Eli Bendersky | bf05df2 | 2013-04-20 05:44:01 -0700 | [diff] [blame] | 1 | /*-------------------------------------------------------------------- |
| 2 | * Licensed to PSF under a Contributor Agreement. |
| 3 | * See http://www.python.org/psf/license for licensing details. |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 4 | * |
Eli Bendersky | bf05df2 | 2013-04-20 05:44:01 -0700 | [diff] [blame] | 5 | * _elementtree - C accelerator for xml.etree.ElementTree |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 6 | * Copyright (c) 1999-2009 by Secret Labs AB. All rights reserved. |
| 7 | * Copyright (c) 1999-2009 by Fredrik Lundh. |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 8 | * |
| 9 | * info@pythonware.com |
| 10 | * http://www.pythonware.com |
Eli Bendersky | bf05df2 | 2013-04-20 05:44:01 -0700 | [diff] [blame] | 11 | *-------------------------------------------------------------------- |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 12 | */ |
| 13 | |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 14 | #define PY_SSIZE_T_CLEAN |
| 15 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 16 | #include "Python.h" |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 17 | #include "structmember.h" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 18 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 19 | /* -------------------------------------------------------------------- */ |
| 20 | /* configuration */ |
| 21 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 22 | /* An element can hold this many children without extra memory |
| 23 | allocations. */ |
| 24 | #define STATIC_CHILDREN 4 |
| 25 | |
| 26 | /* For best performance, chose a value so that 80-90% of all nodes |
| 27 | have no more than the given number of children. Set this to zero |
| 28 | to minimize the size of the element structure itself (this only |
| 29 | helps if you have lots of leaf nodes with attributes). */ |
| 30 | |
| 31 | /* Also note that pymalloc always allocates blocks in multiples of |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 32 | eight bytes. For the current C version of ElementTree, this means |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 33 | that the number of children should be an even number, at least on |
| 34 | 32-bit platforms. */ |
| 35 | |
| 36 | /* -------------------------------------------------------------------- */ |
| 37 | |
| 38 | #if 0 |
| 39 | static int memory = 0; |
| 40 | #define ALLOC(size, comment)\ |
| 41 | do { memory += size; printf("%8d - %s\n", memory, comment); } while (0) |
| 42 | #define RELEASE(size, comment)\ |
| 43 | do { memory -= size; printf("%8d - %s\n", memory, comment); } while (0) |
| 44 | #else |
| 45 | #define ALLOC(size, comment) |
| 46 | #define RELEASE(size, comment) |
| 47 | #endif |
| 48 | |
| 49 | /* compiler tweaks */ |
| 50 | #if defined(_MSC_VER) |
| 51 | #define LOCAL(type) static __inline type __fastcall |
| 52 | #else |
| 53 | #define LOCAL(type) static type |
| 54 | #endif |
| 55 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 56 | /* macros used to store 'join' flags in string object pointers. note |
| 57 | that all use of text and tail as object pointers must be wrapped in |
| 58 | JOIN_OBJ. see comments in the ElementObject definition for more |
| 59 | info. */ |
Benjamin Peterson | ca47063 | 2016-09-06 13:47:26 -0700 | [diff] [blame] | 60 | #define JOIN_GET(p) ((uintptr_t) (p) & 1) |
| 61 | #define JOIN_SET(p, flag) ((void*) ((uintptr_t) (JOIN_OBJ(p)) | (flag))) |
| 62 | #define JOIN_OBJ(p) ((PyObject*) ((uintptr_t) (p) & ~(uintptr_t)1)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 63 | |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 64 | /* Py_SETREF for a PyObject* that uses a join flag. */ |
| 65 | Py_LOCAL_INLINE(void) |
| 66 | _set_joined_ptr(PyObject **p, PyObject *new_joined_ptr) |
| 67 | { |
| 68 | PyObject *tmp = JOIN_OBJ(*p); |
| 69 | *p = new_joined_ptr; |
| 70 | Py_DECREF(tmp); |
| 71 | } |
| 72 | |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 73 | /* Py_CLEAR for a PyObject* that uses a join flag. Pass the pointer by |
| 74 | * reference since this function sets it to NULL. |
| 75 | */ |
doko@ubuntu.com | 0648bf7 | 2013-09-18 12:12:28 +0200 | [diff] [blame] | 76 | static void _clear_joined_ptr(PyObject **p) |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 77 | { |
| 78 | if (*p) { |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 79 | _set_joined_ptr(p, NULL); |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 80 | } |
| 81 | } |
| 82 | |
Ronald Oussoren | 138d080 | 2013-07-19 11:11:25 +0200 | [diff] [blame] | 83 | /* Types defined by this extension */ |
| 84 | static PyTypeObject Element_Type; |
| 85 | static PyTypeObject ElementIter_Type; |
| 86 | static PyTypeObject TreeBuilder_Type; |
| 87 | static PyTypeObject XMLParser_Type; |
| 88 | |
| 89 | |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 90 | /* Per-module state; PEP 3121 */ |
| 91 | typedef struct { |
| 92 | PyObject *parseerror_obj; |
| 93 | PyObject *deepcopy_obj; |
| 94 | PyObject *elementpath_obj; |
| 95 | } elementtreestate; |
| 96 | |
| 97 | static struct PyModuleDef elementtreemodule; |
| 98 | |
| 99 | /* Given a module object (assumed to be _elementtree), get its per-module |
| 100 | * state. |
| 101 | */ |
| 102 | #define ET_STATE(mod) ((elementtreestate *) PyModule_GetState(mod)) |
| 103 | |
| 104 | /* Find the module instance imported in the currently running sub-interpreter |
| 105 | * and get its state. |
| 106 | */ |
| 107 | #define ET_STATE_GLOBAL \ |
| 108 | ((elementtreestate *) PyModule_GetState(PyState_FindModule(&elementtreemodule))) |
| 109 | |
| 110 | static int |
| 111 | elementtree_clear(PyObject *m) |
| 112 | { |
| 113 | elementtreestate *st = ET_STATE(m); |
| 114 | Py_CLEAR(st->parseerror_obj); |
| 115 | Py_CLEAR(st->deepcopy_obj); |
| 116 | Py_CLEAR(st->elementpath_obj); |
| 117 | return 0; |
| 118 | } |
| 119 | |
| 120 | static int |
| 121 | elementtree_traverse(PyObject *m, visitproc visit, void *arg) |
| 122 | { |
| 123 | elementtreestate *st = ET_STATE(m); |
| 124 | Py_VISIT(st->parseerror_obj); |
| 125 | Py_VISIT(st->deepcopy_obj); |
| 126 | Py_VISIT(st->elementpath_obj); |
| 127 | return 0; |
| 128 | } |
| 129 | |
| 130 | static void |
| 131 | elementtree_free(void *m) |
| 132 | { |
| 133 | elementtree_clear((PyObject *)m); |
| 134 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 135 | |
| 136 | /* helpers */ |
| 137 | |
| 138 | LOCAL(PyObject*) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 139 | list_join(PyObject* list) |
| 140 | { |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 141 | /* join list elements */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 142 | PyObject* joiner; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 143 | PyObject* result; |
| 144 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 145 | joiner = PyUnicode_FromStringAndSize("", 0); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 146 | if (!joiner) |
| 147 | return NULL; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 148 | result = PyUnicode_Join(joiner, list); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 149 | Py_DECREF(joiner); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 150 | return result; |
| 151 | } |
| 152 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 153 | /* Is the given object an empty dictionary? |
| 154 | */ |
| 155 | static int |
| 156 | is_empty_dict(PyObject *obj) |
| 157 | { |
Serhiy Storchaka | 5ab81d7 | 2016-12-16 16:18:57 +0200 | [diff] [blame] | 158 | return PyDict_CheckExact(obj) && PyDict_GET_SIZE(obj) == 0; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 159 | } |
| 160 | |
| 161 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 162 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 163 | /* the Element type */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 164 | |
| 165 | typedef struct { |
| 166 | |
| 167 | /* attributes (a dictionary object), or None if no attributes */ |
| 168 | PyObject* attrib; |
| 169 | |
| 170 | /* child elements */ |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 171 | Py_ssize_t length; /* actual number of items */ |
| 172 | Py_ssize_t allocated; /* allocated items */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 173 | |
| 174 | /* this either points to _children or to a malloced buffer */ |
| 175 | PyObject* *children; |
| 176 | |
| 177 | PyObject* _children[STATIC_CHILDREN]; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 178 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 179 | } ElementObjectExtra; |
| 180 | |
| 181 | typedef struct { |
| 182 | PyObject_HEAD |
| 183 | |
| 184 | /* element tag (a string). */ |
| 185 | PyObject* tag; |
| 186 | |
| 187 | /* text before first child. note that this is a tagged pointer; |
| 188 | use JOIN_OBJ to get the object pointer. the join flag is used |
| 189 | to distinguish lists created by the tree builder from lists |
| 190 | assigned to the attribute by application code; the former |
| 191 | should be joined before being returned to the user, the latter |
| 192 | should be left intact. */ |
| 193 | PyObject* text; |
| 194 | |
| 195 | /* text after this element, in parent. note that this is a tagged |
| 196 | pointer; use JOIN_OBJ to get the object pointer. */ |
| 197 | PyObject* tail; |
| 198 | |
| 199 | ElementObjectExtra* extra; |
| 200 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 201 | PyObject *weakreflist; /* For tp_weaklistoffset */ |
| 202 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 203 | } ElementObject; |
| 204 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 205 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 206 | #define Element_CheckExact(op) (Py_TYPE(op) == &Element_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 207 | |
| 208 | /* -------------------------------------------------------------------- */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 209 | /* Element constructors and destructor */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 210 | |
| 211 | LOCAL(int) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 212 | create_extra(ElementObject* self, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 213 | { |
| 214 | self->extra = PyObject_Malloc(sizeof(ElementObjectExtra)); |
Victor Stinner | 81aac73 | 2013-07-12 02:03:34 +0200 | [diff] [blame] | 215 | if (!self->extra) { |
| 216 | PyErr_NoMemory(); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 217 | return -1; |
Victor Stinner | 81aac73 | 2013-07-12 02:03:34 +0200 | [diff] [blame] | 218 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 219 | |
| 220 | if (!attrib) |
| 221 | attrib = Py_None; |
| 222 | |
| 223 | Py_INCREF(attrib); |
| 224 | self->extra->attrib = attrib; |
| 225 | |
| 226 | self->extra->length = 0; |
| 227 | self->extra->allocated = STATIC_CHILDREN; |
| 228 | self->extra->children = self->extra->_children; |
| 229 | |
| 230 | return 0; |
| 231 | } |
| 232 | |
| 233 | LOCAL(void) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 234 | dealloc_extra(ElementObject* self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 235 | { |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 236 | ElementObjectExtra *myextra; |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 237 | Py_ssize_t i; |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 238 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 239 | if (!self->extra) |
| 240 | return; |
| 241 | |
| 242 | /* Avoid DECREFs calling into this code again (cycles, etc.) |
| 243 | */ |
Eli Bendersky | 08b8529 | 2012-04-04 15:55:07 +0300 | [diff] [blame] | 244 | myextra = self->extra; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 245 | self->extra = NULL; |
| 246 | |
| 247 | Py_DECREF(myextra->attrib); |
| 248 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 249 | for (i = 0; i < myextra->length; i++) |
| 250 | Py_DECREF(myextra->children[i]); |
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 | if (myextra->children != myextra->_children) |
| 253 | PyObject_Free(myextra->children); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 254 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 255 | PyObject_Free(myextra); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 256 | } |
| 257 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 258 | /* Convenience internal function to create new Element objects with the given |
| 259 | * tag and attributes. |
| 260 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 261 | LOCAL(PyObject*) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 262 | create_new_element(PyObject* tag, PyObject* attrib) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 263 | { |
| 264 | ElementObject* self; |
| 265 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 266 | self = PyObject_GC_New(ElementObject, &Element_Type); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 267 | if (self == NULL) |
| 268 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 269 | self->extra = NULL; |
| 270 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 271 | Py_INCREF(tag); |
| 272 | self->tag = tag; |
| 273 | |
| 274 | Py_INCREF(Py_None); |
| 275 | self->text = Py_None; |
| 276 | |
| 277 | Py_INCREF(Py_None); |
| 278 | self->tail = Py_None; |
| 279 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 280 | self->weakreflist = NULL; |
| 281 | |
Victor Stinner | d917dcb | 2013-07-12 02:05:17 +0200 | [diff] [blame] | 282 | ALLOC(sizeof(ElementObject), "create element"); |
| 283 | PyObject_GC_Track(self); |
| 284 | |
Victor Stinner | 71c8b7e | 2013-07-11 23:08:39 +0200 | [diff] [blame] | 285 | if (attrib != Py_None && !is_empty_dict(attrib)) { |
| 286 | if (create_extra(self, attrib) < 0) { |
Victor Stinner | d917dcb | 2013-07-12 02:05:17 +0200 | [diff] [blame] | 287 | Py_DECREF(self); |
Victor Stinner | 71c8b7e | 2013-07-11 23:08:39 +0200 | [diff] [blame] | 288 | return NULL; |
| 289 | } |
| 290 | } |
| 291 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 292 | return (PyObject*) self; |
| 293 | } |
| 294 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 295 | static PyObject * |
| 296 | element_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 297 | { |
| 298 | ElementObject *e = (ElementObject *)type->tp_alloc(type, 0); |
| 299 | if (e != NULL) { |
| 300 | Py_INCREF(Py_None); |
| 301 | e->tag = Py_None; |
| 302 | |
| 303 | Py_INCREF(Py_None); |
| 304 | e->text = Py_None; |
| 305 | |
| 306 | Py_INCREF(Py_None); |
| 307 | e->tail = Py_None; |
| 308 | |
| 309 | e->extra = NULL; |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 310 | e->weakreflist = NULL; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 311 | } |
| 312 | return (PyObject *)e; |
| 313 | } |
| 314 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 315 | /* Helper function for extracting the attrib dictionary from a keywords dict. |
| 316 | * This is required by some constructors/functions in this module that can |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 317 | * either accept attrib as a keyword argument or all attributes splashed |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 318 | * directly into *kwds. |
Eli Bendersky | d4cb4b7 | 2013-04-22 05:25:25 -0700 | [diff] [blame] | 319 | * |
| 320 | * Return a dictionary with the content of kwds merged into the content of |
| 321 | * attrib. If there is no attrib keyword, return a copy of kwds. |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 322 | */ |
| 323 | static PyObject* |
| 324 | get_attrib_from_keywords(PyObject *kwds) |
| 325 | { |
Eli Bendersky | 45f3d2f | 2013-04-24 05:34:07 -0700 | [diff] [blame] | 326 | PyObject *attrib_str = PyUnicode_FromString("attrib"); |
| 327 | PyObject *attrib = PyDict_GetItem(kwds, attrib_str); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 328 | |
| 329 | if (attrib) { |
| 330 | /* If attrib was found in kwds, copy its value and remove it from |
| 331 | * kwds |
| 332 | */ |
| 333 | if (!PyDict_Check(attrib)) { |
Eli Bendersky | 45f3d2f | 2013-04-24 05:34:07 -0700 | [diff] [blame] | 334 | Py_DECREF(attrib_str); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 335 | PyErr_Format(PyExc_TypeError, "attrib must be dict, not %.100s", |
| 336 | Py_TYPE(attrib)->tp_name); |
| 337 | return NULL; |
| 338 | } |
| 339 | attrib = PyDict_Copy(attrib); |
Eli Bendersky | 45f3d2f | 2013-04-24 05:34:07 -0700 | [diff] [blame] | 340 | PyDict_DelItem(kwds, attrib_str); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 341 | } else { |
| 342 | attrib = PyDict_New(); |
| 343 | } |
Eli Bendersky | 45f3d2f | 2013-04-24 05:34:07 -0700 | [diff] [blame] | 344 | |
| 345 | Py_DECREF(attrib_str); |
| 346 | |
| 347 | /* attrib can be NULL if PyDict_New failed */ |
| 348 | if (attrib) |
Christian Heimes | 7ed4294 | 2013-07-20 15:12:09 +0200 | [diff] [blame] | 349 | if (PyDict_Update(attrib, kwds) < 0) |
| 350 | return NULL; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 351 | return attrib; |
| 352 | } |
| 353 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 354 | /*[clinic input] |
| 355 | module _elementtree |
| 356 | class _elementtree.Element "ElementObject *" "&Element_Type" |
| 357 | class _elementtree.TreeBuilder "TreeBuilderObject *" "&TreeBuilder_Type" |
| 358 | class _elementtree.XMLParser "XMLParserObject *" "&XMLParser_Type" |
| 359 | [clinic start generated code]*/ |
| 360 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=159aa50a54061c22]*/ |
| 361 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 362 | static int |
| 363 | element_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 364 | { |
| 365 | PyObject *tag; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 366 | PyObject *attrib = NULL; |
| 367 | ElementObject *self_elem; |
| 368 | |
| 369 | if (!PyArg_ParseTuple(args, "O|O!:Element", &tag, &PyDict_Type, &attrib)) |
| 370 | return -1; |
| 371 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 372 | if (attrib) { |
| 373 | /* attrib passed as positional arg */ |
| 374 | attrib = PyDict_Copy(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 375 | if (!attrib) |
| 376 | return -1; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 377 | if (kwds) { |
| 378 | if (PyDict_Update(attrib, kwds) < 0) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 379 | Py_DECREF(attrib); |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 380 | return -1; |
| 381 | } |
| 382 | } |
| 383 | } else if (kwds) { |
| 384 | /* have keywords args */ |
| 385 | attrib = get_attrib_from_keywords(kwds); |
| 386 | if (!attrib) |
| 387 | return -1; |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 388 | } |
| 389 | |
| 390 | self_elem = (ElementObject *)self; |
| 391 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 392 | if (attrib != NULL && !is_empty_dict(attrib)) { |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 393 | if (create_extra(self_elem, attrib) < 0) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 394 | Py_DECREF(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 395 | return -1; |
| 396 | } |
| 397 | } |
| 398 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 399 | /* 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] | 400 | Py_XDECREF(attrib); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 401 | |
| 402 | /* Replace the objects already pointed to by tag, text and tail. */ |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 403 | Py_INCREF(tag); |
Serhiy Storchaka | ec39756 | 2016-04-06 09:50:03 +0300 | [diff] [blame] | 404 | Py_XSETREF(self_elem->tag, tag); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 405 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 406 | Py_INCREF(Py_None); |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 407 | _set_joined_ptr(&self_elem->text, Py_None); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 408 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 409 | Py_INCREF(Py_None); |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 410 | _set_joined_ptr(&self_elem->tail, Py_None); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 411 | |
| 412 | return 0; |
| 413 | } |
| 414 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 415 | LOCAL(int) |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 416 | element_resize(ElementObject* self, Py_ssize_t extra) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 417 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 418 | Py_ssize_t size; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 419 | PyObject* *children; |
| 420 | |
| 421 | /* make sure self->children can hold the given number of extra |
| 422 | elements. set an exception and return -1 if allocation failed */ |
| 423 | |
Victor Stinner | 5f0af23 | 2013-07-11 23:01:36 +0200 | [diff] [blame] | 424 | if (!self->extra) { |
| 425 | if (create_extra(self, NULL) < 0) |
| 426 | return -1; |
| 427 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 428 | |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 429 | size = self->extra->length + extra; /* never overflows */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 430 | |
| 431 | if (size > self->extra->allocated) { |
| 432 | /* use Python 2.4's list growth strategy */ |
| 433 | size = (size >> 3) + (size < 9 ? 3 : 6) + size; |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 434 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer "children" |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 435 | * which needs at least 4 bytes. |
| 436 | * 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] | 437 | * be safe. |
| 438 | */ |
| 439 | size = size ? size : 1; |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 440 | if ((size_t)size > PY_SSIZE_T_MAX/sizeof(PyObject*)) |
| 441 | goto nomemory; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 442 | if (self->extra->children != self->extra->_children) { |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 443 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 444 | * "children", which needs at least 4 bytes. Although it's a |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 445 | * false alarm always assume at least one child to be safe. |
| 446 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 447 | children = PyObject_Realloc(self->extra->children, |
| 448 | size * sizeof(PyObject*)); |
| 449 | if (!children) |
| 450 | goto nomemory; |
| 451 | } else { |
| 452 | children = PyObject_Malloc(size * sizeof(PyObject*)); |
| 453 | if (!children) |
| 454 | goto nomemory; |
| 455 | /* copy existing children from static area to malloc buffer */ |
| 456 | memcpy(children, self->extra->children, |
| 457 | self->extra->length * sizeof(PyObject*)); |
| 458 | } |
| 459 | self->extra->children = children; |
| 460 | self->extra->allocated = size; |
| 461 | } |
| 462 | |
| 463 | return 0; |
| 464 | |
| 465 | nomemory: |
| 466 | PyErr_NoMemory(); |
| 467 | return -1; |
| 468 | } |
| 469 | |
| 470 | LOCAL(int) |
| 471 | element_add_subelement(ElementObject* self, PyObject* element) |
| 472 | { |
| 473 | /* add a child element to a parent */ |
| 474 | |
| 475 | if (element_resize(self, 1) < 0) |
| 476 | return -1; |
| 477 | |
| 478 | Py_INCREF(element); |
| 479 | self->extra->children[self->extra->length] = element; |
| 480 | |
| 481 | self->extra->length++; |
| 482 | |
| 483 | return 0; |
| 484 | } |
| 485 | |
| 486 | LOCAL(PyObject*) |
| 487 | element_get_attrib(ElementObject* self) |
| 488 | { |
| 489 | /* return borrowed reference to attrib dictionary */ |
| 490 | /* note: this function assumes that the extra section exists */ |
| 491 | |
| 492 | PyObject* res = self->extra->attrib; |
| 493 | |
| 494 | if (res == Py_None) { |
| 495 | /* create missing dictionary */ |
| 496 | res = PyDict_New(); |
| 497 | if (!res) |
| 498 | return NULL; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 499 | Py_DECREF(Py_None); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 500 | self->extra->attrib = res; |
| 501 | } |
| 502 | |
| 503 | return res; |
| 504 | } |
| 505 | |
| 506 | LOCAL(PyObject*) |
| 507 | element_get_text(ElementObject* self) |
| 508 | { |
| 509 | /* return borrowed reference to text attribute */ |
| 510 | |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 511 | PyObject *res = self->text; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 512 | |
| 513 | if (JOIN_GET(res)) { |
| 514 | res = JOIN_OBJ(res); |
| 515 | if (PyList_CheckExact(res)) { |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 516 | PyObject *tmp = list_join(res); |
| 517 | if (!tmp) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 518 | return NULL; |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 519 | self->text = tmp; |
| 520 | Py_DECREF(res); |
| 521 | res = tmp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 522 | } |
| 523 | } |
| 524 | |
| 525 | return res; |
| 526 | } |
| 527 | |
| 528 | LOCAL(PyObject*) |
| 529 | element_get_tail(ElementObject* self) |
| 530 | { |
| 531 | /* return borrowed reference to text attribute */ |
| 532 | |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 533 | PyObject *res = self->tail; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 534 | |
| 535 | if (JOIN_GET(res)) { |
| 536 | res = JOIN_OBJ(res); |
| 537 | if (PyList_CheckExact(res)) { |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 538 | PyObject *tmp = list_join(res); |
| 539 | if (!tmp) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 540 | return NULL; |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 541 | self->tail = tmp; |
| 542 | Py_DECREF(res); |
| 543 | res = tmp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 544 | } |
| 545 | } |
| 546 | |
| 547 | return res; |
| 548 | } |
| 549 | |
| 550 | static PyObject* |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 551 | subelement(PyObject *self, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 552 | { |
| 553 | PyObject* elem; |
| 554 | |
| 555 | ElementObject* parent; |
| 556 | PyObject* tag; |
| 557 | PyObject* attrib = NULL; |
| 558 | if (!PyArg_ParseTuple(args, "O!O|O!:SubElement", |
| 559 | &Element_Type, &parent, &tag, |
Eli Bendersky | 163d7f0 | 2013-11-24 06:55:04 -0800 | [diff] [blame] | 560 | &PyDict_Type, &attrib)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 561 | return NULL; |
Eli Bendersky | 163d7f0 | 2013-11-24 06:55:04 -0800 | [diff] [blame] | 562 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 563 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 564 | if (attrib) { |
| 565 | /* attrib passed as positional arg */ |
| 566 | attrib = PyDict_Copy(attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 567 | if (!attrib) |
| 568 | return NULL; |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 569 | if (kwds) { |
| 570 | if (PyDict_Update(attrib, kwds) < 0) { |
| 571 | return NULL; |
| 572 | } |
| 573 | } |
| 574 | } else if (kwds) { |
| 575 | /* have keyword args */ |
| 576 | attrib = get_attrib_from_keywords(kwds); |
| 577 | if (!attrib) |
| 578 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 579 | } else { |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 580 | /* no attrib arg, no kwds, so no attribute */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 581 | Py_INCREF(Py_None); |
| 582 | attrib = Py_None; |
| 583 | } |
| 584 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 585 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 586 | Py_DECREF(attrib); |
Victor Stinner | 71c8b7e | 2013-07-11 23:08:39 +0200 | [diff] [blame] | 587 | if (elem == NULL) |
| 588 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 589 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 590 | if (element_add_subelement(parent, elem) < 0) { |
| 591 | Py_DECREF(elem); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 592 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 593 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 594 | |
| 595 | return elem; |
| 596 | } |
| 597 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 598 | static int |
| 599 | element_gc_traverse(ElementObject *self, visitproc visit, void *arg) |
| 600 | { |
| 601 | Py_VISIT(self->tag); |
| 602 | Py_VISIT(JOIN_OBJ(self->text)); |
| 603 | Py_VISIT(JOIN_OBJ(self->tail)); |
| 604 | |
| 605 | if (self->extra) { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 606 | Py_ssize_t i; |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 607 | Py_VISIT(self->extra->attrib); |
| 608 | |
| 609 | for (i = 0; i < self->extra->length; ++i) |
| 610 | Py_VISIT(self->extra->children[i]); |
| 611 | } |
| 612 | return 0; |
| 613 | } |
| 614 | |
| 615 | static int |
| 616 | element_gc_clear(ElementObject *self) |
| 617 | { |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 618 | Py_CLEAR(self->tag); |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 619 | _clear_joined_ptr(&self->text); |
| 620 | _clear_joined_ptr(&self->tail); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 621 | |
| 622 | /* After dropping all references from extra, it's no longer valid anyway, |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 623 | * so fully deallocate it. |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 624 | */ |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 625 | dealloc_extra(self); |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 626 | return 0; |
| 627 | } |
| 628 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 629 | static void |
| 630 | element_dealloc(ElementObject* self) |
| 631 | { |
INADA Naoki | a6296d3 | 2017-08-24 14:55:17 +0900 | [diff] [blame] | 632 | /* bpo-31095: UnTrack is needed before calling any callbacks */ |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 633 | PyObject_GC_UnTrack(self); |
Serhiy Storchaka | 18f018c | 2016-12-21 12:32:56 +0200 | [diff] [blame] | 634 | Py_TRASHCAN_SAFE_BEGIN(self) |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 635 | |
| 636 | if (self->weakreflist != NULL) |
| 637 | PyObject_ClearWeakRefs((PyObject *) self); |
| 638 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 639 | /* element_gc_clear clears all references and deallocates extra |
| 640 | */ |
| 641 | element_gc_clear(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 642 | |
| 643 | RELEASE(sizeof(ElementObject), "destroy element"); |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 644 | Py_TYPE(self)->tp_free((PyObject *)self); |
Serhiy Storchaka | 18f018c | 2016-12-21 12:32:56 +0200 | [diff] [blame] | 645 | Py_TRASHCAN_SAFE_END(self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 646 | } |
| 647 | |
| 648 | /* -------------------------------------------------------------------- */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 649 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 650 | /*[clinic input] |
| 651 | _elementtree.Element.append |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 652 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 653 | subelement: object(subclass_of='&Element_Type') |
| 654 | / |
| 655 | |
| 656 | [clinic start generated code]*/ |
| 657 | |
| 658 | static PyObject * |
| 659 | _elementtree_Element_append_impl(ElementObject *self, PyObject *subelement) |
| 660 | /*[clinic end generated code: output=54a884b7cf2295f4 input=3ed648beb5bfa22a]*/ |
| 661 | { |
| 662 | if (element_add_subelement(self, subelement) < 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 663 | return NULL; |
| 664 | |
| 665 | Py_RETURN_NONE; |
| 666 | } |
| 667 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 668 | /*[clinic input] |
| 669 | _elementtree.Element.clear |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 670 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 671 | [clinic start generated code]*/ |
| 672 | |
| 673 | static PyObject * |
| 674 | _elementtree_Element_clear_impl(ElementObject *self) |
| 675 | /*[clinic end generated code: output=8bcd7a51f94cfff6 input=3c719ff94bf45dd6]*/ |
| 676 | { |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 677 | dealloc_extra(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 678 | |
| 679 | Py_INCREF(Py_None); |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 680 | _set_joined_ptr(&self->text, Py_None); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 681 | |
| 682 | Py_INCREF(Py_None); |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 683 | _set_joined_ptr(&self->tail, Py_None); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 684 | |
| 685 | Py_RETURN_NONE; |
| 686 | } |
| 687 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 688 | /*[clinic input] |
| 689 | _elementtree.Element.__copy__ |
| 690 | |
| 691 | [clinic start generated code]*/ |
| 692 | |
| 693 | static PyObject * |
| 694 | _elementtree_Element___copy___impl(ElementObject *self) |
| 695 | /*[clinic end generated code: output=2c701ebff7247781 input=ad87aaebe95675bf]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 696 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 697 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 698 | ElementObject* element; |
| 699 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 700 | element = (ElementObject*) create_new_element( |
Eli Bendersky | 163d7f0 | 2013-11-24 06:55:04 -0800 | [diff] [blame] | 701 | self->tag, (self->extra) ? self->extra->attrib : Py_None); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 702 | if (!element) |
| 703 | return NULL; |
| 704 | |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 705 | Py_INCREF(JOIN_OBJ(self->text)); |
| 706 | _set_joined_ptr(&element->text, self->text); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 707 | |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 708 | Py_INCREF(JOIN_OBJ(self->tail)); |
| 709 | _set_joined_ptr(&element->tail, self->tail); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 710 | |
| 711 | if (self->extra) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 712 | if (element_resize(element, self->extra->length) < 0) { |
| 713 | Py_DECREF(element); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 714 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 715 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 716 | |
| 717 | for (i = 0; i < self->extra->length; i++) { |
| 718 | Py_INCREF(self->extra->children[i]); |
| 719 | element->extra->children[i] = self->extra->children[i]; |
| 720 | } |
| 721 | |
| 722 | element->extra->length = self->extra->length; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 723 | } |
| 724 | |
| 725 | return (PyObject*) element; |
| 726 | } |
| 727 | |
Serhiy Storchaka | 060ed71 | 2015-12-21 12:57:27 +0200 | [diff] [blame] | 728 | /* Helper for a deep copy. */ |
| 729 | LOCAL(PyObject *) deepcopy(PyObject *, PyObject *); |
| 730 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 731 | /*[clinic input] |
| 732 | _elementtree.Element.__deepcopy__ |
| 733 | |
Oren Milman | d056818 | 2017-09-12 17:39:15 +0300 | [diff] [blame] | 734 | memo: object(subclass_of="&PyDict_Type") |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 735 | / |
| 736 | |
| 737 | [clinic start generated code]*/ |
| 738 | |
| 739 | static PyObject * |
Oren Milman | d056818 | 2017-09-12 17:39:15 +0300 | [diff] [blame] | 740 | _elementtree_Element___deepcopy___impl(ElementObject *self, PyObject *memo) |
| 741 | /*[clinic end generated code: output=eefc3df50465b642 input=a2d40348c0aade10]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 742 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 743 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 744 | ElementObject* element; |
| 745 | PyObject* tag; |
| 746 | PyObject* attrib; |
| 747 | PyObject* text; |
| 748 | PyObject* tail; |
| 749 | PyObject* id; |
| 750 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 751 | tag = deepcopy(self->tag, memo); |
| 752 | if (!tag) |
| 753 | return NULL; |
| 754 | |
| 755 | if (self->extra) { |
| 756 | attrib = deepcopy(self->extra->attrib, memo); |
| 757 | if (!attrib) { |
| 758 | Py_DECREF(tag); |
| 759 | return NULL; |
| 760 | } |
| 761 | } else { |
| 762 | Py_INCREF(Py_None); |
| 763 | attrib = Py_None; |
| 764 | } |
| 765 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 766 | element = (ElementObject*) create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 767 | |
| 768 | Py_DECREF(tag); |
| 769 | Py_DECREF(attrib); |
| 770 | |
| 771 | if (!element) |
| 772 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 773 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 774 | text = deepcopy(JOIN_OBJ(self->text), memo); |
| 775 | if (!text) |
| 776 | goto error; |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 777 | _set_joined_ptr(&element->text, JOIN_SET(text, JOIN_GET(self->text))); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 778 | |
| 779 | tail = deepcopy(JOIN_OBJ(self->tail), memo); |
| 780 | if (!tail) |
| 781 | goto error; |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 782 | _set_joined_ptr(&element->tail, JOIN_SET(tail, JOIN_GET(self->tail))); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 783 | |
| 784 | if (self->extra) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 785 | if (element_resize(element, self->extra->length) < 0) |
| 786 | goto error; |
| 787 | |
| 788 | for (i = 0; i < self->extra->length; i++) { |
| 789 | PyObject* child = deepcopy(self->extra->children[i], memo); |
| 790 | if (!child) { |
| 791 | element->extra->length = i; |
| 792 | goto error; |
| 793 | } |
| 794 | element->extra->children[i] = child; |
| 795 | } |
| 796 | |
| 797 | element->extra->length = self->extra->length; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 798 | } |
| 799 | |
| 800 | /* add object to memo dictionary (so deepcopy won't visit it again) */ |
Benjamin Peterson | ca47063 | 2016-09-06 13:47:26 -0700 | [diff] [blame] | 801 | id = PyLong_FromSsize_t((uintptr_t) self); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 802 | if (!id) |
| 803 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 804 | |
| 805 | i = PyDict_SetItem(memo, id, (PyObject*) element); |
| 806 | |
| 807 | Py_DECREF(id); |
| 808 | |
| 809 | if (i < 0) |
| 810 | goto error; |
| 811 | |
| 812 | return (PyObject*) element; |
| 813 | |
| 814 | error: |
| 815 | Py_DECREF(element); |
| 816 | return NULL; |
| 817 | } |
| 818 | |
Serhiy Storchaka | 060ed71 | 2015-12-21 12:57:27 +0200 | [diff] [blame] | 819 | LOCAL(PyObject *) |
| 820 | deepcopy(PyObject *object, PyObject *memo) |
| 821 | { |
| 822 | /* do a deep copy of the given object */ |
Serhiy Storchaka | 060ed71 | 2015-12-21 12:57:27 +0200 | [diff] [blame] | 823 | elementtreestate *st; |
Victor Stinner | 7fbac45 | 2016-08-20 01:34:44 +0200 | [diff] [blame] | 824 | PyObject *stack[2]; |
Serhiy Storchaka | 060ed71 | 2015-12-21 12:57:27 +0200 | [diff] [blame] | 825 | |
| 826 | /* Fast paths */ |
| 827 | if (object == Py_None || PyUnicode_CheckExact(object)) { |
| 828 | Py_INCREF(object); |
| 829 | return object; |
| 830 | } |
| 831 | |
| 832 | if (Py_REFCNT(object) == 1) { |
| 833 | if (PyDict_CheckExact(object)) { |
| 834 | PyObject *key, *value; |
| 835 | Py_ssize_t pos = 0; |
| 836 | int simple = 1; |
| 837 | while (PyDict_Next(object, &pos, &key, &value)) { |
| 838 | if (!PyUnicode_CheckExact(key) || !PyUnicode_CheckExact(value)) { |
| 839 | simple = 0; |
| 840 | break; |
| 841 | } |
| 842 | } |
| 843 | if (simple) |
| 844 | return PyDict_Copy(object); |
| 845 | /* Fall through to general case */ |
| 846 | } |
| 847 | else if (Element_CheckExact(object)) { |
Oren Milman | d056818 | 2017-09-12 17:39:15 +0300 | [diff] [blame] | 848 | return _elementtree_Element___deepcopy___impl( |
| 849 | (ElementObject *)object, memo); |
Serhiy Storchaka | 060ed71 | 2015-12-21 12:57:27 +0200 | [diff] [blame] | 850 | } |
| 851 | } |
| 852 | |
| 853 | /* General case */ |
| 854 | st = ET_STATE_GLOBAL; |
| 855 | if (!st->deepcopy_obj) { |
| 856 | PyErr_SetString(PyExc_RuntimeError, |
| 857 | "deepcopy helper not found"); |
| 858 | return NULL; |
| 859 | } |
| 860 | |
Victor Stinner | 7fbac45 | 2016-08-20 01:34:44 +0200 | [diff] [blame] | 861 | stack[0] = object; |
| 862 | stack[1] = memo; |
Victor Stinner | 559bb6a | 2016-08-22 22:48:54 +0200 | [diff] [blame] | 863 | return _PyObject_FastCall(st->deepcopy_obj, stack, 2); |
Serhiy Storchaka | 060ed71 | 2015-12-21 12:57:27 +0200 | [diff] [blame] | 864 | } |
| 865 | |
| 866 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 867 | /*[clinic input] |
| 868 | _elementtree.Element.__sizeof__ -> Py_ssize_t |
| 869 | |
| 870 | [clinic start generated code]*/ |
| 871 | |
| 872 | static Py_ssize_t |
| 873 | _elementtree_Element___sizeof___impl(ElementObject *self) |
| 874 | /*[clinic end generated code: output=bf73867721008000 input=70f4b323d55a17c1]*/ |
Martin v. Löwis | bce1666 | 2012-06-17 10:41:22 +0200 | [diff] [blame] | 875 | { |
Serhiy Storchaka | 5c4064e | 2015-12-19 20:05:25 +0200 | [diff] [blame] | 876 | Py_ssize_t result = _PyObject_SIZE(Py_TYPE(self)); |
Martin v. Löwis | bce1666 | 2012-06-17 10:41:22 +0200 | [diff] [blame] | 877 | if (self->extra) { |
| 878 | result += sizeof(ElementObjectExtra); |
| 879 | if (self->extra->children != self->extra->_children) |
| 880 | result += sizeof(PyObject*) * self->extra->allocated; |
| 881 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 882 | return result; |
Martin v. Löwis | bce1666 | 2012-06-17 10:41:22 +0200 | [diff] [blame] | 883 | } |
| 884 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 885 | /* dict keys for getstate/setstate. */ |
| 886 | #define PICKLED_TAG "tag" |
| 887 | #define PICKLED_CHILDREN "_children" |
| 888 | #define PICKLED_ATTRIB "attrib" |
| 889 | #define PICKLED_TAIL "tail" |
| 890 | #define PICKLED_TEXT "text" |
| 891 | |
| 892 | /* __getstate__ returns a fabricated instance dict as in the pure-Python |
| 893 | * Element implementation, for interoperability/interchangeability. This |
| 894 | * makes the pure-Python implementation details an API, but (a) there aren't |
| 895 | * any unnecessary structures there; and (b) it buys compatibility with 3.2 |
| 896 | * pickles. See issue #16076. |
| 897 | */ |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 898 | /*[clinic input] |
| 899 | _elementtree.Element.__getstate__ |
| 900 | |
| 901 | [clinic start generated code]*/ |
| 902 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 903 | static PyObject * |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 904 | _elementtree_Element___getstate___impl(ElementObject *self) |
| 905 | /*[clinic end generated code: output=37279aeeb6bb5b04 input=f0d16d7ec2f7adc1]*/ |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 906 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 907 | Py_ssize_t i, noattrib; |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 908 | PyObject *instancedict = NULL, *children; |
| 909 | |
| 910 | /* Build a list of children. */ |
| 911 | children = PyList_New(self->extra ? self->extra->length : 0); |
| 912 | if (!children) |
| 913 | return NULL; |
| 914 | for (i = 0; i < PyList_GET_SIZE(children); i++) { |
| 915 | PyObject *child = self->extra->children[i]; |
| 916 | Py_INCREF(child); |
| 917 | PyList_SET_ITEM(children, i, child); |
| 918 | } |
| 919 | |
| 920 | /* Construct the state object. */ |
| 921 | noattrib = (self->extra == NULL || self->extra->attrib == Py_None); |
| 922 | if (noattrib) |
| 923 | instancedict = Py_BuildValue("{sOsOs{}sOsO}", |
| 924 | PICKLED_TAG, self->tag, |
| 925 | PICKLED_CHILDREN, children, |
| 926 | PICKLED_ATTRIB, |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 927 | PICKLED_TEXT, JOIN_OBJ(self->text), |
| 928 | PICKLED_TAIL, JOIN_OBJ(self->tail)); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 929 | else |
| 930 | instancedict = Py_BuildValue("{sOsOsOsOsO}", |
| 931 | PICKLED_TAG, self->tag, |
| 932 | PICKLED_CHILDREN, children, |
| 933 | PICKLED_ATTRIB, self->extra->attrib, |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 934 | PICKLED_TEXT, JOIN_OBJ(self->text), |
| 935 | PICKLED_TAIL, JOIN_OBJ(self->tail)); |
Eli Bendersky | b8f6dc8 | 2013-01-12 05:20:16 -0800 | [diff] [blame] | 936 | if (instancedict) { |
| 937 | Py_DECREF(children); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 938 | return instancedict; |
Eli Bendersky | b8f6dc8 | 2013-01-12 05:20:16 -0800 | [diff] [blame] | 939 | } |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 940 | else { |
| 941 | for (i = 0; i < PyList_GET_SIZE(children); i++) |
| 942 | Py_DECREF(PyList_GET_ITEM(children, i)); |
| 943 | Py_DECREF(children); |
| 944 | |
| 945 | return NULL; |
| 946 | } |
| 947 | } |
| 948 | |
| 949 | static PyObject * |
| 950 | element_setstate_from_attributes(ElementObject *self, |
| 951 | PyObject *tag, |
| 952 | PyObject *attrib, |
| 953 | PyObject *text, |
| 954 | PyObject *tail, |
| 955 | PyObject *children) |
| 956 | { |
| 957 | Py_ssize_t i, nchildren; |
| 958 | |
| 959 | if (!tag) { |
| 960 | PyErr_SetString(PyExc_TypeError, "tag may not be NULL"); |
| 961 | return NULL; |
| 962 | } |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 963 | |
Serhiy Storchaka | 191321d | 2015-12-27 15:41:34 +0200 | [diff] [blame] | 964 | Py_INCREF(tag); |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 965 | Py_XSETREF(self->tag, tag); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 966 | |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 967 | text = text ? JOIN_SET(text, PyList_CheckExact(text)) : Py_None; |
| 968 | Py_INCREF(JOIN_OBJ(text)); |
| 969 | _set_joined_ptr(&self->text, text); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 970 | |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 971 | tail = tail ? JOIN_SET(tail, PyList_CheckExact(tail)) : Py_None; |
| 972 | Py_INCREF(JOIN_OBJ(tail)); |
| 973 | _set_joined_ptr(&self->tail, tail); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 974 | |
| 975 | /* Handle ATTRIB and CHILDREN. */ |
| 976 | if (!children && !attrib) |
| 977 | Py_RETURN_NONE; |
| 978 | |
| 979 | /* Compute 'nchildren'. */ |
| 980 | if (children) { |
| 981 | if (!PyList_Check(children)) { |
| 982 | PyErr_SetString(PyExc_TypeError, "'_children' is not a list"); |
| 983 | return NULL; |
| 984 | } |
| 985 | nchildren = PyList_Size(children); |
| 986 | } |
| 987 | else { |
| 988 | nchildren = 0; |
| 989 | } |
| 990 | |
| 991 | /* Allocate 'extra'. */ |
| 992 | if (element_resize(self, nchildren)) { |
| 993 | return NULL; |
| 994 | } |
| 995 | assert(self->extra && self->extra->allocated >= nchildren); |
| 996 | |
| 997 | /* Copy children */ |
| 998 | for (i = 0; i < nchildren; i++) { |
| 999 | self->extra->children[i] = PyList_GET_ITEM(children, i); |
| 1000 | Py_INCREF(self->extra->children[i]); |
| 1001 | } |
| 1002 | |
| 1003 | self->extra->length = nchildren; |
| 1004 | self->extra->allocated = nchildren; |
| 1005 | |
| 1006 | /* Stash attrib. */ |
| 1007 | if (attrib) { |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1008 | Py_INCREF(attrib); |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 1009 | Py_XSETREF(self->extra->attrib, attrib); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1010 | } |
| 1011 | |
| 1012 | Py_RETURN_NONE; |
| 1013 | } |
| 1014 | |
| 1015 | /* __setstate__ for Element instance from the Python implementation. |
| 1016 | * 'state' should be the instance dict. |
| 1017 | */ |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1018 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1019 | static PyObject * |
| 1020 | element_setstate_from_Python(ElementObject *self, PyObject *state) |
| 1021 | { |
| 1022 | static char *kwlist[] = {PICKLED_TAG, PICKLED_ATTRIB, PICKLED_TEXT, |
| 1023 | PICKLED_TAIL, PICKLED_CHILDREN, 0}; |
| 1024 | PyObject *args; |
| 1025 | PyObject *tag, *attrib, *text, *tail, *children; |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 1026 | PyObject *retval; |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1027 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1028 | tag = attrib = text = tail = children = NULL; |
| 1029 | args = PyTuple_New(0); |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 1030 | if (!args) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1031 | return NULL; |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 1032 | |
| 1033 | if (PyArg_ParseTupleAndKeywords(args, state, "|$OOOOO", kwlist, &tag, |
| 1034 | &attrib, &text, &tail, &children)) |
| 1035 | retval = element_setstate_from_attributes(self, tag, attrib, text, |
| 1036 | tail, children); |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1037 | else |
Eli Bendersky | 799e3ed | 2013-01-12 05:42:38 -0800 | [diff] [blame] | 1038 | retval = NULL; |
| 1039 | |
| 1040 | Py_DECREF(args); |
| 1041 | return retval; |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1042 | } |
| 1043 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1044 | /*[clinic input] |
| 1045 | _elementtree.Element.__setstate__ |
| 1046 | |
| 1047 | state: object |
| 1048 | / |
| 1049 | |
| 1050 | [clinic start generated code]*/ |
| 1051 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1052 | static PyObject * |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1053 | _elementtree_Element___setstate__(ElementObject *self, PyObject *state) |
| 1054 | /*[clinic end generated code: output=ea28bf3491b1f75e input=aaf80abea7c1e3b9]*/ |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1055 | { |
| 1056 | if (!PyDict_CheckExact(state)) { |
| 1057 | PyErr_Format(PyExc_TypeError, |
| 1058 | "Don't know how to unpickle \"%.200R\" as an Element", |
| 1059 | state); |
| 1060 | return NULL; |
| 1061 | } |
| 1062 | else |
| 1063 | return element_setstate_from_Python(self, state); |
| 1064 | } |
| 1065 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1066 | LOCAL(int) |
| 1067 | checkpath(PyObject* tag) |
| 1068 | { |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1069 | Py_ssize_t i; |
| 1070 | int check = 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1071 | |
| 1072 | /* check if a tag contains an xpath character */ |
| 1073 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1074 | #define PATHCHAR(ch) \ |
| 1075 | (ch == '/' || ch == '*' || ch == '[' || ch == '@' || ch == '.') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1076 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1077 | if (PyUnicode_Check(tag)) { |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 1078 | const Py_ssize_t len = PyUnicode_GET_LENGTH(tag); |
| 1079 | void *data = PyUnicode_DATA(tag); |
| 1080 | unsigned int kind = PyUnicode_KIND(tag); |
| 1081 | for (i = 0; i < len; i++) { |
| 1082 | Py_UCS4 ch = PyUnicode_READ(kind, data, i); |
| 1083 | if (ch == '{') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1084 | check = 0; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 1085 | else if (ch == '}') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1086 | check = 1; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 1087 | else if (check && PATHCHAR(ch)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1088 | return 1; |
| 1089 | } |
| 1090 | return 0; |
| 1091 | } |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 1092 | if (PyBytes_Check(tag)) { |
| 1093 | char *p = PyBytes_AS_STRING(tag); |
| 1094 | for (i = 0; i < PyBytes_GET_SIZE(tag); i++) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1095 | if (p[i] == '{') |
| 1096 | check = 0; |
| 1097 | else if (p[i] == '}') |
| 1098 | check = 1; |
| 1099 | else if (check && PATHCHAR(p[i])) |
| 1100 | return 1; |
| 1101 | } |
| 1102 | return 0; |
| 1103 | } |
| 1104 | |
| 1105 | return 1; /* unknown type; might be path expression */ |
| 1106 | } |
| 1107 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1108 | /*[clinic input] |
| 1109 | _elementtree.Element.extend |
| 1110 | |
| 1111 | elements: object |
| 1112 | / |
| 1113 | |
| 1114 | [clinic start generated code]*/ |
| 1115 | |
| 1116 | static PyObject * |
| 1117 | _elementtree_Element_extend(ElementObject *self, PyObject *elements) |
| 1118 | /*[clinic end generated code: output=f6e67fc2ff529191 input=807bc4f31c69f7c0]*/ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1119 | { |
| 1120 | PyObject* seq; |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1121 | Py_ssize_t i; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1122 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1123 | seq = PySequence_Fast(elements, ""); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1124 | if (!seq) { |
| 1125 | PyErr_Format( |
| 1126 | PyExc_TypeError, |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1127 | "expected sequence, not \"%.200s\"", Py_TYPE(elements)->tp_name |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1128 | ); |
| 1129 | return NULL; |
| 1130 | } |
| 1131 | |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1132 | for (i = 0; i < PySequence_Fast_GET_SIZE(seq); i++) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1133 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1134 | Py_INCREF(element); |
| 1135 | if (!PyObject_TypeCheck(element, (PyTypeObject *)&Element_Type)) { |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 1136 | PyErr_Format( |
| 1137 | PyExc_TypeError, |
| 1138 | "expected an Element, not \"%.200s\"", |
| 1139 | Py_TYPE(element)->tp_name); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1140 | Py_DECREF(seq); |
| 1141 | Py_DECREF(element); |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 1142 | return NULL; |
| 1143 | } |
| 1144 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1145 | if (element_add_subelement(self, element) < 0) { |
| 1146 | Py_DECREF(seq); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1147 | Py_DECREF(element); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1148 | return NULL; |
| 1149 | } |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1150 | Py_DECREF(element); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1151 | } |
| 1152 | |
| 1153 | Py_DECREF(seq); |
| 1154 | |
| 1155 | Py_RETURN_NONE; |
| 1156 | } |
| 1157 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1158 | /*[clinic input] |
| 1159 | _elementtree.Element.find |
| 1160 | |
| 1161 | path: object |
| 1162 | namespaces: object = None |
| 1163 | |
| 1164 | [clinic start generated code]*/ |
| 1165 | |
| 1166 | static PyObject * |
| 1167 | _elementtree_Element_find_impl(ElementObject *self, PyObject *path, |
| 1168 | PyObject *namespaces) |
| 1169 | /*[clinic end generated code: output=41b43f0f0becafae input=359b6985f6489d2e]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1170 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 1171 | Py_ssize_t i; |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 1172 | elementtreestate *st = ET_STATE_GLOBAL; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1173 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1174 | if (checkpath(path) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1175 | _Py_IDENTIFIER(find); |
Victor Stinner | f561634 | 2016-12-09 15:26:00 +0100 | [diff] [blame] | 1176 | return _PyObject_CallMethodIdObjArgs( |
| 1177 | st->elementpath_obj, &PyId_find, self, path, namespaces, NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1178 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1179 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1180 | |
| 1181 | if (!self->extra) |
| 1182 | Py_RETURN_NONE; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1183 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1184 | for (i = 0; i < self->extra->length; i++) { |
| 1185 | PyObject* item = self->extra->children[i]; |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1186 | int rc; |
| 1187 | if (!Element_CheckExact(item)) |
| 1188 | continue; |
| 1189 | Py_INCREF(item); |
Serhiy Storchaka | a2c145c | 2015-05-18 18:33:31 +0300 | [diff] [blame] | 1190 | rc = PyObject_RichCompareBool(((ElementObject*)item)->tag, path, Py_EQ); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1191 | if (rc > 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1192 | return item; |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1193 | Py_DECREF(item); |
| 1194 | if (rc < 0) |
| 1195 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1196 | } |
| 1197 | |
| 1198 | Py_RETURN_NONE; |
| 1199 | } |
| 1200 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1201 | /*[clinic input] |
| 1202 | _elementtree.Element.findtext |
| 1203 | |
| 1204 | path: object |
| 1205 | default: object = None |
| 1206 | namespaces: object = None |
| 1207 | |
| 1208 | [clinic start generated code]*/ |
| 1209 | |
| 1210 | static PyObject * |
| 1211 | _elementtree_Element_findtext_impl(ElementObject *self, PyObject *path, |
| 1212 | PyObject *default_value, |
| 1213 | PyObject *namespaces) |
| 1214 | /*[clinic end generated code: output=83b3ba4535d308d2 input=b53a85aa5aa2a916]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1215 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 1216 | Py_ssize_t i; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1217 | _Py_IDENTIFIER(findtext); |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 1218 | elementtreestate *st = ET_STATE_GLOBAL; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1219 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1220 | if (checkpath(path) || namespaces != Py_None) |
Victor Stinner | f561634 | 2016-12-09 15:26:00 +0100 | [diff] [blame] | 1221 | return _PyObject_CallMethodIdObjArgs( |
| 1222 | st->elementpath_obj, &PyId_findtext, |
| 1223 | self, path, default_value, namespaces, NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1224 | ); |
| 1225 | |
| 1226 | if (!self->extra) { |
| 1227 | Py_INCREF(default_value); |
| 1228 | return default_value; |
| 1229 | } |
| 1230 | |
| 1231 | for (i = 0; i < self->extra->length; i++) { |
| 1232 | ElementObject* item = (ElementObject*) self->extra->children[i]; |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1233 | int rc; |
| 1234 | if (!Element_CheckExact(item)) |
| 1235 | continue; |
| 1236 | Py_INCREF(item); |
Serhiy Storchaka | a2c145c | 2015-05-18 18:33:31 +0300 | [diff] [blame] | 1237 | rc = PyObject_RichCompareBool(item->tag, path, Py_EQ); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1238 | if (rc > 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1239 | PyObject* text = element_get_text(item); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1240 | if (text == Py_None) { |
| 1241 | Py_DECREF(item); |
Eli Bendersky | 25771b3 | 2013-01-13 05:26:07 -0800 | [diff] [blame] | 1242 | return PyUnicode_New(0, 0); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1243 | } |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 1244 | Py_XINCREF(text); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1245 | Py_DECREF(item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1246 | return text; |
| 1247 | } |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1248 | Py_DECREF(item); |
| 1249 | if (rc < 0) |
| 1250 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1251 | } |
| 1252 | |
| 1253 | Py_INCREF(default_value); |
| 1254 | return default_value; |
| 1255 | } |
| 1256 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1257 | /*[clinic input] |
| 1258 | _elementtree.Element.findall |
| 1259 | |
| 1260 | path: object |
| 1261 | namespaces: object = None |
| 1262 | |
| 1263 | [clinic start generated code]*/ |
| 1264 | |
| 1265 | static PyObject * |
| 1266 | _elementtree_Element_findall_impl(ElementObject *self, PyObject *path, |
| 1267 | PyObject *namespaces) |
| 1268 | /*[clinic end generated code: output=1a0bd9f5541b711d input=4d9e6505a638550c]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1269 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 1270 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1271 | PyObject* out; |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1272 | PyObject* tag = path; |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 1273 | elementtreestate *st = ET_STATE_GLOBAL; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1274 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1275 | if (checkpath(tag) || namespaces != Py_None) { |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1276 | _Py_IDENTIFIER(findall); |
Victor Stinner | f561634 | 2016-12-09 15:26:00 +0100 | [diff] [blame] | 1277 | return _PyObject_CallMethodIdObjArgs( |
| 1278 | st->elementpath_obj, &PyId_findall, self, tag, namespaces, NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1279 | ); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1280 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1281 | |
| 1282 | out = PyList_New(0); |
| 1283 | if (!out) |
| 1284 | return NULL; |
| 1285 | |
| 1286 | if (!self->extra) |
| 1287 | return out; |
| 1288 | |
| 1289 | for (i = 0; i < self->extra->length; i++) { |
| 1290 | PyObject* item = self->extra->children[i]; |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1291 | int rc; |
| 1292 | if (!Element_CheckExact(item)) |
| 1293 | continue; |
| 1294 | Py_INCREF(item); |
| 1295 | rc = PyObject_RichCompareBool(((ElementObject*)item)->tag, tag, Py_EQ); |
| 1296 | if (rc != 0 && (rc < 0 || PyList_Append(out, item) < 0)) { |
| 1297 | Py_DECREF(item); |
| 1298 | Py_DECREF(out); |
| 1299 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1300 | } |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1301 | Py_DECREF(item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1302 | } |
| 1303 | |
| 1304 | return out; |
| 1305 | } |
| 1306 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1307 | /*[clinic input] |
| 1308 | _elementtree.Element.iterfind |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1309 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1310 | path: object |
| 1311 | namespaces: object = None |
| 1312 | |
| 1313 | [clinic start generated code]*/ |
| 1314 | |
| 1315 | static PyObject * |
| 1316 | _elementtree_Element_iterfind_impl(ElementObject *self, PyObject *path, |
| 1317 | PyObject *namespaces) |
| 1318 | /*[clinic end generated code: output=ecdd56d63b19d40f input=abb974e350fb65c7]*/ |
| 1319 | { |
| 1320 | PyObject* tag = path; |
| 1321 | _Py_IDENTIFIER(iterfind); |
| 1322 | elementtreestate *st = ET_STATE_GLOBAL; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1323 | |
Victor Stinner | f561634 | 2016-12-09 15:26:00 +0100 | [diff] [blame] | 1324 | return _PyObject_CallMethodIdObjArgs( |
| 1325 | st->elementpath_obj, &PyId_iterfind, self, tag, namespaces, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1326 | } |
| 1327 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1328 | /*[clinic input] |
| 1329 | _elementtree.Element.get |
| 1330 | |
| 1331 | key: object |
| 1332 | default: object = None |
| 1333 | |
| 1334 | [clinic start generated code]*/ |
| 1335 | |
| 1336 | static PyObject * |
| 1337 | _elementtree_Element_get_impl(ElementObject *self, PyObject *key, |
| 1338 | PyObject *default_value) |
| 1339 | /*[clinic end generated code: output=523c614142595d75 input=ee153bbf8cdb246e]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1340 | { |
| 1341 | PyObject* value; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1342 | |
| 1343 | if (!self->extra || self->extra->attrib == Py_None) |
| 1344 | value = default_value; |
| 1345 | else { |
| 1346 | value = PyDict_GetItem(self->extra->attrib, key); |
| 1347 | if (!value) |
| 1348 | value = default_value; |
| 1349 | } |
| 1350 | |
| 1351 | Py_INCREF(value); |
| 1352 | return value; |
| 1353 | } |
| 1354 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1355 | /*[clinic input] |
| 1356 | _elementtree.Element.getchildren |
| 1357 | |
| 1358 | [clinic start generated code]*/ |
| 1359 | |
| 1360 | static PyObject * |
| 1361 | _elementtree_Element_getchildren_impl(ElementObject *self) |
| 1362 | /*[clinic end generated code: output=e50ffe118637b14f input=0f754dfded150d5f]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1363 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 1364 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1365 | PyObject* list; |
| 1366 | |
Serhiy Storchaka | 762ec97 | 2017-03-30 18:12:06 +0300 | [diff] [blame] | 1367 | if (PyErr_WarnEx(PyExc_DeprecationWarning, |
| 1368 | "This method will be removed in future versions. " |
| 1369 | "Use 'list(elem)' or iteration over elem instead.", |
| 1370 | 1) < 0) { |
| 1371 | return NULL; |
| 1372 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1373 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1374 | if (!self->extra) |
| 1375 | return PyList_New(0); |
| 1376 | |
| 1377 | list = PyList_New(self->extra->length); |
| 1378 | if (!list) |
| 1379 | return NULL; |
| 1380 | |
| 1381 | for (i = 0; i < self->extra->length; i++) { |
| 1382 | PyObject* item = self->extra->children[i]; |
| 1383 | Py_INCREF(item); |
| 1384 | PyList_SET_ITEM(list, i, item); |
| 1385 | } |
| 1386 | |
| 1387 | return list; |
| 1388 | } |
| 1389 | |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 1390 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1391 | static PyObject * |
| 1392 | create_elementiter(ElementObject *self, PyObject *tag, int gettext); |
| 1393 | |
| 1394 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1395 | /*[clinic input] |
| 1396 | _elementtree.Element.iter |
| 1397 | |
| 1398 | tag: object = None |
| 1399 | |
| 1400 | [clinic start generated code]*/ |
| 1401 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1402 | static PyObject * |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1403 | _elementtree_Element_iter_impl(ElementObject *self, PyObject *tag) |
| 1404 | /*[clinic end generated code: output=3f49f9a862941cc5 input=774d5b12e573aedd]*/ |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1405 | { |
Serhiy Storchaka | d6a69d8 | 2015-12-09 11:27:07 +0200 | [diff] [blame] | 1406 | if (PyUnicode_Check(tag)) { |
| 1407 | if (PyUnicode_READY(tag) < 0) |
| 1408 | return NULL; |
| 1409 | if (PyUnicode_GET_LENGTH(tag) == 1 && PyUnicode_READ_CHAR(tag, 0) == '*') |
| 1410 | tag = Py_None; |
| 1411 | } |
| 1412 | else if (PyBytes_Check(tag)) { |
| 1413 | if (PyBytes_GET_SIZE(tag) == 1 && *PyBytes_AS_STRING(tag) == '*') |
| 1414 | tag = Py_None; |
| 1415 | } |
| 1416 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1417 | return create_elementiter(self, tag, 0); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1418 | } |
| 1419 | |
| 1420 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1421 | /*[clinic input] |
Serhiy Storchaka | 762ec97 | 2017-03-30 18:12:06 +0300 | [diff] [blame] | 1422 | _elementtree.Element.getiterator |
| 1423 | |
| 1424 | tag: object = None |
| 1425 | |
| 1426 | [clinic start generated code]*/ |
| 1427 | |
| 1428 | static PyObject * |
| 1429 | _elementtree_Element_getiterator_impl(ElementObject *self, PyObject *tag) |
| 1430 | /*[clinic end generated code: output=cb69ff4a3742dfa1 input=500da1a03f7b9e28]*/ |
| 1431 | { |
| 1432 | /* Change for a DeprecationWarning in 1.4 */ |
| 1433 | if (PyErr_WarnEx(PyExc_PendingDeprecationWarning, |
| 1434 | "This method will be removed in future versions. " |
| 1435 | "Use 'tree.iter()' or 'list(tree.iter())' instead.", |
| 1436 | 1) < 0) { |
| 1437 | return NULL; |
| 1438 | } |
| 1439 | return _elementtree_Element_iter_impl(self, tag); |
| 1440 | } |
| 1441 | |
| 1442 | |
| 1443 | /*[clinic input] |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1444 | _elementtree.Element.itertext |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1445 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1446 | [clinic start generated code]*/ |
| 1447 | |
| 1448 | static PyObject * |
| 1449 | _elementtree_Element_itertext_impl(ElementObject *self) |
| 1450 | /*[clinic end generated code: output=5fa34b2fbcb65df6 input=af8f0e42cb239c89]*/ |
| 1451 | { |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1452 | return create_elementiter(self, Py_None, 1); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1453 | } |
| 1454 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1455 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1456 | static PyObject* |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1457 | element_getitem(PyObject* self_, Py_ssize_t index) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1458 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1459 | ElementObject* self = (ElementObject*) self_; |
| 1460 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1461 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1462 | PyErr_SetString( |
| 1463 | PyExc_IndexError, |
| 1464 | "child index out of range" |
| 1465 | ); |
| 1466 | return NULL; |
| 1467 | } |
| 1468 | |
| 1469 | Py_INCREF(self->extra->children[index]); |
| 1470 | return self->extra->children[index]; |
| 1471 | } |
| 1472 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1473 | /*[clinic input] |
| 1474 | _elementtree.Element.insert |
| 1475 | |
| 1476 | index: Py_ssize_t |
| 1477 | subelement: object(subclass_of='&Element_Type') |
| 1478 | / |
| 1479 | |
| 1480 | [clinic start generated code]*/ |
| 1481 | |
| 1482 | static PyObject * |
| 1483 | _elementtree_Element_insert_impl(ElementObject *self, Py_ssize_t index, |
| 1484 | PyObject *subelement) |
| 1485 | /*[clinic end generated code: output=990adfef4d424c0b input=cd6fbfcdab52d7a8]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1486 | { |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1487 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1488 | |
Victor Stinner | 5f0af23 | 2013-07-11 23:01:36 +0200 | [diff] [blame] | 1489 | if (!self->extra) { |
| 1490 | if (create_extra(self, NULL) < 0) |
| 1491 | return NULL; |
| 1492 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1493 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1494 | if (index < 0) { |
| 1495 | index += self->extra->length; |
| 1496 | if (index < 0) |
| 1497 | index = 0; |
| 1498 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1499 | if (index > self->extra->length) |
| 1500 | index = self->extra->length; |
| 1501 | |
| 1502 | if (element_resize(self, 1) < 0) |
| 1503 | return NULL; |
| 1504 | |
| 1505 | for (i = self->extra->length; i > index; i--) |
| 1506 | self->extra->children[i] = self->extra->children[i-1]; |
| 1507 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1508 | Py_INCREF(subelement); |
| 1509 | self->extra->children[index] = subelement; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1510 | |
| 1511 | self->extra->length++; |
| 1512 | |
| 1513 | Py_RETURN_NONE; |
| 1514 | } |
| 1515 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1516 | /*[clinic input] |
| 1517 | _elementtree.Element.items |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1518 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1519 | [clinic start generated code]*/ |
| 1520 | |
| 1521 | static PyObject * |
| 1522 | _elementtree_Element_items_impl(ElementObject *self) |
| 1523 | /*[clinic end generated code: output=6db2c778ce3f5a4d input=adbe09aaea474447]*/ |
| 1524 | { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1525 | if (!self->extra || self->extra->attrib == Py_None) |
| 1526 | return PyList_New(0); |
| 1527 | |
| 1528 | return PyDict_Items(self->extra->attrib); |
| 1529 | } |
| 1530 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1531 | /*[clinic input] |
| 1532 | _elementtree.Element.keys |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1533 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1534 | [clinic start generated code]*/ |
| 1535 | |
| 1536 | static PyObject * |
| 1537 | _elementtree_Element_keys_impl(ElementObject *self) |
| 1538 | /*[clinic end generated code: output=bc5bfabbf20eeb3c input=f02caf5b496b5b0b]*/ |
| 1539 | { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1540 | if (!self->extra || self->extra->attrib == Py_None) |
| 1541 | return PyList_New(0); |
| 1542 | |
| 1543 | return PyDict_Keys(self->extra->attrib); |
| 1544 | } |
| 1545 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1546 | static Py_ssize_t |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1547 | element_length(ElementObject* self) |
| 1548 | { |
| 1549 | if (!self->extra) |
| 1550 | return 0; |
| 1551 | |
| 1552 | return self->extra->length; |
| 1553 | } |
| 1554 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1555 | /*[clinic input] |
| 1556 | _elementtree.Element.makeelement |
| 1557 | |
| 1558 | tag: object |
| 1559 | attrib: object |
| 1560 | / |
| 1561 | |
| 1562 | [clinic start generated code]*/ |
| 1563 | |
| 1564 | static PyObject * |
| 1565 | _elementtree_Element_makeelement_impl(ElementObject *self, PyObject *tag, |
| 1566 | PyObject *attrib) |
| 1567 | /*[clinic end generated code: output=4109832d5bb789ef input=9480d1d2e3e68235]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1568 | { |
| 1569 | PyObject* elem; |
| 1570 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1571 | attrib = PyDict_Copy(attrib); |
| 1572 | if (!attrib) |
| 1573 | return NULL; |
| 1574 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1575 | elem = create_new_element(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1576 | |
| 1577 | Py_DECREF(attrib); |
| 1578 | |
| 1579 | return elem; |
| 1580 | } |
| 1581 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1582 | /*[clinic input] |
| 1583 | _elementtree.Element.remove |
| 1584 | |
| 1585 | subelement: object(subclass_of='&Element_Type') |
| 1586 | / |
| 1587 | |
| 1588 | [clinic start generated code]*/ |
| 1589 | |
| 1590 | static PyObject * |
| 1591 | _elementtree_Element_remove_impl(ElementObject *self, PyObject *subelement) |
| 1592 | /*[clinic end generated code: output=38fe6c07d6d87d1f input=d52fc28ededc0bd8]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1593 | { |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 1594 | Py_ssize_t i; |
Serhiy Storchaka | a2c145c | 2015-05-18 18:33:31 +0300 | [diff] [blame] | 1595 | int rc; |
| 1596 | PyObject *found; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1597 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1598 | if (!self->extra) { |
| 1599 | /* element has no children, so raise exception */ |
| 1600 | PyErr_SetString( |
| 1601 | PyExc_ValueError, |
| 1602 | "list.remove(x): x not in list" |
| 1603 | ); |
| 1604 | return NULL; |
| 1605 | } |
| 1606 | |
| 1607 | for (i = 0; i < self->extra->length; i++) { |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1608 | if (self->extra->children[i] == subelement) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1609 | break; |
Serhiy Storchaka | a2c145c | 2015-05-18 18:33:31 +0300 | [diff] [blame] | 1610 | rc = PyObject_RichCompareBool(self->extra->children[i], subelement, Py_EQ); |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1611 | if (rc > 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1612 | break; |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1613 | if (rc < 0) |
| 1614 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1615 | } |
| 1616 | |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1617 | if (i >= self->extra->length) { |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1618 | /* subelement is not in children, so raise exception */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1619 | PyErr_SetString( |
| 1620 | PyExc_ValueError, |
| 1621 | "list.remove(x): x not in list" |
| 1622 | ); |
| 1623 | return NULL; |
| 1624 | } |
| 1625 | |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1626 | found = self->extra->children[i]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1627 | |
| 1628 | self->extra->length--; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1629 | for (; i < self->extra->length; i++) |
| 1630 | self->extra->children[i] = self->extra->children[i+1]; |
| 1631 | |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1632 | Py_DECREF(found); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1633 | Py_RETURN_NONE; |
| 1634 | } |
| 1635 | |
| 1636 | static PyObject* |
| 1637 | element_repr(ElementObject* self) |
| 1638 | { |
Serhiy Storchaka | 9062c26 | 2016-06-12 09:43:55 +0300 | [diff] [blame] | 1639 | int status; |
| 1640 | |
| 1641 | if (self->tag == NULL) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1642 | return PyUnicode_FromFormat("<Element at %p>", self); |
Serhiy Storchaka | 9062c26 | 2016-06-12 09:43:55 +0300 | [diff] [blame] | 1643 | |
| 1644 | status = Py_ReprEnter((PyObject *)self); |
| 1645 | if (status == 0) { |
| 1646 | PyObject *res; |
| 1647 | res = PyUnicode_FromFormat("<Element %R at %p>", self->tag, self); |
| 1648 | Py_ReprLeave((PyObject *)self); |
| 1649 | return res; |
| 1650 | } |
| 1651 | if (status > 0) |
| 1652 | PyErr_Format(PyExc_RuntimeError, |
| 1653 | "reentrant call inside %s.__repr__", |
| 1654 | Py_TYPE(self)->tp_name); |
| 1655 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1656 | } |
| 1657 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 1658 | /*[clinic input] |
| 1659 | _elementtree.Element.set |
| 1660 | |
| 1661 | key: object |
| 1662 | value: object |
| 1663 | / |
| 1664 | |
| 1665 | [clinic start generated code]*/ |
| 1666 | |
| 1667 | static PyObject * |
| 1668 | _elementtree_Element_set_impl(ElementObject *self, PyObject *key, |
| 1669 | PyObject *value) |
| 1670 | /*[clinic end generated code: output=fb938806be3c5656 input=1efe90f7d82b3fe9]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1671 | { |
| 1672 | PyObject* attrib; |
| 1673 | |
Victor Stinner | 5f0af23 | 2013-07-11 23:01:36 +0200 | [diff] [blame] | 1674 | if (!self->extra) { |
| 1675 | if (create_extra(self, NULL) < 0) |
| 1676 | return NULL; |
| 1677 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1678 | |
| 1679 | attrib = element_get_attrib(self); |
| 1680 | if (!attrib) |
| 1681 | return NULL; |
| 1682 | |
| 1683 | if (PyDict_SetItem(attrib, key, value) < 0) |
| 1684 | return NULL; |
| 1685 | |
| 1686 | Py_RETURN_NONE; |
| 1687 | } |
| 1688 | |
| 1689 | static int |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1690 | element_setitem(PyObject* self_, Py_ssize_t index, PyObject* item) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1691 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1692 | ElementObject* self = (ElementObject*) self_; |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 1693 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1694 | PyObject* old; |
| 1695 | |
| 1696 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1697 | PyErr_SetString( |
| 1698 | PyExc_IndexError, |
| 1699 | "child assignment index out of range"); |
| 1700 | return -1; |
| 1701 | } |
| 1702 | |
| 1703 | old = self->extra->children[index]; |
| 1704 | |
| 1705 | if (item) { |
| 1706 | Py_INCREF(item); |
| 1707 | self->extra->children[index] = item; |
| 1708 | } else { |
| 1709 | self->extra->length--; |
| 1710 | for (i = index; i < self->extra->length; i++) |
| 1711 | self->extra->children[i] = self->extra->children[i+1]; |
| 1712 | } |
| 1713 | |
| 1714 | Py_DECREF(old); |
| 1715 | |
| 1716 | return 0; |
| 1717 | } |
| 1718 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1719 | static PyObject* |
| 1720 | element_subscr(PyObject* self_, PyObject* item) |
| 1721 | { |
| 1722 | ElementObject* self = (ElementObject*) self_; |
| 1723 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1724 | if (PyIndex_Check(item)) { |
| 1725 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1726 | |
| 1727 | if (i == -1 && PyErr_Occurred()) { |
| 1728 | return NULL; |
| 1729 | } |
| 1730 | if (i < 0 && self->extra) |
| 1731 | i += self->extra->length; |
| 1732 | return element_getitem(self_, i); |
| 1733 | } |
| 1734 | else if (PySlice_Check(item)) { |
| 1735 | Py_ssize_t start, stop, step, slicelen, cur, i; |
| 1736 | PyObject* list; |
| 1737 | |
| 1738 | if (!self->extra) |
| 1739 | return PyList_New(0); |
| 1740 | |
Serhiy Storchaka | b879fe8 | 2017-04-08 09:53:51 +0300 | [diff] [blame] | 1741 | if (PySlice_Unpack(item, &start, &stop, &step) < 0) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1742 | return NULL; |
| 1743 | } |
Serhiy Storchaka | b879fe8 | 2017-04-08 09:53:51 +0300 | [diff] [blame] | 1744 | slicelen = PySlice_AdjustIndices(self->extra->length, &start, &stop, |
| 1745 | step); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1746 | |
| 1747 | if (slicelen <= 0) |
| 1748 | return PyList_New(0); |
| 1749 | else { |
| 1750 | list = PyList_New(slicelen); |
| 1751 | if (!list) |
| 1752 | return NULL; |
| 1753 | |
| 1754 | for (cur = start, i = 0; i < slicelen; |
| 1755 | cur += step, i++) { |
| 1756 | PyObject* item = self->extra->children[cur]; |
| 1757 | Py_INCREF(item); |
| 1758 | PyList_SET_ITEM(list, i, item); |
| 1759 | } |
| 1760 | |
| 1761 | return list; |
| 1762 | } |
| 1763 | } |
| 1764 | else { |
| 1765 | PyErr_SetString(PyExc_TypeError, |
| 1766 | "element indices must be integers"); |
| 1767 | return NULL; |
| 1768 | } |
| 1769 | } |
| 1770 | |
| 1771 | static int |
| 1772 | element_ass_subscr(PyObject* self_, PyObject* item, PyObject* value) |
| 1773 | { |
| 1774 | ElementObject* self = (ElementObject*) self_; |
| 1775 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1776 | if (PyIndex_Check(item)) { |
| 1777 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1778 | |
| 1779 | if (i == -1 && PyErr_Occurred()) { |
| 1780 | return -1; |
| 1781 | } |
| 1782 | if (i < 0 && self->extra) |
| 1783 | i += self->extra->length; |
| 1784 | return element_setitem(self_, i, value); |
| 1785 | } |
| 1786 | else if (PySlice_Check(item)) { |
| 1787 | Py_ssize_t start, stop, step, slicelen, newlen, cur, i; |
| 1788 | |
| 1789 | PyObject* recycle = NULL; |
Serhiy Storchaka | 04d759b | 2015-11-22 12:18:38 +0200 | [diff] [blame] | 1790 | PyObject* seq; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1791 | |
Victor Stinner | 5f0af23 | 2013-07-11 23:01:36 +0200 | [diff] [blame] | 1792 | if (!self->extra) { |
| 1793 | if (create_extra(self, NULL) < 0) |
| 1794 | return -1; |
| 1795 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1796 | |
Serhiy Storchaka | b879fe8 | 2017-04-08 09:53:51 +0300 | [diff] [blame] | 1797 | if (PySlice_Unpack(item, &start, &stop, &step) < 0) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1798 | return -1; |
| 1799 | } |
Serhiy Storchaka | b879fe8 | 2017-04-08 09:53:51 +0300 | [diff] [blame] | 1800 | slicelen = PySlice_AdjustIndices(self->extra->length, &start, &stop, |
| 1801 | step); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1802 | |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1803 | if (value == NULL) { |
| 1804 | /* Delete slice */ |
| 1805 | size_t cur; |
| 1806 | Py_ssize_t i; |
| 1807 | |
| 1808 | if (slicelen <= 0) |
| 1809 | return 0; |
| 1810 | |
| 1811 | /* Since we're deleting, the direction of the range doesn't matter, |
| 1812 | * so for simplicity make it always ascending. |
| 1813 | */ |
| 1814 | if (step < 0) { |
| 1815 | stop = start + 1; |
| 1816 | start = stop + step * (slicelen - 1) - 1; |
| 1817 | step = -step; |
| 1818 | } |
| 1819 | |
Benjamin Peterson | 2f8bfef | 2016-09-07 09:26:18 -0700 | [diff] [blame] | 1820 | assert((size_t)slicelen <= SIZE_MAX / sizeof(PyObject *)); |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1821 | |
| 1822 | /* recycle is a list that will contain all the children |
| 1823 | * scheduled for removal. |
| 1824 | */ |
| 1825 | if (!(recycle = PyList_New(slicelen))) { |
| 1826 | PyErr_NoMemory(); |
| 1827 | return -1; |
| 1828 | } |
| 1829 | |
| 1830 | /* This loop walks over all the children that have to be deleted, |
| 1831 | * with cur pointing at them. num_moved is the amount of children |
| 1832 | * until the next deleted child that have to be "shifted down" to |
| 1833 | * occupy the deleted's places. |
| 1834 | * Note that in the ith iteration, shifting is done i+i places down |
| 1835 | * because i children were already removed. |
| 1836 | */ |
| 1837 | for (cur = start, i = 0; cur < (size_t)stop; cur += step, ++i) { |
| 1838 | /* Compute how many children have to be moved, clipping at the |
| 1839 | * list end. |
| 1840 | */ |
| 1841 | Py_ssize_t num_moved = step - 1; |
| 1842 | if (cur + step >= (size_t)self->extra->length) { |
| 1843 | num_moved = self->extra->length - cur - 1; |
| 1844 | } |
| 1845 | |
| 1846 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1847 | |
| 1848 | memmove( |
| 1849 | self->extra->children + cur - i, |
| 1850 | self->extra->children + cur + 1, |
| 1851 | num_moved * sizeof(PyObject *)); |
| 1852 | } |
| 1853 | |
| 1854 | /* Leftover "tail" after the last removed child */ |
| 1855 | cur = start + (size_t)slicelen * step; |
| 1856 | if (cur < (size_t)self->extra->length) { |
| 1857 | memmove( |
| 1858 | self->extra->children + cur - slicelen, |
| 1859 | self->extra->children + cur, |
| 1860 | (self->extra->length - cur) * sizeof(PyObject *)); |
| 1861 | } |
| 1862 | |
| 1863 | self->extra->length -= slicelen; |
| 1864 | |
| 1865 | /* Discard the recycle list with all the deleted sub-elements */ |
| 1866 | Py_XDECREF(recycle); |
| 1867 | return 0; |
| 1868 | } |
Serhiy Storchaka | 04d759b | 2015-11-22 12:18:38 +0200 | [diff] [blame] | 1869 | |
| 1870 | /* A new slice is actually being assigned */ |
| 1871 | seq = PySequence_Fast(value, ""); |
| 1872 | if (!seq) { |
| 1873 | PyErr_Format( |
| 1874 | PyExc_TypeError, |
| 1875 | "expected sequence, not \"%.200s\"", Py_TYPE(value)->tp_name |
| 1876 | ); |
| 1877 | return -1; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1878 | } |
Serhiy Storchaka | bf623ae | 2017-04-19 20:03:52 +0300 | [diff] [blame] | 1879 | newlen = PySequence_Fast_GET_SIZE(seq); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1880 | |
| 1881 | if (step != 1 && newlen != slicelen) |
| 1882 | { |
Serhiy Storchaka | 04d759b | 2015-11-22 12:18:38 +0200 | [diff] [blame] | 1883 | Py_DECREF(seq); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1884 | PyErr_Format(PyExc_ValueError, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1885 | "attempt to assign sequence of size %zd " |
| 1886 | "to extended slice of size %zd", |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1887 | newlen, slicelen |
| 1888 | ); |
| 1889 | return -1; |
| 1890 | } |
| 1891 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1892 | /* Resize before creating the recycle bin, to prevent refleaks. */ |
| 1893 | if (newlen > slicelen) { |
| 1894 | if (element_resize(self, newlen - slicelen) < 0) { |
Serhiy Storchaka | 04d759b | 2015-11-22 12:18:38 +0200 | [diff] [blame] | 1895 | Py_DECREF(seq); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1896 | return -1; |
| 1897 | } |
| 1898 | } |
| 1899 | |
| 1900 | if (slicelen > 0) { |
| 1901 | /* to avoid recursive calls to this method (via decref), move |
| 1902 | old items to the recycle bin here, and get rid of them when |
| 1903 | we're done modifying the element */ |
| 1904 | recycle = PyList_New(slicelen); |
| 1905 | if (!recycle) { |
Serhiy Storchaka | 04d759b | 2015-11-22 12:18:38 +0200 | [diff] [blame] | 1906 | Py_DECREF(seq); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1907 | return -1; |
| 1908 | } |
| 1909 | for (cur = start, i = 0; i < slicelen; |
| 1910 | cur += step, i++) |
| 1911 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1912 | } |
| 1913 | |
| 1914 | if (newlen < slicelen) { |
| 1915 | /* delete slice */ |
| 1916 | for (i = stop; i < self->extra->length; i++) |
| 1917 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1918 | } else if (newlen > slicelen) { |
| 1919 | /* insert slice */ |
| 1920 | for (i = self->extra->length-1; i >= stop; i--) |
| 1921 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1922 | } |
| 1923 | |
| 1924 | /* replace the slice */ |
| 1925 | for (cur = start, i = 0; i < newlen; |
| 1926 | cur += step, i++) { |
| 1927 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
| 1928 | Py_INCREF(element); |
| 1929 | self->extra->children[cur] = element; |
| 1930 | } |
| 1931 | |
| 1932 | self->extra->length += newlen - slicelen; |
| 1933 | |
Serhiy Storchaka | 04d759b | 2015-11-22 12:18:38 +0200 | [diff] [blame] | 1934 | Py_DECREF(seq); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1935 | |
| 1936 | /* discard the recycle bin, and everything in it */ |
| 1937 | Py_XDECREF(recycle); |
| 1938 | |
| 1939 | return 0; |
| 1940 | } |
| 1941 | else { |
| 1942 | PyErr_SetString(PyExc_TypeError, |
| 1943 | "element indices must be integers"); |
| 1944 | return -1; |
| 1945 | } |
| 1946 | } |
| 1947 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1948 | static PyObject* |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1949 | element_tag_getter(ElementObject *self, void *closure) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1950 | { |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1951 | PyObject *res = self->tag; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1952 | Py_INCREF(res); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1953 | return res; |
| 1954 | } |
| 1955 | |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1956 | static PyObject* |
| 1957 | element_text_getter(ElementObject *self, void *closure) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1958 | { |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1959 | PyObject *res = element_get_text(self); |
| 1960 | Py_XINCREF(res); |
| 1961 | return res; |
| 1962 | } |
Serhiy Storchaka | b6aa537 | 2015-11-23 08:42:25 +0200 | [diff] [blame] | 1963 | |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1964 | static PyObject* |
| 1965 | element_tail_getter(ElementObject *self, void *closure) |
| 1966 | { |
| 1967 | PyObject *res = element_get_tail(self); |
| 1968 | Py_XINCREF(res); |
| 1969 | return res; |
| 1970 | } |
| 1971 | |
| 1972 | static PyObject* |
| 1973 | element_attrib_getter(ElementObject *self, void *closure) |
| 1974 | { |
| 1975 | PyObject *res; |
| 1976 | if (!self->extra) { |
| 1977 | if (create_extra(self, NULL) < 0) |
| 1978 | return NULL; |
Serhiy Storchaka | b6aa537 | 2015-11-23 08:42:25 +0200 | [diff] [blame] | 1979 | } |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1980 | res = element_get_attrib(self); |
| 1981 | Py_XINCREF(res); |
| 1982 | return res; |
| 1983 | } |
Victor Stinner | 4d46343 | 2013-07-11 23:05:03 +0200 | [diff] [blame] | 1984 | |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1985 | /* macro for setter validation */ |
| 1986 | #define _VALIDATE_ATTR_VALUE(V) \ |
| 1987 | if ((V) == NULL) { \ |
| 1988 | PyErr_SetString( \ |
| 1989 | PyExc_AttributeError, \ |
| 1990 | "can't delete element attribute"); \ |
| 1991 | return -1; \ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1992 | } |
| 1993 | |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 1994 | static int |
| 1995 | element_tag_setter(ElementObject *self, PyObject *value, void *closure) |
| 1996 | { |
| 1997 | _VALIDATE_ATTR_VALUE(value); |
| 1998 | Py_INCREF(value); |
Serhiy Storchaka | f01e408 | 2016-04-10 18:12:01 +0300 | [diff] [blame] | 1999 | Py_SETREF(self->tag, value); |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 2000 | return 0; |
| 2001 | } |
| 2002 | |
| 2003 | static int |
| 2004 | element_text_setter(ElementObject *self, PyObject *value, void *closure) |
| 2005 | { |
| 2006 | _VALIDATE_ATTR_VALUE(value); |
| 2007 | Py_INCREF(value); |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 2008 | _set_joined_ptr(&self->text, value); |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 2009 | return 0; |
| 2010 | } |
| 2011 | |
| 2012 | static int |
| 2013 | element_tail_setter(ElementObject *self, PyObject *value, void *closure) |
| 2014 | { |
| 2015 | _VALIDATE_ATTR_VALUE(value); |
| 2016 | Py_INCREF(value); |
Oren Milman | 39ecb9c | 2017-10-10 23:26:24 +0300 | [diff] [blame] | 2017 | _set_joined_ptr(&self->tail, value); |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 2018 | return 0; |
| 2019 | } |
| 2020 | |
| 2021 | static int |
| 2022 | element_attrib_setter(ElementObject *self, PyObject *value, void *closure) |
| 2023 | { |
| 2024 | _VALIDATE_ATTR_VALUE(value); |
| 2025 | if (!self->extra) { |
| 2026 | if (create_extra(self, NULL) < 0) |
| 2027 | return -1; |
| 2028 | } |
| 2029 | Py_INCREF(value); |
Serhiy Storchaka | f01e408 | 2016-04-10 18:12:01 +0300 | [diff] [blame] | 2030 | Py_SETREF(self->extra->attrib, value); |
Eli Bendersky | ef9683b | 2013-05-18 07:52:34 -0700 | [diff] [blame] | 2031 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2032 | } |
| 2033 | |
| 2034 | static PySequenceMethods element_as_sequence = { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 2035 | (lenfunc) element_length, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2036 | 0, /* sq_concat */ |
| 2037 | 0, /* sq_repeat */ |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 2038 | element_getitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2039 | 0, |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 2040 | element_setitem, |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2041 | 0, |
| 2042 | }; |
| 2043 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2044 | /******************************* Element iterator ****************************/ |
| 2045 | |
| 2046 | /* ElementIterObject represents the iteration state over an XML element in |
| 2047 | * pre-order traversal. To keep track of which sub-element should be returned |
| 2048 | * next, a stack of parents is maintained. This is a standard stack-based |
| 2049 | * iterative pre-order traversal of a tree. |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2050 | * The stack is managed using a continuous array. |
| 2051 | * Each stack item contains the saved parent to which we should return after |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2052 | * the current one is exhausted, and the next child to examine in that parent. |
| 2053 | */ |
| 2054 | typedef struct ParentLocator_t { |
| 2055 | ElementObject *parent; |
| 2056 | Py_ssize_t child_index; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2057 | } ParentLocator; |
| 2058 | |
| 2059 | typedef struct { |
| 2060 | PyObject_HEAD |
| 2061 | ParentLocator *parent_stack; |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2062 | Py_ssize_t parent_stack_used; |
| 2063 | Py_ssize_t parent_stack_size; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2064 | ElementObject *root_element; |
| 2065 | PyObject *sought_tag; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2066 | int gettext; |
| 2067 | } ElementIterObject; |
| 2068 | |
| 2069 | |
| 2070 | static void |
| 2071 | elementiter_dealloc(ElementIterObject *it) |
| 2072 | { |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2073 | Py_ssize_t i = it->parent_stack_used; |
| 2074 | it->parent_stack_used = 0; |
INADA Naoki | a6296d3 | 2017-08-24 14:55:17 +0900 | [diff] [blame] | 2075 | /* bpo-31095: UnTrack is needed before calling any callbacks */ |
| 2076 | PyObject_GC_UnTrack(it); |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2077 | while (i--) |
| 2078 | Py_XDECREF(it->parent_stack[i].parent); |
| 2079 | PyMem_Free(it->parent_stack); |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2080 | |
| 2081 | Py_XDECREF(it->sought_tag); |
| 2082 | Py_XDECREF(it->root_element); |
| 2083 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2084 | PyObject_GC_Del(it); |
| 2085 | } |
| 2086 | |
| 2087 | static int |
| 2088 | elementiter_traverse(ElementIterObject *it, visitproc visit, void *arg) |
| 2089 | { |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2090 | Py_ssize_t i = it->parent_stack_used; |
| 2091 | while (i--) |
| 2092 | Py_VISIT(it->parent_stack[i].parent); |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2093 | |
| 2094 | Py_VISIT(it->root_element); |
| 2095 | Py_VISIT(it->sought_tag); |
| 2096 | return 0; |
| 2097 | } |
| 2098 | |
| 2099 | /* Helper function for elementiter_next. Add a new parent to the parent stack. |
| 2100 | */ |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2101 | static int |
| 2102 | parent_stack_push_new(ElementIterObject *it, ElementObject *parent) |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2103 | { |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2104 | ParentLocator *item; |
| 2105 | |
| 2106 | if (it->parent_stack_used >= it->parent_stack_size) { |
| 2107 | Py_ssize_t new_size = it->parent_stack_size * 2; /* never overflow */ |
| 2108 | ParentLocator *parent_stack = it->parent_stack; |
| 2109 | PyMem_Resize(parent_stack, ParentLocator, new_size); |
| 2110 | if (parent_stack == NULL) |
| 2111 | return -1; |
| 2112 | it->parent_stack = parent_stack; |
| 2113 | it->parent_stack_size = new_size; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2114 | } |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2115 | item = it->parent_stack + it->parent_stack_used++; |
| 2116 | Py_INCREF(parent); |
| 2117 | item->parent = parent; |
| 2118 | item->child_index = 0; |
| 2119 | return 0; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2120 | } |
| 2121 | |
| 2122 | static PyObject * |
| 2123 | elementiter_next(ElementIterObject *it) |
| 2124 | { |
| 2125 | /* Sub-element iterator. |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 2126 | * |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2127 | * A short note on gettext: this function serves both the iter() and |
| 2128 | * itertext() methods to avoid code duplication. However, there are a few |
| 2129 | * small differences in the way these iterations work. Namely: |
| 2130 | * - itertext() only yields text from nodes that have it, and continues |
| 2131 | * iterating when a node doesn't have text (so it doesn't return any |
| 2132 | * node like iter()) |
| 2133 | * - itertext() also has to handle tail, after finishing with all the |
| 2134 | * children of a node. |
| 2135 | */ |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 2136 | int rc; |
Serhiy Storchaka | 66c08d9 | 2015-12-21 11:09:48 +0200 | [diff] [blame] | 2137 | ElementObject *elem; |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2138 | PyObject *text; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2139 | |
| 2140 | while (1) { |
| 2141 | /* Handle the case reached in the beginning and end of iteration, where |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2142 | * the parent stack is empty. If root_element is NULL and we're here, the |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2143 | * iterator is exhausted. |
| 2144 | */ |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2145 | if (!it->parent_stack_used) { |
| 2146 | if (!it->root_element) { |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2147 | PyErr_SetNone(PyExc_StopIteration); |
| 2148 | return NULL; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2149 | } |
| 2150 | |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2151 | elem = it->root_element; /* steals a reference */ |
| 2152 | it->root_element = NULL; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2153 | } |
| 2154 | else { |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2155 | /* See if there are children left to traverse in the current parent. If |
| 2156 | * yes, visit the next child. If not, pop the stack and try again. |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2157 | */ |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2158 | ParentLocator *item = &it->parent_stack[it->parent_stack_used - 1]; |
| 2159 | Py_ssize_t child_index = item->child_index; |
| 2160 | ElementObjectExtra *extra; |
| 2161 | elem = item->parent; |
| 2162 | extra = elem->extra; |
| 2163 | if (!extra || child_index >= extra->length) { |
| 2164 | it->parent_stack_used--; |
| 2165 | /* Note that extra condition on it->parent_stack_used here; |
| 2166 | * this is because itertext() is supposed to only return *inner* |
| 2167 | * text, not text following the element it began iteration with. |
| 2168 | */ |
| 2169 | if (it->gettext && it->parent_stack_used) { |
| 2170 | text = element_get_tail(elem); |
| 2171 | goto gettext; |
| 2172 | } |
| 2173 | Py_DECREF(elem); |
| 2174 | continue; |
Serhiy Storchaka | 66c08d9 | 2015-12-21 11:09:48 +0200 | [diff] [blame] | 2175 | } |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2176 | |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2177 | if (!PyObject_TypeCheck(extra->children[child_index], &Element_Type)) { |
| 2178 | PyErr_Format(PyExc_AttributeError, |
| 2179 | "'%.100s' object has no attribute 'iter'", |
| 2180 | Py_TYPE(extra->children[child_index])->tp_name); |
| 2181 | return NULL; |
| 2182 | } |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2183 | elem = (ElementObject *)extra->children[child_index]; |
| 2184 | item->child_index++; |
| 2185 | Py_INCREF(elem); |
| 2186 | } |
| 2187 | |
| 2188 | if (parent_stack_push_new(it, elem) < 0) { |
| 2189 | Py_DECREF(elem); |
| 2190 | PyErr_NoMemory(); |
| 2191 | return NULL; |
| 2192 | } |
| 2193 | if (it->gettext) { |
| 2194 | text = element_get_text(elem); |
| 2195 | goto gettext; |
| 2196 | } |
| 2197 | |
| 2198 | if (it->sought_tag == Py_None) |
| 2199 | return (PyObject *)elem; |
| 2200 | |
| 2201 | rc = PyObject_RichCompareBool(elem->tag, it->sought_tag, Py_EQ); |
| 2202 | if (rc > 0) |
| 2203 | return (PyObject *)elem; |
| 2204 | |
| 2205 | Py_DECREF(elem); |
| 2206 | if (rc < 0) |
| 2207 | return NULL; |
| 2208 | continue; |
| 2209 | |
| 2210 | gettext: |
| 2211 | if (!text) { |
| 2212 | Py_DECREF(elem); |
| 2213 | return NULL; |
| 2214 | } |
| 2215 | if (text == Py_None) { |
| 2216 | Py_DECREF(elem); |
| 2217 | } |
| 2218 | else { |
| 2219 | Py_INCREF(text); |
| 2220 | Py_DECREF(elem); |
| 2221 | rc = PyObject_IsTrue(text); |
| 2222 | if (rc > 0) |
| 2223 | return text; |
| 2224 | Py_DECREF(text); |
| 2225 | if (rc < 0) |
| 2226 | return NULL; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2227 | } |
| 2228 | } |
| 2229 | |
| 2230 | return NULL; |
| 2231 | } |
| 2232 | |
| 2233 | |
| 2234 | static PyTypeObject ElementIter_Type = { |
| 2235 | PyVarObject_HEAD_INIT(NULL, 0) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 2236 | /* Using the module's name since the pure-Python implementation does not |
| 2237 | have such a type. */ |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2238 | "_elementtree._element_iterator", /* tp_name */ |
| 2239 | sizeof(ElementIterObject), /* tp_basicsize */ |
| 2240 | 0, /* tp_itemsize */ |
| 2241 | /* methods */ |
| 2242 | (destructor)elementiter_dealloc, /* tp_dealloc */ |
| 2243 | 0, /* tp_print */ |
| 2244 | 0, /* tp_getattr */ |
| 2245 | 0, /* tp_setattr */ |
| 2246 | 0, /* tp_reserved */ |
| 2247 | 0, /* tp_repr */ |
| 2248 | 0, /* tp_as_number */ |
| 2249 | 0, /* tp_as_sequence */ |
| 2250 | 0, /* tp_as_mapping */ |
| 2251 | 0, /* tp_hash */ |
| 2252 | 0, /* tp_call */ |
| 2253 | 0, /* tp_str */ |
| 2254 | 0, /* tp_getattro */ |
| 2255 | 0, /* tp_setattro */ |
| 2256 | 0, /* tp_as_buffer */ |
| 2257 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /* tp_flags */ |
| 2258 | 0, /* tp_doc */ |
| 2259 | (traverseproc)elementiter_traverse, /* tp_traverse */ |
| 2260 | 0, /* tp_clear */ |
| 2261 | 0, /* tp_richcompare */ |
| 2262 | 0, /* tp_weaklistoffset */ |
| 2263 | PyObject_SelfIter, /* tp_iter */ |
| 2264 | (iternextfunc)elementiter_next, /* tp_iternext */ |
| 2265 | 0, /* tp_methods */ |
| 2266 | 0, /* tp_members */ |
| 2267 | 0, /* tp_getset */ |
| 2268 | 0, /* tp_base */ |
| 2269 | 0, /* tp_dict */ |
| 2270 | 0, /* tp_descr_get */ |
| 2271 | 0, /* tp_descr_set */ |
| 2272 | 0, /* tp_dictoffset */ |
| 2273 | 0, /* tp_init */ |
| 2274 | 0, /* tp_alloc */ |
| 2275 | 0, /* tp_new */ |
| 2276 | }; |
| 2277 | |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2278 | #define INIT_PARENT_STACK_SIZE 8 |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2279 | |
| 2280 | static PyObject * |
| 2281 | create_elementiter(ElementObject *self, PyObject *tag, int gettext) |
| 2282 | { |
| 2283 | ElementIterObject *it; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2284 | |
| 2285 | it = PyObject_GC_New(ElementIterObject, &ElementIter_Type); |
| 2286 | if (!it) |
| 2287 | return NULL; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2288 | |
Victor Stinner | 4d46343 | 2013-07-11 23:05:03 +0200 | [diff] [blame] | 2289 | Py_INCREF(tag); |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2290 | it->sought_tag = tag; |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2291 | it->gettext = gettext; |
Victor Stinner | 4d46343 | 2013-07-11 23:05:03 +0200 | [diff] [blame] | 2292 | Py_INCREF(self); |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2293 | it->root_element = self; |
| 2294 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2295 | PyObject_GC_Track(it); |
Victor Stinner | d917dcb | 2013-07-12 02:05:17 +0200 | [diff] [blame] | 2296 | |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2297 | it->parent_stack = PyMem_New(ParentLocator, INIT_PARENT_STACK_SIZE); |
Victor Stinner | d917dcb | 2013-07-12 02:05:17 +0200 | [diff] [blame] | 2298 | if (it->parent_stack == NULL) { |
| 2299 | Py_DECREF(it); |
| 2300 | PyErr_NoMemory(); |
| 2301 | return NULL; |
| 2302 | } |
Serhiy Storchaka | 22adf2a | 2015-12-21 12:43:54 +0200 | [diff] [blame] | 2303 | it->parent_stack_used = 0; |
| 2304 | it->parent_stack_size = INIT_PARENT_STACK_SIZE; |
Victor Stinner | d917dcb | 2013-07-12 02:05:17 +0200 | [diff] [blame] | 2305 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2306 | return (PyObject *)it; |
| 2307 | } |
| 2308 | |
| 2309 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2310 | /* ==================================================================== */ |
| 2311 | /* the tree builder type */ |
| 2312 | |
| 2313 | typedef struct { |
| 2314 | PyObject_HEAD |
| 2315 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2316 | PyObject *root; /* root node (first created node) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2317 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2318 | PyObject *this; /* current node */ |
| 2319 | PyObject *last; /* most recently created node */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2320 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2321 | PyObject *data; /* data collector (string or list), or NULL */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2322 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2323 | PyObject *stack; /* element stack */ |
| 2324 | Py_ssize_t index; /* current stack size (0 means empty) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2325 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2326 | PyObject *element_factory; |
| 2327 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2328 | /* element tracing */ |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 2329 | PyObject *events_append; /* the append method of the list of events, or NULL */ |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2330 | PyObject *start_event_obj; /* event objects (NULL to ignore) */ |
| 2331 | PyObject *end_event_obj; |
| 2332 | PyObject *start_ns_event_obj; |
| 2333 | PyObject *end_ns_event_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2334 | } TreeBuilderObject; |
| 2335 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 2336 | #define TreeBuilder_CheckExact(op) (Py_TYPE(op) == &TreeBuilder_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2337 | |
| 2338 | /* -------------------------------------------------------------------- */ |
| 2339 | /* constructor and destructor */ |
| 2340 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2341 | static PyObject * |
| 2342 | treebuilder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2343 | { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2344 | TreeBuilderObject *t = (TreeBuilderObject *)type->tp_alloc(type, 0); |
| 2345 | if (t != NULL) { |
| 2346 | t->root = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2347 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2348 | Py_INCREF(Py_None); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2349 | t->this = Py_None; |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2350 | Py_INCREF(Py_None); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2351 | t->last = Py_None; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2352 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2353 | t->data = NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2354 | t->element_factory = NULL; |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2355 | t->stack = PyList_New(20); |
| 2356 | if (!t->stack) { |
| 2357 | Py_DECREF(t->this); |
| 2358 | Py_DECREF(t->last); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2359 | Py_DECREF((PyObject *) t); |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2360 | return NULL; |
| 2361 | } |
| 2362 | t->index = 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2363 | |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 2364 | t->events_append = NULL; |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2365 | t->start_event_obj = t->end_event_obj = NULL; |
| 2366 | t->start_ns_event_obj = t->end_ns_event_obj = NULL; |
| 2367 | } |
| 2368 | return (PyObject *)t; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2369 | } |
| 2370 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2371 | /*[clinic input] |
| 2372 | _elementtree.TreeBuilder.__init__ |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2373 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2374 | element_factory: object = NULL |
| 2375 | |
| 2376 | [clinic start generated code]*/ |
| 2377 | |
| 2378 | static int |
| 2379 | _elementtree_TreeBuilder___init___impl(TreeBuilderObject *self, |
| 2380 | PyObject *element_factory) |
| 2381 | /*[clinic end generated code: output=91cfa7558970ee96 input=1b424eeefc35249c]*/ |
| 2382 | { |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2383 | if (element_factory) { |
| 2384 | Py_INCREF(element_factory); |
Serhiy Storchaka | ec39756 | 2016-04-06 09:50:03 +0300 | [diff] [blame] | 2385 | Py_XSETREF(self->element_factory, element_factory); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2386 | } |
| 2387 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2388 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2389 | } |
| 2390 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2391 | static int |
| 2392 | treebuilder_gc_traverse(TreeBuilderObject *self, visitproc visit, void *arg) |
| 2393 | { |
| 2394 | Py_VISIT(self->root); |
| 2395 | Py_VISIT(self->this); |
| 2396 | Py_VISIT(self->last); |
| 2397 | Py_VISIT(self->data); |
| 2398 | Py_VISIT(self->stack); |
| 2399 | Py_VISIT(self->element_factory); |
| 2400 | return 0; |
| 2401 | } |
| 2402 | |
| 2403 | static int |
| 2404 | treebuilder_gc_clear(TreeBuilderObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2405 | { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2406 | Py_CLEAR(self->end_ns_event_obj); |
| 2407 | Py_CLEAR(self->start_ns_event_obj); |
| 2408 | Py_CLEAR(self->end_event_obj); |
| 2409 | Py_CLEAR(self->start_event_obj); |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 2410 | Py_CLEAR(self->events_append); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2411 | Py_CLEAR(self->stack); |
| 2412 | Py_CLEAR(self->data); |
| 2413 | Py_CLEAR(self->last); |
| 2414 | Py_CLEAR(self->this); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2415 | Py_CLEAR(self->element_factory); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2416 | Py_CLEAR(self->root); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2417 | return 0; |
| 2418 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2419 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2420 | static void |
| 2421 | treebuilder_dealloc(TreeBuilderObject *self) |
| 2422 | { |
| 2423 | PyObject_GC_UnTrack(self); |
| 2424 | treebuilder_gc_clear(self); |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2425 | Py_TYPE(self)->tp_free((PyObject *)self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2426 | } |
| 2427 | |
| 2428 | /* -------------------------------------------------------------------- */ |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2429 | /* helpers for handling of arbitrary element-like objects */ |
| 2430 | |
| 2431 | static int |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2432 | treebuilder_set_element_text_or_tail(PyObject *element, PyObject **data, |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2433 | PyObject **dest, _Py_Identifier *name) |
| 2434 | { |
| 2435 | if (Element_CheckExact(element)) { |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2436 | PyObject *tmp = JOIN_OBJ(*dest); |
| 2437 | *dest = JOIN_SET(*data, PyList_CheckExact(*data)); |
| 2438 | *data = NULL; |
| 2439 | Py_DECREF(tmp); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2440 | return 0; |
| 2441 | } |
| 2442 | else { |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2443 | PyObject *joined = list_join(*data); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2444 | int r; |
| 2445 | if (joined == NULL) |
| 2446 | return -1; |
| 2447 | r = _PyObject_SetAttrId(element, name, joined); |
| 2448 | Py_DECREF(joined); |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2449 | if (r < 0) |
| 2450 | return -1; |
| 2451 | Py_CLEAR(*data); |
| 2452 | return 0; |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2453 | } |
| 2454 | } |
| 2455 | |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2456 | LOCAL(int) |
| 2457 | treebuilder_flush_data(TreeBuilderObject* self) |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2458 | { |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2459 | PyObject *element = self->last; |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2460 | |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2461 | if (!self->data) { |
| 2462 | return 0; |
| 2463 | } |
| 2464 | |
| 2465 | if (self->this == element) { |
| 2466 | _Py_IDENTIFIER(text); |
| 2467 | return treebuilder_set_element_text_or_tail( |
| 2468 | element, &self->data, |
| 2469 | &((ElementObject *) element)->text, &PyId_text); |
| 2470 | } |
| 2471 | else { |
| 2472 | _Py_IDENTIFIER(tail); |
| 2473 | return treebuilder_set_element_text_or_tail( |
| 2474 | element, &self->data, |
| 2475 | &((ElementObject *) element)->tail, &PyId_tail); |
| 2476 | } |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2477 | } |
| 2478 | |
| 2479 | static int |
| 2480 | treebuilder_add_subelement(PyObject *element, PyObject *child) |
| 2481 | { |
| 2482 | _Py_IDENTIFIER(append); |
| 2483 | if (Element_CheckExact(element)) { |
| 2484 | ElementObject *elem = (ElementObject *) element; |
| 2485 | return element_add_subelement(elem, child); |
| 2486 | } |
| 2487 | else { |
| 2488 | PyObject *res; |
Victor Stinner | f561634 | 2016-12-09 15:26:00 +0100 | [diff] [blame] | 2489 | res = _PyObject_CallMethodIdObjArgs(element, &PyId_append, child, NULL); |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2490 | if (res == NULL) |
| 2491 | return -1; |
| 2492 | Py_DECREF(res); |
| 2493 | return 0; |
| 2494 | } |
| 2495 | } |
| 2496 | |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 2497 | LOCAL(int) |
| 2498 | treebuilder_append_event(TreeBuilderObject *self, PyObject *action, |
| 2499 | PyObject *node) |
| 2500 | { |
| 2501 | if (action != NULL) { |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 2502 | PyObject *res; |
| 2503 | PyObject *event = PyTuple_Pack(2, action, node); |
| 2504 | if (event == NULL) |
| 2505 | return -1; |
Victor Stinner | de4ae3d | 2016-12-04 22:59:09 +0100 | [diff] [blame] | 2506 | res = PyObject_CallFunctionObjArgs(self->events_append, event, NULL); |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 2507 | Py_DECREF(event); |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 2508 | if (res == NULL) |
| 2509 | return -1; |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 2510 | Py_DECREF(res); |
| 2511 | } |
| 2512 | return 0; |
| 2513 | } |
| 2514 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2515 | /* -------------------------------------------------------------------- */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2516 | /* handlers */ |
| 2517 | |
| 2518 | LOCAL(PyObject*) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2519 | treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag, |
| 2520 | PyObject* attrib) |
| 2521 | { |
| 2522 | PyObject* node; |
| 2523 | PyObject* this; |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 2524 | elementtreestate *st = ET_STATE_GLOBAL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2525 | |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2526 | if (treebuilder_flush_data(self) < 0) { |
| 2527 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2528 | } |
| 2529 | |
Serhiy Storchaka | 36ff997 | 2015-12-10 09:51:53 +0200 | [diff] [blame] | 2530 | if (!self->element_factory || self->element_factory == Py_None) { |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2531 | node = create_new_element(tag, attrib); |
Serhiy Storchaka | 36ff997 | 2015-12-10 09:51:53 +0200 | [diff] [blame] | 2532 | } else if (attrib == Py_None) { |
| 2533 | attrib = PyDict_New(); |
| 2534 | if (!attrib) |
| 2535 | return NULL; |
Victor Stinner | 5abaa2b | 2016-12-09 16:22:32 +0100 | [diff] [blame] | 2536 | node = PyObject_CallFunctionObjArgs(self->element_factory, |
| 2537 | tag, attrib, NULL); |
Serhiy Storchaka | 36ff997 | 2015-12-10 09:51:53 +0200 | [diff] [blame] | 2538 | Py_DECREF(attrib); |
| 2539 | } |
| 2540 | else { |
Victor Stinner | 5abaa2b | 2016-12-09 16:22:32 +0100 | [diff] [blame] | 2541 | node = PyObject_CallFunctionObjArgs(self->element_factory, |
| 2542 | tag, attrib, NULL); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2543 | } |
| 2544 | if (!node) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2545 | return NULL; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2546 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2547 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2548 | this = self->this; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2549 | |
| 2550 | if (this != Py_None) { |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2551 | if (treebuilder_add_subelement(this, node) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2552 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2553 | } else { |
| 2554 | if (self->root) { |
| 2555 | PyErr_SetString( |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 2556 | st->parseerror_obj, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2557 | "multiple elements on top level" |
| 2558 | ); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2559 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2560 | } |
| 2561 | Py_INCREF(node); |
| 2562 | self->root = node; |
| 2563 | } |
| 2564 | |
| 2565 | if (self->index < PyList_GET_SIZE(self->stack)) { |
| 2566 | if (PyList_SetItem(self->stack, self->index, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2567 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2568 | Py_INCREF(this); |
| 2569 | } else { |
| 2570 | if (PyList_Append(self->stack, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2571 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2572 | } |
| 2573 | self->index++; |
| 2574 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2575 | Py_INCREF(node); |
Serhiy Storchaka | 57a01d3 | 2016-04-10 18:05:40 +0300 | [diff] [blame] | 2576 | Py_SETREF(self->this, node); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2577 | Py_INCREF(node); |
Serhiy Storchaka | 57a01d3 | 2016-04-10 18:05:40 +0300 | [diff] [blame] | 2578 | Py_SETREF(self->last, node); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2579 | |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 2580 | if (treebuilder_append_event(self, self->start_event_obj, node) < 0) |
| 2581 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2582 | |
| 2583 | return node; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2584 | |
| 2585 | error: |
| 2586 | Py_DECREF(node); |
| 2587 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2588 | } |
| 2589 | |
| 2590 | LOCAL(PyObject*) |
| 2591 | treebuilder_handle_data(TreeBuilderObject* self, PyObject* data) |
| 2592 | { |
| 2593 | if (!self->data) { |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2594 | if (self->last == Py_None) { |
Thomas Wouters | 00ee7ba | 2006-08-21 19:07:27 +0000 | [diff] [blame] | 2595 | /* ignore calls to data before the first call to start */ |
| 2596 | Py_RETURN_NONE; |
| 2597 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2598 | /* store the first item as is */ |
| 2599 | Py_INCREF(data); self->data = data; |
| 2600 | } else { |
| 2601 | /* more than one item; use a list to collect items */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2602 | if (PyBytes_CheckExact(self->data) && Py_REFCNT(self->data) == 1 && |
| 2603 | PyBytes_CheckExact(data) && PyBytes_GET_SIZE(data) == 1) { |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2604 | /* XXX this code path unused in Python 3? */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2605 | /* expat often generates single character data sections; handle |
| 2606 | the most common case by resizing the existing string... */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2607 | Py_ssize_t size = PyBytes_GET_SIZE(self->data); |
| 2608 | if (_PyBytes_Resize(&self->data, size + 1) < 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2609 | return NULL; |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2610 | PyBytes_AS_STRING(self->data)[size] = PyBytes_AS_STRING(data)[0]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2611 | } else if (PyList_CheckExact(self->data)) { |
| 2612 | if (PyList_Append(self->data, data) < 0) |
| 2613 | return NULL; |
| 2614 | } else { |
| 2615 | PyObject* list = PyList_New(2); |
| 2616 | if (!list) |
| 2617 | return NULL; |
| 2618 | PyList_SET_ITEM(list, 0, self->data); |
| 2619 | Py_INCREF(data); PyList_SET_ITEM(list, 1, data); |
| 2620 | self->data = list; |
| 2621 | } |
| 2622 | } |
| 2623 | |
| 2624 | Py_RETURN_NONE; |
| 2625 | } |
| 2626 | |
| 2627 | LOCAL(PyObject*) |
| 2628 | treebuilder_handle_end(TreeBuilderObject* self, PyObject* tag) |
| 2629 | { |
| 2630 | PyObject* item; |
| 2631 | |
Serhiy Storchaka | 576def0 | 2017-03-30 09:47:31 +0300 | [diff] [blame] | 2632 | if (treebuilder_flush_data(self) < 0) { |
| 2633 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2634 | } |
| 2635 | |
| 2636 | if (self->index == 0) { |
| 2637 | PyErr_SetString( |
| 2638 | PyExc_IndexError, |
| 2639 | "pop from empty stack" |
| 2640 | ); |
| 2641 | return NULL; |
| 2642 | } |
| 2643 | |
Serhiy Storchaka | 191321d | 2015-12-27 15:41:34 +0200 | [diff] [blame] | 2644 | item = self->last; |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2645 | self->last = self->this; |
Serhiy Storchaka | 191321d | 2015-12-27 15:41:34 +0200 | [diff] [blame] | 2646 | self->index--; |
| 2647 | self->this = PyList_GET_ITEM(self->stack, self->index); |
| 2648 | Py_INCREF(self->this); |
| 2649 | Py_DECREF(item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2650 | |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 2651 | if (treebuilder_append_event(self, self->end_event_obj, self->last) < 0) |
| 2652 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2653 | |
| 2654 | Py_INCREF(self->last); |
| 2655 | return (PyObject*) self->last; |
| 2656 | } |
| 2657 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2658 | /* -------------------------------------------------------------------- */ |
| 2659 | /* methods (in alphabetical order) */ |
| 2660 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2661 | /*[clinic input] |
| 2662 | _elementtree.TreeBuilder.data |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2663 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2664 | data: object |
| 2665 | / |
| 2666 | |
| 2667 | [clinic start generated code]*/ |
| 2668 | |
| 2669 | static PyObject * |
| 2670 | _elementtree_TreeBuilder_data(TreeBuilderObject *self, PyObject *data) |
| 2671 | /*[clinic end generated code: output=69144c7100795bb2 input=a0540c532b284d29]*/ |
| 2672 | { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2673 | return treebuilder_handle_data(self, data); |
| 2674 | } |
| 2675 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2676 | /*[clinic input] |
| 2677 | _elementtree.TreeBuilder.end |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2678 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2679 | tag: object |
| 2680 | / |
| 2681 | |
| 2682 | [clinic start generated code]*/ |
| 2683 | |
| 2684 | static PyObject * |
| 2685 | _elementtree_TreeBuilder_end(TreeBuilderObject *self, PyObject *tag) |
| 2686 | /*[clinic end generated code: output=9a98727cc691cd9d input=22dc3674236f5745]*/ |
| 2687 | { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2688 | return treebuilder_handle_end(self, tag); |
| 2689 | } |
| 2690 | |
| 2691 | LOCAL(PyObject*) |
| 2692 | treebuilder_done(TreeBuilderObject* self) |
| 2693 | { |
| 2694 | PyObject* res; |
| 2695 | |
| 2696 | /* FIXME: check stack size? */ |
| 2697 | |
| 2698 | if (self->root) |
| 2699 | res = self->root; |
| 2700 | else |
| 2701 | res = Py_None; |
| 2702 | |
| 2703 | Py_INCREF(res); |
| 2704 | return res; |
| 2705 | } |
| 2706 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2707 | /*[clinic input] |
| 2708 | _elementtree.TreeBuilder.close |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2709 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2710 | [clinic start generated code]*/ |
| 2711 | |
| 2712 | static PyObject * |
| 2713 | _elementtree_TreeBuilder_close_impl(TreeBuilderObject *self) |
| 2714 | /*[clinic end generated code: output=b441fee3202f61ee input=f7c9c65dc718de14]*/ |
| 2715 | { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2716 | return treebuilder_done(self); |
| 2717 | } |
| 2718 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2719 | /*[clinic input] |
| 2720 | _elementtree.TreeBuilder.start |
| 2721 | |
| 2722 | tag: object |
| 2723 | attrs: object = None |
| 2724 | / |
| 2725 | |
| 2726 | [clinic start generated code]*/ |
| 2727 | |
| 2728 | static PyObject * |
| 2729 | _elementtree_TreeBuilder_start_impl(TreeBuilderObject *self, PyObject *tag, |
| 2730 | PyObject *attrs) |
| 2731 | /*[clinic end generated code: output=e7e9dc2861349411 input=95fc1758dd042c65]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2732 | { |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 2733 | return treebuilder_handle_start(self, tag, attrs); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2734 | } |
| 2735 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2736 | /* ==================================================================== */ |
| 2737 | /* the expat interface */ |
| 2738 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2739 | #include "expat.h" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2740 | #include "pyexpat.h" |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 2741 | |
| 2742 | /* The PyExpat_CAPI structure is an immutable dispatch table, so it can be |
| 2743 | * cached globally without being in per-module state. |
| 2744 | */ |
Eli Bendersky | 20d4174 | 2012-06-01 09:48:37 +0300 | [diff] [blame] | 2745 | static struct PyExpat_CAPI *expat_capi; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2746 | #define EXPAT(func) (expat_capi->func) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2747 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2748 | static XML_Memory_Handling_Suite ExpatMemoryHandler = { |
| 2749 | PyObject_Malloc, PyObject_Realloc, PyObject_Free}; |
| 2750 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2751 | typedef struct { |
| 2752 | PyObject_HEAD |
| 2753 | |
| 2754 | XML_Parser parser; |
| 2755 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2756 | PyObject *target; |
| 2757 | PyObject *entity; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2758 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2759 | PyObject *names; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2760 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2761 | PyObject *handle_start; |
| 2762 | PyObject *handle_data; |
| 2763 | PyObject *handle_end; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2764 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2765 | PyObject *handle_comment; |
| 2766 | PyObject *handle_pi; |
| 2767 | PyObject *handle_doctype; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2768 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2769 | PyObject *handle_close; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2770 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2771 | } XMLParserObject; |
| 2772 | |
Serhiy Storchaka | 4a01cab | 2015-06-29 23:08:52 +0300 | [diff] [blame] | 2773 | static PyObject* |
Serhiy Storchaka | a5552f0 | 2017-12-15 13:11:11 +0200 | [diff] [blame] | 2774 | _elementtree_XMLParser_doctype(XMLParserObject *self, PyObject *const *args, Py_ssize_t nargs); |
Serhiy Storchaka | 4a01cab | 2015-06-29 23:08:52 +0300 | [diff] [blame] | 2775 | static PyObject * |
| 2776 | _elementtree_XMLParser_doctype_impl(XMLParserObject *self, PyObject *name, |
| 2777 | PyObject *pubid, PyObject *system); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2778 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2779 | /* helpers */ |
| 2780 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2781 | LOCAL(PyObject*) |
| 2782 | makeuniversal(XMLParserObject* self, const char* string) |
| 2783 | { |
| 2784 | /* convert a UTF-8 tag/attribute name from the expat parser |
| 2785 | to a universal name string */ |
| 2786 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2787 | Py_ssize_t size = (Py_ssize_t) strlen(string); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2788 | PyObject* key; |
| 2789 | PyObject* value; |
| 2790 | |
| 2791 | /* look the 'raw' name up in the names dictionary */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2792 | key = PyBytes_FromStringAndSize(string, size); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2793 | if (!key) |
| 2794 | return NULL; |
| 2795 | |
| 2796 | value = PyDict_GetItem(self->names, key); |
| 2797 | |
| 2798 | if (value) { |
| 2799 | Py_INCREF(value); |
| 2800 | } else { |
| 2801 | /* new name. convert to universal name, and decode as |
| 2802 | necessary */ |
| 2803 | |
| 2804 | PyObject* tag; |
| 2805 | char* p; |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 2806 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2807 | |
| 2808 | /* look for namespace separator */ |
| 2809 | for (i = 0; i < size; i++) |
| 2810 | if (string[i] == '}') |
| 2811 | break; |
| 2812 | if (i != size) { |
| 2813 | /* convert to universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2814 | tag = PyBytes_FromStringAndSize(NULL, size+1); |
Victor Stinner | 71c8b7e | 2013-07-11 23:08:39 +0200 | [diff] [blame] | 2815 | if (tag == NULL) { |
| 2816 | Py_DECREF(key); |
| 2817 | return NULL; |
| 2818 | } |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2819 | p = PyBytes_AS_STRING(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2820 | p[0] = '{'; |
| 2821 | memcpy(p+1, string, size); |
| 2822 | size++; |
| 2823 | } else { |
| 2824 | /* plain name; use key as tag */ |
| 2825 | Py_INCREF(key); |
| 2826 | tag = key; |
| 2827 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 2828 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2829 | /* decode universal name */ |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 2830 | p = PyBytes_AS_STRING(tag); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2831 | value = PyUnicode_DecodeUTF8(p, size, "strict"); |
| 2832 | Py_DECREF(tag); |
| 2833 | if (!value) { |
| 2834 | Py_DECREF(key); |
| 2835 | return NULL; |
| 2836 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2837 | |
| 2838 | /* add to names dictionary */ |
| 2839 | if (PyDict_SetItem(self->names, key, value) < 0) { |
| 2840 | Py_DECREF(key); |
| 2841 | Py_DECREF(value); |
| 2842 | return NULL; |
| 2843 | } |
| 2844 | } |
| 2845 | |
| 2846 | Py_DECREF(key); |
| 2847 | return value; |
| 2848 | } |
| 2849 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2850 | /* Set the ParseError exception with the given parameters. |
| 2851 | * If message is not NULL, it's used as the error string. Otherwise, the |
| 2852 | * message string is the default for the given error_code. |
| 2853 | */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2854 | static void |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 2855 | expat_set_error(enum XML_Error error_code, Py_ssize_t line, Py_ssize_t column, |
| 2856 | const char *message) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2857 | { |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2858 | PyObject *errmsg, *error, *position, *code; |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 2859 | elementtreestate *st = ET_STATE_GLOBAL; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2860 | |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 2861 | errmsg = PyUnicode_FromFormat("%s: line %zd, column %zd", |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2862 | message ? message : EXPAT(ErrorString)(error_code), |
| 2863 | line, column); |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2864 | if (errmsg == NULL) |
| 2865 | return; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2866 | |
Victor Stinner | 7bfb42d | 2016-12-05 17:04:32 +0100 | [diff] [blame] | 2867 | error = PyObject_CallFunctionObjArgs(st->parseerror_obj, errmsg, NULL); |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 2868 | Py_DECREF(errmsg); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2869 | if (!error) |
| 2870 | return; |
| 2871 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2872 | /* Add code and position attributes */ |
| 2873 | code = PyLong_FromLong((long)error_code); |
| 2874 | if (!code) { |
| 2875 | Py_DECREF(error); |
| 2876 | return; |
| 2877 | } |
| 2878 | if (PyObject_SetAttrString(error, "code", code) == -1) { |
| 2879 | Py_DECREF(error); |
| 2880 | Py_DECREF(code); |
| 2881 | return; |
| 2882 | } |
| 2883 | Py_DECREF(code); |
| 2884 | |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 2885 | position = Py_BuildValue("(nn)", line, column); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2886 | if (!position) { |
| 2887 | Py_DECREF(error); |
| 2888 | return; |
| 2889 | } |
| 2890 | if (PyObject_SetAttrString(error, "position", position) == -1) { |
| 2891 | Py_DECREF(error); |
| 2892 | Py_DECREF(position); |
| 2893 | return; |
| 2894 | } |
| 2895 | Py_DECREF(position); |
| 2896 | |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 2897 | PyErr_SetObject(st->parseerror_obj, error); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2898 | Py_DECREF(error); |
| 2899 | } |
| 2900 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2901 | /* -------------------------------------------------------------------- */ |
| 2902 | /* handlers */ |
| 2903 | |
| 2904 | static void |
| 2905 | expat_default_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2906 | int data_len) |
| 2907 | { |
| 2908 | PyObject* key; |
| 2909 | PyObject* value; |
| 2910 | PyObject* res; |
| 2911 | |
| 2912 | if (data_len < 2 || data_in[0] != '&') |
| 2913 | return; |
| 2914 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 2915 | if (PyErr_Occurred()) |
| 2916 | return; |
| 2917 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2918 | key = PyUnicode_DecodeUTF8(data_in + 1, data_len - 2, "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2919 | if (!key) |
| 2920 | return; |
| 2921 | |
| 2922 | value = PyDict_GetItem(self->entity, key); |
| 2923 | |
| 2924 | if (value) { |
| 2925 | if (TreeBuilder_CheckExact(self->target)) |
| 2926 | res = treebuilder_handle_data( |
| 2927 | (TreeBuilderObject*) self->target, value |
| 2928 | ); |
| 2929 | else if (self->handle_data) |
Victor Stinner | 7bfb42d | 2016-12-05 17:04:32 +0100 | [diff] [blame] | 2930 | res = PyObject_CallFunctionObjArgs(self->handle_data, value, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2931 | else |
| 2932 | res = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2933 | Py_XDECREF(res); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2934 | } else if (!PyErr_Occurred()) { |
| 2935 | /* Report the first error, not the last */ |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 2936 | char message[128] = "undefined entity "; |
| 2937 | strncat(message, data_in, data_len < 100?data_len:100); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2938 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2939 | XML_ERROR_UNDEFINED_ENTITY, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2940 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2941 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 2942 | message |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2943 | ); |
| 2944 | } |
| 2945 | |
| 2946 | Py_DECREF(key); |
| 2947 | } |
| 2948 | |
| 2949 | static void |
| 2950 | expat_start_handler(XMLParserObject* self, const XML_Char* tag_in, |
| 2951 | const XML_Char **attrib_in) |
| 2952 | { |
| 2953 | PyObject* res; |
| 2954 | PyObject* tag; |
| 2955 | PyObject* attrib; |
| 2956 | int ok; |
| 2957 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 2958 | if (PyErr_Occurred()) |
| 2959 | return; |
| 2960 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2961 | /* tag name */ |
| 2962 | tag = makeuniversal(self, tag_in); |
| 2963 | if (!tag) |
| 2964 | return; /* parser will look for errors */ |
| 2965 | |
| 2966 | /* attributes */ |
| 2967 | if (attrib_in[0]) { |
| 2968 | attrib = PyDict_New(); |
Serhiy Storchaka | a29eb08 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2969 | if (!attrib) { |
| 2970 | Py_DECREF(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2971 | return; |
Serhiy Storchaka | a29eb08 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2972 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2973 | while (attrib_in[0] && attrib_in[1]) { |
| 2974 | PyObject* key = makeuniversal(self, attrib_in[0]); |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 2975 | PyObject* value = PyUnicode_DecodeUTF8(attrib_in[1], strlen(attrib_in[1]), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2976 | if (!key || !value) { |
| 2977 | Py_XDECREF(value); |
| 2978 | Py_XDECREF(key); |
| 2979 | Py_DECREF(attrib); |
Serhiy Storchaka | a29eb08 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2980 | Py_DECREF(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2981 | return; |
| 2982 | } |
| 2983 | ok = PyDict_SetItem(attrib, key, value); |
| 2984 | Py_DECREF(value); |
| 2985 | Py_DECREF(key); |
| 2986 | if (ok < 0) { |
| 2987 | Py_DECREF(attrib); |
Serhiy Storchaka | a29eb08 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2988 | Py_DECREF(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2989 | return; |
| 2990 | } |
| 2991 | attrib_in += 2; |
| 2992 | } |
| 2993 | } else { |
Serhiy Storchaka | 36ff997 | 2015-12-10 09:51:53 +0200 | [diff] [blame] | 2994 | Py_INCREF(Py_None); |
| 2995 | attrib = Py_None; |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2996 | } |
| 2997 | |
| 2998 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2999 | /* shortcut */ |
| 3000 | res = treebuilder_handle_start((TreeBuilderObject*) self->target, |
| 3001 | tag, attrib); |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 3002 | } |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3003 | else if (self->handle_start) { |
Serhiy Storchaka | 36ff997 | 2015-12-10 09:51:53 +0200 | [diff] [blame] | 3004 | if (attrib == Py_None) { |
| 3005 | Py_DECREF(attrib); |
| 3006 | attrib = PyDict_New(); |
| 3007 | if (!attrib) { |
| 3008 | Py_DECREF(tag); |
| 3009 | return; |
| 3010 | } |
| 3011 | } |
Victor Stinner | 5abaa2b | 2016-12-09 16:22:32 +0100 | [diff] [blame] | 3012 | res = PyObject_CallFunctionObjArgs(self->handle_start, |
| 3013 | tag, attrib, NULL); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3014 | } else |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3015 | res = NULL; |
| 3016 | |
| 3017 | Py_DECREF(tag); |
| 3018 | Py_DECREF(attrib); |
| 3019 | |
| 3020 | Py_XDECREF(res); |
| 3021 | } |
| 3022 | |
| 3023 | static void |
| 3024 | expat_data_handler(XMLParserObject* self, const XML_Char* data_in, |
| 3025 | int data_len) |
| 3026 | { |
| 3027 | PyObject* data; |
| 3028 | PyObject* res; |
| 3029 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3030 | if (PyErr_Occurred()) |
| 3031 | return; |
| 3032 | |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 3033 | data = PyUnicode_DecodeUTF8(data_in, data_len, "strict"); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 3034 | if (!data) |
| 3035 | return; /* parser will look for errors */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3036 | |
| 3037 | if (TreeBuilder_CheckExact(self->target)) |
| 3038 | /* shortcut */ |
| 3039 | res = treebuilder_handle_data((TreeBuilderObject*) self->target, data); |
| 3040 | else if (self->handle_data) |
Victor Stinner | 7bfb42d | 2016-12-05 17:04:32 +0100 | [diff] [blame] | 3041 | res = PyObject_CallFunctionObjArgs(self->handle_data, data, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3042 | else |
| 3043 | res = NULL; |
| 3044 | |
| 3045 | Py_DECREF(data); |
| 3046 | |
| 3047 | Py_XDECREF(res); |
| 3048 | } |
| 3049 | |
| 3050 | static void |
| 3051 | expat_end_handler(XMLParserObject* self, const XML_Char* tag_in) |
| 3052 | { |
| 3053 | PyObject* tag; |
| 3054 | PyObject* res = NULL; |
| 3055 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3056 | if (PyErr_Occurred()) |
| 3057 | return; |
| 3058 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3059 | if (TreeBuilder_CheckExact(self->target)) |
| 3060 | /* shortcut */ |
| 3061 | /* the standard tree builder doesn't look at the end tag */ |
| 3062 | res = treebuilder_handle_end( |
| 3063 | (TreeBuilderObject*) self->target, Py_None |
| 3064 | ); |
| 3065 | else if (self->handle_end) { |
| 3066 | tag = makeuniversal(self, tag_in); |
| 3067 | if (tag) { |
Victor Stinner | 7bfb42d | 2016-12-05 17:04:32 +0100 | [diff] [blame] | 3068 | res = PyObject_CallFunctionObjArgs(self->handle_end, tag, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3069 | Py_DECREF(tag); |
| 3070 | } |
| 3071 | } |
| 3072 | |
| 3073 | Py_XDECREF(res); |
| 3074 | } |
| 3075 | |
| 3076 | static void |
| 3077 | expat_start_ns_handler(XMLParserObject* self, const XML_Char* prefix, |
| 3078 | const XML_Char *uri) |
| 3079 | { |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 3080 | TreeBuilderObject *target = (TreeBuilderObject*) self->target; |
| 3081 | PyObject *parcel; |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3082 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3083 | if (PyErr_Occurred()) |
| 3084 | return; |
| 3085 | |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 3086 | if (!target->events_append || !target->start_ns_event_obj) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3087 | return; |
| 3088 | |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 3089 | if (!uri) |
| 3090 | uri = ""; |
| 3091 | if (!prefix) |
| 3092 | prefix = ""; |
| 3093 | |
| 3094 | parcel = Py_BuildValue("ss", prefix, uri); |
| 3095 | if (!parcel) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3096 | return; |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 3097 | treebuilder_append_event(target, target->start_ns_event_obj, parcel); |
| 3098 | Py_DECREF(parcel); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3099 | } |
| 3100 | |
| 3101 | static void |
| 3102 | expat_end_ns_handler(XMLParserObject* self, const XML_Char* prefix_in) |
| 3103 | { |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 3104 | TreeBuilderObject *target = (TreeBuilderObject*) self->target; |
| 3105 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3106 | if (PyErr_Occurred()) |
| 3107 | return; |
| 3108 | |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 3109 | if (!target->events_append) |
Serhiy Storchaka | 7efaf95 | 2015-12-06 23:51:44 +0200 | [diff] [blame] | 3110 | return; |
| 3111 | |
| 3112 | treebuilder_append_event(target, target->end_ns_event_obj, Py_None); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3113 | } |
| 3114 | |
| 3115 | static void |
| 3116 | expat_comment_handler(XMLParserObject* self, const XML_Char* comment_in) |
| 3117 | { |
| 3118 | PyObject* comment; |
| 3119 | PyObject* res; |
| 3120 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3121 | if (PyErr_Occurred()) |
| 3122 | return; |
| 3123 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3124 | if (self->handle_comment) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 3125 | comment = PyUnicode_DecodeUTF8(comment_in, strlen(comment_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3126 | if (comment) { |
Victor Stinner | 7bfb42d | 2016-12-05 17:04:32 +0100 | [diff] [blame] | 3127 | res = PyObject_CallFunctionObjArgs(self->handle_comment, |
| 3128 | comment, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3129 | Py_XDECREF(res); |
| 3130 | Py_DECREF(comment); |
| 3131 | } |
| 3132 | } |
| 3133 | } |
| 3134 | |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 3135 | static void |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3136 | expat_start_doctype_handler(XMLParserObject *self, |
| 3137 | const XML_Char *doctype_name, |
| 3138 | const XML_Char *sysid, |
| 3139 | const XML_Char *pubid, |
| 3140 | int has_internal_subset) |
| 3141 | { |
| 3142 | PyObject *self_pyobj = (PyObject *)self; |
| 3143 | PyObject *doctype_name_obj, *sysid_obj, *pubid_obj; |
| 3144 | PyObject *parser_doctype = NULL; |
| 3145 | PyObject *res = NULL; |
| 3146 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3147 | if (PyErr_Occurred()) |
| 3148 | return; |
| 3149 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3150 | doctype_name_obj = makeuniversal(self, doctype_name); |
| 3151 | if (!doctype_name_obj) |
| 3152 | return; |
| 3153 | |
| 3154 | if (sysid) { |
| 3155 | sysid_obj = makeuniversal(self, sysid); |
| 3156 | if (!sysid_obj) { |
| 3157 | Py_DECREF(doctype_name_obj); |
| 3158 | return; |
| 3159 | } |
| 3160 | } else { |
| 3161 | Py_INCREF(Py_None); |
| 3162 | sysid_obj = Py_None; |
| 3163 | } |
| 3164 | |
| 3165 | if (pubid) { |
| 3166 | pubid_obj = makeuniversal(self, pubid); |
| 3167 | if (!pubid_obj) { |
| 3168 | Py_DECREF(doctype_name_obj); |
| 3169 | Py_DECREF(sysid_obj); |
| 3170 | return; |
| 3171 | } |
| 3172 | } else { |
| 3173 | Py_INCREF(Py_None); |
| 3174 | pubid_obj = Py_None; |
| 3175 | } |
| 3176 | |
| 3177 | /* If the target has a handler for doctype, call it. */ |
| 3178 | if (self->handle_doctype) { |
Victor Stinner | 5abaa2b | 2016-12-09 16:22:32 +0100 | [diff] [blame] | 3179 | res = PyObject_CallFunctionObjArgs(self->handle_doctype, |
| 3180 | doctype_name_obj, pubid_obj, |
| 3181 | sysid_obj, NULL); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3182 | Py_CLEAR(res); |
| 3183 | } |
Serhiy Storchaka | 05744ac | 2015-06-29 22:35:58 +0300 | [diff] [blame] | 3184 | else { |
| 3185 | /* Now see if the parser itself has a doctype method. If yes and it's |
| 3186 | * a custom method, call it but warn about deprecation. If it's only |
| 3187 | * the vanilla XMLParser method, do nothing. |
| 3188 | */ |
| 3189 | parser_doctype = PyObject_GetAttrString(self_pyobj, "doctype"); |
| 3190 | if (parser_doctype && |
| 3191 | !(PyCFunction_Check(parser_doctype) && |
| 3192 | PyCFunction_GET_SELF(parser_doctype) == self_pyobj && |
| 3193 | PyCFunction_GET_FUNCTION(parser_doctype) == |
Serhiy Storchaka | 4a01cab | 2015-06-29 23:08:52 +0300 | [diff] [blame] | 3194 | (PyCFunction) _elementtree_XMLParser_doctype)) { |
| 3195 | res = _elementtree_XMLParser_doctype_impl(self, doctype_name_obj, |
| 3196 | pubid_obj, sysid_obj); |
Serhiy Storchaka | 05744ac | 2015-06-29 22:35:58 +0300 | [diff] [blame] | 3197 | if (!res) |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3198 | goto clear; |
Serhiy Storchaka | 05744ac | 2015-06-29 22:35:58 +0300 | [diff] [blame] | 3199 | Py_DECREF(res); |
Victor Stinner | 5abaa2b | 2016-12-09 16:22:32 +0100 | [diff] [blame] | 3200 | res = PyObject_CallFunctionObjArgs(parser_doctype, |
| 3201 | doctype_name_obj, pubid_obj, |
| 3202 | sysid_obj, NULL); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3203 | Py_CLEAR(res); |
| 3204 | } |
| 3205 | } |
| 3206 | |
| 3207 | clear: |
| 3208 | Py_XDECREF(parser_doctype); |
| 3209 | Py_DECREF(doctype_name_obj); |
| 3210 | Py_DECREF(pubid_obj); |
| 3211 | Py_DECREF(sysid_obj); |
| 3212 | } |
| 3213 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3214 | static void |
| 3215 | expat_pi_handler(XMLParserObject* self, const XML_Char* target_in, |
| 3216 | const XML_Char* data_in) |
| 3217 | { |
| 3218 | PyObject* target; |
| 3219 | PyObject* data; |
| 3220 | PyObject* res; |
| 3221 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3222 | if (PyErr_Occurred()) |
| 3223 | return; |
| 3224 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3225 | if (self->handle_pi) { |
Neal Norwitz | 0269b91 | 2007-08-08 06:56:02 +0000 | [diff] [blame] | 3226 | target = PyUnicode_DecodeUTF8(target_in, strlen(target_in), "strict"); |
| 3227 | data = PyUnicode_DecodeUTF8(data_in, strlen(data_in), "strict"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3228 | if (target && data) { |
Victor Stinner | 5abaa2b | 2016-12-09 16:22:32 +0100 | [diff] [blame] | 3229 | res = PyObject_CallFunctionObjArgs(self->handle_pi, |
| 3230 | target, data, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3231 | Py_XDECREF(res); |
| 3232 | Py_DECREF(data); |
| 3233 | Py_DECREF(target); |
| 3234 | } else { |
| 3235 | Py_XDECREF(data); |
| 3236 | Py_XDECREF(target); |
| 3237 | } |
| 3238 | } |
| 3239 | } |
| 3240 | |
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 | static PyObject * |
| 3244 | xmlparser_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3245 | { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3246 | XMLParserObject *self = (XMLParserObject *)type->tp_alloc(type, 0); |
| 3247 | if (self) { |
| 3248 | self->parser = NULL; |
| 3249 | self->target = self->entity = self->names = NULL; |
| 3250 | self->handle_start = self->handle_data = self->handle_end = NULL; |
| 3251 | self->handle_comment = self->handle_pi = self->handle_close = NULL; |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3252 | self->handle_doctype = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3253 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3254 | return (PyObject *)self; |
| 3255 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3256 | |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3257 | static int |
| 3258 | ignore_attribute_error(PyObject *value) |
| 3259 | { |
| 3260 | if (value == NULL) { |
| 3261 | if (!PyErr_ExceptionMatches(PyExc_AttributeError)) { |
| 3262 | return -1; |
| 3263 | } |
| 3264 | PyErr_Clear(); |
| 3265 | } |
| 3266 | return 0; |
| 3267 | } |
| 3268 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3269 | /*[clinic input] |
| 3270 | _elementtree.XMLParser.__init__ |
| 3271 | |
| 3272 | html: object = NULL |
| 3273 | target: object = NULL |
Larry Hastings | dbfdc38 | 2015-05-04 06:59:46 -0700 | [diff] [blame] | 3274 | encoding: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3275 | |
| 3276 | [clinic start generated code]*/ |
| 3277 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3278 | static int |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3279 | _elementtree_XMLParser___init___impl(XMLParserObject *self, PyObject *html, |
| 3280 | PyObject *target, const char *encoding) |
Larry Hastings | dbfdc38 | 2015-05-04 06:59:46 -0700 | [diff] [blame] | 3281 | /*[clinic end generated code: output=d6a16c63dda54441 input=155bc5695baafffd]*/ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3282 | { |
Serhiy Storchaka | 762ec97 | 2017-03-30 18:12:06 +0300 | [diff] [blame] | 3283 | if (html != NULL) { |
| 3284 | if (PyErr_WarnEx(PyExc_DeprecationWarning, |
| 3285 | "The html argument of XMLParser() is deprecated", |
| 3286 | 1) < 0) { |
| 3287 | return -1; |
| 3288 | } |
| 3289 | } |
| 3290 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3291 | self->entity = PyDict_New(); |
| 3292 | if (!self->entity) |
| 3293 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3294 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3295 | self->names = PyDict_New(); |
| 3296 | if (!self->names) { |
| 3297 | Py_CLEAR(self->entity); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3298 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3299 | } |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3300 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3301 | self->parser = EXPAT(ParserCreate_MM)(encoding, &ExpatMemoryHandler, "}"); |
| 3302 | if (!self->parser) { |
| 3303 | Py_CLEAR(self->entity); |
| 3304 | Py_CLEAR(self->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3305 | PyErr_NoMemory(); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3306 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3307 | } |
| 3308 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3309 | if (target) { |
| 3310 | Py_INCREF(target); |
| 3311 | } else { |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 3312 | target = treebuilder_new(&TreeBuilder_Type, NULL, NULL); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3313 | if (!target) { |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3314 | Py_CLEAR(self->entity); |
| 3315 | Py_CLEAR(self->names); |
| 3316 | EXPAT(ParserFree)(self->parser); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3317 | return -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3318 | } |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3319 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3320 | self->target = target; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3321 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3322 | self->handle_start = PyObject_GetAttrString(target, "start"); |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3323 | if (ignore_attribute_error(self->handle_start)) { |
| 3324 | return -1; |
| 3325 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3326 | self->handle_data = PyObject_GetAttrString(target, "data"); |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3327 | if (ignore_attribute_error(self->handle_data)) { |
| 3328 | return -1; |
| 3329 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3330 | self->handle_end = PyObject_GetAttrString(target, "end"); |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3331 | if (ignore_attribute_error(self->handle_end)) { |
| 3332 | return -1; |
| 3333 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3334 | self->handle_comment = PyObject_GetAttrString(target, "comment"); |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3335 | if (ignore_attribute_error(self->handle_comment)) { |
| 3336 | return -1; |
| 3337 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3338 | self->handle_pi = PyObject_GetAttrString(target, "pi"); |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3339 | if (ignore_attribute_error(self->handle_pi)) { |
| 3340 | return -1; |
| 3341 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3342 | self->handle_close = PyObject_GetAttrString(target, "close"); |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3343 | if (ignore_attribute_error(self->handle_close)) { |
| 3344 | return -1; |
| 3345 | } |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3346 | self->handle_doctype = PyObject_GetAttrString(target, "doctype"); |
scoder | c8d8e15 | 2017-09-14 22:00:03 +0200 | [diff] [blame] | 3347 | if (ignore_attribute_error(self->handle_doctype)) { |
| 3348 | return -1; |
| 3349 | } |
Eli Bendersky | 4583990 | 2013-01-13 05:14:47 -0800 | [diff] [blame] | 3350 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3351 | /* configure parser */ |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3352 | EXPAT(SetUserData)(self->parser, self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3353 | EXPAT(SetElementHandler)( |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3354 | self->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3355 | (XML_StartElementHandler) expat_start_handler, |
| 3356 | (XML_EndElementHandler) expat_end_handler |
| 3357 | ); |
| 3358 | EXPAT(SetDefaultHandlerExpand)( |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3359 | self->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3360 | (XML_DefaultHandler) expat_default_handler |
| 3361 | ); |
| 3362 | EXPAT(SetCharacterDataHandler)( |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3363 | self->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3364 | (XML_CharacterDataHandler) expat_data_handler |
| 3365 | ); |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3366 | if (self->handle_comment) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3367 | EXPAT(SetCommentHandler)( |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3368 | self->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3369 | (XML_CommentHandler) expat_comment_handler |
| 3370 | ); |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3371 | if (self->handle_pi) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3372 | EXPAT(SetProcessingInstructionHandler)( |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3373 | self->parser, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3374 | (XML_ProcessingInstructionHandler) expat_pi_handler |
| 3375 | ); |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3376 | EXPAT(SetStartDoctypeDeclHandler)( |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3377 | self->parser, |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3378 | (XML_StartDoctypeDeclHandler) expat_start_doctype_handler |
| 3379 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3380 | EXPAT(SetUnknownEncodingHandler)( |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3381 | self->parser, |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 3382 | EXPAT(DefaultUnknownEncodingHandler), NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3383 | ); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3384 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3385 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3386 | } |
| 3387 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3388 | static int |
| 3389 | xmlparser_gc_traverse(XMLParserObject *self, visitproc visit, void *arg) |
| 3390 | { |
| 3391 | Py_VISIT(self->handle_close); |
| 3392 | Py_VISIT(self->handle_pi); |
| 3393 | Py_VISIT(self->handle_comment); |
| 3394 | Py_VISIT(self->handle_end); |
| 3395 | Py_VISIT(self->handle_data); |
| 3396 | Py_VISIT(self->handle_start); |
| 3397 | |
| 3398 | Py_VISIT(self->target); |
| 3399 | Py_VISIT(self->entity); |
| 3400 | Py_VISIT(self->names); |
| 3401 | |
| 3402 | return 0; |
| 3403 | } |
| 3404 | |
| 3405 | static int |
| 3406 | xmlparser_gc_clear(XMLParserObject *self) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3407 | { |
Victor Stinner | e727d41 | 2017-09-18 05:29:37 -0700 | [diff] [blame] | 3408 | if (self->parser != NULL) { |
| 3409 | XML_Parser parser = self->parser; |
| 3410 | self->parser = NULL; |
| 3411 | EXPAT(ParserFree)(parser); |
| 3412 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3413 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3414 | Py_CLEAR(self->handle_close); |
| 3415 | Py_CLEAR(self->handle_pi); |
| 3416 | Py_CLEAR(self->handle_comment); |
| 3417 | Py_CLEAR(self->handle_end); |
| 3418 | Py_CLEAR(self->handle_data); |
| 3419 | Py_CLEAR(self->handle_start); |
| 3420 | Py_CLEAR(self->handle_doctype); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3421 | |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3422 | Py_CLEAR(self->target); |
| 3423 | Py_CLEAR(self->entity); |
| 3424 | Py_CLEAR(self->names); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3425 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3426 | return 0; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3427 | } |
| 3428 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3429 | static void |
| 3430 | xmlparser_dealloc(XMLParserObject* self) |
| 3431 | { |
| 3432 | PyObject_GC_UnTrack(self); |
| 3433 | xmlparser_gc_clear(self); |
| 3434 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 3435 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3436 | |
| 3437 | LOCAL(PyObject*) |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3438 | expat_parse(XMLParserObject* self, const char* data, int data_len, int final) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3439 | { |
| 3440 | int ok; |
| 3441 | |
Victor Stinner | 3fd8cbd | 2013-07-18 22:46:14 +0200 | [diff] [blame] | 3442 | assert(!PyErr_Occurred()); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3443 | ok = EXPAT(Parse)(self->parser, data, data_len, final); |
| 3444 | |
| 3445 | if (PyErr_Occurred()) |
| 3446 | return NULL; |
| 3447 | |
| 3448 | if (!ok) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3449 | expat_set_error( |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 3450 | EXPAT(GetErrorCode)(self->parser), |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3451 | EXPAT(GetErrorLineNumber)(self->parser), |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 3452 | EXPAT(GetErrorColumnNumber)(self->parser), |
| 3453 | NULL |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3454 | ); |
| 3455 | return NULL; |
| 3456 | } |
| 3457 | |
| 3458 | Py_RETURN_NONE; |
| 3459 | } |
| 3460 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3461 | /*[clinic input] |
| 3462 | _elementtree.XMLParser.close |
| 3463 | |
| 3464 | [clinic start generated code]*/ |
| 3465 | |
| 3466 | static PyObject * |
| 3467 | _elementtree_XMLParser_close_impl(XMLParserObject *self) |
| 3468 | /*[clinic end generated code: output=d68d375dd23bc7fb input=ca7909ca78c3abfe]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3469 | { |
| 3470 | /* end feeding data to parser */ |
| 3471 | |
| 3472 | PyObject* res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3473 | res = expat_parse(self, "", 0, 1); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3474 | if (!res) |
| 3475 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3476 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3477 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3478 | Py_DECREF(res); |
| 3479 | return treebuilder_done((TreeBuilderObject*) self->target); |
Eli Bendersky | 6eb50b1 | 2013-08-24 15:17:08 -0700 | [diff] [blame] | 3480 | } |
| 3481 | else if (self->handle_close) { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3482 | Py_DECREF(res); |
Victor Stinner | 3466bde | 2016-09-05 18:16:01 -0700 | [diff] [blame] | 3483 | return _PyObject_CallNoArg(self->handle_close); |
Eli Bendersky | 6eb50b1 | 2013-08-24 15:17:08 -0700 | [diff] [blame] | 3484 | } |
| 3485 | else { |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3486 | return res; |
Eli Bendersky | 6eb50b1 | 2013-08-24 15:17:08 -0700 | [diff] [blame] | 3487 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3488 | } |
| 3489 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3490 | /*[clinic input] |
| 3491 | _elementtree.XMLParser.feed |
| 3492 | |
| 3493 | data: object |
| 3494 | / |
| 3495 | |
| 3496 | [clinic start generated code]*/ |
| 3497 | |
| 3498 | static PyObject * |
| 3499 | _elementtree_XMLParser_feed(XMLParserObject *self, PyObject *data) |
| 3500 | /*[clinic end generated code: output=e42b6a78eec7446d input=fe231b6b8de3ce1f]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3501 | { |
| 3502 | /* feed data to parser */ |
| 3503 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3504 | if (PyUnicode_Check(data)) { |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3505 | Py_ssize_t data_len; |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3506 | const char *data_ptr = PyUnicode_AsUTF8AndSize(data, &data_len); |
| 3507 | if (data_ptr == NULL) |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3508 | return NULL; |
| 3509 | if (data_len > INT_MAX) { |
| 3510 | PyErr_SetString(PyExc_OverflowError, "size does not fit in an int"); |
| 3511 | return NULL; |
| 3512 | } |
| 3513 | /* Explicitly set UTF-8 encoding. Return code ignored. */ |
| 3514 | (void)EXPAT(SetEncoding)(self->parser, "utf-8"); |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3515 | return expat_parse(self, data_ptr, (int)data_len, 0); |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3516 | } |
| 3517 | else { |
| 3518 | Py_buffer view; |
| 3519 | PyObject *res; |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3520 | if (PyObject_GetBuffer(data, &view, PyBUF_SIMPLE) < 0) |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 3521 | return NULL; |
| 3522 | if (view.len > INT_MAX) { |
| 3523 | PyBuffer_Release(&view); |
| 3524 | PyErr_SetString(PyExc_OverflowError, "size does not fit in an int"); |
| 3525 | return NULL; |
| 3526 | } |
| 3527 | res = expat_parse(self, view.buf, (int)view.len, 0); |
| 3528 | PyBuffer_Release(&view); |
| 3529 | return res; |
| 3530 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3531 | } |
| 3532 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3533 | /*[clinic input] |
| 3534 | _elementtree.XMLParser._parse_whole |
| 3535 | |
| 3536 | file: object |
| 3537 | / |
| 3538 | |
| 3539 | [clinic start generated code]*/ |
| 3540 | |
| 3541 | static PyObject * |
| 3542 | _elementtree_XMLParser__parse_whole(XMLParserObject *self, PyObject *file) |
| 3543 | /*[clinic end generated code: output=f797197bb818dda3 input=19ecc893b6f3e752]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3544 | { |
Eli Bendersky | a369923 | 2013-05-19 18:47:23 -0700 | [diff] [blame] | 3545 | /* (internal) parse the whole input, until end of stream */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3546 | PyObject* reader; |
| 3547 | PyObject* buffer; |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3548 | PyObject* temp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3549 | PyObject* res; |
| 3550 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3551 | reader = PyObject_GetAttrString(file, "read"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3552 | if (!reader) |
| 3553 | return NULL; |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3554 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3555 | /* read from open file object */ |
| 3556 | for (;;) { |
| 3557 | |
| 3558 | buffer = PyObject_CallFunction(reader, "i", 64*1024); |
| 3559 | |
| 3560 | if (!buffer) { |
| 3561 | /* read failed (e.g. due to KeyboardInterrupt) */ |
| 3562 | Py_DECREF(reader); |
| 3563 | return NULL; |
| 3564 | } |
| 3565 | |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3566 | if (PyUnicode_CheckExact(buffer)) { |
| 3567 | /* A unicode object is encoded into bytes using UTF-8 */ |
Victor Stinner | 59799a8 | 2013-11-13 14:17:30 +0100 | [diff] [blame] | 3568 | if (PyUnicode_GET_LENGTH(buffer) == 0) { |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3569 | Py_DECREF(buffer); |
| 3570 | break; |
| 3571 | } |
| 3572 | temp = PyUnicode_AsEncodedString(buffer, "utf-8", "surrogatepass"); |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3573 | Py_DECREF(buffer); |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3574 | if (!temp) { |
| 3575 | /* Propagate exception from PyUnicode_AsEncodedString */ |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3576 | Py_DECREF(reader); |
| 3577 | return NULL; |
| 3578 | } |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 3579 | buffer = temp; |
| 3580 | } |
| 3581 | else if (!PyBytes_CheckExact(buffer) || PyBytes_GET_SIZE(buffer) == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3582 | Py_DECREF(buffer); |
| 3583 | break; |
| 3584 | } |
| 3585 | |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 3586 | if (PyBytes_GET_SIZE(buffer) > INT_MAX) { |
| 3587 | Py_DECREF(buffer); |
| 3588 | Py_DECREF(reader); |
| 3589 | PyErr_SetString(PyExc_OverflowError, "size does not fit in an int"); |
| 3590 | return NULL; |
| 3591 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3592 | res = expat_parse( |
Serhiy Storchaka | 26861b0 | 2015-02-16 20:52:17 +0200 | [diff] [blame] | 3593 | self, PyBytes_AS_STRING(buffer), (int)PyBytes_GET_SIZE(buffer), 0 |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3594 | ); |
| 3595 | |
| 3596 | Py_DECREF(buffer); |
| 3597 | |
| 3598 | if (!res) { |
| 3599 | Py_DECREF(reader); |
| 3600 | return NULL; |
| 3601 | } |
| 3602 | Py_DECREF(res); |
| 3603 | |
| 3604 | } |
| 3605 | |
| 3606 | Py_DECREF(reader); |
| 3607 | |
| 3608 | res = expat_parse(self, "", 0, 1); |
| 3609 | |
| 3610 | if (res && TreeBuilder_CheckExact(self->target)) { |
| 3611 | Py_DECREF(res); |
| 3612 | return treebuilder_done((TreeBuilderObject*) self->target); |
| 3613 | } |
| 3614 | |
| 3615 | return res; |
| 3616 | } |
| 3617 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3618 | /*[clinic input] |
| 3619 | _elementtree.XMLParser.doctype |
| 3620 | |
Serhiy Storchaka | 4a01cab | 2015-06-29 23:08:52 +0300 | [diff] [blame] | 3621 | name: object |
| 3622 | pubid: object |
| 3623 | system: object |
| 3624 | / |
| 3625 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3626 | [clinic start generated code]*/ |
| 3627 | |
| 3628 | static PyObject * |
Serhiy Storchaka | 4a01cab | 2015-06-29 23:08:52 +0300 | [diff] [blame] | 3629 | _elementtree_XMLParser_doctype_impl(XMLParserObject *self, PyObject *name, |
| 3630 | PyObject *pubid, PyObject *system) |
| 3631 | /*[clinic end generated code: output=10fb50c2afded88d input=84050276cca045e1]*/ |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3632 | { |
Serhiy Storchaka | 05744ac | 2015-06-29 22:35:58 +0300 | [diff] [blame] | 3633 | if (PyErr_WarnEx(PyExc_DeprecationWarning, |
| 3634 | "This method of XMLParser is deprecated. Define" |
| 3635 | " doctype() method on the TreeBuilder target.", |
| 3636 | 1) < 0) { |
| 3637 | return NULL; |
| 3638 | } |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 3639 | Py_RETURN_NONE; |
| 3640 | } |
| 3641 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3642 | /*[clinic input] |
| 3643 | _elementtree.XMLParser._setevents |
| 3644 | |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 3645 | events_queue: object |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3646 | events_to_report: object = None |
| 3647 | / |
| 3648 | |
| 3649 | [clinic start generated code]*/ |
| 3650 | |
| 3651 | static PyObject * |
| 3652 | _elementtree_XMLParser__setevents_impl(XMLParserObject *self, |
| 3653 | PyObject *events_queue, |
| 3654 | PyObject *events_to_report) |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 3655 | /*[clinic end generated code: output=1440092922b13ed1 input=abf90830a1c3b0fc]*/ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3656 | { |
| 3657 | /* activate element event reporting */ |
Serhiy Storchaka | bc4ded95 | 2015-12-24 11:51:57 +0200 | [diff] [blame] | 3658 | Py_ssize_t i; |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3659 | TreeBuilderObject *target; |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 3660 | PyObject *events_append, *events_seq; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3661 | |
| 3662 | if (!TreeBuilder_CheckExact(self->target)) { |
| 3663 | PyErr_SetString( |
| 3664 | PyExc_TypeError, |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 3665 | "event handling only supported for ElementTree.TreeBuilder " |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3666 | "targets" |
| 3667 | ); |
| 3668 | return NULL; |
| 3669 | } |
| 3670 | |
| 3671 | target = (TreeBuilderObject*) self->target; |
| 3672 | |
Serhiy Storchaka | 9ec5e25 | 2015-12-07 02:31:11 +0200 | [diff] [blame] | 3673 | events_append = PyObject_GetAttrString(events_queue, "append"); |
| 3674 | if (events_append == NULL) |
| 3675 | return NULL; |
Serhiy Storchaka | ec39756 | 2016-04-06 09:50:03 +0300 | [diff] [blame] | 3676 | Py_XSETREF(target->events_append, events_append); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3677 | |
| 3678 | /* clear out existing events */ |
Antoine Pitrou | c194884 | 2012-10-01 23:40:37 +0200 | [diff] [blame] | 3679 | Py_CLEAR(target->start_event_obj); |
| 3680 | Py_CLEAR(target->end_event_obj); |
| 3681 | Py_CLEAR(target->start_ns_event_obj); |
| 3682 | Py_CLEAR(target->end_ns_event_obj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3683 | |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3684 | if (events_to_report == Py_None) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3685 | /* default is "end" only */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3686 | target->end_event_obj = PyUnicode_FromString("end"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3687 | Py_RETURN_NONE; |
| 3688 | } |
| 3689 | |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3690 | if (!(events_seq = PySequence_Fast(events_to_report, |
| 3691 | "events must be a sequence"))) { |
| 3692 | return NULL; |
| 3693 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3694 | |
Serhiy Storchaka | bf623ae | 2017-04-19 20:03:52 +0300 | [diff] [blame] | 3695 | for (i = 0; i < PySequence_Fast_GET_SIZE(events_seq); ++i) { |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3696 | PyObject *event_name_obj = PySequence_Fast_GET_ITEM(events_seq, i); |
Serhiy Storchaka | 85b0f5b | 2016-11-20 10:16:47 +0200 | [diff] [blame] | 3697 | const char *event_name = NULL; |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3698 | if (PyUnicode_Check(event_name_obj)) { |
Serhiy Storchaka | bc4ded95 | 2015-12-24 11:51:57 +0200 | [diff] [blame] | 3699 | event_name = PyUnicode_AsUTF8(event_name_obj); |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3700 | } else if (PyBytes_Check(event_name_obj)) { |
| 3701 | event_name = PyBytes_AS_STRING(event_name_obj); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 3702 | } |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3703 | if (event_name == NULL) { |
| 3704 | Py_DECREF(events_seq); |
| 3705 | PyErr_Format(PyExc_ValueError, "invalid events sequence"); |
| 3706 | return NULL; |
Serhiy Storchaka | bc4ded95 | 2015-12-24 11:51:57 +0200 | [diff] [blame] | 3707 | } |
| 3708 | |
| 3709 | Py_INCREF(event_name_obj); |
| 3710 | if (strcmp(event_name, "start") == 0) { |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 3711 | Py_XSETREF(target->start_event_obj, event_name_obj); |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3712 | } else if (strcmp(event_name, "end") == 0) { |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 3713 | Py_XSETREF(target->end_event_obj, event_name_obj); |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3714 | } else if (strcmp(event_name, "start-ns") == 0) { |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 3715 | Py_XSETREF(target->start_ns_event_obj, event_name_obj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3716 | EXPAT(SetNamespaceDeclHandler)( |
| 3717 | self->parser, |
| 3718 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3719 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3720 | ); |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3721 | } else if (strcmp(event_name, "end-ns") == 0) { |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 3722 | Py_XSETREF(target->end_ns_event_obj, event_name_obj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3723 | EXPAT(SetNamespaceDeclHandler)( |
| 3724 | self->parser, |
| 3725 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 3726 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 3727 | ); |
| 3728 | } else { |
Serhiy Storchaka | bc4ded95 | 2015-12-24 11:51:57 +0200 | [diff] [blame] | 3729 | Py_DECREF(event_name_obj); |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3730 | Py_DECREF(events_seq); |
| 3731 | PyErr_Format(PyExc_ValueError, "unknown event '%s'", event_name); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3732 | return NULL; |
| 3733 | } |
| 3734 | } |
| 3735 | |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 3736 | Py_DECREF(events_seq); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3737 | Py_RETURN_NONE; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3738 | } |
| 3739 | |
Victor Stinner | bfc7bf0 | 2011-03-21 13:23:42 +0100 | [diff] [blame] | 3740 | static PyObject* |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 3741 | xmlparser_getattro(XMLParserObject* self, PyObject* nameobj) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3742 | { |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3743 | if (PyUnicode_Check(nameobj)) { |
| 3744 | PyObject* res; |
Serhiy Storchaka | f4934ea | 2016-11-16 10:17:58 +0200 | [diff] [blame] | 3745 | if (_PyUnicode_EqualToASCIIString(nameobj, "entity")) |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3746 | res = self->entity; |
Serhiy Storchaka | f4934ea | 2016-11-16 10:17:58 +0200 | [diff] [blame] | 3747 | else if (_PyUnicode_EqualToASCIIString(nameobj, "target")) |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3748 | res = self->target; |
Serhiy Storchaka | f4934ea | 2016-11-16 10:17:58 +0200 | [diff] [blame] | 3749 | else if (_PyUnicode_EqualToASCIIString(nameobj, "version")) { |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3750 | return PyUnicode_FromFormat( |
| 3751 | "Expat %d.%d.%d", XML_MAJOR_VERSION, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3752 | XML_MINOR_VERSION, XML_MICRO_VERSION); |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3753 | } |
| 3754 | else |
| 3755 | goto generic; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3756 | |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 3757 | Py_INCREF(res); |
| 3758 | return res; |
| 3759 | } |
| 3760 | generic: |
| 3761 | return PyObject_GenericGetAttr((PyObject*) self, nameobj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3762 | } |
| 3763 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3764 | #include "clinic/_elementtree.c.h" |
| 3765 | |
| 3766 | static PyMethodDef element_methods[] = { |
| 3767 | |
| 3768 | _ELEMENTTREE_ELEMENT_CLEAR_METHODDEF |
| 3769 | |
| 3770 | _ELEMENTTREE_ELEMENT_GET_METHODDEF |
| 3771 | _ELEMENTTREE_ELEMENT_SET_METHODDEF |
| 3772 | |
| 3773 | _ELEMENTTREE_ELEMENT_FIND_METHODDEF |
| 3774 | _ELEMENTTREE_ELEMENT_FINDTEXT_METHODDEF |
| 3775 | _ELEMENTTREE_ELEMENT_FINDALL_METHODDEF |
| 3776 | |
| 3777 | _ELEMENTTREE_ELEMENT_APPEND_METHODDEF |
| 3778 | _ELEMENTTREE_ELEMENT_EXTEND_METHODDEF |
| 3779 | _ELEMENTTREE_ELEMENT_INSERT_METHODDEF |
| 3780 | _ELEMENTTREE_ELEMENT_REMOVE_METHODDEF |
| 3781 | |
| 3782 | _ELEMENTTREE_ELEMENT_ITER_METHODDEF |
| 3783 | _ELEMENTTREE_ELEMENT_ITERTEXT_METHODDEF |
| 3784 | _ELEMENTTREE_ELEMENT_ITERFIND_METHODDEF |
| 3785 | |
Serhiy Storchaka | 762ec97 | 2017-03-30 18:12:06 +0300 | [diff] [blame] | 3786 | _ELEMENTTREE_ELEMENT_GETITERATOR_METHODDEF |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3787 | _ELEMENTTREE_ELEMENT_GETCHILDREN_METHODDEF |
| 3788 | |
| 3789 | _ELEMENTTREE_ELEMENT_ITEMS_METHODDEF |
| 3790 | _ELEMENTTREE_ELEMENT_KEYS_METHODDEF |
| 3791 | |
| 3792 | _ELEMENTTREE_ELEMENT_MAKEELEMENT_METHODDEF |
| 3793 | |
| 3794 | _ELEMENTTREE_ELEMENT___COPY___METHODDEF |
| 3795 | _ELEMENTTREE_ELEMENT___DEEPCOPY___METHODDEF |
| 3796 | _ELEMENTTREE_ELEMENT___SIZEOF___METHODDEF |
| 3797 | _ELEMENTTREE_ELEMENT___GETSTATE___METHODDEF |
| 3798 | _ELEMENTTREE_ELEMENT___SETSTATE___METHODDEF |
| 3799 | |
| 3800 | {NULL, NULL} |
| 3801 | }; |
| 3802 | |
| 3803 | static PyMappingMethods element_as_mapping = { |
| 3804 | (lenfunc) element_length, |
| 3805 | (binaryfunc) element_subscr, |
| 3806 | (objobjargproc) element_ass_subscr, |
| 3807 | }; |
| 3808 | |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 3809 | static PyGetSetDef element_getsetlist[] = { |
| 3810 | {"tag", |
| 3811 | (getter)element_tag_getter, |
| 3812 | (setter)element_tag_setter, |
| 3813 | "A string identifying what kind of data this element represents"}, |
| 3814 | {"text", |
| 3815 | (getter)element_text_getter, |
| 3816 | (setter)element_text_setter, |
| 3817 | "A string of text directly after the start tag, or None"}, |
| 3818 | {"tail", |
| 3819 | (getter)element_tail_getter, |
| 3820 | (setter)element_tail_setter, |
| 3821 | "A string of text directly after the end tag, or None"}, |
| 3822 | {"attrib", |
| 3823 | (getter)element_attrib_getter, |
| 3824 | (setter)element_attrib_setter, |
| 3825 | "A dictionary containing the element's attributes"}, |
| 3826 | {NULL}, |
| 3827 | }; |
| 3828 | |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3829 | static PyTypeObject Element_Type = { |
| 3830 | PyVarObject_HEAD_INIT(NULL, 0) |
| 3831 | "xml.etree.ElementTree.Element", sizeof(ElementObject), 0, |
| 3832 | /* methods */ |
| 3833 | (destructor)element_dealloc, /* tp_dealloc */ |
| 3834 | 0, /* tp_print */ |
| 3835 | 0, /* tp_getattr */ |
| 3836 | 0, /* tp_setattr */ |
| 3837 | 0, /* tp_reserved */ |
| 3838 | (reprfunc)element_repr, /* tp_repr */ |
| 3839 | 0, /* tp_as_number */ |
| 3840 | &element_as_sequence, /* tp_as_sequence */ |
| 3841 | &element_as_mapping, /* tp_as_mapping */ |
| 3842 | 0, /* tp_hash */ |
| 3843 | 0, /* tp_call */ |
| 3844 | 0, /* tp_str */ |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 3845 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 3846 | 0, /* tp_setattro */ |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3847 | 0, /* tp_as_buffer */ |
| 3848 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 3849 | /* tp_flags */ |
| 3850 | 0, /* tp_doc */ |
| 3851 | (traverseproc)element_gc_traverse, /* tp_traverse */ |
| 3852 | (inquiry)element_gc_clear, /* tp_clear */ |
| 3853 | 0, /* tp_richcompare */ |
| 3854 | offsetof(ElementObject, weakreflist), /* tp_weaklistoffset */ |
| 3855 | 0, /* tp_iter */ |
| 3856 | 0, /* tp_iternext */ |
| 3857 | element_methods, /* tp_methods */ |
| 3858 | 0, /* tp_members */ |
Serhiy Storchaka | dde0815 | 2015-11-25 15:28:13 +0200 | [diff] [blame] | 3859 | element_getsetlist, /* tp_getset */ |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3860 | 0, /* tp_base */ |
| 3861 | 0, /* tp_dict */ |
| 3862 | 0, /* tp_descr_get */ |
| 3863 | 0, /* tp_descr_set */ |
| 3864 | 0, /* tp_dictoffset */ |
| 3865 | (initproc)element_init, /* tp_init */ |
| 3866 | PyType_GenericAlloc, /* tp_alloc */ |
| 3867 | element_new, /* tp_new */ |
| 3868 | 0, /* tp_free */ |
| 3869 | }; |
| 3870 | |
| 3871 | static PyMethodDef treebuilder_methods[] = { |
| 3872 | _ELEMENTTREE_TREEBUILDER_DATA_METHODDEF |
| 3873 | _ELEMENTTREE_TREEBUILDER_START_METHODDEF |
| 3874 | _ELEMENTTREE_TREEBUILDER_END_METHODDEF |
| 3875 | _ELEMENTTREE_TREEBUILDER_CLOSE_METHODDEF |
| 3876 | {NULL, NULL} |
| 3877 | }; |
| 3878 | |
| 3879 | static PyTypeObject TreeBuilder_Type = { |
| 3880 | PyVarObject_HEAD_INIT(NULL, 0) |
| 3881 | "xml.etree.ElementTree.TreeBuilder", sizeof(TreeBuilderObject), 0, |
| 3882 | /* methods */ |
| 3883 | (destructor)treebuilder_dealloc, /* tp_dealloc */ |
| 3884 | 0, /* tp_print */ |
| 3885 | 0, /* tp_getattr */ |
| 3886 | 0, /* tp_setattr */ |
| 3887 | 0, /* tp_reserved */ |
| 3888 | 0, /* tp_repr */ |
| 3889 | 0, /* tp_as_number */ |
| 3890 | 0, /* tp_as_sequence */ |
| 3891 | 0, /* tp_as_mapping */ |
| 3892 | 0, /* tp_hash */ |
| 3893 | 0, /* tp_call */ |
| 3894 | 0, /* tp_str */ |
| 3895 | 0, /* tp_getattro */ |
| 3896 | 0, /* tp_setattro */ |
| 3897 | 0, /* tp_as_buffer */ |
| 3898 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 3899 | /* tp_flags */ |
| 3900 | 0, /* tp_doc */ |
| 3901 | (traverseproc)treebuilder_gc_traverse, /* tp_traverse */ |
| 3902 | (inquiry)treebuilder_gc_clear, /* tp_clear */ |
| 3903 | 0, /* tp_richcompare */ |
| 3904 | 0, /* tp_weaklistoffset */ |
| 3905 | 0, /* tp_iter */ |
| 3906 | 0, /* tp_iternext */ |
| 3907 | treebuilder_methods, /* tp_methods */ |
| 3908 | 0, /* tp_members */ |
| 3909 | 0, /* tp_getset */ |
| 3910 | 0, /* tp_base */ |
| 3911 | 0, /* tp_dict */ |
| 3912 | 0, /* tp_descr_get */ |
| 3913 | 0, /* tp_descr_set */ |
| 3914 | 0, /* tp_dictoffset */ |
| 3915 | _elementtree_TreeBuilder___init__, /* tp_init */ |
| 3916 | PyType_GenericAlloc, /* tp_alloc */ |
| 3917 | treebuilder_new, /* tp_new */ |
| 3918 | 0, /* tp_free */ |
| 3919 | }; |
| 3920 | |
| 3921 | static PyMethodDef xmlparser_methods[] = { |
| 3922 | _ELEMENTTREE_XMLPARSER_FEED_METHODDEF |
| 3923 | _ELEMENTTREE_XMLPARSER_CLOSE_METHODDEF |
| 3924 | _ELEMENTTREE_XMLPARSER__PARSE_WHOLE_METHODDEF |
| 3925 | _ELEMENTTREE_XMLPARSER__SETEVENTS_METHODDEF |
| 3926 | _ELEMENTTREE_XMLPARSER_DOCTYPE_METHODDEF |
| 3927 | {NULL, NULL} |
| 3928 | }; |
| 3929 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 3930 | static PyTypeObject XMLParser_Type = { |
Martin v. Löwis | 9f2e346 | 2007-07-21 17:22:18 +0000 | [diff] [blame] | 3931 | PyVarObject_HEAD_INIT(NULL, 0) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 3932 | "xml.etree.ElementTree.XMLParser", sizeof(XMLParserObject), 0, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3933 | /* methods */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3934 | (destructor)xmlparser_dealloc, /* tp_dealloc */ |
| 3935 | 0, /* tp_print */ |
| 3936 | 0, /* tp_getattr */ |
| 3937 | 0, /* tp_setattr */ |
| 3938 | 0, /* tp_reserved */ |
| 3939 | 0, /* tp_repr */ |
| 3940 | 0, /* tp_as_number */ |
| 3941 | 0, /* tp_as_sequence */ |
| 3942 | 0, /* tp_as_mapping */ |
| 3943 | 0, /* tp_hash */ |
| 3944 | 0, /* tp_call */ |
| 3945 | 0, /* tp_str */ |
| 3946 | (getattrofunc)xmlparser_getattro, /* tp_getattro */ |
| 3947 | 0, /* tp_setattro */ |
| 3948 | 0, /* tp_as_buffer */ |
| 3949 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC, |
| 3950 | /* tp_flags */ |
| 3951 | 0, /* tp_doc */ |
| 3952 | (traverseproc)xmlparser_gc_traverse, /* tp_traverse */ |
| 3953 | (inquiry)xmlparser_gc_clear, /* tp_clear */ |
| 3954 | 0, /* tp_richcompare */ |
| 3955 | 0, /* tp_weaklistoffset */ |
| 3956 | 0, /* tp_iter */ |
| 3957 | 0, /* tp_iternext */ |
| 3958 | xmlparser_methods, /* tp_methods */ |
| 3959 | 0, /* tp_members */ |
| 3960 | 0, /* tp_getset */ |
| 3961 | 0, /* tp_base */ |
| 3962 | 0, /* tp_dict */ |
| 3963 | 0, /* tp_descr_get */ |
| 3964 | 0, /* tp_descr_set */ |
| 3965 | 0, /* tp_dictoffset */ |
Serhiy Storchaka | cb98556 | 2015-05-04 15:32:48 +0300 | [diff] [blame] | 3966 | _elementtree_XMLParser___init__, /* tp_init */ |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 3967 | PyType_GenericAlloc, /* tp_alloc */ |
| 3968 | xmlparser_new, /* tp_new */ |
| 3969 | 0, /* tp_free */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3970 | }; |
| 3971 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3972 | /* ==================================================================== */ |
| 3973 | /* python module interface */ |
| 3974 | |
| 3975 | static PyMethodDef _functions[] = { |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 3976 | {"SubElement", (PyCFunction) subelement, METH_VARARGS | METH_KEYWORDS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3977 | {NULL, NULL} |
| 3978 | }; |
| 3979 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3980 | |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 3981 | static struct PyModuleDef elementtreemodule = { |
| 3982 | PyModuleDef_HEAD_INIT, |
| 3983 | "_elementtree", |
| 3984 | NULL, |
| 3985 | sizeof(elementtreestate), |
| 3986 | _functions, |
| 3987 | NULL, |
| 3988 | elementtree_traverse, |
| 3989 | elementtree_clear, |
| 3990 | elementtree_free |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3991 | }; |
| 3992 | |
Neal Norwitz | f6657e6 | 2006-12-28 04:47:50 +0000 | [diff] [blame] | 3993 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 3994 | PyInit__elementtree(void) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3995 | { |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 3996 | PyObject *m, *temp; |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 3997 | elementtreestate *st; |
| 3998 | |
| 3999 | m = PyState_FindModule(&elementtreemodule); |
| 4000 | if (m) { |
| 4001 | Py_INCREF(m); |
| 4002 | return m; |
| 4003 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 4004 | |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 4005 | /* Initialize object types */ |
Ronald Oussoren | 138d080 | 2013-07-19 11:11:25 +0200 | [diff] [blame] | 4006 | if (PyType_Ready(&ElementIter_Type) < 0) |
| 4007 | return NULL; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 4008 | if (PyType_Ready(&TreeBuilder_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 4009 | return NULL; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 4010 | if (PyType_Ready(&Element_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 4011 | return NULL; |
Amaury Forgeot d'Arc | ba4105c | 2008-07-02 21:41:01 +0000 | [diff] [blame] | 4012 | if (PyType_Ready(&XMLParser_Type) < 0) |
Alexander Belopolsky | f0f4514 | 2010-08-11 17:31:17 +0000 | [diff] [blame] | 4013 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 4014 | |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 4015 | m = PyModule_Create(&elementtreemodule); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 4016 | if (!m) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 4017 | return NULL; |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 4018 | st = ET_STATE(m); |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 4019 | |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 4020 | if (!(temp = PyImport_ImportModule("copy"))) |
| 4021 | return NULL; |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 4022 | st->deepcopy_obj = PyObject_GetAttrString(temp, "deepcopy"); |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 4023 | Py_XDECREF(temp); |
| 4024 | |
Victor Stinner | b136f11 | 2017-07-10 22:28:02 +0200 | [diff] [blame] | 4025 | if (st->deepcopy_obj == NULL) { |
| 4026 | return NULL; |
| 4027 | } |
| 4028 | |
| 4029 | assert(!PyErr_Occurred()); |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 4030 | if (!(st->elementpath_obj = PyImport_ImportModule("xml.etree.ElementPath"))) |
Eli Bendersky | 828efde | 2012-04-05 05:40:58 +0300 | [diff] [blame] | 4031 | return NULL; |
| 4032 | |
Eli Bendersky | 20d4174 | 2012-06-01 09:48:37 +0300 | [diff] [blame] | 4033 | /* link against pyexpat */ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 4034 | expat_capi = PyCapsule_Import(PyExpat_CAPSULE_NAME, 0); |
| 4035 | if (expat_capi) { |
| 4036 | /* check that it's usable */ |
| 4037 | if (strcmp(expat_capi->magic, PyExpat_CAPI_MAGIC) != 0 || |
Victor Stinner | 706768c | 2014-08-16 01:03:39 +0200 | [diff] [blame] | 4038 | (size_t)expat_capi->size < sizeof(struct PyExpat_CAPI) || |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 4039 | expat_capi->MAJOR_VERSION != XML_MAJOR_VERSION || |
| 4040 | expat_capi->MINOR_VERSION != XML_MINOR_VERSION || |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 4041 | expat_capi->MICRO_VERSION != XML_MICRO_VERSION) { |
Eli Bendersky | ef391ac | 2012-07-21 20:28:46 +0300 | [diff] [blame] | 4042 | PyErr_SetString(PyExc_ImportError, |
| 4043 | "pyexpat version is incompatible"); |
| 4044 | return NULL; |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 4045 | } |
Eli Bendersky | ef391ac | 2012-07-21 20:28:46 +0300 | [diff] [blame] | 4046 | } else { |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 4047 | return NULL; |
Eli Bendersky | ef391ac | 2012-07-21 20:28:46 +0300 | [diff] [blame] | 4048 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 4049 | |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 4050 | st->parseerror_obj = PyErr_NewException( |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 4051 | "xml.etree.ElementTree.ParseError", PyExc_SyntaxError, NULL |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 4052 | ); |
Eli Bendersky | 532d03e | 2013-08-10 08:00:39 -0700 | [diff] [blame] | 4053 | Py_INCREF(st->parseerror_obj); |
| 4054 | PyModule_AddObject(m, "ParseError", st->parseerror_obj); |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 4055 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 4056 | Py_INCREF((PyObject *)&Element_Type); |
| 4057 | PyModule_AddObject(m, "Element", (PyObject *)&Element_Type); |
| 4058 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 4059 | Py_INCREF((PyObject *)&TreeBuilder_Type); |
| 4060 | PyModule_AddObject(m, "TreeBuilder", (PyObject *)&TreeBuilder_Type); |
| 4061 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 4062 | Py_INCREF((PyObject *)&XMLParser_Type); |
| 4063 | PyModule_AddObject(m, "XMLParser", (PyObject *)&XMLParser_Type); |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 4064 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 4065 | return m; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 4066 | } |