Daniel Veillard | 5e5c2d0 | 2002-02-09 18:03:01 +0000 | [diff] [blame] | 1 | import libxml2mod |
Daniel Veillard | 85bb5b0 | 2003-12-04 14:12:05 +0000 | [diff] [blame] | 2 | import types |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 3 | |
Daniel Veillard | a81355e | 2004-09-28 11:08:27 +0000 | [diff] [blame] | 4 | # The root of all libxml2 errors. |
| 5 | class libxmlError(Exception): pass |
| 6 | |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 7 | # |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 8 | # Errors raised by the wrappers when some tree handling failed. |
| 9 | # |
Daniel Veillard | a81355e | 2004-09-28 11:08:27 +0000 | [diff] [blame] | 10 | class treeError(libxmlError): |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 11 | def __init__(self, msg): |
| 12 | self.msg = msg |
| 13 | def __str__(self): |
| 14 | return self.msg |
| 15 | |
Daniel Veillard | a81355e | 2004-09-28 11:08:27 +0000 | [diff] [blame] | 16 | class parserError(libxmlError): |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 17 | def __init__(self, msg): |
| 18 | self.msg = msg |
| 19 | def __str__(self): |
| 20 | return self.msg |
| 21 | |
Daniel Veillard | a81355e | 2004-09-28 11:08:27 +0000 | [diff] [blame] | 22 | class uriError(libxmlError): |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 23 | def __init__(self, msg): |
| 24 | self.msg = msg |
| 25 | def __str__(self): |
| 26 | return self.msg |
| 27 | |
Daniel Veillard | a81355e | 2004-09-28 11:08:27 +0000 | [diff] [blame] | 28 | class xpathError(libxmlError): |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 29 | def __init__(self, msg): |
| 30 | self.msg = msg |
| 31 | def __str__(self): |
| 32 | return self.msg |
| 33 | |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 34 | class ioWrapper: |
| 35 | def __init__(self, _obj): |
| 36 | self.__io = _obj |
| 37 | self._o = None |
| 38 | |
| 39 | def io_close(self): |
| 40 | if self.__io == None: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 41 | return(-1) |
| 42 | self.__io.close() |
| 43 | self.__io = None |
| 44 | return(0) |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 45 | |
| 46 | def io_flush(self): |
| 47 | if self.__io == None: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 48 | return(-1) |
| 49 | self.__io.flush() |
| 50 | return(0) |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 51 | |
| 52 | def io_read(self, len = -1): |
| 53 | if self.__io == None: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 54 | return(-1) |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 55 | if len < 0: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 56 | return(self.__io.read()) |
| 57 | return(self.__io.read(len)) |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 58 | |
| 59 | def io_write(self, str, len = -1): |
| 60 | if self.__io == None: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 61 | return(-1) |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 62 | if len < 0: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 63 | return(self.__io.write(str)) |
| 64 | return(self.__io.write(str, len)) |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 65 | |
| 66 | class ioReadWrapper(ioWrapper): |
| 67 | def __init__(self, _obj, enc = ""): |
| 68 | ioWrapper.__init__(self, _obj) |
| 69 | self._o = libxml2mod.xmlCreateInputBuffer(self, enc) |
| 70 | |
| 71 | def __del__(self): |
| 72 | print "__del__" |
| 73 | self.io_close() |
| 74 | if self._o != None: |
| 75 | libxml2mod.xmlFreeParserInputBuffer(self._o) |
| 76 | self._o = None |
| 77 | |
| 78 | def close(self): |
| 79 | self.io_close() |
| 80 | if self._o != None: |
| 81 | libxml2mod.xmlFreeParserInputBuffer(self._o) |
| 82 | self._o = None |
| 83 | |
| 84 | class ioWriteWrapper(ioWrapper): |
| 85 | def __init__(self, _obj, enc = ""): |
Daniel Veillard | 85bb5b0 | 2003-12-04 14:12:05 +0000 | [diff] [blame] | 86 | # print "ioWriteWrapper.__init__", _obj |
| 87 | if type(_obj) == type(''): |
William M. Brack | 37e6394 | 2004-07-12 16:27:37 +0000 | [diff] [blame] | 88 | print "write io from a string" |
| 89 | self.o = None |
| 90 | elif type(_obj) == types.InstanceType: |
| 91 | print "write io from instance of %s" % (_obj.__class__) |
| 92 | ioWrapper.__init__(self, _obj) |
| 93 | self._o = libxml2mod.xmlCreateOutputBuffer(self, enc) |
| 94 | else: |
| 95 | file = libxml2mod.outputBufferGetPythonFile(_obj) |
| 96 | if file != None: |
| 97 | ioWrapper.__init__(self, file) |
| 98 | else: |
| 99 | ioWrapper.__init__(self, _obj) |
| 100 | self._o = _obj |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 101 | |
| 102 | def __del__(self): |
Daniel Veillard | 85bb5b0 | 2003-12-04 14:12:05 +0000 | [diff] [blame] | 103 | # print "__del__" |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 104 | self.io_close() |
| 105 | if self._o != None: |
| 106 | libxml2mod.xmlOutputBufferClose(self._o) |
| 107 | self._o = None |
| 108 | |
Daniel Veillard | 85bb5b0 | 2003-12-04 14:12:05 +0000 | [diff] [blame] | 109 | def flush(self): |
| 110 | self.io_flush() |
| 111 | if self._o != None: |
| 112 | libxml2mod.xmlOutputBufferClose(self._o) |
| 113 | self._o = None |
| 114 | |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 115 | def close(self): |
Daniel Veillard | 85bb5b0 | 2003-12-04 14:12:05 +0000 | [diff] [blame] | 116 | self.io_flush() |
Daniel Veillard | c6d4a93 | 2002-09-12 15:00:57 +0000 | [diff] [blame] | 117 | if self._o != None: |
| 118 | libxml2mod.xmlOutputBufferClose(self._o) |
| 119 | self._o = None |
| 120 | |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 121 | # |
| 122 | # Example of a class to handle SAX events |
| 123 | # |
| 124 | class SAXCallback: |
| 125 | """Base class for SAX handlers""" |
| 126 | def startDocument(self): |
| 127 | """called at the start of the document""" |
| 128 | pass |
| 129 | |
| 130 | def endDocument(self): |
| 131 | """called at the end of the document""" |
| 132 | pass |
| 133 | |
| 134 | def startElement(self, tag, attrs): |
| 135 | """called at the start of every element, tag is the name of |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 136 | the element, attrs is a dictionary of the element's attributes""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 137 | pass |
| 138 | |
| 139 | def endElement(self, tag): |
| 140 | """called at the start of every element, tag is the name of |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 141 | the element""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 142 | pass |
| 143 | |
| 144 | def characters(self, data): |
| 145 | """called when character data have been read, data is the string |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 146 | containing the data, multiple consecutive characters() callback |
| 147 | are possible.""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 148 | pass |
| 149 | |
| 150 | def cdataBlock(self, data): |
| 151 | """called when CDATA section have been read, data is the string |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 152 | containing the data, multiple consecutive cdataBlock() callback |
| 153 | are possible.""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 154 | pass |
| 155 | |
| 156 | def reference(self, name): |
| 157 | """called when an entity reference has been found""" |
| 158 | pass |
| 159 | |
| 160 | def ignorableWhitespace(self, data): |
| 161 | """called when potentially ignorable white spaces have been found""" |
| 162 | pass |
| 163 | |
| 164 | def processingInstruction(self, target, data): |
| 165 | """called when a PI has been found, target contains the PI name and |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 166 | data is the associated data in the PI""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 167 | pass |
| 168 | |
| 169 | def comment(self, content): |
| 170 | """called when a comment has been found, content contains the comment""" |
| 171 | pass |
| 172 | |
| 173 | def externalSubset(self, name, externalID, systemID): |
| 174 | """called when a DOCTYPE declaration has been found, name is the |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 175 | DTD name and externalID, systemID are the DTD public and system |
| 176 | identifier for that DTd if available""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 177 | pass |
| 178 | |
| 179 | def internalSubset(self, name, externalID, systemID): |
| 180 | """called when a DOCTYPE declaration has been found, name is the |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 181 | DTD name and externalID, systemID are the DTD public and system |
| 182 | identifier for that DTD if available""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 183 | pass |
| 184 | |
| 185 | def entityDecl(self, name, type, externalID, systemID, content): |
| 186 | """called when an ENTITY declaration has been found, name is the |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 187 | entity name and externalID, systemID are the entity public and |
| 188 | system identifier for that entity if available, type indicates |
| 189 | the entity type, and content reports it's string content""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 190 | pass |
| 191 | |
| 192 | def notationDecl(self, name, externalID, systemID): |
| 193 | """called when an NOTATION declaration has been found, name is the |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 194 | notation name and externalID, systemID are the notation public and |
| 195 | system identifier for that notation if available""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 196 | pass |
| 197 | |
| 198 | def attributeDecl(self, elem, name, type, defi, defaultValue, nameList): |
| 199 | """called when an ATTRIBUTE definition has been found""" |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 200 | pass |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 201 | |
| 202 | def elementDecl(self, name, type, content): |
| 203 | """called when an ELEMENT definition has been found""" |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 204 | pass |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 205 | |
| 206 | def entityDecl(self, name, publicId, systemID, notationName): |
| 207 | """called when an unparsed ENTITY declaration has been found, |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 208 | name is the entity name and publicId,, systemID are the entity |
| 209 | public and system identifier for that entity if available, |
| 210 | and notationName indicate the associated NOTATION""" |
Daniel Veillard | 8d24cc1 | 2002-03-05 15:41:29 +0000 | [diff] [blame] | 211 | pass |
| 212 | |
| 213 | def warning(self, msg): |
| 214 | print msg |
| 215 | |
| 216 | def error(self, msg): |
| 217 | raise parserError(msg) |
| 218 | |
| 219 | def fatalError(self, msg): |
| 220 | raise parserError(msg) |
| 221 | |
| 222 | # |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 223 | # This class is the ancestor of all the Node classes. It provides |
| 224 | # the basic functionalities shared by all nodes (and handle |
| 225 | # gracefylly the exception), like name, navigation in the tree, |
Daniel Veillard | 1e77438 | 2002-03-06 17:35:40 +0000 | [diff] [blame] | 226 | # doc reference, content access and serializing to a string or URI |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 227 | # |
Daniel Veillard | 36ed529 | 2002-01-30 23:49:06 +0000 | [diff] [blame] | 228 | class xmlCore: |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 229 | def __init__(self, _obj=None): |
| 230 | if _obj != None: |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 231 | self._o = _obj; |
| 232 | return |
| 233 | self._o = None |
Daniel Veillard | 1cd4dae | 2005-01-15 17:45:28 +0000 | [diff] [blame] | 234 | def __str__(self): |
| 235 | return self.serialize() |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 236 | def get_parent(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 237 | ret = libxml2mod.parent(self._o) |
| 238 | if ret == None: |
| 239 | return None |
| 240 | return xmlNode(_obj=ret) |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 241 | def get_children(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 242 | ret = libxml2mod.children(self._o) |
| 243 | if ret == None: |
| 244 | return None |
| 245 | return xmlNode(_obj=ret) |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 246 | def get_last(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 247 | ret = libxml2mod.last(self._o) |
| 248 | if ret == None: |
| 249 | return None |
| 250 | return xmlNode(_obj=ret) |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 251 | def get_next(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 252 | ret = libxml2mod.next(self._o) |
| 253 | if ret == None: |
| 254 | return None |
| 255 | return xmlNode(_obj=ret) |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 256 | def get_properties(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 257 | ret = libxml2mod.properties(self._o) |
| 258 | if ret == None: |
| 259 | return None |
| 260 | return xmlAttr(_obj=ret) |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 261 | def get_prev(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 262 | ret = libxml2mod.prev(self._o) |
| 263 | if ret == None: |
| 264 | return None |
| 265 | return xmlNode(_obj=ret) |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 266 | def get_content(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 267 | return libxml2mod.xmlNodeGetContent(self._o) |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 268 | getContent = get_content # why is this duplicate naming needed ? |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 269 | def get_name(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 270 | return libxml2mod.name(self._o) |
Daniel Veillard | d2897fd | 2002-01-30 16:37:32 +0000 | [diff] [blame] | 271 | def get_type(self): |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 272 | return libxml2mod.type(self._o) |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 273 | def get_doc(self): |
| 274 | ret = libxml2mod.doc(self._o) |
| 275 | if ret == None: |
| 276 | if self.type in ["document_xml", "document_html"]: |
| 277 | return xmlDoc(_obj=self._o) |
| 278 | else: |
| 279 | return None |
| 280 | return xmlDoc(_obj=ret) |
| 281 | # |
| 282 | # Those are common attributes to nearly all type of nodes |
| 283 | # defined as python2 properties |
| 284 | # |
| 285 | import sys |
| 286 | if float(sys.version[0:3]) < 2.2: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 287 | def __getattr__(self, attr): |
| 288 | if attr == "parent": |
| 289 | ret = libxml2mod.parent(self._o) |
| 290 | if ret == None: |
| 291 | return None |
| 292 | return xmlNode(_obj=ret) |
| 293 | elif attr == "properties": |
| 294 | ret = libxml2mod.properties(self._o) |
| 295 | if ret == None: |
| 296 | return None |
| 297 | return xmlAttr(_obj=ret) |
| 298 | elif attr == "children": |
| 299 | ret = libxml2mod.children(self._o) |
| 300 | if ret == None: |
| 301 | return None |
| 302 | return xmlNode(_obj=ret) |
| 303 | elif attr == "last": |
| 304 | ret = libxml2mod.last(self._o) |
| 305 | if ret == None: |
| 306 | return None |
| 307 | return xmlNode(_obj=ret) |
| 308 | elif attr == "next": |
| 309 | ret = libxml2mod.next(self._o) |
| 310 | if ret == None: |
| 311 | return None |
| 312 | return xmlNode(_obj=ret) |
| 313 | elif attr == "prev": |
| 314 | ret = libxml2mod.prev(self._o) |
| 315 | if ret == None: |
| 316 | return None |
| 317 | return xmlNode(_obj=ret) |
| 318 | elif attr == "content": |
| 319 | return libxml2mod.xmlNodeGetContent(self._o) |
| 320 | elif attr == "name": |
| 321 | return libxml2mod.name(self._o) |
| 322 | elif attr == "type": |
| 323 | return libxml2mod.type(self._o) |
| 324 | elif attr == "doc": |
| 325 | ret = libxml2mod.doc(self._o) |
| 326 | if ret == None: |
| 327 | if self.type == "document_xml" or self.type == "document_html": |
| 328 | return xmlDoc(_obj=self._o) |
| 329 | else: |
| 330 | return None |
| 331 | return xmlDoc(_obj=ret) |
| 332 | raise AttributeError,attr |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 333 | else: |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 334 | parent = property(get_parent, None, None, "Parent node") |
| 335 | children = property(get_children, None, None, "First child node") |
| 336 | last = property(get_last, None, None, "Last sibling node") |
| 337 | next = property(get_next, None, None, "Next sibling node") |
| 338 | prev = property(get_prev, None, None, "Previous sibling node") |
| 339 | properties = property(get_properties, None, None, "List of properies") |
| 340 | content = property(get_content, None, None, "Content of this node") |
| 341 | name = property(get_name, None, None, "Node name") |
| 342 | type = property(get_type, None, None, "Node type") |
| 343 | doc = property(get_doc, None, None, "The document this node belongs to") |
Daniel Veillard | 1e77438 | 2002-03-06 17:35:40 +0000 | [diff] [blame] | 344 | |
| 345 | # |
| 346 | # Serialization routines, the optional arguments have the following |
| 347 | # meaning: |
| 348 | # encoding: string to ask saving in a specific encoding |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 349 | # indent: if 1 the serializer is asked to indent the output |
Daniel Veillard | 1e77438 | 2002-03-06 17:35:40 +0000 | [diff] [blame] | 350 | # |
| 351 | def serialize(self, encoding = None, format = 0): |
| 352 | return libxml2mod.serializeNode(self._o, encoding, format) |
| 353 | def saveTo(self, file, encoding = None, format = 0): |
| 354 | return libxml2mod.saveNodeTo(self._o, file, encoding, format) |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 355 | |
Daniel Veillard | f742d34 | 2002-03-07 00:05:35 +0000 | [diff] [blame] | 356 | # |
Daniel Veillard | d5e198a | 2004-03-09 09:03:28 +0000 | [diff] [blame] | 357 | # Canonicalization routines: |
| 358 | # |
| 359 | # nodes: the node set (tuple or list) to be included in the |
| 360 | # canonized image or None if all document nodes should be |
| 361 | # included. |
| 362 | # exclusive: the exclusive flag (0 - non-exclusive |
| 363 | # canonicalization; otherwise - exclusive canonicalization) |
| 364 | # prefixes: the list of inclusive namespace prefixes (strings), |
| 365 | # or None if there is no inclusive namespaces (only for |
| 366 | # exclusive canonicalization, ignored otherwise) |
| 367 | # with_comments: include comments in the result (!=0) or not |
| 368 | # (==0) |
| 369 | def c14nMemory(self, |
| 370 | nodes=None, |
| 371 | exclusive=0, |
| 372 | prefixes=None, |
| 373 | with_comments=0): |
| 374 | if nodes: |
| 375 | nodes = map(lambda n: n._o, nodes) |
| 376 | return libxml2mod.xmlC14NDocDumpMemory( |
| 377 | self.get_doc()._o, |
| 378 | nodes, |
| 379 | exclusive != 0, |
| 380 | prefixes, |
| 381 | with_comments != 0) |
| 382 | def c14nSaveTo(self, |
| 383 | file, |
| 384 | nodes=None, |
| 385 | exclusive=0, |
| 386 | prefixes=None, |
| 387 | with_comments=0): |
| 388 | if nodes: |
| 389 | nodes = map(lambda n: n._o, nodes) |
| 390 | return libxml2mod.xmlC14NDocSaveTo( |
| 391 | self.get_doc()._o, |
| 392 | nodes, |
| 393 | exclusive != 0, |
| 394 | prefixes, |
| 395 | with_comments != 0, |
| 396 | file) |
| 397 | |
| 398 | # |
Daniel Veillard | f742d34 | 2002-03-07 00:05:35 +0000 | [diff] [blame] | 399 | # Selecting nodes using XPath, a bit slow because the context |
| 400 | # is allocated/freed every time but convenient. |
| 401 | # |
| 402 | def xpathEval(self, expr): |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 403 | doc = self.doc |
| 404 | if doc == None: |
| 405 | return None |
| 406 | ctxt = doc.xpathNewContext() |
| 407 | ctxt.setContextNode(self) |
| 408 | res = ctxt.xpathEval(expr) |
| 409 | ctxt.xpathFreeContext() |
| 410 | return res |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 411 | |
Daniel Veillard | f88d8cf | 2003-12-08 10:25:02 +0000 | [diff] [blame] | 412 | # # |
| 413 | # # Selecting nodes using XPath, faster because the context |
| 414 | # # is allocated just once per xmlDoc. |
| 415 | # # |
| 416 | # # Removed: DV memleaks c.f. #126735 |
| 417 | # # |
| 418 | # def xpathEval2(self, expr): |
| 419 | # doc = self.doc |
| 420 | # if doc == None: |
| 421 | # return None |
| 422 | # try: |
| 423 | # doc._ctxt.setContextNode(self) |
| 424 | # except: |
| 425 | # doc._ctxt = doc.xpathNewContext() |
| 426 | # doc._ctxt.setContextNode(self) |
| 427 | # res = doc._ctxt.xpathEval(expr) |
| 428 | # return res |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 429 | def xpathEval2(self, expr): |
Daniel Veillard | f88d8cf | 2003-12-08 10:25:02 +0000 | [diff] [blame] | 430 | return self.xpathEval(expr) |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 431 | |
| 432 | # support for python2 iterators |
| 433 | def walk_depth_first(self): |
| 434 | return xmlCoreDepthFirstItertor(self) |
| 435 | def walk_breadth_first(self): |
| 436 | return xmlCoreBreadthFirstItertor(self) |
| 437 | __iter__ = walk_depth_first |
| 438 | |
| 439 | def free(self): |
| 440 | try: |
| 441 | self.doc._ctxt.xpathFreeContext() |
| 442 | except: |
| 443 | pass |
Daniel Veillard | f88d8cf | 2003-12-08 10:25:02 +0000 | [diff] [blame] | 444 | libxml2mod.xmlFreeDoc(self._o) |
Daniel Veillard | 51a447a | 2003-01-04 19:42:46 +0000 | [diff] [blame] | 445 | |
| 446 | |
| 447 | # |
| 448 | # implements the depth-first iterator for libxml2 DOM tree |
| 449 | # |
| 450 | class xmlCoreDepthFirstItertor: |
| 451 | def __init__(self, node): |
| 452 | self.node = node |
| 453 | self.parents = [] |
| 454 | def __iter__(self): |
| 455 | return self |
| 456 | def next(self): |
| 457 | while 1: |
| 458 | if self.node: |
| 459 | ret = self.node |
| 460 | self.parents.append(self.node) |
| 461 | self.node = self.node.children |
| 462 | return ret |
| 463 | try: |
| 464 | parent = self.parents.pop() |
| 465 | except IndexError: |
| 466 | raise StopIteration |
| 467 | self.node = parent.next |
| 468 | |
| 469 | # |
| 470 | # implements the breadth-first iterator for libxml2 DOM tree |
| 471 | # |
| 472 | class xmlCoreBreadthFirstItertor: |
| 473 | def __init__(self, node): |
| 474 | self.node = node |
| 475 | self.parents = [] |
| 476 | def __iter__(self): |
| 477 | return self |
| 478 | def next(self): |
| 479 | while 1: |
| 480 | if self.node: |
| 481 | ret = self.node |
| 482 | self.parents.append(self.node) |
| 483 | self.node = self.node.next |
| 484 | return ret |
| 485 | try: |
| 486 | parent = self.parents.pop() |
| 487 | except IndexError: |
| 488 | raise StopIteration |
| 489 | self.node = parent.children |
| 490 | |
Daniel Veillard | 36ed529 | 2002-01-30 23:49:06 +0000 | [diff] [blame] | 491 | # |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 492 | # converters to present a nicer view of the XPath returns |
| 493 | # |
| 494 | def nodeWrap(o): |
| 495 | # TODO try to cast to the most appropriate node class |
Daniel Veillard | 1f8658a | 2004-08-14 21:46:31 +0000 | [diff] [blame] | 496 | name = libxml2mod.type(o) |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 497 | if name == "element" or name == "text": |
| 498 | return xmlNode(_obj=o) |
| 499 | if name == "attribute": |
| 500 | return xmlAttr(_obj=o) |
| 501 | if name[0:8] == "document": |
| 502 | return xmlDoc(_obj=o) |
Daniel Veillard | 1f8658a | 2004-08-14 21:46:31 +0000 | [diff] [blame] | 503 | if name == "namespace": |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 504 | return xmlNs(_obj=o) |
| 505 | if name == "elem_decl": |
| 506 | return xmlElement(_obj=o) |
| 507 | if name == "attribute_decl": |
Daniel Veillard | 1f8658a | 2004-08-14 21:46:31 +0000 | [diff] [blame] | 508 | return xmlAttribute(_obj=o) |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 509 | if name == "entity_decl": |
| 510 | return xmlEntity(_obj=o) |
| 511 | if name == "dtd": |
Daniel Veillard | e59494f | 2003-01-04 16:35:29 +0000 | [diff] [blame] | 512 | return xmlDtd(_obj=o) |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 513 | return xmlNode(_obj=o) |
| 514 | |
| 515 | def xpathObjectRet(o): |
| 516 | if type(o) == type([]) or type(o) == type(()): |
| 517 | ret = map(lambda x: nodeWrap(x), o) |
Daniel Veillard | 01a6d41 | 2002-02-11 18:42:20 +0000 | [diff] [blame] | 518 | return ret |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 519 | return o |
| 520 | |
| 521 | # |
Daniel Veillard | a7340c8 | 2002-02-01 17:56:45 +0000 | [diff] [blame] | 522 | # register an XPath function |
| 523 | # |
| 524 | def registerXPathFunction(ctxt, name, ns_uri, f): |
Daniel Veillard | 5e5c2d0 | 2002-02-09 18:03:01 +0000 | [diff] [blame] | 525 | ret = libxml2mod.xmlRegisterXPathFunction(ctxt, name, ns_uri, f) |
Daniel Veillard | a7340c8 | 2002-02-01 17:56:45 +0000 | [diff] [blame] | 526 | |
| 527 | # |
Daniel Veillard | f25b4ca | 2002-12-27 15:18:35 +0000 | [diff] [blame] | 528 | # For the xmlTextReader parser configuration |
| 529 | # |
| 530 | PARSER_LOADDTD=1 |
| 531 | PARSER_DEFAULTATTRS=2 |
| 532 | PARSER_VALIDATE=3 |
Daniel Veillard | e18fc18 | 2002-12-28 22:56:33 +0000 | [diff] [blame] | 533 | PARSER_SUBST_ENTITIES=4 |
Daniel Veillard | f25b4ca | 2002-12-27 15:18:35 +0000 | [diff] [blame] | 534 | |
| 535 | # |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 536 | # For the error callback severities |
Daniel Veillard | 26f7026 | 2003-01-16 22:45:08 +0000 | [diff] [blame] | 537 | # |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 538 | PARSER_SEVERITY_VALIDITY_WARNING=1 |
| 539 | PARSER_SEVERITY_VALIDITY_ERROR=2 |
| 540 | PARSER_SEVERITY_WARNING=3 |
| 541 | PARSER_SEVERITY_ERROR=4 |
Daniel Veillard | 26f7026 | 2003-01-16 22:45:08 +0000 | [diff] [blame] | 542 | |
| 543 | # |
Daniel Veillard | 3e20a29 | 2003-01-10 13:14:40 +0000 | [diff] [blame] | 544 | # register the libxml2 error handler |
Daniel Veillard | 36ed529 | 2002-01-30 23:49:06 +0000 | [diff] [blame] | 545 | # |
Daniel Veillard | 3e20a29 | 2003-01-10 13:14:40 +0000 | [diff] [blame] | 546 | def registerErrorHandler(f, ctx): |
| 547 | """Register a Python written function to for error reporting. |
| 548 | The function is called back as f(ctx, error). """ |
| 549 | import sys |
| 550 | if not sys.modules.has_key('libxslt'): |
| 551 | # normal behaviour when libxslt is not imported |
| 552 | ret = libxml2mod.xmlRegisterErrorHandler(f,ctx) |
| 553 | else: |
| 554 | # when libxslt is already imported, one must |
| 555 | # use libxst's error handler instead |
| 556 | import libxslt |
| 557 | ret = libxslt.registerErrorHandler(f,ctx) |
| 558 | return ret |
| 559 | |
Daniel Veillard | e6227e0 | 2003-01-14 11:42:39 +0000 | [diff] [blame] | 560 | class parserCtxtCore: |
| 561 | |
| 562 | def __init__(self, _obj=None): |
| 563 | if _obj != None: |
| 564 | self._o = _obj; |
| 565 | return |
| 566 | self._o = None |
| 567 | |
| 568 | def __del__(self): |
| 569 | if self._o != None: |
| 570 | libxml2mod.xmlFreeParserCtxt(self._o) |
William M. Brack | 1d75c8a | 2003-10-27 13:48:16 +0000 | [diff] [blame] | 571 | self._o = None |
Daniel Veillard | e6227e0 | 2003-01-14 11:42:39 +0000 | [diff] [blame] | 572 | |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 573 | def setErrorHandler(self,f,arg): |
| 574 | """Register an error handler that will be called back as |
| 575 | f(arg,msg,severity,reserved). |
| 576 | |
| 577 | @reserved is currently always None.""" |
| 578 | libxml2mod.xmlParserCtxtSetErrorHandler(self._o,f,arg) |
Daniel Veillard | e6227e0 | 2003-01-14 11:42:39 +0000 | [diff] [blame] | 579 | |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 580 | def getErrorHandler(self): |
| 581 | """Return (f,arg) as previously registered with setErrorHandler |
| 582 | or (None,None).""" |
| 583 | return libxml2mod.xmlParserCtxtGetErrorHandler(self._o) |
| 584 | |
Daniel Veillard | 5439624 | 2003-04-23 07:36:50 +0000 | [diff] [blame] | 585 | def addLocalCatalog(self, uri): |
| 586 | """Register a local catalog with the parser""" |
| 587 | return libxml2mod.addLocalCatalog(self._o, uri) |
| 588 | |
| 589 | |
Daniel Veillard | 0e460da | 2005-03-30 22:47:10 +0000 | [diff] [blame^] | 590 | class ValidCtxtCore: |
| 591 | |
| 592 | def __init__(self, *args, **kw): |
| 593 | pass |
| 594 | |
| 595 | def setValidityErrorHandler(self, err_func, warn_func, arg=None): |
| 596 | """ |
| 597 | Register error and warning handlers for DTD validation. |
| 598 | These will be called back as f(msg,arg) |
| 599 | """ |
| 600 | libxml2mod.xmlSetValidErrors(self._o, err_func, warn_func, arg) |
| 601 | |
| 602 | |
| 603 | class SchemaValidCtxtCore: |
| 604 | |
| 605 | def __init__(self, *args, **kw): |
| 606 | pass |
| 607 | |
| 608 | def setValidityErrorHandler(self, err_func, warn_func, arg=None): |
| 609 | """ |
| 610 | Register error and warning handlers for Schema validation. |
| 611 | These will be called back as f(msg,arg) |
| 612 | """ |
| 613 | libxml2mod.xmlSchemaSetValidErrors(self._o, err_func, warn_func, arg) |
| 614 | |
| 615 | |
| 616 | class relaxNgValidCtxtCore: |
| 617 | |
| 618 | def __init__(self, *args, **kw): |
| 619 | pass |
| 620 | |
| 621 | def setValidityErrorHandler(self, err_func, warn_func, arg=None): |
| 622 | """ |
| 623 | Register error and warning handlers for RelaxNG validation. |
| 624 | These will be called back as f(msg,arg) |
| 625 | """ |
| 626 | libxml2mod.xmlRelaxNGSetValidErrors(self._o, err_func, warn_func, arg) |
| 627 | |
| 628 | |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 629 | def _xmlTextReaderErrorFunc((f,arg),msg,severity,locator): |
| 630 | """Intermediate callback to wrap the locator""" |
| 631 | return f(arg,msg,severity,xmlTextReaderLocator(locator)) |
Daniel Veillard | e6227e0 | 2003-01-14 11:42:39 +0000 | [diff] [blame] | 632 | |
Daniel Veillard | 26f7026 | 2003-01-16 22:45:08 +0000 | [diff] [blame] | 633 | class xmlTextReaderCore: |
| 634 | |
| 635 | def __init__(self, _obj=None): |
| 636 | self.input = None |
| 637 | if _obj != None:self._o = _obj;return |
| 638 | self._o = None |
| 639 | |
| 640 | def __del__(self): |
| 641 | if self._o != None: |
| 642 | libxml2mod.xmlFreeTextReader(self._o) |
| 643 | self._o = None |
| 644 | |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 645 | def SetErrorHandler(self,f,arg): |
Daniel Veillard | 26f7026 | 2003-01-16 22:45:08 +0000 | [diff] [blame] | 646 | """Register an error handler that will be called back as |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 647 | f(arg,msg,severity,locator).""" |
| 648 | if f is None: |
| 649 | libxml2mod.xmlTextReaderSetErrorHandler(\ |
| 650 | self._o,None,None) |
| 651 | else: |
| 652 | libxml2mod.xmlTextReaderSetErrorHandler(\ |
| 653 | self._o,_xmlTextReaderErrorFunc,(f,arg)) |
Daniel Veillard | 26f7026 | 2003-01-16 22:45:08 +0000 | [diff] [blame] | 654 | |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 655 | def GetErrorHandler(self): |
Daniel Veillard | 26f7026 | 2003-01-16 22:45:08 +0000 | [diff] [blame] | 656 | """Return (f,arg) as previously registered with setErrorHandler |
| 657 | or (None,None).""" |
Daniel Veillard | 417be3a | 2003-01-20 21:26:34 +0000 | [diff] [blame] | 658 | f,arg = libxml2mod.xmlTextReaderGetErrorHandler(self._o) |
| 659 | if f is None: |
| 660 | return None,None |
| 661 | else: |
| 662 | # assert f is _xmlTextReaderErrorFunc |
| 663 | return arg |
Daniel Veillard | 26f7026 | 2003-01-16 22:45:08 +0000 | [diff] [blame] | 664 | |
Daniel Veillard | f93a866 | 2004-07-01 12:56:30 +0000 | [diff] [blame] | 665 | # |
| 666 | # The cleanup now goes though a wrappe in libxml.c |
| 667 | # |
| 668 | def cleanupParser(): |
| 669 | libxml2mod.xmlPythonCleanupParser() |
Daniel Veillard | 87ab1c1 | 2003-12-21 13:01:56 +0000 | [diff] [blame] | 670 | |
Daniel Veillard | 3e20a29 | 2003-01-10 13:14:40 +0000 | [diff] [blame] | 671 | # WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING |
| 672 | # |
| 673 | # Everything before this line comes from libxml.py |
| 674 | # Everything after this line is automatically generated |
| 675 | # |
| 676 | # WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING |
Daniel Veillard | 1971ee2 | 2002-01-31 20:29:19 +0000 | [diff] [blame] | 677 | |