blob: 47c3e86c209d9c1a1d1d1627610ed6d9585e65a7 [file] [log] [blame]
Martin v. Löwis7090ed12001-09-19 10:37:50 +00001#include "Python.h"
Fred Drake4113b132001-03-24 19:58:26 +00002#include <ctype.h>
3
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00004#include "frameobject.h"
Fred Drakea77254a2000-09-29 19:23:29 +00005#include "expat.h"
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00006
Fredrik Lundhc3345042005-12-13 19:49:55 +00007#include "pyexpat.h"
8
Brett Cannond0aeda82014-08-22 14:23:20 -04009/* Do not emit Clinic output to a file as that wreaks havoc with conditionally
10 included methods. */
11/*[clinic input]
12module pyexpat
13[clinic start generated code]*/
14/*[clinic end generated code: output=da39a3ee5e6b4b0d input=b168d503a4490c15]*/
15
Martin v. Löwisc847f402003-01-21 11:09:21 +000016#define XML_COMBINED_VERSION (10000*XML_MAJOR_VERSION+100*XML_MINOR_VERSION+XML_MICRO_VERSION)
17
Christian Heimesfa535f52013-07-07 17:35:11 +020018static XML_Memory_Handling_Suite ExpatMemoryHandler = {
19 PyObject_Malloc, PyObject_Realloc, PyObject_Free};
20
Fred Drake0582df92000-07-12 04:49:00 +000021enum HandlerTypes {
22 StartElement,
23 EndElement,
24 ProcessingInstruction,
25 CharacterData,
26 UnparsedEntityDecl,
27 NotationDecl,
28 StartNamespaceDecl,
29 EndNamespaceDecl,
30 Comment,
31 StartCdataSection,
32 EndCdataSection,
33 Default,
34 DefaultHandlerExpand,
35 NotStandalone,
Martin v. Löwis0078f6c2001-01-21 10:18:10 +000036 ExternalEntityRef,
37 StartDoctypeDecl,
38 EndDoctypeDecl,
Fred Drake85d835f2001-02-08 15:39:08 +000039 EntityDecl,
40 XmlDecl,
41 ElementDecl,
42 AttlistDecl,
Martin v. Löwisc847f402003-01-21 11:09:21 +000043#if XML_COMBINED_VERSION >= 19504
Martin v. Löwis069dde22003-01-21 10:58:18 +000044 SkippedEntity,
Martin v. Löwisc847f402003-01-21 11:09:21 +000045#endif
Fred Drake85d835f2001-02-08 15:39:08 +000046 _DummyDecl
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +000047};
48
49static PyObject *ErrorObject;
50
51/* ----------------------------------------------------- */
52
53/* Declarations for objects of type xmlparser */
54
55typedef struct {
Fred Drake0582df92000-07-12 04:49:00 +000056 PyObject_HEAD
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +000057
Fred Drake0582df92000-07-12 04:49:00 +000058 XML_Parser itself;
Fred Drake85d835f2001-02-08 15:39:08 +000059 int ordered_attributes; /* Return attributes as a list. */
60 int specified_attributes; /* Report only specified attributes. */
Fred Drakebd6101c2001-02-14 18:29:45 +000061 int in_callback; /* Is a callback active? */
Martin v. Löwis069dde22003-01-21 10:58:18 +000062 int ns_prefixes; /* Namespace-triplets mode? */
Fred Drake2a3d7db2002-06-28 22:56:48 +000063 XML_Char *buffer; /* Buffer used when accumulating characters */
64 /* NULL if not enabled */
65 int buffer_size; /* Size of buffer, in XML_Char units */
66 int buffer_used; /* Buffer units in use */
Fred Drakeb91a36b2002-06-27 19:40:48 +000067 PyObject *intern; /* Dictionary to intern strings */
Fred Drake0582df92000-07-12 04:49:00 +000068 PyObject **handlers;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +000069} xmlparseobject;
70
Serhiy Storchaka1009bf12015-04-03 23:53:51 +030071#include "clinic/pyexpat.c.h"
72
Fred Drake2a3d7db2002-06-28 22:56:48 +000073#define CHARACTER_DATA_BUFFER_SIZE 8192
74
Jeremy Hylton938ace62002-07-17 16:30:39 +000075static PyTypeObject Xmlparsetype;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +000076
Fred Drake117ac852002-09-24 16:24:54 +000077typedef void (*xmlhandlersetter)(XML_Parser self, void *meth);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +000078typedef void* xmlhandler;
79
Andrew M. Kuchlingbeba0562000-06-27 00:33:30 +000080struct HandlerInfo {
Fred Drake0582df92000-07-12 04:49:00 +000081 const char *name;
82 xmlhandlersetter setter;
83 xmlhandler handler;
Martin v. Löwis0078f6c2001-01-21 10:18:10 +000084 PyCodeObject *tb_code;
Fred Drake71b63ff2002-06-28 22:29:01 +000085 PyObject *nameobj;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +000086};
87
Jeremy Hylton938ace62002-07-17 16:30:39 +000088static struct HandlerInfo handler_info[64];
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +000089
Fred Drakebd6101c2001-02-14 18:29:45 +000090/* Set an integer attribute on the error object; return true on success,
91 * false on an exception.
92 */
93static int
Serhiy Storchakaef1585e2015-12-25 20:01:53 +020094set_error_attr(PyObject *err, const char *name, int value)
Fred Drakebd6101c2001-02-14 18:29:45 +000095{
Christian Heimes217cfd12007-12-02 14:31:20 +000096 PyObject *v = PyLong_FromLong(value);
Fred Drake85d835f2001-02-08 15:39:08 +000097
Neal Norwitz2f5e9902006-03-08 06:36:45 +000098 if (v == NULL || PyObject_SetAttrString(err, name, v) == -1) {
99 Py_XDECREF(v);
Fred Drakebd6101c2001-02-14 18:29:45 +0000100 return 0;
101 }
Michael W. Hudson0bb84542004-08-03 11:31:31 +0000102 Py_DECREF(v);
Fred Drakebd6101c2001-02-14 18:29:45 +0000103 return 1;
104}
105
106/* Build and set an Expat exception, including positioning
107 * information. Always returns NULL.
108 */
Fred Drake85d835f2001-02-08 15:39:08 +0000109static PyObject *
Martin v. Löwis069dde22003-01-21 10:58:18 +0000110set_error(xmlparseobject *self, enum XML_Error code)
Fred Drake85d835f2001-02-08 15:39:08 +0000111{
112 PyObject *err;
Victor Stinner499dfcf2011-03-21 13:26:24 +0100113 PyObject *buffer;
Fred Drake85d835f2001-02-08 15:39:08 +0000114 XML_Parser parser = self->itself;
Fred Drakebd6101c2001-02-14 18:29:45 +0000115 int lineno = XML_GetErrorLineNumber(parser);
116 int column = XML_GetErrorColumnNumber(parser);
Fred Drake85d835f2001-02-08 15:39:08 +0000117
Victor Stinner499dfcf2011-03-21 13:26:24 +0100118 buffer = PyUnicode_FromFormat("%s: line %i, column %i",
119 XML_ErrorString(code), lineno, column);
120 if (buffer == NULL)
121 return NULL;
122 err = PyObject_CallFunction(ErrorObject, "O", buffer);
123 Py_DECREF(buffer);
Fred Drakebd6101c2001-02-14 18:29:45 +0000124 if ( err != NULL
125 && set_error_attr(err, "code", code)
126 && set_error_attr(err, "offset", column)
127 && set_error_attr(err, "lineno", lineno)) {
128 PyErr_SetObject(ErrorObject, err);
Fred Drake85d835f2001-02-08 15:39:08 +0000129 }
Neal Norwitz2f5e9902006-03-08 06:36:45 +0000130 Py_XDECREF(err);
Fred Drake85d835f2001-02-08 15:39:08 +0000131 return NULL;
132}
133
Fred Drake71b63ff2002-06-28 22:29:01 +0000134static int
135have_handler(xmlparseobject *self, int type)
136{
137 PyObject *handler = self->handlers[type];
138 return handler != NULL;
139}
140
141static PyObject *
142get_handler_name(struct HandlerInfo *hinfo)
143{
144 PyObject *name = hinfo->nameobj;
145 if (name == NULL) {
Neal Norwitz392c5be2007-08-25 17:20:32 +0000146 name = PyUnicode_FromString(hinfo->name);
Fred Drake71b63ff2002-06-28 22:29:01 +0000147 hinfo->nameobj = name;
148 }
149 Py_XINCREF(name);
150 return name;
151}
152
Fred Drake85d835f2001-02-08 15:39:08 +0000153
Andrew M. Kuchlingbeba0562000-06-27 00:33:30 +0000154/* Convert a string of XML_Chars into a Unicode string.
155 Returns None if str is a null pointer. */
156
Fred Drake0582df92000-07-12 04:49:00 +0000157static PyObject *
Fred Drakeb91a36b2002-06-27 19:40:48 +0000158conv_string_to_unicode(const XML_Char *str)
Fred Drake0582df92000-07-12 04:49:00 +0000159{
Fred Drake71b63ff2002-06-28 22:29:01 +0000160 /* XXX currently this code assumes that XML_Char is 8-bit,
Fred Drake0582df92000-07-12 04:49:00 +0000161 and hence in UTF-8. */
162 /* UTF-8 from Expat, Unicode desired */
163 if (str == NULL) {
164 Py_INCREF(Py_None);
165 return Py_None;
166 }
Fred Drake71b63ff2002-06-28 22:29:01 +0000167 return PyUnicode_DecodeUTF8(str, strlen(str), "strict");
Andrew M. Kuchlingbeba0562000-06-27 00:33:30 +0000168}
169
Fred Drake0582df92000-07-12 04:49:00 +0000170static PyObject *
171conv_string_len_to_unicode(const XML_Char *str, int len)
172{
Fred Drake71b63ff2002-06-28 22:29:01 +0000173 /* XXX currently this code assumes that XML_Char is 8-bit,
Fred Drake0582df92000-07-12 04:49:00 +0000174 and hence in UTF-8. */
175 /* UTF-8 from Expat, Unicode desired */
176 if (str == NULL) {
177 Py_INCREF(Py_None);
178 return Py_None;
179 }
Fred Drake6f987622000-08-25 18:03:30 +0000180 return PyUnicode_DecodeUTF8((const char *)str, len, "strict");
Andrew M. Kuchlingbeba0562000-06-27 00:33:30 +0000181}
Andrew M. Kuchlingbeba0562000-06-27 00:33:30 +0000182
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000183/* Callback routines */
184
Martin v. Löwis5b68ce32001-10-21 08:53:52 +0000185static void clear_handlers(xmlparseobject *self, int initial);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000186
Martin v. Löwis069dde22003-01-21 10:58:18 +0000187/* This handler is used when an error has been detected, in the hope
188 that actual parsing can be terminated early. This will only help
189 if an external entity reference is encountered. */
190static int
191error_external_entity_ref_handler(XML_Parser parser,
192 const XML_Char *context,
193 const XML_Char *base,
194 const XML_Char *systemId,
195 const XML_Char *publicId)
196{
197 return 0;
198}
199
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000200/* Dummy character data handler used when an error (exception) has
201 been detected, and the actual parsing can be terminated early.
202 This is needed since character data handler can't be safely removed
203 from within the character data handler, but can be replaced. It is
204 used only from the character data handler trampoline, and must be
205 used right after `flag_error()` is called. */
206static void
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000207noop_character_data_handler(void *userData, const XML_Char *data, int len)
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000208{
209 /* Do nothing. */
210}
211
Fred Drake6f987622000-08-25 18:03:30 +0000212static void
213flag_error(xmlparseobject *self)
214{
Martin v. Löwis5b68ce32001-10-21 08:53:52 +0000215 clear_handlers(self, 0);
Martin v. Löwis069dde22003-01-21 10:58:18 +0000216 XML_SetExternalEntityRefHandler(self->itself,
217 error_external_entity_ref_handler);
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000218}
219
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000220static PyObject*
Serhiy Storchakaef1585e2015-12-25 20:01:53 +0200221call_with_frame(const char *funcname, int lineno, PyObject* func, PyObject* args,
Fred Drake39689c52004-08-13 03:12:57 +0000222 xmlparseobject *self)
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000223{
Antoine Pitrou0ddbf472014-10-08 20:00:09 +0200224 PyObject *res;
Fred Drakebd6101c2001-02-14 18:29:45 +0000225
Fred Drakebd6101c2001-02-14 18:29:45 +0000226 res = PyEval_CallObject(func, args);
Jeremy Hylton9263f572003-06-27 16:13:17 +0000227 if (res == NULL) {
Antoine Pitrou0ddbf472014-10-08 20:00:09 +0200228 _PyTraceback_Add(funcname, __FILE__, lineno);
Fred Drake39689c52004-08-13 03:12:57 +0000229 XML_StopParser(self->itself, XML_FALSE);
Jeremy Hylton9263f572003-06-27 16:13:17 +0000230 }
Fred Drakebd6101c2001-02-14 18:29:45 +0000231 return res;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000232}
233
Fred Drakeb91a36b2002-06-27 19:40:48 +0000234static PyObject*
235string_intern(xmlparseobject *self, const char* str)
236{
Guido van Rossum4ca94712007-07-23 17:42:32 +0000237 PyObject *result = conv_string_to_unicode(str);
Fred Drakeb91a36b2002-06-27 19:40:48 +0000238 PyObject *value;
Neal Norwitz484d9a42005-09-30 04:46:49 +0000239 /* result can be NULL if the unicode conversion failed. */
240 if (!result)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000241 return result;
Fred Drakeb91a36b2002-06-27 19:40:48 +0000242 if (!self->intern)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000243 return result;
Fred Drakeb91a36b2002-06-27 19:40:48 +0000244 value = PyDict_GetItem(self->intern, result);
245 if (!value) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000246 if (PyDict_SetItem(self->intern, result, result) == 0)
Fred Drakeb91a36b2002-06-27 19:40:48 +0000247 return result;
248 else
249 return NULL;
250 }
251 Py_INCREF(value);
252 Py_DECREF(result);
253 return value;
254}
255
Fred Drake2a3d7db2002-06-28 22:56:48 +0000256/* Return 0 on success, -1 on exception.
257 * flag_error() will be called before return if needed.
258 */
259static int
260call_character_handler(xmlparseobject *self, const XML_Char *buffer, int len)
261{
262 PyObject *args;
263 PyObject *temp;
264
Georg Brandlc01537f2010-10-15 16:26:08 +0000265 if (!have_handler(self, CharacterData))
266 return -1;
267
Fred Drake2a3d7db2002-06-28 22:56:48 +0000268 args = PyTuple_New(1);
269 if (args == NULL)
270 return -1;
Guido van Rossum4ca94712007-07-23 17:42:32 +0000271 temp = (conv_string_len_to_unicode(buffer, len));
Fred Drake2a3d7db2002-06-28 22:56:48 +0000272 if (temp == NULL) {
273 Py_DECREF(args);
274 flag_error(self);
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000275 XML_SetCharacterDataHandler(self->itself,
276 noop_character_data_handler);
Fred Drake2a3d7db2002-06-28 22:56:48 +0000277 return -1;
278 }
279 PyTuple_SET_ITEM(args, 0, temp);
280 /* temp is now a borrowed reference; consider it unused. */
281 self->in_callback = 1;
Antoine Pitrou0ddbf472014-10-08 20:00:09 +0200282 temp = call_with_frame("CharacterData", __LINE__,
Fred Drake39689c52004-08-13 03:12:57 +0000283 self->handlers[CharacterData], args, self);
Fred Drake2a3d7db2002-06-28 22:56:48 +0000284 /* temp is an owned reference again, or NULL */
285 self->in_callback = 0;
286 Py_DECREF(args);
287 if (temp == NULL) {
288 flag_error(self);
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000289 XML_SetCharacterDataHandler(self->itself,
290 noop_character_data_handler);
Fred Drake2a3d7db2002-06-28 22:56:48 +0000291 return -1;
292 }
293 Py_DECREF(temp);
294 return 0;
295}
296
297static int
298flush_character_buffer(xmlparseobject *self)
299{
300 int rc;
301 if (self->buffer == NULL || self->buffer_used == 0)
302 return 0;
303 rc = call_character_handler(self, self->buffer, self->buffer_used);
304 self->buffer_used = 0;
305 return rc;
306}
307
308static void
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000309my_CharacterDataHandler(void *userData, const XML_Char *data, int len)
Fred Drake2a3d7db2002-06-28 22:56:48 +0000310{
311 xmlparseobject *self = (xmlparseobject *) userData;
Victor Stinner9e09c262013-07-18 23:17:01 +0200312
313 if (PyErr_Occurred())
314 return;
315
Fred Drake2a3d7db2002-06-28 22:56:48 +0000316 if (self->buffer == NULL)
317 call_character_handler(self, data, len);
318 else {
319 if ((self->buffer_used + len) > self->buffer_size) {
320 if (flush_character_buffer(self) < 0)
321 return;
322 /* handler might have changed; drop the rest on the floor
323 * if there isn't a handler anymore
324 */
325 if (!have_handler(self, CharacterData))
326 return;
327 }
328 if (len > self->buffer_size) {
329 call_character_handler(self, data, len);
330 self->buffer_used = 0;
331 }
332 else {
333 memcpy(self->buffer + self->buffer_used,
334 data, len * sizeof(XML_Char));
335 self->buffer_used += len;
336 }
337 }
338}
339
Fred Drake85d835f2001-02-08 15:39:08 +0000340static void
341my_StartElementHandler(void *userData,
Fred Drake71b63ff2002-06-28 22:29:01 +0000342 const XML_Char *name, const XML_Char *atts[])
Fred Drake85d835f2001-02-08 15:39:08 +0000343{
344 xmlparseobject *self = (xmlparseobject *)userData;
345
Fred Drake71b63ff2002-06-28 22:29:01 +0000346 if (have_handler(self, StartElement)) {
Fred Drake85d835f2001-02-08 15:39:08 +0000347 PyObject *container, *rv, *args;
348 int i, max;
349
Victor Stinner9e09c262013-07-18 23:17:01 +0200350 if (PyErr_Occurred())
351 return;
352
Fred Drake2a3d7db2002-06-28 22:56:48 +0000353 if (flush_character_buffer(self) < 0)
354 return;
Fred Drake85d835f2001-02-08 15:39:08 +0000355 /* Set max to the number of slots filled in atts[]; max/2 is
356 * the number of attributes we need to process.
357 */
358 if (self->specified_attributes) {
359 max = XML_GetSpecifiedAttributeCount(self->itself);
360 }
361 else {
362 max = 0;
363 while (atts[max] != NULL)
364 max += 2;
365 }
366 /* Build the container. */
367 if (self->ordered_attributes)
368 container = PyList_New(max);
369 else
370 container = PyDict_New();
371 if (container == NULL) {
372 flag_error(self);
373 return;
374 }
375 for (i = 0; i < max; i += 2) {
Fred Drakeb91a36b2002-06-27 19:40:48 +0000376 PyObject *n = string_intern(self, (XML_Char *) atts[i]);
Fred Drake85d835f2001-02-08 15:39:08 +0000377 PyObject *v;
378 if (n == NULL) {
379 flag_error(self);
380 Py_DECREF(container);
381 return;
382 }
Guido van Rossum4ca94712007-07-23 17:42:32 +0000383 v = conv_string_to_unicode((XML_Char *) atts[i+1]);
Fred Drake85d835f2001-02-08 15:39:08 +0000384 if (v == NULL) {
385 flag_error(self);
386 Py_DECREF(container);
387 Py_DECREF(n);
388 return;
389 }
390 if (self->ordered_attributes) {
391 PyList_SET_ITEM(container, i, n);
392 PyList_SET_ITEM(container, i+1, v);
393 }
394 else if (PyDict_SetItem(container, n, v)) {
395 flag_error(self);
396 Py_DECREF(n);
397 Py_DECREF(v);
398 return;
399 }
400 else {
401 Py_DECREF(n);
402 Py_DECREF(v);
403 }
404 }
Neal Norwitz484d9a42005-09-30 04:46:49 +0000405 args = string_intern(self, name);
406 if (args != NULL)
407 args = Py_BuildValue("(NN)", args, container);
Fred Drake85d835f2001-02-08 15:39:08 +0000408 if (args == NULL) {
409 Py_DECREF(container);
410 return;
411 }
412 /* Container is now a borrowed reference; ignore it. */
Fred Drakebd6101c2001-02-14 18:29:45 +0000413 self->in_callback = 1;
Antoine Pitrou0ddbf472014-10-08 20:00:09 +0200414 rv = call_with_frame("StartElement", __LINE__,
Fred Drake39689c52004-08-13 03:12:57 +0000415 self->handlers[StartElement], args, self);
Fred Drakebd6101c2001-02-14 18:29:45 +0000416 self->in_callback = 0;
417 Py_DECREF(args);
Fred Drake85d835f2001-02-08 15:39:08 +0000418 if (rv == NULL) {
419 flag_error(self);
420 return;
Fred Drakebd6101c2001-02-14 18:29:45 +0000421 }
Fred Drake85d835f2001-02-08 15:39:08 +0000422 Py_DECREF(rv);
423 }
424}
425
426#define RC_HANDLER(RC, NAME, PARAMS, INIT, PARAM_FORMAT, CONVERSION, \
427 RETURN, GETUSERDATA) \
428static RC \
429my_##NAME##Handler PARAMS {\
430 xmlparseobject *self = GETUSERDATA ; \
431 PyObject *args = NULL; \
432 PyObject *rv = NULL; \
433 INIT \
434\
Fred Drake71b63ff2002-06-28 22:29:01 +0000435 if (have_handler(self, NAME)) { \
Victor Stinner9e09c262013-07-18 23:17:01 +0200436 if (PyErr_Occurred()) \
437 return RETURN; \
Fred Drake2a3d7db2002-06-28 22:56:48 +0000438 if (flush_character_buffer(self) < 0) \
439 return RETURN; \
Fred Drake85d835f2001-02-08 15:39:08 +0000440 args = Py_BuildValue PARAM_FORMAT ;\
Martin v. Löwis1d7c55f2001-11-10 13:57:55 +0000441 if (!args) { flag_error(self); return RETURN;} \
Fred Drakebd6101c2001-02-14 18:29:45 +0000442 self->in_callback = 1; \
Antoine Pitrou0ddbf472014-10-08 20:00:09 +0200443 rv = call_with_frame(#NAME,__LINE__, \
Fred Drake39689c52004-08-13 03:12:57 +0000444 self->handlers[NAME], args, self); \
Fred Drakebd6101c2001-02-14 18:29:45 +0000445 self->in_callback = 0; \
Fred Drake85d835f2001-02-08 15:39:08 +0000446 Py_DECREF(args); \
447 if (rv == NULL) { \
448 flag_error(self); \
449 return RETURN; \
450 } \
451 CONVERSION \
452 Py_DECREF(rv); \
453 } \
454 return RETURN; \
455}
456
Fred Drake6f987622000-08-25 18:03:30 +0000457#define VOID_HANDLER(NAME, PARAMS, PARAM_FORMAT) \
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000458 RC_HANDLER(void, NAME, PARAMS, ;, PARAM_FORMAT, ;, ;,\
459 (xmlparseobject *)userData)
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000460
Fred Drake6f987622000-08-25 18:03:30 +0000461#define INT_HANDLER(NAME, PARAMS, PARAM_FORMAT)\
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000462 RC_HANDLER(int, NAME, PARAMS, int rc=0;, PARAM_FORMAT, \
463 rc = PyLong_AsLong(rv);, rc, \
464 (xmlparseobject *)userData)
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000465
Fred Drake71b63ff2002-06-28 22:29:01 +0000466VOID_HANDLER(EndElement,
467 (void *userData, const XML_Char *name),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000468 ("(N)", string_intern(self, name)))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000469
Fred Drake6f987622000-08-25 18:03:30 +0000470VOID_HANDLER(ProcessingInstruction,
Fred Drake71b63ff2002-06-28 22:29:01 +0000471 (void *userData,
472 const XML_Char *target,
Fred Drake85d835f2001-02-08 15:39:08 +0000473 const XML_Char *data),
Guido van Rossum4ca94712007-07-23 17:42:32 +0000474 ("(NO&)", string_intern(self, target), conv_string_to_unicode ,data))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000475
Fred Drake6f987622000-08-25 18:03:30 +0000476VOID_HANDLER(UnparsedEntityDecl,
Fred Drake71b63ff2002-06-28 22:29:01 +0000477 (void *userData,
Fred Drake85d835f2001-02-08 15:39:08 +0000478 const XML_Char *entityName,
479 const XML_Char *base,
480 const XML_Char *systemId,
481 const XML_Char *publicId,
482 const XML_Char *notationName),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000483 ("(NNNNN)",
Fred Drake71b63ff2002-06-28 22:29:01 +0000484 string_intern(self, entityName), string_intern(self, base),
485 string_intern(self, systemId), string_intern(self, publicId),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000486 string_intern(self, notationName)))
Fred Drake85d835f2001-02-08 15:39:08 +0000487
Fred Drake85d835f2001-02-08 15:39:08 +0000488VOID_HANDLER(EntityDecl,
489 (void *userData,
490 const XML_Char *entityName,
491 int is_parameter_entity,
492 const XML_Char *value,
493 int value_length,
494 const XML_Char *base,
495 const XML_Char *systemId,
496 const XML_Char *publicId,
497 const XML_Char *notationName),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000498 ("NiNNNNN",
499 string_intern(self, entityName), is_parameter_entity,
Guido van Rossum4ca94712007-07-23 17:42:32 +0000500 (conv_string_len_to_unicode(value, value_length)),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000501 string_intern(self, base), string_intern(self, systemId),
502 string_intern(self, publicId),
503 string_intern(self, notationName)))
Fred Drake85d835f2001-02-08 15:39:08 +0000504
505VOID_HANDLER(XmlDecl,
506 (void *userData,
507 const XML_Char *version,
508 const XML_Char *encoding,
509 int standalone),
510 ("(O&O&i)",
Guido van Rossum4ca94712007-07-23 17:42:32 +0000511 conv_string_to_unicode ,version, conv_string_to_unicode ,encoding,
Fred Drake85d835f2001-02-08 15:39:08 +0000512 standalone))
513
514static PyObject *
515conv_content_model(XML_Content * const model,
Fred Drakeb91a36b2002-06-27 19:40:48 +0000516 PyObject *(*conv_string)(const XML_Char *))
Fred Drake85d835f2001-02-08 15:39:08 +0000517{
518 PyObject *result = NULL;
519 PyObject *children = PyTuple_New(model->numchildren);
520 int i;
521
522 if (children != NULL) {
Tim Peters9544fc52001-07-28 09:36:36 +0000523 assert(model->numchildren < INT_MAX);
524 for (i = 0; i < (int)model->numchildren; ++i) {
Fred Drake85d835f2001-02-08 15:39:08 +0000525 PyObject *child = conv_content_model(&model->children[i],
526 conv_string);
527 if (child == NULL) {
528 Py_XDECREF(children);
529 return NULL;
530 }
531 PyTuple_SET_ITEM(children, i, child);
532 }
533 result = Py_BuildValue("(iiO&N)",
534 model->type, model->quant,
535 conv_string,model->name, children);
536 }
537 return result;
538}
539
Fred Drake06dd8cf2003-02-02 03:54:17 +0000540static void
541my_ElementDeclHandler(void *userData,
542 const XML_Char *name,
543 XML_Content *model)
Fred Drake85d835f2001-02-08 15:39:08 +0000544{
Fred Drake06dd8cf2003-02-02 03:54:17 +0000545 xmlparseobject *self = (xmlparseobject *)userData;
546 PyObject *args = NULL;
Fred Drake85d835f2001-02-08 15:39:08 +0000547
Fred Drake06dd8cf2003-02-02 03:54:17 +0000548 if (have_handler(self, ElementDecl)) {
549 PyObject *rv = NULL;
550 PyObject *modelobj, *nameobj;
551
Victor Stinner9e09c262013-07-18 23:17:01 +0200552 if (PyErr_Occurred())
553 return;
554
Fred Drake06dd8cf2003-02-02 03:54:17 +0000555 if (flush_character_buffer(self) < 0)
556 goto finally;
Guido van Rossum4ca94712007-07-23 17:42:32 +0000557 modelobj = conv_content_model(model, (conv_string_to_unicode));
Fred Drake06dd8cf2003-02-02 03:54:17 +0000558 if (modelobj == NULL) {
559 flag_error(self);
560 goto finally;
561 }
562 nameobj = string_intern(self, name);
563 if (nameobj == NULL) {
564 Py_DECREF(modelobj);
565 flag_error(self);
566 goto finally;
567 }
Michael W. Hudson0bb84542004-08-03 11:31:31 +0000568 args = Py_BuildValue("NN", nameobj, modelobj);
Fred Drake06dd8cf2003-02-02 03:54:17 +0000569 if (args == NULL) {
570 Py_DECREF(modelobj);
571 flag_error(self);
572 goto finally;
573 }
574 self->in_callback = 1;
Antoine Pitrou0ddbf472014-10-08 20:00:09 +0200575 rv = call_with_frame("ElementDecl", __LINE__,
Fred Drake39689c52004-08-13 03:12:57 +0000576 self->handlers[ElementDecl], args, self);
Fred Drake06dd8cf2003-02-02 03:54:17 +0000577 self->in_callback = 0;
578 if (rv == NULL) {
579 flag_error(self);
580 goto finally;
581 }
582 Py_DECREF(rv);
583 }
584 finally:
585 Py_XDECREF(args);
586 XML_FreeContentModel(self->itself, model);
587 return;
588}
Fred Drake85d835f2001-02-08 15:39:08 +0000589
590VOID_HANDLER(AttlistDecl,
591 (void *userData,
592 const XML_Char *elname,
593 const XML_Char *attname,
594 const XML_Char *att_type,
595 const XML_Char *dflt,
596 int isrequired),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000597 ("(NNO&O&i)",
598 string_intern(self, elname), string_intern(self, attname),
Guido van Rossum4ca94712007-07-23 17:42:32 +0000599 conv_string_to_unicode ,att_type, conv_string_to_unicode ,dflt,
Fred Drake85d835f2001-02-08 15:39:08 +0000600 isrequired))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000601
Martin v. Löwisc847f402003-01-21 11:09:21 +0000602#if XML_COMBINED_VERSION >= 19504
Martin v. Löwis069dde22003-01-21 10:58:18 +0000603VOID_HANDLER(SkippedEntity,
604 (void *userData,
605 const XML_Char *entityName,
606 int is_parameter_entity),
607 ("Ni",
608 string_intern(self, entityName), is_parameter_entity))
Martin v. Löwisc847f402003-01-21 11:09:21 +0000609#endif
Martin v. Löwis069dde22003-01-21 10:58:18 +0000610
Fred Drake71b63ff2002-06-28 22:29:01 +0000611VOID_HANDLER(NotationDecl,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000612 (void *userData,
613 const XML_Char *notationName,
614 const XML_Char *base,
615 const XML_Char *systemId,
616 const XML_Char *publicId),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000617 ("(NNNN)",
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000618 string_intern(self, notationName), string_intern(self, base),
619 string_intern(self, systemId), string_intern(self, publicId)))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000620
Fred Drake6f987622000-08-25 18:03:30 +0000621VOID_HANDLER(StartNamespaceDecl,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000622 (void *userData,
623 const XML_Char *prefix,
624 const XML_Char *uri),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000625 ("(NN)",
626 string_intern(self, prefix), string_intern(self, uri)))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000627
Fred Drake6f987622000-08-25 18:03:30 +0000628VOID_HANDLER(EndNamespaceDecl,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000629 (void *userData,
630 const XML_Char *prefix),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000631 ("(N)", string_intern(self, prefix)))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000632
Fred Drake6f987622000-08-25 18:03:30 +0000633VOID_HANDLER(Comment,
Fred Drakeb91a36b2002-06-27 19:40:48 +0000634 (void *userData, const XML_Char *data),
Guido van Rossum4ca94712007-07-23 17:42:32 +0000635 ("(O&)", conv_string_to_unicode ,data))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000636
Fred Drake6f987622000-08-25 18:03:30 +0000637VOID_HANDLER(StartCdataSection,
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000638 (void *userData),
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000639 ("()"))
Fred Drake71b63ff2002-06-28 22:29:01 +0000640
Fred Drake6f987622000-08-25 18:03:30 +0000641VOID_HANDLER(EndCdataSection,
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000642 (void *userData),
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000643 ("()"))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000644
Fred Drake6f987622000-08-25 18:03:30 +0000645VOID_HANDLER(Default,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000646 (void *userData, const XML_Char *s, int len),
647 ("(N)", (conv_string_len_to_unicode(s,len))))
Andrew M. Kuchlingbeba0562000-06-27 00:33:30 +0000648
Fred Drake6f987622000-08-25 18:03:30 +0000649VOID_HANDLER(DefaultHandlerExpand,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000650 (void *userData, const XML_Char *s, int len),
651 ("(N)", (conv_string_len_to_unicode(s,len))))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000652
Fred Drake71b63ff2002-06-28 22:29:01 +0000653INT_HANDLER(NotStandalone,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000654 (void *userData),
655 ("()"))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000656
Fred Drake6f987622000-08-25 18:03:30 +0000657RC_HANDLER(int, ExternalEntityRef,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000658 (XML_Parser parser,
659 const XML_Char *context,
660 const XML_Char *base,
661 const XML_Char *systemId,
662 const XML_Char *publicId),
663 int rc=0;,
Fred Drakeb91a36b2002-06-27 19:40:48 +0000664 ("(O&NNN)",
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000665 conv_string_to_unicode ,context, string_intern(self, base),
666 string_intern(self, systemId), string_intern(self, publicId)),
667 rc = PyLong_AsLong(rv);, rc,
668 XML_GetUserData(parser))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000669
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000670/* XXX UnknownEncodingHandler */
671
Fred Drake85d835f2001-02-08 15:39:08 +0000672VOID_HANDLER(StartDoctypeDecl,
673 (void *userData, const XML_Char *doctypeName,
674 const XML_Char *sysid, const XML_Char *pubid,
675 int has_internal_subset),
Fred Drakeb91a36b2002-06-27 19:40:48 +0000676 ("(NNNi)", string_intern(self, doctypeName),
677 string_intern(self, sysid), string_intern(self, pubid),
Fred Drake85d835f2001-02-08 15:39:08 +0000678 has_internal_subset))
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000679
680VOID_HANDLER(EndDoctypeDecl, (void *userData), ("()"))
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000681
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000682/* ---------------------------------------------------------------- */
Brett Cannond0aeda82014-08-22 14:23:20 -0400683/*[clinic input]
684class pyexpat.xmlparser "xmlparseobject *" "&Xmlparsetype"
685[clinic start generated code]*/
686/*[clinic end generated code: output=da39a3ee5e6b4b0d input=2393162385232e1c]*/
687
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000688
Fred Drake71b63ff2002-06-28 22:29:01 +0000689static PyObject *
690get_parse_result(xmlparseobject *self, int rv)
691{
692 if (PyErr_Occurred()) {
693 return NULL;
694 }
695 if (rv == 0) {
Martin v. Löwis069dde22003-01-21 10:58:18 +0000696 return set_error(self, XML_GetErrorCode(self->itself));
Fred Drake71b63ff2002-06-28 22:29:01 +0000697 }
Fred Drake2a3d7db2002-06-28 22:56:48 +0000698 if (flush_character_buffer(self) < 0) {
699 return NULL;
700 }
Christian Heimes217cfd12007-12-02 14:31:20 +0000701 return PyLong_FromLong(rv);
Fred Drake71b63ff2002-06-28 22:29:01 +0000702}
703
Serhiy Storchaka43536e92013-02-04 18:26:15 +0200704#define MAX_CHUNK_SIZE (1 << 20)
705
Brett Cannond0aeda82014-08-22 14:23:20 -0400706/*[clinic input]
707pyexpat.xmlparser.Parse
708
709 data: object
Serhiy Storchaka8b2e8b62015-05-30 11:30:39 +0300710 isfinal: int(c_default="0") = False
Brett Cannond0aeda82014-08-22 14:23:20 -0400711 /
712
713Parse XML data.
714
715`isfinal' should be true at end of input.
716[clinic start generated code]*/
717
Brett Cannond0aeda82014-08-22 14:23:20 -0400718static PyObject *
Larry Hastings89964c42015-04-14 18:07:59 -0400719pyexpat_xmlparser_Parse_impl(xmlparseobject *self, PyObject *data,
Serhiy Storchaka8b2e8b62015-05-30 11:30:39 +0300720 int isfinal)
721/*[clinic end generated code: output=f4db843dd1f4ed4b input=199d9e8e92ebbb4b]*/
Brett Cannond0aeda82014-08-22 14:23:20 -0400722{
Serhiy Storchaka43536e92013-02-04 18:26:15 +0200723 const char *s;
724 Py_ssize_t slen;
725 Py_buffer view;
726 int rc;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000727
Serhiy Storchaka43536e92013-02-04 18:26:15 +0200728 if (PyUnicode_Check(data)) {
Serhiy Storchaka43536e92013-02-04 18:26:15 +0200729 view.buf = NULL;
Serhiy Storchaka36b365c2013-02-04 18:28:01 +0200730 s = PyUnicode_AsUTF8AndSize(data, &slen);
731 if (s == NULL)
732 return NULL;
Serhiy Storchaka43536e92013-02-04 18:26:15 +0200733 /* Explicitly set UTF-8 encoding. Return code ignored. */
734 (void)XML_SetEncoding(self->itself, "utf-8");
735 }
736 else {
737 if (PyObject_GetBuffer(data, &view, PyBUF_SIMPLE) < 0)
738 return NULL;
739 s = view.buf;
740 slen = view.len;
741 }
742
743 while (slen > MAX_CHUNK_SIZE) {
744 rc = XML_Parse(self->itself, s, MAX_CHUNK_SIZE, 0);
745 if (!rc)
746 goto done;
747 s += MAX_CHUNK_SIZE;
748 slen -= MAX_CHUNK_SIZE;
749 }
Serhiy Storchakafad85aa2015-11-07 15:42:38 +0200750 Py_BUILD_ASSERT(MAX_CHUNK_SIZE <= INT_MAX);
751 assert(slen <= INT_MAX);
Serhiy Storchaka8b2e8b62015-05-30 11:30:39 +0300752 rc = XML_Parse(self->itself, s, (int)slen, isfinal);
Serhiy Storchaka43536e92013-02-04 18:26:15 +0200753
754done:
755 if (view.buf != NULL)
756 PyBuffer_Release(&view);
757 return get_parse_result(self, rc);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000758}
759
Fred Drakeca1f4262000-09-21 20:10:23 +0000760/* File reading copied from cPickle */
761
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000762#define BUF_SIZE 2048
763
Fred Drake0582df92000-07-12 04:49:00 +0000764static int
765readinst(char *buf, int buf_size, PyObject *meth)
766{
Victor Stinner95f1dfc2011-01-10 23:00:36 +0000767 PyObject *str;
768 Py_ssize_t len;
Serhiy Storchakaef1585e2015-12-25 20:01:53 +0200769 const char *ptr;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000770
Victor Stinner95f1dfc2011-01-10 23:00:36 +0000771 str = PyObject_CallFunction(meth, "n", buf_size);
Martin v. Löwis9171f022004-10-13 19:50:11 +0000772 if (str == NULL)
Victor Stinner95f1dfc2011-01-10 23:00:36 +0000773 goto error;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000774
Christian Heimes72b710a2008-05-26 13:28:38 +0000775 if (PyBytes_Check(str))
776 ptr = PyBytes_AS_STRING(str);
Christian Heimes9c4756e2008-05-26 13:22:05 +0000777 else if (PyByteArray_Check(str))
778 ptr = PyByteArray_AS_STRING(str);
Guido van Rossum98297ee2007-11-06 21:34:58 +0000779 else {
Fred Drake71b63ff2002-06-28 22:29:01 +0000780 PyErr_Format(PyExc_TypeError,
Guido van Rossum4ca94712007-07-23 17:42:32 +0000781 "read() did not return a bytes object (type=%.400s)",
Christian Heimes90aa7642007-12-19 02:45:37 +0000782 Py_TYPE(str)->tp_name);
Victor Stinner95f1dfc2011-01-10 23:00:36 +0000783 goto error;
Fred Drake0582df92000-07-12 04:49:00 +0000784 }
Christian Heimes90aa7642007-12-19 02:45:37 +0000785 len = Py_SIZE(str);
Fred Drake0582df92000-07-12 04:49:00 +0000786 if (len > buf_size) {
787 PyErr_Format(PyExc_ValueError,
788 "read() returned too much data: "
Victor Stinner9d6f9362011-01-04 22:00:04 +0000789 "%i bytes requested, %zd returned",
Fred Drake0582df92000-07-12 04:49:00 +0000790 buf_size, len);
Victor Stinner95f1dfc2011-01-10 23:00:36 +0000791 goto error;
Fred Drake0582df92000-07-12 04:49:00 +0000792 }
Guido van Rossum98297ee2007-11-06 21:34:58 +0000793 memcpy(buf, ptr, len);
Victor Stinner95f1dfc2011-01-10 23:00:36 +0000794 Py_DECREF(str);
795 /* len <= buf_size <= INT_MAX */
Victor Stinner0fcab4a2011-01-04 12:59:15 +0000796 return (int)len;
Victor Stinner95f1dfc2011-01-10 23:00:36 +0000797
798error:
799 Py_XDECREF(str);
800 return -1;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000801}
802
Brett Cannond0aeda82014-08-22 14:23:20 -0400803/*[clinic input]
804pyexpat.xmlparser.ParseFile
805
806 file: object
807 /
808
809Parse XML data from file-like object.
810[clinic start generated code]*/
811
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000812static PyObject *
Brett Cannond0aeda82014-08-22 14:23:20 -0400813pyexpat_xmlparser_ParseFile(xmlparseobject *self, PyObject *file)
Serhiy Storchaka1009bf12015-04-03 23:53:51 +0300814/*[clinic end generated code: output=2adc6a13100cc42b input=fbb5a12b6038d735]*/
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000815{
Fred Drake0582df92000-07-12 04:49:00 +0000816 int rv = 1;
Fred Drake0582df92000-07-12 04:49:00 +0000817 PyObject *readmethod = NULL;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +0200818 _Py_IDENTIFIER(read);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000819
Brett Cannond0aeda82014-08-22 14:23:20 -0400820 readmethod = _PyObject_GetAttrId(file, &PyId_read);
Benjamin Peterson4e7f2852010-08-08 16:54:58 +0000821 if (readmethod == NULL) {
Benjamin Peterson4e7f2852010-08-08 16:54:58 +0000822 PyErr_SetString(PyExc_TypeError,
823 "argument must have 'read' attribute");
824 return NULL;
Fred Drake0582df92000-07-12 04:49:00 +0000825 }
826 for (;;) {
827 int bytes_read;
828 void *buf = XML_GetBuffer(self->itself, BUF_SIZE);
Fred Drake7b6caff2003-07-21 17:05:56 +0000829 if (buf == NULL) {
Fred Drakef239c6d2003-07-21 17:22:43 +0000830 Py_XDECREF(readmethod);
Ned Deilye7d532f2014-03-27 16:39:58 -0700831 return get_parse_result(self, 0);
Fred Drake7b6caff2003-07-21 17:05:56 +0000832 }
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000833
Benjamin Peterson4e7f2852010-08-08 16:54:58 +0000834 bytes_read = readinst(buf, BUF_SIZE, readmethod);
835 if (bytes_read < 0) {
836 Py_DECREF(readmethod);
837 return NULL;
Fred Drake0582df92000-07-12 04:49:00 +0000838 }
839 rv = XML_ParseBuffer(self->itself, bytes_read, bytes_read == 0);
Fred Drake7b6caff2003-07-21 17:05:56 +0000840 if (PyErr_Occurred()) {
841 Py_XDECREF(readmethod);
Fred Drake0582df92000-07-12 04:49:00 +0000842 return NULL;
Fred Drake7b6caff2003-07-21 17:05:56 +0000843 }
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000844
Fred Drake0582df92000-07-12 04:49:00 +0000845 if (!rv || bytes_read == 0)
846 break;
847 }
Fred Drake7b6caff2003-07-21 17:05:56 +0000848 Py_XDECREF(readmethod);
Fred Drake71b63ff2002-06-28 22:29:01 +0000849 return get_parse_result(self, rv);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000850}
851
Brett Cannond0aeda82014-08-22 14:23:20 -0400852/*[clinic input]
853pyexpat.xmlparser.SetBase
854
855 base: str
856 /
857
858Set the base URL for the parser.
859[clinic start generated code]*/
860
Brett Cannond0aeda82014-08-22 14:23:20 -0400861static PyObject *
862pyexpat_xmlparser_SetBase_impl(xmlparseobject *self, const char *base)
Serhiy Storchaka1009bf12015-04-03 23:53:51 +0300863/*[clinic end generated code: output=c212ddceb607b539 input=c684e5de895ee1a8]*/
Brett Cannond0aeda82014-08-22 14:23:20 -0400864{
Fred Drake0582df92000-07-12 04:49:00 +0000865 if (!XML_SetBase(self->itself, base)) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000866 return PyErr_NoMemory();
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000867 }
Brett Cannond0aeda82014-08-22 14:23:20 -0400868 Py_RETURN_NONE;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000869}
870
Brett Cannond0aeda82014-08-22 14:23:20 -0400871/*[clinic input]
872pyexpat.xmlparser.GetBase
873
874Return base URL string for the parser.
875[clinic start generated code]*/
876
Brett Cannond0aeda82014-08-22 14:23:20 -0400877static PyObject *
878pyexpat_xmlparser_GetBase_impl(xmlparseobject *self)
Serhiy Storchaka1009bf12015-04-03 23:53:51 +0300879/*[clinic end generated code: output=2886cb21f9a8739a input=918d71c38009620e]*/
Fred Drake0582df92000-07-12 04:49:00 +0000880{
Fred Drake0582df92000-07-12 04:49:00 +0000881 return Py_BuildValue("z", XML_GetBase(self->itself));
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +0000882}
883
Brett Cannond0aeda82014-08-22 14:23:20 -0400884/*[clinic input]
885pyexpat.xmlparser.GetInputContext
886
887Return the untranslated text of the input that caused the current event.
888
889If the event was generated by a large amount of text (such as a start tag
890for an element with many attributes), not all of the text may be available.
891[clinic start generated code]*/
892
Brett Cannond0aeda82014-08-22 14:23:20 -0400893static PyObject *
894pyexpat_xmlparser_GetInputContext_impl(xmlparseobject *self)
Serhiy Storchaka1009bf12015-04-03 23:53:51 +0300895/*[clinic end generated code: output=a88026d683fc22cc input=034df8712db68379]*/
Fred Drakebd6101c2001-02-14 18:29:45 +0000896{
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000897 if (self->in_callback) {
898 int offset, size;
899 const char *buffer
900 = XML_GetInputContext(self->itself, &offset, &size);
Fred Drakebd6101c2001-02-14 18:29:45 +0000901
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000902 if (buffer != NULL)
Christian Heimes72b710a2008-05-26 13:28:38 +0000903 return PyBytes_FromStringAndSize(buffer + offset,
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000904 size - offset);
905 else
906 Py_RETURN_NONE;
Fred Drakebd6101c2001-02-14 18:29:45 +0000907 }
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000908 else
909 Py_RETURN_NONE;
Fred Drakebd6101c2001-02-14 18:29:45 +0000910}
Fred Drakebd6101c2001-02-14 18:29:45 +0000911
Brett Cannond0aeda82014-08-22 14:23:20 -0400912/*[clinic input]
913pyexpat.xmlparser.ExternalEntityParserCreate
914
Larry Hastingsdbfdc382015-05-04 06:59:46 -0700915 context: str(accept={str, NoneType})
Brett Cannond0aeda82014-08-22 14:23:20 -0400916 encoding: str = NULL
917 /
918
919Create a parser for parsing an external entity based on the information passed to the ExternalEntityRefHandler.
920[clinic start generated code]*/
921
Brett Cannond0aeda82014-08-22 14:23:20 -0400922static PyObject *
Larry Hastings89964c42015-04-14 18:07:59 -0400923pyexpat_xmlparser_ExternalEntityParserCreate_impl(xmlparseobject *self,
924 const char *context,
925 const char *encoding)
Larry Hastingsdbfdc382015-05-04 06:59:46 -0700926/*[clinic end generated code: output=535cda9d7a0fbcd6 input=b906714cc122c322]*/
Brett Cannond0aeda82014-08-22 14:23:20 -0400927{
Lars Gustäbel4a30a072000-09-24 20:50:52 +0000928 xmlparseobject *new_parser;
929 int i;
930
Martin v. Löwis894258c2001-09-23 10:20:10 +0000931 new_parser = PyObject_GC_New(xmlparseobject, &Xmlparsetype);
Fred Drake85d835f2001-02-08 15:39:08 +0000932 if (new_parser == NULL)
933 return NULL;
Fred Drake2a3d7db2002-06-28 22:56:48 +0000934 new_parser->buffer_size = self->buffer_size;
935 new_parser->buffer_used = 0;
Victor Stinnerb4ba9862010-09-10 22:25:19 +0000936 new_parser->buffer = NULL;
Fred Drake85d835f2001-02-08 15:39:08 +0000937 new_parser->ordered_attributes = self->ordered_attributes;
938 new_parser->specified_attributes = self->specified_attributes;
Fred Drakebd6101c2001-02-14 18:29:45 +0000939 new_parser->in_callback = 0;
Martin v. Löwis069dde22003-01-21 10:58:18 +0000940 new_parser->ns_prefixes = self->ns_prefixes;
Lars Gustäbel4a30a072000-09-24 20:50:52 +0000941 new_parser->itself = XML_ExternalEntityParserCreate(self->itself, context,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000942 encoding);
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000943 new_parser->handlers = 0;
Fred Drakeb91a36b2002-06-27 19:40:48 +0000944 new_parser->intern = self->intern;
945 Py_XINCREF(new_parser->intern);
Martin v. Löwis894258c2001-09-23 10:20:10 +0000946 PyObject_GC_Track(new_parser);
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000947
Victor Stinnerb4ba9862010-09-10 22:25:19 +0000948 if (self->buffer != NULL) {
Victor Stinnerb6404912013-07-07 16:21:41 +0200949 new_parser->buffer = PyMem_Malloc(new_parser->buffer_size);
Victor Stinnerb4ba9862010-09-10 22:25:19 +0000950 if (new_parser->buffer == NULL) {
951 Py_DECREF(new_parser);
952 return PyErr_NoMemory();
953 }
954 }
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000955 if (!new_parser->itself) {
Fred Drake85d835f2001-02-08 15:39:08 +0000956 Py_DECREF(new_parser);
957 return PyErr_NoMemory();
Lars Gustäbel4a30a072000-09-24 20:50:52 +0000958 }
959
960 XML_SetUserData(new_parser->itself, (void *)new_parser);
961
962 /* allocate and clear handlers first */
Fred Drake2a3d7db2002-06-28 22:56:48 +0000963 for (i = 0; handler_info[i].name != NULL; i++)
Fred Drake85d835f2001-02-08 15:39:08 +0000964 /* do nothing */;
Lars Gustäbel4a30a072000-09-24 20:50:52 +0000965
Serhiy Storchaka1a1ff292015-02-16 13:28:22 +0200966 new_parser->handlers = PyMem_New(PyObject *, i);
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000967 if (!new_parser->handlers) {
Fred Drake85d835f2001-02-08 15:39:08 +0000968 Py_DECREF(new_parser);
969 return PyErr_NoMemory();
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000970 }
Martin v. Löwis5b68ce32001-10-21 08:53:52 +0000971 clear_handlers(new_parser, 1);
Lars Gustäbel4a30a072000-09-24 20:50:52 +0000972
973 /* then copy handlers from self */
974 for (i = 0; handler_info[i].name != NULL; i++) {
Fred Drake71b63ff2002-06-28 22:29:01 +0000975 PyObject *handler = self->handlers[i];
976 if (handler != NULL) {
977 Py_INCREF(handler);
978 new_parser->handlers[i] = handler;
979 handler_info[i].setter(new_parser->itself,
Fred Drake85d835f2001-02-08 15:39:08 +0000980 handler_info[i].handler);
981 }
Lars Gustäbel4a30a072000-09-24 20:50:52 +0000982 }
Fred Drake71b63ff2002-06-28 22:29:01 +0000983 return (PyObject *)new_parser;
Lars Gustäbel4a30a072000-09-24 20:50:52 +0000984}
985
Brett Cannond0aeda82014-08-22 14:23:20 -0400986/*[clinic input]
987pyexpat.xmlparser.SetParamEntityParsing
Martin v. Löwis0078f6c2001-01-21 10:18:10 +0000988
Brett Cannond0aeda82014-08-22 14:23:20 -0400989 flag: int
990 /
991
992Controls parsing of parameter entities (including the external DTD subset).
993
994Possible flag values are XML_PARAM_ENTITY_PARSING_NEVER,
995XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE and
996XML_PARAM_ENTITY_PARSING_ALWAYS. Returns true if setting the flag
997was successful.
998[clinic start generated code]*/
999
Brett Cannond0aeda82014-08-22 14:23:20 -04001000static PyObject *
1001pyexpat_xmlparser_SetParamEntityParsing_impl(xmlparseobject *self, int flag)
Serhiy Storchaka1009bf12015-04-03 23:53:51 +03001002/*[clinic end generated code: output=18668ee8e760d64c input=8aea19b4b15e9af1]*/
Brett Cannond0aeda82014-08-22 14:23:20 -04001003{
1004 flag = XML_SetParamEntityParsing(self->itself, flag);
Christian Heimes217cfd12007-12-02 14:31:20 +00001005 return PyLong_FromLong(flag);
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001006}
1007
Martin v. Löwisc847f402003-01-21 11:09:21 +00001008
1009#if XML_COMBINED_VERSION >= 19505
Brett Cannond0aeda82014-08-22 14:23:20 -04001010/*[clinic input]
1011pyexpat.xmlparser.UseForeignDTD
1012
1013 flag: bool = True
1014 /
1015
1016Allows the application to provide an artificial external subset if one is not specified as part of the document instance.
1017
1018This readily allows the use of a 'default' document type controlled by the
1019application, while still getting the advantage of providing document type
1020information to the parser. 'flag' defaults to True if not provided.
1021[clinic start generated code]*/
1022
Brett Cannond0aeda82014-08-22 14:23:20 -04001023static PyObject *
1024pyexpat_xmlparser_UseForeignDTD_impl(xmlparseobject *self, int flag)
Serhiy Storchaka1009bf12015-04-03 23:53:51 +03001025/*[clinic end generated code: output=cfaa9aa50bb0f65c input=78144c519d116a6e]*/
Brett Cannond0aeda82014-08-22 14:23:20 -04001026{
Martin v. Löwis069dde22003-01-21 10:58:18 +00001027 enum XML_Error rc;
Brett Cannond0aeda82014-08-22 14:23:20 -04001028
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001029 rc = XML_UseForeignDTD(self->itself, flag ? XML_TRUE : XML_FALSE);
Martin v. Löwis069dde22003-01-21 10:58:18 +00001030 if (rc != XML_ERROR_NONE) {
1031 return set_error(self, rc);
1032 }
1033 Py_INCREF(Py_None);
1034 return Py_None;
1035}
Martin v. Löwisc847f402003-01-21 11:09:21 +00001036#endif
Martin v. Löwis069dde22003-01-21 10:58:18 +00001037
Brett Cannond0aeda82014-08-22 14:23:20 -04001038/*[clinic input]
1039pyexpat.xmlparser.__dir__
1040[clinic start generated code]*/
1041
Brett Cannond0aeda82014-08-22 14:23:20 -04001042static PyObject *
1043pyexpat_xmlparser___dir___impl(xmlparseobject *self)
Serhiy Storchaka1009bf12015-04-03 23:53:51 +03001044/*[clinic end generated code: output=bc22451efb9e4d17 input=76aa455f2a661384]*/
Brett Cannond0aeda82014-08-22 14:23:20 -04001045{
1046#define APPEND(list, str) \
1047 do { \
1048 PyObject *o = PyUnicode_FromString(str); \
1049 if (o != NULL) \
1050 PyList_Append(list, o); \
1051 Py_XDECREF(o); \
1052 } while (0)
1053
1054 int i;
1055 PyObject *rc = PyList_New(0);
1056 if (!rc)
1057 return NULL;
1058 for (i = 0; handler_info[i].name != NULL; i++) {
1059 PyObject *o = get_handler_name(&handler_info[i]);
1060 if (o != NULL)
1061 PyList_Append(rc, o);
1062 Py_XDECREF(o);
1063 }
1064 APPEND(rc, "ErrorCode");
1065 APPEND(rc, "ErrorLineNumber");
1066 APPEND(rc, "ErrorColumnNumber");
1067 APPEND(rc, "ErrorByteIndex");
1068 APPEND(rc, "CurrentLineNumber");
1069 APPEND(rc, "CurrentColumnNumber");
1070 APPEND(rc, "CurrentByteIndex");
1071 APPEND(rc, "buffer_size");
1072 APPEND(rc, "buffer_text");
1073 APPEND(rc, "buffer_used");
1074 APPEND(rc, "namespace_prefixes");
1075 APPEND(rc, "ordered_attributes");
1076 APPEND(rc, "specified_attributes");
1077 APPEND(rc, "intern");
1078
1079#undef APPEND
1080
1081 if (PyErr_Occurred()) {
1082 Py_DECREF(rc);
1083 rc = NULL;
1084 }
1085
1086 return rc;
1087}
Amaury Forgeot d'Arcba4105c2008-07-02 21:41:01 +00001088
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001089static struct PyMethodDef xmlparse_methods[] = {
Brett Cannond0aeda82014-08-22 14:23:20 -04001090 PYEXPAT_XMLPARSER_PARSE_METHODDEF
1091 PYEXPAT_XMLPARSER_PARSEFILE_METHODDEF
1092 PYEXPAT_XMLPARSER_SETBASE_METHODDEF
1093 PYEXPAT_XMLPARSER_GETBASE_METHODDEF
1094 PYEXPAT_XMLPARSER_GETINPUTCONTEXT_METHODDEF
1095 PYEXPAT_XMLPARSER_EXTERNALENTITYPARSERCREATE_METHODDEF
1096 PYEXPAT_XMLPARSER_SETPARAMENTITYPARSING_METHODDEF
Martin v. Löwisc847f402003-01-21 11:09:21 +00001097#if XML_COMBINED_VERSION >= 19505
Brett Cannond0aeda82014-08-22 14:23:20 -04001098 PYEXPAT_XMLPARSER_USEFOREIGNDTD_METHODDEF
Martin v. Löwisc847f402003-01-21 11:09:21 +00001099#endif
Brett Cannond0aeda82014-08-22 14:23:20 -04001100 PYEXPAT_XMLPARSER___DIR___METHODDEF
1101 {NULL, NULL} /* sentinel */
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001102};
1103
1104/* ---------- */
1105
1106
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001107
Fred Drake71b63ff2002-06-28 22:29:01 +00001108/* pyexpat international encoding support.
1109 Make it as simple as possible.
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001110*/
1111
Fred Drake71b63ff2002-06-28 22:29:01 +00001112static int
1113PyUnknownEncodingHandler(void *encodingHandlerData,
1114 const XML_Char *name,
1115 XML_Encoding *info)
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001116{
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001117 static unsigned char template_buffer[256] = {0};
1118 PyObject* u;
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001119 int i;
Martin v. Löwisd63a3b82011-09-28 07:41:54 +02001120 void *data;
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001121 unsigned int kind;
Fred Drake71b63ff2002-06-28 22:29:01 +00001122
Victor Stinner9e09c262013-07-18 23:17:01 +02001123 if (PyErr_Occurred())
1124 return XML_STATUS_ERROR;
1125
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001126 if (template_buffer[1] == 0) {
1127 for (i = 0; i < 256; i++)
1128 template_buffer[i] = i;
Tim Peters63cb99e2001-02-17 18:12:50 +00001129 }
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001130
1131 u = PyUnicode_Decode((char*) template_buffer, 256, name, "replace");
Christian Heimesb5821552013-06-29 20:43:13 +02001132 if (u == NULL || PyUnicode_READY(u)) {
Christian Heimes72172422013-06-29 21:49:27 +02001133 Py_XDECREF(u);
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001134 return XML_STATUS_ERROR;
Christian Heimesb5821552013-06-29 20:43:13 +02001135 }
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001136
1137 if (PyUnicode_GET_LENGTH(u) != 256) {
1138 Py_DECREF(u);
1139 PyErr_SetString(PyExc_ValueError,
1140 "multi-byte encodings are not supported");
1141 return XML_STATUS_ERROR;
1142 }
1143
1144 kind = PyUnicode_KIND(u);
1145 data = PyUnicode_DATA(u);
1146 for (i = 0; i < 256; i++) {
1147 Py_UCS4 ch = PyUnicode_READ(kind, data, i);
1148 if (ch != Py_UNICODE_REPLACEMENT_CHARACTER)
1149 info->map[i] = ch;
1150 else
1151 info->map[i] = -1;
1152 }
1153
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001154 info->data = NULL;
1155 info->convert = NULL;
1156 info->release = NULL;
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001157 Py_DECREF(u);
1158
1159 return XML_STATUS_OK;
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001160}
1161
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001162
1163static PyObject *
Serhiy Storchaka20b39b22014-09-28 11:27:24 +03001164newxmlparseobject(const char *encoding, const char *namespace_separator, PyObject *intern)
Fred Drake0582df92000-07-12 04:49:00 +00001165{
1166 int i;
1167 xmlparseobject *self;
Fred Drake71b63ff2002-06-28 22:29:01 +00001168
Martin v. Löwis894258c2001-09-23 10:20:10 +00001169 self = PyObject_GC_New(xmlparseobject, &Xmlparsetype);
Fred Drake0582df92000-07-12 04:49:00 +00001170 if (self == NULL)
1171 return NULL;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001172
Fred Drake2a3d7db2002-06-28 22:56:48 +00001173 self->buffer = NULL;
1174 self->buffer_size = CHARACTER_DATA_BUFFER_SIZE;
1175 self->buffer_used = 0;
Fred Drake85d835f2001-02-08 15:39:08 +00001176 self->ordered_attributes = 0;
1177 self->specified_attributes = 0;
Fred Drakebd6101c2001-02-14 18:29:45 +00001178 self->in_callback = 0;
Martin v. Löwis069dde22003-01-21 10:58:18 +00001179 self->ns_prefixes = 0;
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001180 self->handlers = NULL;
Victor Stinner54b2d2e2013-07-15 17:15:57 +02001181 self->intern = intern;
1182 Py_XINCREF(self->intern);
1183 PyObject_GC_Track(self);
1184
Christian Heimesfa535f52013-07-07 17:35:11 +02001185 /* namespace_separator is either NULL or contains one char + \0 */
1186 self->itself = XML_ParserCreate_MM(encoding, &ExpatMemoryHandler,
1187 namespace_separator);
Victor Stinner54b2d2e2013-07-15 17:15:57 +02001188 if (self->itself == NULL) {
1189 PyErr_SetString(PyExc_RuntimeError,
1190 "XML_ParserCreate failed");
1191 Py_DECREF(self);
1192 return NULL;
1193 }
Victor Stinner86b95372017-06-15 01:26:16 +02001194#if XML_COMBINED_VERSION >= 20100
1195 /* This feature was added upstream in libexpat 2.1.0. */
Gregory P. Smith8e91cf62012-03-14 14:26:55 -07001196 XML_SetHashSalt(self->itself,
Christian Heimes985ecdc2013-11-20 11:46:18 +01001197 (unsigned long)_Py_HashSecret.expat.hashsalt);
Gregory P. Smith25227712012-03-14 18:10:37 -07001198#endif
Fred Drake0582df92000-07-12 04:49:00 +00001199 XML_SetUserData(self->itself, (void *)self);
Fred Drake7c75bf22002-07-01 14:02:31 +00001200 XML_SetUnknownEncodingHandler(self->itself,
1201 (XML_UnknownEncodingHandler) PyUnknownEncodingHandler, NULL);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001202
Fred Drake2a3d7db2002-06-28 22:56:48 +00001203 for (i = 0; handler_info[i].name != NULL; i++)
Fred Drake0582df92000-07-12 04:49:00 +00001204 /* do nothing */;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001205
Serhiy Storchaka1a1ff292015-02-16 13:28:22 +02001206 self->handlers = PyMem_New(PyObject *, i);
Fred Drake7c75bf22002-07-01 14:02:31 +00001207 if (!self->handlers) {
Fred Drake71b63ff2002-06-28 22:29:01 +00001208 Py_DECREF(self);
1209 return PyErr_NoMemory();
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001210 }
Martin v. Löwis5b68ce32001-10-21 08:53:52 +00001211 clear_handlers(self, 1);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001212
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001213 return (PyObject*)self;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001214}
1215
1216
1217static void
Fred Drake0582df92000-07-12 04:49:00 +00001218xmlparse_dealloc(xmlparseobject *self)
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001219{
Fred Drake0582df92000-07-12 04:49:00 +00001220 int i;
Martin v. Löwis894258c2001-09-23 10:20:10 +00001221 PyObject_GC_UnTrack(self);
Fred Drake85d835f2001-02-08 15:39:08 +00001222 if (self->itself != NULL)
Fred Drake0582df92000-07-12 04:49:00 +00001223 XML_ParserFree(self->itself);
1224 self->itself = NULL;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001225
Fred Drake85d835f2001-02-08 15:39:08 +00001226 if (self->handlers != NULL) {
Serhiy Storchaka1ed017a2015-12-27 15:51:32 +02001227 for (i = 0; handler_info[i].name != NULL; i++)
1228 Py_CLEAR(self->handlers[i]);
Victor Stinnerb6404912013-07-07 16:21:41 +02001229 PyMem_Free(self->handlers);
Fred Drake71b63ff2002-06-28 22:29:01 +00001230 self->handlers = NULL;
Fred Drake0582df92000-07-12 04:49:00 +00001231 }
Fred Drake2a3d7db2002-06-28 22:56:48 +00001232 if (self->buffer != NULL) {
Victor Stinnerb6404912013-07-07 16:21:41 +02001233 PyMem_Free(self->buffer);
Fred Drake2a3d7db2002-06-28 22:56:48 +00001234 self->buffer = NULL;
1235 }
Fred Drakeb91a36b2002-06-27 19:40:48 +00001236 Py_XDECREF(self->intern);
Martin v. Löwis894258c2001-09-23 10:20:10 +00001237 PyObject_GC_Del(self);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001238}
1239
Fred Drake0582df92000-07-12 04:49:00 +00001240static int
Alexander Belopolskye239d232010-12-08 23:31:48 +00001241handlername2int(PyObject *name)
Fred Drake0582df92000-07-12 04:49:00 +00001242{
1243 int i;
Fred Drake71b63ff2002-06-28 22:29:01 +00001244 for (i = 0; handler_info[i].name != NULL; i++) {
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001245 if (_PyUnicode_EqualToASCIIString(name, handler_info[i].name)) {
Fred Drake0582df92000-07-12 04:49:00 +00001246 return i;
1247 }
1248 }
1249 return -1;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001250}
1251
1252static PyObject *
Fred Drake71b63ff2002-06-28 22:29:01 +00001253get_pybool(int istrue)
1254{
1255 PyObject *result = istrue ? Py_True : Py_False;
1256 Py_INCREF(result);
1257 return result;
1258}
1259
1260static PyObject *
Amaury Forgeot d'Arcba4105c2008-07-02 21:41:01 +00001261xmlparse_getattro(xmlparseobject *self, PyObject *nameobj)
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001262{
Victor Stinner9e5bd6c2011-10-01 01:05:40 +02001263 Py_UCS4 first_char;
Amaury Forgeot d'Arcba4105c2008-07-02 21:41:01 +00001264 int handlernum = -1;
1265
Alexander Belopolskye239d232010-12-08 23:31:48 +00001266 if (!PyUnicode_Check(nameobj))
1267 goto generic;
Victor Stinner9e5bd6c2011-10-01 01:05:40 +02001268 if (PyUnicode_READY(nameobj))
1269 return NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001270
Alexander Belopolskye239d232010-12-08 23:31:48 +00001271 handlernum = handlername2int(nameobj);
Fred Drake71b63ff2002-06-28 22:29:01 +00001272
1273 if (handlernum != -1) {
1274 PyObject *result = self->handlers[handlernum];
1275 if (result == NULL)
1276 result = Py_None;
1277 Py_INCREF(result);
1278 return result;
1279 }
Alexander Belopolskye239d232010-12-08 23:31:48 +00001280
Victor Stinner9e5bd6c2011-10-01 01:05:40 +02001281 first_char = PyUnicode_READ_CHAR(nameobj, 0);
1282 if (first_char == 'E') {
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001283 if (_PyUnicode_EqualToASCIIString(nameobj, "ErrorCode"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001284 return PyLong_FromLong((long)
Fred Drake71b63ff2002-06-28 22:29:01 +00001285 XML_GetErrorCode(self->itself));
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001286 if (_PyUnicode_EqualToASCIIString(nameobj, "ErrorLineNumber"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001287 return PyLong_FromLong((long)
Fred Drake71b63ff2002-06-28 22:29:01 +00001288 XML_GetErrorLineNumber(self->itself));
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001289 if (_PyUnicode_EqualToASCIIString(nameobj, "ErrorColumnNumber"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001290 return PyLong_FromLong((long)
Fred Drake71b63ff2002-06-28 22:29:01 +00001291 XML_GetErrorColumnNumber(self->itself));
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001292 if (_PyUnicode_EqualToASCIIString(nameobj, "ErrorByteIndex"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001293 return PyLong_FromLong((long)
Fred Drake71b63ff2002-06-28 22:29:01 +00001294 XML_GetErrorByteIndex(self->itself));
1295 }
Victor Stinner9e5bd6c2011-10-01 01:05:40 +02001296 if (first_char == 'C') {
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001297 if (_PyUnicode_EqualToASCIIString(nameobj, "CurrentLineNumber"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001298 return PyLong_FromLong((long)
Dave Cole3203efb2004-08-26 00:37:31 +00001299 XML_GetCurrentLineNumber(self->itself));
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001300 if (_PyUnicode_EqualToASCIIString(nameobj, "CurrentColumnNumber"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001301 return PyLong_FromLong((long)
Dave Cole3203efb2004-08-26 00:37:31 +00001302 XML_GetCurrentColumnNumber(self->itself));
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001303 if (_PyUnicode_EqualToASCIIString(nameobj, "CurrentByteIndex"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001304 return PyLong_FromLong((long)
Dave Cole3203efb2004-08-26 00:37:31 +00001305 XML_GetCurrentByteIndex(self->itself));
1306 }
Victor Stinner9e5bd6c2011-10-01 01:05:40 +02001307 if (first_char == 'b') {
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001308 if (_PyUnicode_EqualToASCIIString(nameobj, "buffer_size"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001309 return PyLong_FromLong((long) self->buffer_size);
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001310 if (_PyUnicode_EqualToASCIIString(nameobj, "buffer_text"))
Fred Drake2a3d7db2002-06-28 22:56:48 +00001311 return get_pybool(self->buffer != NULL);
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001312 if (_PyUnicode_EqualToASCIIString(nameobj, "buffer_used"))
Christian Heimes217cfd12007-12-02 14:31:20 +00001313 return PyLong_FromLong((long) self->buffer_used);
Fred Drake2a3d7db2002-06-28 22:56:48 +00001314 }
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001315 if (_PyUnicode_EqualToASCIIString(nameobj, "namespace_prefixes"))
Martin v. Löwis069dde22003-01-21 10:58:18 +00001316 return get_pybool(self->ns_prefixes);
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001317 if (_PyUnicode_EqualToASCIIString(nameobj, "ordered_attributes"))
Fred Drake71b63ff2002-06-28 22:29:01 +00001318 return get_pybool(self->ordered_attributes);
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001319 if (_PyUnicode_EqualToASCIIString(nameobj, "specified_attributes"))
Fred Drake71b63ff2002-06-28 22:29:01 +00001320 return get_pybool((long) self->specified_attributes);
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001321 if (_PyUnicode_EqualToASCIIString(nameobj, "intern")) {
Fred Drakeb91a36b2002-06-27 19:40:48 +00001322 if (self->intern == NULL) {
1323 Py_INCREF(Py_None);
1324 return Py_None;
1325 }
1326 else {
1327 Py_INCREF(self->intern);
1328 return self->intern;
1329 }
1330 }
Alexander Belopolskye239d232010-12-08 23:31:48 +00001331 generic:
Amaury Forgeot d'Arcba4105c2008-07-02 21:41:01 +00001332 return PyObject_GenericGetAttr((PyObject*)self, nameobj);
Neal Norwitz8dfc4a92007-08-11 06:39:53 +00001333}
1334
Fred Drake6f987622000-08-25 18:03:30 +00001335static int
Alexander Belopolskye239d232010-12-08 23:31:48 +00001336sethandler(xmlparseobject *self, PyObject *name, PyObject* v)
Fred Drake0582df92000-07-12 04:49:00 +00001337{
1338 int handlernum = handlername2int(name);
Fred Drake71b63ff2002-06-28 22:29:01 +00001339 if (handlernum >= 0) {
1340 xmlhandler c_handler = NULL;
Fred Drake71b63ff2002-06-28 22:29:01 +00001341
Thomas Wouters0e3f5912006-08-11 14:57:12 +00001342 if (v == Py_None) {
1343 /* If this is the character data handler, and a character
1344 data handler is already active, we need to be more
1345 careful. What we can safely do is replace the existing
1346 character data handler callback function with a no-op
1347 function that will refuse to call Python. The downside
1348 is that this doesn't completely remove the character
1349 data handler from the C layer if there's any callback
1350 active, so Expat does a little more work than it
1351 otherwise would, but that's really an odd case. A more
1352 elaborate system of handlers and state could remove the
1353 C handler more effectively. */
1354 if (handlernum == CharacterData && self->in_callback)
1355 c_handler = noop_character_data_handler;
Fred Drake71b63ff2002-06-28 22:29:01 +00001356 v = NULL;
Thomas Wouters0e3f5912006-08-11 14:57:12 +00001357 }
Fred Drake71b63ff2002-06-28 22:29:01 +00001358 else if (v != NULL) {
1359 Py_INCREF(v);
1360 c_handler = handler_info[handlernum].handler;
1361 }
Serhiy Storchakaec397562016-04-06 09:50:03 +03001362 Py_XSETREF(self->handlers[handlernum], v);
Fred Drake71b63ff2002-06-28 22:29:01 +00001363 handler_info[handlernum].setter(self->itself, c_handler);
Fred Drake0582df92000-07-12 04:49:00 +00001364 return 1;
1365 }
1366 return 0;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001367}
1368
1369static int
Alexander Belopolskye239d232010-12-08 23:31:48 +00001370xmlparse_setattro(xmlparseobject *self, PyObject *name, PyObject *v)
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001371{
Fred Drake6f987622000-08-25 18:03:30 +00001372 /* Set attribute 'name' to value 'v'. v==NULL means delete */
Serhiy Storchaka931331a2015-09-07 22:37:02 +03001373 if (!PyUnicode_Check(name)) {
1374 PyErr_Format(PyExc_TypeError,
1375 "attribute name must be string, not '%.200s'",
1376 name->ob_type->tp_name);
1377 return -1;
1378 }
Fred Drake85d835f2001-02-08 15:39:08 +00001379 if (v == NULL) {
Fred Drake6f987622000-08-25 18:03:30 +00001380 PyErr_SetString(PyExc_RuntimeError, "Cannot delete attribute");
1381 return -1;
1382 }
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001383 if (_PyUnicode_EqualToASCIIString(name, "buffer_text")) {
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001384 int b = PyObject_IsTrue(v);
1385 if (b < 0)
1386 return -1;
1387 if (b) {
Fred Drake2a3d7db2002-06-28 22:56:48 +00001388 if (self->buffer == NULL) {
Victor Stinnerb6404912013-07-07 16:21:41 +02001389 self->buffer = PyMem_Malloc(self->buffer_size);
Fred Drake2a3d7db2002-06-28 22:56:48 +00001390 if (self->buffer == NULL) {
1391 PyErr_NoMemory();
1392 return -1;
1393 }
1394 self->buffer_used = 0;
1395 }
1396 }
1397 else if (self->buffer != NULL) {
1398 if (flush_character_buffer(self) < 0)
1399 return -1;
Victor Stinnerb6404912013-07-07 16:21:41 +02001400 PyMem_Free(self->buffer);
Fred Drake2a3d7db2002-06-28 22:56:48 +00001401 self->buffer = NULL;
1402 }
1403 return 0;
1404 }
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001405 if (_PyUnicode_EqualToASCIIString(name, "namespace_prefixes")) {
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001406 int b = PyObject_IsTrue(v);
1407 if (b < 0)
1408 return -1;
1409 self->ns_prefixes = b;
Martin v. Löwis069dde22003-01-21 10:58:18 +00001410 XML_SetReturnNSTriplet(self->itself, self->ns_prefixes);
1411 return 0;
1412 }
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001413 if (_PyUnicode_EqualToASCIIString(name, "ordered_attributes")) {
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001414 int b = PyObject_IsTrue(v);
1415 if (b < 0)
1416 return -1;
1417 self->ordered_attributes = b;
Fred Drake85d835f2001-02-08 15:39:08 +00001418 return 0;
1419 }
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001420 if (_PyUnicode_EqualToASCIIString(name, "specified_attributes")) {
Antoine Pitrou6f430e42012-08-15 23:18:25 +02001421 int b = PyObject_IsTrue(v);
1422 if (b < 0)
1423 return -1;
1424 self->specified_attributes = b;
Fred Drake6f987622000-08-25 18:03:30 +00001425 return 0;
1426 }
Christian Heimes2380ac72008-01-09 00:17:24 +00001427
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001428 if (_PyUnicode_EqualToASCIIString(name, "buffer_size")) {
Christian Heimes2380ac72008-01-09 00:17:24 +00001429 long new_buffer_size;
1430 if (!PyLong_Check(v)) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001431 PyErr_SetString(PyExc_TypeError, "buffer_size must be an integer");
1432 return -1;
Christian Heimes2380ac72008-01-09 00:17:24 +00001433 }
1434
Serhiy Storchakade5f9f42015-09-07 22:51:56 +03001435 new_buffer_size = PyLong_AsLong(v);
1436 if (new_buffer_size <= 0) {
1437 if (!PyErr_Occurred())
1438 PyErr_SetString(PyExc_ValueError, "buffer_size must be greater than zero");
1439 return -1;
1440 }
1441
Christian Heimes2380ac72008-01-09 00:17:24 +00001442 /* trivial case -- no change */
1443 if (new_buffer_size == self->buffer_size) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001444 return 0;
Christian Heimes2380ac72008-01-09 00:17:24 +00001445 }
1446
Christian Heimes2380ac72008-01-09 00:17:24 +00001447 /* check maximum */
1448 if (new_buffer_size > INT_MAX) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001449 char errmsg[100];
1450 sprintf(errmsg, "buffer_size must not be greater than %i", INT_MAX);
1451 PyErr_SetString(PyExc_ValueError, errmsg);
1452 return -1;
Christian Heimes2380ac72008-01-09 00:17:24 +00001453 }
1454
1455 if (self->buffer != NULL) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001456 /* there is already a buffer */
1457 if (self->buffer_used != 0) {
Christian Heimes09994a92013-07-20 22:41:58 +02001458 if (flush_character_buffer(self) < 0) {
1459 return -1;
1460 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001461 }
1462 /* free existing buffer */
Victor Stinnerb6404912013-07-07 16:21:41 +02001463 PyMem_Free(self->buffer);
Christian Heimes2380ac72008-01-09 00:17:24 +00001464 }
Victor Stinnerb6404912013-07-07 16:21:41 +02001465 self->buffer = PyMem_Malloc(new_buffer_size);
Christian Heimes2380ac72008-01-09 00:17:24 +00001466 if (self->buffer == NULL) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001467 PyErr_NoMemory();
1468 return -1;
1469 }
Christian Heimes2380ac72008-01-09 00:17:24 +00001470 self->buffer_size = new_buffer_size;
1471 return 0;
1472 }
1473
Serhiy Storchakaf4934ea2016-11-16 10:17:58 +02001474 if (_PyUnicode_EqualToASCIIString(name, "CharacterDataHandler")) {
Fred Drake2a3d7db2002-06-28 22:56:48 +00001475 /* If we're changing the character data handler, flush all
1476 * cached data with the old handler. Not sure there's a
1477 * "right" thing to do, though, but this probably won't
1478 * happen.
1479 */
1480 if (flush_character_buffer(self) < 0)
1481 return -1;
1482 }
Fred Drake6f987622000-08-25 18:03:30 +00001483 if (sethandler(self, name, v)) {
1484 return 0;
1485 }
Alexander Belopolskye239d232010-12-08 23:31:48 +00001486 PyErr_SetObject(PyExc_AttributeError, name);
Fred Drake6f987622000-08-25 18:03:30 +00001487 return -1;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001488}
1489
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001490static int
1491xmlparse_traverse(xmlparseobject *op, visitproc visit, void *arg)
1492{
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001493 int i;
1494 for (i = 0; handler_info[i].name != NULL; i++)
1495 Py_VISIT(op->handlers[i]);
Fred Drakecde79132001-04-25 16:01:30 +00001496 return 0;
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001497}
1498
1499static int
1500xmlparse_clear(xmlparseobject *op)
1501{
Martin v. Löwis5b68ce32001-10-21 08:53:52 +00001502 clear_handlers(op, 0);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001503 Py_CLEAR(op->intern);
Fred Drakecde79132001-04-25 16:01:30 +00001504 return 0;
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001505}
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001506
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001507PyDoc_STRVAR(Xmlparsetype__doc__, "XML parser");
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001508
1509static PyTypeObject Xmlparsetype = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001510 PyVarObject_HEAD_INIT(NULL, 0)
1511 "pyexpat.xmlparser", /*tp_name*/
Antoine Pitrou23683ef2011-01-04 00:00:31 +00001512 sizeof(xmlparseobject), /*tp_basicsize*/
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001513 0, /*tp_itemsize*/
1514 /* methods */
1515 (destructor)xmlparse_dealloc, /*tp_dealloc*/
1516 (printfunc)0, /*tp_print*/
1517 0, /*tp_getattr*/
Alexander Belopolskye239d232010-12-08 23:31:48 +00001518 0, /*tp_setattr*/
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001519 0, /*tp_reserved*/
1520 (reprfunc)0, /*tp_repr*/
1521 0, /*tp_as_number*/
1522 0, /*tp_as_sequence*/
1523 0, /*tp_as_mapping*/
1524 (hashfunc)0, /*tp_hash*/
1525 (ternaryfunc)0, /*tp_call*/
1526 (reprfunc)0, /*tp_str*/
1527 (getattrofunc)xmlparse_getattro, /* tp_getattro */
Alexander Belopolskye239d232010-12-08 23:31:48 +00001528 (setattrofunc)xmlparse_setattro, /* tp_setattro */
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001529 0, /* tp_as_buffer */
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001530 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /*tp_flags*/
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001531 Xmlparsetype__doc__, /* tp_doc - Documentation string */
1532 (traverseproc)xmlparse_traverse, /* tp_traverse */
1533 (inquiry)xmlparse_clear, /* tp_clear */
1534 0, /* tp_richcompare */
1535 0, /* tp_weaklistoffset */
1536 0, /* tp_iter */
1537 0, /* tp_iternext */
1538 xmlparse_methods, /* tp_methods */
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001539};
1540
1541/* End of code for xmlparser objects */
1542/* -------------------------------------------------------- */
1543
Brett Cannond0aeda82014-08-22 14:23:20 -04001544/*[clinic input]
1545pyexpat.ParserCreate
1546
Larry Hastingsdbfdc382015-05-04 06:59:46 -07001547 encoding: str(accept={str, NoneType}) = NULL
1548 namespace_separator: str(accept={str, NoneType}) = NULL
Brett Cannond0aeda82014-08-22 14:23:20 -04001549 intern: object = NULL
1550
1551Return a new XML parser object.
1552[clinic start generated code]*/
1553
Brett Cannond0aeda82014-08-22 14:23:20 -04001554static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03001555pyexpat_ParserCreate_impl(PyObject *module, const char *encoding,
Larry Hastings89964c42015-04-14 18:07:59 -04001556 const char *namespace_separator, PyObject *intern)
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03001557/*[clinic end generated code: output=295c0cf01ab1146c input=23d29704acad385d]*/
Brett Cannond0aeda82014-08-22 14:23:20 -04001558{
Fred Drakeb91a36b2002-06-27 19:40:48 +00001559 PyObject *result;
1560 int intern_decref = 0;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001561
Fred Drakecde79132001-04-25 16:01:30 +00001562 if (namespace_separator != NULL
1563 && strlen(namespace_separator) > 1) {
1564 PyErr_SetString(PyExc_ValueError,
1565 "namespace_separator must be at most one"
1566 " character, omitted, or None");
1567 return NULL;
1568 }
Fred Drakeb91a36b2002-06-27 19:40:48 +00001569 /* Explicitly passing None means no interning is desired.
1570 Not passing anything means that a new dictionary is used. */
1571 if (intern == Py_None)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001572 intern = NULL;
Fred Drakeb91a36b2002-06-27 19:40:48 +00001573 else if (intern == NULL) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001574 intern = PyDict_New();
1575 if (!intern)
1576 return NULL;
1577 intern_decref = 1;
Fred Drake71b63ff2002-06-28 22:29:01 +00001578 }
Fred Drakeb91a36b2002-06-27 19:40:48 +00001579 else if (!PyDict_Check(intern)) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001580 PyErr_SetString(PyExc_TypeError, "intern must be a dictionary");
1581 return NULL;
Fred Drakeb91a36b2002-06-27 19:40:48 +00001582 }
1583
Serhiy Storchaka20b39b22014-09-28 11:27:24 +03001584 result = newxmlparseobject(encoding, namespace_separator, intern);
Fred Drakeb91a36b2002-06-27 19:40:48 +00001585 if (intern_decref) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001586 Py_DECREF(intern);
Fred Drakeb91a36b2002-06-27 19:40:48 +00001587 }
1588 return result;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001589}
1590
Brett Cannond0aeda82014-08-22 14:23:20 -04001591/*[clinic input]
1592pyexpat.ErrorString
1593
1594 code: long
1595 /
1596
1597Returns string error for given number.
1598[clinic start generated code]*/
1599
Brett Cannond0aeda82014-08-22 14:23:20 -04001600static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03001601pyexpat_ErrorString_impl(PyObject *module, long code)
1602/*[clinic end generated code: output=2feae50d166f2174 input=cc67de010d9e62b3]*/
Brett Cannond0aeda82014-08-22 14:23:20 -04001603{
Fred Drake0582df92000-07-12 04:49:00 +00001604 return Py_BuildValue("z", XML_ErrorString((int)code));
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001605}
1606
1607/* List of methods defined in the module */
1608
1609static struct PyMethodDef pyexpat_methods[] = {
Brett Cannond0aeda82014-08-22 14:23:20 -04001610 PYEXPAT_PARSERCREATE_METHODDEF
1611 PYEXPAT_ERRORSTRING_METHODDEF
1612 {NULL, NULL} /* sentinel */
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001613};
1614
Andrew M. Kuchlingbeba0562000-06-27 00:33:30 +00001615/* Module docstring */
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001616
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001617PyDoc_STRVAR(pyexpat_module_documentation,
1618"Python wrapper for Expat parser.");
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001619
Fred Drakecde79132001-04-25 16:01:30 +00001620/* Initialization function for the module */
1621
1622#ifndef MODULE_NAME
1623#define MODULE_NAME "pyexpat"
1624#endif
1625
1626#ifndef MODULE_INITFUNC
Martin v. Löwis1a214512008-06-11 05:26:20 +00001627#define MODULE_INITFUNC PyInit_pyexpat
Fred Drakecde79132001-04-25 16:01:30 +00001628#endif
1629
Martin v. Löwis1a214512008-06-11 05:26:20 +00001630static struct PyModuleDef pyexpatmodule = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001631 PyModuleDef_HEAD_INIT,
1632 MODULE_NAME,
1633 pyexpat_module_documentation,
1634 -1,
1635 pyexpat_methods,
1636 NULL,
1637 NULL,
1638 NULL,
1639 NULL
Martin v. Löwis1a214512008-06-11 05:26:20 +00001640};
1641
Martin v. Löwis069dde22003-01-21 10:58:18 +00001642PyMODINIT_FUNC
1643MODULE_INITFUNC(void)
Fred Drake0582df92000-07-12 04:49:00 +00001644{
1645 PyObject *m, *d;
Neal Norwitz392c5be2007-08-25 17:20:32 +00001646 PyObject *errmod_name = PyUnicode_FromString(MODULE_NAME ".errors");
Fred Drake85d835f2001-02-08 15:39:08 +00001647 PyObject *errors_module;
1648 PyObject *modelmod_name;
1649 PyObject *model_module;
Fred Drake0582df92000-07-12 04:49:00 +00001650 PyObject *sys_modules;
Georg Brandlb4dac712010-10-15 14:46:48 +00001651 PyObject *tmpnum, *tmpstr;
1652 PyObject *codes_dict;
1653 PyObject *rev_codes_dict;
1654 int res;
Fredrik Lundhd7a42882005-12-13 20:43:04 +00001655 static struct PyExpat_CAPI capi;
Georg Brandlb4dac712010-10-15 14:46:48 +00001656 PyObject *capi_object;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001657
Fred Drake6f987622000-08-25 18:03:30 +00001658 if (errmod_name == NULL)
Martin v. Löwis1a214512008-06-11 05:26:20 +00001659 return NULL;
Neal Norwitz392c5be2007-08-25 17:20:32 +00001660 modelmod_name = PyUnicode_FromString(MODULE_NAME ".model");
Fred Drake85d835f2001-02-08 15:39:08 +00001661 if (modelmod_name == NULL)
Martin v. Löwis1a214512008-06-11 05:26:20 +00001662 return NULL;
Fred Drake6f987622000-08-25 18:03:30 +00001663
Amaury Forgeot d'Arcba4105c2008-07-02 21:41:01 +00001664 if (PyType_Ready(&Xmlparsetype) < 0)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001665 return NULL;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001666
Fred Drake0582df92000-07-12 04:49:00 +00001667 /* Create the module and add the functions */
Martin v. Löwis1a214512008-06-11 05:26:20 +00001668 m = PyModule_Create(&pyexpatmodule);
Neal Norwitz1ac754f2006-01-19 06:09:39 +00001669 if (m == NULL)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001670 return NULL;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001671
Fred Drake0582df92000-07-12 04:49:00 +00001672 /* Add some symbolic constants to the module */
Fred Drakebd6101c2001-02-14 18:29:45 +00001673 if (ErrorObject == NULL) {
1674 ErrorObject = PyErr_NewException("xml.parsers.expat.ExpatError",
Fred Drake93adb692000-09-23 04:55:48 +00001675 NULL, NULL);
Fred Drakebd6101c2001-02-14 18:29:45 +00001676 if (ErrorObject == NULL)
Martin v. Löwis1a214512008-06-11 05:26:20 +00001677 return NULL;
Fred Drakebd6101c2001-02-14 18:29:45 +00001678 }
1679 Py_INCREF(ErrorObject);
Fred Drake93adb692000-09-23 04:55:48 +00001680 PyModule_AddObject(m, "error", ErrorObject);
Fred Drakebd6101c2001-02-14 18:29:45 +00001681 Py_INCREF(ErrorObject);
1682 PyModule_AddObject(m, "ExpatError", ErrorObject);
Fred Drake4ba298c2000-10-29 04:57:53 +00001683 Py_INCREF(&Xmlparsetype);
1684 PyModule_AddObject(m, "XMLParserType", (PyObject *) &Xmlparsetype);
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001685
Fred Drake738293d2000-12-21 17:25:07 +00001686 PyModule_AddStringConstant(m, "EXPAT_VERSION",
Serhiy Storchaka20b39b22014-09-28 11:27:24 +03001687 XML_ExpatVersion());
Fred Drake85d835f2001-02-08 15:39:08 +00001688 {
1689 XML_Expat_Version info = XML_ExpatVersionInfo();
1690 PyModule_AddObject(m, "version_info",
1691 Py_BuildValue("(iii)", info.major,
1692 info.minor, info.micro));
1693 }
Fred Drake0582df92000-07-12 04:49:00 +00001694 /* XXX When Expat supports some way of figuring out how it was
Fred Drake71b63ff2002-06-28 22:29:01 +00001695 compiled, this should check and set native_encoding
1696 appropriately.
Fred Drake0582df92000-07-12 04:49:00 +00001697 */
Fred Drake93adb692000-09-23 04:55:48 +00001698 PyModule_AddStringConstant(m, "native_encoding", "UTF-8");
Fred Drakec23b5232000-08-24 21:57:43 +00001699
Fred Drake85d835f2001-02-08 15:39:08 +00001700 sys_modules = PySys_GetObject("modules");
Christian Heimes7a5457b2016-09-09 00:13:35 +02001701 if (sys_modules == NULL) {
1702 Py_DECREF(m);
1703 return NULL;
1704 }
Fred Drake93adb692000-09-23 04:55:48 +00001705 d = PyModule_GetDict(m);
Christian Heimes7a5457b2016-09-09 00:13:35 +02001706 if (d == NULL) {
1707 Py_DECREF(m);
1708 return NULL;
1709 }
Fred Drake6f987622000-08-25 18:03:30 +00001710 errors_module = PyDict_GetItem(d, errmod_name);
1711 if (errors_module == NULL) {
Fred Drakecde79132001-04-25 16:01:30 +00001712 errors_module = PyModule_New(MODULE_NAME ".errors");
Fred Drake6f987622000-08-25 18:03:30 +00001713 if (errors_module != NULL) {
Fred Drake6f987622000-08-25 18:03:30 +00001714 PyDict_SetItem(sys_modules, errmod_name, errors_module);
Fred Drake93adb692000-09-23 04:55:48 +00001715 /* gives away the reference to errors_module */
1716 PyModule_AddObject(m, "errors", errors_module);
Fred Drakec23b5232000-08-24 21:57:43 +00001717 }
1718 }
Fred Drake6f987622000-08-25 18:03:30 +00001719 Py_DECREF(errmod_name);
Fred Drake85d835f2001-02-08 15:39:08 +00001720 model_module = PyDict_GetItem(d, modelmod_name);
1721 if (model_module == NULL) {
Fred Drakecde79132001-04-25 16:01:30 +00001722 model_module = PyModule_New(MODULE_NAME ".model");
Fred Drake85d835f2001-02-08 15:39:08 +00001723 if (model_module != NULL) {
1724 PyDict_SetItem(sys_modules, modelmod_name, model_module);
1725 /* gives away the reference to model_module */
1726 PyModule_AddObject(m, "model", model_module);
1727 }
1728 }
1729 Py_DECREF(modelmod_name);
Christian Heimes7a5457b2016-09-09 00:13:35 +02001730 if (errors_module == NULL || model_module == NULL) {
Fred Drake85d835f2001-02-08 15:39:08 +00001731 /* Don't core dump later! */
Christian Heimes7a5457b2016-09-09 00:13:35 +02001732 Py_DECREF(m);
Martin v. Löwis1a214512008-06-11 05:26:20 +00001733 return NULL;
Christian Heimes7a5457b2016-09-09 00:13:35 +02001734 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001735
Martin v. Löwisc847f402003-01-21 11:09:21 +00001736#if XML_COMBINED_VERSION > 19505
Martin v. Löwis069dde22003-01-21 10:58:18 +00001737 {
1738 const XML_Feature *features = XML_GetFeatureList();
1739 PyObject *list = PyList_New(0);
1740 if (list == NULL)
1741 /* just ignore it */
1742 PyErr_Clear();
1743 else {
1744 int i = 0;
1745 for (; features[i].feature != XML_FEATURE_END; ++i) {
1746 int ok;
1747 PyObject *item = Py_BuildValue("si", features[i].name,
1748 features[i].value);
1749 if (item == NULL) {
1750 Py_DECREF(list);
1751 list = NULL;
1752 break;
1753 }
1754 ok = PyList_Append(list, item);
1755 Py_DECREF(item);
1756 if (ok < 0) {
1757 PyErr_Clear();
1758 break;
1759 }
1760 }
1761 if (list != NULL)
1762 PyModule_AddObject(m, "features", list);
1763 }
1764 }
Martin v. Löwisc847f402003-01-21 11:09:21 +00001765#endif
Fred Drake6f987622000-08-25 18:03:30 +00001766
Georg Brandlb4dac712010-10-15 14:46:48 +00001767 codes_dict = PyDict_New();
1768 rev_codes_dict = PyDict_New();
1769 if (codes_dict == NULL || rev_codes_dict == NULL) {
1770 Py_XDECREF(codes_dict);
1771 Py_XDECREF(rev_codes_dict);
1772 return NULL;
1773 }
Victor Stinner0fcab4a2011-01-04 12:59:15 +00001774
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001775#define MYCONST(name) \
Georg Brandlb4dac712010-10-15 14:46:48 +00001776 if (PyModule_AddStringConstant(errors_module, #name, \
Serhiy Storchaka20b39b22014-09-28 11:27:24 +03001777 XML_ErrorString(name)) < 0) \
Georg Brandlb4dac712010-10-15 14:46:48 +00001778 return NULL; \
1779 tmpnum = PyLong_FromLong(name); \
1780 if (tmpnum == NULL) return NULL; \
1781 res = PyDict_SetItemString(codes_dict, \
1782 XML_ErrorString(name), tmpnum); \
1783 if (res < 0) return NULL; \
1784 tmpstr = PyUnicode_FromString(XML_ErrorString(name)); \
1785 if (tmpstr == NULL) return NULL; \
1786 res = PyDict_SetItem(rev_codes_dict, tmpnum, tmpstr); \
1787 Py_DECREF(tmpstr); \
1788 Py_DECREF(tmpnum); \
1789 if (res < 0) return NULL; \
Fred Drake7bd9f412000-07-04 23:51:31 +00001790
Fred Drake0582df92000-07-12 04:49:00 +00001791 MYCONST(XML_ERROR_NO_MEMORY);
1792 MYCONST(XML_ERROR_SYNTAX);
1793 MYCONST(XML_ERROR_NO_ELEMENTS);
1794 MYCONST(XML_ERROR_INVALID_TOKEN);
1795 MYCONST(XML_ERROR_UNCLOSED_TOKEN);
1796 MYCONST(XML_ERROR_PARTIAL_CHAR);
1797 MYCONST(XML_ERROR_TAG_MISMATCH);
1798 MYCONST(XML_ERROR_DUPLICATE_ATTRIBUTE);
1799 MYCONST(XML_ERROR_JUNK_AFTER_DOC_ELEMENT);
1800 MYCONST(XML_ERROR_PARAM_ENTITY_REF);
1801 MYCONST(XML_ERROR_UNDEFINED_ENTITY);
1802 MYCONST(XML_ERROR_RECURSIVE_ENTITY_REF);
1803 MYCONST(XML_ERROR_ASYNC_ENTITY);
1804 MYCONST(XML_ERROR_BAD_CHAR_REF);
1805 MYCONST(XML_ERROR_BINARY_ENTITY_REF);
1806 MYCONST(XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF);
1807 MYCONST(XML_ERROR_MISPLACED_XML_PI);
1808 MYCONST(XML_ERROR_UNKNOWN_ENCODING);
1809 MYCONST(XML_ERROR_INCORRECT_ENCODING);
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001810 MYCONST(XML_ERROR_UNCLOSED_CDATA_SECTION);
1811 MYCONST(XML_ERROR_EXTERNAL_ENTITY_HANDLING);
1812 MYCONST(XML_ERROR_NOT_STANDALONE);
Fred Drake283b6702004-08-04 22:28:16 +00001813 MYCONST(XML_ERROR_UNEXPECTED_STATE);
1814 MYCONST(XML_ERROR_ENTITY_DECLARED_IN_PE);
1815 MYCONST(XML_ERROR_FEATURE_REQUIRES_XML_DTD);
1816 MYCONST(XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING);
1817 /* Added in Expat 1.95.7. */
1818 MYCONST(XML_ERROR_UNBOUND_PREFIX);
1819 /* Added in Expat 1.95.8. */
1820 MYCONST(XML_ERROR_UNDECLARING_PREFIX);
1821 MYCONST(XML_ERROR_INCOMPLETE_PE);
1822 MYCONST(XML_ERROR_XML_DECL);
1823 MYCONST(XML_ERROR_TEXT_DECL);
1824 MYCONST(XML_ERROR_PUBLICID);
1825 MYCONST(XML_ERROR_SUSPENDED);
1826 MYCONST(XML_ERROR_NOT_SUSPENDED);
1827 MYCONST(XML_ERROR_ABORTED);
1828 MYCONST(XML_ERROR_FINISHED);
1829 MYCONST(XML_ERROR_SUSPEND_PE);
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001830
Georg Brandlb4dac712010-10-15 14:46:48 +00001831 if (PyModule_AddStringConstant(errors_module, "__doc__",
1832 "Constants used to describe "
1833 "error conditions.") < 0)
1834 return NULL;
Fred Drake85d835f2001-02-08 15:39:08 +00001835
Georg Brandlb4dac712010-10-15 14:46:48 +00001836 if (PyModule_AddObject(errors_module, "codes", codes_dict) < 0)
1837 return NULL;
1838 if (PyModule_AddObject(errors_module, "messages", rev_codes_dict) < 0)
1839 return NULL;
Victor Stinner0fcab4a2011-01-04 12:59:15 +00001840
Fred Drake93adb692000-09-23 04:55:48 +00001841#undef MYCONST
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001842
Fred Drake85d835f2001-02-08 15:39:08 +00001843#define MYCONST(c) PyModule_AddIntConstant(m, #c, c)
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001844 MYCONST(XML_PARAM_ENTITY_PARSING_NEVER);
1845 MYCONST(XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE);
1846 MYCONST(XML_PARAM_ENTITY_PARSING_ALWAYS);
Fred Drake85d835f2001-02-08 15:39:08 +00001847#undef MYCONST
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001848
Fred Drake85d835f2001-02-08 15:39:08 +00001849#define MYCONST(c) PyModule_AddIntConstant(model_module, #c, c)
1850 PyModule_AddStringConstant(model_module, "__doc__",
1851 "Constants used to interpret content model information.");
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001852
Fred Drake85d835f2001-02-08 15:39:08 +00001853 MYCONST(XML_CTYPE_EMPTY);
1854 MYCONST(XML_CTYPE_ANY);
1855 MYCONST(XML_CTYPE_MIXED);
1856 MYCONST(XML_CTYPE_NAME);
1857 MYCONST(XML_CTYPE_CHOICE);
1858 MYCONST(XML_CTYPE_SEQ);
1859
1860 MYCONST(XML_CQUANT_NONE);
1861 MYCONST(XML_CQUANT_OPT);
1862 MYCONST(XML_CQUANT_REP);
1863 MYCONST(XML_CQUANT_PLUS);
1864#undef MYCONST
Fredrik Lundhc3345042005-12-13 19:49:55 +00001865
1866 /* initialize pyexpat dispatch table */
Fredrik Lundhd7a42882005-12-13 20:43:04 +00001867 capi.size = sizeof(capi);
Fredrik Lundhcc117db2005-12-13 21:55:36 +00001868 capi.magic = PyExpat_CAPI_MAGIC;
Fredrik Lundhd7a42882005-12-13 20:43:04 +00001869 capi.MAJOR_VERSION = XML_MAJOR_VERSION;
1870 capi.MINOR_VERSION = XML_MINOR_VERSION;
1871 capi.MICRO_VERSION = XML_MICRO_VERSION;
1872 capi.ErrorString = XML_ErrorString;
Fredrik Lundhcc117db2005-12-13 21:55:36 +00001873 capi.GetErrorCode = XML_GetErrorCode;
1874 capi.GetErrorColumnNumber = XML_GetErrorColumnNumber;
1875 capi.GetErrorLineNumber = XML_GetErrorLineNumber;
Fredrik Lundhd7a42882005-12-13 20:43:04 +00001876 capi.Parse = XML_Parse;
1877 capi.ParserCreate_MM = XML_ParserCreate_MM;
1878 capi.ParserFree = XML_ParserFree;
1879 capi.SetCharacterDataHandler = XML_SetCharacterDataHandler;
1880 capi.SetCommentHandler = XML_SetCommentHandler;
1881 capi.SetDefaultHandlerExpand = XML_SetDefaultHandlerExpand;
1882 capi.SetElementHandler = XML_SetElementHandler;
1883 capi.SetNamespaceDeclHandler = XML_SetNamespaceDeclHandler;
1884 capi.SetProcessingInstructionHandler = XML_SetProcessingInstructionHandler;
1885 capi.SetUnknownEncodingHandler = XML_SetUnknownEncodingHandler;
1886 capi.SetUserData = XML_SetUserData;
Eli Bendersky2b6b73e2012-06-01 11:32:34 +03001887 capi.SetStartDoctypeDeclHandler = XML_SetStartDoctypeDeclHandler;
Serhiy Storchaka66d53fa2013-05-22 17:07:51 +03001888 capi.SetEncoding = XML_SetEncoding;
Eli Bendersky6dc32b32013-05-25 05:25:48 -07001889 capi.DefaultUnknownEncodingHandler = PyUnknownEncodingHandler;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001890
Benjamin Petersonb173f782009-05-05 22:31:58 +00001891 /* export using capsule */
1892 capi_object = PyCapsule_New(&capi, PyExpat_CAPSULE_NAME, NULL);
Fredrik Lundhd7a42882005-12-13 20:43:04 +00001893 if (capi_object)
1894 PyModule_AddObject(m, "expat_CAPI", capi_object);
Martin v. Löwis1a214512008-06-11 05:26:20 +00001895 return m;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001896}
1897
Fred Drake6f987622000-08-25 18:03:30 +00001898static void
Martin v. Löwis5b68ce32001-10-21 08:53:52 +00001899clear_handlers(xmlparseobject *self, int initial)
Fred Drake0582df92000-07-12 04:49:00 +00001900{
Fred Drakecde79132001-04-25 16:01:30 +00001901 int i = 0;
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001902
Fred Drake71b63ff2002-06-28 22:29:01 +00001903 for (; handler_info[i].name != NULL; i++) {
Martin v. Löwis5b68ce32001-10-21 08:53:52 +00001904 if (initial)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001905 self->handlers[i] = NULL;
1906 else {
Serhiy Storchaka1ed017a2015-12-27 15:51:32 +02001907 Py_CLEAR(self->handlers[i]);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001908 handler_info[i].setter(self->itself, NULL);
Fred Drakecde79132001-04-25 16:01:30 +00001909 }
Fred Drakecde79132001-04-25 16:01:30 +00001910 }
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001911}
1912
Tim Peters0c322792002-07-17 16:49:03 +00001913static struct HandlerInfo handler_info[] = {
Fred Drake71b63ff2002-06-28 22:29:01 +00001914 {"StartElementHandler",
1915 (xmlhandlersetter)XML_SetStartElementHandler,
Fred Drake0582df92000-07-12 04:49:00 +00001916 (xmlhandler)my_StartElementHandler},
Fred Drake71b63ff2002-06-28 22:29:01 +00001917 {"EndElementHandler",
1918 (xmlhandlersetter)XML_SetEndElementHandler,
Fred Drake0582df92000-07-12 04:49:00 +00001919 (xmlhandler)my_EndElementHandler},
Fred Drake71b63ff2002-06-28 22:29:01 +00001920 {"ProcessingInstructionHandler",
Fred Drake0582df92000-07-12 04:49:00 +00001921 (xmlhandlersetter)XML_SetProcessingInstructionHandler,
1922 (xmlhandler)my_ProcessingInstructionHandler},
Fred Drake71b63ff2002-06-28 22:29:01 +00001923 {"CharacterDataHandler",
Fred Drake0582df92000-07-12 04:49:00 +00001924 (xmlhandlersetter)XML_SetCharacterDataHandler,
1925 (xmlhandler)my_CharacterDataHandler},
Fred Drake71b63ff2002-06-28 22:29:01 +00001926 {"UnparsedEntityDeclHandler",
Fred Drake0582df92000-07-12 04:49:00 +00001927 (xmlhandlersetter)XML_SetUnparsedEntityDeclHandler,
Fred Drake2a3d7db2002-06-28 22:56:48 +00001928 (xmlhandler)my_UnparsedEntityDeclHandler},
Fred Drake71b63ff2002-06-28 22:29:01 +00001929 {"NotationDeclHandler",
Fred Drake0582df92000-07-12 04:49:00 +00001930 (xmlhandlersetter)XML_SetNotationDeclHandler,
Fred Drake2a3d7db2002-06-28 22:56:48 +00001931 (xmlhandler)my_NotationDeclHandler},
Fred Drake71b63ff2002-06-28 22:29:01 +00001932 {"StartNamespaceDeclHandler",
1933 (xmlhandlersetter)XML_SetStartNamespaceDeclHandler,
Fred Drake2a3d7db2002-06-28 22:56:48 +00001934 (xmlhandler)my_StartNamespaceDeclHandler},
Fred Drake71b63ff2002-06-28 22:29:01 +00001935 {"EndNamespaceDeclHandler",
1936 (xmlhandlersetter)XML_SetEndNamespaceDeclHandler,
Fred Drake2a3d7db2002-06-28 22:56:48 +00001937 (xmlhandler)my_EndNamespaceDeclHandler},
Fred Drake0582df92000-07-12 04:49:00 +00001938 {"CommentHandler",
1939 (xmlhandlersetter)XML_SetCommentHandler,
1940 (xmlhandler)my_CommentHandler},
1941 {"StartCdataSectionHandler",
Fred Drake71b63ff2002-06-28 22:29:01 +00001942 (xmlhandlersetter)XML_SetStartCdataSectionHandler,
Fred Drake0582df92000-07-12 04:49:00 +00001943 (xmlhandler)my_StartCdataSectionHandler},
1944 {"EndCdataSectionHandler",
Fred Drake71b63ff2002-06-28 22:29:01 +00001945 (xmlhandlersetter)XML_SetEndCdataSectionHandler,
Fred Drake0582df92000-07-12 04:49:00 +00001946 (xmlhandler)my_EndCdataSectionHandler},
1947 {"DefaultHandler",
1948 (xmlhandlersetter)XML_SetDefaultHandler,
1949 (xmlhandler)my_DefaultHandler},
1950 {"DefaultHandlerExpand",
1951 (xmlhandlersetter)XML_SetDefaultHandlerExpand,
1952 (xmlhandler)my_DefaultHandlerExpandHandler},
1953 {"NotStandaloneHandler",
1954 (xmlhandlersetter)XML_SetNotStandaloneHandler,
1955 (xmlhandler)my_NotStandaloneHandler},
1956 {"ExternalEntityRefHandler",
1957 (xmlhandlersetter)XML_SetExternalEntityRefHandler,
Fred Drake2a3d7db2002-06-28 22:56:48 +00001958 (xmlhandler)my_ExternalEntityRefHandler},
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001959 {"StartDoctypeDeclHandler",
Fred Drake71b63ff2002-06-28 22:29:01 +00001960 (xmlhandlersetter)XML_SetStartDoctypeDeclHandler,
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001961 (xmlhandler)my_StartDoctypeDeclHandler},
1962 {"EndDoctypeDeclHandler",
Fred Drake71b63ff2002-06-28 22:29:01 +00001963 (xmlhandlersetter)XML_SetEndDoctypeDeclHandler,
Martin v. Löwis0078f6c2001-01-21 10:18:10 +00001964 (xmlhandler)my_EndDoctypeDeclHandler},
Fred Drake85d835f2001-02-08 15:39:08 +00001965 {"EntityDeclHandler",
1966 (xmlhandlersetter)XML_SetEntityDeclHandler,
1967 (xmlhandler)my_EntityDeclHandler},
1968 {"XmlDeclHandler",
1969 (xmlhandlersetter)XML_SetXmlDeclHandler,
1970 (xmlhandler)my_XmlDeclHandler},
1971 {"ElementDeclHandler",
1972 (xmlhandlersetter)XML_SetElementDeclHandler,
1973 (xmlhandler)my_ElementDeclHandler},
1974 {"AttlistDeclHandler",
1975 (xmlhandlersetter)XML_SetAttlistDeclHandler,
1976 (xmlhandler)my_AttlistDeclHandler},
Martin v. Löwisc847f402003-01-21 11:09:21 +00001977#if XML_COMBINED_VERSION >= 19504
Martin v. Löwis069dde22003-01-21 10:58:18 +00001978 {"SkippedEntityHandler",
1979 (xmlhandlersetter)XML_SetSkippedEntityHandler,
1980 (xmlhandler)my_SkippedEntityHandler},
Martin v. Löwisc847f402003-01-21 11:09:21 +00001981#endif
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001982
Fred Drake0582df92000-07-12 04:49:00 +00001983 {NULL, NULL, NULL} /* sentinel */
Andrew M. Kuchlingb7f10532000-03-31 15:43:31 +00001984};
Brett Cannond0aeda82014-08-22 14:23:20 -04001985
1986/*[clinic input]
1987dump buffer
1988[clinic start generated code]*/
Serhiy Storchaka1009bf12015-04-03 23:53:51 +03001989/*[clinic end generated code: output=da39a3ee5e6b4b0d input=524ce2e021e4eba6]*/