ |
| 909 | if (children) { |
| 910 | if (!PyList_Check(children)) { |
| 911 | PyErr_SetString(PyExc_TypeError, "'_children' is not a list"); |
| 912 | return NULL; |
| 913 | } |
| 914 | nchildren = PyList_Size(children); |
| 915 | } |
| 916 | else { |
| 917 | nchildren = 0; |
| 918 | } |
| 919 | |
| 920 | /* Allocate 'extra'. */ |
| 921 | if (element_resize(self, nchildren)) { |
| 922 | return NULL; |
| 923 | } |
| 924 | assert(self->extra && self->extra->allocated >= nchildren); |
| 925 | |
| 926 | /* Copy children */ |
| 927 | for (i = 0; i < nchildren; i++) { |
| 928 | self->extra->children[i] = PyList_GET_ITEM(children, i); |
| 929 | Py_INCREF(self->extra->children[i]); |
| 930 | } |
| 931 | |
| 932 | self->extra->length = nchildren; |
| 933 | self->extra->allocated = nchildren; |
| 934 | |
| 935 | /* Stash attrib. */ |
| 936 | if (attrib) { |
| 937 | Py_CLEAR(self->extra->attrib); |
| 938 | self->extra->attrib = attrib; |
| 939 | Py_INCREF(attrib); |
| 940 | } |
| 941 | |
| 942 | Py_RETURN_NONE; |
| 943 | } |
| 944 | |
| 945 | /* __setstate__ for Element instance from the Python implementation. |
| 946 | * 'state' should be the instance dict. |
| 947 | */ |
| 948 | static PyObject * |
| 949 | element_setstate_from_Python(ElementObject *self, PyObject *state) |
| 950 | { |
| 951 | static char *kwlist[] = {PICKLED_TAG, PICKLED_ATTRIB, PICKLED_TEXT, |
| 952 | PICKLED_TAIL, PICKLED_CHILDREN, 0}; |
| 953 | PyObject *args; |
| 954 | PyObject *tag, *attrib, *text, *tail, *children; |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 955 | PyObject *retval; |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 956 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 957 | tag = attrib = text = tail = children = NULL; |
| 958 | args = PyTuple_New(0); |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 959 | if (!args) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 960 | return NULL; |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 961 | |
| 962 | if (PyArg_ParseTupleAndKeywords(args, state, "|$OOOOO", kwlist, &tag, |
| 963 | &attrib, &text, &tail, &children)) |
| 964 | retval = element_setstate_from_attributes(self, tag, attrib, text, |
| 965 | tail, children); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 966 | else |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 967 | retval = NULL; |
| 968 | |
| 969 | Py_DECREF(args); |
| 970 | return retval; |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 971 | } |
| 972 | |
| 973 | static PyObject * |
| 974 | element_setstate(ElementObject *self, PyObject *state) |
| 975 | { |
| 976 | if (!PyDict_CheckExact(state)) { |
| 977 | PyErr_Format(PyExc_TypeError, |
| 978 | "Don't know how to unpickle \"%.200R\" as an Element", |
| 979 | state); |
| 980 | return NULL; |
| 981 | } |
| 982 | else |
| 983 | return element_setstate_from_Python(self, state); |
| 984 | } |
| 985 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 986 | LOCAL(int) |
| 987 | checkpath(PyObject* tag) |
| 988 | { |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 989 | Py_ssize_t i; |
| 990 | int check = 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 991 | |
| 992 | /* check if a tag contains an xpath character */ |
| 993 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 994 | #define PATHCHAR(ch) \ |
| 995 | (ch == '/' || ch == '*' || ch == '[' || ch == '@' || ch == '.') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 996 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 997 | if (PyUnicode_Check(tag)) { |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 998 | const Py_ssize_t len = PyUnicode_GET_LENGTH(tag); |
| 999 | void *data = PyUnicode_DATA(tag); |
| 1000 | unsigned int kind = PyUnicode_KIND(tag); |
| 1001 | for (i = 0; i < len; i++) { |
| 1002 | Py_UCS4 ch = PyUnicode_READ(kind, data, i); |
| 1003 | if (ch == '{') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1004 | check = 0; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 1005 | else if (ch == '}') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1006 | check = 1; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 1007 | else if (check && PATHCHAR(ch)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1008 | return 1; |
| 1009 | } |
| 1010 | return 0; |
| 1011 | } |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 1012 | if (PyBytes_Check(tag)) { |
| 1013 | char *p = PyBytes_AS_STRING(tag); |
| 1014 | for (i = 0; i < PyBytes_GET_SIZE(tag); i++) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1015 | if (p[i] == '{') |
| 1016 | check = 0; |
| 1017 | else if (p[i] == '}') |
| 1018 | check = 1; |
| 1019 | else if (check && PATHCHAR(p[i])) |
| 1020 | return 1; |
| 1021 | } |
| 1022 | return 0; |
| 1023 | } |
| 1024 | |
| 1025 | return 1; /* unknown type; might be path expression */ |
| 1026 | } |
| 1027 | |
| 1028 | static PyObject* |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1029 | element_extend(ElementObject* self, PyObject* args) |
| 1030 | { |
| 1031 | PyObject* seq; |
| 1032 | Py_ssize_t i, seqlen = 0; |
| 1033 | |
| 1034 | PyObject* seq_in; |
| 1035 | if (!PyArg_ParseTuple(args, "O:extend", &seq_in)) |
| 1036 | return NULL; |
| 1037 | |
| 1038 | seq = PySequence_Fast(seq_in, ""); |
| 1039 | if (!seq) { |
| 1040 | PyErr_Format( |
| 1041 | PyExc_TypeError, |
| 1042 | "expected sequence, not \"%.200s\"", Py_TYPE(seq_in)->tp_name |
| 1043 | ); |
| 1044 | return NULL; |
| 1045 | } |
| 1046 | |
| 1047 | seqlen = PySequence_Size(seq); |
| 1048 | for (i = 0; i < seqlen; i++) { |
| 1049 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 1050 | if (!PyObject_IsInstance(element, (PyObject *)&Element_Type)) { |
| 1051 | Py_DECREF(seq); |
| 1052 | PyErr_Format( |
| 1053 | PyExc_TypeError, |
| 1054 | "expected an Element, not \"%.200s\"", |
| 1055 | Py_TYPE(element)->tp_name); |
| 1056 | return NULL; |
| 1057 | } |
| 1058 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1059 | if (element_add_subelement(self, element) < 0) { |
| 1060 | Py_DECREF(seq); |
| 1061 | return NULL; |
| 1062 | } |
| 1063 | } |
| 1064 | |
| 1065 | Py_DECREF(seq); |
| 1066 | |
| 1067 | Py_RETURN_NONE; |
| 1068 | } |
| 1069 | |
| 1070 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1071 | element_find(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1072 | { |
| 1073 | int i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1074 | PyObject* tag; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1075 | PyObject* namespaces = Py_None; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1076 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1077 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1078 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:find", kwlist, |
| 1079 | &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1080 | return NULL; |
| 1081 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1082 | if (checkpath(tag) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1083 | _Py_IDENTIFIER(find); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1084 | return _PyObject_CallMethodId( |
| 1085 | elementpath_obj, &PyId_find, "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1086 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1087 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1088 | |
| 1089 | if (!self->extra) |
| 1090 | Py_RETURN_NONE; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1091 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1092 | for (i = 0; i < self->extra->length; i++) { |
| 1093 | PyObject* item = self->extra->children[i]; |
| 1094 | if (Element_CheckExact(item) && |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1095 | PyObject_RichCompareBool(((ElementObject*)item)->tag, tag, Py_EQ) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1096 | Py_INCREF(item); |
| 1097 | return item; |
| 1098 | } |
| 1099 | } |
| 1100 | |
| 1101 | Py_RETURN_NONE; |
| 1102 | } |
| 1103 | |
| 1104 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1105 | element_findtext(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1106 | { |
| 1107 | int i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1108 | PyObject* tag; |
| 1109 | PyObject* default_value = Py_None; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1110 | PyObject* namespaces = Py_None; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1111 | _Py_IDENTIFIER(findtext); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1112 | static char *kwlist[] = {"path", "default", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1113 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1114 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|OO:findtext", kwlist, |
| 1115 | &tag, &default_value, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1116 | return NULL; |
| 1117 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1118 | if (checkpath(tag) || namespaces != Py_None) |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1119 | return _PyObject_CallMethodId( |
| 1120 | elementpath_obj, &PyId_findtext, "OOOO", self, tag, default_value, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1121 | ); |
| 1122 | |
| 1123 | if (!self->extra) { |
| 1124 | Py_INCREF(default_value); |
| 1125 | return default_value; |
| 1126 | } |
| 1127 | |
| 1128 | for (i = 0; i < self->extra->length; i++) { |
| 1129 | ElementObject* item = (ElementObject*) self->extra->children[i]; |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1130 | if (Element_CheckExact(item) && (PyObject_RichCompareBool(item->tag, tag, Py_EQ) == 1)) { |
| 1131 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1132 | PyObject* text = element_get_text(item); |
| 1133 | if (text == Py_None) |
Eli Bendersky | 25771b3 | 2013-01-13 05:26:07 -0800 | [diff] [blame] | 1134 | return PyUnicode_New(0, 0); |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 1135 | Py_XINCREF(text); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1136 | return text; |
| 1137 | } |
| 1138 | } |
| 1139 | |
| 1140 | Py_INCREF(default_value); |
| 1141 | return default_value; |
| 1142 | } |
| 1143 | |
| 1144 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1145 | element_findall(ElementObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1146 | { |
| 1147 | int i; |
| 1148 | PyObject* out; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1149 | PyObject* tag; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1150 | PyObject* namespaces = Py_None; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1151 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1152 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1153 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:findall", kwlist, |
| 1154 | &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1155 | return NULL; |
| 1156 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1157 | if (checkpath(tag) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1158 | _Py_IDENTIFIER(findall); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1159 | return _PyObject_CallMethodId( |
| 1160 | elementpath_obj, &PyId_findall, "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1161 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1162 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1163 | |
| 1164 | out = PyList_New(0); |
| 1165 | if (!out) |
| 1166 | return NULL; |
| 1167 | |
| 1168 | if (!self->extra) |
| 1169 | return out; |
| 1170 | |
| 1171 | for (i = 0; i < self->extra->length; i++) { |
| 1172 | PyObject* item = self->extra->children[i]; |
| 1173 | if (Element_CheckExact(item) && |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1174 | PyObject_RichCompareBool(((ElementObject*)item)->tag, tag, Py_EQ) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1175 | if (PyList_Append(out, item) < 0) { |
| 1176 | Py_DECREF(out); |
| 1177 | return NULL; |
| 1178 | } |
| 1179 | } |
| 1180 | } |
| 1181 | |
| 1182 | return out; |
| 1183 | } |
| 1184 | |
| 1185 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1186 | element_iterfind(ElementObject *self, PyObject *args, PyObject *kwds) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1187 | { |
| 1188 | PyObject* tag; |
| 1189 | PyObject* namespaces = Py_None; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1190 | _Py_IDENTIFIER(iterfind); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1191 | static char *kwlist[] = {"path", "namespaces", 0}; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1192 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1193 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:iterfind", kwlist, |
| 1194 | &tag, &namespaces)) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1195 | return NULL; |
| 1196 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1197 | return _PyObject_CallMethodId( |
| 1198 | elementpath_obj, &PyId_iterfind, "OOO", self, tag, namespaces |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1199 | ); |
| 1200 | } |
| 1201 | |
| 1202 | static PyObject* |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1203 | element_get(ElementObject* self, PyObject* args, PyObject* kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1204 | { |
| 1205 | PyObject* value; |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1206 | static char* kwlist[] = {"key", "default", 0}; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1207 | |
| 1208 | PyObject* key; |
| 1209 | PyObject* default_value = Py_None; |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1210 | |
| 1211 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|O:get", kwlist, &key, |
| 1212 | &default_value)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1213 | return NULL; |
| 1214 | |
| 1215 | if (!self->extra || self->extra->attrib == Py_None) |
| 1216 | value = default_value; |
| 1217 | else { |
| 1218 | value = PyDict_GetItem(self->extra->attrib, key); |
| 1219 | if (!value) |
| 1220 | value = default_value; |
| 1221 | } |
| 1222 | |
| 1223 | Py_INCREF(value); |
| 1224 | return value; |
| 1225 | } |
| 1226 | |
| 1227 | static PyObject* |
| 1228 | element_getchildren(ElementObject* self, PyObject* args) |
| 1229 | { |
| 1230 | int i; |
| 1231 | PyObject* list; |
| 1232 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1233 | /* FIXME: report as deprecated? */ |
| 1234 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1235 | if (!PyArg_ParseTuple(args, ":getchildren")) |
| 1236 | return NULL; |
| 1237 | |
| 1238 | if (!self->extra) |
| 1239 | return PyList_New(0); |
| 1240 | |
| 1241 | list = PyList_New(self->extra->length); |
| 1242 | if (!list) |
| 1243 | return NULL; |
| 1244 | |
| 1245 | for (i = 0; i < self->extra->length; i++) { |
| 1246 | PyObject* item = self->extra->children[i]; |
| 1247 | Py_INCREF(item); |
| 1248 | PyList_SET_ITEM(list, i, item); |
| 1249 | } |
| 1250 | |
| 1251 | return list; |
| 1252 | } |
| 1253 | |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1254 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1255 | static PyObject * |
| 1256 | create_elementiter(ElementObject *self, PyObject *tag, int gettext); |
| 1257 | |
| 1258 | |
| 1259 | static PyObject * |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1260 | element_iter(ElementObject *self, PyObject *args, PyObject *kwds) |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1261 | { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1262 | PyObject* tag = Py_None; |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1263 | static char* kwlist[] = {"tag", 0}; |
| 1264 | |
| 1265 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O:iter", kwlist, &tag)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1266 | return NULL; |
| 1267 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1268 | return create_elementiter(self, tag, 0); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1269 | } |
| 1270 | |
| 1271 | |
| 1272 | static PyObject* |
| 1273 | element_itertext(ElementObject* self, PyObject* args) |
| 1274 | { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1275 | if (!PyArg_ParseTuple(args, ":itertext")) |
| 1276 | return NULL; |
| 1277 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1278 | return create_elementiter(self, Py_None, 1); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1279 | } |
| 1280 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1281 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1282 | static PyObject* |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1283 | element_getitem(PyObject* self_, Py_ssize_t index) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1284 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1285 | ElementObject* self = (ElementObject*) self_; |
| 1286 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1287 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1288 | PyErr_SetString( |
| 1289 | PyExc_IndexError, |
| 1290 | "child index out of range" |
| 1291 | ); |
| 1292 | return NULL; |
| 1293 | } |
| 1294 | |
| 1295 | Py_INCREF(self->extra->children[index]); |
| 1296 | return self->extra->children[index]; |
| 1297 | } |
| 1298 | |
| 1299 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1300 | element_insert(ElementObject* self, PyObject* args) |
| 1301 | { |
| 1302 | int i; |
| 1303 | |
| 1304 | int index; |
| 1305 | PyObject* element; |
| 1306 | if (!PyArg_ParseTuple(args, "iO!:insert", &index, |
| 1307 | &Element_Type, &element)) |
| 1308 | return NULL; |
| 1309 | |
| 1310 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1311 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1312 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1313 | if (index < 0) { |
| 1314 | index += self->extra->length; |
| 1315 | if (index < 0) |
| 1316 | index = 0; |
| 1317 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1318 | if (index > self->extra->length) |
| 1319 | index = self->extra->length; |
| 1320 | |
| 1321 | if (element_resize(self, 1) < 0) |
| 1322 | return NULL; |
| 1323 | |
| 1324 | for (i = self->extra->length; i > index; i--) |
| 1325 | self->extra->children[i] = self->extra->children[i-1]; |
| 1326 | |
| 1327 | Py_INCREF(element); |
| 1328 | self->extra->children[index] = element; |
| 1329 | |
| 1330 | self->extra->length++; |
| 1331 | |
| 1332 | Py_RETURN_NONE; |
| 1333 | } |
| 1334 | |
| 1335 | static PyObject* |
| 1336 | element_items(ElementObject* self, PyObject* args) |
| 1337 | { |
| 1338 | if (!PyArg_ParseTuple(args, ":items")) |
| 1339 | return NULL; |
| 1340 | |
| 1341 | if (!self->extra || self->extra->attrib == Py_None) |
| 1342 | return PyList_New(0); |
| 1343 | |
| 1344 | return PyDict_Items(self->extra->attrib); |
| 1345 | } |
| 1346 | |
| 1347 | static PyObject* |
| 1348 | element_keys(ElementObject* self, PyObject* args) |
| 1349 | { |
| 1350 | if (!PyArg_ParseTuple(args, ":keys")) |
| 1351 | return NULL; |
| 1352 | |
| 1353 | if (!self->extra || self->extra->attrib == Py_None) |
| 1354 | return PyList_New(0); |
| 1355 | |
| 1356 | return PyDict_Keys(self->extra->attrib); |
| 1357 | } |
| 1358 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1359 | static Py_ssize_t |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1360 | element_length(ElementObject* self) |
| 1361 | { |
| 1362 | if (!self->extra) |
| 1363 | return 0; |
| 1364 | |
| 1365 | return self->extra->length; |
| 1366 | } |
| 1367 | |
| 1368 | static PyObject* |
| 1369 | element_makeelement(PyObject* self, PyObject* args, PyObject* kw) |
| 1370 | { |
| 1371 | PyObject* elem; |
| 1372 | |
| 1373 | PyObject* tag; |
| 1374 | PyObject* attrib; |
| 1375 | if (!PyArg_ParseTuple(args, "OO:makeelement", &tag, &attrib)) |
| 1376 | return NULL; |
| 1377 | |
| 1378 | attrib = PyDict_Copy(attrib); |
| 1379 | if (!attrib) |
| 1380 | return NULL; |
| 1381 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1382 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1383 | |
| 1384 | Py_DECREF(attrib); |
| 1385 | |
| 1386 | return elem; |
| 1387 | } |
| 1388 | |
| 1389 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1390 | element_remove(ElementObject* self, PyObject* args) |
| 1391 | { |
| 1392 | int i; |
| 1393 | |
| 1394 | PyObject* element; |
| 1395 | if (!PyArg_ParseTuple(args, "O!:remove", &Element_Type, &element)) |
| 1396 | return NULL; |
| 1397 | |
| 1398 | if (!self->extra) { |
| 1399 | /* element has no children, so raise exception */ |
| 1400 | PyErr_SetString( |
| 1401 | PyExc_ValueError, |
| 1402 | "list.remove(x): x not in list" |
| 1403 | ); |
| 1404 | return NULL; |
| 1405 | } |
| 1406 | |
| 1407 | for (i = 0; i < self->extra->length; i++) { |
| 1408 | if (self->extra->children[i] == element) |
| 1409 | break; |
Mark Dickinson | 211c625 | 2009-02-01 10:28:51 +0000 | [diff] [blame] | 1410 | if (PyObject_RichCompareBool(self->extra->children[i], element, Py_EQ) == 1) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1411 | break; |
| 1412 | } |
| 1413 | |
| 1414 | if (i == self->extra->length) { |
| 1415 | /* element is not in children, so raise exception */ |
| 1416 | PyErr_SetString( |
| 1417 | PyExc_ValueError, |
| 1418 | "list.remove(x): x not in list" |
| 1419 | ); |
| 1420 | return NULL; |
| 1421 | } |
| 1422 | |
| 1423 | Py_DECREF(self->extra->children[i]); |
| 1424 | |
| 1425 | self->extra->length--; |
| 1426 | |
| 1427 | for (; i < self->extra->length; i++) |
| 1428 | self->extra->children[i] = self->extra->children[i+1]; |
| 1429 | |
| 1430 | Py_RETURN_NONE; |
| 1431 | } |
| 1432 | |
| 1433 | static PyObject* |
| 1434 | element_repr(ElementObject* self) |
| 1435 | { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1436 | if (self->tag) |
| 1437 | return PyUnicode_FromFormat("<Element %R at %p>", self->tag, self); |
| 1438 | else |
| 1439 | return PyUnicode_FromFormat("<Element at %p>", self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1440 | } |
| 1441 | |
| 1442 | static PyObject* |
| 1443 | element_set(ElementObject* self, PyObject* args) |
| 1444 | { |
| 1445 | PyObject* attrib; |
| 1446 | |
| 1447 | PyObject* key; |
| 1448 | PyObject* value; |
| 1449 | if (!PyArg_ParseTuple(args, "OO:set", &key, &value)) |
| 1450 | return NULL; |
| 1451 | |
| 1452 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1453 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1454 | |
| 1455 | attrib = element_get_attrib(self); |
| 1456 | if (!attrib) |
| 1457 | return NULL; |
| 1458 | |
| 1459 | if (PyDict_SetItem(attrib, key, value) < 0) |
| 1460 | return NULL; |
| 1461 | |
| 1462 | Py_RETURN_NONE; |
| 1463 | } |
| 1464 | |
| 1465 | static int |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1466 | element_setitem(PyObject* self_, Py_ssize_t index, PyObject* item) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1467 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1468 | ElementObject* self = (ElementObject*) self_; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1469 | int i; |
| 1470 | PyObject* old; |
| 1471 | |
| 1472 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1473 | PyErr_SetString( |
| 1474 | PyExc_IndexError, |
| 1475 | "child assignment index out of range"); |
| 1476 | return -1; |
| 1477 | } |
| 1478 | |
| 1479 | old = self->extra->children[index]; |
| 1480 | |
| 1481 | if (item) { |
| 1482 | Py_INCREF(item); |
| 1483 | self->extra->children[index] = item; |
| 1484 | } else { |
| 1485 | self->extra->length--; |
| 1486 | for (i = index; i < self->extra->length; i++) |
| 1487 | self->extra->children[i] = self->extra->children[i+1]; |
| 1488 | } |
| 1489 | |
| 1490 | Py_DECREF(old); |
| 1491 | |
| 1492 | return 0; |
| 1493 | } |
| 1494 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1495 | static PyObject* |
| 1496 | element_subscr(PyObject* self_, PyObject* item) |
| 1497 | { |
| 1498 | ElementObject* self = (ElementObject*) self_; |
| 1499 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1500 | if (PyIndex_Check(item)) { |
| 1501 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1502 | |
| 1503 | if (i == -1 && PyErr_Occurred()) { |
| 1504 | return NULL; |
| 1505 | } |
| 1506 | if (i < 0 && self->extra) |
| 1507 | i += self->extra->length; |
| 1508 | return element_getitem(self_, i); |
| 1509 | } |
| 1510 | else if (PySlice_Check(item)) { |
| 1511 | Py_ssize_t start, stop, step, slicelen, cur, i; |
| 1512 | PyObject* list; |
| 1513 | |
| 1514 | if (!self->extra) |
| 1515 | return PyList_New(0); |
| 1516 | |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 1517 | if (PySlice_GetIndicesEx(item, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1518 | self->extra->length, |
| 1519 | &start, &stop, &step, &slicelen) < 0) { |
| 1520 | return NULL; |
| 1521 | } |
| 1522 | |
| 1523 | if (slicelen <= 0) |
| 1524 | return PyList_New(0); |
| 1525 | else { |
| 1526 | list = PyList_New(slicelen); |
| 1527 | if (!list) |
| 1528 | return NULL; |
| 1529 | |
| 1530 | for (cur = start, i = 0; i < slicelen; |
| 1531 | cur += step, i++) { |
| 1532 | PyObject* item = self->extra->children[cur]; |
| 1533 | Py_INCREF(item); |
| 1534 | PyList_SET_ITEM(list, i, item); |
| 1535 | } |
| 1536 | |
| 1537 | return list; |
| 1538 | } |
| 1539 | } |
| 1540 | else { |
| 1541 | PyErr_SetString(PyExc_TypeError, |
| 1542 | "element indices must be integers"); |
| 1543 | return NULL; |
| 1544 | } |
| 1545 | } |
| 1546 | |
| 1547 | static int |
| 1548 | element_ass_subscr(PyObject* self_, PyObject* item, PyObject* value) |
| 1549 | { |
| 1550 | ElementObject* self = (ElementObject*) self_; |
| 1551 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1552 | if (PyIndex_Check(item)) { |
| 1553 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1554 | |
| 1555 | if (i == -1 && PyErr_Occurred()) { |
| 1556 | return -1; |
| 1557 | } |
| 1558 | if (i < 0 && self->extra) |
| 1559 | i += self->extra->length; |
| 1560 | return element_setitem(self_, i, value); |
| 1561 | } |
| 1562 | else if (PySlice_Check(item)) { |
| 1563 | Py_ssize_t start, stop, step, slicelen, newlen, cur, i; |
| 1564 | |
| 1565 | PyObject* recycle = NULL; |
| 1566 | PyObject* seq = NULL; |
| 1567 | |
| 1568 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1569 | create_extra(self, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1570 | |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 1571 | if (PySlice_GetIndicesEx(item, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1572 | self->extra->length, |
| 1573 | &start, &stop, &step, &slicelen) < 0) { |
| 1574 | return -1; |
| 1575 | } |
| 1576 | |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1577 | if (value == NULL) { |
| 1578 | /* Delete slice */ |
| 1579 | size_t cur; |
| 1580 | Py_ssize_t i; |
| 1581 | |
| 1582 | if (slicelen <= 0) |
| 1583 | return 0; |
| 1584 | |
| 1585 | /* Since we're deleting, the direction of the range doesn't matter, |
| 1586 | * so for simplicity make it always ascending. |
| 1587 | */ |
| 1588 | if (step < 0) { |
| 1589 | stop = start + 1; |
| 1590 | start = stop + step * (slicelen - 1) - 1; |
| 1591 | step = -step; |
| 1592 | } |
| 1593 | |
| 1594 | assert((size_t)slicelen <= PY_SIZE_MAX / sizeof(PyObject *)); |
| 1595 | |
| 1596 | /* recycle is a list that will contain all the children |
| 1597 | * scheduled for removal. |
| 1598 | */ |
| 1599 | if (!(recycle = PyList_New(slicelen))) { |
| 1600 | PyErr_NoMemory(); |
| 1601 | return -1; |
| 1602 | } |
| 1603 | |
| 1604 | /* This loop walks over all the children that have to be deleted, |
| 1605 | * with cur pointing at them. num_moved is the amount of children |
| 1606 | * until the next deleted child that have to be "shifted down" to |
| 1607 | * occupy the deleted's places. |
| 1608 | * Note that in the ith iteration, shifting is done i+i places down |
| 1609 | * because i children were already removed. |
| 1610 | */ |
| 1611 | for (cur = start, i = 0; cur < (size_t)stop; cur += step, ++i) { |
| 1612 | /* Compute how many children have to be moved, clipping at the |
| 1613 | * list end. |
| 1614 | */ |
| 1615 | Py_ssize_t num_moved = step - 1; |
| 1616 | if (cur + step >= (size_t)self->extra->length) { |
| 1617 | num_moved = self->extra->length - cur - 1; |
| 1618 | } |
| 1619 | |
| 1620 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1621 | |
| 1622 | memmove( |
| 1623 | self->extra->children + cur - i, |
| 1624 | self->extra->children + cur + 1, |
| 1625 | num_moved * sizeof(PyObject *)); |
| 1626 | } |
| 1627 | |
| 1628 | /* Leftover "tail" after the last removed child */ |
| 1629 | cur = start + (size_t)slicelen * step; |
| 1630 | if (cur < (size_t)self->extra->length) { |
| 1631 | memmove( |
| 1632 | self->extra->children + cur - slicelen, |
| 1633 | self->extra->children + cur, |
| 1634 | (self->extra->length - cur) * sizeof(PyObject *)); |
| 1635 | } |
| 1636 | |
| 1637 | self->extra->length -= slicelen; |
| 1638 | |
| 1639 | /* Discard the recycle list with all the deleted sub-elements */ |
| 1640 | Py_XDECREF(recycle); |
| 1641 | return 0; |
| 1642 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1643 | else { |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1644 | /* A new slice is actually being assigned */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1645 | seq = PySequence_Fast(value, ""); |
| 1646 | if (!seq) { |
| 1647 | PyErr_Format( |
| 1648 | PyExc_TypeError, |
| 1649 | "expected sequence, not \"%.200s\"", Py_TYPE(value)->tp_name |
| 1650 | ); |
| 1651 | return -1; |
| 1652 | } |
| 1653 | newlen = PySequence_Size(seq); |
| 1654 | } |
| 1655 | |
| 1656 | if (step != 1 && newlen != slicelen) |
| 1657 | { |
| 1658 | PyErr_Format(PyExc_ValueError, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1659 | "attempt to assign sequence of size %zd " |
| 1660 | "to extended slice of size %zd", |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1661 | newlen, slicelen |
| 1662 | ); |
| 1663 | return -1; |
| 1664 | } |
| 1665 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1666 | /* Resize before creating the recycle bin, to prevent refleaks. */ |
| 1667 | if (newlen > slicelen) { |
| 1668 | if (element_resize(self, newlen - slicelen) < 0) { |
| 1669 | if (seq) { |
| 1670 | Py_DECREF(seq); |
| 1671 | } |
| 1672 | return -1; |
| 1673 | } |
| 1674 | } |
| 1675 | |
| 1676 | if (slicelen > 0) { |
| 1677 | /* to avoid recursive calls to this method (via decref), move |
| 1678 | old items to the recycle bin here, and get rid of them when |
| 1679 | we're done modifying the element */ |
| 1680 | recycle = PyList_New(slicelen); |
| 1681 | if (!recycle) { |
| 1682 | if (seq) { |
| 1683 | Py_DECREF(seq); |
| 1684 | } |
| 1685 | return -1; |
| 1686 | } |
| 1687 | for (cur = start, i = 0; i < slicelen; |
| 1688 | cur += step, i++) |
| 1689 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1690 | } |
| 1691 | |
| 1692 | if (newlen < slicelen) { |
| 1693 | /* delete slice */ |
| 1694 | for (i = stop; i < self->extra->length; i++) |
| 1695 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1696 | } else if (newlen > slicelen) { |
| 1697 | /* insert slice */ |
| 1698 | for (i = self->extra->length-1; i >= stop; i--) |
| 1699 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1700 | } |
| 1701 | |
| 1702 | /* replace the slice */ |
| 1703 | for (cur = start, i = 0; i < newlen; |
| 1704 | cur += step, i++) { |
| 1705 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
| 1706 | Py_INCREF(element); |
| 1707 | self->extra->children[cur] = element; |
| 1708 | } |
| 1709 | |
| 1710 | self->extra->length += newlen - slicelen; |
| 1711 | |
| 1712 | if (seq) { |
| 1713 | Py_DECREF(seq); |
| 1714 | } |
| 1715 | |
| 1716 | /* discard the recycle bin, and everything in it */ |
| 1717 | Py_XDECREF(recycle); |
| 1718 | |
| 1719 | return 0; |
| 1720 | } |
| 1721 | else { |
| 1722 | PyErr_SetString(PyExc_TypeError, |
| 1723 | "element indices must be integers"); |
| 1724 | return -1; |
| 1725 | } |
| 1726 | } |
| 1727 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1728 | static PyMethodDef element_methods[] = { |
| 1729 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1730 | {"clear", (PyCFunction) element_clearmethod, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1731 | |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1732 | {"get", (PyCFunction) element_get, METH_VARARGS | METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1733 | {"set", (PyCFunction) element_set, METH_VARARGS}, |
| 1734 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1735 | {"find", (PyCFunction) element_find, METH_VARARGS | METH_KEYWORDS}, |
| 1736 | {"findtext", (PyCFunction) element_findtext, METH_VARARGS | METH_KEYWORDS}, |
| 1737 | {"findall", (PyCFunction) element_findall, METH_VARARGS | METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1738 | |
| 1739 | {"append", (PyCFunction) element_append, METH_VARARGS}, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1740 | {"extend", (PyCFunction) element_extend, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1741 | {"insert", (PyCFunction) element_insert, METH_VARARGS}, |
| 1742 | {"remove", (PyCFunction) element_remove, METH_VARARGS}, |
| 1743 | |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1744 | {"iter", (PyCFunction) element_iter, METH_VARARGS | METH_KEYWORDS}, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1745 | {"itertext", (PyCFunction) element_itertext, METH_VARARGS}, |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 1746 | {"iterfind", (PyCFunction) element_iterfind, METH_VARARGS | METH_KEYWORDS}, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1747 | |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1748 | {"getiterator", (PyCFunction) element_iter, METH_VARARGS | METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1749 | {"getchildren", (PyCFunction) element_getchildren, METH_VARARGS}, |
| 1750 | |
| 1751 | {"items", (PyCFunction) element_items, METH_VARARGS}, |
| 1752 | {"keys", (PyCFunction) element_keys, METH_VARARGS}, |
| 1753 | |
| 1754 | {"makeelement", (PyCFunction) element_makeelement, METH_VARARGS}, |
| 1755 | |
| 1756 | {"__copy__", (PyCFunction) element_copy, METH_VARARGS}, |
| 1757 | {"__deepcopy__", (PyCFunction) element_deepcopy, METH_VARARGS}, |
Martin v. Löwis | bce1666 | 2012-06-17 10:41:22 +0200 | [diff] [blame] | 1758 | {"__sizeof__", element_sizeof, METH_NOARGS}, |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1759 | {"__getstate__", (PyCFunction)element_getstate, METH_NOARGS}, |
| 1760 | {"__setstate__", (PyCFunction)element_setstate, METH_O}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1761 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1762 | {NULL, NULL} |
| 1763 | }; |
| 1764 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1765 | static PyObject* |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1766 | element_getattro(ElementObject* self, PyObject* nameobj) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1767 | { |
| 1768 | PyObject* res; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1769 | char *name = ""; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1770 | |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 1771 | if (PyUnicode_Check(nameobj)) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1772 | name = _PyUnicode_AsString(nameobj); |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1773 | |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 1774 | if (name == NULL) |
| 1775 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1776 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1777 | /* handle common attributes first */ |
| 1778 | if (strcmp(name, "tag") == 0) { |
| 1779 | res = self->tag; |
| 1780 | Py_INCREF(res); |
| 1781 | return res; |
| 1782 | } else if (strcmp(name, "text") == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1783 | res = element_get_text(self); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1784 | Py_INCREF(res); |
| 1785 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1786 | } |
| 1787 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1788 | /* methods */ |
| 1789 | res = PyObject_GenericGetAttr((PyObject*) self, nameobj); |
| 1790 | if (res) |
| 1791 | return res; |
| 1792 | |
| 1793 | /* less common attributes */ |
| 1794 | if (strcmp(name, "tail") == 0) { |
| 1795 | PyErr_Clear(); |
| 1796 | res = element_get_tail(self); |
| 1797 | } else if (strcmp(name, "attrib") == 0) { |
| 1798 | PyErr_Clear(); |
| 1799 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1800 | create_extra(self, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1801 | res = element_get_attrib(self); |
| 1802 | } |
| 1803 | |
| 1804 | if (!res) |
| 1805 | return NULL; |
| 1806 | |
| 1807 | Py_INCREF(res); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1808 | return res; |
| 1809 | } |
| 1810 | |
Eli Bendersky | ef9683b | 2013-05-18 07:52:34 -0700 | [diff] [blame] | 1811 | static int |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1812 | element_setattro(ElementObject* self, PyObject* nameobj, PyObject* value) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1813 | { |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1814 | char *name = ""; |
| 1815 | if (PyUnicode_Check(nameobj)) |
| 1816 | name = _PyUnicode_AsString(nameobj); |
| 1817 | |
Eli Bendersky | ef9683b | 2013-05-18 07:52:34 -0700 | [diff] [blame] | 1818 | if (name == NULL) { |
| 1819 | return -1; |
| 1820 | } else if (strcmp(name, "tag") == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1821 | Py_DECREF(self->tag); |
| 1822 | self->tag = value; |
| 1823 | Py_INCREF(self->tag); |
| 1824 | } else if (strcmp(name, "text") == 0) { |
| 1825 | Py_DECREF(JOIN_OBJ(self->text)); |
| 1826 | self->text = value; |
| 1827 | Py_INCREF(self->text); |
| 1828 | } else if (strcmp(name, "tail") == 0) { |
| 1829 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 1830 | self->tail = value; |
| 1831 | Py_INCREF(self->tail); |
| 1832 | } else if (strcmp(name, "attrib") == 0) { |
| 1833 | if (!self->extra) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1834 | create_extra(self, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1835 | Py_DECREF(self->extra->attrib); |
| 1836 | self->extra->attrib = value; |
| 1837 | Py_INCREF(self->extra->attrib); |
| 1838 | } else { |
Eli Bendersky | ef9683b | 2013-05-18 07:52:34 -0700 | [diff] [blame] | 1839 | PyErr_SetString(PyExc_AttributeError, |
Eli Bendersky | 6a55dc3 | 2013-05-19 16:59:59 -0700 | [diff] [blame] | 1840 | "Can't set arbitrary attributes on Element"); |
Eli Bendersky | ef9683b | 2013-05-18 07:52:34 -0700 | [diff] [blame] | 1841 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1842 | } |
| 1843 | |
Eli Bendersky | ef9683b | 2013-05-18 07:52:34 -0700 | [diff] [blame] | 1844 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1845 | } |
| 1846 | |
| 1847 | static PySequenceMethods element_as_sequence = { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1848 | (lenfunc) element_length, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1849 | 0, /* sq_concat */ |
| 1850 | 0, /* sq_repeat */ |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1851 | element_getitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1852 | 0, |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1853 | element_setitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1854 | 0, |
| 1855 | }; |
| 1856 | |
| 1857 | static PyMappingMethods element_as_mapping = { |
| 1858 | (lenfunc) element_length, |
| 1859 | (binaryfunc) element_subscr, |
| 1860 | (objobjargproc) element_ass_subscr, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1861 | }; |
| 1862 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 1863 | static PyTypeObject Element_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 1864 | PyVarObject_HEAD_INIT(NULL, 0) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1865 | "xml.etree.ElementTree.Element", sizeof(ElementObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1866 | /* methods */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1867 | (destructor)element_dealloc, /* tp_dealloc */ |
| 1868 | 0, /* tp_print */ |
| 1869 | 0, /* tp_getattr */ |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1870 | 0, /* tp_setattr */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1871 | 0, /* tp_reserved */ |
| 1872 | (reprfunc)element_repr, /* tp_repr */ |
| 1873 | 0, /* tp_as_number */ |
| 1874 | &element_as_sequence, /* tp_as_sequence */ |
| 1875 | &element_as_mapping, /* tp_as_mapping */ |
| 1876 | 0, /* tp_hash */ |
| 1877 | 0, /* tp_call */ |
| 1878 | 0, /* tp_str */ |
| 1879 | (getattrofunc)element_getattro, /* tp_getattro */ |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1880 | (setattrofunc)element_setattro, /* tp_setattro */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1881 | 0, /* tp_as_buffer */ |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1882 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 1883 | /* tp_flags */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1884 | 0, /* tp_doc */ |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1885 | (traverseproc)element_gc_traverse, /* tp_traverse */ |
| 1886 | (inquiry)element_gc_clear, /* tp_clear */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1887 | 0, /* tp_richcompare */ |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 1888 | offsetof(ElementObject, weakreflist), /* tp_weaklistoffset */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1889 | 0, /* tp_iter */ |
| 1890 | 0, /* tp_iternext */ |
| 1891 | element_methods, /* tp_methods */ |
| 1892 | 0, /* tp_members */ |
| 1893 | 0, /* tp_getset */ |
| 1894 | 0, /* tp_base */ |
| 1895 | 0, /* tp_dict */ |
| 1896 | 0, /* tp_descr_get */ |
| 1897 | 0, /* tp_descr_set */ |
| 1898 | 0, /* tp_dictoffset */ |
| 1899 | (initproc)element_init, /* tp_init */ |
| 1900 | PyType_GenericAlloc, /* tp_alloc */ |
| 1901 | element_new, /* tp_new */ |
| 1902 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1903 | }; |
| 1904 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1905 | /******************************* Element iterator ****************************/ |
| 1906 | |
| 1907 | /* ElementIterObject represents the iteration state over an XML element in |
| 1908 | * pre-order traversal. To keep track of which sub-element should be returned |
| 1909 | * next, a stack of parents is maintained. This is a standard stack-based |
| 1910 | * iterative pre-order traversal of a tree. |
| 1911 | * The stack is managed using a single-linked list starting at parent_stack. |
| 1912 | * Each stack node contains the saved parent to which we should return after |
| 1913 | * the current one is exhausted, and the next child to examine in that parent. |
| 1914 | */ |
| 1915 | typedef struct ParentLocator_t { |
| 1916 | ElementObject *parent; |
| 1917 | Py_ssize_t child_index; |
| 1918 | struct ParentLocator_t *next; |
| 1919 | } ParentLocator; |
| 1920 | |
| 1921 | typedef struct { |
| 1922 | PyObject_HEAD |
| 1923 | ParentLocator *parent_stack; |
| 1924 | ElementObject *root_element; |
| 1925 | PyObject *sought_tag; |
| 1926 | int root_done; |
| 1927 | int gettext; |
| 1928 | } ElementIterObject; |
| 1929 | |
| 1930 | |
| 1931 | static void |
| 1932 | elementiter_dealloc(ElementIterObject *it) |
| 1933 | { |
| 1934 | ParentLocator *p = it->parent_stack; |
| 1935 | while (p) { |
| 1936 | ParentLocator *temp = p; |
| 1937 | Py_XDECREF(p->parent); |
| 1938 | p = p->next; |
| 1939 | PyObject_Free(temp); |
| 1940 | } |
| 1941 | |
| 1942 | Py_XDECREF(it->sought_tag); |
| 1943 | Py_XDECREF(it->root_element); |
| 1944 | |
| 1945 | PyObject_GC_UnTrack(it); |
| 1946 | PyObject_GC_Del(it); |
| 1947 | } |
| 1948 | |
| 1949 | static int |
| 1950 | elementiter_traverse(ElementIterObject *it, visitproc visit, void *arg) |
| 1951 | { |
| 1952 | ParentLocator *p = it->parent_stack; |
| 1953 | while (p) { |
| 1954 | Py_VISIT(p->parent); |
| 1955 | p = p->next; |
| 1956 | } |
| 1957 | |
| 1958 | Py_VISIT(it->root_element); |
| 1959 | Py_VISIT(it->sought_tag); |
| 1960 | return 0; |
| 1961 | } |
| 1962 | |
| 1963 | /* Helper function for elementiter_next. Add a new parent to the parent stack. |
| 1964 | */ |
| 1965 | static ParentLocator * |
| 1966 | parent_stack_push_new(ParentLocator *stack, ElementObject *parent) |
| 1967 | { |
| 1968 | ParentLocator *new_node = PyObject_Malloc(sizeof(ParentLocator)); |
| 1969 | if (new_node) { |
| 1970 | new_node->parent = parent; |
| 1971 | Py_INCREF(parent); |
| 1972 | new_node->child_index = 0; |
| 1973 | new_node->next = stack; |
| 1974 | } |
| 1975 | return new_node; |
| 1976 | } |
| 1977 | |
| 1978 | static PyObject * |
| 1979 | elementiter_next(ElementIterObject *it) |
| 1980 | { |
| 1981 | /* Sub-element iterator. |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 1982 | * |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1983 | * A short note on gettext: this function serves both the iter() and |
| 1984 | * itertext() methods to avoid code duplication. However, there are a few |
| 1985 | * small differences in the way these iterations work. Namely: |
| 1986 | * - itertext() only yields text from nodes that have it, and continues |
| 1987 | * iterating when a node doesn't have text (so it doesn't return any |
| 1988 | * node like iter()) |
| 1989 | * - itertext() also has to handle tail, after finishing with all the |
| 1990 | * children of a node. |
| 1991 | */ |
Eli Bendersky | 113da64 | 2012-06-15 07:52:49 +0300 | [diff] [blame] | 1992 | ElementObject *cur_parent; |
| 1993 | Py_ssize_t child_index; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1994 | |
| 1995 | while (1) { |
| 1996 | /* Handle the case reached in the beginning and end of iteration, where |
| 1997 | * the parent stack is empty. The root_done flag gives us indication |
| 1998 | * whether we've just started iterating (so root_done is 0), in which |
| 1999 | * case the root is returned. If root_done is 1 and we're here, the |
| 2000 | * iterator is exhausted. |
| 2001 | */ |
| 2002 | if (!it->parent_stack->parent) { |
| 2003 | if (it->root_done) { |
| 2004 | PyErr_SetNone(PyExc_StopIteration); |
| 2005 | return NULL; |
| 2006 | } else { |
| 2007 | it->parent_stack = parent_stack_push_new(it->parent_stack, |
| 2008 | it->root_element); |
| 2009 | if (!it->parent_stack) { |
| 2010 | PyErr_NoMemory(); |
| 2011 | return NULL; |
| 2012 | } |
| 2013 | |
| 2014 | it->root_done = 1; |
| 2015 | if (it->sought_tag == Py_None || |
| 2016 | PyObject_RichCompareBool(it->root_element->tag, |
| 2017 | it->sought_tag, Py_EQ) == 1) { |
| 2018 | if (it->gettext) { |
Eli Bendersky | e6174ca | 2013-01-10 06:27:53 -0800 | [diff] [blame] | 2019 | PyObject *text = element_get_text(it->root_element); |
| 2020 | if (!text) |
| 2021 | return NULL; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2022 | if (PyObject_IsTrue(text)) { |
| 2023 | Py_INCREF(text); |
| 2024 | return text; |
| 2025 | } |
| 2026 | } else { |
| 2027 | Py_INCREF(it->root_element); |
| 2028 | return (PyObject *)it->root_element; |
| 2029 | } |
| 2030 | } |
| 2031 | } |
| 2032 | } |
| 2033 | |
| 2034 | /* See if there are children left to traverse in the current parent. If |
| 2035 | * yes, visit the next child. If not, pop the stack and try again. |
| 2036 | */ |
Eli Bendersky | 113da64 | 2012-06-15 07:52:49 +0300 | [diff] [blame] | 2037 | cur_parent = it->parent_stack->parent; |
| 2038 | child_index = it->parent_stack->child_index; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2039 | if (cur_parent->extra && child_index < cur_parent->extra->length) { |
| 2040 | ElementObject *child = (ElementObject *) |
| 2041 | cur_parent->extra->children[child_index]; |
| 2042 | it->parent_stack->child_index++; |
| 2043 | it->parent_stack = parent_stack_push_new(it->parent_stack, |
| 2044 | child); |
| 2045 | if (!it->parent_stack) { |
| 2046 | PyErr_NoMemory(); |
| 2047 | return NULL; |
| 2048 | } |
| 2049 | |
| 2050 | if (it->gettext) { |
Eli Bendersky | e6174ca | 2013-01-10 06:27:53 -0800 | [diff] [blame] | 2051 | PyObject *text = element_get_text(child); |
| 2052 | if (!text) |
| 2053 | return NULL; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2054 | if (PyObject_IsTrue(text)) { |
| 2055 | Py_INCREF(text); |
| 2056 | return text; |
| 2057 | } |
| 2058 | } else if (it->sought_tag == Py_None || |
| 2059 | PyObject_RichCompareBool(child->tag, |
| 2060 | it->sought_tag, Py_EQ) == 1) { |
| 2061 | Py_INCREF(child); |
| 2062 | return (PyObject *)child; |
| 2063 | } |
| 2064 | else |
| 2065 | continue; |
| 2066 | } |
| 2067 | else { |
Eli Bendersky | e6174ca | 2013-01-10 06:27:53 -0800 | [diff] [blame] | 2068 | PyObject *tail; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2069 | ParentLocator *next = it->parent_stack->next; |
Eli Bendersky | e6174ca | 2013-01-10 06:27:53 -0800 | [diff] [blame] | 2070 | if (it->gettext) { |
| 2071 | tail = element_get_tail(cur_parent); |
| 2072 | if (!tail) |
| 2073 | return NULL; |
| 2074 | } |
| 2075 | else |
| 2076 | tail = Py_None; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2077 | Py_XDECREF(it->parent_stack->parent); |
| 2078 | PyObject_Free(it->parent_stack); |
| 2079 | it->parent_stack = next; |
| 2080 | |
| 2081 | /* Note that extra condition on it->parent_stack->parent here; |
| 2082 | * this is because itertext() is supposed to only return *inner* |
| 2083 | * text, not text following the element it began iteration with. |
| 2084 | */ |
| 2085 | if (it->parent_stack->parent && PyObject_IsTrue(tail)) { |
| 2086 | Py_INCREF(tail); |
| 2087 | return tail; |
| 2088 | } |
| 2089 | } |
| 2090 | } |
| 2091 | |
| 2092 | return NULL; |
| 2093 | } |
| 2094 | |
| 2095 | |
| 2096 | static PyTypeObject ElementIter_Type = { |
| 2097 | PyVarObject_HEAD_INIT(NULL, 0) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 2098 | /* Using the module's name since the pure-Python implementation does not |
| 2099 | have such a type. */ |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2100 | "_elementtree._element_iterator", /* tp_name */ |
| 2101 | sizeof(ElementIterObject), /* tp_basicsize */ |
| 2102 | 0, /* tp_itemsize */ |
| 2103 | /* methods */ |
| 2104 | (destructor)elementiter_dealloc, /* tp_dealloc */ |
| 2105 | 0, /* tp_print */ |
| 2106 | 0, /* tp_getattr */ |
| 2107 | 0, /* tp_setattr */ |
| 2108 | 0, /* tp_reserved */ |
| 2109 | 0, /* tp_repr */ |
| 2110 | 0, /* tp_as_number */ |
| 2111 | 0, /* tp_as_sequence */ |
| 2112 | 0, /* tp_as_mapping */ |
| 2113 | 0, /* tp_hash */ |
| 2114 | 0, /* tp_call */ |
| 2115 | 0, /* tp_str */ |
| 2116 | 0, /* tp_getattro */ |
| 2117 | 0, /* tp_setattro */ |
| 2118 | 0, /* tp_as_buffer */ |
| 2119 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */ |
| 2120 | 0, /* tp_doc */ |
| 2121 | (traverseproc)elementiter_traverse, /* tp_traverse */ |
| 2122 | 0, /* tp_clear */ |
| 2123 | 0, /* tp_richcompare */ |
| 2124 | 0, /* tp_weaklistoffset */ |
| 2125 | PyObject_SelfIter, /* tp_iter */ |
| 2126 | (iternextfunc)elementiter_next, /* tp_iternext */ |
| 2127 | 0, /* tp_methods */ |
| 2128 | 0, /* tp_members */ |
| 2129 | 0, /* tp_getset */ |
| 2130 | 0, /* tp_base */ |
| 2131 | 0, /* tp_dict */ |
| 2132 | 0, /* tp_descr_get */ |
| 2133 | 0, /* tp_descr_set */ |
| 2134 | 0, /* tp_dictoffset */ |
| 2135 | 0, /* tp_init */ |
| 2136 | 0, /* tp_alloc */ |
| 2137 | 0, /* tp_new */ |
| 2138 | }; |
| 2139 | |
| 2140 | |
| 2141 | static PyObject * |
| 2142 | create_elementiter(ElementObject *self, PyObject *tag, int gettext) |
| 2143 | { |
| 2144 | ElementIterObject *it; |
| 2145 | PyObject *star = NULL; |
| 2146 | |
| 2147 | it = PyObject_GC_New(ElementIterObject, &ElementIter_Type); |
| 2148 | if (!it) |
| 2149 | return NULL; |
| 2150 | if (!(it->parent_stack = PyObject_Malloc(sizeof(ParentLocator)))) { |
| 2151 | PyObject_GC_Del(it); |
| 2152 | return NULL; |
| 2153 | } |
| 2154 | |
| 2155 | it->parent_stack->parent = NULL; |
| 2156 | it->parent_stack->child_index = 0; |
| 2157 | it->parent_stack->next = NULL; |
| 2158 | |
| 2159 | if (PyUnicode_Check(tag)) |
| 2160 | star = PyUnicode_FromString("*"); |
| 2161 | else if (PyBytes_Check(tag)) |
| 2162 | star = PyBytes_FromString("*"); |
| 2163 | |
| 2164 | if (star && PyObject_RichCompareBool(tag, star, Py_EQ) == 1) |
| 2165 | tag = Py_None; |
| 2166 | |
| 2167 | Py_XDECREF(star); |
| 2168 | it->sought_tag = tag; |
| 2169 | it->root_done = 0; |
| 2170 | it->gettext = gettext; |
| 2171 | it->root_element = self; |
| 2172 | |
| 2173 | Py_INCREF(self); |
| 2174 | Py_INCREF(tag); |
| 2175 | |
| 2176 | PyObject_GC_Track(it); |
| 2177 | return (PyObject *)it; |
| 2178 | } |
| 2179 | |
| 2180 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2181 | /* ==================================================================== */ |
| 2182 | /* the tree builder type */ |
| 2183 | |
| 2184 | typedef struct { |
| 2185 | PyObject_HEAD |
| 2186 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2187 | PyObject *root; /* root node (first created node) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2188 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2189 | PyObject *this; /* current node */ |
| 2190 | PyObject *last; /* most recently created node */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2191 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2192 | PyObject *data; /* data collector (string or list), or NULL */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2193 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2194 | PyObject *stack; /* element stack */ |
| 2195 | Py_ssize_t index; /* current stack size (0 means empty) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2196 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2197 | PyObject *element_factory; |
| 2198 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2199 | /* element tracing */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2200 | PyObject *events; /* list of events, or NULL if not collecting */ |
| 2201 | PyObject *start_event_obj; /* event objects (NULL to ignore) */ |
| 2202 | PyObject *end_event_obj; |
| 2203 | PyObject *start_ns_event_obj; |
| 2204 | PyObject *end_ns_event_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2205 | } TreeBuilderObject; |
| 2206 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 2207 | static PyTypeObject TreeBuilder_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2208 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 2209 | #define TreeBuilder_CheckExact(op) (Py_TYPE(op) == &TreeBuilder_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2210 | |
| 2211 | /* -------------------------------------------------------------------- */ |
| 2212 | /* constructor and destructor */ |
| 2213 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2214 | static PyObject * |
| 2215 | treebuilder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2216 | { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2217 | TreeBuilderObject *t = (TreeBuilderObject *)type->tp_alloc(type, 0); |
| 2218 | if (t != NULL) { |
| 2219 | t->root = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2220 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2221 | Py_INCREF(Py_None); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2222 | t->this = Py_None; |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2223 | Py_INCREF(Py_None); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2224 | t->last = Py_None; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2225 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2226 | t->data = NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2227 | t->element_factory = NULL; |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2228 | t->stack = PyList_New(20); |
| 2229 | if (!t->stack) { |
| 2230 | Py_DECREF(t->this); |
| 2231 | Py_DECREF(t->last); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2232 | Py_DECREF((PyObject *) t); |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2233 | return NULL; |
| 2234 | } |
| 2235 | t->index = 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2236 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2237 | t->events = NULL; |
| 2238 | t->start_event_obj = t->end_event_obj = NULL; |
| 2239 | t->start_ns_event_obj = t->end_ns_event_obj = NULL; |
| 2240 | } |
| 2241 | return (PyObject *)t; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2242 | } |
| 2243 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2244 | static int |
| 2245 | treebuilder_init(PyObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2246 | { |
Eli Bendersky | c68e136 | 2012-06-03 06:09:42 +0300 | [diff] [blame] | 2247 | static char *kwlist[] = {"element_factory", 0}; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2248 | PyObject *element_factory = NULL; |
| 2249 | TreeBuilderObject *self_tb = (TreeBuilderObject *)self; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2250 | PyObject *tmp; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2251 | |
| 2252 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O:TreeBuilder", kwlist, |
| 2253 | &element_factory)) { |
| 2254 | return -1; |
| 2255 | } |
| 2256 | |
| 2257 | if (element_factory) { |
| 2258 | Py_INCREF(element_factory); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2259 | tmp = self_tb->element_factory; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2260 | self_tb->element_factory = element_factory; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2261 | Py_XDECREF(tmp); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2262 | } |
| 2263 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2264 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2265 | } |
| 2266 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2267 | static int |
| 2268 | treebuilder_gc_traverse(TreeBuilderObject *self, visitproc visit, void *arg) |
| 2269 | { |
| 2270 | Py_VISIT(self->root); |
| 2271 | Py_VISIT(self->this); |
| 2272 | Py_VISIT(self->last); |
| 2273 | Py_VISIT(self->data); |
| 2274 | Py_VISIT(self->stack); |
| 2275 | Py_VISIT(self->element_factory); |
| 2276 | return 0; |
| 2277 | } |
| 2278 | |
| 2279 | static int |
| 2280 | treebuilder_gc_clear(TreeBuilderObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2281 | { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2282 | Py_CLEAR(self->end_ns_event_obj); |
| 2283 | Py_CLEAR(self->start_ns_event_obj); |
| 2284 | Py_CLEAR(self->end_event_obj); |
| 2285 | Py_CLEAR(self->start_event_obj); |
| 2286 | Py_CLEAR(self->events); |
| 2287 | Py_CLEAR(self->stack); |
| 2288 | Py_CLEAR(self->data); |
| 2289 | Py_CLEAR(self->last); |
| 2290 | Py_CLEAR(self->this); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2291 | Py_CLEAR(self->element_factory); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2292 | Py_CLEAR(self->root); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2293 | return 0; |
| 2294 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2295 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2296 | static void |
| 2297 | treebuilder_dealloc(TreeBuilderObject *self) |
| 2298 | { |
| 2299 | PyObject_GC_UnTrack(self); |
| 2300 | treebuilder_gc_clear(self); |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2301 | Py_TYPE(self)->tp_free((PyObject *)self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2302 | } |
| 2303 | |
| 2304 | /* -------------------------------------------------------------------- */ |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2305 | /* helpers for handling of arbitrary element-like objects */ |
| 2306 | |
| 2307 | static int |
| 2308 | treebuilder_set_element_text_or_tail(PyObject *element, PyObject *data, |
| 2309 | PyObject **dest, _Py_Identifier *name) |
| 2310 | { |
| 2311 | if (Element_CheckExact(element)) { |
| 2312 | Py_DECREF(JOIN_OBJ(*dest)); |
| 2313 | *dest = JOIN_SET(data, PyList_CheckExact(data)); |
| 2314 | return 0; |
| 2315 | } |
| 2316 | else { |
| 2317 | PyObject *joined = list_join(data); |
| 2318 | int r; |
| 2319 | if (joined == NULL) |
| 2320 | return -1; |
| 2321 | r = _PyObject_SetAttrId(element, name, joined); |
| 2322 | Py_DECREF(joined); |
| 2323 | return r; |
| 2324 | } |
| 2325 | } |
| 2326 | |
| 2327 | /* These two functions steal a reference to data */ |
| 2328 | static int |
| 2329 | treebuilder_set_element_text(PyObject *element, PyObject *data) |
| 2330 | { |
| 2331 | _Py_IDENTIFIER(text); |
| 2332 | return treebuilder_set_element_text_or_tail( |
| 2333 | element, data, &((ElementObject *) element)->text, &PyId_text); |
| 2334 | } |
| 2335 | |
| 2336 | static int |
| 2337 | treebuilder_set_element_tail(PyObject *element, PyObject *data) |
| 2338 | { |
| 2339 | _Py_IDENTIFIER(tail); |
| 2340 | return treebuilder_set_element_text_or_tail( |
| 2341 | element, data, &((ElementObject *) element)->tail, &PyId_tail); |
| 2342 | } |
| 2343 | |
| 2344 | static int |
| 2345 | treebuilder_add_subelement(PyObject *element, PyObject *child) |
| 2346 | { |
| 2347 | _Py_IDENTIFIER(append); |
| 2348 | if (Element_CheckExact(element)) { |
| 2349 | ElementObject *elem = (ElementObject *) element; |
| 2350 | return element_add_subelement(elem, child); |
| 2351 | } |
| 2352 | else { |
| 2353 | PyObject *res; |
| 2354 | res = _PyObject_CallMethodId(element, &PyId_append, "O", child); |
| 2355 | if (res == NULL) |
| 2356 | return -1; |
| 2357 | Py_DECREF(res); |
| 2358 | return 0; |
| 2359 | } |
| 2360 | } |
| 2361 | |
| 2362 | /* -------------------------------------------------------------------- */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2363 | /* handlers */ |
| 2364 | |
| 2365 | LOCAL(PyObject*) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2366 | treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag, |
| 2367 | PyObject* attrib) |
| 2368 | { |
| 2369 | PyObject* node; |
| 2370 | PyObject* this; |
| 2371 | |
| 2372 | if (self->data) { |
| 2373 | if (self->this == self->last) { |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2374 | if (treebuilder_set_element_text(self->last, self->data)) |
| 2375 | return NULL; |
| 2376 | } |
| 2377 | else { |
| 2378 | if (treebuilder_set_element_tail(self->last, self->data)) |
| 2379 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2380 | } |
| 2381 | self->data = NULL; |
| 2382 | } |
| 2383 | |
Eli Bendersky | 08231a9 | 2013-05-18 15:47:16 -0700 | [diff] [blame] | 2384 | if (self->element_factory && self->element_factory != Py_None) { |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2385 | node = PyObject_CallFunction(self->element_factory, "OO", tag, attrib); |
| 2386 | } else { |
| 2387 | node = create_new_element(tag, attrib); |
| 2388 | } |
| 2389 | if (!node) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2390 | return NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2391 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2392 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2393 | this = self->this; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2394 | |
| 2395 | if (this != Py_None) { |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2396 | if (treebuilder_add_subelement(this, node) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2397 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2398 | } else { |
| 2399 | if (self->root) { |
| 2400 | PyErr_SetString( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2401 | elementtree_parseerror_obj, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2402 | "multiple elements on top level" |
| 2403 | ); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2404 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2405 | } |
| 2406 | Py_INCREF(node); |
| 2407 | self->root = node; |
| 2408 | } |
| 2409 | |
| 2410 | if (self->index < PyList_GET_SIZE(self->stack)) { |
| 2411 | if (PyList_SetItem(self->stack, self->index, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2412 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2413 | Py_INCREF(this); |
| 2414 | } else { |
| 2415 | if (PyList_Append(self->stack, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2416 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2417 | } |
| 2418 | self->index++; |
| 2419 | |
| 2420 | Py_DECREF(this); |
| 2421 | Py_INCREF(node); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2422 | self->this = node; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2423 | |
| 2424 | Py_DECREF(self->last); |
| 2425 | Py_INCREF(node); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2426 | self->last = node; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2427 | |
| 2428 | if (self->start_event_obj) { |
| 2429 | PyObject* res; |
| 2430 | PyObject* action = self->start_event_obj; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2431 | res = PyTuple_Pack(2, action, node); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2432 | if (res) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2433 | PyList_Append(self->events, res); |
| 2434 | Py_DECREF(res); |
| 2435 | } else |
| 2436 | PyErr_Clear(); /* FIXME: propagate error */ |
| 2437 | } |
| 2438 | |
| 2439 | return node; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2440 | |
| 2441 | error: |
| 2442 | Py_DECREF(node); |
| 2443 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2444 | } |
| 2445 | |
| 2446 | LOCAL(PyObject*) |
| 2447 | treebuilder_handle_data(TreeBuilderObject* self, PyObject* data) |
| 2448 | { |
| 2449 | if (!self->data) { |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2450 | if (self->last == Py_None) { |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 2451 | /* ignore calls to data before the first call to start */ |
| 2452 | Py_RETURN_NONE; |
| 2453 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2454 | /* store the first item as is */ |
| 2455 | Py_INCREF(data); self->data = data; |
| 2456 | } else { |
| 2457 | /* more than one item; use a list to collect items */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2458 | if (PyBytes_CheckExact(self->data) && Py_REFCNT(self->data) == 1 && |
| 2459 | PyBytes_CheckExact(data) && PyBytes_GET_SIZE(data) == 1) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2460 | /* XXX this code path unused in Python 3? */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2461 | /* expat often generates single character data sections; handle |
| 2462 | the most common case by resizing the existing string... */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2463 | Py_ssize_t size = PyBytes_GET_SIZE(self->data); |
| 2464 | if (_PyBytes_Resize(&self->data, size + 1) < 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2465 | return NULL; |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2466 | PyBytes_AS_STRING(self->data)[size] = PyBytes_AS_STRING(data)[0]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2467 | } else if (PyList_CheckExact(self->data)) { |
| 2468 | if (PyList_Append(self->data, data) < 0) |
| 2469 | return NULL; |
| 2470 | } else { |
| 2471 | PyObject* list = PyList_New(2); |
| 2472 | if (!list) |
| 2473 | return NULL; |
| 2474 | PyList_SET_ITEM(list, 0, self->data); |
| 2475 | Py_INCREF(data); PyList_SET_ITEM(list, 1, data); |
| 2476 | self->data = list; |
| 2477 | } |
| 2478 | } |
| 2479 | |
| 2480 | Py_RETURN_NONE; |
| 2481 | } |
| 2482 | |
| 2483 | LOCAL(PyObject*) |
| 2484 | treebuilder_handle_end(TreeBuilderObject* self, PyObject* tag) |
| 2485 | { |
| 2486 | PyObject* item; |
| 2487 | |
| 2488 | if (self->data) { |
| 2489 | if (self->this == self->last) { |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2490 | if (treebuilder_set_element_text(self->last, self->data)) |
| 2491 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2492 | } else { |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2493 | if (treebuilder_set_element_tail(self->last, self->data)) |
| 2494 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2495 | } |
| 2496 | self->data = NULL; |
| 2497 | } |
| 2498 | |
| 2499 | if (self->index == 0) { |
| 2500 | PyErr_SetString( |
| 2501 | PyExc_IndexError, |
| 2502 | "pop from empty stack" |
| 2503 | ); |
| 2504 | return NULL; |
| 2505 | } |
| 2506 | |
| 2507 | self->index--; |
| 2508 | |
| 2509 | item = PyList_GET_ITEM(self->stack, self->index); |
| 2510 | Py_INCREF(item); |
| 2511 | |
| 2512 | Py_DECREF(self->last); |
| 2513 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2514 | self->last = self->this; |
| 2515 | self->this = item; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2516 | |
| 2517 | if (self->end_event_obj) { |
| 2518 | PyObject* res; |
| 2519 | PyObject* action = self->end_event_obj; |
| 2520 | PyObject* node = (PyObject*) self->last; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2521 | res = PyTuple_Pack(2, action, node); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2522 | if (res) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2523 | PyList_Append(self->events, res); |
| 2524 | Py_DECREF(res); |
| 2525 | } else |
| 2526 | PyErr_Clear(); /* FIXME: propagate error */ |
| 2527 | } |
| 2528 | |
| 2529 | Py_INCREF(self->last); |
| 2530 | return (PyObject*) self->last; |
| 2531 | } |
| 2532 | |
| 2533 | LOCAL(void) |
| 2534 | treebuilder_handle_namespace(TreeBuilderObject* self, int start, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2535 | PyObject *prefix, PyObject *uri) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2536 | { |
| 2537 | PyObject* res; |
| 2538 | PyObject* action; |
| 2539 | PyObject* parcel; |
| 2540 | |
| 2541 | if (!self->events) |
| 2542 | return; |
| 2543 | |
| 2544 | if (start) { |
| 2545 | if (!self->start_ns_event_obj) |
| 2546 | return; |
| 2547 | action = self->start_ns_event_obj; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2548 | parcel = Py_BuildValue("OO", prefix, uri); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2549 | if (!parcel) |
| 2550 | return; |
| 2551 | Py_INCREF(action); |
| 2552 | } else { |
| 2553 | if (!self->end_ns_event_obj) |
| 2554 | return; |
| 2555 | action = self->end_ns_event_obj; |
| 2556 | Py_INCREF(action); |
| 2557 | parcel = Py_None; |
| 2558 | Py_INCREF(parcel); |
| 2559 | } |
| 2560 | |
| 2561 | res = PyTuple_New(2); |
| 2562 | |
| 2563 | if (res) { |
| 2564 | PyTuple_SET_ITEM(res, 0, action); |
| 2565 | PyTuple_SET_ITEM(res, 1, parcel); |
| 2566 | PyList_Append(self->events, res); |
| 2567 | Py_DECREF(res); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2568 | } |
| 2569 | else { |
| 2570 | Py_DECREF(action); |
| 2571 | Py_DECREF(parcel); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2572 | PyErr_Clear(); /* FIXME: propagate error */ |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2573 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2574 | } |
| 2575 | |
| 2576 | /* -------------------------------------------------------------------- */ |
| 2577 | /* methods (in alphabetical order) */ |
| 2578 | |
| 2579 | static PyObject* |
| 2580 | treebuilder_data(TreeBuilderObject* self, PyObject* args) |
| 2581 | { |
| 2582 | PyObject* data; |
| 2583 | if (!PyArg_ParseTuple(args, "O:data", &data)) |
| 2584 | return NULL; |
| 2585 | |
| 2586 | return treebuilder_handle_data(self, data); |
| 2587 | } |
| 2588 | |
| 2589 | static PyObject* |
| 2590 | treebuilder_end(TreeBuilderObject* self, PyObject* args) |
| 2591 | { |
| 2592 | PyObject* tag; |
| 2593 | if (!PyArg_ParseTuple(args, "O:end", &tag)) |
| 2594 | return NULL; |
| 2595 | |
| 2596 | return treebuilder_handle_end(self, tag); |
| 2597 | } |
| 2598 | |
| 2599 | LOCAL(PyObject*) |
| 2600 | treebuilder_done(TreeBuilderObject* self) |
| 2601 | { |
| 2602 | PyObject* res; |
| 2603 | |
| 2604 | /* FIXME: check stack size? */ |
| 2605 | |
| 2606 | if (self->root) |
| 2607 | res = self->root; |
| 2608 | else |
| 2609 | res = Py_None; |
| 2610 | |
| 2611 | Py_INCREF(res); |
| 2612 | return res; |
| 2613 | } |
| 2614 | |
| 2615 | static PyObject* |
| 2616 | treebuilder_close(TreeBuilderObject* self, PyObject* args) |
| 2617 | { |
| 2618 | if (!PyArg_ParseTuple(args, ":close")) |
| 2619 | return NULL; |
| 2620 | |
| 2621 | return treebuilder_done(self); |
| 2622 | } |
| 2623 | |
| 2624 | static PyObject* |
| 2625 | treebuilder_start(TreeBuilderObject* self, PyObject* args) |
| 2626 | { |
| 2627 | PyObject* tag; |
| 2628 | PyObject* attrib = Py_None; |
| 2629 | if (!PyArg_ParseTuple(args, "O|O:start", &tag, &attrib)) |
| 2630 | return NULL; |
| 2631 | |
| 2632 | return treebuilder_handle_start(self, tag, attrib); |
| 2633 | } |
| 2634 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2635 | static PyMethodDef treebuilder_methods[] = { |
| 2636 | {"data", (PyCFunction) treebuilder_data, METH_VARARGS}, |
| 2637 | {"start", (PyCFunction) treebuilder_start, METH_VARARGS}, |
| 2638 | {"end", (PyCFunction) treebuilder_end, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2639 | {"close", (PyCFunction) treebuilder_close, METH_VARARGS}, |
| 2640 | {NULL, NULL} |
| 2641 | }; |
| 2642 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 2643 | static PyTypeObject TreeBuilder_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 2644 | PyVarObject_HEAD_INIT(NULL, 0) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 2645 | "xml.etree.ElementTree.TreeBuilder", sizeof(TreeBuilderObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2646 | /* methods */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2647 | (destructor)treebuilder_dealloc, /* tp_dealloc */ |
| 2648 | 0, /* tp_print */ |
| 2649 | 0, /* tp_getattr */ |
| 2650 | 0, /* tp_setattr */ |
| 2651 | 0, /* tp_reserved */ |
| 2652 | 0, /* tp_repr */ |
| 2653 | 0, /* tp_as_number */ |
| 2654 | 0, /* tp_as_sequence */ |
| 2655 | 0, /* tp_as_mapping */ |
| 2656 | 0, /* tp_hash */ |
| 2657 | 0, /* tp_call */ |
| 2658 | 0, /* tp_str */ |
| 2659 | 0, /* tp_getattro */ |
| 2660 | 0, /* tp_setattro */ |
| 2661 | 0, /* tp_as_buffer */ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2662 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 2663 | /* tp_flags */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2664 | 0, /* tp_doc */ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2665 | (traverseproc)treebuilder_gc_traverse, /* tp_traverse */ |
| 2666 | (inquiry)treebuilder_gc_clear, /* tp_clear */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2667 | 0, /* tp_richcompare */ |
| 2668 | 0, /* tp_weaklistoffset */ |
| 2669 | 0, /* tp_iter */ |
| 2670 | 0, /* tp_iternext */ |
| 2671 | treebuilder_methods, /* tp_methods */ |
| 2672 | 0, /* tp_members */ |
| 2673 | 0, /* tp_getset */ |
| 2674 | 0, /* tp_base */ |
| 2675 | 0, /* tp_dict */ |
| 2676 | 0, /* tp_descr_get */ |
| 2677 | 0, /* tp_descr_set */ |
| 2678 | 0, /* tp_dictoffset */ |
| 2679 | (initproc)treebuilder_init, /* tp_init */ |
| 2680 | PyType_GenericAlloc, /* tp_alloc */ |
| 2681 | treebuilder_new, /* tp_new */ |
| 2682 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2683 | }; |
| 2684 | |
| 2685 | /* ==================================================================== */ |
| 2686 | /* the expat interface */ |
| 2687 | |
| 2688 | #if defined(USE_EXPAT) |
| 2689 | |
| 2690 | #include "expat.h" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2691 | #include "pyexpat.h" |
Eli Bendersky | 20d4174 | 2012-06-01 09:48:37 +0300 | [diff] [blame] | 2692 | static struct PyExpat_CAPI *expat_capi; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2693 | #define EXPAT(func) (expat_capi->func) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2694 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2695 | static XML_Memory_Handling_Suite ExpatMemoryHandler = { |
| 2696 | PyObject_Malloc, PyObject_Realloc, PyObject_Free}; |
| 2697 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2698 | typedef struct { |
| 2699 | PyObject_HEAD |
| 2700 | |
| 2701 | XML_Parser parser; |
| 2702 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2703 | PyObject *target; |
| 2704 | PyObject *entity; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2705 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2706 | PyObject *names; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2707 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2708 | PyObject *handle_start; |
| 2709 | PyObject *handle_data; |
| 2710 | PyObject *handle_end; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2711 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2712 | PyObject *handle_comment; |
| 2713 | PyObject *handle_pi; |
| 2714 | PyObject *handle_doctype; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2715 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2716 | PyObject *handle_close; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2717 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2718 | } XMLParserObject; |
| 2719 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 2720 | static PyTypeObject XMLParser_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2721 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2722 | #define XMLParser_CheckExact(op) (Py_TYPE(op) == &XMLParser_Type) |
| 2723 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2724 | /* helpers */ |
| 2725 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2726 | LOCAL(PyObject*) |
| 2727 | makeuniversal(XMLParserObject* self, const char* string) |
| 2728 | { |
| 2729 | /* convert a UTF-8 tag/attribute name from the expat parser |
| 2730 | to a universal name string */ |
| 2731 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2732 | Py_ssize_t size = (Py_ssize_t) strlen(string); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2733 | PyObject* key; |
| 2734 | PyObject* value; |
| 2735 | |
| 2736 | /* look the 'raw' name up in the names dictionary */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2737 | key = PyBytes_FromStringAndSize(string, size); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2738 | if (!key) |
| 2739 | return NULL; |
| 2740 | |
| 2741 | value = PyDict_GetItem(self->names, key); |
| 2742 | |
| 2743 | if (value) { |
| 2744 | Py_INCREF(value); |
| 2745 | } else { |
| 2746 | /* new name. convert to universal name, and decode as |
| 2747 | necessary */ |
| 2748 | |
| 2749 | PyObject* tag; |
| 2750 | char* p; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2751 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2752 | |
| 2753 | /* look for namespace separator */ |
| 2754 | for (i = 0; i < size; i++) |
| 2755 | if (string[i] == '}') |
| 2756 | break; |
| 2757 | if (i != size) { |
| 2758 | /* convert to universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2759 | tag = PyBytes_FromStringAndSize(NULL, size+1); |
| 2760 | p = PyBytes_AS_STRING(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2761 | p[0] = '{'; |
| 2762 | memcpy(p+1, string, size); |
| 2763 | size++; |
| 2764 | } else { |
| 2765 | /* plain name; use key as tag */ |
| 2766 | Py_INCREF(key); |
| 2767 | tag = key; |
| 2768 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2769 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2770 | /* decode universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2771 | p = PyBytes_AS_STRING(tag); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2772 | value = PyUnicode_DecodeUTF8(p, size, "strict"); |
| 2773 | Py_DECREF(tag); |
| 2774 | if (!value) { |
| 2775 | Py_DECREF(key); |
| 2776 | return NULL; |
| 2777 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2778 | |
| 2779 | /* add to names dictionary */ |
| 2780 | if (PyDict_SetItem(self->names, key, value) < 0) { |
| 2781 | Py_DECREF(key); |
| 2782 | Py_DECREF(value); |
| 2783 | return NULL; |
| 2784 | } |
| 2785 | } |
| 2786 | |
| 2787 | Py_DECREF(key); |
| 2788 | return value; |
| 2789 | } |
| 2790 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2791 | /* Set the ParseError exception with the given parameters. |
| 2792 | * If message is not NULL, it's used as the error string. Otherwise, the |
| 2793 | * message string is the default for the given error_code. |
| 2794 | */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2795 | static void |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2796 | 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] | 2797 | { |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2798 | PyObject *errmsg, *error, *position, *code; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2799 | |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2800 | errmsg = PyUnicode_FromFormat("%s: line %d, column %d", |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2801 | message ? message : EXPAT(ErrorString)(error_code), |
| 2802 | line, column); |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2803 | if (errmsg == NULL) |
| 2804 | return; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2805 | |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2806 | error = PyObject_CallFunction(elementtree_parseerror_obj, "O", errmsg); |
| 2807 | Py_DECREF(errmsg); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2808 | if (!error) |
| 2809 | return; |
| 2810 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2811 | /* Add code and position attributes */ |
| 2812 | code = PyLong_FromLong((long)error_code); |
| 2813 | if (!code) { |
| 2814 | Py_DECREF(error); |
| 2815 | return; |
| 2816 | } |
| 2817 | if (PyObject_SetAttrString(error, "code", code) == -1) { |
| 2818 | Py_DECREF(error); |
| 2819 | Py_DECREF(code); |
| 2820 | return; |
| 2821 | } |
| 2822 | Py_DECREF(code); |
| 2823 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2824 | position = Py_BuildValue("(ii)", line, column); |
| 2825 | if (!position) { |
| 2826 | Py_DECREF(error); |
| 2827 | return; |
| 2828 | } |
| 2829 | if (PyObject_SetAttrString(error, "position", position) == -1) { |
| 2830 | Py_DECREF(error); |
| 2831 | Py_DECREF(position); |
| 2832 | return; |
| 2833 | } |
| 2834 | Py_DECREF(position); |
| 2835 | |
| 2836 | PyErr_SetObject(elementtree_parseerror_obj, error); |
| 2837 | Py_DECREF(error); |
| 2838 | } |
| 2839 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2840 | /* -------------------------------------------------------------------- */ |
| 2841 | /* handlers */ |
| 2842 | |
| 2843 | static void |
| 2844 | expat_default_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2845 | int data_len) |
| 2846 | { |
| 2847 | PyObject* key; |
| 2848 | PyObject* value; |
| 2849 | PyObject* res; |
| 2850 | |
| 2851 | if (data_len < 2 || data_in[0] != '&') |
| 2852 | return; |
| 2853 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2854 | key = PyUnicode_DecodeUTF8(data_in + 1, data_len - 2, "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2855 | if (!key) |
| 2856 | return; |
| 2857 | |
| 2858 | value = PyDict_GetItem(self->entity, key); |
| 2859 | |
| 2860 | if (value) { |
| 2861 | if (TreeBuilder_CheckExact(self->target)) |
| 2862 | res = treebuilder_handle_data( |
| 2863 | (TreeBuilderObject*) self->target, value |
| 2864 | ); |
| 2865 | else if (self->handle_data) |
| 2866 | res = PyObject_CallFunction(self->handle_data, "O", value); |
| 2867 | else |
| 2868 | res = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2869 | Py_XDECREF(res); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2870 | } else if (!PyErr_Occurred()) { |
| 2871 | /* Report the first error, not the last */ |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 2872 | char message[128] = "undefined entity "; |
| 2873 | strncat(message, data_in, data_len < 100?data_len:100); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2874 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2875 | XML_ERROR_UNDEFINED_ENTITY, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2876 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2877 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 2878 | message |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2879 | ); |
| 2880 | } |
| 2881 | |
| 2882 | Py_DECREF(key); |
| 2883 | } |
| 2884 | |
| 2885 | static void |
| 2886 | expat_start_handler(XMLParserObject* self, const XML_Char* tag_in, |
| 2887 | const XML_Char **attrib_in) |
| 2888 | { |
| 2889 | PyObject* res; |
| 2890 | PyObject* tag; |
| 2891 | PyObject* attrib; |
| 2892 | int ok; |
| 2893 | |
| 2894 | /* tag name */ |
| 2895 | tag = makeuniversal(self, tag_in); |
| 2896 | if (!tag) |
| 2897 | return; /* parser will look for errors */ |
| 2898 | |
| 2899 | /* attributes */ |
| 2900 | if (attrib_in[0]) { |
| 2901 | attrib = PyDict_New(); |
| 2902 | if (!attrib) |
| 2903 | return; |
| 2904 | while (attrib_in[0] && attrib_in[1]) { |
| 2905 | PyObject* key = makeuniversal(self, attrib_in[0]); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2906 | PyObject* value = PyUnicode_DecodeUTF8(attrib_in[1], strlen(attrib_in[1]), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2907 | if (!key || !value) { |
| 2908 | Py_XDECREF(value); |
| 2909 | Py_XDECREF(key); |
| 2910 | Py_DECREF(attrib); |
| 2911 | return; |
| 2912 | } |
| 2913 | ok = PyDict_SetItem(attrib, key, value); |
| 2914 | Py_DECREF(value); |
| 2915 | Py_DECREF(key); |
| 2916 | if (ok < 0) { |
| 2917 | Py_DECREF(attrib); |
| 2918 | return; |
| 2919 | } |
| 2920 | attrib_in += 2; |
| 2921 | } |
| 2922 | } else { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2923 | /* Pass an empty dictionary on */ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2924 | attrib = PyDict_New(); |
| 2925 | if (!attrib) |
| 2926 | return; |
| 2927 | } |
| 2928 | |
| 2929 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2930 | /* shortcut */ |
| 2931 | res = treebuilder_handle_start((TreeBuilderObject*) self->target, |
| 2932 | tag, attrib); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2933 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2934 | else if (self->handle_start) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2935 | res = PyObject_CallFunction(self->handle_start, "OO", tag, attrib); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2936 | } else |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2937 | res = NULL; |
| 2938 | |
| 2939 | Py_DECREF(tag); |
| 2940 | Py_DECREF(attrib); |
| 2941 | |
| 2942 | Py_XDECREF(res); |
| 2943 | } |
| 2944 | |
| 2945 | static void |
| 2946 | expat_data_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2947 | int data_len) |
| 2948 | { |
| 2949 | PyObject* data; |
| 2950 | PyObject* res; |
| 2951 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2952 | data = PyUnicode_DecodeUTF8(data_in, data_len, "strict"); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2953 | if (!data) |
| 2954 | return; /* parser will look for errors */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2955 | |
| 2956 | if (TreeBuilder_CheckExact(self->target)) |
| 2957 | /* shortcut */ |
| 2958 | res = treebuilder_handle_data((TreeBuilderObject*) self->target, data); |
| 2959 | else if (self->handle_data) |
| 2960 | res = PyObject_CallFunction(self->handle_data, "O", data); |
| 2961 | else |
| 2962 | res = NULL; |
| 2963 | |
| 2964 | Py_DECREF(data); |
| 2965 | |
| 2966 | Py_XDECREF(res); |
| 2967 | } |
| 2968 | |
| 2969 | static void |
| 2970 | expat_end_handler(XMLParserObject* self, const XML_Char* tag_in) |
| 2971 | { |
| 2972 | PyObject* tag; |
| 2973 | PyObject* res = NULL; |
| 2974 | |
| 2975 | if (TreeBuilder_CheckExact(self->target)) |
| 2976 | /* shortcut */ |
| 2977 | /* the standard tree builder doesn't look at the end tag */ |
| 2978 | res = treebuilder_handle_end( |
| 2979 | (TreeBuilderObject*) self->target, Py_None |
| 2980 | ); |
| 2981 | else if (self->handle_end) { |
| 2982 | tag = makeuniversal(self, tag_in); |
| 2983 | if (tag) { |
| 2984 | res = PyObject_CallFunction(self->handle_end, "O", tag); |
| 2985 | Py_DECREF(tag); |
| 2986 | } |
| 2987 | } |
| 2988 | |
| 2989 | Py_XDECREF(res); |
| 2990 | } |
| 2991 | |
| 2992 | static void |
| 2993 | expat_start_ns_handler(XMLParserObject* self, const XML_Char* prefix, |
| 2994 | const XML_Char *uri) |
| 2995 | { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2996 | PyObject* sprefix = NULL; |
| 2997 | PyObject* suri = NULL; |
| 2998 | |
| 2999 | suri = PyUnicode_DecodeUTF8(uri, strlen(uri), "strict"); |
| 3000 | if (!suri) |
| 3001 | return; |
| 3002 | |
| 3003 | if (prefix) |
| 3004 | sprefix = PyUnicode_DecodeUTF8(prefix, strlen(prefix), "strict"); |
| 3005 | else |
| 3006 | sprefix = PyUnicode_FromString(""); |
| 3007 | if (!sprefix) { |
| 3008 | Py_DECREF(suri); |
| 3009 | return; |
| 3010 | } |
| 3011 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3012 | treebuilder_handle_namespace( |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3013 | (TreeBuilderObject*) self->target, 1, sprefix, suri |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3014 | ); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3015 | |
| 3016 | Py_DECREF(sprefix); |
| 3017 | Py_DECREF(suri); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3018 | } |
| 3019 | |
| 3020 | static void |
| 3021 | expat_end_ns_handler(XMLParserObject* self, const XML_Char* prefix_in) |
| 3022 | { |
| 3023 | treebuilder_handle_namespace( |
| 3024 | (TreeBuilderObject*) self->target, 0, NULL, NULL |
| 3025 | ); |
| 3026 | } |
| 3027 | |
| 3028 | static void |
| 3029 | expat_comment_handler(XMLParserObject* self, const XML_Char* comment_in) |
| 3030 | { |
| 3031 | PyObject* comment; |
| 3032 | PyObject* res; |
| 3033 | |
| 3034 | if (self->handle_comment) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 3035 | comment = PyUnicode_DecodeUTF8(comment_in, strlen(comment_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3036 | if (comment) { |
| 3037 | res = PyObject_CallFunction(self->handle_comment, "O", comment); |
| 3038 | Py_XDECREF(res); |
| 3039 | Py_DECREF(comment); |
| 3040 | } |
| 3041 | } |
| 3042 | } |
| 3043 | |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 3044 | static void |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3045 | expat_start_doctype_handler(XMLParserObject *self, |
| 3046 | const XML_Char *doctype_name, |
| 3047 | const XML_Char *sysid, |
| 3048 | const XML_Char *pubid, |
| 3049 | int has_internal_subset) |
| 3050 | { |
| 3051 | PyObject *self_pyobj = (PyObject *)self; |
| 3052 | PyObject *doctype_name_obj, *sysid_obj, *pubid_obj; |
| 3053 | PyObject *parser_doctype = NULL; |
| 3054 | PyObject *res = NULL; |
| 3055 | |
| 3056 | doctype_name_obj = makeuniversal(self, doctype_name); |
| 3057 | if (!doctype_name_obj) |
| 3058 | return; |
| 3059 | |
| 3060 | if (sysid) { |
| 3061 | sysid_obj = makeuniversal(self, sysid); |
| 3062 | if (!sysid_obj) { |
| 3063 | Py_DECREF(doctype_name_obj); |
| 3064 | return; |
| 3065 | } |
| 3066 | } else { |
| 3067 | Py_INCREF(Py_None); |
| 3068 | sysid_obj = Py_None; |
| 3069 | } |
| 3070 | |
| 3071 | if (pubid) { |
| 3072 | pubid_obj = makeuniversal(self, pubid); |
| 3073 | if (!pubid_obj) { |
| 3074 | Py_DECREF(doctype_name_obj); |
| 3075 | Py_DECREF(sysid_obj); |
| 3076 | return; |
| 3077 | } |
| 3078 | } else { |
| 3079 | Py_INCREF(Py_None); |
| 3080 | pubid_obj = Py_None; |
| 3081 | } |
| 3082 | |
| 3083 | /* If the target has a handler for doctype, call it. */ |
| 3084 | if (self->handle_doctype) { |
| 3085 | res = PyObject_CallFunction(self->handle_doctype, "OOO", |
| 3086 | doctype_name_obj, pubid_obj, sysid_obj); |
| 3087 | Py_CLEAR(res); |
| 3088 | } |
| 3089 | |
| 3090 | /* Now see if the parser itself has a doctype method. If yes and it's |
| 3091 | * a subclass, call it but warn about deprecation. If it's not a subclass |
| 3092 | * (i.e. vanilla XMLParser), do nothing. |
| 3093 | */ |
| 3094 | parser_doctype = PyObject_GetAttrString(self_pyobj, "doctype"); |
| 3095 | if (parser_doctype) { |
| 3096 | if (!XMLParser_CheckExact(self_pyobj)) { |
| 3097 | if (PyErr_WarnEx(PyExc_DeprecationWarning, |
| 3098 | "This method of XMLParser is deprecated. Define" |
| 3099 | " doctype() method on the TreeBuilder target.", |
| 3100 | 1) < 0) { |
| 3101 | goto clear; |
| 3102 | } |
| 3103 | res = PyObject_CallFunction(parser_doctype, "OOO", |
| 3104 | doctype_name_obj, pubid_obj, sysid_obj); |
| 3105 | Py_CLEAR(res); |
| 3106 | } |
| 3107 | } |
| 3108 | |
| 3109 | clear: |
| 3110 | Py_XDECREF(parser_doctype); |
| 3111 | Py_DECREF(doctype_name_obj); |
| 3112 | Py_DECREF(pubid_obj); |
| 3113 | Py_DECREF(sysid_obj); |
| 3114 | } |
| 3115 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3116 | static void |
| 3117 | expat_pi_handler(XMLParserObject* self, const XML_Char* target_in, |
| 3118 | const XML_Char* data_in) |
| 3119 | { |
| 3120 | PyObject* target; |
| 3121 | PyObject* data; |
| 3122 | PyObject* res; |
| 3123 | |
| 3124 | if (self->handle_pi) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 3125 | target = PyUnicode_DecodeUTF8(target_in, strlen(target_in), "strict"); |
| 3126 | data = PyUnicode_DecodeUTF8(data_in, strlen(data_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3127 | if (target && data) { |
| 3128 | res = PyObject_CallFunction(self->handle_pi, "OO", target, data); |
| 3129 | Py_XDECREF(res); |
| 3130 | Py_DECREF(data); |
| 3131 | Py_DECREF(target); |
| 3132 | } else { |
| 3133 | Py_XDECREF(data); |
| 3134 | Py_XDECREF(target); |
| 3135 | } |
| 3136 | } |
| 3137 | } |
| 3138 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3139 | /* -------------------------------------------------------------------- */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3140 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3141 | static PyObject * |
| 3142 | xmlparser_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3143 | { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3144 | XMLParserObject *self = (XMLParserObject *)type->tp_alloc(type, 0); |
| 3145 | if (self) { |
| 3146 | self->parser = NULL; |
| 3147 | self->target = self->entity = self->names = NULL; |
| 3148 | self->handle_start = self->handle_data = self->handle_end = NULL; |
| 3149 | self->handle_comment = self->handle_pi = self->handle_close = NULL; |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3150 | self->handle_doctype = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3151 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3152 | return (PyObject *)self; |
| 3153 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3154 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3155 | static int |
| 3156 | xmlparser_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 3157 | { |
| 3158 | XMLParserObject *self_xp = (XMLParserObject *)self; |
| 3159 | PyObject *target = NULL, *html = NULL; |
| 3160 | char *encoding = NULL; |
Eli Bendersky | c68e136 | 2012-06-03 06:09:42 +0300 | [diff] [blame] | 3161 | static char *kwlist[] = {"html", "target", "encoding", 0}; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3162 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3163 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|OOz:XMLParser", kwlist, |
| 3164 | &html, &target, &encoding)) { |
| 3165 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3166 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3167 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3168 | self_xp->entity = PyDict_New(); |
| 3169 | if (!self_xp->entity) |
| 3170 | return -1; |
| 3171 | |
| 3172 | self_xp->names = PyDict_New(); |
| 3173 | if (!self_xp->names) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3174 | Py_CLEAR(self_xp->entity); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3175 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3176 | } |
| 3177 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3178 | self_xp->parser = EXPAT(ParserCreate_MM)(encoding, &ExpatMemoryHandler, "}"); |
| 3179 | if (!self_xp->parser) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3180 | Py_CLEAR(self_xp->entity); |
| 3181 | Py_CLEAR(self_xp->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3182 | PyErr_NoMemory(); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3183 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3184 | } |
| 3185 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3186 | if (target) { |
| 3187 | Py_INCREF(target); |
| 3188 | } else { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 3189 | target = treebuilder_new(&TreeBuilder_Type, NULL, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3190 | if (!target) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3191 | Py_CLEAR(self_xp->entity); |
| 3192 | Py_CLEAR(self_xp->names); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3193 | EXPAT(ParserFree)(self_xp->parser); |
| 3194 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3195 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3196 | } |
| 3197 | self_xp->target = target; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3198 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3199 | self_xp->handle_start = PyObject_GetAttrString(target, "start"); |
| 3200 | self_xp->handle_data = PyObject_GetAttrString(target, "data"); |
| 3201 | self_xp->handle_end = PyObject_GetAttrString(target, "end"); |
| 3202 | self_xp->handle_comment = PyObject_GetAttrString(target, "comment"); |
| 3203 | self_xp->handle_pi = PyObject_GetAttrString(target, "pi"); |
| 3204 | self_xp->handle_close = PyObject_GetAttrString(target, "close"); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3205 | self_xp->handle_doctype = PyObject_GetAttrString(target, "doctype"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3206 | |
| 3207 | PyErr_Clear(); |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 3208 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3209 | /* configure parser */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3210 | EXPAT(SetUserData)(self_xp->parser, self_xp); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3211 | EXPAT(SetElementHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3212 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3213 | (XML_StartElementHandler) expat_start_handler, |
| 3214 | (XML_EndElementHandler) expat_end_handler |
| 3215 | ); |
| 3216 | EXPAT(SetDefaultHandlerExpand)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3217 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3218 | (XML_DefaultHandler) expat_default_handler |
| 3219 | ); |
| 3220 | EXPAT(SetCharacterDataHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3221 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3222 | (XML_CharacterDataHandler) expat_data_handler |
| 3223 | ); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3224 | if (self_xp->handle_comment) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3225 | EXPAT(SetCommentHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3226 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3227 | (XML_CommentHandler) expat_comment_handler |
| 3228 | ); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3229 | if (self_xp->handle_pi) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3230 | EXPAT(SetProcessingInstructionHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3231 | self_xp->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3232 | (XML_ProcessingInstructionHandler) expat_pi_handler |
| 3233 | ); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3234 | EXPAT(SetStartDoctypeDeclHandler)( |
| 3235 | self_xp->parser, |
| 3236 | (XML_StartDoctypeDeclHandler) expat_start_doctype_handler |
| 3237 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3238 | EXPAT(SetUnknownEncodingHandler)( |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3239 | self_xp->parser, |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 3240 | EXPAT(DefaultUnknownEncodingHandler), NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3241 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3242 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3243 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3244 | } |
| 3245 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3246 | static int |
| 3247 | xmlparser_gc_traverse(XMLParserObject *self, visitproc visit, void *arg) |
| 3248 | { |
| 3249 | Py_VISIT(self->handle_close); |
| 3250 | Py_VISIT(self->handle_pi); |
| 3251 | Py_VISIT(self->handle_comment); |
| 3252 | Py_VISIT(self->handle_end); |
| 3253 | Py_VISIT(self->handle_data); |
| 3254 | Py_VISIT(self->handle_start); |
| 3255 | |
| 3256 | Py_VISIT(self->target); |
| 3257 | Py_VISIT(self->entity); |
| 3258 | Py_VISIT(self->names); |
| 3259 | |
| 3260 | return 0; |
| 3261 | } |
| 3262 | |
| 3263 | static int |
| 3264 | xmlparser_gc_clear(XMLParserObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3265 | { |
| 3266 | EXPAT(ParserFree)(self->parser); |
| 3267 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3268 | Py_CLEAR(self->handle_close); |
| 3269 | Py_CLEAR(self->handle_pi); |
| 3270 | Py_CLEAR(self->handle_comment); |
| 3271 | Py_CLEAR(self->handle_end); |
| 3272 | Py_CLEAR(self->handle_data); |
| 3273 | Py_CLEAR(self->handle_start); |
| 3274 | Py_CLEAR(self->handle_doctype); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3275 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3276 | Py_CLEAR(self->target); |
| 3277 | Py_CLEAR(self->entity); |
| 3278 | Py_CLEAR(self->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3279 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3280 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3281 | } |
| 3282 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3283 | static void |
| 3284 | xmlparser_dealloc(XMLParserObject* self) |
| 3285 | { |
| 3286 | PyObject_GC_UnTrack(self); |
| 3287 | xmlparser_gc_clear(self); |
| 3288 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 3289 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3290 | |
| 3291 | LOCAL(PyObject*) |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3292 | expat_parse(XMLParserObject* self, const char* data, int data_len, int final) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3293 | { |
| 3294 | int ok; |
| 3295 | |
| 3296 | ok = EXPAT(Parse)(self->parser, data, data_len, final); |
| 3297 | |
| 3298 | if (PyErr_Occurred()) |
| 3299 | return NULL; |
| 3300 | |
| 3301 | if (!ok) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3302 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 3303 | EXPAT(GetErrorCode)(self->parser), |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3304 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 3305 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 3306 | NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3307 | ); |
| 3308 | return NULL; |
| 3309 | } |
| 3310 | |
| 3311 | Py_RETURN_NONE; |
| 3312 | } |
| 3313 | |
| 3314 | static PyObject* |
| 3315 | xmlparser_close(XMLParserObject* self, PyObject* args) |
| 3316 | { |
| 3317 | /* end feeding data to parser */ |
| 3318 | |
| 3319 | PyObject* res; |
| 3320 | if (!PyArg_ParseTuple(args, ":close")) |
| 3321 | return NULL; |
| 3322 | |
| 3323 | res = expat_parse(self, "", 0, 1); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3324 | if (!res) |
| 3325 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3326 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3327 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3328 | Py_DECREF(res); |
| 3329 | return treebuilder_done((TreeBuilderObject*) self->target); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3330 | } if (self->handle_close) { |
| 3331 | Py_DECREF(res); |
| 3332 | return PyObject_CallFunction(self->handle_close, ""); |
| 3333 | } else |
| 3334 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3335 | } |
| 3336 | |
| 3337 | static PyObject* |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3338 | xmlparser_feed(XMLParserObject* self, PyObject* arg) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3339 | { |
| 3340 | /* feed data to parser */ |
| 3341 | |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3342 | if (PyUnicode_Check(arg)) { |
| 3343 | Py_ssize_t data_len; |
| 3344 | const char *data = PyUnicode_AsUTF8AndSize(arg, &data_len); |
| 3345 | if (data == NULL) |
| 3346 | return NULL; |
| 3347 | if (data_len > INT_MAX) { |
| 3348 | PyErr_SetString(PyExc_OverflowError, "size does not fit in an int"); |
| 3349 | return NULL; |
| 3350 | } |
| 3351 | /* Explicitly set UTF-8 encoding. Return code ignored. */ |
| 3352 | (void)EXPAT(SetEncoding)(self->parser, "utf-8"); |
| 3353 | return expat_parse(self, data, (int)data_len, 0); |
| 3354 | } |
| 3355 | else { |
| 3356 | Py_buffer view; |
| 3357 | PyObject *res; |
| 3358 | if (PyObject_GetBuffer(arg, &view, PyBUF_SIMPLE) < 0) |
| 3359 | return NULL; |
| 3360 | if (view.len > INT_MAX) { |
| 3361 | PyBuffer_Release(&view); |
| 3362 | PyErr_SetString(PyExc_OverflowError, "size does not fit in an int"); |
| 3363 | return NULL; |
| 3364 | } |
| 3365 | res = expat_parse(self, view.buf, (int)view.len, 0); |
| 3366 | PyBuffer_Release(&view); |
| 3367 | return res; |
| 3368 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3369 | } |
| 3370 | |
| 3371 | static PyObject* |
| 3372 | xmlparser_parse(XMLParserObject* self, PyObject* args) |
| 3373 | { |
| 3374 | /* (internal) parse until end of input stream */ |
| 3375 | |
| 3376 | PyObject* reader; |
| 3377 | PyObject* buffer; |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3378 | PyObject* temp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3379 | PyObject* res; |
| 3380 | |
| 3381 | PyObject* fileobj; |
| 3382 | if (!PyArg_ParseTuple(args, "O:_parse", &fileobj)) |
| 3383 | return NULL; |
| 3384 | |
| 3385 | reader = PyObject_GetAttrString(fileobj, "read"); |
| 3386 | if (!reader) |
| 3387 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3388 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3389 | /* read from open file object */ |
| 3390 | for (;;) { |
| 3391 | |
| 3392 | buffer = PyObject_CallFunction(reader, "i", 64*1024); |
| 3393 | |
| 3394 | if (!buffer) { |
| 3395 | /* read failed (e.g. due to KeyboardInterrupt) */ |
| 3396 | Py_DECREF(reader); |
| 3397 | return NULL; |
| 3398 | } |
| 3399 | |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3400 | if (PyUnicode_CheckExact(buffer)) { |
| 3401 | /* A unicode object is encoded into bytes using UTF-8 */ |
| 3402 | if (PyUnicode_GET_SIZE(buffer) == 0) { |
| 3403 | Py_DECREF(buffer); |
| 3404 | break; |
| 3405 | } |
| 3406 | temp = PyUnicode_AsEncodedString(buffer, "utf-8", "surrogatepass"); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3407 | Py_DECREF(buffer); |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3408 | if (!temp) { |
| 3409 | /* Propagate exception from PyUnicode_AsEncodedString */ |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3410 | Py_DECREF(reader); |
| 3411 | return NULL; |
| 3412 | } |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3413 | buffer = temp; |
| 3414 | } |
| 3415 | else if (!PyBytes_CheckExact(buffer) || PyBytes_GET_SIZE(buffer) == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3416 | Py_DECREF(buffer); |
| 3417 | break; |
| 3418 | } |
| 3419 | |
| 3420 | res = expat_parse( |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 3421 | self, PyBytes_AS_STRING(buffer), PyBytes_GET_SIZE(buffer), 0 |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3422 | ); |
| 3423 | |
| 3424 | Py_DECREF(buffer); |
| 3425 | |
| 3426 | if (!res) { |
| 3427 | Py_DECREF(reader); |
| 3428 | return NULL; |
| 3429 | } |
| 3430 | Py_DECREF(res); |
| 3431 | |
| 3432 | } |
| 3433 | |
| 3434 | Py_DECREF(reader); |
| 3435 | |
| 3436 | res = expat_parse(self, "", 0, 1); |
| 3437 | |
| 3438 | if (res && TreeBuilder_CheckExact(self->target)) { |
| 3439 | Py_DECREF(res); |
| 3440 | return treebuilder_done((TreeBuilderObject*) self->target); |
| 3441 | } |
| 3442 | |
| 3443 | return res; |
| 3444 | } |
| 3445 | |
| 3446 | static PyObject* |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3447 | xmlparser_doctype(XMLParserObject *self, PyObject *args) |
| 3448 | { |
| 3449 | Py_RETURN_NONE; |
| 3450 | } |
| 3451 | |
| 3452 | static PyObject* |
| 3453 | xmlparser_setevents(XMLParserObject *self, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3454 | { |
| 3455 | /* activate element event reporting */ |
| 3456 | |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 3457 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3458 | TreeBuilderObject* target; |
| 3459 | |
| 3460 | PyObject* events; /* event collector */ |
| 3461 | PyObject* event_set = Py_None; |
| 3462 | if (!PyArg_ParseTuple(args, "O!|O:_setevents", &PyList_Type, &events, |
| 3463 | &event_set)) |
| 3464 | return NULL; |
| 3465 | |
| 3466 | if (!TreeBuilder_CheckExact(self->target)) { |
| 3467 | PyErr_SetString( |
| 3468 | PyExc_TypeError, |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 3469 | "event handling only supported for ElementTree.TreeBuilder " |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3470 | "targets" |
| 3471 | ); |
| 3472 | return NULL; |
| 3473 | } |
| 3474 | |
| 3475 | target = (TreeBuilderObject*) self->target; |
| 3476 | |
| 3477 | Py_INCREF(events); |
| 3478 | Py_XDECREF(target->events); |
| 3479 | target->events = events; |
| 3480 | |
| 3481 | /* clear out existing events */ |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3482 | Py_CLEAR(target->start_event_obj); |
| 3483 | Py_CLEAR(target->end_event_obj); |
| 3484 | Py_CLEAR(target->start_ns_event_obj); |
| 3485 | Py_CLEAR(target->end_ns_event_obj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3486 | |
| 3487 | if (event_set == Py_None) { |
| 3488 | /* default is "end" only */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3489 | target->end_event_obj = PyUnicode_FromString("end"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3490 | Py_RETURN_NONE; |
| 3491 | } |
| 3492 | |
| 3493 | if (!PyTuple_Check(event_set)) /* FIXME: handle arbitrary sequences */ |
| 3494 | goto error; |
| 3495 | |
| 3496 | for (i = 0; i < PyTuple_GET_SIZE(event_set); i++) { |
| 3497 | PyObject* item = PyTuple_GET_ITEM(event_set, i); |
| 3498 | char* event; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3499 | if (PyUnicode_Check(item)) { |
| 3500 | event = _PyUnicode_AsString(item); |
Victor Stinner | 0477bf3 | 2010-03-22 12:11:44 +0000 | [diff] [blame] | 3501 | if (event == NULL) |
| 3502 | goto error; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3503 | } else if (PyBytes_Check(item)) |
| 3504 | event = PyBytes_AS_STRING(item); |
| 3505 | else { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3506 | goto error; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3507 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3508 | if (strcmp(event, "start") == 0) { |
| 3509 | Py_INCREF(item); |
| 3510 | target->start_event_obj = item; |
| 3511 | } else if (strcmp(event, "end") == 0) { |
| 3512 | Py_INCREF(item); |
| 3513 | Py_XDECREF(target->end_event_obj); |
| 3514 | target->end_event_obj = item; |
| 3515 | } else if (strcmp(event, "start-ns") == 0) { |
| 3516 | Py_INCREF(item); |
| 3517 | Py_XDECREF(target->start_ns_event_obj); |
| 3518 | target->start_ns_event_obj = item; |
| 3519 | EXPAT(SetNamespaceDeclHandler)( |
| 3520 | self->parser, |
| 3521 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3522 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3523 | ); |
| 3524 | } else if (strcmp(event, "end-ns") == 0) { |
| 3525 | Py_INCREF(item); |
| 3526 | Py_XDECREF(target->end_ns_event_obj); |
| 3527 | target->end_ns_event_obj = item; |
| 3528 | EXPAT(SetNamespaceDeclHandler)( |
| 3529 | self->parser, |
| 3530 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3531 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3532 | ); |
| 3533 | } else { |
| 3534 | PyErr_Format( |
| 3535 | PyExc_ValueError, |
| 3536 | "unknown event '%s'", event |
| 3537 | ); |
| 3538 | return NULL; |
| 3539 | } |
| 3540 | } |
| 3541 | |
| 3542 | Py_RETURN_NONE; |
| 3543 | |
| 3544 | error: |
| 3545 | PyErr_SetString( |
| 3546 | PyExc_TypeError, |
| 3547 | "invalid event tuple" |
| 3548 | ); |
| 3549 | return NULL; |
| 3550 | } |
| 3551 | |
| 3552 | static PyMethodDef xmlparser_methods[] = { |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3553 | {"feed", (PyCFunction) xmlparser_feed, METH_O}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3554 | {"close", (PyCFunction) xmlparser_close, METH_VARARGS}, |
| 3555 | {"_parse", (PyCFunction) xmlparser_parse, METH_VARARGS}, |
| 3556 | {"_setevents", (PyCFunction) xmlparser_setevents, METH_VARARGS}, |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3557 | {"doctype", (PyCFunction) xmlparser_doctype, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3558 | {NULL, NULL} |
| 3559 | }; |
| 3560 | |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3561 | static PyObject* |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3562 | xmlparser_getattro(XMLParserObject* self, PyObject* nameobj) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3563 | { |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3564 | if (PyUnicode_Check(nameobj)) { |
| 3565 | PyObject* res; |
| 3566 | if (PyUnicode_CompareWithASCIIString(nameobj, "entity") == 0) |
| 3567 | res = self->entity; |
| 3568 | else if (PyUnicode_CompareWithASCIIString(nameobj, "target") == 0) |
| 3569 | res = self->target; |
| 3570 | else if (PyUnicode_CompareWithASCIIString(nameobj, "version") == 0) { |
| 3571 | return PyUnicode_FromFormat( |
| 3572 | "Expat %d.%d.%d", XML_MAJOR_VERSION, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3573 | XML_MINOR_VERSION, XML_MICRO_VERSION); |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3574 | } |
| 3575 | else |
| 3576 | goto generic; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3577 | |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3578 | Py_INCREF(res); |
| 3579 | return res; |
| 3580 | } |
| 3581 | generic: |
| 3582 | return PyObject_GenericGetAttr((PyObject*) self, nameobj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3583 | } |
| 3584 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 3585 | static PyTypeObject XMLParser_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 3586 | PyVarObject_HEAD_INIT(NULL, 0) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 3587 | "xml.etree.ElementTree.XMLParser", sizeof(XMLParserObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3588 | /* methods */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3589 | (destructor)xmlparser_dealloc, /* tp_dealloc */ |
| 3590 | 0, /* tp_print */ |
| 3591 | 0, /* tp_getattr */ |
| 3592 | 0, /* tp_setattr */ |
| 3593 | 0, /* tp_reserved */ |
| 3594 | 0, /* tp_repr */ |
| 3595 | 0, /* tp_as_number */ |
| 3596 | 0, /* tp_as_sequence */ |
| 3597 | 0, /* tp_as_mapping */ |
| 3598 | 0, /* tp_hash */ |
| 3599 | 0, /* tp_call */ |
| 3600 | 0, /* tp_str */ |
| 3601 | (getattrofunc)xmlparser_getattro, /* tp_getattro */ |
| 3602 | 0, /* tp_setattro */ |
| 3603 | 0, /* tp_as_buffer */ |
| 3604 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 3605 | /* tp_flags */ |
| 3606 | 0, /* tp_doc */ |
| 3607 | (traverseproc)xmlparser_gc_traverse, /* tp_traverse */ |
| 3608 | (inquiry)xmlparser_gc_clear, /* tp_clear */ |
| 3609 | 0, /* tp_richcompare */ |
| 3610 | 0, /* tp_weaklistoffset */ |
| 3611 | 0, /* tp_iter */ |
| 3612 | 0, /* tp_iternext */ |
| 3613 | xmlparser_methods, /* tp_methods */ |
| 3614 | 0, /* tp_members */ |
| 3615 | 0, /* tp_getset */ |
| 3616 | 0, /* tp_base */ |
| 3617 | 0, /* tp_dict */ |
| 3618 | 0, /* tp_descr_get */ |
| 3619 | 0, /* tp_descr_set */ |
| 3620 | 0, /* tp_dictoffset */ |
| 3621 | (initproc)xmlparser_init, /* tp_init */ |
| 3622 | PyType_GenericAlloc, /* tp_alloc */ |
| 3623 | xmlparser_new, /* tp_new */ |
| 3624 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3625 | }; |
| 3626 | |
| 3627 | #endif |
| 3628 | |
| 3629 | /* ==================================================================== */ |
| 3630 | /* python module interface */ |
| 3631 | |
| 3632 | static PyMethodDef _functions[] = { |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 3633 | {"SubElement", (PyCFunction) subelement, METH_VARARGS | METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3634 | {NULL, NULL} |
| 3635 | }; |
| 3636 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3637 | |
| 3638 | static struct PyModuleDef _elementtreemodule = { |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3639 | PyModuleDef_HEAD_INIT, |
| 3640 | "_elementtree", |
| 3641 | NULL, |
| 3642 | -1, |
| 3643 | _functions, |
| 3644 | NULL, |
| 3645 | NULL, |
| 3646 | NULL, |
| 3647 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3648 | }; |
| 3649 | |
Neal Norwitz | f6657e6 | 2006-12-28 04:47:50 +0000 | [diff] [blame] | 3650 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3651 | PyInit__elementtree(void) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3652 | { |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 3653 | PyObject *m, *temp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3654 | |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3655 | /* Initialize object types */ |
| 3656 | if (PyType_Ready(&TreeBuilder_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3657 | return NULL; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3658 | if (PyType_Ready(&Element_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3659 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3660 | #if defined(USE_EXPAT) |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3661 | if (PyType_Ready(&XMLParser_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 3662 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3663 | #endif |
| 3664 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3665 | m = PyModule_Create(&_elementtreemodule); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 3666 | if (!m) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3667 | return NULL; |
| 3668 | |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 3669 | if (!(temp = PyImport_ImportModule("copy"))) |
| 3670 | return NULL; |
| 3671 | elementtree_deepcopy_obj = PyObject_GetAttrString(temp, "deepcopy"); |
| 3672 | Py_XDECREF(temp); |
| 3673 | |
| 3674 | if (!(elementpath_obj = PyImport_ImportModule("xml.etree.ElementPath"))) |
| 3675 | return NULL; |
| 3676 | |
Eli Bendersky | 20d4174 | 2012-06-01 09:48:37 +0300 | [diff] [blame] | 3677 | /* link against pyexpat */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3678 | expat_capi = PyCapsule_Import(PyExpat_CAPSULE_NAME, 0); |
| 3679 | if (expat_capi) { |
| 3680 | /* check that it's usable */ |
| 3681 | if (strcmp(expat_capi->magic, PyExpat_CAPI_MAGIC) != 0 || |
| 3682 | expat_capi->size < sizeof(struct PyExpat_CAPI) || |
| 3683 | expat_capi->MAJOR_VERSION != XML_MAJOR_VERSION || |
| 3684 | expat_capi->MINOR_VERSION != XML_MINOR_VERSION || |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3685 | expat_capi->MICRO_VERSION != XML_MICRO_VERSION) { |
Eli Bendersky | ef391ac | 2012-07-21 20:28:46 +0300 | [diff] [blame] | 3686 | PyErr_SetString(PyExc_ImportError, |
| 3687 | "pyexpat version is incompatible"); |
| 3688 | return NULL; |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3689 | } |
Eli Bendersky | ef391ac | 2012-07-21 20:28:46 +0300 | [diff] [blame] | 3690 | } else { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3691 | return NULL; |
Eli Bendersky | ef391ac | 2012-07-21 20:28:46 +0300 | [diff] [blame] | 3692 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3693 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3694 | elementtree_parseerror_obj = PyErr_NewException( |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 3695 | "xml.etree.ElementTree.ParseError", PyExc_SyntaxError, NULL |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3696 | ); |
| 3697 | Py_INCREF(elementtree_parseerror_obj); |
| 3698 | PyModule_AddObject(m, "ParseError", elementtree_parseerror_obj); |
| 3699 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 3700 | Py_INCREF((PyObject *)&Element_Type); |
| 3701 | PyModule_AddObject(m, "Element", (PyObject *)&Element_Type); |
| 3702 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 3703 | Py_INCREF((PyObject *)&TreeBuilder_Type); |
| 3704 | PyModule_AddObject(m, "TreeBuilder", (PyObject *)&TreeBuilder_Type); |
| 3705 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3706 | #if defined(USE_EXPAT) |
| 3707 | Py_INCREF((PyObject *)&XMLParser_Type); |
| 3708 | PyModule_AddObject(m, "XMLParser", (PyObject *)&XMLParser_Type); |
| 3709 | #endif |
| 3710 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3711 | return m; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3712 | } |