Martin v. Löwis | 7090ed1 | 2001-09-19 10:37:50 +0000 | [diff] [blame] | 1 | #include "Python.h" |
Fred Drake | 4113b13 | 2001-03-24 19:58:26 +0000 | [diff] [blame] | 2 | #include <ctype.h> |
| 3 | |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 4 | #include "structmember.h" |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 5 | #include "frameobject.h" |
Fred Drake | a77254a | 2000-09-29 19:23:29 +0000 | [diff] [blame] | 6 | #include "expat.h" |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 7 | |
Fredrik Lundh | c334504 | 2005-12-13 19:49:55 +0000 | [diff] [blame] | 8 | #include "pyexpat.h" |
| 9 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 10 | /* Do not emit Clinic output to a file as that wreaks havoc with conditionally |
| 11 | included methods. */ |
| 12 | /*[clinic input] |
| 13 | module pyexpat |
| 14 | [clinic start generated code]*/ |
| 15 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=b168d503a4490c15]*/ |
| 16 | |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 17 | #define XML_COMBINED_VERSION (10000*XML_MAJOR_VERSION+100*XML_MINOR_VERSION+XML_MICRO_VERSION) |
| 18 | |
Christian Heimes | fa535f5 | 2013-07-07 17:35:11 +0200 | [diff] [blame] | 19 | static XML_Memory_Handling_Suite ExpatMemoryHandler = { |
| 20 | PyObject_Malloc, PyObject_Realloc, PyObject_Free}; |
| 21 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 22 | enum HandlerTypes { |
| 23 | StartElement, |
| 24 | EndElement, |
| 25 | ProcessingInstruction, |
| 26 | CharacterData, |
| 27 | UnparsedEntityDecl, |
| 28 | NotationDecl, |
| 29 | StartNamespaceDecl, |
| 30 | EndNamespaceDecl, |
| 31 | Comment, |
| 32 | StartCdataSection, |
| 33 | EndCdataSection, |
| 34 | Default, |
| 35 | DefaultHandlerExpand, |
| 36 | NotStandalone, |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 37 | ExternalEntityRef, |
| 38 | StartDoctypeDecl, |
| 39 | EndDoctypeDecl, |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 40 | EntityDecl, |
| 41 | XmlDecl, |
| 42 | ElementDecl, |
| 43 | AttlistDecl, |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 44 | #if XML_COMBINED_VERSION >= 19504 |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 45 | SkippedEntity, |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 46 | #endif |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 47 | _DummyDecl |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 48 | }; |
| 49 | |
| 50 | static PyObject *ErrorObject; |
| 51 | |
| 52 | /* ----------------------------------------------------- */ |
| 53 | |
| 54 | /* Declarations for objects of type xmlparser */ |
| 55 | |
| 56 | typedef struct { |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 57 | PyObject_HEAD |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 58 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 59 | XML_Parser itself; |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 60 | int ordered_attributes; /* Return attributes as a list. */ |
| 61 | int specified_attributes; /* Report only specified attributes. */ |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 62 | int in_callback; /* Is a callback active? */ |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 63 | int ns_prefixes; /* Namespace-triplets mode? */ |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 64 | XML_Char *buffer; /* Buffer used when accumulating characters */ |
| 65 | /* NULL if not enabled */ |
| 66 | int buffer_size; /* Size of buffer, in XML_Char units */ |
| 67 | int buffer_used; /* Buffer units in use */ |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 68 | PyObject *intern; /* Dictionary to intern strings */ |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 69 | PyObject **handlers; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 70 | } xmlparseobject; |
| 71 | |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 72 | #include "clinic/pyexpat.c.h" |
| 73 | |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 74 | #define CHARACTER_DATA_BUFFER_SIZE 8192 |
| 75 | |
Jeremy Hylton | 938ace6 | 2002-07-17 16:30:39 +0000 | [diff] [blame] | 76 | static PyTypeObject Xmlparsetype; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 77 | |
Fred Drake | 117ac85 | 2002-09-24 16:24:54 +0000 | [diff] [blame] | 78 | typedef void (*xmlhandlersetter)(XML_Parser self, void *meth); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 79 | typedef void* xmlhandler; |
| 80 | |
Andrew M. Kuchling | beba056 | 2000-06-27 00:33:30 +0000 | [diff] [blame] | 81 | struct HandlerInfo { |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 82 | const char *name; |
| 83 | xmlhandlersetter setter; |
| 84 | xmlhandler handler; |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 85 | PyGetSetDef getset; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 86 | }; |
| 87 | |
Jeremy Hylton | 938ace6 | 2002-07-17 16:30:39 +0000 | [diff] [blame] | 88 | static struct HandlerInfo handler_info[64]; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 89 | |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 90 | /* Set an integer attribute on the error object; return true on success, |
| 91 | * false on an exception. |
| 92 | */ |
| 93 | static int |
Serhiy Storchaka | ef1585e | 2015-12-25 20:01:53 +0200 | [diff] [blame] | 94 | set_error_attr(PyObject *err, const char *name, int value) |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 95 | { |
Christian Heimes | 217cfd1 | 2007-12-02 14:31:20 +0000 | [diff] [blame] | 96 | PyObject *v = PyLong_FromLong(value); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 97 | |
Neal Norwitz | 2f5e990 | 2006-03-08 06:36:45 +0000 | [diff] [blame] | 98 | if (v == NULL || PyObject_SetAttrString(err, name, v) == -1) { |
| 99 | Py_XDECREF(v); |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 100 | return 0; |
| 101 | } |
Michael W. Hudson | 0bb8454 | 2004-08-03 11:31:31 +0000 | [diff] [blame] | 102 | Py_DECREF(v); |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 103 | return 1; |
| 104 | } |
| 105 | |
| 106 | /* Build and set an Expat exception, including positioning |
| 107 | * information. Always returns NULL. |
| 108 | */ |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 109 | static PyObject * |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 110 | set_error(xmlparseobject *self, enum XML_Error code) |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 111 | { |
| 112 | PyObject *err; |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 113 | PyObject *buffer; |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 114 | XML_Parser parser = self->itself; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 115 | int lineno = XML_GetErrorLineNumber(parser); |
| 116 | int column = XML_GetErrorColumnNumber(parser); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 117 | |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 118 | buffer = PyUnicode_FromFormat("%s: line %i, column %i", |
| 119 | XML_ErrorString(code), lineno, column); |
| 120 | if (buffer == NULL) |
| 121 | return NULL; |
Jeroen Demeyer | 196a530 | 2019-07-04 12:31:34 +0200 | [diff] [blame] | 122 | err = _PyObject_CallOneArg(ErrorObject, buffer); |
Victor Stinner | 499dfcf | 2011-03-21 13:26:24 +0100 | [diff] [blame] | 123 | Py_DECREF(buffer); |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 124 | 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 Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 129 | } |
Neal Norwitz | 2f5e990 | 2006-03-08 06:36:45 +0000 | [diff] [blame] | 130 | Py_XDECREF(err); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 131 | return NULL; |
| 132 | } |
| 133 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 134 | static int |
| 135 | have_handler(xmlparseobject *self, int type) |
| 136 | { |
| 137 | PyObject *handler = self->handlers[type]; |
| 138 | return handler != NULL; |
| 139 | } |
| 140 | |
Andrew M. Kuchling | beba056 | 2000-06-27 00:33:30 +0000 | [diff] [blame] | 141 | /* Convert a string of XML_Chars into a Unicode string. |
| 142 | Returns None if str is a null pointer. */ |
| 143 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 144 | static PyObject * |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 145 | conv_string_to_unicode(const XML_Char *str) |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 146 | { |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 147 | /* XXX currently this code assumes that XML_Char is 8-bit, |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 148 | and hence in UTF-8. */ |
| 149 | /* UTF-8 from Expat, Unicode desired */ |
| 150 | if (str == NULL) { |
Serhiy Storchaka | 228b12e | 2017-01-23 09:47:21 +0200 | [diff] [blame] | 151 | Py_RETURN_NONE; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 152 | } |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 153 | return PyUnicode_DecodeUTF8(str, strlen(str), "strict"); |
Andrew M. Kuchling | beba056 | 2000-06-27 00:33:30 +0000 | [diff] [blame] | 154 | } |
| 155 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 156 | static PyObject * |
| 157 | conv_string_len_to_unicode(const XML_Char *str, int len) |
| 158 | { |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 159 | /* XXX currently this code assumes that XML_Char is 8-bit, |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 160 | and hence in UTF-8. */ |
| 161 | /* UTF-8 from Expat, Unicode desired */ |
| 162 | if (str == NULL) { |
Serhiy Storchaka | 228b12e | 2017-01-23 09:47:21 +0200 | [diff] [blame] | 163 | Py_RETURN_NONE; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 164 | } |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 165 | return PyUnicode_DecodeUTF8((const char *)str, len, "strict"); |
Andrew M. Kuchling | beba056 | 2000-06-27 00:33:30 +0000 | [diff] [blame] | 166 | } |
Andrew M. Kuchling | beba056 | 2000-06-27 00:33:30 +0000 | [diff] [blame] | 167 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 168 | /* Callback routines */ |
| 169 | |
Martin v. Löwis | 5b68ce3 | 2001-10-21 08:53:52 +0000 | [diff] [blame] | 170 | static void clear_handlers(xmlparseobject *self, int initial); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 171 | |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 172 | /* This handler is used when an error has been detected, in the hope |
| 173 | that actual parsing can be terminated early. This will only help |
| 174 | if an external entity reference is encountered. */ |
| 175 | static int |
| 176 | error_external_entity_ref_handler(XML_Parser parser, |
| 177 | const XML_Char *context, |
| 178 | const XML_Char *base, |
| 179 | const XML_Char *systemId, |
| 180 | const XML_Char *publicId) |
| 181 | { |
| 182 | return 0; |
| 183 | } |
| 184 | |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 185 | /* Dummy character data handler used when an error (exception) has |
| 186 | been detected, and the actual parsing can be terminated early. |
| 187 | This is needed since character data handler can't be safely removed |
| 188 | from within the character data handler, but can be replaced. It is |
| 189 | used only from the character data handler trampoline, and must be |
| 190 | used right after `flag_error()` is called. */ |
| 191 | static void |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 192 | noop_character_data_handler(void *userData, const XML_Char *data, int len) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 193 | { |
| 194 | /* Do nothing. */ |
| 195 | } |
| 196 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 197 | static void |
| 198 | flag_error(xmlparseobject *self) |
| 199 | { |
Martin v. Löwis | 5b68ce3 | 2001-10-21 08:53:52 +0000 | [diff] [blame] | 200 | clear_handlers(self, 0); |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 201 | XML_SetExternalEntityRefHandler(self->itself, |
| 202 | error_external_entity_ref_handler); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 203 | } |
| 204 | |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 205 | static PyObject* |
Serhiy Storchaka | ef1585e | 2015-12-25 20:01:53 +0200 | [diff] [blame] | 206 | call_with_frame(const char *funcname, int lineno, PyObject* func, PyObject* args, |
Fred Drake | 39689c5 | 2004-08-13 03:12:57 +0000 | [diff] [blame] | 207 | xmlparseobject *self) |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 208 | { |
Antoine Pitrou | 0ddbf47 | 2014-10-08 20:00:09 +0200 | [diff] [blame] | 209 | PyObject *res; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 210 | |
Jeroen Demeyer | 1dbd084 | 2019-07-11 17:57:32 +0200 | [diff] [blame] | 211 | res = PyObject_Call(func, args, NULL); |
Jeremy Hylton | 9263f57 | 2003-06-27 16:13:17 +0000 | [diff] [blame] | 212 | if (res == NULL) { |
Antoine Pitrou | 0ddbf47 | 2014-10-08 20:00:09 +0200 | [diff] [blame] | 213 | _PyTraceback_Add(funcname, __FILE__, lineno); |
Fred Drake | 39689c5 | 2004-08-13 03:12:57 +0000 | [diff] [blame] | 214 | XML_StopParser(self->itself, XML_FALSE); |
Jeremy Hylton | 9263f57 | 2003-06-27 16:13:17 +0000 | [diff] [blame] | 215 | } |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 216 | return res; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 217 | } |
| 218 | |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 219 | static PyObject* |
| 220 | string_intern(xmlparseobject *self, const char* str) |
| 221 | { |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 222 | PyObject *result = conv_string_to_unicode(str); |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 223 | PyObject *value; |
Neal Norwitz | 484d9a4 | 2005-09-30 04:46:49 +0000 | [diff] [blame] | 224 | /* result can be NULL if the unicode conversion failed. */ |
| 225 | if (!result) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 226 | return result; |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 227 | if (!self->intern) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 228 | return result; |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 229 | value = PyDict_GetItemWithError(self->intern, result); |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 230 | if (!value) { |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 231 | if (!PyErr_Occurred() && |
| 232 | PyDict_SetItem(self->intern, result, result) == 0) |
| 233 | { |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 234 | return result; |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 235 | } |
Zackery Spytz | 68def05 | 2018-10-19 00:57:38 -0600 | [diff] [blame] | 236 | else { |
| 237 | Py_DECREF(result); |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 238 | return NULL; |
Zackery Spytz | 68def05 | 2018-10-19 00:57:38 -0600 | [diff] [blame] | 239 | } |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 240 | } |
| 241 | Py_INCREF(value); |
| 242 | Py_DECREF(result); |
| 243 | return value; |
| 244 | } |
| 245 | |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 246 | /* Return 0 on success, -1 on exception. |
| 247 | * flag_error() will be called before return if needed. |
| 248 | */ |
| 249 | static int |
| 250 | call_character_handler(xmlparseobject *self, const XML_Char *buffer, int len) |
| 251 | { |
| 252 | PyObject *args; |
| 253 | PyObject *temp; |
| 254 | |
Georg Brandl | c01537f | 2010-10-15 16:26:08 +0000 | [diff] [blame] | 255 | if (!have_handler(self, CharacterData)) |
| 256 | return -1; |
| 257 | |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 258 | args = PyTuple_New(1); |
| 259 | if (args == NULL) |
| 260 | return -1; |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 261 | temp = (conv_string_len_to_unicode(buffer, len)); |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 262 | if (temp == NULL) { |
| 263 | Py_DECREF(args); |
| 264 | flag_error(self); |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 265 | XML_SetCharacterDataHandler(self->itself, |
| 266 | noop_character_data_handler); |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 267 | return -1; |
| 268 | } |
| 269 | PyTuple_SET_ITEM(args, 0, temp); |
| 270 | /* temp is now a borrowed reference; consider it unused. */ |
| 271 | self->in_callback = 1; |
Antoine Pitrou | 0ddbf47 | 2014-10-08 20:00:09 +0200 | [diff] [blame] | 272 | temp = call_with_frame("CharacterData", __LINE__, |
Fred Drake | 39689c5 | 2004-08-13 03:12:57 +0000 | [diff] [blame] | 273 | self->handlers[CharacterData], args, self); |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 274 | /* temp is an owned reference again, or NULL */ |
| 275 | self->in_callback = 0; |
| 276 | Py_DECREF(args); |
| 277 | if (temp == NULL) { |
| 278 | flag_error(self); |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 279 | XML_SetCharacterDataHandler(self->itself, |
| 280 | noop_character_data_handler); |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 281 | return -1; |
| 282 | } |
| 283 | Py_DECREF(temp); |
| 284 | return 0; |
| 285 | } |
| 286 | |
| 287 | static int |
| 288 | flush_character_buffer(xmlparseobject *self) |
| 289 | { |
| 290 | int rc; |
| 291 | if (self->buffer == NULL || self->buffer_used == 0) |
| 292 | return 0; |
| 293 | rc = call_character_handler(self, self->buffer, self->buffer_used); |
| 294 | self->buffer_used = 0; |
| 295 | return rc; |
| 296 | } |
| 297 | |
| 298 | static void |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 299 | my_CharacterDataHandler(void *userData, const XML_Char *data, int len) |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 300 | { |
| 301 | xmlparseobject *self = (xmlparseobject *) userData; |
Victor Stinner | 9e09c26 | 2013-07-18 23:17:01 +0200 | [diff] [blame] | 302 | |
| 303 | if (PyErr_Occurred()) |
| 304 | return; |
| 305 | |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 306 | if (self->buffer == NULL) |
| 307 | call_character_handler(self, data, len); |
| 308 | else { |
| 309 | if ((self->buffer_used + len) > self->buffer_size) { |
| 310 | if (flush_character_buffer(self) < 0) |
| 311 | return; |
| 312 | /* handler might have changed; drop the rest on the floor |
| 313 | * if there isn't a handler anymore |
| 314 | */ |
| 315 | if (!have_handler(self, CharacterData)) |
| 316 | return; |
| 317 | } |
| 318 | if (len > self->buffer_size) { |
| 319 | call_character_handler(self, data, len); |
| 320 | self->buffer_used = 0; |
| 321 | } |
| 322 | else { |
| 323 | memcpy(self->buffer + self->buffer_used, |
| 324 | data, len * sizeof(XML_Char)); |
| 325 | self->buffer_used += len; |
| 326 | } |
| 327 | } |
| 328 | } |
| 329 | |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 330 | static void |
| 331 | my_StartElementHandler(void *userData, |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 332 | const XML_Char *name, const XML_Char *atts[]) |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 333 | { |
| 334 | xmlparseobject *self = (xmlparseobject *)userData; |
| 335 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 336 | if (have_handler(self, StartElement)) { |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 337 | PyObject *container, *rv, *args; |
| 338 | int i, max; |
| 339 | |
Victor Stinner | 9e09c26 | 2013-07-18 23:17:01 +0200 | [diff] [blame] | 340 | if (PyErr_Occurred()) |
| 341 | return; |
| 342 | |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 343 | if (flush_character_buffer(self) < 0) |
| 344 | return; |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 345 | /* Set max to the number of slots filled in atts[]; max/2 is |
| 346 | * the number of attributes we need to process. |
| 347 | */ |
| 348 | if (self->specified_attributes) { |
| 349 | max = XML_GetSpecifiedAttributeCount(self->itself); |
| 350 | } |
| 351 | else { |
| 352 | max = 0; |
| 353 | while (atts[max] != NULL) |
| 354 | max += 2; |
| 355 | } |
| 356 | /* Build the container. */ |
| 357 | if (self->ordered_attributes) |
| 358 | container = PyList_New(max); |
| 359 | else |
| 360 | container = PyDict_New(); |
| 361 | if (container == NULL) { |
| 362 | flag_error(self); |
| 363 | return; |
| 364 | } |
| 365 | for (i = 0; i < max; i += 2) { |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 366 | PyObject *n = string_intern(self, (XML_Char *) atts[i]); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 367 | PyObject *v; |
| 368 | if (n == NULL) { |
| 369 | flag_error(self); |
| 370 | Py_DECREF(container); |
| 371 | return; |
| 372 | } |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 373 | v = conv_string_to_unicode((XML_Char *) atts[i+1]); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 374 | if (v == NULL) { |
| 375 | flag_error(self); |
| 376 | Py_DECREF(container); |
| 377 | Py_DECREF(n); |
| 378 | return; |
| 379 | } |
| 380 | if (self->ordered_attributes) { |
| 381 | PyList_SET_ITEM(container, i, n); |
| 382 | PyList_SET_ITEM(container, i+1, v); |
| 383 | } |
| 384 | else if (PyDict_SetItem(container, n, v)) { |
| 385 | flag_error(self); |
| 386 | Py_DECREF(n); |
| 387 | Py_DECREF(v); |
Zackery Spytz | 68def05 | 2018-10-19 00:57:38 -0600 | [diff] [blame] | 388 | Py_DECREF(container); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 389 | return; |
| 390 | } |
| 391 | else { |
| 392 | Py_DECREF(n); |
| 393 | Py_DECREF(v); |
| 394 | } |
| 395 | } |
Neal Norwitz | 484d9a4 | 2005-09-30 04:46:49 +0000 | [diff] [blame] | 396 | args = string_intern(self, name); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 397 | if (args == NULL) { |
| 398 | Py_DECREF(container); |
| 399 | return; |
| 400 | } |
Zackery Spytz | 68def05 | 2018-10-19 00:57:38 -0600 | [diff] [blame] | 401 | args = Py_BuildValue("(NN)", args, container); |
| 402 | if (args == NULL) { |
| 403 | return; |
| 404 | } |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 405 | /* Container is now a borrowed reference; ignore it. */ |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 406 | self->in_callback = 1; |
Antoine Pitrou | 0ddbf47 | 2014-10-08 20:00:09 +0200 | [diff] [blame] | 407 | rv = call_with_frame("StartElement", __LINE__, |
Fred Drake | 39689c5 | 2004-08-13 03:12:57 +0000 | [diff] [blame] | 408 | self->handlers[StartElement], args, self); |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 409 | self->in_callback = 0; |
| 410 | Py_DECREF(args); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 411 | if (rv == NULL) { |
| 412 | flag_error(self); |
| 413 | return; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 414 | } |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 415 | Py_DECREF(rv); |
| 416 | } |
| 417 | } |
| 418 | |
| 419 | #define RC_HANDLER(RC, NAME, PARAMS, INIT, PARAM_FORMAT, CONVERSION, \ |
| 420 | RETURN, GETUSERDATA) \ |
| 421 | static RC \ |
| 422 | my_##NAME##Handler PARAMS {\ |
| 423 | xmlparseobject *self = GETUSERDATA ; \ |
| 424 | PyObject *args = NULL; \ |
| 425 | PyObject *rv = NULL; \ |
| 426 | INIT \ |
| 427 | \ |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 428 | if (have_handler(self, NAME)) { \ |
Victor Stinner | 9e09c26 | 2013-07-18 23:17:01 +0200 | [diff] [blame] | 429 | if (PyErr_Occurred()) \ |
| 430 | return RETURN; \ |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 431 | if (flush_character_buffer(self) < 0) \ |
| 432 | return RETURN; \ |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 433 | args = Py_BuildValue PARAM_FORMAT ;\ |
Martin v. Löwis | 1d7c55f | 2001-11-10 13:57:55 +0000 | [diff] [blame] | 434 | if (!args) { flag_error(self); return RETURN;} \ |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 435 | self->in_callback = 1; \ |
Antoine Pitrou | 0ddbf47 | 2014-10-08 20:00:09 +0200 | [diff] [blame] | 436 | rv = call_with_frame(#NAME,__LINE__, \ |
Fred Drake | 39689c5 | 2004-08-13 03:12:57 +0000 | [diff] [blame] | 437 | self->handlers[NAME], args, self); \ |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 438 | self->in_callback = 0; \ |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 439 | Py_DECREF(args); \ |
| 440 | if (rv == NULL) { \ |
| 441 | flag_error(self); \ |
| 442 | return RETURN; \ |
| 443 | } \ |
| 444 | CONVERSION \ |
| 445 | Py_DECREF(rv); \ |
| 446 | } \ |
| 447 | return RETURN; \ |
| 448 | } |
| 449 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 450 | #define VOID_HANDLER(NAME, PARAMS, PARAM_FORMAT) \ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 451 | RC_HANDLER(void, NAME, PARAMS, ;, PARAM_FORMAT, ;, ;,\ |
| 452 | (xmlparseobject *)userData) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 453 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 454 | #define INT_HANDLER(NAME, PARAMS, PARAM_FORMAT)\ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 455 | RC_HANDLER(int, NAME, PARAMS, int rc=0;, PARAM_FORMAT, \ |
| 456 | rc = PyLong_AsLong(rv);, rc, \ |
| 457 | (xmlparseobject *)userData) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 458 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 459 | VOID_HANDLER(EndElement, |
| 460 | (void *userData, const XML_Char *name), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 461 | ("(N)", string_intern(self, name))) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 462 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 463 | VOID_HANDLER(ProcessingInstruction, |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 464 | (void *userData, |
| 465 | const XML_Char *target, |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 466 | const XML_Char *data), |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 467 | ("(NO&)", string_intern(self, target), conv_string_to_unicode ,data)) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 468 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 469 | VOID_HANDLER(UnparsedEntityDecl, |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 470 | (void *userData, |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 471 | const XML_Char *entityName, |
| 472 | const XML_Char *base, |
| 473 | const XML_Char *systemId, |
| 474 | const XML_Char *publicId, |
| 475 | const XML_Char *notationName), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 476 | ("(NNNNN)", |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 477 | string_intern(self, entityName), string_intern(self, base), |
| 478 | string_intern(self, systemId), string_intern(self, publicId), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 479 | string_intern(self, notationName))) |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 480 | |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 481 | VOID_HANDLER(EntityDecl, |
| 482 | (void *userData, |
| 483 | const XML_Char *entityName, |
| 484 | int is_parameter_entity, |
| 485 | const XML_Char *value, |
| 486 | int value_length, |
| 487 | const XML_Char *base, |
| 488 | const XML_Char *systemId, |
| 489 | const XML_Char *publicId, |
| 490 | const XML_Char *notationName), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 491 | ("NiNNNNN", |
| 492 | string_intern(self, entityName), is_parameter_entity, |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 493 | (conv_string_len_to_unicode(value, value_length)), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 494 | string_intern(self, base), string_intern(self, systemId), |
| 495 | string_intern(self, publicId), |
| 496 | string_intern(self, notationName))) |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 497 | |
| 498 | VOID_HANDLER(XmlDecl, |
| 499 | (void *userData, |
| 500 | const XML_Char *version, |
| 501 | const XML_Char *encoding, |
| 502 | int standalone), |
| 503 | ("(O&O&i)", |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 504 | conv_string_to_unicode ,version, conv_string_to_unicode ,encoding, |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 505 | standalone)) |
| 506 | |
| 507 | static PyObject * |
| 508 | conv_content_model(XML_Content * const model, |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 509 | PyObject *(*conv_string)(const XML_Char *)) |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 510 | { |
| 511 | PyObject *result = NULL; |
| 512 | PyObject *children = PyTuple_New(model->numchildren); |
| 513 | int i; |
| 514 | |
| 515 | if (children != NULL) { |
Tim Peters | 9544fc5 | 2001-07-28 09:36:36 +0000 | [diff] [blame] | 516 | assert(model->numchildren < INT_MAX); |
| 517 | for (i = 0; i < (int)model->numchildren; ++i) { |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 518 | PyObject *child = conv_content_model(&model->children[i], |
| 519 | conv_string); |
| 520 | if (child == NULL) { |
| 521 | Py_XDECREF(children); |
| 522 | return NULL; |
| 523 | } |
| 524 | PyTuple_SET_ITEM(children, i, child); |
| 525 | } |
| 526 | result = Py_BuildValue("(iiO&N)", |
| 527 | model->type, model->quant, |
| 528 | conv_string,model->name, children); |
| 529 | } |
| 530 | return result; |
| 531 | } |
| 532 | |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 533 | static void |
| 534 | my_ElementDeclHandler(void *userData, |
| 535 | const XML_Char *name, |
| 536 | XML_Content *model) |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 537 | { |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 538 | xmlparseobject *self = (xmlparseobject *)userData; |
| 539 | PyObject *args = NULL; |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 540 | |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 541 | if (have_handler(self, ElementDecl)) { |
| 542 | PyObject *rv = NULL; |
| 543 | PyObject *modelobj, *nameobj; |
| 544 | |
Victor Stinner | 9e09c26 | 2013-07-18 23:17:01 +0200 | [diff] [blame] | 545 | if (PyErr_Occurred()) |
| 546 | return; |
| 547 | |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 548 | if (flush_character_buffer(self) < 0) |
| 549 | goto finally; |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 550 | modelobj = conv_content_model(model, (conv_string_to_unicode)); |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 551 | if (modelobj == NULL) { |
| 552 | flag_error(self); |
| 553 | goto finally; |
| 554 | } |
| 555 | nameobj = string_intern(self, name); |
| 556 | if (nameobj == NULL) { |
| 557 | Py_DECREF(modelobj); |
| 558 | flag_error(self); |
| 559 | goto finally; |
| 560 | } |
Michael W. Hudson | 0bb8454 | 2004-08-03 11:31:31 +0000 | [diff] [blame] | 561 | args = Py_BuildValue("NN", nameobj, modelobj); |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 562 | if (args == NULL) { |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 563 | flag_error(self); |
| 564 | goto finally; |
| 565 | } |
| 566 | self->in_callback = 1; |
Antoine Pitrou | 0ddbf47 | 2014-10-08 20:00:09 +0200 | [diff] [blame] | 567 | rv = call_with_frame("ElementDecl", __LINE__, |
Fred Drake | 39689c5 | 2004-08-13 03:12:57 +0000 | [diff] [blame] | 568 | self->handlers[ElementDecl], args, self); |
Fred Drake | 06dd8cf | 2003-02-02 03:54:17 +0000 | [diff] [blame] | 569 | self->in_callback = 0; |
| 570 | if (rv == NULL) { |
| 571 | flag_error(self); |
| 572 | goto finally; |
| 573 | } |
| 574 | Py_DECREF(rv); |
| 575 | } |
| 576 | finally: |
| 577 | Py_XDECREF(args); |
| 578 | XML_FreeContentModel(self->itself, model); |
| 579 | return; |
| 580 | } |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 581 | |
| 582 | VOID_HANDLER(AttlistDecl, |
| 583 | (void *userData, |
| 584 | const XML_Char *elname, |
| 585 | const XML_Char *attname, |
| 586 | const XML_Char *att_type, |
| 587 | const XML_Char *dflt, |
| 588 | int isrequired), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 589 | ("(NNO&O&i)", |
| 590 | string_intern(self, elname), string_intern(self, attname), |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 591 | conv_string_to_unicode ,att_type, conv_string_to_unicode ,dflt, |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 592 | isrequired)) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 593 | |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 594 | #if XML_COMBINED_VERSION >= 19504 |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 595 | VOID_HANDLER(SkippedEntity, |
| 596 | (void *userData, |
| 597 | const XML_Char *entityName, |
| 598 | int is_parameter_entity), |
| 599 | ("Ni", |
| 600 | string_intern(self, entityName), is_parameter_entity)) |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 601 | #endif |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 602 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 603 | VOID_HANDLER(NotationDecl, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 604 | (void *userData, |
| 605 | const XML_Char *notationName, |
| 606 | const XML_Char *base, |
| 607 | const XML_Char *systemId, |
| 608 | const XML_Char *publicId), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 609 | ("(NNNN)", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 610 | string_intern(self, notationName), string_intern(self, base), |
| 611 | string_intern(self, systemId), string_intern(self, publicId))) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 612 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 613 | VOID_HANDLER(StartNamespaceDecl, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 614 | (void *userData, |
| 615 | const XML_Char *prefix, |
| 616 | const XML_Char *uri), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 617 | ("(NN)", |
| 618 | string_intern(self, prefix), string_intern(self, uri))) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 619 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 620 | VOID_HANDLER(EndNamespaceDecl, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 621 | (void *userData, |
| 622 | const XML_Char *prefix), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 623 | ("(N)", string_intern(self, prefix))) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 624 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 625 | VOID_HANDLER(Comment, |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 626 | (void *userData, const XML_Char *data), |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 627 | ("(O&)", conv_string_to_unicode ,data)) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 628 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 629 | VOID_HANDLER(StartCdataSection, |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 630 | (void *userData), |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 631 | ("()")) |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 632 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 633 | VOID_HANDLER(EndCdataSection, |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 634 | (void *userData), |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 635 | ("()")) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 636 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 637 | VOID_HANDLER(Default, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 638 | (void *userData, const XML_Char *s, int len), |
| 639 | ("(N)", (conv_string_len_to_unicode(s,len)))) |
Andrew M. Kuchling | beba056 | 2000-06-27 00:33:30 +0000 | [diff] [blame] | 640 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 641 | VOID_HANDLER(DefaultHandlerExpand, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 642 | (void *userData, const XML_Char *s, int len), |
| 643 | ("(N)", (conv_string_len_to_unicode(s,len)))) |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 644 | #define my_DefaultHandlerExpand my_DefaultHandlerExpandHandler |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 645 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 646 | INT_HANDLER(NotStandalone, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 647 | (void *userData), |
| 648 | ("()")) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 649 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 650 | RC_HANDLER(int, ExternalEntityRef, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 651 | (XML_Parser parser, |
| 652 | const XML_Char *context, |
| 653 | const XML_Char *base, |
| 654 | const XML_Char *systemId, |
| 655 | const XML_Char *publicId), |
| 656 | int rc=0;, |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 657 | ("(O&NNN)", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 658 | conv_string_to_unicode ,context, string_intern(self, base), |
| 659 | string_intern(self, systemId), string_intern(self, publicId)), |
| 660 | rc = PyLong_AsLong(rv);, rc, |
| 661 | XML_GetUserData(parser)) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 662 | |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 663 | /* XXX UnknownEncodingHandler */ |
| 664 | |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 665 | VOID_HANDLER(StartDoctypeDecl, |
| 666 | (void *userData, const XML_Char *doctypeName, |
| 667 | const XML_Char *sysid, const XML_Char *pubid, |
| 668 | int has_internal_subset), |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 669 | ("(NNNi)", string_intern(self, doctypeName), |
| 670 | string_intern(self, sysid), string_intern(self, pubid), |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 671 | has_internal_subset)) |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 672 | |
| 673 | VOID_HANDLER(EndDoctypeDecl, (void *userData), ("()")) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 674 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 675 | /* ---------------------------------------------------------------- */ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 676 | /*[clinic input] |
| 677 | class pyexpat.xmlparser "xmlparseobject *" "&Xmlparsetype" |
| 678 | [clinic start generated code]*/ |
| 679 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=2393162385232e1c]*/ |
| 680 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 681 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 682 | static PyObject * |
| 683 | get_parse_result(xmlparseobject *self, int rv) |
| 684 | { |
| 685 | if (PyErr_Occurred()) { |
| 686 | return NULL; |
| 687 | } |
| 688 | if (rv == 0) { |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 689 | return set_error(self, XML_GetErrorCode(self->itself)); |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 690 | } |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 691 | if (flush_character_buffer(self) < 0) { |
| 692 | return NULL; |
| 693 | } |
Christian Heimes | 217cfd1 | 2007-12-02 14:31:20 +0000 | [diff] [blame] | 694 | return PyLong_FromLong(rv); |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 695 | } |
| 696 | |
Serhiy Storchaka | 43536e9 | 2013-02-04 18:26:15 +0200 | [diff] [blame] | 697 | #define MAX_CHUNK_SIZE (1 << 20) |
| 698 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 699 | /*[clinic input] |
| 700 | pyexpat.xmlparser.Parse |
| 701 | |
| 702 | data: object |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 703 | isfinal: bool(accept={int}) = False |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 704 | / |
| 705 | |
| 706 | Parse XML data. |
| 707 | |
| 708 | `isfinal' should be true at end of input. |
| 709 | [clinic start generated code]*/ |
| 710 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 711 | static PyObject * |
Larry Hastings | 89964c4 | 2015-04-14 18:07:59 -0400 | [diff] [blame] | 712 | pyexpat_xmlparser_Parse_impl(xmlparseobject *self, PyObject *data, |
Serhiy Storchaka | 8b2e8b6 | 2015-05-30 11:30:39 +0300 | [diff] [blame] | 713 | int isfinal) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 714 | /*[clinic end generated code: output=f4db843dd1f4ed4b input=eb616027bfa9847f]*/ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 715 | { |
Serhiy Storchaka | 43536e9 | 2013-02-04 18:26:15 +0200 | [diff] [blame] | 716 | const char *s; |
| 717 | Py_ssize_t slen; |
| 718 | Py_buffer view; |
| 719 | int rc; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 720 | |
Serhiy Storchaka | 43536e9 | 2013-02-04 18:26:15 +0200 | [diff] [blame] | 721 | if (PyUnicode_Check(data)) { |
Serhiy Storchaka | 43536e9 | 2013-02-04 18:26:15 +0200 | [diff] [blame] | 722 | view.buf = NULL; |
Serhiy Storchaka | 36b365c | 2013-02-04 18:28:01 +0200 | [diff] [blame] | 723 | s = PyUnicode_AsUTF8AndSize(data, &slen); |
| 724 | if (s == NULL) |
| 725 | return NULL; |
Serhiy Storchaka | 43536e9 | 2013-02-04 18:26:15 +0200 | [diff] [blame] | 726 | /* Explicitly set UTF-8 encoding. Return code ignored. */ |
| 727 | (void)XML_SetEncoding(self->itself, "utf-8"); |
| 728 | } |
| 729 | else { |
| 730 | if (PyObject_GetBuffer(data, &view, PyBUF_SIMPLE) < 0) |
| 731 | return NULL; |
| 732 | s = view.buf; |
| 733 | slen = view.len; |
| 734 | } |
| 735 | |
| 736 | while (slen > MAX_CHUNK_SIZE) { |
| 737 | rc = XML_Parse(self->itself, s, MAX_CHUNK_SIZE, 0); |
| 738 | if (!rc) |
| 739 | goto done; |
| 740 | s += MAX_CHUNK_SIZE; |
| 741 | slen -= MAX_CHUNK_SIZE; |
| 742 | } |
Serhiy Storchaka | fad85aa | 2015-11-07 15:42:38 +0200 | [diff] [blame] | 743 | Py_BUILD_ASSERT(MAX_CHUNK_SIZE <= INT_MAX); |
| 744 | assert(slen <= INT_MAX); |
Serhiy Storchaka | 8b2e8b6 | 2015-05-30 11:30:39 +0300 | [diff] [blame] | 745 | rc = XML_Parse(self->itself, s, (int)slen, isfinal); |
Serhiy Storchaka | 43536e9 | 2013-02-04 18:26:15 +0200 | [diff] [blame] | 746 | |
| 747 | done: |
| 748 | if (view.buf != NULL) |
| 749 | PyBuffer_Release(&view); |
| 750 | return get_parse_result(self, rc); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 751 | } |
| 752 | |
Fred Drake | ca1f426 | 2000-09-21 20:10:23 +0000 | [diff] [blame] | 753 | /* File reading copied from cPickle */ |
| 754 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 755 | #define BUF_SIZE 2048 |
| 756 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 757 | static int |
| 758 | readinst(char *buf, int buf_size, PyObject *meth) |
| 759 | { |
Victor Stinner | 95f1dfc | 2011-01-10 23:00:36 +0000 | [diff] [blame] | 760 | PyObject *str; |
| 761 | Py_ssize_t len; |
Serhiy Storchaka | ef1585e | 2015-12-25 20:01:53 +0200 | [diff] [blame] | 762 | const char *ptr; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 763 | |
Victor Stinner | 95f1dfc | 2011-01-10 23:00:36 +0000 | [diff] [blame] | 764 | str = PyObject_CallFunction(meth, "n", buf_size); |
Martin v. Löwis | 9171f02 | 2004-10-13 19:50:11 +0000 | [diff] [blame] | 765 | if (str == NULL) |
Victor Stinner | 95f1dfc | 2011-01-10 23:00:36 +0000 | [diff] [blame] | 766 | goto error; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 767 | |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 768 | if (PyBytes_Check(str)) |
| 769 | ptr = PyBytes_AS_STRING(str); |
Christian Heimes | 9c4756e | 2008-05-26 13:22:05 +0000 | [diff] [blame] | 770 | else if (PyByteArray_Check(str)) |
| 771 | ptr = PyByteArray_AS_STRING(str); |
Guido van Rossum | 98297ee | 2007-11-06 21:34:58 +0000 | [diff] [blame] | 772 | else { |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 773 | PyErr_Format(PyExc_TypeError, |
Guido van Rossum | 4ca9471 | 2007-07-23 17:42:32 +0000 | [diff] [blame] | 774 | "read() did not return a bytes object (type=%.400s)", |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 775 | Py_TYPE(str)->tp_name); |
Victor Stinner | 95f1dfc | 2011-01-10 23:00:36 +0000 | [diff] [blame] | 776 | goto error; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 777 | } |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 778 | len = Py_SIZE(str); |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 779 | if (len > buf_size) { |
| 780 | PyErr_Format(PyExc_ValueError, |
| 781 | "read() returned too much data: " |
Victor Stinner | 9d6f936 | 2011-01-04 22:00:04 +0000 | [diff] [blame] | 782 | "%i bytes requested, %zd returned", |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 783 | buf_size, len); |
Victor Stinner | 95f1dfc | 2011-01-10 23:00:36 +0000 | [diff] [blame] | 784 | goto error; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 785 | } |
Guido van Rossum | 98297ee | 2007-11-06 21:34:58 +0000 | [diff] [blame] | 786 | memcpy(buf, ptr, len); |
Victor Stinner | 95f1dfc | 2011-01-10 23:00:36 +0000 | [diff] [blame] | 787 | Py_DECREF(str); |
| 788 | /* len <= buf_size <= INT_MAX */ |
Victor Stinner | 0fcab4a | 2011-01-04 12:59:15 +0000 | [diff] [blame] | 789 | return (int)len; |
Victor Stinner | 95f1dfc | 2011-01-10 23:00:36 +0000 | [diff] [blame] | 790 | |
| 791 | error: |
| 792 | Py_XDECREF(str); |
| 793 | return -1; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 794 | } |
| 795 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 796 | /*[clinic input] |
| 797 | pyexpat.xmlparser.ParseFile |
| 798 | |
| 799 | file: object |
| 800 | / |
| 801 | |
| 802 | Parse XML data from file-like object. |
| 803 | [clinic start generated code]*/ |
| 804 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 805 | static PyObject * |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 806 | pyexpat_xmlparser_ParseFile(xmlparseobject *self, PyObject *file) |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 807 | /*[clinic end generated code: output=2adc6a13100cc42b input=fbb5a12b6038d735]*/ |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 808 | { |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 809 | int rv = 1; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 810 | PyObject *readmethod = NULL; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 811 | _Py_IDENTIFIER(read); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 812 | |
Serhiy Storchaka | 41c57b3 | 2019-09-01 12:03:39 +0300 | [diff] [blame] | 813 | if (_PyObject_LookupAttrId(file, &PyId_read, &readmethod) < 0) { |
| 814 | return NULL; |
| 815 | } |
Benjamin Peterson | 4e7f285 | 2010-08-08 16:54:58 +0000 | [diff] [blame] | 816 | if (readmethod == NULL) { |
Benjamin Peterson | 4e7f285 | 2010-08-08 16:54:58 +0000 | [diff] [blame] | 817 | PyErr_SetString(PyExc_TypeError, |
| 818 | "argument must have 'read' attribute"); |
| 819 | return NULL; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 820 | } |
| 821 | for (;;) { |
| 822 | int bytes_read; |
| 823 | void *buf = XML_GetBuffer(self->itself, BUF_SIZE); |
Fred Drake | 7b6caff | 2003-07-21 17:05:56 +0000 | [diff] [blame] | 824 | if (buf == NULL) { |
Fred Drake | f239c6d | 2003-07-21 17:22:43 +0000 | [diff] [blame] | 825 | Py_XDECREF(readmethod); |
Ned Deily | e7d532f | 2014-03-27 16:39:58 -0700 | [diff] [blame] | 826 | return get_parse_result(self, 0); |
Fred Drake | 7b6caff | 2003-07-21 17:05:56 +0000 | [diff] [blame] | 827 | } |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 828 | |
Benjamin Peterson | 4e7f285 | 2010-08-08 16:54:58 +0000 | [diff] [blame] | 829 | bytes_read = readinst(buf, BUF_SIZE, readmethod); |
| 830 | if (bytes_read < 0) { |
| 831 | Py_DECREF(readmethod); |
| 832 | return NULL; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 833 | } |
| 834 | rv = XML_ParseBuffer(self->itself, bytes_read, bytes_read == 0); |
Fred Drake | 7b6caff | 2003-07-21 17:05:56 +0000 | [diff] [blame] | 835 | if (PyErr_Occurred()) { |
| 836 | Py_XDECREF(readmethod); |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 837 | return NULL; |
Fred Drake | 7b6caff | 2003-07-21 17:05:56 +0000 | [diff] [blame] | 838 | } |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 839 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 840 | if (!rv || bytes_read == 0) |
| 841 | break; |
| 842 | } |
Fred Drake | 7b6caff | 2003-07-21 17:05:56 +0000 | [diff] [blame] | 843 | Py_XDECREF(readmethod); |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 844 | return get_parse_result(self, rv); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 845 | } |
| 846 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 847 | /*[clinic input] |
| 848 | pyexpat.xmlparser.SetBase |
| 849 | |
| 850 | base: str |
| 851 | / |
| 852 | |
| 853 | Set the base URL for the parser. |
| 854 | [clinic start generated code]*/ |
| 855 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 856 | static PyObject * |
| 857 | pyexpat_xmlparser_SetBase_impl(xmlparseobject *self, const char *base) |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 858 | /*[clinic end generated code: output=c212ddceb607b539 input=c684e5de895ee1a8]*/ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 859 | { |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 860 | if (!XML_SetBase(self->itself, base)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 861 | return PyErr_NoMemory(); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 862 | } |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 863 | Py_RETURN_NONE; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 864 | } |
| 865 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 866 | /*[clinic input] |
| 867 | pyexpat.xmlparser.GetBase |
| 868 | |
| 869 | Return base URL string for the parser. |
| 870 | [clinic start generated code]*/ |
| 871 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 872 | static PyObject * |
| 873 | pyexpat_xmlparser_GetBase_impl(xmlparseobject *self) |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 874 | /*[clinic end generated code: output=2886cb21f9a8739a input=918d71c38009620e]*/ |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 875 | { |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 876 | return Py_BuildValue("z", XML_GetBase(self->itself)); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 877 | } |
| 878 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 879 | /*[clinic input] |
| 880 | pyexpat.xmlparser.GetInputContext |
| 881 | |
| 882 | Return the untranslated text of the input that caused the current event. |
| 883 | |
| 884 | If the event was generated by a large amount of text (such as a start tag |
| 885 | for an element with many attributes), not all of the text may be available. |
| 886 | [clinic start generated code]*/ |
| 887 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 888 | static PyObject * |
| 889 | pyexpat_xmlparser_GetInputContext_impl(xmlparseobject *self) |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 890 | /*[clinic end generated code: output=a88026d683fc22cc input=034df8712db68379]*/ |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 891 | { |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 892 | if (self->in_callback) { |
| 893 | int offset, size; |
| 894 | const char *buffer |
| 895 | = XML_GetInputContext(self->itself, &offset, &size); |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 896 | |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 897 | if (buffer != NULL) |
Christian Heimes | 72b710a | 2008-05-26 13:28:38 +0000 | [diff] [blame] | 898 | return PyBytes_FromStringAndSize(buffer + offset, |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 899 | size - offset); |
| 900 | else |
| 901 | Py_RETURN_NONE; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 902 | } |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 903 | else |
| 904 | Py_RETURN_NONE; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 905 | } |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 906 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 907 | /*[clinic input] |
| 908 | pyexpat.xmlparser.ExternalEntityParserCreate |
| 909 | |
Larry Hastings | dbfdc38 | 2015-05-04 06:59:46 -0700 | [diff] [blame] | 910 | context: str(accept={str, NoneType}) |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 911 | encoding: str = NULL |
| 912 | / |
| 913 | |
| 914 | Create a parser for parsing an external entity based on the information passed to the ExternalEntityRefHandler. |
| 915 | [clinic start generated code]*/ |
| 916 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 917 | static PyObject * |
Larry Hastings | 89964c4 | 2015-04-14 18:07:59 -0400 | [diff] [blame] | 918 | pyexpat_xmlparser_ExternalEntityParserCreate_impl(xmlparseobject *self, |
| 919 | const char *context, |
| 920 | const char *encoding) |
Larry Hastings | dbfdc38 | 2015-05-04 06:59:46 -0700 | [diff] [blame] | 921 | /*[clinic end generated code: output=535cda9d7a0fbcd6 input=b906714cc122c322]*/ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 922 | { |
Lars Gustäbel | 4a30a07 | 2000-09-24 20:50:52 +0000 | [diff] [blame] | 923 | xmlparseobject *new_parser; |
| 924 | int i; |
| 925 | |
Martin v. Löwis | 894258c | 2001-09-23 10:20:10 +0000 | [diff] [blame] | 926 | new_parser = PyObject_GC_New(xmlparseobject, &Xmlparsetype); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 927 | if (new_parser == NULL) |
| 928 | return NULL; |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 929 | new_parser->buffer_size = self->buffer_size; |
| 930 | new_parser->buffer_used = 0; |
Victor Stinner | b4ba986 | 2010-09-10 22:25:19 +0000 | [diff] [blame] | 931 | new_parser->buffer = NULL; |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 932 | new_parser->ordered_attributes = self->ordered_attributes; |
| 933 | new_parser->specified_attributes = self->specified_attributes; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 934 | new_parser->in_callback = 0; |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 935 | new_parser->ns_prefixes = self->ns_prefixes; |
Lars Gustäbel | 4a30a07 | 2000-09-24 20:50:52 +0000 | [diff] [blame] | 936 | new_parser->itself = XML_ExternalEntityParserCreate(self->itself, context, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 937 | encoding); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 938 | new_parser->handlers = 0; |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 939 | new_parser->intern = self->intern; |
| 940 | Py_XINCREF(new_parser->intern); |
Martin v. Löwis | 894258c | 2001-09-23 10:20:10 +0000 | [diff] [blame] | 941 | PyObject_GC_Track(new_parser); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 942 | |
Victor Stinner | b4ba986 | 2010-09-10 22:25:19 +0000 | [diff] [blame] | 943 | if (self->buffer != NULL) { |
Victor Stinner | b640491 | 2013-07-07 16:21:41 +0200 | [diff] [blame] | 944 | new_parser->buffer = PyMem_Malloc(new_parser->buffer_size); |
Victor Stinner | b4ba986 | 2010-09-10 22:25:19 +0000 | [diff] [blame] | 945 | if (new_parser->buffer == NULL) { |
| 946 | Py_DECREF(new_parser); |
| 947 | return PyErr_NoMemory(); |
| 948 | } |
| 949 | } |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 950 | if (!new_parser->itself) { |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 951 | Py_DECREF(new_parser); |
| 952 | return PyErr_NoMemory(); |
Lars Gustäbel | 4a30a07 | 2000-09-24 20:50:52 +0000 | [diff] [blame] | 953 | } |
| 954 | |
| 955 | XML_SetUserData(new_parser->itself, (void *)new_parser); |
| 956 | |
| 957 | /* allocate and clear handlers first */ |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 958 | for (i = 0; handler_info[i].name != NULL; i++) |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 959 | /* do nothing */; |
Lars Gustäbel | 4a30a07 | 2000-09-24 20:50:52 +0000 | [diff] [blame] | 960 | |
Serhiy Storchaka | 1a1ff29 | 2015-02-16 13:28:22 +0200 | [diff] [blame] | 961 | new_parser->handlers = PyMem_New(PyObject *, i); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 962 | if (!new_parser->handlers) { |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 963 | Py_DECREF(new_parser); |
| 964 | return PyErr_NoMemory(); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 965 | } |
Martin v. Löwis | 5b68ce3 | 2001-10-21 08:53:52 +0000 | [diff] [blame] | 966 | clear_handlers(new_parser, 1); |
Lars Gustäbel | 4a30a07 | 2000-09-24 20:50:52 +0000 | [diff] [blame] | 967 | |
| 968 | /* then copy handlers from self */ |
| 969 | for (i = 0; handler_info[i].name != NULL; i++) { |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 970 | PyObject *handler = self->handlers[i]; |
| 971 | if (handler != NULL) { |
| 972 | Py_INCREF(handler); |
| 973 | new_parser->handlers[i] = handler; |
| 974 | handler_info[i].setter(new_parser->itself, |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 975 | handler_info[i].handler); |
| 976 | } |
Lars Gustäbel | 4a30a07 | 2000-09-24 20:50:52 +0000 | [diff] [blame] | 977 | } |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 978 | return (PyObject *)new_parser; |
Lars Gustäbel | 4a30a07 | 2000-09-24 20:50:52 +0000 | [diff] [blame] | 979 | } |
| 980 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 981 | /*[clinic input] |
| 982 | pyexpat.xmlparser.SetParamEntityParsing |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 983 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 984 | flag: int |
| 985 | / |
| 986 | |
| 987 | Controls parsing of parameter entities (including the external DTD subset). |
| 988 | |
| 989 | Possible flag values are XML_PARAM_ENTITY_PARSING_NEVER, |
| 990 | XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE and |
| 991 | XML_PARAM_ENTITY_PARSING_ALWAYS. Returns true if setting the flag |
| 992 | was successful. |
| 993 | [clinic start generated code]*/ |
| 994 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 995 | static PyObject * |
| 996 | pyexpat_xmlparser_SetParamEntityParsing_impl(xmlparseobject *self, int flag) |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 997 | /*[clinic end generated code: output=18668ee8e760d64c input=8aea19b4b15e9af1]*/ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 998 | { |
| 999 | flag = XML_SetParamEntityParsing(self->itself, flag); |
Christian Heimes | 217cfd1 | 2007-12-02 14:31:20 +0000 | [diff] [blame] | 1000 | return PyLong_FromLong(flag); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1001 | } |
| 1002 | |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1003 | |
| 1004 | #if XML_COMBINED_VERSION >= 19505 |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1005 | /*[clinic input] |
| 1006 | pyexpat.xmlparser.UseForeignDTD |
| 1007 | |
| 1008 | flag: bool = True |
| 1009 | / |
| 1010 | |
| 1011 | Allows the application to provide an artificial external subset if one is not specified as part of the document instance. |
| 1012 | |
| 1013 | This readily allows the use of a 'default' document type controlled by the |
| 1014 | application, while still getting the advantage of providing document type |
| 1015 | information to the parser. 'flag' defaults to True if not provided. |
| 1016 | [clinic start generated code]*/ |
| 1017 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1018 | static PyObject * |
| 1019 | pyexpat_xmlparser_UseForeignDTD_impl(xmlparseobject *self, int flag) |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 1020 | /*[clinic end generated code: output=cfaa9aa50bb0f65c input=78144c519d116a6e]*/ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1021 | { |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 1022 | enum XML_Error rc; |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1023 | |
Antoine Pitrou | 6f430e4 | 2012-08-15 23:18:25 +0200 | [diff] [blame] | 1024 | rc = XML_UseForeignDTD(self->itself, flag ? XML_TRUE : XML_FALSE); |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 1025 | if (rc != XML_ERROR_NONE) { |
| 1026 | return set_error(self, rc); |
| 1027 | } |
Serhiy Storchaka | 228b12e | 2017-01-23 09:47:21 +0200 | [diff] [blame] | 1028 | Py_RETURN_NONE; |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 1029 | } |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1030 | #endif |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 1031 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1032 | static struct PyMethodDef xmlparse_methods[] = { |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1033 | PYEXPAT_XMLPARSER_PARSE_METHODDEF |
| 1034 | PYEXPAT_XMLPARSER_PARSEFILE_METHODDEF |
| 1035 | PYEXPAT_XMLPARSER_SETBASE_METHODDEF |
| 1036 | PYEXPAT_XMLPARSER_GETBASE_METHODDEF |
| 1037 | PYEXPAT_XMLPARSER_GETINPUTCONTEXT_METHODDEF |
| 1038 | PYEXPAT_XMLPARSER_EXTERNALENTITYPARSERCREATE_METHODDEF |
| 1039 | PYEXPAT_XMLPARSER_SETPARAMENTITYPARSING_METHODDEF |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1040 | #if XML_COMBINED_VERSION >= 19505 |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1041 | PYEXPAT_XMLPARSER_USEFOREIGNDTD_METHODDEF |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1042 | #endif |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1043 | {NULL, NULL} /* sentinel */ |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1044 | }; |
| 1045 | |
| 1046 | /* ---------- */ |
| 1047 | |
| 1048 | |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1049 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1050 | /* pyexpat international encoding support. |
| 1051 | Make it as simple as possible. |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1052 | */ |
| 1053 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1054 | static int |
| 1055 | PyUnknownEncodingHandler(void *encodingHandlerData, |
| 1056 | const XML_Char *name, |
| 1057 | XML_Encoding *info) |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1058 | { |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1059 | static unsigned char template_buffer[256] = {0}; |
| 1060 | PyObject* u; |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1061 | int i; |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 1062 | void *data; |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1063 | unsigned int kind; |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1064 | |
Victor Stinner | 9e09c26 | 2013-07-18 23:17:01 +0200 | [diff] [blame] | 1065 | if (PyErr_Occurred()) |
| 1066 | return XML_STATUS_ERROR; |
| 1067 | |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1068 | if (template_buffer[1] == 0) { |
| 1069 | for (i = 0; i < 256; i++) |
| 1070 | template_buffer[i] = i; |
Tim Peters | 63cb99e | 2001-02-17 18:12:50 +0000 | [diff] [blame] | 1071 | } |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1072 | |
| 1073 | u = PyUnicode_Decode((char*) template_buffer, 256, name, "replace"); |
Christian Heimes | b582155 | 2013-06-29 20:43:13 +0200 | [diff] [blame] | 1074 | if (u == NULL || PyUnicode_READY(u)) { |
Christian Heimes | 7217242 | 2013-06-29 21:49:27 +0200 | [diff] [blame] | 1075 | Py_XDECREF(u); |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1076 | return XML_STATUS_ERROR; |
Christian Heimes | b582155 | 2013-06-29 20:43:13 +0200 | [diff] [blame] | 1077 | } |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1078 | |
| 1079 | if (PyUnicode_GET_LENGTH(u) != 256) { |
| 1080 | Py_DECREF(u); |
| 1081 | PyErr_SetString(PyExc_ValueError, |
| 1082 | "multi-byte encodings are not supported"); |
| 1083 | return XML_STATUS_ERROR; |
| 1084 | } |
| 1085 | |
| 1086 | kind = PyUnicode_KIND(u); |
| 1087 | data = PyUnicode_DATA(u); |
| 1088 | for (i = 0; i < 256; i++) { |
| 1089 | Py_UCS4 ch = PyUnicode_READ(kind, data, i); |
| 1090 | if (ch != Py_UNICODE_REPLACEMENT_CHARACTER) |
| 1091 | info->map[i] = ch; |
| 1092 | else |
| 1093 | info->map[i] = -1; |
| 1094 | } |
| 1095 | |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1096 | info->data = NULL; |
| 1097 | info->convert = NULL; |
| 1098 | info->release = NULL; |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1099 | Py_DECREF(u); |
| 1100 | |
| 1101 | return XML_STATUS_OK; |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1102 | } |
| 1103 | |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1104 | |
| 1105 | static PyObject * |
Serhiy Storchaka | 20b39b2 | 2014-09-28 11:27:24 +0300 | [diff] [blame] | 1106 | newxmlparseobject(const char *encoding, const char *namespace_separator, PyObject *intern) |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1107 | { |
| 1108 | int i; |
| 1109 | xmlparseobject *self; |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1110 | |
Martin v. Löwis | 894258c | 2001-09-23 10:20:10 +0000 | [diff] [blame] | 1111 | self = PyObject_GC_New(xmlparseobject, &Xmlparsetype); |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1112 | if (self == NULL) |
| 1113 | return NULL; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1114 | |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 1115 | self->buffer = NULL; |
| 1116 | self->buffer_size = CHARACTER_DATA_BUFFER_SIZE; |
| 1117 | self->buffer_used = 0; |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1118 | self->ordered_attributes = 0; |
| 1119 | self->specified_attributes = 0; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 1120 | self->in_callback = 0; |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 1121 | self->ns_prefixes = 0; |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1122 | self->handlers = NULL; |
Victor Stinner | 54b2d2e | 2013-07-15 17:15:57 +0200 | [diff] [blame] | 1123 | self->intern = intern; |
| 1124 | Py_XINCREF(self->intern); |
| 1125 | PyObject_GC_Track(self); |
| 1126 | |
Christian Heimes | fa535f5 | 2013-07-07 17:35:11 +0200 | [diff] [blame] | 1127 | /* namespace_separator is either NULL or contains one char + \0 */ |
| 1128 | self->itself = XML_ParserCreate_MM(encoding, &ExpatMemoryHandler, |
| 1129 | namespace_separator); |
Victor Stinner | 54b2d2e | 2013-07-15 17:15:57 +0200 | [diff] [blame] | 1130 | if (self->itself == NULL) { |
| 1131 | PyErr_SetString(PyExc_RuntimeError, |
| 1132 | "XML_ParserCreate failed"); |
| 1133 | Py_DECREF(self); |
| 1134 | return NULL; |
| 1135 | } |
Victor Stinner | 23ec4b5 | 2017-06-15 00:54:36 +0200 | [diff] [blame] | 1136 | #if XML_COMBINED_VERSION >= 20100 |
| 1137 | /* This feature was added upstream in libexpat 2.1.0. */ |
Gregory P. Smith | 8e91cf6 | 2012-03-14 14:26:55 -0700 | [diff] [blame] | 1138 | XML_SetHashSalt(self->itself, |
Christian Heimes | 985ecdc | 2013-11-20 11:46:18 +0100 | [diff] [blame] | 1139 | (unsigned long)_Py_HashSecret.expat.hashsalt); |
Gregory P. Smith | 2522771 | 2012-03-14 18:10:37 -0700 | [diff] [blame] | 1140 | #endif |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1141 | XML_SetUserData(self->itself, (void *)self); |
Fred Drake | 7c75bf2 | 2002-07-01 14:02:31 +0000 | [diff] [blame] | 1142 | XML_SetUnknownEncodingHandler(self->itself, |
| 1143 | (XML_UnknownEncodingHandler) PyUnknownEncodingHandler, NULL); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1144 | |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 1145 | for (i = 0; handler_info[i].name != NULL; i++) |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1146 | /* do nothing */; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1147 | |
Serhiy Storchaka | 1a1ff29 | 2015-02-16 13:28:22 +0200 | [diff] [blame] | 1148 | self->handlers = PyMem_New(PyObject *, i); |
Fred Drake | 7c75bf2 | 2002-07-01 14:02:31 +0000 | [diff] [blame] | 1149 | if (!self->handlers) { |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1150 | Py_DECREF(self); |
| 1151 | return PyErr_NoMemory(); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1152 | } |
Martin v. Löwis | 5b68ce3 | 2001-10-21 08:53:52 +0000 | [diff] [blame] | 1153 | clear_handlers(self, 1); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1154 | |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1155 | return (PyObject*)self; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1156 | } |
| 1157 | |
| 1158 | |
| 1159 | static void |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1160 | xmlparse_dealloc(xmlparseobject *self) |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1161 | { |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1162 | int i; |
Martin v. Löwis | 894258c | 2001-09-23 10:20:10 +0000 | [diff] [blame] | 1163 | PyObject_GC_UnTrack(self); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1164 | if (self->itself != NULL) |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1165 | XML_ParserFree(self->itself); |
| 1166 | self->itself = NULL; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1167 | |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1168 | if (self->handlers != NULL) { |
Serhiy Storchaka | 1ed017a | 2015-12-27 15:51:32 +0200 | [diff] [blame] | 1169 | for (i = 0; handler_info[i].name != NULL; i++) |
| 1170 | Py_CLEAR(self->handlers[i]); |
Victor Stinner | b640491 | 2013-07-07 16:21:41 +0200 | [diff] [blame] | 1171 | PyMem_Free(self->handlers); |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1172 | self->handlers = NULL; |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1173 | } |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 1174 | if (self->buffer != NULL) { |
Victor Stinner | b640491 | 2013-07-07 16:21:41 +0200 | [diff] [blame] | 1175 | PyMem_Free(self->buffer); |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 1176 | self->buffer = NULL; |
| 1177 | } |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1178 | Py_XDECREF(self->intern); |
Martin v. Löwis | 894258c | 2001-09-23 10:20:10 +0000 | [diff] [blame] | 1179 | PyObject_GC_Del(self); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1180 | } |
| 1181 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1182 | |
| 1183 | static PyObject * |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1184 | xmlparse_handler_getter(xmlparseobject *self, struct HandlerInfo *hi) |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1185 | { |
Victor Stinner | 28f468c | 2018-11-22 13:21:43 +0100 | [diff] [blame] | 1186 | assert((hi - handler_info) < (Py_ssize_t)Py_ARRAY_LENGTH(handler_info)); |
| 1187 | int handlernum = (int)(hi - handler_info); |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1188 | PyObject *result = self->handlers[handlernum]; |
| 1189 | if (result == NULL) |
| 1190 | result = Py_None; |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1191 | Py_INCREF(result); |
| 1192 | return result; |
| 1193 | } |
| 1194 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1195 | static int |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1196 | xmlparse_handler_setter(xmlparseobject *self, PyObject *v, struct HandlerInfo *hi) |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1197 | { |
Victor Stinner | 28f468c | 2018-11-22 13:21:43 +0100 | [diff] [blame] | 1198 | assert((hi - handler_info) < (Py_ssize_t)Py_ARRAY_LENGTH(handler_info)); |
| 1199 | int handlernum = (int)(hi - handler_info); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1200 | if (v == NULL) { |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1201 | PyErr_SetString(PyExc_RuntimeError, "Cannot delete attribute"); |
| 1202 | return -1; |
| 1203 | } |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1204 | if (handlernum == CharacterData) { |
Fred Drake | 2a3d7db | 2002-06-28 22:56:48 +0000 | [diff] [blame] | 1205 | /* If we're changing the character data handler, flush all |
| 1206 | * cached data with the old handler. Not sure there's a |
| 1207 | * "right" thing to do, though, but this probably won't |
| 1208 | * happen. |
| 1209 | */ |
| 1210 | if (flush_character_buffer(self) < 0) |
| 1211 | return -1; |
| 1212 | } |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1213 | |
| 1214 | xmlhandler c_handler = NULL; |
| 1215 | if (v == Py_None) { |
| 1216 | /* If this is the character data handler, and a character |
| 1217 | data handler is already active, we need to be more |
| 1218 | careful. What we can safely do is replace the existing |
| 1219 | character data handler callback function with a no-op |
| 1220 | function that will refuse to call Python. The downside |
| 1221 | is that this doesn't completely remove the character |
| 1222 | data handler from the C layer if there's any callback |
| 1223 | active, so Expat does a little more work than it |
| 1224 | otherwise would, but that's really an odd case. A more |
| 1225 | elaborate system of handlers and state could remove the |
| 1226 | C handler more effectively. */ |
| 1227 | if (handlernum == CharacterData && self->in_callback) |
| 1228 | c_handler = noop_character_data_handler; |
| 1229 | v = NULL; |
| 1230 | } |
| 1231 | else if (v != NULL) { |
| 1232 | Py_INCREF(v); |
| 1233 | c_handler = handler_info[handlernum].handler; |
| 1234 | } |
| 1235 | Py_XSETREF(self->handlers[handlernum], v); |
| 1236 | handler_info[handlernum].setter(self->itself, c_handler); |
| 1237 | return 0; |
| 1238 | } |
| 1239 | |
| 1240 | #define INT_GETTER(name) \ |
| 1241 | static PyObject * \ |
| 1242 | xmlparse_##name##_getter(xmlparseobject *self, void *closure) \ |
| 1243 | { \ |
| 1244 | return PyLong_FromLong((long) XML_Get##name(self->itself)); \ |
| 1245 | } |
| 1246 | INT_GETTER(ErrorCode) |
| 1247 | INT_GETTER(ErrorLineNumber) |
| 1248 | INT_GETTER(ErrorColumnNumber) |
| 1249 | INT_GETTER(ErrorByteIndex) |
| 1250 | INT_GETTER(CurrentLineNumber) |
| 1251 | INT_GETTER(CurrentColumnNumber) |
| 1252 | INT_GETTER(CurrentByteIndex) |
| 1253 | |
| 1254 | #undef INT_GETTER |
| 1255 | |
| 1256 | static PyObject * |
| 1257 | xmlparse_buffer_text_getter(xmlparseobject *self, void *closure) |
| 1258 | { |
| 1259 | return PyBool_FromLong(self->buffer != NULL); |
| 1260 | } |
| 1261 | |
| 1262 | static int |
| 1263 | xmlparse_buffer_text_setter(xmlparseobject *self, PyObject *v, void *closure) |
| 1264 | { |
| 1265 | if (v == NULL) { |
| 1266 | PyErr_SetString(PyExc_RuntimeError, "Cannot delete attribute"); |
| 1267 | return -1; |
| 1268 | } |
| 1269 | int b = PyObject_IsTrue(v); |
| 1270 | if (b < 0) |
| 1271 | return -1; |
| 1272 | if (b) { |
| 1273 | if (self->buffer == NULL) { |
| 1274 | self->buffer = PyMem_Malloc(self->buffer_size); |
| 1275 | if (self->buffer == NULL) { |
| 1276 | PyErr_NoMemory(); |
| 1277 | return -1; |
| 1278 | } |
| 1279 | self->buffer_used = 0; |
| 1280 | } |
| 1281 | } |
| 1282 | else if (self->buffer != NULL) { |
| 1283 | if (flush_character_buffer(self) < 0) |
| 1284 | return -1; |
| 1285 | PyMem_Free(self->buffer); |
| 1286 | self->buffer = NULL; |
| 1287 | } |
| 1288 | return 0; |
| 1289 | } |
| 1290 | |
| 1291 | static PyObject * |
| 1292 | xmlparse_buffer_size_getter(xmlparseobject *self, void *closure) |
| 1293 | { |
| 1294 | return PyLong_FromLong((long) self->buffer_size); |
| 1295 | } |
| 1296 | |
| 1297 | static int |
| 1298 | xmlparse_buffer_size_setter(xmlparseobject *self, PyObject *v, void *closure) |
| 1299 | { |
| 1300 | if (v == NULL) { |
| 1301 | PyErr_SetString(PyExc_RuntimeError, "Cannot delete attribute"); |
| 1302 | return -1; |
| 1303 | } |
| 1304 | long new_buffer_size; |
| 1305 | if (!PyLong_Check(v)) { |
| 1306 | PyErr_SetString(PyExc_TypeError, "buffer_size must be an integer"); |
| 1307 | return -1; |
| 1308 | } |
| 1309 | |
| 1310 | new_buffer_size = PyLong_AsLong(v); |
| 1311 | if (new_buffer_size <= 0) { |
| 1312 | if (!PyErr_Occurred()) |
| 1313 | PyErr_SetString(PyExc_ValueError, "buffer_size must be greater than zero"); |
| 1314 | return -1; |
| 1315 | } |
| 1316 | |
| 1317 | /* trivial case -- no change */ |
| 1318 | if (new_buffer_size == self->buffer_size) { |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1319 | return 0; |
| 1320 | } |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1321 | |
| 1322 | /* check maximum */ |
| 1323 | if (new_buffer_size > INT_MAX) { |
| 1324 | char errmsg[100]; |
| 1325 | sprintf(errmsg, "buffer_size must not be greater than %i", INT_MAX); |
| 1326 | PyErr_SetString(PyExc_ValueError, errmsg); |
| 1327 | return -1; |
| 1328 | } |
| 1329 | |
| 1330 | if (self->buffer != NULL) { |
| 1331 | /* there is already a buffer */ |
| 1332 | if (self->buffer_used != 0) { |
| 1333 | if (flush_character_buffer(self) < 0) { |
| 1334 | return -1; |
| 1335 | } |
| 1336 | } |
| 1337 | /* free existing buffer */ |
| 1338 | PyMem_Free(self->buffer); |
| 1339 | } |
| 1340 | self->buffer = PyMem_Malloc(new_buffer_size); |
| 1341 | if (self->buffer == NULL) { |
| 1342 | PyErr_NoMemory(); |
| 1343 | return -1; |
| 1344 | } |
| 1345 | self->buffer_size = new_buffer_size; |
| 1346 | return 0; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1347 | } |
| 1348 | |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1349 | static PyObject * |
| 1350 | xmlparse_buffer_used_getter(xmlparseobject *self, void *closure) |
| 1351 | { |
| 1352 | return PyLong_FromLong((long) self->buffer_used); |
| 1353 | } |
| 1354 | |
| 1355 | static PyObject * |
| 1356 | xmlparse_namespace_prefixes_getter(xmlparseobject *self, void *closure) |
| 1357 | { |
| 1358 | return PyBool_FromLong(self->ns_prefixes); |
| 1359 | } |
| 1360 | |
| 1361 | static int |
| 1362 | xmlparse_namespace_prefixes_setter(xmlparseobject *self, PyObject *v, void *closure) |
| 1363 | { |
| 1364 | if (v == NULL) { |
| 1365 | PyErr_SetString(PyExc_RuntimeError, "Cannot delete attribute"); |
| 1366 | return -1; |
| 1367 | } |
| 1368 | int b = PyObject_IsTrue(v); |
| 1369 | if (b < 0) |
| 1370 | return -1; |
| 1371 | self->ns_prefixes = b; |
| 1372 | XML_SetReturnNSTriplet(self->itself, self->ns_prefixes); |
| 1373 | return 0; |
| 1374 | } |
| 1375 | |
| 1376 | static PyObject * |
| 1377 | xmlparse_ordered_attributes_getter(xmlparseobject *self, void *closure) |
| 1378 | { |
| 1379 | return PyBool_FromLong(self->ordered_attributes); |
| 1380 | } |
| 1381 | |
| 1382 | static int |
| 1383 | xmlparse_ordered_attributes_setter(xmlparseobject *self, PyObject *v, void *closure) |
| 1384 | { |
| 1385 | if (v == NULL) { |
| 1386 | PyErr_SetString(PyExc_RuntimeError, "Cannot delete attribute"); |
| 1387 | return -1; |
| 1388 | } |
| 1389 | int b = PyObject_IsTrue(v); |
| 1390 | if (b < 0) |
| 1391 | return -1; |
| 1392 | self->ordered_attributes = b; |
| 1393 | return 0; |
| 1394 | } |
| 1395 | |
| 1396 | static PyObject * |
| 1397 | xmlparse_specified_attributes_getter(xmlparseobject *self, void *closure) |
| 1398 | { |
| 1399 | return PyBool_FromLong((long) self->specified_attributes); |
| 1400 | } |
| 1401 | |
| 1402 | static int |
| 1403 | xmlparse_specified_attributes_setter(xmlparseobject *self, PyObject *v, void *closure) |
| 1404 | { |
| 1405 | if (v == NULL) { |
| 1406 | PyErr_SetString(PyExc_RuntimeError, "Cannot delete attribute"); |
| 1407 | return -1; |
| 1408 | } |
| 1409 | int b = PyObject_IsTrue(v); |
| 1410 | if (b < 0) |
| 1411 | return -1; |
| 1412 | self->specified_attributes = b; |
| 1413 | return 0; |
| 1414 | } |
| 1415 | |
| 1416 | static PyMemberDef xmlparse_members[] = { |
| 1417 | {"intern", T_OBJECT, offsetof(xmlparseobject, intern), READONLY, NULL}, |
| 1418 | {NULL} |
| 1419 | }; |
| 1420 | |
| 1421 | #define XMLPARSE_GETTER_DEF(name) \ |
| 1422 | {#name, (getter)xmlparse_##name##_getter, NULL, NULL}, |
| 1423 | #define XMLPARSE_GETTER_SETTER_DEF(name) \ |
| 1424 | {#name, (getter)xmlparse_##name##_getter, \ |
| 1425 | (setter)xmlparse_##name##_setter, NULL}, |
| 1426 | |
| 1427 | static PyGetSetDef xmlparse_getsetlist[] = { |
| 1428 | XMLPARSE_GETTER_DEF(ErrorCode) |
| 1429 | XMLPARSE_GETTER_DEF(ErrorLineNumber) |
| 1430 | XMLPARSE_GETTER_DEF(ErrorColumnNumber) |
| 1431 | XMLPARSE_GETTER_DEF(ErrorByteIndex) |
| 1432 | XMLPARSE_GETTER_DEF(CurrentLineNumber) |
| 1433 | XMLPARSE_GETTER_DEF(CurrentColumnNumber) |
| 1434 | XMLPARSE_GETTER_DEF(CurrentByteIndex) |
| 1435 | XMLPARSE_GETTER_SETTER_DEF(buffer_size) |
| 1436 | XMLPARSE_GETTER_SETTER_DEF(buffer_text) |
| 1437 | XMLPARSE_GETTER_DEF(buffer_used) |
| 1438 | XMLPARSE_GETTER_SETTER_DEF(namespace_prefixes) |
| 1439 | XMLPARSE_GETTER_SETTER_DEF(ordered_attributes) |
| 1440 | XMLPARSE_GETTER_SETTER_DEF(specified_attributes) |
| 1441 | {NULL}, |
| 1442 | }; |
| 1443 | |
| 1444 | #undef XMLPARSE_GETTER_DEF |
| 1445 | #undef XMLPARSE_GETTER_SETTER_DEF |
| 1446 | |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1447 | static int |
| 1448 | xmlparse_traverse(xmlparseobject *op, visitproc visit, void *arg) |
| 1449 | { |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1450 | int i; |
| 1451 | for (i = 0; handler_info[i].name != NULL; i++) |
| 1452 | Py_VISIT(op->handlers[i]); |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1453 | return 0; |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1454 | } |
| 1455 | |
| 1456 | static int |
| 1457 | xmlparse_clear(xmlparseobject *op) |
| 1458 | { |
Martin v. Löwis | 5b68ce3 | 2001-10-21 08:53:52 +0000 | [diff] [blame] | 1459 | clear_handlers(op, 0); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1460 | Py_CLEAR(op->intern); |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1461 | return 0; |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1462 | } |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1463 | |
Martin v. Löwis | 14f8b4c | 2002-06-13 20:33:02 +0000 | [diff] [blame] | 1464 | PyDoc_STRVAR(Xmlparsetype__doc__, "XML parser"); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1465 | |
| 1466 | static PyTypeObject Xmlparsetype = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1467 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1468 | "pyexpat.xmlparser", /*tp_name*/ |
Antoine Pitrou | 23683ef | 2011-01-04 00:00:31 +0000 | [diff] [blame] | 1469 | sizeof(xmlparseobject), /*tp_basicsize*/ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1470 | 0, /*tp_itemsize*/ |
| 1471 | /* methods */ |
| 1472 | (destructor)xmlparse_dealloc, /*tp_dealloc*/ |
Jeroen Demeyer | 530f506 | 2019-05-31 04:13:39 +0200 | [diff] [blame] | 1473 | 0, /*tp_vectorcall_offset*/ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1474 | 0, /*tp_getattr*/ |
Alexander Belopolsky | e239d23 | 2010-12-08 23:31:48 +0000 | [diff] [blame] | 1475 | 0, /*tp_setattr*/ |
Jeroen Demeyer | 530f506 | 2019-05-31 04:13:39 +0200 | [diff] [blame] | 1476 | 0, /*tp_as_async*/ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1477 | (reprfunc)0, /*tp_repr*/ |
| 1478 | 0, /*tp_as_number*/ |
| 1479 | 0, /*tp_as_sequence*/ |
| 1480 | 0, /*tp_as_mapping*/ |
| 1481 | (hashfunc)0, /*tp_hash*/ |
| 1482 | (ternaryfunc)0, /*tp_call*/ |
| 1483 | (reprfunc)0, /*tp_str*/ |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1484 | (getattrofunc)0, /* tp_getattro */ |
| 1485 | (setattrofunc)0, /* tp_setattro */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1486 | 0, /* tp_as_buffer */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1487 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC, /*tp_flags*/ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1488 | Xmlparsetype__doc__, /* tp_doc - Documentation string */ |
| 1489 | (traverseproc)xmlparse_traverse, /* tp_traverse */ |
| 1490 | (inquiry)xmlparse_clear, /* tp_clear */ |
| 1491 | 0, /* tp_richcompare */ |
| 1492 | 0, /* tp_weaklistoffset */ |
| 1493 | 0, /* tp_iter */ |
| 1494 | 0, /* tp_iternext */ |
| 1495 | xmlparse_methods, /* tp_methods */ |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1496 | xmlparse_members, /* tp_members */ |
| 1497 | xmlparse_getsetlist, /* tp_getset */ |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1498 | }; |
| 1499 | |
| 1500 | /* End of code for xmlparser objects */ |
| 1501 | /* -------------------------------------------------------- */ |
| 1502 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1503 | /*[clinic input] |
| 1504 | pyexpat.ParserCreate |
| 1505 | |
Serhiy Storchaka | 279f446 | 2019-09-14 12:24:05 +0300 | [diff] [blame^] | 1506 | encoding: str(accept={str, NoneType}) = None |
| 1507 | namespace_separator: str(accept={str, NoneType}) = None |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1508 | intern: object = NULL |
| 1509 | |
| 1510 | Return a new XML parser object. |
| 1511 | [clinic start generated code]*/ |
| 1512 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1513 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 1514 | pyexpat_ParserCreate_impl(PyObject *module, const char *encoding, |
Larry Hastings | 89964c4 | 2015-04-14 18:07:59 -0400 | [diff] [blame] | 1515 | const char *namespace_separator, PyObject *intern) |
Serhiy Storchaka | 279f446 | 2019-09-14 12:24:05 +0300 | [diff] [blame^] | 1516 | /*[clinic end generated code: output=295c0cf01ab1146c input=e8da8e8d7122cb5d]*/ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1517 | { |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1518 | PyObject *result; |
| 1519 | int intern_decref = 0; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1520 | |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1521 | if (namespace_separator != NULL |
| 1522 | && strlen(namespace_separator) > 1) { |
| 1523 | PyErr_SetString(PyExc_ValueError, |
| 1524 | "namespace_separator must be at most one" |
| 1525 | " character, omitted, or None"); |
| 1526 | return NULL; |
| 1527 | } |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1528 | /* Explicitly passing None means no interning is desired. |
| 1529 | Not passing anything means that a new dictionary is used. */ |
| 1530 | if (intern == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1531 | intern = NULL; |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1532 | else if (intern == NULL) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1533 | intern = PyDict_New(); |
| 1534 | if (!intern) |
| 1535 | return NULL; |
| 1536 | intern_decref = 1; |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1537 | } |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1538 | else if (!PyDict_Check(intern)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1539 | PyErr_SetString(PyExc_TypeError, "intern must be a dictionary"); |
| 1540 | return NULL; |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1541 | } |
| 1542 | |
Serhiy Storchaka | 20b39b2 | 2014-09-28 11:27:24 +0300 | [diff] [blame] | 1543 | result = newxmlparseobject(encoding, namespace_separator, intern); |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1544 | if (intern_decref) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1545 | Py_DECREF(intern); |
Fred Drake | b91a36b | 2002-06-27 19:40:48 +0000 | [diff] [blame] | 1546 | } |
| 1547 | return result; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1548 | } |
| 1549 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1550 | /*[clinic input] |
| 1551 | pyexpat.ErrorString |
| 1552 | |
| 1553 | code: long |
| 1554 | / |
| 1555 | |
| 1556 | Returns string error for given number. |
| 1557 | [clinic start generated code]*/ |
| 1558 | |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1559 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 1560 | pyexpat_ErrorString_impl(PyObject *module, long code) |
| 1561 | /*[clinic end generated code: output=2feae50d166f2174 input=cc67de010d9e62b3]*/ |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1562 | { |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1563 | return Py_BuildValue("z", XML_ErrorString((int)code)); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1564 | } |
| 1565 | |
| 1566 | /* List of methods defined in the module */ |
| 1567 | |
| 1568 | static struct PyMethodDef pyexpat_methods[] = { |
Brett Cannon | d0aeda8 | 2014-08-22 14:23:20 -0400 | [diff] [blame] | 1569 | PYEXPAT_PARSERCREATE_METHODDEF |
| 1570 | PYEXPAT_ERRORSTRING_METHODDEF |
| 1571 | {NULL, NULL} /* sentinel */ |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1572 | }; |
| 1573 | |
Andrew M. Kuchling | beba056 | 2000-06-27 00:33:30 +0000 | [diff] [blame] | 1574 | /* Module docstring */ |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1575 | |
Martin v. Löwis | 14f8b4c | 2002-06-13 20:33:02 +0000 | [diff] [blame] | 1576 | PyDoc_STRVAR(pyexpat_module_documentation, |
| 1577 | "Python wrapper for Expat parser."); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1578 | |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1579 | /* Initialization function for the module */ |
| 1580 | |
| 1581 | #ifndef MODULE_NAME |
| 1582 | #define MODULE_NAME "pyexpat" |
| 1583 | #endif |
| 1584 | |
| 1585 | #ifndef MODULE_INITFUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1586 | #define MODULE_INITFUNC PyInit_pyexpat |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1587 | #endif |
| 1588 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1589 | static struct PyModuleDef pyexpatmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1590 | PyModuleDef_HEAD_INIT, |
| 1591 | MODULE_NAME, |
| 1592 | pyexpat_module_documentation, |
| 1593 | -1, |
| 1594 | pyexpat_methods, |
| 1595 | NULL, |
| 1596 | NULL, |
| 1597 | NULL, |
| 1598 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1599 | }; |
| 1600 | |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1601 | static int init_handler_descrs(void) |
| 1602 | { |
| 1603 | int i; |
| 1604 | assert(!PyType_HasFeature(&Xmlparsetype, Py_TPFLAGS_VALID_VERSION_TAG)); |
| 1605 | for (i = 0; handler_info[i].name != NULL; i++) { |
| 1606 | struct HandlerInfo *hi = &handler_info[i]; |
| 1607 | hi->getset.name = hi->name; |
| 1608 | hi->getset.get = (getter)xmlparse_handler_getter; |
| 1609 | hi->getset.set = (setter)xmlparse_handler_setter; |
| 1610 | hi->getset.closure = &handler_info[i]; |
| 1611 | |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 1612 | PyObject *descr = PyDescr_NewGetSet(&Xmlparsetype, &hi->getset); |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1613 | if (descr == NULL) |
| 1614 | return -1; |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 1615 | |
| 1616 | if (PyDict_GetItemWithError(Xmlparsetype.tp_dict, PyDescr_NAME(descr))) { |
| 1617 | Py_DECREF(descr); |
| 1618 | continue; |
| 1619 | } |
| 1620 | else if (PyErr_Occurred()) { |
| 1621 | Py_DECREF(descr); |
| 1622 | return -1; |
| 1623 | } |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1624 | if (PyDict_SetItem(Xmlparsetype.tp_dict, PyDescr_NAME(descr), descr) < 0) { |
| 1625 | Py_DECREF(descr); |
| 1626 | return -1; |
| 1627 | } |
| 1628 | Py_DECREF(descr); |
| 1629 | } |
| 1630 | return 0; |
| 1631 | } |
| 1632 | |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 1633 | PyMODINIT_FUNC |
| 1634 | MODULE_INITFUNC(void) |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1635 | { |
| 1636 | PyObject *m, *d; |
Neal Norwitz | 392c5be | 2007-08-25 17:20:32 +0000 | [diff] [blame] | 1637 | PyObject *errmod_name = PyUnicode_FromString(MODULE_NAME ".errors"); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1638 | PyObject *errors_module; |
| 1639 | PyObject *modelmod_name; |
| 1640 | PyObject *model_module; |
Georg Brandl | b4dac71 | 2010-10-15 14:46:48 +0000 | [diff] [blame] | 1641 | PyObject *tmpnum, *tmpstr; |
| 1642 | PyObject *codes_dict; |
| 1643 | PyObject *rev_codes_dict; |
| 1644 | int res; |
Fredrik Lundh | d7a4288 | 2005-12-13 20:43:04 +0000 | [diff] [blame] | 1645 | static struct PyExpat_CAPI capi; |
Georg Brandl | b4dac71 | 2010-10-15 14:46:48 +0000 | [diff] [blame] | 1646 | PyObject *capi_object; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1647 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1648 | if (errmod_name == NULL) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1649 | return NULL; |
Neal Norwitz | 392c5be | 2007-08-25 17:20:32 +0000 | [diff] [blame] | 1650 | modelmod_name = PyUnicode_FromString(MODULE_NAME ".model"); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1651 | if (modelmod_name == NULL) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1652 | return NULL; |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1653 | |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1654 | if (PyType_Ready(&Xmlparsetype) < 0 || init_handler_descrs() < 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1655 | return NULL; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1656 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1657 | /* Create the module and add the functions */ |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1658 | m = PyModule_Create(&pyexpatmodule); |
Neal Norwitz | 1ac754f | 2006-01-19 06:09:39 +0000 | [diff] [blame] | 1659 | if (m == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1660 | return NULL; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1661 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1662 | /* Add some symbolic constants to the module */ |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 1663 | if (ErrorObject == NULL) { |
| 1664 | ErrorObject = PyErr_NewException("xml.parsers.expat.ExpatError", |
Fred Drake | 93adb69 | 2000-09-23 04:55:48 +0000 | [diff] [blame] | 1665 | NULL, NULL); |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 1666 | if (ErrorObject == NULL) |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1667 | return NULL; |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 1668 | } |
| 1669 | Py_INCREF(ErrorObject); |
Fred Drake | 93adb69 | 2000-09-23 04:55:48 +0000 | [diff] [blame] | 1670 | PyModule_AddObject(m, "error", ErrorObject); |
Fred Drake | bd6101c | 2001-02-14 18:29:45 +0000 | [diff] [blame] | 1671 | Py_INCREF(ErrorObject); |
| 1672 | PyModule_AddObject(m, "ExpatError", ErrorObject); |
Fred Drake | 4ba298c | 2000-10-29 04:57:53 +0000 | [diff] [blame] | 1673 | Py_INCREF(&Xmlparsetype); |
| 1674 | PyModule_AddObject(m, "XMLParserType", (PyObject *) &Xmlparsetype); |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1675 | |
Fred Drake | 738293d | 2000-12-21 17:25:07 +0000 | [diff] [blame] | 1676 | PyModule_AddStringConstant(m, "EXPAT_VERSION", |
Serhiy Storchaka | 20b39b2 | 2014-09-28 11:27:24 +0300 | [diff] [blame] | 1677 | XML_ExpatVersion()); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1678 | { |
| 1679 | XML_Expat_Version info = XML_ExpatVersionInfo(); |
| 1680 | PyModule_AddObject(m, "version_info", |
| 1681 | Py_BuildValue("(iii)", info.major, |
| 1682 | info.minor, info.micro)); |
| 1683 | } |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1684 | /* XXX When Expat supports some way of figuring out how it was |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1685 | compiled, this should check and set native_encoding |
| 1686 | appropriately. |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1687 | */ |
Fred Drake | 93adb69 | 2000-09-23 04:55:48 +0000 | [diff] [blame] | 1688 | PyModule_AddStringConstant(m, "native_encoding", "UTF-8"); |
Fred Drake | c23b523 | 2000-08-24 21:57:43 +0000 | [diff] [blame] | 1689 | |
Fred Drake | 93adb69 | 2000-09-23 04:55:48 +0000 | [diff] [blame] | 1690 | d = PyModule_GetDict(m); |
Christian Heimes | 7a5457b | 2016-09-09 00:13:35 +0200 | [diff] [blame] | 1691 | if (d == NULL) { |
| 1692 | Py_DECREF(m); |
| 1693 | return NULL; |
| 1694 | } |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 1695 | errors_module = PyDict_GetItemWithError(d, errmod_name); |
| 1696 | if (errors_module == NULL && !PyErr_Occurred()) { |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1697 | errors_module = PyModule_New(MODULE_NAME ".errors"); |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1698 | if (errors_module != NULL) { |
Eric Snow | 3f9eee6 | 2017-09-15 16:35:20 -0600 | [diff] [blame] | 1699 | _PyImport_SetModule(errmod_name, errors_module); |
Fred Drake | 93adb69 | 2000-09-23 04:55:48 +0000 | [diff] [blame] | 1700 | /* gives away the reference to errors_module */ |
| 1701 | PyModule_AddObject(m, "errors", errors_module); |
Fred Drake | c23b523 | 2000-08-24 21:57:43 +0000 | [diff] [blame] | 1702 | } |
| 1703 | } |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1704 | Py_DECREF(errmod_name); |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 1705 | model_module = PyDict_GetItemWithError(d, modelmod_name); |
| 1706 | if (model_module == NULL && !PyErr_Occurred()) { |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1707 | model_module = PyModule_New(MODULE_NAME ".model"); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1708 | if (model_module != NULL) { |
Eric Snow | 3f9eee6 | 2017-09-15 16:35:20 -0600 | [diff] [blame] | 1709 | _PyImport_SetModule(modelmod_name, model_module); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1710 | /* gives away the reference to model_module */ |
| 1711 | PyModule_AddObject(m, "model", model_module); |
| 1712 | } |
| 1713 | } |
| 1714 | Py_DECREF(modelmod_name); |
Christian Heimes | 7a5457b | 2016-09-09 00:13:35 +0200 | [diff] [blame] | 1715 | if (errors_module == NULL || model_module == NULL) { |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1716 | /* Don't core dump later! */ |
Christian Heimes | 7a5457b | 2016-09-09 00:13:35 +0200 | [diff] [blame] | 1717 | Py_DECREF(m); |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1718 | return NULL; |
Christian Heimes | 7a5457b | 2016-09-09 00:13:35 +0200 | [diff] [blame] | 1719 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1720 | |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1721 | #if XML_COMBINED_VERSION > 19505 |
Martin v. Löwis | 069dde2 | 2003-01-21 10:58:18 +0000 | [diff] [blame] | 1722 | { |
| 1723 | const XML_Feature *features = XML_GetFeatureList(); |
| 1724 | PyObject *list = PyList_New(0); |
| 1725 | if (list == NULL) |
| 1726 | /* just ignore it */ |
| 1727 | PyErr_Clear(); |
| 1728 | else { |
| 1729 | int i = 0; |
| 1730 | for (; features[i].feature != XML_FEATURE_END; ++i) { |
| 1731 | int ok; |
| 1732 | PyObject *item = Py_BuildValue("si", features[i].name, |
| 1733 | features[i].value); |
| 1734 | if (item == NULL) { |
| 1735 | Py_DECREF(list); |
| 1736 | list = NULL; |
| 1737 | break; |
| 1738 | } |
| 1739 | ok = PyList_Append(list, item); |
| 1740 | Py_DECREF(item); |
| 1741 | if (ok < 0) { |
| 1742 | PyErr_Clear(); |
| 1743 | break; |
| 1744 | } |
| 1745 | } |
| 1746 | if (list != NULL) |
| 1747 | PyModule_AddObject(m, "features", list); |
| 1748 | } |
| 1749 | } |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1750 | #endif |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1751 | |
Georg Brandl | b4dac71 | 2010-10-15 14:46:48 +0000 | [diff] [blame] | 1752 | codes_dict = PyDict_New(); |
| 1753 | rev_codes_dict = PyDict_New(); |
| 1754 | if (codes_dict == NULL || rev_codes_dict == NULL) { |
| 1755 | Py_XDECREF(codes_dict); |
| 1756 | Py_XDECREF(rev_codes_dict); |
| 1757 | return NULL; |
| 1758 | } |
Victor Stinner | 0fcab4a | 2011-01-04 12:59:15 +0000 | [diff] [blame] | 1759 | |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1760 | #define MYCONST(name) \ |
Georg Brandl | b4dac71 | 2010-10-15 14:46:48 +0000 | [diff] [blame] | 1761 | if (PyModule_AddStringConstant(errors_module, #name, \ |
Serhiy Storchaka | 20b39b2 | 2014-09-28 11:27:24 +0300 | [diff] [blame] | 1762 | XML_ErrorString(name)) < 0) \ |
Georg Brandl | b4dac71 | 2010-10-15 14:46:48 +0000 | [diff] [blame] | 1763 | return NULL; \ |
| 1764 | tmpnum = PyLong_FromLong(name); \ |
| 1765 | if (tmpnum == NULL) return NULL; \ |
| 1766 | res = PyDict_SetItemString(codes_dict, \ |
| 1767 | XML_ErrorString(name), tmpnum); \ |
| 1768 | if (res < 0) return NULL; \ |
| 1769 | tmpstr = PyUnicode_FromString(XML_ErrorString(name)); \ |
| 1770 | if (tmpstr == NULL) return NULL; \ |
| 1771 | res = PyDict_SetItem(rev_codes_dict, tmpnum, tmpstr); \ |
| 1772 | Py_DECREF(tmpstr); \ |
| 1773 | Py_DECREF(tmpnum); \ |
| 1774 | if (res < 0) return NULL; \ |
Fred Drake | 7bd9f41 | 2000-07-04 23:51:31 +0000 | [diff] [blame] | 1775 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1776 | MYCONST(XML_ERROR_NO_MEMORY); |
| 1777 | MYCONST(XML_ERROR_SYNTAX); |
| 1778 | MYCONST(XML_ERROR_NO_ELEMENTS); |
| 1779 | MYCONST(XML_ERROR_INVALID_TOKEN); |
| 1780 | MYCONST(XML_ERROR_UNCLOSED_TOKEN); |
| 1781 | MYCONST(XML_ERROR_PARTIAL_CHAR); |
| 1782 | MYCONST(XML_ERROR_TAG_MISMATCH); |
| 1783 | MYCONST(XML_ERROR_DUPLICATE_ATTRIBUTE); |
| 1784 | MYCONST(XML_ERROR_JUNK_AFTER_DOC_ELEMENT); |
| 1785 | MYCONST(XML_ERROR_PARAM_ENTITY_REF); |
| 1786 | MYCONST(XML_ERROR_UNDEFINED_ENTITY); |
| 1787 | MYCONST(XML_ERROR_RECURSIVE_ENTITY_REF); |
| 1788 | MYCONST(XML_ERROR_ASYNC_ENTITY); |
| 1789 | MYCONST(XML_ERROR_BAD_CHAR_REF); |
| 1790 | MYCONST(XML_ERROR_BINARY_ENTITY_REF); |
| 1791 | MYCONST(XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF); |
| 1792 | MYCONST(XML_ERROR_MISPLACED_XML_PI); |
| 1793 | MYCONST(XML_ERROR_UNKNOWN_ENCODING); |
| 1794 | MYCONST(XML_ERROR_INCORRECT_ENCODING); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1795 | MYCONST(XML_ERROR_UNCLOSED_CDATA_SECTION); |
| 1796 | MYCONST(XML_ERROR_EXTERNAL_ENTITY_HANDLING); |
| 1797 | MYCONST(XML_ERROR_NOT_STANDALONE); |
Fred Drake | 283b670 | 2004-08-04 22:28:16 +0000 | [diff] [blame] | 1798 | MYCONST(XML_ERROR_UNEXPECTED_STATE); |
| 1799 | MYCONST(XML_ERROR_ENTITY_DECLARED_IN_PE); |
| 1800 | MYCONST(XML_ERROR_FEATURE_REQUIRES_XML_DTD); |
| 1801 | MYCONST(XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING); |
| 1802 | /* Added in Expat 1.95.7. */ |
| 1803 | MYCONST(XML_ERROR_UNBOUND_PREFIX); |
| 1804 | /* Added in Expat 1.95.8. */ |
| 1805 | MYCONST(XML_ERROR_UNDECLARING_PREFIX); |
| 1806 | MYCONST(XML_ERROR_INCOMPLETE_PE); |
| 1807 | MYCONST(XML_ERROR_XML_DECL); |
| 1808 | MYCONST(XML_ERROR_TEXT_DECL); |
| 1809 | MYCONST(XML_ERROR_PUBLICID); |
| 1810 | MYCONST(XML_ERROR_SUSPENDED); |
| 1811 | MYCONST(XML_ERROR_NOT_SUSPENDED); |
| 1812 | MYCONST(XML_ERROR_ABORTED); |
| 1813 | MYCONST(XML_ERROR_FINISHED); |
| 1814 | MYCONST(XML_ERROR_SUSPEND_PE); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1815 | |
Georg Brandl | b4dac71 | 2010-10-15 14:46:48 +0000 | [diff] [blame] | 1816 | if (PyModule_AddStringConstant(errors_module, "__doc__", |
| 1817 | "Constants used to describe " |
| 1818 | "error conditions.") < 0) |
| 1819 | return NULL; |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1820 | |
Georg Brandl | b4dac71 | 2010-10-15 14:46:48 +0000 | [diff] [blame] | 1821 | if (PyModule_AddObject(errors_module, "codes", codes_dict) < 0) |
| 1822 | return NULL; |
| 1823 | if (PyModule_AddObject(errors_module, "messages", rev_codes_dict) < 0) |
| 1824 | return NULL; |
Victor Stinner | 0fcab4a | 2011-01-04 12:59:15 +0000 | [diff] [blame] | 1825 | |
Fred Drake | 93adb69 | 2000-09-23 04:55:48 +0000 | [diff] [blame] | 1826 | #undef MYCONST |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1827 | |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1828 | #define MYCONST(c) PyModule_AddIntConstant(m, #c, c) |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1829 | MYCONST(XML_PARAM_ENTITY_PARSING_NEVER); |
| 1830 | MYCONST(XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE); |
| 1831 | MYCONST(XML_PARAM_ENTITY_PARSING_ALWAYS); |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1832 | #undef MYCONST |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1833 | |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1834 | #define MYCONST(c) PyModule_AddIntConstant(model_module, #c, c) |
| 1835 | PyModule_AddStringConstant(model_module, "__doc__", |
| 1836 | "Constants used to interpret content model information."); |
Martin v. Löwis | 0078f6c | 2001-01-21 10:18:10 +0000 | [diff] [blame] | 1837 | |
Fred Drake | 85d835f | 2001-02-08 15:39:08 +0000 | [diff] [blame] | 1838 | MYCONST(XML_CTYPE_EMPTY); |
| 1839 | MYCONST(XML_CTYPE_ANY); |
| 1840 | MYCONST(XML_CTYPE_MIXED); |
| 1841 | MYCONST(XML_CTYPE_NAME); |
| 1842 | MYCONST(XML_CTYPE_CHOICE); |
| 1843 | MYCONST(XML_CTYPE_SEQ); |
| 1844 | |
| 1845 | MYCONST(XML_CQUANT_NONE); |
| 1846 | MYCONST(XML_CQUANT_OPT); |
| 1847 | MYCONST(XML_CQUANT_REP); |
| 1848 | MYCONST(XML_CQUANT_PLUS); |
| 1849 | #undef MYCONST |
Fredrik Lundh | c334504 | 2005-12-13 19:49:55 +0000 | [diff] [blame] | 1850 | |
| 1851 | /* initialize pyexpat dispatch table */ |
Fredrik Lundh | d7a4288 | 2005-12-13 20:43:04 +0000 | [diff] [blame] | 1852 | capi.size = sizeof(capi); |
Fredrik Lundh | cc117db | 2005-12-13 21:55:36 +0000 | [diff] [blame] | 1853 | capi.magic = PyExpat_CAPI_MAGIC; |
Fredrik Lundh | d7a4288 | 2005-12-13 20:43:04 +0000 | [diff] [blame] | 1854 | capi.MAJOR_VERSION = XML_MAJOR_VERSION; |
| 1855 | capi.MINOR_VERSION = XML_MINOR_VERSION; |
| 1856 | capi.MICRO_VERSION = XML_MICRO_VERSION; |
| 1857 | capi.ErrorString = XML_ErrorString; |
Fredrik Lundh | cc117db | 2005-12-13 21:55:36 +0000 | [diff] [blame] | 1858 | capi.GetErrorCode = XML_GetErrorCode; |
| 1859 | capi.GetErrorColumnNumber = XML_GetErrorColumnNumber; |
| 1860 | capi.GetErrorLineNumber = XML_GetErrorLineNumber; |
Fredrik Lundh | d7a4288 | 2005-12-13 20:43:04 +0000 | [diff] [blame] | 1861 | capi.Parse = XML_Parse; |
| 1862 | capi.ParserCreate_MM = XML_ParserCreate_MM; |
| 1863 | capi.ParserFree = XML_ParserFree; |
| 1864 | capi.SetCharacterDataHandler = XML_SetCharacterDataHandler; |
| 1865 | capi.SetCommentHandler = XML_SetCommentHandler; |
| 1866 | capi.SetDefaultHandlerExpand = XML_SetDefaultHandlerExpand; |
| 1867 | capi.SetElementHandler = XML_SetElementHandler; |
| 1868 | capi.SetNamespaceDeclHandler = XML_SetNamespaceDeclHandler; |
| 1869 | capi.SetProcessingInstructionHandler = XML_SetProcessingInstructionHandler; |
| 1870 | capi.SetUnknownEncodingHandler = XML_SetUnknownEncodingHandler; |
| 1871 | capi.SetUserData = XML_SetUserData; |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 1872 | capi.SetStartDoctypeDeclHandler = XML_SetStartDoctypeDeclHandler; |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 1873 | capi.SetEncoding = XML_SetEncoding; |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 1874 | capi.DefaultUnknownEncodingHandler = PyUnknownEncodingHandler; |
Christian Heimes | cb5778f | 2018-09-18 14:38:58 +0200 | [diff] [blame] | 1875 | #if XML_COMBINED_VERSION >= 20100 |
| 1876 | capi.SetHashSalt = XML_SetHashSalt; |
| 1877 | #else |
| 1878 | capi.SetHashSalt = NULL; |
| 1879 | #endif |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1880 | |
Benjamin Peterson | b173f78 | 2009-05-05 22:31:58 +0000 | [diff] [blame] | 1881 | /* export using capsule */ |
| 1882 | capi_object = PyCapsule_New(&capi, PyExpat_CAPSULE_NAME, NULL); |
Fredrik Lundh | d7a4288 | 2005-12-13 20:43:04 +0000 | [diff] [blame] | 1883 | if (capi_object) |
| 1884 | PyModule_AddObject(m, "expat_CAPI", capi_object); |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1885 | return m; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1886 | } |
| 1887 | |
Fred Drake | 6f98762 | 2000-08-25 18:03:30 +0000 | [diff] [blame] | 1888 | static void |
Martin v. Löwis | 5b68ce3 | 2001-10-21 08:53:52 +0000 | [diff] [blame] | 1889 | clear_handlers(xmlparseobject *self, int initial) |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1890 | { |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1891 | int i = 0; |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1892 | |
Fred Drake | 71b63ff | 2002-06-28 22:29:01 +0000 | [diff] [blame] | 1893 | for (; handler_info[i].name != NULL; i++) { |
Martin v. Löwis | 5b68ce3 | 2001-10-21 08:53:52 +0000 | [diff] [blame] | 1894 | if (initial) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1895 | self->handlers[i] = NULL; |
| 1896 | else { |
Serhiy Storchaka | 1ed017a | 2015-12-27 15:51:32 +0200 | [diff] [blame] | 1897 | Py_CLEAR(self->handlers[i]); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1898 | handler_info[i].setter(self->itself, NULL); |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1899 | } |
Fred Drake | cde7913 | 2001-04-25 16:01:30 +0000 | [diff] [blame] | 1900 | } |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1901 | } |
| 1902 | |
Tim Peters | 0c32279 | 2002-07-17 16:49:03 +0000 | [diff] [blame] | 1903 | static struct HandlerInfo handler_info[] = { |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1904 | |
| 1905 | #define HANDLER_INFO(name) \ |
| 1906 | {#name, (xmlhandlersetter)XML_Set##name, (xmlhandler)my_##name}, |
| 1907 | |
| 1908 | HANDLER_INFO(StartElementHandler) |
| 1909 | HANDLER_INFO(EndElementHandler) |
| 1910 | HANDLER_INFO(ProcessingInstructionHandler) |
| 1911 | HANDLER_INFO(CharacterDataHandler) |
| 1912 | HANDLER_INFO(UnparsedEntityDeclHandler) |
| 1913 | HANDLER_INFO(NotationDeclHandler) |
| 1914 | HANDLER_INFO(StartNamespaceDeclHandler) |
| 1915 | HANDLER_INFO(EndNamespaceDeclHandler) |
| 1916 | HANDLER_INFO(CommentHandler) |
| 1917 | HANDLER_INFO(StartCdataSectionHandler) |
| 1918 | HANDLER_INFO(EndCdataSectionHandler) |
| 1919 | HANDLER_INFO(DefaultHandler) |
| 1920 | HANDLER_INFO(DefaultHandlerExpand) |
| 1921 | HANDLER_INFO(NotStandaloneHandler) |
| 1922 | HANDLER_INFO(ExternalEntityRefHandler) |
| 1923 | HANDLER_INFO(StartDoctypeDeclHandler) |
| 1924 | HANDLER_INFO(EndDoctypeDeclHandler) |
| 1925 | HANDLER_INFO(EntityDeclHandler) |
| 1926 | HANDLER_INFO(XmlDeclHandler) |
| 1927 | HANDLER_INFO(ElementDeclHandler) |
| 1928 | HANDLER_INFO(AttlistDeclHandler) |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1929 | #if XML_COMBINED_VERSION >= 19504 |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1930 | HANDLER_INFO(SkippedEntityHandler) |
Martin v. Löwis | c847f40 | 2003-01-21 11:09:21 +0000 | [diff] [blame] | 1931 | #endif |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1932 | |
Serhiy Storchaka | 55f8249 | 2018-10-19 18:00:51 +0300 | [diff] [blame] | 1933 | #undef HANDLER_INFO |
| 1934 | |
Fred Drake | 0582df9 | 2000-07-12 04:49:00 +0000 | [diff] [blame] | 1935 | {NULL, NULL, NULL} /* sentinel */ |
Andrew M. Kuchling | b7f1053 | 2000-03-31 15:43:31 +0000 | [diff] [blame] | 1936 | }; |