Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1 | /* |
| 2 | * ElementTree |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3 | * $Id: _elementtree.c 3473 2009-01-11 22:53:55Z fredrik $ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 4 | * |
| 5 | * elementtree accelerator |
| 6 | * |
| 7 | * History: |
| 8 | * 1999-06-20 fl created (as part of sgmlop) |
| 9 | * 2001-05-29 fl effdom edition |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 10 | * 2003-02-27 fl elementtree edition (alpha) |
| 11 | * 2004-06-03 fl updates for elementtree 1.2 |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 12 | * 2005-01-05 fl major optimization effort |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 13 | * 2005-01-11 fl first public release (cElementTree 0.8) |
| 14 | * 2005-01-12 fl split element object into base and extras |
| 15 | * 2005-01-13 fl use tagged pointers for tail/text (cElementTree 0.9) |
| 16 | * 2005-01-17 fl added treebuilder close method |
| 17 | * 2005-01-17 fl fixed crash in getchildren |
| 18 | * 2005-01-18 fl removed observer api, added iterparse (cElementTree 0.9.3) |
| 19 | * 2005-01-23 fl revised iterparse api; added namespace event support (0.9.8) |
| 20 | * 2005-01-26 fl added VERSION module property (cElementTree 1.0) |
| 21 | * 2005-01-28 fl added remove method (1.0.1) |
| 22 | * 2005-03-01 fl added iselement function; fixed makeelement aliasing (1.0.2) |
| 23 | * 2005-03-13 fl export Comment and ProcessingInstruction/PI helpers |
| 24 | * 2005-03-26 fl added Comment and PI support to XMLParser |
| 25 | * 2005-03-27 fl event optimizations; complain about bogus events |
| 26 | * 2005-08-08 fl fixed read error handling in parse |
| 27 | * 2005-08-11 fl added runtime test for copy workaround (1.0.3) |
| 28 | * 2005-12-13 fl added expat_capi support (for xml.etree) (1.0.4) |
| 29 | * 2005-12-16 fl added support for non-standard encodings |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 30 | * 2006-03-08 fl fixed a couple of potential null-refs and leaks |
| 31 | * 2006-03-12 fl merge in 2.5 ssize_t changes |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 32 | * 2007-08-25 fl call custom builder's close method from XMLParser |
| 33 | * 2007-08-31 fl added iter, extend from ET 1.3 |
| 34 | * 2007-09-01 fl fixed ParseError exception, setslice source type, etc |
| 35 | * 2007-09-03 fl fixed handling of negative insert indexes |
| 36 | * 2007-09-04 fl added itertext from ET 1.3 |
| 37 | * 2007-09-06 fl added position attribute to ParseError exception |
| 38 | * 2008-06-06 fl delay error reporting in iterparse (from Hrvoje Niksic) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 39 | * |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 40 | * Copyright (c) 1999-2009 by Secret Labs AB. All rights reserved. |
| 41 | * Copyright (c) 1999-2009 by Fredrik Lundh. |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 42 | * |
| 43 | * info@pythonware.com |
| 44 | * http://www.pythonware.com |
| 45 | */ |
| 46 | |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 47 | /* Licensed to PSF under a Contributor Agreement. */ |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 48 | /* See http://www.python.org/psf/license for licensing details. */ |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 49 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 50 | #include "Python.h" |
| 51 | |
Fredrik Lundh | dc075b9 | 2006-08-16 16:47:07 +0000 | [diff] [blame] | 52 | #define VERSION "1.0.6" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 53 | |
| 54 | /* -------------------------------------------------------------------- */ |
| 55 | /* configuration */ |
| 56 | |
| 57 | /* Leave defined to include the expat-based XMLParser type */ |
| 58 | #define USE_EXPAT |
| 59 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 60 | /* Define to do all expat calls via pyexpat's embedded expat library */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 61 | /* #define USE_PYEXPAT_CAPI */ |
| 62 | |
| 63 | /* An element can hold this many children without extra memory |
| 64 | allocations. */ |
| 65 | #define STATIC_CHILDREN 4 |
| 66 | |
| 67 | /* For best performance, chose a value so that 80-90% of all nodes |
| 68 | have no more than the given number of children. Set this to zero |
| 69 | to minimize the size of the element structure itself (this only |
| 70 | helps if you have lots of leaf nodes with attributes). */ |
| 71 | |
| 72 | /* Also note that pymalloc always allocates blocks in multiples of |
| 73 | eight bytes. For the current version of cElementTree, this means |
| 74 | that the number of children should be an even number, at least on |
| 75 | 32-bit platforms. */ |
| 76 | |
| 77 | /* -------------------------------------------------------------------- */ |
| 78 | |
| 79 | #if 0 |
| 80 | static int memory = 0; |
| 81 | #define ALLOC(size, comment)\ |
| 82 | do { memory += size; printf("%8d - %s\n", memory, comment); } while (0) |
| 83 | #define RELEASE(size, comment)\ |
| 84 | do { memory -= size; printf("%8d - %s\n", memory, comment); } while (0) |
| 85 | #else |
| 86 | #define ALLOC(size, comment) |
| 87 | #define RELEASE(size, comment) |
| 88 | #endif |
| 89 | |
| 90 | /* compiler tweaks */ |
| 91 | #if defined(_MSC_VER) |
| 92 | #define LOCAL(type) static __inline type __fastcall |
| 93 | #else |
| 94 | #define LOCAL(type) static type |
| 95 | #endif |
| 96 | |
| 97 | /* compatibility macros */ |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 98 | #if (PY_VERSION_HEX < 0x02060000) |
| 99 | #define Py_REFCNT(ob) (((PyObject*)(ob))->ob_refcnt) |
| 100 | #define Py_TYPE(ob) (((PyObject*)(ob))->ob_type) |
| 101 | #endif |
| 102 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 103 | #if (PY_VERSION_HEX < 0x02050000) |
| 104 | typedef int Py_ssize_t; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 105 | #define lenfunc inquiry |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 106 | #endif |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 107 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 108 | #if (PY_VERSION_HEX < 0x02040000) |
| 109 | #define PyDict_CheckExact PyDict_Check |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 110 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 111 | #if !defined(Py_RETURN_NONE) |
| 112 | #define Py_RETURN_NONE return Py_INCREF(Py_None), Py_None |
| 113 | #endif |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 114 | #endif |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 115 | |
| 116 | /* macros used to store 'join' flags in string object pointers. note |
| 117 | that all use of text and tail as object pointers must be wrapped in |
| 118 | JOIN_OBJ. see comments in the ElementObject definition for more |
| 119 | info. */ |
| 120 | #define JOIN_GET(p) ((Py_uintptr_t) (p) & 1) |
| 121 | #define JOIN_SET(p, flag) ((void*) ((Py_uintptr_t) (JOIN_OBJ(p)) | (flag))) |
| 122 | #define JOIN_OBJ(p) ((PyObject*) ((Py_uintptr_t) (p) & ~1)) |
| 123 | |
| 124 | /* glue functions (see the init function for details) */ |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 125 | static PyObject* elementtree_parseerror_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 126 | static PyObject* elementtree_copyelement_obj; |
| 127 | static PyObject* elementtree_deepcopy_obj; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 128 | static PyObject* elementtree_iter_obj; |
| 129 | static PyObject* elementtree_itertext_obj; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 130 | static PyObject* elementpath_obj; |
| 131 | |
| 132 | /* helpers */ |
| 133 | |
| 134 | LOCAL(PyObject*) |
| 135 | deepcopy(PyObject* object, PyObject* memo) |
| 136 | { |
| 137 | /* do a deep copy of the given object */ |
| 138 | |
| 139 | PyObject* args; |
| 140 | PyObject* result; |
| 141 | |
| 142 | if (!elementtree_deepcopy_obj) { |
| 143 | PyErr_SetString( |
| 144 | PyExc_RuntimeError, |
| 145 | "deepcopy helper not found" |
| 146 | ); |
| 147 | return NULL; |
| 148 | } |
| 149 | |
| 150 | args = PyTuple_New(2); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 151 | if (!args) |
| 152 | return NULL; |
| 153 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 154 | Py_INCREF(object); PyTuple_SET_ITEM(args, 0, (PyObject*) object); |
| 155 | Py_INCREF(memo); PyTuple_SET_ITEM(args, 1, (PyObject*) memo); |
| 156 | |
| 157 | result = PyObject_CallObject(elementtree_deepcopy_obj, args); |
| 158 | |
| 159 | Py_DECREF(args); |
| 160 | |
| 161 | return result; |
| 162 | } |
| 163 | |
| 164 | LOCAL(PyObject*) |
| 165 | list_join(PyObject* list) |
| 166 | { |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 167 | /* join list elements */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 168 | PyObject* joiner; |
| 169 | PyObject* function; |
| 170 | PyObject* args; |
| 171 | PyObject* result; |
| 172 | |
| 173 | switch (PyList_GET_SIZE(list)) { |
| 174 | case 0: |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 175 | return PyString_FromString(""); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 176 | case 1: |
| 177 | result = PyList_GET_ITEM(list, 0); |
| 178 | Py_INCREF(result); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 179 | return result; |
| 180 | } |
| 181 | |
| 182 | /* two or more elements: slice out a suitable separator from the |
| 183 | first member, and use that to join the entire list */ |
| 184 | |
| 185 | joiner = PySequence_GetSlice(PyList_GET_ITEM(list, 0), 0, 0); |
| 186 | if (!joiner) |
| 187 | return NULL; |
| 188 | |
| 189 | function = PyObject_GetAttrString(joiner, "join"); |
| 190 | if (!function) { |
| 191 | Py_DECREF(joiner); |
| 192 | return NULL; |
| 193 | } |
| 194 | |
| 195 | args = PyTuple_New(1); |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 196 | if (!args) { |
| 197 | Py_DECREF(function); |
| 198 | Py_DECREF(joiner); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 199 | return NULL; |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 200 | } |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 201 | |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 202 | Py_INCREF(list); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 203 | PyTuple_SET_ITEM(args, 0, list); |
| 204 | |
| 205 | result = PyObject_CallObject(function, args); |
| 206 | |
| 207 | Py_DECREF(args); /* also removes list */ |
| 208 | Py_DECREF(function); |
| 209 | Py_DECREF(joiner); |
| 210 | |
| 211 | return result; |
| 212 | } |
| 213 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 214 | /* -------------------------------------------------------------------- */ |
| 215 | /* the element type */ |
| 216 | |
| 217 | typedef struct { |
| 218 | |
| 219 | /* attributes (a dictionary object), or None if no attributes */ |
| 220 | PyObject* attrib; |
| 221 | |
| 222 | /* child elements */ |
| 223 | int length; /* actual number of items */ |
| 224 | int allocated; /* allocated items */ |
| 225 | |
| 226 | /* this either points to _children or to a malloced buffer */ |
| 227 | PyObject* *children; |
| 228 | |
| 229 | PyObject* _children[STATIC_CHILDREN]; |
| 230 | |
| 231 | } ElementObjectExtra; |
| 232 | |
| 233 | typedef struct { |
| 234 | PyObject_HEAD |
| 235 | |
| 236 | /* element tag (a string). */ |
| 237 | PyObject* tag; |
| 238 | |
| 239 | /* text before first child. note that this is a tagged pointer; |
| 240 | use JOIN_OBJ to get the object pointer. the join flag is used |
| 241 | to distinguish lists created by the tree builder from lists |
| 242 | assigned to the attribute by application code; the former |
| 243 | should be joined before being returned to the user, the latter |
| 244 | should be left intact. */ |
| 245 | PyObject* text; |
| 246 | |
| 247 | /* text after this element, in parent. note that this is a tagged |
| 248 | pointer; use JOIN_OBJ to get the object pointer. */ |
| 249 | PyObject* tail; |
| 250 | |
| 251 | ElementObjectExtra* extra; |
| 252 | |
| 253 | } ElementObject; |
| 254 | |
| 255 | staticforward PyTypeObject Element_Type; |
| 256 | |
Christian Heimes | e93237d | 2007-12-19 02:37:44 +0000 | [diff] [blame] | 257 | #define Element_CheckExact(op) (Py_TYPE(op) == &Element_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 258 | |
| 259 | /* -------------------------------------------------------------------- */ |
| 260 | /* element constructor and destructor */ |
| 261 | |
| 262 | LOCAL(int) |
| 263 | element_new_extra(ElementObject* self, PyObject* attrib) |
| 264 | { |
| 265 | self->extra = PyObject_Malloc(sizeof(ElementObjectExtra)); |
| 266 | if (!self->extra) |
| 267 | return -1; |
| 268 | |
| 269 | if (!attrib) |
| 270 | attrib = Py_None; |
| 271 | |
| 272 | Py_INCREF(attrib); |
| 273 | self->extra->attrib = attrib; |
| 274 | |
| 275 | self->extra->length = 0; |
| 276 | self->extra->allocated = STATIC_CHILDREN; |
| 277 | self->extra->children = self->extra->_children; |
| 278 | |
| 279 | return 0; |
| 280 | } |
| 281 | |
| 282 | LOCAL(void) |
| 283 | element_dealloc_extra(ElementObject* self) |
| 284 | { |
| 285 | int i; |
| 286 | |
| 287 | Py_DECREF(self->extra->attrib); |
| 288 | |
| 289 | for (i = 0; i < self->extra->length; i++) |
| 290 | Py_DECREF(self->extra->children[i]); |
| 291 | |
| 292 | if (self->extra->children != self->extra->_children) |
| 293 | PyObject_Free(self->extra->children); |
| 294 | |
| 295 | PyObject_Free(self->extra); |
| 296 | } |
| 297 | |
| 298 | LOCAL(PyObject*) |
| 299 | element_new(PyObject* tag, PyObject* attrib) |
| 300 | { |
| 301 | ElementObject* self; |
| 302 | |
| 303 | self = PyObject_New(ElementObject, &Element_Type); |
| 304 | if (self == NULL) |
| 305 | return NULL; |
| 306 | |
| 307 | /* use None for empty dictionaries */ |
| 308 | if (PyDict_CheckExact(attrib) && !PyDict_Size(attrib)) |
| 309 | attrib = Py_None; |
| 310 | |
| 311 | self->extra = NULL; |
| 312 | |
| 313 | if (attrib != Py_None) { |
| 314 | |
Neal Norwitz | c6a989a | 2006-05-10 06:57:58 +0000 | [diff] [blame] | 315 | if (element_new_extra(self, attrib) < 0) { |
| 316 | PyObject_Del(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 317 | return NULL; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 318 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 319 | |
| 320 | self->extra->length = 0; |
| 321 | self->extra->allocated = STATIC_CHILDREN; |
| 322 | self->extra->children = self->extra->_children; |
| 323 | |
| 324 | } |
| 325 | |
| 326 | Py_INCREF(tag); |
| 327 | self->tag = tag; |
| 328 | |
| 329 | Py_INCREF(Py_None); |
| 330 | self->text = Py_None; |
| 331 | |
| 332 | Py_INCREF(Py_None); |
| 333 | self->tail = Py_None; |
| 334 | |
| 335 | ALLOC(sizeof(ElementObject), "create element"); |
| 336 | |
| 337 | return (PyObject*) self; |
| 338 | } |
| 339 | |
| 340 | LOCAL(int) |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 341 | element_resize(ElementObject* self, Py_ssize_t extra) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 342 | { |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 343 | Py_ssize_t size; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 344 | PyObject* *children; |
| 345 | |
| 346 | /* make sure self->children can hold the given number of extra |
| 347 | elements. set an exception and return -1 if allocation failed */ |
| 348 | |
| 349 | if (!self->extra) |
| 350 | element_new_extra(self, NULL); |
| 351 | |
| 352 | size = self->extra->length + extra; |
| 353 | |
| 354 | if (size > self->extra->allocated) { |
| 355 | /* use Python 2.4's list growth strategy */ |
| 356 | size = (size >> 3) + (size < 9 ? 3 : 6) + size; |
Christian Heimes | 87dcf3d | 2008-01-18 08:04:57 +0000 | [diff] [blame] | 357 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer "children" |
| 358 | * which needs at least 4 bytes. |
| 359 | * Although it's a false alarm always assume at least one child to |
| 360 | * be safe. |
| 361 | */ |
| 362 | size = size ? size : 1; |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 363 | if ((size_t)size > PY_SSIZE_T_MAX/sizeof(PyObject*)) |
| 364 | goto nomemory; |
| 365 | if (size > INT_MAX) { |
| 366 | PyErr_SetString(PyExc_OverflowError, |
| 367 | "too many children"); |
| 368 | return -1; |
| 369 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 370 | if (self->extra->children != self->extra->_children) { |
Christian Heimes | 87dcf3d | 2008-01-18 08:04:57 +0000 | [diff] [blame] | 371 | /* Coverity CID #182 size_error: Allocating 1 bytes to pointer |
| 372 | * "children", which needs at least 4 bytes. Although it's a |
| 373 | * false alarm always assume at least one child to be safe. |
| 374 | */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 375 | children = PyObject_Realloc(self->extra->children, |
| 376 | size * sizeof(PyObject*)); |
| 377 | if (!children) |
| 378 | goto nomemory; |
| 379 | } else { |
| 380 | children = PyObject_Malloc(size * sizeof(PyObject*)); |
| 381 | if (!children) |
| 382 | goto nomemory; |
| 383 | /* copy existing children from static area to malloc buffer */ |
| 384 | memcpy(children, self->extra->children, |
| 385 | self->extra->length * sizeof(PyObject*)); |
| 386 | } |
| 387 | self->extra->children = children; |
| 388 | self->extra->allocated = size; |
| 389 | } |
| 390 | |
| 391 | return 0; |
| 392 | |
| 393 | nomemory: |
| 394 | PyErr_NoMemory(); |
| 395 | return -1; |
| 396 | } |
| 397 | |
| 398 | LOCAL(int) |
| 399 | element_add_subelement(ElementObject* self, PyObject* element) |
| 400 | { |
| 401 | /* add a child element to a parent */ |
| 402 | |
| 403 | if (element_resize(self, 1) < 0) |
| 404 | return -1; |
| 405 | |
| 406 | Py_INCREF(element); |
| 407 | self->extra->children[self->extra->length] = element; |
| 408 | |
| 409 | self->extra->length++; |
| 410 | |
| 411 | return 0; |
| 412 | } |
| 413 | |
| 414 | LOCAL(PyObject*) |
| 415 | element_get_attrib(ElementObject* self) |
| 416 | { |
| 417 | /* return borrowed reference to attrib dictionary */ |
| 418 | /* note: this function assumes that the extra section exists */ |
| 419 | |
| 420 | PyObject* res = self->extra->attrib; |
| 421 | |
| 422 | if (res == Py_None) { |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 423 | Py_DECREF(res); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 424 | /* create missing dictionary */ |
| 425 | res = PyDict_New(); |
| 426 | if (!res) |
| 427 | return NULL; |
| 428 | self->extra->attrib = res; |
| 429 | } |
| 430 | |
| 431 | return res; |
| 432 | } |
| 433 | |
| 434 | LOCAL(PyObject*) |
| 435 | element_get_text(ElementObject* self) |
| 436 | { |
| 437 | /* return borrowed reference to text attribute */ |
| 438 | |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 439 | PyObject *res = self->text; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 440 | |
| 441 | if (JOIN_GET(res)) { |
| 442 | res = JOIN_OBJ(res); |
| 443 | if (PyList_CheckExact(res)) { |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 444 | PyObject *tmp = list_join(res); |
| 445 | if (!tmp) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 446 | return NULL; |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 447 | self->text = tmp; |
| 448 | Py_DECREF(res); |
| 449 | res = tmp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 450 | } |
| 451 | } |
| 452 | |
| 453 | return res; |
| 454 | } |
| 455 | |
| 456 | LOCAL(PyObject*) |
| 457 | element_get_tail(ElementObject* self) |
| 458 | { |
| 459 | /* return borrowed reference to text attribute */ |
| 460 | |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 461 | PyObject *res = self->tail; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 462 | |
| 463 | if (JOIN_GET(res)) { |
| 464 | res = JOIN_OBJ(res); |
| 465 | if (PyList_CheckExact(res)) { |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 466 | PyObject *tmp = list_join(res); |
| 467 | if (!tmp) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 468 | return NULL; |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 469 | self->tail = tmp; |
| 470 | Py_DECREF(res); |
| 471 | res = tmp; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 472 | } |
| 473 | } |
| 474 | |
| 475 | return res; |
| 476 | } |
| 477 | |
| 478 | static PyObject* |
| 479 | element(PyObject* self, PyObject* args, PyObject* kw) |
| 480 | { |
| 481 | PyObject* elem; |
| 482 | |
| 483 | PyObject* tag; |
| 484 | PyObject* attrib = NULL; |
| 485 | if (!PyArg_ParseTuple(args, "O|O!:Element", &tag, |
| 486 | &PyDict_Type, &attrib)) |
| 487 | return NULL; |
| 488 | |
| 489 | if (attrib || kw) { |
| 490 | attrib = (attrib) ? PyDict_Copy(attrib) : PyDict_New(); |
| 491 | if (!attrib) |
| 492 | return NULL; |
| 493 | if (kw) |
| 494 | PyDict_Update(attrib, kw); |
| 495 | } else { |
| 496 | Py_INCREF(Py_None); |
| 497 | attrib = Py_None; |
| 498 | } |
| 499 | |
| 500 | elem = element_new(tag, attrib); |
| 501 | |
| 502 | Py_DECREF(attrib); |
| 503 | |
| 504 | return elem; |
| 505 | } |
| 506 | |
| 507 | static PyObject* |
| 508 | subelement(PyObject* self, PyObject* args, PyObject* kw) |
| 509 | { |
| 510 | PyObject* elem; |
| 511 | |
| 512 | ElementObject* parent; |
| 513 | PyObject* tag; |
| 514 | PyObject* attrib = NULL; |
| 515 | if (!PyArg_ParseTuple(args, "O!O|O!:SubElement", |
| 516 | &Element_Type, &parent, &tag, |
| 517 | &PyDict_Type, &attrib)) |
| 518 | return NULL; |
| 519 | |
| 520 | if (attrib || kw) { |
| 521 | attrib = (attrib) ? PyDict_Copy(attrib) : PyDict_New(); |
| 522 | if (!attrib) |
| 523 | return NULL; |
| 524 | if (kw) |
| 525 | PyDict_Update(attrib, kw); |
| 526 | } else { |
| 527 | Py_INCREF(Py_None); |
| 528 | attrib = Py_None; |
| 529 | } |
| 530 | |
| 531 | elem = element_new(tag, attrib); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 532 | Py_DECREF(attrib); |
Xiang Zhang | 9c0408d | 2017-03-22 14:32:52 +0800 | [diff] [blame] | 533 | if (elem == NULL) |
| 534 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 535 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 536 | if (element_add_subelement(parent, elem) < 0) { |
| 537 | Py_DECREF(elem); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 538 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 539 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 540 | |
| 541 | return elem; |
| 542 | } |
| 543 | |
| 544 | static void |
| 545 | element_dealloc(ElementObject* self) |
| 546 | { |
Serhiy Storchaka | 1451874 | 2016-12-28 09:23:17 +0200 | [diff] [blame] | 547 | if (self->extra) |
| 548 | element_dealloc_extra(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 549 | |
| 550 | /* discard attributes */ |
| 551 | Py_DECREF(self->tag); |
Serhiy Storchaka | 1451874 | 2016-12-28 09:23:17 +0200 | [diff] [blame] | 552 | Py_DECREF(JOIN_OBJ(self->text)); |
| 553 | Py_DECREF(JOIN_OBJ(self->tail)); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 554 | |
| 555 | RELEASE(sizeof(ElementObject), "destroy element"); |
| 556 | |
| 557 | PyObject_Del(self); |
| 558 | } |
| 559 | |
| 560 | /* -------------------------------------------------------------------- */ |
| 561 | /* methods (in alphabetical order) */ |
| 562 | |
| 563 | static PyObject* |
| 564 | element_append(ElementObject* self, PyObject* args) |
| 565 | { |
| 566 | PyObject* element; |
| 567 | if (!PyArg_ParseTuple(args, "O!:append", &Element_Type, &element)) |
| 568 | return NULL; |
| 569 | |
| 570 | if (element_add_subelement(self, element) < 0) |
| 571 | return NULL; |
| 572 | |
| 573 | Py_RETURN_NONE; |
| 574 | } |
| 575 | |
| 576 | static PyObject* |
| 577 | element_clear(ElementObject* self, PyObject* args) |
| 578 | { |
| 579 | if (!PyArg_ParseTuple(args, ":clear")) |
| 580 | return NULL; |
| 581 | |
| 582 | if (self->extra) { |
| 583 | element_dealloc_extra(self); |
| 584 | self->extra = NULL; |
| 585 | } |
| 586 | |
| 587 | Py_INCREF(Py_None); |
| 588 | Py_DECREF(JOIN_OBJ(self->text)); |
| 589 | self->text = Py_None; |
| 590 | |
| 591 | Py_INCREF(Py_None); |
| 592 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 593 | self->tail = Py_None; |
| 594 | |
| 595 | Py_RETURN_NONE; |
| 596 | } |
| 597 | |
| 598 | static PyObject* |
| 599 | element_copy(ElementObject* self, PyObject* args) |
| 600 | { |
| 601 | int i; |
| 602 | ElementObject* element; |
| 603 | |
| 604 | if (!PyArg_ParseTuple(args, ":__copy__")) |
| 605 | return NULL; |
| 606 | |
| 607 | element = (ElementObject*) element_new( |
| 608 | self->tag, (self->extra) ? self->extra->attrib : Py_None |
| 609 | ); |
| 610 | if (!element) |
| 611 | return NULL; |
| 612 | |
| 613 | Py_DECREF(JOIN_OBJ(element->text)); |
| 614 | element->text = self->text; |
| 615 | Py_INCREF(JOIN_OBJ(element->text)); |
| 616 | |
| 617 | Py_DECREF(JOIN_OBJ(element->tail)); |
| 618 | element->tail = self->tail; |
| 619 | Py_INCREF(JOIN_OBJ(element->tail)); |
| 620 | |
| 621 | if (self->extra) { |
| 622 | |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 623 | if (element_resize(element, self->extra->length) < 0) { |
| 624 | Py_DECREF(element); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 625 | return NULL; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 626 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 627 | |
| 628 | for (i = 0; i < self->extra->length; i++) { |
| 629 | Py_INCREF(self->extra->children[i]); |
| 630 | element->extra->children[i] = self->extra->children[i]; |
| 631 | } |
| 632 | |
| 633 | element->extra->length = self->extra->length; |
| 634 | |
| 635 | } |
| 636 | |
| 637 | return (PyObject*) element; |
| 638 | } |
| 639 | |
| 640 | static PyObject* |
| 641 | element_deepcopy(ElementObject* self, PyObject* args) |
| 642 | { |
| 643 | int i; |
| 644 | ElementObject* element; |
| 645 | PyObject* tag; |
| 646 | PyObject* attrib; |
| 647 | PyObject* text; |
| 648 | PyObject* tail; |
| 649 | PyObject* id; |
| 650 | |
| 651 | PyObject* memo; |
| 652 | if (!PyArg_ParseTuple(args, "O:__deepcopy__", &memo)) |
| 653 | return NULL; |
| 654 | |
| 655 | tag = deepcopy(self->tag, memo); |
| 656 | if (!tag) |
| 657 | return NULL; |
| 658 | |
| 659 | if (self->extra) { |
| 660 | attrib = deepcopy(self->extra->attrib, memo); |
| 661 | if (!attrib) { |
| 662 | Py_DECREF(tag); |
| 663 | return NULL; |
| 664 | } |
| 665 | } else { |
| 666 | Py_INCREF(Py_None); |
| 667 | attrib = Py_None; |
| 668 | } |
| 669 | |
| 670 | element = (ElementObject*) element_new(tag, attrib); |
| 671 | |
| 672 | Py_DECREF(tag); |
| 673 | Py_DECREF(attrib); |
| 674 | |
| 675 | if (!element) |
| 676 | return NULL; |
| 677 | |
| 678 | text = deepcopy(JOIN_OBJ(self->text), memo); |
| 679 | if (!text) |
| 680 | goto error; |
| 681 | Py_DECREF(element->text); |
| 682 | element->text = JOIN_SET(text, JOIN_GET(self->text)); |
| 683 | |
| 684 | tail = deepcopy(JOIN_OBJ(self->tail), memo); |
| 685 | if (!tail) |
| 686 | goto error; |
| 687 | Py_DECREF(element->tail); |
| 688 | element->tail = JOIN_SET(tail, JOIN_GET(self->tail)); |
| 689 | |
| 690 | if (self->extra) { |
| 691 | |
| 692 | if (element_resize(element, self->extra->length) < 0) |
| 693 | goto error; |
| 694 | |
| 695 | for (i = 0; i < self->extra->length; i++) { |
| 696 | PyObject* child = deepcopy(self->extra->children[i], memo); |
| 697 | if (!child) { |
| 698 | element->extra->length = i; |
| 699 | goto error; |
| 700 | } |
| 701 | element->extra->children[i] = child; |
| 702 | } |
| 703 | |
| 704 | element->extra->length = self->extra->length; |
| 705 | |
| 706 | } |
| 707 | |
| 708 | /* add object to memo dictionary (so deepcopy won't visit it again) */ |
| 709 | id = PyInt_FromLong((Py_uintptr_t) self); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 710 | if (!id) |
| 711 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 712 | |
| 713 | i = PyDict_SetItem(memo, id, (PyObject*) element); |
| 714 | |
| 715 | Py_DECREF(id); |
| 716 | |
| 717 | if (i < 0) |
| 718 | goto error; |
| 719 | |
| 720 | return (PyObject*) element; |
| 721 | |
| 722 | error: |
| 723 | Py_DECREF(element); |
| 724 | return NULL; |
| 725 | } |
| 726 | |
| 727 | LOCAL(int) |
| 728 | checkpath(PyObject* tag) |
| 729 | { |
Neal Norwitz | c707438 | 2006-06-12 02:06:17 +0000 | [diff] [blame] | 730 | Py_ssize_t i; |
| 731 | int check = 1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 732 | |
| 733 | /* check if a tag contains an xpath character */ |
| 734 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 735 | #define PATHCHAR(ch) \ |
| 736 | (ch == '/' || ch == '*' || ch == '[' || ch == '@' || ch == '.') |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 737 | |
| 738 | #if defined(Py_USING_UNICODE) |
| 739 | if (PyUnicode_Check(tag)) { |
| 740 | Py_UNICODE *p = PyUnicode_AS_UNICODE(tag); |
| 741 | for (i = 0; i < PyUnicode_GET_SIZE(tag); i++) { |
| 742 | if (p[i] == '{') |
| 743 | check = 0; |
| 744 | else if (p[i] == '}') |
| 745 | check = 1; |
| 746 | else if (check && PATHCHAR(p[i])) |
| 747 | return 1; |
| 748 | } |
| 749 | return 0; |
| 750 | } |
| 751 | #endif |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 752 | if (PyString_Check(tag)) { |
| 753 | char *p = PyString_AS_STRING(tag); |
| 754 | for (i = 0; i < PyString_GET_SIZE(tag); i++) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 755 | if (p[i] == '{') |
| 756 | check = 0; |
| 757 | else if (p[i] == '}') |
| 758 | check = 1; |
| 759 | else if (check && PATHCHAR(p[i])) |
| 760 | return 1; |
| 761 | } |
| 762 | return 0; |
| 763 | } |
| 764 | |
| 765 | return 1; /* unknown type; might be path expression */ |
| 766 | } |
| 767 | |
| 768 | static PyObject* |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 769 | element_extend(ElementObject* self, PyObject* args) |
| 770 | { |
| 771 | PyObject* seq; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 772 | Py_ssize_t i; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 773 | |
| 774 | PyObject* seq_in; |
| 775 | if (!PyArg_ParseTuple(args, "O:extend", &seq_in)) |
| 776 | return NULL; |
| 777 | |
| 778 | seq = PySequence_Fast(seq_in, ""); |
| 779 | if (!seq) { |
| 780 | PyErr_Format( |
| 781 | PyExc_TypeError, |
| 782 | "expected sequence, not \"%.200s\"", Py_TYPE(seq_in)->tp_name |
| 783 | ); |
| 784 | return NULL; |
| 785 | } |
| 786 | |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 787 | for (i = 0; i < PySequence_Fast_GET_SIZE(seq); i++) { |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 788 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
| 789 | if (element_add_subelement(self, element) < 0) { |
| 790 | Py_DECREF(seq); |
| 791 | return NULL; |
| 792 | } |
| 793 | } |
| 794 | |
| 795 | Py_DECREF(seq); |
| 796 | |
| 797 | Py_RETURN_NONE; |
| 798 | } |
| 799 | |
| 800 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 801 | element_find(ElementObject* self, PyObject* args) |
| 802 | { |
| 803 | int i; |
| 804 | |
| 805 | PyObject* tag; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 806 | PyObject* namespaces = Py_None; |
| 807 | if (!PyArg_ParseTuple(args, "O|O:find", &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 808 | return NULL; |
| 809 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 810 | if (checkpath(tag) || namespaces != Py_None) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 811 | return PyObject_CallMethod( |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 812 | elementpath_obj, "find", "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 813 | ); |
| 814 | |
| 815 | if (!self->extra) |
| 816 | Py_RETURN_NONE; |
| 817 | |
| 818 | for (i = 0; i < self->extra->length; i++) { |
| 819 | PyObject* item = self->extra->children[i]; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 820 | int rc; |
| 821 | if (!Element_CheckExact(item)) |
| 822 | continue; |
| 823 | Py_INCREF(item); |
| 824 | rc = PyObject_Compare(((ElementObject*)item)->tag, tag); |
| 825 | if (rc == 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 826 | return item; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 827 | Py_DECREF(item); |
| 828 | if (rc < 0 && PyErr_Occurred()) |
| 829 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 830 | } |
| 831 | |
| 832 | Py_RETURN_NONE; |
| 833 | } |
| 834 | |
| 835 | static PyObject* |
| 836 | element_findtext(ElementObject* self, PyObject* args) |
| 837 | { |
| 838 | int i; |
| 839 | |
| 840 | PyObject* tag; |
| 841 | PyObject* default_value = Py_None; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 842 | PyObject* namespaces = Py_None; |
| 843 | if (!PyArg_ParseTuple(args, "O|OO:findtext", &tag, &default_value, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 844 | return NULL; |
| 845 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 846 | if (checkpath(tag) || namespaces != Py_None) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 847 | return PyObject_CallMethod( |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 848 | elementpath_obj, "findtext", "OOOO", self, tag, default_value, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 849 | ); |
| 850 | |
| 851 | if (!self->extra) { |
| 852 | Py_INCREF(default_value); |
| 853 | return default_value; |
| 854 | } |
| 855 | |
| 856 | for (i = 0; i < self->extra->length; i++) { |
| 857 | ElementObject* item = (ElementObject*) self->extra->children[i]; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 858 | int rc; |
| 859 | if (!Element_CheckExact(item)) |
| 860 | continue; |
| 861 | Py_INCREF(item); |
| 862 | rc = PyObject_Compare(item->tag, tag); |
| 863 | if (rc == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 864 | PyObject* text = element_get_text(item); |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 865 | if (text == Py_None) { |
| 866 | Py_DECREF(item); |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 867 | return PyString_FromString(""); |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 868 | } |
Neal Norwitz | 6f5ff3f | 2006-08-12 01:43:40 +0000 | [diff] [blame] | 869 | Py_XINCREF(text); |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 870 | Py_DECREF(item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 871 | return text; |
| 872 | } |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 873 | Py_DECREF(item); |
| 874 | if (rc < 0 && PyErr_Occurred()) |
| 875 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 876 | } |
| 877 | |
| 878 | Py_INCREF(default_value); |
| 879 | return default_value; |
| 880 | } |
| 881 | |
| 882 | static PyObject* |
| 883 | element_findall(ElementObject* self, PyObject* args) |
| 884 | { |
| 885 | int i; |
| 886 | PyObject* out; |
| 887 | |
| 888 | PyObject* tag; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 889 | PyObject* namespaces = Py_None; |
| 890 | if (!PyArg_ParseTuple(args, "O|O:findall", &tag, &namespaces)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 891 | return NULL; |
| 892 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 893 | if (checkpath(tag) || namespaces != Py_None) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 894 | return PyObject_CallMethod( |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 895 | elementpath_obj, "findall", "OOO", self, tag, namespaces |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 896 | ); |
| 897 | |
| 898 | out = PyList_New(0); |
| 899 | if (!out) |
| 900 | return NULL; |
| 901 | |
| 902 | if (!self->extra) |
| 903 | return out; |
| 904 | |
| 905 | for (i = 0; i < self->extra->length; i++) { |
| 906 | PyObject* item = self->extra->children[i]; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 907 | int rc; |
| 908 | if (!Element_CheckExact(item)) |
| 909 | continue; |
| 910 | Py_INCREF(item); |
| 911 | rc = PyObject_Compare(((ElementObject*)item)->tag, tag); |
| 912 | if (rc == 0) |
| 913 | rc = PyList_Append(out, item); |
| 914 | Py_DECREF(item); |
| 915 | if (rc < 0 && PyErr_Occurred()) { |
| 916 | Py_DECREF(out); |
| 917 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 918 | } |
| 919 | } |
| 920 | |
| 921 | return out; |
| 922 | } |
| 923 | |
| 924 | static PyObject* |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 925 | element_iterfind(ElementObject* self, PyObject* args) |
| 926 | { |
| 927 | PyObject* tag; |
| 928 | PyObject* namespaces = Py_None; |
| 929 | if (!PyArg_ParseTuple(args, "O|O:iterfind", &tag, &namespaces)) |
| 930 | return NULL; |
| 931 | |
| 932 | return PyObject_CallMethod( |
| 933 | elementpath_obj, "iterfind", "OOO", self, tag, namespaces |
| 934 | ); |
| 935 | } |
| 936 | |
| 937 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 938 | element_get(ElementObject* self, PyObject* args) |
| 939 | { |
| 940 | PyObject* value; |
| 941 | |
| 942 | PyObject* key; |
| 943 | PyObject* default_value = Py_None; |
| 944 | if (!PyArg_ParseTuple(args, "O|O:get", &key, &default_value)) |
| 945 | return NULL; |
| 946 | |
| 947 | if (!self->extra || self->extra->attrib == Py_None) |
| 948 | value = default_value; |
| 949 | else { |
| 950 | value = PyDict_GetItem(self->extra->attrib, key); |
| 951 | if (!value) |
| 952 | value = default_value; |
| 953 | } |
| 954 | |
| 955 | Py_INCREF(value); |
| 956 | return value; |
| 957 | } |
| 958 | |
| 959 | static PyObject* |
| 960 | element_getchildren(ElementObject* self, PyObject* args) |
| 961 | { |
| 962 | int i; |
| 963 | PyObject* list; |
| 964 | |
Serhiy Storchaka | 09b5247 | 2017-05-17 10:08:11 +0300 | [diff] [blame] | 965 | if (PyErr_WarnPy3k("This method will be removed in future versions. " |
| 966 | "Use 'list(elem)' or iteration over elem instead.", |
| 967 | 1) < 0) { |
| 968 | return NULL; |
| 969 | } |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 970 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 971 | if (!PyArg_ParseTuple(args, ":getchildren")) |
| 972 | return NULL; |
| 973 | |
| 974 | if (!self->extra) |
| 975 | return PyList_New(0); |
| 976 | |
| 977 | list = PyList_New(self->extra->length); |
| 978 | if (!list) |
| 979 | return NULL; |
| 980 | |
| 981 | for (i = 0; i < self->extra->length; i++) { |
| 982 | PyObject* item = self->extra->children[i]; |
| 983 | Py_INCREF(item); |
| 984 | PyList_SET_ITEM(list, i, item); |
| 985 | } |
| 986 | |
| 987 | return list; |
| 988 | } |
| 989 | |
| 990 | static PyObject* |
Serhiy Storchaka | 09b5247 | 2017-05-17 10:08:11 +0300 | [diff] [blame] | 991 | element_iter_impl(ElementObject* self, PyObject* tag) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 992 | { |
Serhiy Storchaka | 09b5247 | 2017-05-17 10:08:11 +0300 | [diff] [blame] | 993 | PyObject* args; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 994 | PyObject* result; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 995 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 996 | if (!elementtree_iter_obj) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 997 | PyErr_SetString( |
| 998 | PyExc_RuntimeError, |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 999 | "iter helper not found" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1000 | ); |
| 1001 | return NULL; |
| 1002 | } |
| 1003 | |
| 1004 | args = PyTuple_New(2); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1005 | if (!args) |
| 1006 | return NULL; |
Neal Norwitz | 02876df | 2006-02-07 06:58:52 +0000 | [diff] [blame] | 1007 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1008 | Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self); |
| 1009 | Py_INCREF(tag); PyTuple_SET_ITEM(args, 1, (PyObject*) tag); |
| 1010 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1011 | result = PyObject_CallObject(elementtree_iter_obj, args); |
| 1012 | |
| 1013 | Py_DECREF(args); |
| 1014 | |
| 1015 | return result; |
| 1016 | } |
| 1017 | |
Serhiy Storchaka | 09b5247 | 2017-05-17 10:08:11 +0300 | [diff] [blame] | 1018 | static PyObject* |
| 1019 | element_iter(ElementObject* self, PyObject* args) |
| 1020 | { |
| 1021 | PyObject* tag = Py_None; |
| 1022 | if (!PyArg_ParseTuple(args, "|O:iter", &tag)) |
| 1023 | return NULL; |
| 1024 | |
| 1025 | return element_iter_impl(self, tag); |
| 1026 | } |
| 1027 | |
| 1028 | static PyObject* |
| 1029 | element_getiterator(ElementObject* self, PyObject* args) |
| 1030 | { |
| 1031 | PyObject* tag = Py_None; |
| 1032 | if (!PyArg_ParseTuple(args, "|O:getiterator", &tag)) |
| 1033 | return NULL; |
| 1034 | |
| 1035 | /* Change for a DeprecationWarning in 1.4 */ |
| 1036 | if (Py_Py3kWarningFlag && |
| 1037 | PyErr_WarnEx(PyExc_PendingDeprecationWarning, |
| 1038 | "This method will be removed in future versions. " |
| 1039 | "Use 'tree.iter()' or 'list(tree.iter())' instead.", |
| 1040 | 1) < 0) { |
| 1041 | return NULL; |
| 1042 | } |
| 1043 | return element_iter_impl(self, tag); |
| 1044 | } |
| 1045 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1046 | |
| 1047 | static PyObject* |
| 1048 | element_itertext(ElementObject* self, PyObject* args) |
| 1049 | { |
| 1050 | PyObject* result; |
| 1051 | |
| 1052 | if (!PyArg_ParseTuple(args, ":itertext")) |
| 1053 | return NULL; |
| 1054 | |
| 1055 | if (!elementtree_itertext_obj) { |
| 1056 | PyErr_SetString( |
| 1057 | PyExc_RuntimeError, |
| 1058 | "itertext helper not found" |
| 1059 | ); |
| 1060 | return NULL; |
| 1061 | } |
| 1062 | |
| 1063 | args = PyTuple_New(1); |
| 1064 | if (!args) |
| 1065 | return NULL; |
| 1066 | |
| 1067 | Py_INCREF(self); PyTuple_SET_ITEM(args, 0, (PyObject*) self); |
| 1068 | |
| 1069 | result = PyObject_CallObject(elementtree_itertext_obj, args); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1070 | |
| 1071 | Py_DECREF(args); |
| 1072 | |
| 1073 | return result; |
| 1074 | } |
| 1075 | |
| 1076 | static PyObject* |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1077 | element_getitem(PyObject* self_, Py_ssize_t index) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1078 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1079 | ElementObject* self = (ElementObject*) self_; |
| 1080 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1081 | if (!self->extra || index < 0 || index >= self->extra->length) { |
| 1082 | PyErr_SetString( |
| 1083 | PyExc_IndexError, |
| 1084 | "child index out of range" |
| 1085 | ); |
| 1086 | return NULL; |
| 1087 | } |
| 1088 | |
| 1089 | Py_INCREF(self->extra->children[index]); |
| 1090 | return self->extra->children[index]; |
| 1091 | } |
| 1092 | |
| 1093 | static PyObject* |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1094 | element_insert(ElementObject* self, PyObject* args) |
| 1095 | { |
| 1096 | int i; |
| 1097 | |
| 1098 | int index; |
| 1099 | PyObject* element; |
| 1100 | if (!PyArg_ParseTuple(args, "iO!:insert", &index, |
| 1101 | &Element_Type, &element)) |
| 1102 | return NULL; |
| 1103 | |
| 1104 | if (!self->extra) |
| 1105 | element_new_extra(self, NULL); |
| 1106 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1107 | if (index < 0) { |
| 1108 | index += self->extra->length; |
| 1109 | if (index < 0) |
| 1110 | index = 0; |
| 1111 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1112 | if (index > self->extra->length) |
| 1113 | index = self->extra->length; |
| 1114 | |
| 1115 | if (element_resize(self, 1) < 0) |
| 1116 | return NULL; |
| 1117 | |
| 1118 | for (i = self->extra->length; i > index; i--) |
| 1119 | self->extra->children[i] = self->extra->children[i-1]; |
| 1120 | |
| 1121 | Py_INCREF(element); |
| 1122 | self->extra->children[index] = element; |
| 1123 | |
| 1124 | self->extra->length++; |
| 1125 | |
| 1126 | Py_RETURN_NONE; |
| 1127 | } |
| 1128 | |
| 1129 | static PyObject* |
| 1130 | element_items(ElementObject* self, PyObject* args) |
| 1131 | { |
| 1132 | if (!PyArg_ParseTuple(args, ":items")) |
| 1133 | return NULL; |
| 1134 | |
| 1135 | if (!self->extra || self->extra->attrib == Py_None) |
| 1136 | return PyList_New(0); |
| 1137 | |
| 1138 | return PyDict_Items(self->extra->attrib); |
| 1139 | } |
| 1140 | |
| 1141 | static PyObject* |
| 1142 | element_keys(ElementObject* self, PyObject* args) |
| 1143 | { |
| 1144 | if (!PyArg_ParseTuple(args, ":keys")) |
| 1145 | return NULL; |
| 1146 | |
| 1147 | if (!self->extra || self->extra->attrib == Py_None) |
| 1148 | return PyList_New(0); |
| 1149 | |
| 1150 | return PyDict_Keys(self->extra->attrib); |
| 1151 | } |
| 1152 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1153 | static Py_ssize_t |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1154 | element_length(ElementObject* self) |
| 1155 | { |
| 1156 | if (!self->extra) |
| 1157 | return 0; |
| 1158 | |
| 1159 | return self->extra->length; |
| 1160 | } |
| 1161 | |
| 1162 | static PyObject* |
| 1163 | element_makeelement(PyObject* self, PyObject* args, PyObject* kw) |
| 1164 | { |
| 1165 | PyObject* elem; |
| 1166 | |
| 1167 | PyObject* tag; |
| 1168 | PyObject* attrib; |
| 1169 | if (!PyArg_ParseTuple(args, "OO:makeelement", &tag, &attrib)) |
| 1170 | return NULL; |
| 1171 | |
| 1172 | attrib = PyDict_Copy(attrib); |
| 1173 | if (!attrib) |
| 1174 | return NULL; |
| 1175 | |
| 1176 | elem = element_new(tag, attrib); |
| 1177 | |
| 1178 | Py_DECREF(attrib); |
| 1179 | |
| 1180 | return elem; |
| 1181 | } |
| 1182 | |
| 1183 | static PyObject* |
| 1184 | element_reduce(ElementObject* self, PyObject* args) |
| 1185 | { |
| 1186 | if (!PyArg_ParseTuple(args, ":__reduce__")) |
| 1187 | return NULL; |
| 1188 | |
| 1189 | /* Hack alert: This method is used to work around a __copy__ |
| 1190 | problem on certain 2.3 and 2.4 versions. To save time and |
| 1191 | simplify the code, we create the copy in here, and use a dummy |
| 1192 | copyelement helper to trick the copy module into doing the |
| 1193 | right thing. */ |
| 1194 | |
| 1195 | if (!elementtree_copyelement_obj) { |
| 1196 | PyErr_SetString( |
| 1197 | PyExc_RuntimeError, |
| 1198 | "copyelement helper not found" |
| 1199 | ); |
| 1200 | return NULL; |
| 1201 | } |
| 1202 | |
| 1203 | return Py_BuildValue( |
| 1204 | "O(N)", elementtree_copyelement_obj, element_copy(self, args) |
| 1205 | ); |
| 1206 | } |
| 1207 | |
| 1208 | static PyObject* |
| 1209 | element_remove(ElementObject* self, PyObject* args) |
| 1210 | { |
| 1211 | int i; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 1212 | int rc; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1213 | PyObject* element; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 1214 | PyObject* found; |
| 1215 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1216 | if (!PyArg_ParseTuple(args, "O!:remove", &Element_Type, &element)) |
| 1217 | return NULL; |
| 1218 | |
| 1219 | if (!self->extra) { |
| 1220 | /* element has no children, so raise exception */ |
| 1221 | PyErr_SetString( |
| 1222 | PyExc_ValueError, |
| 1223 | "list.remove(x): x not in list" |
| 1224 | ); |
| 1225 | return NULL; |
| 1226 | } |
| 1227 | |
| 1228 | for (i = 0; i < self->extra->length; i++) { |
| 1229 | if (self->extra->children[i] == element) |
| 1230 | break; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 1231 | rc = PyObject_Compare(self->extra->children[i], element); |
| 1232 | if (rc == 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1233 | break; |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 1234 | if (rc < 0 && PyErr_Occurred()) |
| 1235 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1236 | } |
| 1237 | |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 1238 | if (i >= self->extra->length) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1239 | /* element is not in children, so raise exception */ |
| 1240 | PyErr_SetString( |
| 1241 | PyExc_ValueError, |
| 1242 | "list.remove(x): x not in list" |
| 1243 | ); |
| 1244 | return NULL; |
| 1245 | } |
| 1246 | |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 1247 | found = self->extra->children[i]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1248 | |
| 1249 | self->extra->length--; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1250 | for (; i < self->extra->length; i++) |
| 1251 | self->extra->children[i] = self->extra->children[i+1]; |
| 1252 | |
Serhiy Storchaka | 25598f3 | 2015-05-18 18:28:57 +0300 | [diff] [blame] | 1253 | Py_DECREF(found); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1254 | Py_RETURN_NONE; |
| 1255 | } |
| 1256 | |
| 1257 | static PyObject* |
| 1258 | element_repr(ElementObject* self) |
| 1259 | { |
Serhiy Storchaka | 1f7586e | 2016-06-12 10:06:32 +0300 | [diff] [blame] | 1260 | int status; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1261 | |
Serhiy Storchaka | 1f7586e | 2016-06-12 10:06:32 +0300 | [diff] [blame] | 1262 | if (self->tag == NULL) |
| 1263 | return PyUnicode_FromFormat("<Element at %p>", self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1264 | |
Serhiy Storchaka | 1f7586e | 2016-06-12 10:06:32 +0300 | [diff] [blame] | 1265 | status = Py_ReprEnter((PyObject *)self); |
| 1266 | if (status == 0) { |
| 1267 | PyObject *repr, *tag; |
| 1268 | tag = PyObject_Repr(self->tag); |
Stéphane Wirtel | 41af942 | 2017-06-12 15:30:48 +0200 | [diff] [blame] | 1269 | if (!tag) { |
| 1270 | Py_ReprLeave((PyObject *)self); |
Serhiy Storchaka | 1f7586e | 2016-06-12 10:06:32 +0300 | [diff] [blame] | 1271 | return NULL; |
Stéphane Wirtel | 41af942 | 2017-06-12 15:30:48 +0200 | [diff] [blame] | 1272 | } |
Florent Xicluna | e2e81e8 | 2010-03-11 15:55:11 +0000 | [diff] [blame] | 1273 | |
Serhiy Storchaka | 1f7586e | 2016-06-12 10:06:32 +0300 | [diff] [blame] | 1274 | repr = PyString_FromFormat("<Element %s at %p>", |
| 1275 | PyString_AS_STRING(tag), self); |
Benjamin Peterson | d7324bc | 2016-12-03 11:30:04 -0800 | [diff] [blame] | 1276 | Py_ReprLeave((PyObject *)self); |
Serhiy Storchaka | 1f7586e | 2016-06-12 10:06:32 +0300 | [diff] [blame] | 1277 | Py_DECREF(tag); |
| 1278 | return repr; |
| 1279 | } |
| 1280 | if (status > 0) |
| 1281 | PyErr_Format(PyExc_RuntimeError, |
| 1282 | "reentrant call inside %s.__repr__", |
| 1283 | Py_TYPE(self)->tp_name); |
| 1284 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1285 | } |
| 1286 | |
| 1287 | static PyObject* |
| 1288 | element_set(ElementObject* self, PyObject* args) |
| 1289 | { |
| 1290 | PyObject* attrib; |
| 1291 | |
| 1292 | PyObject* key; |
| 1293 | PyObject* value; |
| 1294 | if (!PyArg_ParseTuple(args, "OO:set", &key, &value)) |
| 1295 | return NULL; |
| 1296 | |
| 1297 | if (!self->extra) |
| 1298 | element_new_extra(self, NULL); |
| 1299 | |
| 1300 | attrib = element_get_attrib(self); |
| 1301 | if (!attrib) |
| 1302 | return NULL; |
| 1303 | |
| 1304 | if (PyDict_SetItem(attrib, key, value) < 0) |
| 1305 | return NULL; |
| 1306 | |
| 1307 | Py_RETURN_NONE; |
| 1308 | } |
| 1309 | |
| 1310 | static int |
Serhiy Storchaka | b5b76c3 | 2015-11-26 11:21:47 +0200 | [diff] [blame] | 1311 | element_setitem(PyObject* self_, Py_ssize_t index_, PyObject* item) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1312 | { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1313 | ElementObject* self = (ElementObject*) self_; |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 1314 | int i, index; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1315 | PyObject* old; |
| 1316 | |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 1317 | if (!self->extra || index_ < 0 || index_ >= self->extra->length) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1318 | PyErr_SetString( |
| 1319 | PyExc_IndexError, |
| 1320 | "child assignment index out of range"); |
| 1321 | return -1; |
| 1322 | } |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 1323 | index = (int)index_; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1324 | |
| 1325 | old = self->extra->children[index]; |
| 1326 | |
| 1327 | if (item) { |
| 1328 | Py_INCREF(item); |
| 1329 | self->extra->children[index] = item; |
| 1330 | } else { |
| 1331 | self->extra->length--; |
| 1332 | for (i = index; i < self->extra->length; i++) |
| 1333 | self->extra->children[i] = self->extra->children[i+1]; |
| 1334 | } |
| 1335 | |
| 1336 | Py_DECREF(old); |
| 1337 | |
| 1338 | return 0; |
| 1339 | } |
| 1340 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1341 | static PyObject* |
| 1342 | element_subscr(PyObject* self_, PyObject* item) |
| 1343 | { |
| 1344 | ElementObject* self = (ElementObject*) self_; |
| 1345 | |
| 1346 | #if (PY_VERSION_HEX < 0x02050000) |
| 1347 | if (PyInt_Check(item) || PyLong_Check(item)) { |
| 1348 | long i = PyInt_AsLong(item); |
| 1349 | #else |
| 1350 | if (PyIndex_Check(item)) { |
| 1351 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
| 1352 | #endif |
| 1353 | |
| 1354 | if (i == -1 && PyErr_Occurred()) { |
| 1355 | return NULL; |
| 1356 | } |
| 1357 | if (i < 0 && self->extra) |
| 1358 | i += self->extra->length; |
| 1359 | return element_getitem(self_, i); |
| 1360 | } |
| 1361 | else if (PySlice_Check(item)) { |
| 1362 | Py_ssize_t start, stop, step, slicelen, cur, i; |
| 1363 | PyObject* list; |
| 1364 | |
| 1365 | if (!self->extra) |
| 1366 | return PyList_New(0); |
| 1367 | |
Serhiy Storchaka | 5e79321 | 2017-04-15 20:11:12 +0300 | [diff] [blame] | 1368 | if (_PySlice_Unpack(item, &start, &stop, &step) < 0) { |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1369 | return NULL; |
| 1370 | } |
Serhiy Storchaka | e41390a | 2017-04-08 11:48:57 +0300 | [diff] [blame] | 1371 | slicelen = _PySlice_AdjustIndices(self->extra->length, &start, &stop, |
| 1372 | step); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1373 | |
| 1374 | if (slicelen <= 0) |
| 1375 | return PyList_New(0); |
| 1376 | else { |
| 1377 | list = PyList_New(slicelen); |
| 1378 | if (!list) |
| 1379 | return NULL; |
| 1380 | |
| 1381 | for (cur = start, i = 0; i < slicelen; |
| 1382 | cur += step, i++) { |
| 1383 | PyObject* item = self->extra->children[cur]; |
| 1384 | Py_INCREF(item); |
| 1385 | PyList_SET_ITEM(list, i, item); |
| 1386 | } |
| 1387 | |
| 1388 | return list; |
| 1389 | } |
| 1390 | } |
| 1391 | else { |
| 1392 | PyErr_SetString(PyExc_TypeError, |
| 1393 | "element indices must be integers"); |
| 1394 | return NULL; |
| 1395 | } |
| 1396 | } |
| 1397 | |
| 1398 | static int |
| 1399 | element_ass_subscr(PyObject* self_, PyObject* item, PyObject* value) |
| 1400 | { |
| 1401 | ElementObject* self = (ElementObject*) self_; |
| 1402 | |
| 1403 | #if (PY_VERSION_HEX < 0x02050000) |
| 1404 | if (PyInt_Check(item) || PyLong_Check(item)) { |
| 1405 | long i = PyInt_AsLong(item); |
| 1406 | #else |
| 1407 | if (PyIndex_Check(item)) { |
| 1408 | Py_ssize_t i = PyNumber_AsSsize_t(item, PyExc_IndexError); |
| 1409 | #endif |
| 1410 | |
| 1411 | if (i == -1 && PyErr_Occurred()) { |
| 1412 | return -1; |
| 1413 | } |
| 1414 | if (i < 0 && self->extra) |
| 1415 | i += self->extra->length; |
| 1416 | return element_setitem(self_, i, value); |
| 1417 | } |
| 1418 | else if (PySlice_Check(item)) { |
| 1419 | Py_ssize_t start, stop, step, slicelen, newlen, cur, i; |
| 1420 | |
| 1421 | PyObject* recycle = NULL; |
| 1422 | PyObject* seq = NULL; |
| 1423 | |
| 1424 | if (!self->extra) |
| 1425 | element_new_extra(self, NULL); |
| 1426 | |
Serhiy Storchaka | 5e79321 | 2017-04-15 20:11:12 +0300 | [diff] [blame] | 1427 | if (_PySlice_Unpack(item, &start, &stop, &step) < 0) { |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1428 | return -1; |
| 1429 | } |
Serhiy Storchaka | e41390a | 2017-04-08 11:48:57 +0300 | [diff] [blame] | 1430 | slicelen = _PySlice_AdjustIndices(self->extra->length, &start, &stop, |
| 1431 | step); |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 1432 | assert(slicelen <= self->extra->length); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1433 | |
| 1434 | if (value == NULL) |
| 1435 | newlen = 0; |
| 1436 | else { |
| 1437 | seq = PySequence_Fast(value, ""); |
| 1438 | if (!seq) { |
| 1439 | PyErr_Format( |
| 1440 | PyExc_TypeError, |
| 1441 | "expected sequence, not \"%.200s\"", Py_TYPE(value)->tp_name |
| 1442 | ); |
| 1443 | return -1; |
| 1444 | } |
| 1445 | newlen = PySequence_Size(seq); |
| 1446 | } |
| 1447 | |
| 1448 | if (step != 1 && newlen != slicelen) |
| 1449 | { |
Serhiy Storchaka | a0ae9ff | 2015-11-22 12:31:11 +0200 | [diff] [blame] | 1450 | Py_XDECREF(seq); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1451 | PyErr_Format(PyExc_ValueError, |
| 1452 | #if (PY_VERSION_HEX < 0x02050000) |
| 1453 | "attempt to assign sequence of size %d " |
| 1454 | "to extended slice of size %d", |
Serhiy Storchaka | a0ae9ff | 2015-11-22 12:31:11 +0200 | [diff] [blame] | 1455 | (int)newlen, (int)slicelen |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1456 | #else |
| 1457 | "attempt to assign sequence of size %zd " |
| 1458 | "to extended slice of size %zd", |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1459 | newlen, slicelen |
Serhiy Storchaka | a0ae9ff | 2015-11-22 12:31:11 +0200 | [diff] [blame] | 1460 | #endif |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1461 | ); |
| 1462 | return -1; |
| 1463 | } |
| 1464 | |
| 1465 | |
| 1466 | /* Resize before creating the recycle bin, to prevent refleaks. */ |
| 1467 | if (newlen > slicelen) { |
| 1468 | if (element_resize(self, newlen - slicelen) < 0) { |
Serhiy Storchaka | a0ae9ff | 2015-11-22 12:31:11 +0200 | [diff] [blame] | 1469 | Py_XDECREF(seq); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1470 | return -1; |
| 1471 | } |
| 1472 | } |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 1473 | assert(newlen - slicelen <= INT_MAX - self->extra->length); |
| 1474 | assert(newlen - slicelen >= -self->extra->length); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1475 | |
| 1476 | if (slicelen > 0) { |
| 1477 | /* to avoid recursive calls to this method (via decref), move |
| 1478 | old items to the recycle bin here, and get rid of them when |
| 1479 | we're done modifying the element */ |
| 1480 | recycle = PyList_New(slicelen); |
| 1481 | if (!recycle) { |
Serhiy Storchaka | a0ae9ff | 2015-11-22 12:31:11 +0200 | [diff] [blame] | 1482 | Py_XDECREF(seq); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1483 | return -1; |
| 1484 | } |
| 1485 | for (cur = start, i = 0; i < slicelen; |
| 1486 | cur += step, i++) |
| 1487 | PyList_SET_ITEM(recycle, i, self->extra->children[cur]); |
| 1488 | } |
| 1489 | |
| 1490 | if (newlen < slicelen) { |
| 1491 | /* delete slice */ |
| 1492 | for (i = stop; i < self->extra->length; i++) |
| 1493 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1494 | } else if (newlen > slicelen) { |
| 1495 | /* insert slice */ |
| 1496 | for (i = self->extra->length-1; i >= stop; i--) |
| 1497 | self->extra->children[i + newlen - slicelen] = self->extra->children[i]; |
| 1498 | } |
| 1499 | |
| 1500 | /* replace the slice */ |
| 1501 | for (cur = start, i = 0; i < newlen; |
| 1502 | cur += step, i++) { |
| 1503 | PyObject* element = PySequence_Fast_GET_ITEM(seq, i); |
| 1504 | Py_INCREF(element); |
| 1505 | self->extra->children[cur] = element; |
| 1506 | } |
| 1507 | |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 1508 | self->extra->length += (int)(newlen - slicelen); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1509 | |
Serhiy Storchaka | a0ae9ff | 2015-11-22 12:31:11 +0200 | [diff] [blame] | 1510 | Py_XDECREF(seq); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1511 | |
| 1512 | /* discard the recycle bin, and everything in it */ |
| 1513 | Py_XDECREF(recycle); |
| 1514 | |
| 1515 | return 0; |
| 1516 | } |
| 1517 | else { |
| 1518 | PyErr_SetString(PyExc_TypeError, |
| 1519 | "element indices must be integers"); |
| 1520 | return -1; |
| 1521 | } |
| 1522 | } |
| 1523 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1524 | static PyMethodDef element_methods[] = { |
| 1525 | |
| 1526 | {"clear", (PyCFunction) element_clear, METH_VARARGS}, |
| 1527 | |
| 1528 | {"get", (PyCFunction) element_get, METH_VARARGS}, |
| 1529 | {"set", (PyCFunction) element_set, METH_VARARGS}, |
| 1530 | |
| 1531 | {"find", (PyCFunction) element_find, METH_VARARGS}, |
| 1532 | {"findtext", (PyCFunction) element_findtext, METH_VARARGS}, |
| 1533 | {"findall", (PyCFunction) element_findall, METH_VARARGS}, |
| 1534 | |
| 1535 | {"append", (PyCFunction) element_append, METH_VARARGS}, |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1536 | {"extend", (PyCFunction) element_extend, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1537 | {"insert", (PyCFunction) element_insert, METH_VARARGS}, |
| 1538 | {"remove", (PyCFunction) element_remove, METH_VARARGS}, |
| 1539 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1540 | {"iter", (PyCFunction) element_iter, METH_VARARGS}, |
| 1541 | {"itertext", (PyCFunction) element_itertext, METH_VARARGS}, |
| 1542 | {"iterfind", (PyCFunction) element_iterfind, METH_VARARGS}, |
| 1543 | |
Serhiy Storchaka | 09b5247 | 2017-05-17 10:08:11 +0300 | [diff] [blame] | 1544 | {"getiterator", (PyCFunction) element_getiterator, METH_VARARGS}, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1545 | {"getchildren", (PyCFunction) element_getchildren, METH_VARARGS}, |
| 1546 | |
| 1547 | {"items", (PyCFunction) element_items, METH_VARARGS}, |
| 1548 | {"keys", (PyCFunction) element_keys, METH_VARARGS}, |
| 1549 | |
| 1550 | {"makeelement", (PyCFunction) element_makeelement, METH_VARARGS}, |
| 1551 | |
| 1552 | {"__copy__", (PyCFunction) element_copy, METH_VARARGS}, |
| 1553 | {"__deepcopy__", (PyCFunction) element_deepcopy, METH_VARARGS}, |
| 1554 | |
| 1555 | /* Some 2.3 and 2.4 versions do not handle the __copy__ method on |
| 1556 | C objects correctly, so we have to fake it using a __reduce__- |
| 1557 | based hack (see the element_reduce implementation above for |
| 1558 | details). */ |
| 1559 | |
| 1560 | /* The behaviour has been changed in 2.3.5 and 2.4.1, so we're |
| 1561 | using a runtime test to figure out if we need to fake things |
| 1562 | or now (see the init code below). The following entry is |
| 1563 | enabled only if the hack is needed. */ |
| 1564 | |
| 1565 | {"!__reduce__", (PyCFunction) element_reduce, METH_VARARGS}, |
| 1566 | |
| 1567 | {NULL, NULL} |
| 1568 | }; |
| 1569 | |
| 1570 | static PyObject* |
| 1571 | element_getattr(ElementObject* self, char* name) |
| 1572 | { |
| 1573 | PyObject* res; |
| 1574 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1575 | /* handle common attributes first */ |
| 1576 | if (strcmp(name, "tag") == 0) { |
| 1577 | res = self->tag; |
| 1578 | Py_INCREF(res); |
| 1579 | return res; |
| 1580 | } else if (strcmp(name, "text") == 0) { |
| 1581 | res = element_get_text(self); |
Xiang Zhang | 827c783 | 2017-03-22 12:25:51 +0800 | [diff] [blame] | 1582 | Py_XINCREF(res); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1583 | return res; |
| 1584 | } |
| 1585 | |
| 1586 | /* methods */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1587 | res = Py_FindMethod(element_methods, (PyObject*) self, name); |
| 1588 | if (res) |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1589 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1590 | |
| 1591 | PyErr_Clear(); |
| 1592 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1593 | /* less common attributes */ |
| 1594 | if (strcmp(name, "tail") == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1595 | res = element_get_tail(self); |
| 1596 | } else if (strcmp(name, "attrib") == 0) { |
| 1597 | if (!self->extra) |
| 1598 | element_new_extra(self, NULL); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1599 | res = element_get_attrib(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1600 | } else { |
| 1601 | PyErr_SetString(PyExc_AttributeError, name); |
| 1602 | return NULL; |
| 1603 | } |
| 1604 | |
| 1605 | if (!res) |
| 1606 | return NULL; |
| 1607 | |
| 1608 | Py_INCREF(res); |
| 1609 | return res; |
| 1610 | } |
| 1611 | |
| 1612 | static int |
| 1613 | element_setattr(ElementObject* self, const char* name, PyObject* value) |
| 1614 | { |
| 1615 | if (value == NULL) { |
| 1616 | PyErr_SetString( |
| 1617 | PyExc_AttributeError, |
| 1618 | "can't delete element attributes" |
| 1619 | ); |
| 1620 | return -1; |
| 1621 | } |
| 1622 | |
| 1623 | if (strcmp(name, "tag") == 0) { |
Serhiy Storchaka | 2e6c829 | 2015-12-27 15:41:58 +0200 | [diff] [blame] | 1624 | Py_INCREF(value); |
Serhiy Storchaka | 763a61c | 2016-04-10 18:05:12 +0300 | [diff] [blame] | 1625 | Py_SETREF(self->tag, value); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1626 | } else if (strcmp(name, "text") == 0) { |
| 1627 | Py_DECREF(JOIN_OBJ(self->text)); |
| 1628 | self->text = value; |
| 1629 | Py_INCREF(self->text); |
| 1630 | } else if (strcmp(name, "tail") == 0) { |
| 1631 | Py_DECREF(JOIN_OBJ(self->tail)); |
| 1632 | self->tail = value; |
| 1633 | Py_INCREF(self->tail); |
| 1634 | } else if (strcmp(name, "attrib") == 0) { |
| 1635 | if (!self->extra) |
| 1636 | element_new_extra(self, NULL); |
Serhiy Storchaka | 2e6c829 | 2015-12-27 15:41:58 +0200 | [diff] [blame] | 1637 | Py_INCREF(value); |
Serhiy Storchaka | 763a61c | 2016-04-10 18:05:12 +0300 | [diff] [blame] | 1638 | Py_SETREF(self->extra->attrib, value); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1639 | } else { |
| 1640 | PyErr_SetString(PyExc_AttributeError, name); |
| 1641 | return -1; |
| 1642 | } |
| 1643 | |
| 1644 | return 0; |
| 1645 | } |
| 1646 | |
| 1647 | static PySequenceMethods element_as_sequence = { |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1648 | (lenfunc) element_length, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1649 | 0, /* sq_concat */ |
| 1650 | 0, /* sq_repeat */ |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1651 | element_getitem, |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1652 | 0, |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 1653 | element_setitem, |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1654 | 0, |
| 1655 | }; |
| 1656 | |
| 1657 | static PyMappingMethods element_as_mapping = { |
| 1658 | (lenfunc) element_length, |
| 1659 | (binaryfunc) element_subscr, |
| 1660 | (objobjargproc) element_ass_subscr, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1661 | }; |
| 1662 | |
| 1663 | statichere PyTypeObject Element_Type = { |
| 1664 | PyObject_HEAD_INIT(NULL) |
| 1665 | 0, "Element", sizeof(ElementObject), 0, |
| 1666 | /* methods */ |
| 1667 | (destructor)element_dealloc, /* tp_dealloc */ |
| 1668 | 0, /* tp_print */ |
| 1669 | (getattrfunc)element_getattr, /* tp_getattr */ |
| 1670 | (setattrfunc)element_setattr, /* tp_setattr */ |
| 1671 | 0, /* tp_compare */ |
| 1672 | (reprfunc)element_repr, /* tp_repr */ |
| 1673 | 0, /* tp_as_number */ |
| 1674 | &element_as_sequence, /* tp_as_sequence */ |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1675 | &element_as_mapping, /* tp_as_mapping */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1676 | }; |
| 1677 | |
| 1678 | /* ==================================================================== */ |
| 1679 | /* the tree builder type */ |
| 1680 | |
| 1681 | typedef struct { |
| 1682 | PyObject_HEAD |
| 1683 | |
| 1684 | PyObject* root; /* root node (first created node) */ |
| 1685 | |
| 1686 | ElementObject* this; /* current node */ |
| 1687 | ElementObject* last; /* most recently created node */ |
| 1688 | |
| 1689 | PyObject* data; /* data collector (string or list), or NULL */ |
| 1690 | |
| 1691 | PyObject* stack; /* element stack */ |
Neal Norwitz | c707438 | 2006-06-12 02:06:17 +0000 | [diff] [blame] | 1692 | Py_ssize_t index; /* current stack size (0=empty) */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1693 | |
| 1694 | /* element tracing */ |
| 1695 | PyObject* events; /* list of events, or NULL if not collecting */ |
| 1696 | PyObject* start_event_obj; /* event objects (NULL to ignore) */ |
| 1697 | PyObject* end_event_obj; |
| 1698 | PyObject* start_ns_event_obj; |
| 1699 | PyObject* end_ns_event_obj; |
| 1700 | |
| 1701 | } TreeBuilderObject; |
| 1702 | |
| 1703 | staticforward PyTypeObject TreeBuilder_Type; |
| 1704 | |
Christian Heimes | e93237d | 2007-12-19 02:37:44 +0000 | [diff] [blame] | 1705 | #define TreeBuilder_CheckExact(op) (Py_TYPE(op) == &TreeBuilder_Type) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1706 | |
| 1707 | /* -------------------------------------------------------------------- */ |
| 1708 | /* constructor and destructor */ |
| 1709 | |
| 1710 | LOCAL(PyObject*) |
| 1711 | treebuilder_new(void) |
| 1712 | { |
| 1713 | TreeBuilderObject* self; |
| 1714 | |
| 1715 | self = PyObject_New(TreeBuilderObject, &TreeBuilder_Type); |
| 1716 | if (self == NULL) |
| 1717 | return NULL; |
| 1718 | |
| 1719 | self->root = NULL; |
| 1720 | |
| 1721 | Py_INCREF(Py_None); |
| 1722 | self->this = (ElementObject*) Py_None; |
| 1723 | |
| 1724 | Py_INCREF(Py_None); |
| 1725 | self->last = (ElementObject*) Py_None; |
| 1726 | |
| 1727 | self->data = NULL; |
| 1728 | |
| 1729 | self->stack = PyList_New(20); |
| 1730 | self->index = 0; |
| 1731 | |
| 1732 | self->events = NULL; |
| 1733 | self->start_event_obj = self->end_event_obj = NULL; |
| 1734 | self->start_ns_event_obj = self->end_ns_event_obj = NULL; |
| 1735 | |
| 1736 | ALLOC(sizeof(TreeBuilderObject), "create treebuilder"); |
| 1737 | |
| 1738 | return (PyObject*) self; |
| 1739 | } |
| 1740 | |
| 1741 | static PyObject* |
Fredrik Lundh | 81707f1 | 2006-06-03 21:56:05 +0000 | [diff] [blame] | 1742 | treebuilder(PyObject* self_, PyObject* args) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1743 | { |
| 1744 | if (!PyArg_ParseTuple(args, ":TreeBuilder")) |
| 1745 | return NULL; |
| 1746 | |
| 1747 | return treebuilder_new(); |
| 1748 | } |
| 1749 | |
| 1750 | static void |
| 1751 | treebuilder_dealloc(TreeBuilderObject* self) |
| 1752 | { |
| 1753 | Py_XDECREF(self->end_ns_event_obj); |
| 1754 | Py_XDECREF(self->start_ns_event_obj); |
| 1755 | Py_XDECREF(self->end_event_obj); |
| 1756 | Py_XDECREF(self->start_event_obj); |
| 1757 | Py_XDECREF(self->events); |
| 1758 | Py_DECREF(self->stack); |
| 1759 | Py_XDECREF(self->data); |
| 1760 | Py_DECREF(self->last); |
| 1761 | Py_DECREF(self->this); |
| 1762 | Py_XDECREF(self->root); |
| 1763 | |
| 1764 | RELEASE(sizeof(TreeBuilderObject), "destroy treebuilder"); |
| 1765 | |
| 1766 | PyObject_Del(self); |
| 1767 | } |
| 1768 | |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 1769 | /* -------------------------------------------------------------------- */ |
| 1770 | /* helpers for handling of arbitrary element-like objects */ |
| 1771 | |
| 1772 | static void |
| 1773 | treebuilder_set_element_text_or_tail(PyObject **data, PyObject **dest) |
| 1774 | { |
| 1775 | PyObject *tmp = JOIN_OBJ(*dest); |
| 1776 | *dest = JOIN_SET(*data, PyList_CheckExact(*data)); |
| 1777 | *data = NULL; |
| 1778 | Py_DECREF(tmp); |
| 1779 | } |
| 1780 | |
| 1781 | LOCAL(void) |
| 1782 | treebuilder_flush_data(TreeBuilderObject* self) |
| 1783 | { |
| 1784 | ElementObject *element = self->last; |
| 1785 | |
| 1786 | if (self->data) { |
| 1787 | if (self->this == element) { |
| 1788 | treebuilder_set_element_text_or_tail( |
| 1789 | &self->data, |
| 1790 | &element->text); |
| 1791 | } |
| 1792 | else { |
| 1793 | treebuilder_set_element_text_or_tail( |
| 1794 | &self->data, |
| 1795 | &element->tail); |
| 1796 | } |
| 1797 | } |
| 1798 | } |
| 1799 | |
Serhiy Storchaka | 45cf0b7 | 2015-12-06 23:51:53 +0200 | [diff] [blame] | 1800 | LOCAL(int) |
| 1801 | treebuilder_append_event(TreeBuilderObject *self, PyObject *action, |
| 1802 | PyObject *node) |
| 1803 | { |
| 1804 | if (action != NULL) { |
| 1805 | PyObject *res = PyTuple_Pack(2, action, node); |
| 1806 | if (res == NULL) |
| 1807 | return -1; |
| 1808 | if (PyList_Append(self->events, res) < 0) { |
| 1809 | Py_DECREF(res); |
| 1810 | return -1; |
| 1811 | } |
| 1812 | Py_DECREF(res); |
| 1813 | } |
| 1814 | return 0; |
| 1815 | } |
| 1816 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1817 | /* -------------------------------------------------------------------- */ |
| 1818 | /* handlers */ |
| 1819 | |
| 1820 | LOCAL(PyObject*) |
| 1821 | treebuilder_handle_xml(TreeBuilderObject* self, PyObject* encoding, |
| 1822 | PyObject* standalone) |
| 1823 | { |
| 1824 | Py_RETURN_NONE; |
| 1825 | } |
| 1826 | |
| 1827 | LOCAL(PyObject*) |
| 1828 | treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag, |
| 1829 | PyObject* attrib) |
| 1830 | { |
| 1831 | PyObject* node; |
| 1832 | PyObject* this; |
| 1833 | |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 1834 | treebuilder_flush_data(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1835 | |
| 1836 | node = element_new(tag, attrib); |
| 1837 | if (!node) |
| 1838 | return NULL; |
| 1839 | |
| 1840 | this = (PyObject*) self->this; |
| 1841 | |
| 1842 | if (this != Py_None) { |
| 1843 | if (element_add_subelement((ElementObject*) this, node) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1844 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1845 | } else { |
| 1846 | if (self->root) { |
| 1847 | PyErr_SetString( |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 1848 | elementtree_parseerror_obj, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1849 | "multiple elements on top level" |
| 1850 | ); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1851 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1852 | } |
| 1853 | Py_INCREF(node); |
| 1854 | self->root = node; |
| 1855 | } |
| 1856 | |
| 1857 | if (self->index < PyList_GET_SIZE(self->stack)) { |
| 1858 | if (PyList_SetItem(self->stack, self->index, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1859 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1860 | Py_INCREF(this); |
| 1861 | } else { |
| 1862 | if (PyList_Append(self->stack, this) < 0) |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1863 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1864 | } |
| 1865 | self->index++; |
| 1866 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1867 | Py_INCREF(node); |
Serhiy Storchaka | 763a61c | 2016-04-10 18:05:12 +0300 | [diff] [blame] | 1868 | Py_SETREF(self->this, (ElementObject*) node); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1869 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1870 | Py_INCREF(node); |
Serhiy Storchaka | 763a61c | 2016-04-10 18:05:12 +0300 | [diff] [blame] | 1871 | Py_SETREF(self->last, (ElementObject*) node); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1872 | |
Serhiy Storchaka | 45cf0b7 | 2015-12-06 23:51:53 +0200 | [diff] [blame] | 1873 | if (treebuilder_append_event(self, self->start_event_obj, node) < 0) |
| 1874 | goto error; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1875 | |
| 1876 | return node; |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 1877 | |
| 1878 | error: |
| 1879 | Py_DECREF(node); |
| 1880 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1881 | } |
| 1882 | |
| 1883 | LOCAL(PyObject*) |
| 1884 | treebuilder_handle_data(TreeBuilderObject* self, PyObject* data) |
| 1885 | { |
| 1886 | if (!self->data) { |
Fredrik Lundh | dc075b9 | 2006-08-16 16:47:07 +0000 | [diff] [blame] | 1887 | if (self->last == (ElementObject*) Py_None) { |
| 1888 | /* ignore calls to data before the first call to start */ |
| 1889 | Py_RETURN_NONE; |
| 1890 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1891 | /* store the first item as is */ |
| 1892 | Py_INCREF(data); self->data = data; |
| 1893 | } else { |
| 1894 | /* more than one item; use a list to collect items */ |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 1895 | if (PyString_CheckExact(self->data) && Py_REFCNT(self->data) == 1 && |
| 1896 | PyString_CheckExact(data) && PyString_GET_SIZE(data) == 1) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1897 | /* expat often generates single character data sections; handle |
| 1898 | the most common case by resizing the existing string... */ |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 1899 | Py_ssize_t size = PyString_GET_SIZE(self->data); |
| 1900 | if (_PyString_Resize(&self->data, size + 1) < 0) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1901 | return NULL; |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 1902 | PyString_AS_STRING(self->data)[size] = PyString_AS_STRING(data)[0]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1903 | } else if (PyList_CheckExact(self->data)) { |
| 1904 | if (PyList_Append(self->data, data) < 0) |
| 1905 | return NULL; |
| 1906 | } else { |
| 1907 | PyObject* list = PyList_New(2); |
| 1908 | if (!list) |
| 1909 | return NULL; |
| 1910 | PyList_SET_ITEM(list, 0, self->data); |
| 1911 | Py_INCREF(data); PyList_SET_ITEM(list, 1, data); |
| 1912 | self->data = list; |
| 1913 | } |
| 1914 | } |
| 1915 | |
| 1916 | Py_RETURN_NONE; |
| 1917 | } |
| 1918 | |
| 1919 | LOCAL(PyObject*) |
| 1920 | treebuilder_handle_end(TreeBuilderObject* self, PyObject* tag) |
| 1921 | { |
Serhiy Storchaka | 2e6c829 | 2015-12-27 15:41:58 +0200 | [diff] [blame] | 1922 | ElementObject *item; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1923 | |
Serhiy Storchaka | 9c2c42c | 2017-04-02 20:37:03 +0300 | [diff] [blame] | 1924 | treebuilder_flush_data(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1925 | |
| 1926 | if (self->index == 0) { |
| 1927 | PyErr_SetString( |
| 1928 | PyExc_IndexError, |
| 1929 | "pop from empty stack" |
| 1930 | ); |
| 1931 | return NULL; |
| 1932 | } |
| 1933 | |
Serhiy Storchaka | 2e6c829 | 2015-12-27 15:41:58 +0200 | [diff] [blame] | 1934 | item = self->last; |
| 1935 | self->last = self->this; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1936 | self->index--; |
Serhiy Storchaka | 2e6c829 | 2015-12-27 15:41:58 +0200 | [diff] [blame] | 1937 | self->this = (ElementObject *) PyList_GET_ITEM(self->stack, self->index); |
| 1938 | Py_INCREF(self->this); |
| 1939 | Py_DECREF(item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1940 | |
Serhiy Storchaka | 45cf0b7 | 2015-12-06 23:51:53 +0200 | [diff] [blame] | 1941 | if (treebuilder_append_event(self, self->end_event_obj, (PyObject*)self->last) < 0) |
| 1942 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1943 | |
| 1944 | Py_INCREF(self->last); |
| 1945 | return (PyObject*) self->last; |
| 1946 | } |
| 1947 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 1948 | /* -------------------------------------------------------------------- */ |
| 1949 | /* methods (in alphabetical order) */ |
| 1950 | |
| 1951 | static PyObject* |
| 1952 | treebuilder_data(TreeBuilderObject* self, PyObject* args) |
| 1953 | { |
| 1954 | PyObject* data; |
| 1955 | if (!PyArg_ParseTuple(args, "O:data", &data)) |
| 1956 | return NULL; |
| 1957 | |
| 1958 | return treebuilder_handle_data(self, data); |
| 1959 | } |
| 1960 | |
| 1961 | static PyObject* |
| 1962 | treebuilder_end(TreeBuilderObject* self, PyObject* args) |
| 1963 | { |
| 1964 | PyObject* tag; |
| 1965 | if (!PyArg_ParseTuple(args, "O:end", &tag)) |
| 1966 | return NULL; |
| 1967 | |
| 1968 | return treebuilder_handle_end(self, tag); |
| 1969 | } |
| 1970 | |
| 1971 | LOCAL(PyObject*) |
| 1972 | treebuilder_done(TreeBuilderObject* self) |
| 1973 | { |
| 1974 | PyObject* res; |
| 1975 | |
| 1976 | /* FIXME: check stack size? */ |
| 1977 | |
| 1978 | if (self->root) |
| 1979 | res = self->root; |
| 1980 | else |
| 1981 | res = Py_None; |
| 1982 | |
| 1983 | Py_INCREF(res); |
| 1984 | return res; |
| 1985 | } |
| 1986 | |
| 1987 | static PyObject* |
| 1988 | treebuilder_close(TreeBuilderObject* self, PyObject* args) |
| 1989 | { |
| 1990 | if (!PyArg_ParseTuple(args, ":close")) |
| 1991 | return NULL; |
| 1992 | |
| 1993 | return treebuilder_done(self); |
| 1994 | } |
| 1995 | |
| 1996 | static PyObject* |
| 1997 | treebuilder_start(TreeBuilderObject* self, PyObject* args) |
| 1998 | { |
| 1999 | PyObject* tag; |
| 2000 | PyObject* attrib = Py_None; |
| 2001 | if (!PyArg_ParseTuple(args, "O|O:start", &tag, &attrib)) |
| 2002 | return NULL; |
| 2003 | |
| 2004 | return treebuilder_handle_start(self, tag, attrib); |
| 2005 | } |
| 2006 | |
| 2007 | static PyObject* |
| 2008 | treebuilder_xml(TreeBuilderObject* self, PyObject* args) |
| 2009 | { |
| 2010 | PyObject* encoding; |
| 2011 | PyObject* standalone; |
| 2012 | if (!PyArg_ParseTuple(args, "OO:xml", &encoding, &standalone)) |
| 2013 | return NULL; |
| 2014 | |
| 2015 | return treebuilder_handle_xml(self, encoding, standalone); |
| 2016 | } |
| 2017 | |
| 2018 | static PyMethodDef treebuilder_methods[] = { |
| 2019 | {"data", (PyCFunction) treebuilder_data, METH_VARARGS}, |
| 2020 | {"start", (PyCFunction) treebuilder_start, METH_VARARGS}, |
| 2021 | {"end", (PyCFunction) treebuilder_end, METH_VARARGS}, |
| 2022 | {"xml", (PyCFunction) treebuilder_xml, METH_VARARGS}, |
| 2023 | {"close", (PyCFunction) treebuilder_close, METH_VARARGS}, |
| 2024 | {NULL, NULL} |
| 2025 | }; |
| 2026 | |
| 2027 | static PyObject* |
| 2028 | treebuilder_getattr(TreeBuilderObject* self, char* name) |
| 2029 | { |
| 2030 | return Py_FindMethod(treebuilder_methods, (PyObject*) self, name); |
| 2031 | } |
| 2032 | |
| 2033 | statichere PyTypeObject TreeBuilder_Type = { |
| 2034 | PyObject_HEAD_INIT(NULL) |
| 2035 | 0, "TreeBuilder", sizeof(TreeBuilderObject), 0, |
| 2036 | /* methods */ |
| 2037 | (destructor)treebuilder_dealloc, /* tp_dealloc */ |
| 2038 | 0, /* tp_print */ |
| 2039 | (getattrfunc)treebuilder_getattr, /* tp_getattr */ |
| 2040 | }; |
| 2041 | |
| 2042 | /* ==================================================================== */ |
| 2043 | /* the expat interface */ |
| 2044 | |
| 2045 | #if defined(USE_EXPAT) |
| 2046 | |
| 2047 | #include "expat.h" |
| 2048 | |
| 2049 | #if defined(USE_PYEXPAT_CAPI) |
| 2050 | #include "pyexpat.h" |
| 2051 | static struct PyExpat_CAPI* expat_capi; |
| 2052 | #define EXPAT(func) (expat_capi->func) |
| 2053 | #else |
| 2054 | #define EXPAT(func) (XML_##func) |
| 2055 | #endif |
| 2056 | |
| 2057 | typedef struct { |
| 2058 | PyObject_HEAD |
| 2059 | |
| 2060 | XML_Parser parser; |
| 2061 | |
| 2062 | PyObject* target; |
| 2063 | PyObject* entity; |
| 2064 | |
| 2065 | PyObject* names; |
| 2066 | |
| 2067 | PyObject* handle_xml; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2068 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2069 | PyObject* handle_start; |
| 2070 | PyObject* handle_data; |
| 2071 | PyObject* handle_end; |
| 2072 | |
| 2073 | PyObject* handle_comment; |
| 2074 | PyObject* handle_pi; |
| 2075 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2076 | PyObject* handle_close; |
| 2077 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2078 | } XMLParserObject; |
| 2079 | |
| 2080 | staticforward PyTypeObject XMLParser_Type; |
| 2081 | |
| 2082 | /* helpers */ |
| 2083 | |
| 2084 | #if defined(Py_USING_UNICODE) |
| 2085 | LOCAL(int) |
| 2086 | checkstring(const char* string, int size) |
| 2087 | { |
| 2088 | int i; |
| 2089 | |
| 2090 | /* check if an 8-bit string contains UTF-8 characters */ |
| 2091 | for (i = 0; i < size; i++) |
| 2092 | if (string[i] & 0x80) |
| 2093 | return 1; |
| 2094 | |
| 2095 | return 0; |
| 2096 | } |
| 2097 | #endif |
| 2098 | |
| 2099 | LOCAL(PyObject*) |
| 2100 | makestring(const char* string, int size) |
| 2101 | { |
| 2102 | /* convert a UTF-8 string to either a 7-bit ascii string or a |
| 2103 | Unicode string */ |
| 2104 | |
| 2105 | #if defined(Py_USING_UNICODE) |
| 2106 | if (checkstring(string, size)) |
| 2107 | return PyUnicode_DecodeUTF8(string, size, "strict"); |
| 2108 | #endif |
| 2109 | |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2110 | return PyString_FromStringAndSize(string, size); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2111 | } |
| 2112 | |
| 2113 | LOCAL(PyObject*) |
| 2114 | makeuniversal(XMLParserObject* self, const char* string) |
| 2115 | { |
| 2116 | /* convert a UTF-8 tag/attribute name from the expat parser |
| 2117 | to a universal name string */ |
| 2118 | |
| 2119 | int size = strlen(string); |
| 2120 | PyObject* key; |
| 2121 | PyObject* value; |
| 2122 | |
| 2123 | /* look the 'raw' name up in the names dictionary */ |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2124 | key = PyString_FromStringAndSize(string, size); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2125 | if (!key) |
| 2126 | return NULL; |
| 2127 | |
| 2128 | value = PyDict_GetItem(self->names, key); |
| 2129 | |
| 2130 | if (value) { |
| 2131 | Py_INCREF(value); |
| 2132 | } else { |
| 2133 | /* new name. convert to universal name, and decode as |
| 2134 | necessary */ |
| 2135 | |
| 2136 | PyObject* tag; |
| 2137 | char* p; |
| 2138 | int i; |
| 2139 | |
| 2140 | /* look for namespace separator */ |
| 2141 | for (i = 0; i < size; i++) |
| 2142 | if (string[i] == '}') |
| 2143 | break; |
| 2144 | if (i != size) { |
| 2145 | /* convert to universal name */ |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2146 | tag = PyString_FromStringAndSize(NULL, size+1); |
| 2147 | p = PyString_AS_STRING(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2148 | p[0] = '{'; |
| 2149 | memcpy(p+1, string, size); |
| 2150 | size++; |
| 2151 | } else { |
| 2152 | /* plain name; use key as tag */ |
| 2153 | Py_INCREF(key); |
| 2154 | tag = key; |
| 2155 | } |
| 2156 | |
| 2157 | /* decode universal name */ |
| 2158 | #if defined(Py_USING_UNICODE) |
| 2159 | /* inline makestring, to avoid duplicating the source string if |
Martin Panter | 6a8163a | 2016-04-15 02:14:19 +0000 | [diff] [blame] | 2160 | it's not a utf-8 string */ |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2161 | p = PyString_AS_STRING(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2162 | if (checkstring(p, size)) { |
| 2163 | value = PyUnicode_DecodeUTF8(p, size, "strict"); |
| 2164 | Py_DECREF(tag); |
| 2165 | if (!value) { |
| 2166 | Py_DECREF(key); |
| 2167 | return NULL; |
| 2168 | } |
| 2169 | } else |
| 2170 | #endif |
| 2171 | value = tag; /* use tag as is */ |
| 2172 | |
| 2173 | /* add to names dictionary */ |
| 2174 | if (PyDict_SetItem(self->names, key, value) < 0) { |
| 2175 | Py_DECREF(key); |
| 2176 | Py_DECREF(value); |
| 2177 | return NULL; |
| 2178 | } |
| 2179 | } |
| 2180 | |
| 2181 | Py_DECREF(key); |
| 2182 | return value; |
| 2183 | } |
| 2184 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2185 | static void |
| 2186 | expat_set_error(const char* message, int line, int column) |
| 2187 | { |
| 2188 | PyObject *error; |
| 2189 | PyObject *position; |
| 2190 | char buffer[256]; |
| 2191 | |
| 2192 | sprintf(buffer, "%s: line %d, column %d", message, line, column); |
| 2193 | |
| 2194 | error = PyObject_CallFunction(elementtree_parseerror_obj, "s", buffer); |
| 2195 | if (!error) |
| 2196 | return; |
| 2197 | |
| 2198 | /* add position attribute */ |
| 2199 | position = Py_BuildValue("(ii)", line, column); |
| 2200 | if (!position) { |
| 2201 | Py_DECREF(error); |
| 2202 | return; |
| 2203 | } |
| 2204 | if (PyObject_SetAttrString(error, "position", position) == -1) { |
| 2205 | Py_DECREF(error); |
| 2206 | Py_DECREF(position); |
| 2207 | return; |
| 2208 | } |
| 2209 | Py_DECREF(position); |
| 2210 | |
| 2211 | PyErr_SetObject(elementtree_parseerror_obj, error); |
| 2212 | Py_DECREF(error); |
| 2213 | } |
| 2214 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2215 | /* -------------------------------------------------------------------- */ |
| 2216 | /* handlers */ |
| 2217 | |
| 2218 | static void |
| 2219 | expat_default_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2220 | int data_len) |
| 2221 | { |
| 2222 | PyObject* key; |
| 2223 | PyObject* value; |
| 2224 | PyObject* res; |
| 2225 | |
| 2226 | if (data_len < 2 || data_in[0] != '&') |
| 2227 | return; |
| 2228 | |
| 2229 | key = makestring(data_in + 1, data_len - 2); |
| 2230 | if (!key) |
| 2231 | return; |
| 2232 | |
| 2233 | value = PyDict_GetItem(self->entity, key); |
| 2234 | |
| 2235 | if (value) { |
| 2236 | if (TreeBuilder_CheckExact(self->target)) |
| 2237 | res = treebuilder_handle_data( |
| 2238 | (TreeBuilderObject*) self->target, value |
| 2239 | ); |
| 2240 | else if (self->handle_data) |
| 2241 | res = PyObject_CallFunction(self->handle_data, "O", value); |
| 2242 | else |
| 2243 | res = NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2244 | Py_XDECREF(res); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2245 | } else if (!PyErr_Occurred()) { |
| 2246 | /* Report the first error, not the last */ |
| 2247 | char message[128]; |
| 2248 | sprintf(message, "undefined entity &%.100s;", PyString_AS_STRING(key)); |
| 2249 | expat_set_error( |
| 2250 | message, |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2251 | EXPAT(GetErrorLineNumber)(self->parser), |
| 2252 | EXPAT(GetErrorColumnNumber)(self->parser) |
| 2253 | ); |
| 2254 | } |
| 2255 | |
| 2256 | Py_DECREF(key); |
| 2257 | } |
| 2258 | |
| 2259 | static void |
| 2260 | expat_start_handler(XMLParserObject* self, const XML_Char* tag_in, |
| 2261 | const XML_Char **attrib_in) |
| 2262 | { |
| 2263 | PyObject* res; |
| 2264 | PyObject* tag; |
| 2265 | PyObject* attrib; |
| 2266 | int ok; |
| 2267 | |
| 2268 | /* tag name */ |
| 2269 | tag = makeuniversal(self, tag_in); |
| 2270 | if (!tag) |
| 2271 | return; /* parser will look for errors */ |
| 2272 | |
| 2273 | /* attributes */ |
| 2274 | if (attrib_in[0]) { |
| 2275 | attrib = PyDict_New(); |
Serhiy Storchaka | 33ea297 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2276 | if (!attrib) { |
| 2277 | Py_DECREF(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2278 | return; |
Serhiy Storchaka | 33ea297 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2279 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2280 | while (attrib_in[0] && attrib_in[1]) { |
| 2281 | PyObject* key = makeuniversal(self, attrib_in[0]); |
| 2282 | PyObject* value = makestring(attrib_in[1], strlen(attrib_in[1])); |
| 2283 | if (!key || !value) { |
| 2284 | Py_XDECREF(value); |
| 2285 | Py_XDECREF(key); |
| 2286 | Py_DECREF(attrib); |
Serhiy Storchaka | 33ea297 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2287 | Py_DECREF(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2288 | return; |
| 2289 | } |
| 2290 | ok = PyDict_SetItem(attrib, key, value); |
| 2291 | Py_DECREF(value); |
| 2292 | Py_DECREF(key); |
| 2293 | if (ok < 0) { |
| 2294 | Py_DECREF(attrib); |
Serhiy Storchaka | 33ea297 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2295 | Py_DECREF(tag); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2296 | return; |
| 2297 | } |
| 2298 | attrib_in += 2; |
| 2299 | } |
| 2300 | } else { |
| 2301 | Py_INCREF(Py_None); |
| 2302 | attrib = Py_None; |
| 2303 | } |
| 2304 | |
| 2305 | if (TreeBuilder_CheckExact(self->target)) |
| 2306 | /* shortcut */ |
| 2307 | res = treebuilder_handle_start((TreeBuilderObject*) self->target, |
| 2308 | tag, attrib); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2309 | else if (self->handle_start) { |
| 2310 | if (attrib == Py_None) { |
| 2311 | Py_DECREF(attrib); |
| 2312 | attrib = PyDict_New(); |
Serhiy Storchaka | 33ea297 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2313 | if (!attrib) { |
| 2314 | Py_DECREF(tag); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2315 | return; |
Serhiy Storchaka | 33ea297 | 2015-12-09 19:44:30 +0200 | [diff] [blame] | 2316 | } |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2317 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2318 | res = PyObject_CallFunction(self->handle_start, "OO", tag, attrib); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2319 | } else |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2320 | res = NULL; |
| 2321 | |
| 2322 | Py_DECREF(tag); |
| 2323 | Py_DECREF(attrib); |
| 2324 | |
| 2325 | Py_XDECREF(res); |
| 2326 | } |
| 2327 | |
| 2328 | static void |
| 2329 | expat_data_handler(XMLParserObject* self, const XML_Char* data_in, |
| 2330 | int data_len) |
| 2331 | { |
| 2332 | PyObject* data; |
| 2333 | PyObject* res; |
| 2334 | |
| 2335 | data = makestring(data_in, data_len); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2336 | if (!data) |
| 2337 | return; /* parser will look for errors */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2338 | |
| 2339 | if (TreeBuilder_CheckExact(self->target)) |
| 2340 | /* shortcut */ |
| 2341 | res = treebuilder_handle_data((TreeBuilderObject*) self->target, data); |
| 2342 | else if (self->handle_data) |
| 2343 | res = PyObject_CallFunction(self->handle_data, "O", data); |
| 2344 | else |
| 2345 | res = NULL; |
| 2346 | |
| 2347 | Py_DECREF(data); |
| 2348 | |
| 2349 | Py_XDECREF(res); |
| 2350 | } |
| 2351 | |
| 2352 | static void |
| 2353 | expat_end_handler(XMLParserObject* self, const XML_Char* tag_in) |
| 2354 | { |
| 2355 | PyObject* tag; |
| 2356 | PyObject* res = NULL; |
| 2357 | |
| 2358 | if (TreeBuilder_CheckExact(self->target)) |
| 2359 | /* shortcut */ |
| 2360 | /* the standard tree builder doesn't look at the end tag */ |
| 2361 | res = treebuilder_handle_end( |
| 2362 | (TreeBuilderObject*) self->target, Py_None |
| 2363 | ); |
| 2364 | else if (self->handle_end) { |
| 2365 | tag = makeuniversal(self, tag_in); |
| 2366 | if (tag) { |
| 2367 | res = PyObject_CallFunction(self->handle_end, "O", tag); |
| 2368 | Py_DECREF(tag); |
| 2369 | } |
| 2370 | } |
| 2371 | |
| 2372 | Py_XDECREF(res); |
| 2373 | } |
| 2374 | |
| 2375 | static void |
| 2376 | expat_start_ns_handler(XMLParserObject* self, const XML_Char* prefix, |
| 2377 | const XML_Char *uri) |
| 2378 | { |
Serhiy Storchaka | 45cf0b7 | 2015-12-06 23:51:53 +0200 | [diff] [blame] | 2379 | TreeBuilderObject *target = (TreeBuilderObject*) self->target; |
| 2380 | PyObject *parcel; |
| 2381 | PyObject *sprefix = NULL; |
| 2382 | PyObject *suri = NULL; |
| 2383 | |
| 2384 | if (PyErr_Occurred()) |
| 2385 | return; |
| 2386 | |
| 2387 | if (!target->events || !target->start_ns_event_obj) |
| 2388 | return; |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2389 | |
Eli Bendersky | f933e08 | 2013-11-28 06:25:45 -0800 | [diff] [blame] | 2390 | if (uri) |
Eli Bendersky | 71142c4 | 2013-11-28 06:37:25 -0800 | [diff] [blame] | 2391 | suri = makestring(uri, strlen(uri)); |
Eli Bendersky | f933e08 | 2013-11-28 06:25:45 -0800 | [diff] [blame] | 2392 | else |
Eli Bendersky | 71142c4 | 2013-11-28 06:37:25 -0800 | [diff] [blame] | 2393 | suri = PyString_FromStringAndSize("", 0); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2394 | if (!suri) |
| 2395 | return; |
| 2396 | |
| 2397 | if (prefix) |
| 2398 | sprefix = makestring(prefix, strlen(prefix)); |
| 2399 | else |
| 2400 | sprefix = PyString_FromStringAndSize("", 0); |
| 2401 | if (!sprefix) { |
| 2402 | Py_DECREF(suri); |
| 2403 | return; |
| 2404 | } |
| 2405 | |
Serhiy Storchaka | 45cf0b7 | 2015-12-06 23:51:53 +0200 | [diff] [blame] | 2406 | parcel = PyTuple_Pack(2, sprefix, suri); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2407 | Py_DECREF(sprefix); |
| 2408 | Py_DECREF(suri); |
Serhiy Storchaka | 45cf0b7 | 2015-12-06 23:51:53 +0200 | [diff] [blame] | 2409 | if (!parcel) |
| 2410 | return; |
| 2411 | treebuilder_append_event(target, target->start_ns_event_obj, parcel); |
| 2412 | Py_DECREF(parcel); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2413 | } |
| 2414 | |
| 2415 | static void |
| 2416 | expat_end_ns_handler(XMLParserObject* self, const XML_Char* prefix_in) |
| 2417 | { |
Serhiy Storchaka | 45cf0b7 | 2015-12-06 23:51:53 +0200 | [diff] [blame] | 2418 | TreeBuilderObject *target = (TreeBuilderObject*) self->target; |
| 2419 | |
| 2420 | if (PyErr_Occurred()) |
| 2421 | return; |
| 2422 | |
| 2423 | if (!target->events) |
| 2424 | return; |
| 2425 | |
| 2426 | treebuilder_append_event(target, target->end_ns_event_obj, Py_None); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2427 | } |
| 2428 | |
| 2429 | static void |
| 2430 | expat_comment_handler(XMLParserObject* self, const XML_Char* comment_in) |
| 2431 | { |
| 2432 | PyObject* comment; |
| 2433 | PyObject* res; |
| 2434 | |
| 2435 | if (self->handle_comment) { |
| 2436 | comment = makestring(comment_in, strlen(comment_in)); |
| 2437 | if (comment) { |
| 2438 | res = PyObject_CallFunction(self->handle_comment, "O", comment); |
| 2439 | Py_XDECREF(res); |
| 2440 | Py_DECREF(comment); |
| 2441 | } |
| 2442 | } |
| 2443 | } |
| 2444 | |
| 2445 | static void |
| 2446 | expat_pi_handler(XMLParserObject* self, const XML_Char* target_in, |
| 2447 | const XML_Char* data_in) |
| 2448 | { |
| 2449 | PyObject* target; |
| 2450 | PyObject* data; |
| 2451 | PyObject* res; |
| 2452 | |
| 2453 | if (self->handle_pi) { |
| 2454 | target = makestring(target_in, strlen(target_in)); |
| 2455 | data = makestring(data_in, strlen(data_in)); |
| 2456 | if (target && data) { |
| 2457 | res = PyObject_CallFunction(self->handle_pi, "OO", target, data); |
| 2458 | Py_XDECREF(res); |
| 2459 | Py_DECREF(data); |
| 2460 | Py_DECREF(target); |
| 2461 | } else { |
| 2462 | Py_XDECREF(data); |
| 2463 | Py_XDECREF(target); |
| 2464 | } |
| 2465 | } |
| 2466 | } |
| 2467 | |
| 2468 | #if defined(Py_USING_UNICODE) |
| 2469 | static int |
| 2470 | expat_unknown_encoding_handler(XMLParserObject *self, const XML_Char *name, |
| 2471 | XML_Encoding *info) |
| 2472 | { |
| 2473 | PyObject* u; |
| 2474 | Py_UNICODE* p; |
| 2475 | unsigned char s[256]; |
| 2476 | int i; |
| 2477 | |
| 2478 | memset(info, 0, sizeof(XML_Encoding)); |
| 2479 | |
| 2480 | for (i = 0; i < 256; i++) |
| 2481 | s[i] = i; |
| 2482 | |
Fredrik Lundh | c338999 | 2005-12-25 11:40:19 +0000 | [diff] [blame] | 2483 | u = PyUnicode_Decode((char*) s, 256, name, "replace"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2484 | if (!u) |
| 2485 | return XML_STATUS_ERROR; |
| 2486 | |
| 2487 | if (PyUnicode_GET_SIZE(u) != 256) { |
| 2488 | Py_DECREF(u); |
Eli Bendersky | b671701 | 2013-08-04 06:09:49 -0700 | [diff] [blame] | 2489 | PyErr_SetString(PyExc_ValueError, |
| 2490 | "multi-byte encodings are not supported"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2491 | return XML_STATUS_ERROR; |
| 2492 | } |
| 2493 | |
| 2494 | p = PyUnicode_AS_UNICODE(u); |
| 2495 | |
| 2496 | for (i = 0; i < 256; i++) { |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2497 | if (p[i] != Py_UNICODE_REPLACEMENT_CHARACTER) |
| 2498 | info->map[i] = p[i]; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2499 | else |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2500 | info->map[i] = -1; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2501 | } |
| 2502 | |
| 2503 | Py_DECREF(u); |
| 2504 | |
| 2505 | return XML_STATUS_OK; |
| 2506 | } |
| 2507 | #endif |
| 2508 | |
| 2509 | /* -------------------------------------------------------------------- */ |
| 2510 | /* constructor and destructor */ |
| 2511 | |
| 2512 | static PyObject* |
Fredrik Lundh | 81707f1 | 2006-06-03 21:56:05 +0000 | [diff] [blame] | 2513 | xmlparser(PyObject* self_, PyObject* args, PyObject* kw) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2514 | { |
| 2515 | XMLParserObject* self; |
| 2516 | /* FIXME: does this need to be static? */ |
| 2517 | static XML_Memory_Handling_Suite memory_handler; |
| 2518 | |
| 2519 | PyObject* target = NULL; |
| 2520 | char* encoding = NULL; |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 2521 | static char* kwlist[] = { "target", "encoding", NULL }; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2522 | if (!PyArg_ParseTupleAndKeywords(args, kw, "|Oz:XMLParser", kwlist, |
| 2523 | &target, &encoding)) |
| 2524 | return NULL; |
| 2525 | |
| 2526 | #if defined(USE_PYEXPAT_CAPI) |
| 2527 | if (!expat_capi) { |
| 2528 | PyErr_SetString( |
| 2529 | PyExc_RuntimeError, "cannot load dispatch table from pyexpat" |
| 2530 | ); |
| 2531 | return NULL; |
| 2532 | } |
| 2533 | #endif |
| 2534 | |
| 2535 | self = PyObject_New(XMLParserObject, &XMLParser_Type); |
| 2536 | if (self == NULL) |
| 2537 | return NULL; |
| 2538 | |
| 2539 | self->entity = PyDict_New(); |
| 2540 | if (!self->entity) { |
| 2541 | PyObject_Del(self); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2542 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2543 | } |
| 2544 | |
| 2545 | self->names = PyDict_New(); |
| 2546 | if (!self->names) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2547 | PyObject_Del(self->entity); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2548 | PyObject_Del(self); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2549 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2550 | } |
| 2551 | |
| 2552 | memory_handler.malloc_fcn = PyObject_Malloc; |
| 2553 | memory_handler.realloc_fcn = PyObject_Realloc; |
| 2554 | memory_handler.free_fcn = PyObject_Free; |
| 2555 | |
| 2556 | self->parser = EXPAT(ParserCreate_MM)(encoding, &memory_handler, "}"); |
| 2557 | if (!self->parser) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2558 | PyObject_Del(self->names); |
| 2559 | PyObject_Del(self->entity); |
| 2560 | PyObject_Del(self); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2561 | PyErr_NoMemory(); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2562 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2563 | } |
| 2564 | |
| 2565 | /* setup target handlers */ |
| 2566 | if (!target) { |
| 2567 | target = treebuilder_new(); |
| 2568 | if (!target) { |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2569 | EXPAT(ParserFree)(self->parser); |
| 2570 | PyObject_Del(self->names); |
| 2571 | PyObject_Del(self->entity); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2572 | PyObject_Del(self); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2573 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2574 | } |
| 2575 | } else |
| 2576 | Py_INCREF(target); |
| 2577 | self->target = target; |
| 2578 | |
| 2579 | self->handle_xml = PyObject_GetAttrString(target, "xml"); |
| 2580 | self->handle_start = PyObject_GetAttrString(target, "start"); |
| 2581 | self->handle_data = PyObject_GetAttrString(target, "data"); |
| 2582 | self->handle_end = PyObject_GetAttrString(target, "end"); |
| 2583 | self->handle_comment = PyObject_GetAttrString(target, "comment"); |
| 2584 | self->handle_pi = PyObject_GetAttrString(target, "pi"); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2585 | self->handle_close = PyObject_GetAttrString(target, "close"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2586 | |
| 2587 | PyErr_Clear(); |
| 2588 | |
| 2589 | /* configure parser */ |
| 2590 | EXPAT(SetUserData)(self->parser, self); |
| 2591 | EXPAT(SetElementHandler)( |
| 2592 | self->parser, |
| 2593 | (XML_StartElementHandler) expat_start_handler, |
| 2594 | (XML_EndElementHandler) expat_end_handler |
| 2595 | ); |
| 2596 | EXPAT(SetDefaultHandlerExpand)( |
| 2597 | self->parser, |
| 2598 | (XML_DefaultHandler) expat_default_handler |
| 2599 | ); |
| 2600 | EXPAT(SetCharacterDataHandler)( |
| 2601 | self->parser, |
| 2602 | (XML_CharacterDataHandler) expat_data_handler |
| 2603 | ); |
| 2604 | if (self->handle_comment) |
| 2605 | EXPAT(SetCommentHandler)( |
| 2606 | self->parser, |
| 2607 | (XML_CommentHandler) expat_comment_handler |
| 2608 | ); |
| 2609 | if (self->handle_pi) |
| 2610 | EXPAT(SetProcessingInstructionHandler)( |
| 2611 | self->parser, |
| 2612 | (XML_ProcessingInstructionHandler) expat_pi_handler |
| 2613 | ); |
| 2614 | #if defined(Py_USING_UNICODE) |
| 2615 | EXPAT(SetUnknownEncodingHandler)( |
| 2616 | self->parser, |
| 2617 | (XML_UnknownEncodingHandler) expat_unknown_encoding_handler, NULL |
| 2618 | ); |
| 2619 | #endif |
| 2620 | |
| 2621 | ALLOC(sizeof(XMLParserObject), "create expatparser"); |
| 2622 | |
| 2623 | return (PyObject*) self; |
| 2624 | } |
| 2625 | |
| 2626 | static void |
| 2627 | xmlparser_dealloc(XMLParserObject* self) |
| 2628 | { |
| 2629 | EXPAT(ParserFree)(self->parser); |
| 2630 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2631 | Py_XDECREF(self->handle_close); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2632 | Py_XDECREF(self->handle_pi); |
| 2633 | Py_XDECREF(self->handle_comment); |
| 2634 | Py_XDECREF(self->handle_end); |
| 2635 | Py_XDECREF(self->handle_data); |
| 2636 | Py_XDECREF(self->handle_start); |
| 2637 | Py_XDECREF(self->handle_xml); |
| 2638 | |
| 2639 | Py_DECREF(self->target); |
| 2640 | Py_DECREF(self->entity); |
| 2641 | Py_DECREF(self->names); |
| 2642 | |
| 2643 | RELEASE(sizeof(XMLParserObject), "destroy expatparser"); |
| 2644 | |
| 2645 | PyObject_Del(self); |
| 2646 | } |
| 2647 | |
| 2648 | /* -------------------------------------------------------------------- */ |
| 2649 | /* methods (in alphabetical order) */ |
| 2650 | |
| 2651 | LOCAL(PyObject*) |
| 2652 | expat_parse(XMLParserObject* self, char* data, int data_len, int final) |
| 2653 | { |
| 2654 | int ok; |
| 2655 | |
| 2656 | ok = EXPAT(Parse)(self->parser, data, data_len, final); |
| 2657 | |
| 2658 | if (PyErr_Occurred()) |
| 2659 | return NULL; |
| 2660 | |
| 2661 | if (!ok) { |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2662 | expat_set_error( |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2663 | EXPAT(ErrorString)(EXPAT(GetErrorCode)(self->parser)), |
| 2664 | EXPAT(GetErrorLineNumber)(self->parser), |
| 2665 | EXPAT(GetErrorColumnNumber)(self->parser) |
| 2666 | ); |
| 2667 | return NULL; |
| 2668 | } |
| 2669 | |
| 2670 | Py_RETURN_NONE; |
| 2671 | } |
| 2672 | |
| 2673 | static PyObject* |
| 2674 | xmlparser_close(XMLParserObject* self, PyObject* args) |
| 2675 | { |
| 2676 | /* end feeding data to parser */ |
| 2677 | |
| 2678 | PyObject* res; |
| 2679 | if (!PyArg_ParseTuple(args, ":close")) |
| 2680 | return NULL; |
| 2681 | |
| 2682 | res = expat_parse(self, "", 0, 1); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2683 | if (!res) |
| 2684 | return NULL; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2685 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2686 | if (TreeBuilder_CheckExact(self->target)) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2687 | Py_DECREF(res); |
| 2688 | return treebuilder_done((TreeBuilderObject*) self->target); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2689 | } if (self->handle_close) { |
| 2690 | Py_DECREF(res); |
| 2691 | return PyObject_CallFunction(self->handle_close, ""); |
| 2692 | } else |
| 2693 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2694 | } |
| 2695 | |
| 2696 | static PyObject* |
| 2697 | xmlparser_feed(XMLParserObject* self, PyObject* args) |
| 2698 | { |
| 2699 | /* feed data to parser */ |
| 2700 | |
| 2701 | char* data; |
| 2702 | int data_len; |
| 2703 | if (!PyArg_ParseTuple(args, "s#:feed", &data, &data_len)) |
| 2704 | return NULL; |
| 2705 | |
| 2706 | return expat_parse(self, data, data_len, 0); |
| 2707 | } |
| 2708 | |
| 2709 | static PyObject* |
| 2710 | xmlparser_parse(XMLParserObject* self, PyObject* args) |
| 2711 | { |
| 2712 | /* (internal) parse until end of input stream */ |
| 2713 | |
| 2714 | PyObject* reader; |
| 2715 | PyObject* buffer; |
| 2716 | PyObject* res; |
| 2717 | |
| 2718 | PyObject* fileobj; |
| 2719 | if (!PyArg_ParseTuple(args, "O:_parse", &fileobj)) |
| 2720 | return NULL; |
| 2721 | |
| 2722 | reader = PyObject_GetAttrString(fileobj, "read"); |
| 2723 | if (!reader) |
| 2724 | return NULL; |
| 2725 | |
| 2726 | /* read from open file object */ |
| 2727 | for (;;) { |
| 2728 | |
| 2729 | buffer = PyObject_CallFunction(reader, "i", 64*1024); |
| 2730 | |
| 2731 | if (!buffer) { |
| 2732 | /* read failed (e.g. due to KeyboardInterrupt) */ |
| 2733 | Py_DECREF(reader); |
| 2734 | return NULL; |
| 2735 | } |
| 2736 | |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2737 | if (!PyString_CheckExact(buffer) || PyString_GET_SIZE(buffer) == 0) { |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2738 | Py_DECREF(buffer); |
| 2739 | break; |
| 2740 | } |
| 2741 | |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 2742 | if (PyString_GET_SIZE(buffer) > INT_MAX) { |
| 2743 | Py_DECREF(buffer); |
| 2744 | Py_DECREF(reader); |
| 2745 | PyErr_SetString(PyExc_OverflowError, "size does not fit in an int"); |
| 2746 | return NULL; |
| 2747 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2748 | res = expat_parse( |
Serhiy Storchaka | c4c64be | 2015-11-25 20:12:58 +0200 | [diff] [blame] | 2749 | self, PyString_AS_STRING(buffer), (int)PyString_GET_SIZE(buffer), 0 |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2750 | ); |
| 2751 | |
| 2752 | Py_DECREF(buffer); |
| 2753 | |
| 2754 | if (!res) { |
| 2755 | Py_DECREF(reader); |
| 2756 | return NULL; |
| 2757 | } |
| 2758 | Py_DECREF(res); |
| 2759 | |
| 2760 | } |
| 2761 | |
| 2762 | Py_DECREF(reader); |
| 2763 | |
| 2764 | res = expat_parse(self, "", 0, 1); |
| 2765 | |
| 2766 | if (res && TreeBuilder_CheckExact(self->target)) { |
| 2767 | Py_DECREF(res); |
| 2768 | return treebuilder_done((TreeBuilderObject*) self->target); |
| 2769 | } |
| 2770 | |
| 2771 | return res; |
| 2772 | } |
| 2773 | |
| 2774 | static PyObject* |
| 2775 | xmlparser_setevents(XMLParserObject* self, PyObject* args) |
| 2776 | { |
| 2777 | /* activate element event reporting */ |
| 2778 | |
Neal Norwitz | c707438 | 2006-06-12 02:06:17 +0000 | [diff] [blame] | 2779 | Py_ssize_t i; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2780 | TreeBuilderObject* target; |
| 2781 | |
| 2782 | PyObject* events; /* event collector */ |
| 2783 | PyObject* event_set = Py_None; |
| 2784 | if (!PyArg_ParseTuple(args, "O!|O:_setevents", &PyList_Type, &events, |
| 2785 | &event_set)) |
| 2786 | return NULL; |
| 2787 | |
| 2788 | if (!TreeBuilder_CheckExact(self->target)) { |
| 2789 | PyErr_SetString( |
| 2790 | PyExc_TypeError, |
| 2791 | "event handling only supported for cElementTree.Treebuilder " |
| 2792 | "targets" |
| 2793 | ); |
| 2794 | return NULL; |
| 2795 | } |
| 2796 | |
| 2797 | target = (TreeBuilderObject*) self->target; |
| 2798 | |
| 2799 | Py_INCREF(events); |
Serhiy Storchaka | bc62af1 | 2016-04-06 09:51:18 +0300 | [diff] [blame] | 2800 | Py_XSETREF(target->events, events); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2801 | |
| 2802 | /* clear out existing events */ |
Serhiy Storchaka | 98a9722 | 2014-02-09 13:14:04 +0200 | [diff] [blame] | 2803 | Py_CLEAR(target->start_event_obj); |
| 2804 | Py_CLEAR(target->end_event_obj); |
| 2805 | Py_CLEAR(target->start_ns_event_obj); |
| 2806 | Py_CLEAR(target->end_ns_event_obj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2807 | |
| 2808 | if (event_set == Py_None) { |
| 2809 | /* default is "end" only */ |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2810 | target->end_event_obj = PyString_FromString("end"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2811 | Py_RETURN_NONE; |
| 2812 | } |
| 2813 | |
| 2814 | if (!PyTuple_Check(event_set)) /* FIXME: handle arbitrary sequences */ |
| 2815 | goto error; |
| 2816 | |
| 2817 | for (i = 0; i < PyTuple_GET_SIZE(event_set); i++) { |
| 2818 | PyObject* item = PyTuple_GET_ITEM(event_set, i); |
| 2819 | char* event; |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2820 | if (!PyString_Check(item)) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2821 | goto error; |
Serhiy Storchaka | 20a003b | 2015-12-24 11:51:24 +0200 | [diff] [blame] | 2822 | Py_INCREF(item); |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2823 | event = PyString_AS_STRING(item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2824 | if (strcmp(event, "start") == 0) { |
Serhiy Storchaka | bc62af1 | 2016-04-06 09:51:18 +0300 | [diff] [blame] | 2825 | Py_XSETREF(target->start_event_obj, item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2826 | } else if (strcmp(event, "end") == 0) { |
Serhiy Storchaka | bc62af1 | 2016-04-06 09:51:18 +0300 | [diff] [blame] | 2827 | Py_XSETREF(target->end_event_obj, item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2828 | } else if (strcmp(event, "start-ns") == 0) { |
Serhiy Storchaka | bc62af1 | 2016-04-06 09:51:18 +0300 | [diff] [blame] | 2829 | Py_XSETREF(target->start_ns_event_obj, item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2830 | EXPAT(SetNamespaceDeclHandler)( |
| 2831 | self->parser, |
| 2832 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 2833 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 2834 | ); |
| 2835 | } else if (strcmp(event, "end-ns") == 0) { |
Serhiy Storchaka | bc62af1 | 2016-04-06 09:51:18 +0300 | [diff] [blame] | 2836 | Py_XSETREF(target->end_ns_event_obj, item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2837 | EXPAT(SetNamespaceDeclHandler)( |
| 2838 | self->parser, |
| 2839 | (XML_StartNamespaceDeclHandler) expat_start_ns_handler, |
| 2840 | (XML_EndNamespaceDeclHandler) expat_end_ns_handler |
| 2841 | ); |
| 2842 | } else { |
Serhiy Storchaka | 20a003b | 2015-12-24 11:51:24 +0200 | [diff] [blame] | 2843 | Py_DECREF(item); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2844 | PyErr_Format( |
| 2845 | PyExc_ValueError, |
| 2846 | "unknown event '%s'", event |
| 2847 | ); |
| 2848 | return NULL; |
| 2849 | } |
| 2850 | } |
| 2851 | |
| 2852 | Py_RETURN_NONE; |
| 2853 | |
| 2854 | error: |
| 2855 | PyErr_SetString( |
| 2856 | PyExc_TypeError, |
| 2857 | "invalid event tuple" |
| 2858 | ); |
| 2859 | return NULL; |
| 2860 | } |
| 2861 | |
| 2862 | static PyMethodDef xmlparser_methods[] = { |
| 2863 | {"feed", (PyCFunction) xmlparser_feed, METH_VARARGS}, |
| 2864 | {"close", (PyCFunction) xmlparser_close, METH_VARARGS}, |
| 2865 | {"_parse", (PyCFunction) xmlparser_parse, METH_VARARGS}, |
| 2866 | {"_setevents", (PyCFunction) xmlparser_setevents, METH_VARARGS}, |
| 2867 | {NULL, NULL} |
| 2868 | }; |
| 2869 | |
| 2870 | static PyObject* |
| 2871 | xmlparser_getattr(XMLParserObject* self, char* name) |
| 2872 | { |
| 2873 | PyObject* res; |
| 2874 | |
| 2875 | res = Py_FindMethod(xmlparser_methods, (PyObject*) self, name); |
| 2876 | if (res) |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2877 | return res; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2878 | |
| 2879 | PyErr_Clear(); |
| 2880 | |
| 2881 | if (strcmp(name, "entity") == 0) |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2882 | res = self->entity; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2883 | else if (strcmp(name, "target") == 0) |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2884 | res = self->target; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2885 | else if (strcmp(name, "version") == 0) { |
| 2886 | char buffer[100]; |
| 2887 | sprintf(buffer, "Expat %d.%d.%d", XML_MAJOR_VERSION, |
| 2888 | XML_MINOR_VERSION, XML_MICRO_VERSION); |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 2889 | return PyString_FromString(buffer); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2890 | } else { |
| 2891 | PyErr_SetString(PyExc_AttributeError, name); |
| 2892 | return NULL; |
| 2893 | } |
| 2894 | |
| 2895 | Py_INCREF(res); |
| 2896 | return res; |
| 2897 | } |
| 2898 | |
| 2899 | statichere PyTypeObject XMLParser_Type = { |
| 2900 | PyObject_HEAD_INIT(NULL) |
| 2901 | 0, "XMLParser", sizeof(XMLParserObject), 0, |
| 2902 | /* methods */ |
| 2903 | (destructor)xmlparser_dealloc, /* tp_dealloc */ |
| 2904 | 0, /* tp_print */ |
| 2905 | (getattrfunc)xmlparser_getattr, /* tp_getattr */ |
| 2906 | }; |
| 2907 | |
| 2908 | #endif |
| 2909 | |
| 2910 | /* ==================================================================== */ |
| 2911 | /* python module interface */ |
| 2912 | |
| 2913 | static PyMethodDef _functions[] = { |
| 2914 | {"Element", (PyCFunction) element, METH_VARARGS|METH_KEYWORDS}, |
| 2915 | {"SubElement", (PyCFunction) subelement, METH_VARARGS|METH_KEYWORDS}, |
| 2916 | {"TreeBuilder", (PyCFunction) treebuilder, METH_VARARGS}, |
| 2917 | #if defined(USE_EXPAT) |
| 2918 | {"XMLParser", (PyCFunction) xmlparser, METH_VARARGS|METH_KEYWORDS}, |
| 2919 | {"XMLTreeBuilder", (PyCFunction) xmlparser, METH_VARARGS|METH_KEYWORDS}, |
| 2920 | #endif |
| 2921 | {NULL, NULL} |
| 2922 | }; |
| 2923 | |
| 2924 | DL_EXPORT(void) |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2925 | init_elementtree(void) |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2926 | { |
| 2927 | PyObject* m; |
| 2928 | PyObject* g; |
| 2929 | char* bootstrap; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2930 | |
| 2931 | /* Patch object type */ |
Christian Heimes | e93237d | 2007-12-19 02:37:44 +0000 | [diff] [blame] | 2932 | Py_TYPE(&Element_Type) = Py_TYPE(&TreeBuilder_Type) = &PyType_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2933 | #if defined(USE_EXPAT) |
Christian Heimes | e93237d | 2007-12-19 02:37:44 +0000 | [diff] [blame] | 2934 | Py_TYPE(&XMLParser_Type) = &PyType_Type; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2935 | #endif |
| 2936 | |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2937 | m = Py_InitModule("_elementtree", _functions); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2938 | if (!m) |
| 2939 | return; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2940 | |
| 2941 | /* python glue code */ |
| 2942 | |
| 2943 | g = PyDict_New(); |
Fredrik Lundh | 44ed4db | 2006-03-12 21:06:35 +0000 | [diff] [blame] | 2944 | if (!g) |
| 2945 | return; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2946 | |
| 2947 | PyDict_SetItemString(g, "__builtins__", PyEval_GetBuiltins()); |
| 2948 | |
| 2949 | bootstrap = ( |
| 2950 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2951 | "from copy import copy, deepcopy\n" |
| 2952 | |
| 2953 | "try:\n" |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2954 | " from xml.etree import ElementTree\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2955 | "except ImportError:\n" |
| 2956 | " import ElementTree\n" |
| 2957 | "ET = ElementTree\n" |
| 2958 | "del ElementTree\n" |
| 2959 | |
Fredrik Lundh | 6d52b55 | 2005-12-16 22:06:43 +0000 | [diff] [blame] | 2960 | "import _elementtree as cElementTree\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2961 | |
| 2962 | "try:\n" /* check if copy works as is */ |
| 2963 | " copy(cElementTree.Element('x'))\n" |
| 2964 | "except:\n" |
| 2965 | " def copyelement(elem):\n" |
| 2966 | " return elem\n" |
| 2967 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2968 | "class CommentProxy:\n" |
| 2969 | " def __call__(self, text=None):\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2970 | " element = cElementTree.Element(ET.Comment)\n" |
| 2971 | " element.text = text\n" |
| 2972 | " return element\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 2973 | " def __cmp__(self, other):\n" |
| 2974 | " return cmp(ET.Comment, other)\n" |
| 2975 | "cElementTree.Comment = CommentProxy()\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2976 | |
| 2977 | "class ElementTree(ET.ElementTree):\n" /* public */ |
| 2978 | " def parse(self, source, parser=None):\n" |
Florent Xicluna | 67d5d0e | 2011-10-29 03:38:56 +0200 | [diff] [blame] | 2979 | " close_source = False\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2980 | " if not hasattr(source, 'read'):\n" |
| 2981 | " source = open(source, 'rb')\n" |
Florent Xicluna | 67d5d0e | 2011-10-29 03:38:56 +0200 | [diff] [blame] | 2982 | " close_source = False\n" |
| 2983 | " try:\n" |
| 2984 | " if parser is not None:\n" |
| 2985 | " while 1:\n" |
| 2986 | " data = source.read(65536)\n" |
| 2987 | " if not data:\n" |
| 2988 | " break\n" |
| 2989 | " parser.feed(data)\n" |
| 2990 | " self._root = parser.close()\n" |
| 2991 | " else:\n" |
| 2992 | " parser = cElementTree.XMLParser()\n" |
| 2993 | " self._root = parser._parse(source)\n" |
| 2994 | " return self._root\n" |
| 2995 | " finally:\n" |
| 2996 | " if close_source:\n" |
| 2997 | " source.close()\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 2998 | "cElementTree.ElementTree = ElementTree\n" |
| 2999 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3000 | "def iter(node, tag=None):\n" /* helper */ |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3001 | " if tag == '*':\n" |
| 3002 | " tag = None\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3003 | " if tag is None or node.tag == tag:\n" |
| 3004 | " yield node\n" |
| 3005 | " for node in node:\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3006 | " for node in iter(node, tag):\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3007 | " yield node\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3008 | |
| 3009 | "def itertext(node):\n" /* helper */ |
| 3010 | " if node.text:\n" |
| 3011 | " yield node.text\n" |
| 3012 | " for e in node:\n" |
| 3013 | " for s in e.itertext():\n" |
| 3014 | " yield s\n" |
| 3015 | " if e.tail:\n" |
| 3016 | " yield e.tail\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3017 | |
| 3018 | "def parse(source, parser=None):\n" /* public */ |
| 3019 | " tree = ElementTree()\n" |
| 3020 | " tree.parse(source, parser)\n" |
| 3021 | " return tree\n" |
| 3022 | "cElementTree.parse = parse\n" |
| 3023 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3024 | "class iterparse(object):\n" |
| 3025 | " root = None\n" |
| 3026 | " def __init__(self, file, events=None):\n" |
Florent Xicluna | 67d5d0e | 2011-10-29 03:38:56 +0200 | [diff] [blame] | 3027 | " self._close_file = False\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3028 | " if not hasattr(file, 'read'):\n" |
| 3029 | " file = open(file, 'rb')\n" |
Florent Xicluna | 67d5d0e | 2011-10-29 03:38:56 +0200 | [diff] [blame] | 3030 | " self._close_file = True\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3031 | " self._file = file\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3032 | " self._events = []\n" |
| 3033 | " self._index = 0\n" |
Florent Xicluna | 0965ee2 | 2011-11-01 23:34:41 +0100 | [diff] [blame] | 3034 | " self._error = None\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3035 | " self.root = self._root = None\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3036 | " b = cElementTree.TreeBuilder()\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3037 | " self._parser = cElementTree.XMLParser(b)\n" |
| 3038 | " self._parser._setevents(self._events, events)\n" |
| 3039 | " def next(self):\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3040 | " while 1:\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3041 | " try:\n" |
| 3042 | " item = self._events[self._index]\n" |
Florent Xicluna | 0965ee2 | 2011-11-01 23:34:41 +0100 | [diff] [blame] | 3043 | " self._index += 1\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3044 | " return item\n" |
Florent Xicluna | 0965ee2 | 2011-11-01 23:34:41 +0100 | [diff] [blame] | 3045 | " except IndexError:\n" |
| 3046 | " pass\n" |
| 3047 | " if self._error:\n" |
| 3048 | " e = self._error\n" |
| 3049 | " self._error = None\n" |
| 3050 | " raise e\n" |
| 3051 | " if self._parser is None:\n" |
| 3052 | " self.root = self._root\n" |
| 3053 | " if self._close_file:\n" |
| 3054 | " self._file.close()\n" |
| 3055 | " raise StopIteration\n" |
| 3056 | " # load event buffer\n" |
| 3057 | " del self._events[:]\n" |
| 3058 | " self._index = 0\n" |
| 3059 | " data = self._file.read(16384)\n" |
| 3060 | " if data:\n" |
| 3061 | " try:\n" |
| 3062 | " self._parser.feed(data)\n" |
| 3063 | " except SyntaxError as exc:\n" |
| 3064 | " self._error = exc\n" |
| 3065 | " else:\n" |
| 3066 | " self._root = self._parser.close()\n" |
| 3067 | " self._parser = None\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3068 | " def __iter__(self):\n" |
| 3069 | " return self\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3070 | "cElementTree.iterparse = iterparse\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3071 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3072 | "class PIProxy:\n" |
| 3073 | " def __call__(self, target, text=None):\n" |
| 3074 | " element = cElementTree.Element(ET.PI)\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3075 | " element.text = target\n" |
| 3076 | " if text:\n" |
| 3077 | " element.text = element.text + ' ' + text\n" |
| 3078 | " return element\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3079 | " def __cmp__(self, other):\n" |
| 3080 | " return cmp(ET.PI, other)\n" |
| 3081 | "cElementTree.PI = cElementTree.ProcessingInstruction = PIProxy()\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3082 | |
| 3083 | "def XML(text):\n" /* public */ |
| 3084 | " parser = cElementTree.XMLParser()\n" |
| 3085 | " parser.feed(text)\n" |
| 3086 | " return parser.close()\n" |
| 3087 | "cElementTree.XML = cElementTree.fromstring = XML\n" |
| 3088 | |
| 3089 | "def XMLID(text):\n" /* public */ |
| 3090 | " tree = XML(text)\n" |
| 3091 | " ids = {}\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3092 | " for elem in tree.iter():\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3093 | " id = elem.get('id')\n" |
| 3094 | " if id:\n" |
| 3095 | " ids[id] = elem\n" |
| 3096 | " return tree, ids\n" |
| 3097 | "cElementTree.XMLID = XMLID\n" |
| 3098 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3099 | "try:\n" |
| 3100 | " register_namespace = ET.register_namespace\n" |
| 3101 | "except AttributeError:\n" |
| 3102 | " def register_namespace(prefix, uri):\n" |
| 3103 | " ET._namespace_map[uri] = prefix\n" |
| 3104 | "cElementTree.register_namespace = register_namespace\n" |
| 3105 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3106 | "cElementTree.dump = ET.dump\n" |
| 3107 | "cElementTree.ElementPath = ElementPath = ET.ElementPath\n" |
| 3108 | "cElementTree.iselement = ET.iselement\n" |
| 3109 | "cElementTree.QName = ET.QName\n" |
| 3110 | "cElementTree.tostring = ET.tostring\n" |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3111 | "cElementTree.fromstringlist = ET.fromstringlist\n" |
| 3112 | "cElementTree.tostringlist = ET.tostringlist\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3113 | "cElementTree.VERSION = '" VERSION "'\n" |
| 3114 | "cElementTree.__version__ = '" VERSION "'\n" |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3115 | |
| 3116 | ); |
| 3117 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3118 | if (!PyRun_String(bootstrap, Py_file_input, g, NULL)) |
| 3119 | return; |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3120 | |
| 3121 | elementpath_obj = PyDict_GetItemString(g, "ElementPath"); |
| 3122 | |
| 3123 | elementtree_copyelement_obj = PyDict_GetItemString(g, "copyelement"); |
| 3124 | if (elementtree_copyelement_obj) { |
| 3125 | /* reduce hack needed; enable reduce method */ |
| 3126 | PyMethodDef* mp; |
| 3127 | for (mp = element_methods; mp->ml_name; mp++) |
| 3128 | if (mp->ml_meth == (PyCFunction) element_reduce) { |
| 3129 | mp->ml_name = "__reduce__"; |
| 3130 | break; |
| 3131 | } |
| 3132 | } else |
| 3133 | PyErr_Clear(); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3134 | |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3135 | elementtree_deepcopy_obj = PyDict_GetItemString(g, "deepcopy"); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3136 | elementtree_iter_obj = PyDict_GetItemString(g, "iter"); |
| 3137 | elementtree_itertext_obj = PyDict_GetItemString(g, "itertext"); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3138 | |
| 3139 | #if defined(USE_PYEXPAT_CAPI) |
| 3140 | /* link against pyexpat, if possible */ |
Larry Hastings | 402b73f | 2010-03-25 00:54:54 +0000 | [diff] [blame] | 3141 | expat_capi = PyCapsule_Import(PyExpat_CAPSULE_NAME, 0); |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3142 | if (expat_capi) { |
| 3143 | /* check that it's usable */ |
| 3144 | if (strcmp(expat_capi->magic, PyExpat_CAPI_MAGIC) != 0 || |
| 3145 | expat_capi->size < sizeof(struct PyExpat_CAPI) || |
| 3146 | expat_capi->MAJOR_VERSION != XML_MAJOR_VERSION || |
| 3147 | expat_capi->MINOR_VERSION != XML_MINOR_VERSION || |
| 3148 | expat_capi->MICRO_VERSION != XML_MICRO_VERSION) |
| 3149 | expat_capi = NULL; |
| 3150 | } |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3151 | #endif |
| 3152 | |
Florent Xicluna | 3e8c189 | 2010-03-11 14:36:19 +0000 | [diff] [blame] | 3153 | elementtree_parseerror_obj = PyErr_NewException( |
| 3154 | "cElementTree.ParseError", PyExc_SyntaxError, NULL |
| 3155 | ); |
| 3156 | Py_INCREF(elementtree_parseerror_obj); |
| 3157 | PyModule_AddObject(m, "ParseError", elementtree_parseerror_obj); |
Fredrik Lundh | 8c8836b | 2005-12-16 22:06:06 +0000 | [diff] [blame] | 3158 | } |