blob: 5473b364544f27f037b622a221e1c4497f9cf577 [file] [log] [blame]
Fred Drake45cd9de2000-06-29 19:34:54 +00001"""
2SAX driver for the Pyexpat C module. This driver works with
Lars Gustäbelbb757132000-09-24 20:38:18 +00003pyexpat.__version__ == '2.22'.
Fred Drake45cd9de2000-06-29 19:34:54 +00004"""
5
Fred Drake45cd9de2000-06-29 19:34:54 +00006version = "0.20"
7
Lars Gustäbelf43cf312000-09-24 18:29:24 +00008from xml.sax._exceptions import *
Martin v. Löwisfb73bb12001-06-17 07:05:43 +00009
10# xml.parsers.expat does not raise ImportError in Jython
11import sys
Fred Drakec974bf42001-07-30 22:41:23 +000012if sys.platform[:4] == "java":
Martin v. Löwisfb73bb12001-06-17 07:05:43 +000013 raise SAXReaderNotAvailable("expat not available in Java", None)
14del sys
15
Martin v. Löwis962c9e72000-10-06 17:41:52 +000016try:
17 from xml.parsers import expat
18except ImportError:
Fred Drakec974bf42001-07-30 22:41:23 +000019 raise SAXReaderNotAvailable("expat not supported", None)
Jeremy Hyltone3c37d62001-07-30 21:49:22 +000020else:
21 if not hasattr(expat, "ParserCreate"):
Fred Drakec974bf42001-07-30 22:41:23 +000022 raise SAXReaderNotAvailable("expat not supported", None)
Martin v. Löwis2066fa02000-09-24 21:17:39 +000023from xml.sax import xmlreader, saxutils, handler
Fred Drake45cd9de2000-06-29 19:34:54 +000024
Lars Gustäbel32bf12e2000-09-24 18:39:23 +000025AttributesImpl = xmlreader.AttributesImpl
26AttributesNSImpl = xmlreader.AttributesNSImpl
27
Martin v. Löwis3f0969f2000-09-29 19:00:40 +000028import string
29
Fred Drake45cd9de2000-06-29 19:34:54 +000030# --- ExpatParser
31
Fred Drakeddb48672000-09-23 05:32:26 +000032class ExpatParser(xmlreader.IncrementalParser, xmlreader.Locator):
Fred Drake45cd9de2000-06-29 19:34:54 +000033 "SAX driver for the Pyexpat C module."
34
35 def __init__(self, namespaceHandling=0, bufsize=2**16-20):
36 xmlreader.IncrementalParser.__init__(self, bufsize)
Lars Gustäbele292a242000-09-24 20:19:45 +000037 self._source = xmlreader.InputSource()
Fred Drake45cd9de2000-06-29 19:34:54 +000038 self._parser = None
39 self._namespaces = namespaceHandling
Martin v. Löwis05917252001-01-27 08:56:24 +000040 self._lex_handler_prop = None
Fred Drake45cd9de2000-06-29 19:34:54 +000041 self._parsing = 0
Lars Gustäbele292a242000-09-24 20:19:45 +000042 self._entity_stack = []
Fred Drake45cd9de2000-06-29 19:34:54 +000043
44 # XMLReader methods
45
Lars Gustäbel523b0a62000-09-24 18:54:49 +000046 def parse(self, source):
Lars Gustäbelbb757132000-09-24 20:38:18 +000047 "Parse an XML document from a URL or an InputSource."
Lars Gustäbel523b0a62000-09-24 18:54:49 +000048 source = saxutils.prepare_input_source(source)
49
50 self._source = source
Fred Drake45cd9de2000-06-29 19:34:54 +000051 self.reset()
52 self._cont_handler.setDocumentLocator(self)
Fred Drake16f63292000-10-23 18:09:50 +000053 xmlreader.IncrementalParser.parse(self, source)
Fred Drake45cd9de2000-06-29 19:34:54 +000054
Lars Gustäbel523b0a62000-09-24 18:54:49 +000055 def prepareParser(self, source):
56 if source.getSystemId() != None:
57 self._parser.SetBase(source.getSystemId())
Fred Drake16f63292000-10-23 18:09:50 +000058
Martin v. Löwisfb73bb12001-06-17 07:05:43 +000059 # Redefined setContentHandle to allow changing handlers during parsing
60
61 def setContentHandler(self, handler):
62 xmlreader.IncrementalParser.setContentHandler(self, handler)
63 if self._parsing:
64 self._reset_cont_handler()
65
Fred Drake45cd9de2000-06-29 19:34:54 +000066 def getFeature(self, name):
Martin v. Löwis2066fa02000-09-24 21:17:39 +000067 if name == handler.feature_namespaces:
Lars Gustäbelf43cf312000-09-24 18:29:24 +000068 return self._namespaces
Fred Drake45cd9de2000-06-29 19:34:54 +000069 raise SAXNotRecognizedException("Feature '%s' not recognized" % name)
70
71 def setFeature(self, name, state):
Lars Gustäbelf43cf312000-09-24 18:29:24 +000072 if self._parsing:
73 raise SAXNotSupportedException("Cannot set features while parsing")
Martin v. Löwis2066fa02000-09-24 21:17:39 +000074 if name == handler.feature_namespaces:
Lars Gustäbelf43cf312000-09-24 18:29:24 +000075 self._namespaces = state
76 else:
77 raise SAXNotRecognizedException("Feature '%s' not recognized" %
78 name)
Fred Drake45cd9de2000-06-29 19:34:54 +000079
80 def getProperty(self, name):
Martin v. Löwis05917252001-01-27 08:56:24 +000081 if name == handler.property_lexical_handler:
82 return self._lex_handler_prop
Fred Drake45cd9de2000-06-29 19:34:54 +000083 raise SAXNotRecognizedException("Property '%s' not recognized" % name)
84
85 def setProperty(self, name, value):
Martin v. Löwis05917252001-01-27 08:56:24 +000086 if name == handler.property_lexical_handler:
87 self._lex_handler_prop = value
Martin v. Löwisfb73bb12001-06-17 07:05:43 +000088 if self._parsing:
89 self._reset_lex_handler_prop()
Martin v. Löwis05917252001-01-27 08:56:24 +000090 else:
91 raise SAXNotRecognizedException("Property '%s' not recognized" % name)
Fred Drake45cd9de2000-06-29 19:34:54 +000092
93 # IncrementalParser methods
94
Martin v. Löwisee1dc152000-10-06 21:08:59 +000095 def feed(self, data, isFinal = 0):
Fred Drake45cd9de2000-06-29 19:34:54 +000096 if not self._parsing:
Fred Drake45cd9de2000-06-29 19:34:54 +000097 self.reset()
Lars Gustäbel55b4efd2000-10-14 10:28:01 +000098 self._parsing = 1
Fred Drake45cd9de2000-06-29 19:34:54 +000099 self._cont_handler.startDocument()
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000100
Martin v. Löwisee1dc152000-10-06 21:08:59 +0000101 try:
102 # The isFinal parameter is internal to the expat reader.
103 # If it is set to true, expat will check validity of the entire
104 # document. When feeding chunks, they are not normally final -
105 # except when invoked from close.
106 self._parser.Parse(data, isFinal)
107 except expat.error:
108 error_code = self._parser.ErrorCode
Martin v. Löwis04f49432000-10-09 16:45:54 +0000109 exc = SAXParseException(expat.ErrorString(error_code), None, self)
Martin v. Löwis05917252001-01-27 08:56:24 +0000110 # FIXME: when to invoke error()?
Martin v. Löwis04f49432000-10-09 16:45:54 +0000111 self._err_handler.fatalError(exc)
Fred Drake45cd9de2000-06-29 19:34:54 +0000112
113 def close(self):
Martin v. Löwisee1dc152000-10-06 21:08:59 +0000114 if self._entity_stack:
115 # If we are completing an external entity, do nothing here
116 return
117 self.feed("", isFinal = 1)
118 self._cont_handler.endDocument()
119 self._parsing = 0
Martin v. Löwis05917252001-01-27 08:56:24 +0000120 # break cycle created by expat handlers pointing to our methods
121 self._parser = None
Fred Drake16f63292000-10-23 18:09:50 +0000122
Martin v. Löwisfb73bb12001-06-17 07:05:43 +0000123 def _reset_cont_handler(self):
124 self._parser.ProcessingInstructionHandler = \
125 self._cont_handler.processingInstruction
126 self._parser.CharacterDataHandler = self._cont_handler.characters
127
128 def _reset_lex_handler_prop(self):
129 self._parser.CommentHandler = self._lex_handler_prop.comment
130 self._parser.StartCdataSectionHandler = self._lex_handler_prop.startCDATA
131 self._parser.EndCdataSectionHandler = self._lex_handler_prop.endCDATA
132
Fred Drake45cd9de2000-06-29 19:34:54 +0000133 def reset(self):
134 if self._namespaces:
Fred Drake96ea1962000-09-23 04:49:30 +0000135 self._parser = expat.ParserCreate(None, " ")
Fred Drake45cd9de2000-06-29 19:34:54 +0000136 self._parser.StartElementHandler = self.start_element_ns
137 self._parser.EndElementHandler = self.end_element_ns
138 else:
Fred Drake96ea1962000-09-23 04:49:30 +0000139 self._parser = expat.ParserCreate()
Paul Prescod6c4753f2000-07-04 03:39:33 +0000140 self._parser.StartElementHandler = self.start_element
141 self._parser.EndElementHandler = self.end_element
Fred Drake45cd9de2000-06-29 19:34:54 +0000142
Martin v. Löwisfb73bb12001-06-17 07:05:43 +0000143 self._reset_cont_handler()
Fred Drake45cd9de2000-06-29 19:34:54 +0000144 self._parser.UnparsedEntityDeclHandler = self.unparsed_entity_decl
145 self._parser.NotationDeclHandler = self.notation_decl
146 self._parser.StartNamespaceDeclHandler = self.start_namespace_decl
147 self._parser.EndNamespaceDeclHandler = self.end_namespace_decl
Martin v. Löwis70d39a62001-01-27 09:01:20 +0000148
Martin v. Löwis05917252001-01-27 08:56:24 +0000149 self._decl_handler_prop = None
150 if self._lex_handler_prop:
Martin v. Löwisfb73bb12001-06-17 07:05:43 +0000151 self._reset_lex_handler_prop()
Martin v. Löwis70d39a62001-01-27 09:01:20 +0000152# self._parser.DefaultHandler =
153# self._parser.DefaultHandlerExpand =
154# self._parser.NotStandaloneHandler =
Fred Drake45cd9de2000-06-29 19:34:54 +0000155 self._parser.ExternalEntityRefHandler = self.external_entity_ref
Lars Gustäbelbb757132000-09-24 20:38:18 +0000156
Lars Gustäbel55b4efd2000-10-14 10:28:01 +0000157 self._parsing = 0
Lars Gustäbelbb757132000-09-24 20:38:18 +0000158 self._entity_stack = []
Fred Drake16f63292000-10-23 18:09:50 +0000159
Fred Drake45cd9de2000-06-29 19:34:54 +0000160 # Locator methods
161
162 def getColumnNumber(self):
Martin v. Löwis05917252001-01-27 08:56:24 +0000163 if self._parser is None:
164 return None
Fred Drake45cd9de2000-06-29 19:34:54 +0000165 return self._parser.ErrorColumnNumber
166
167 def getLineNumber(self):
Martin v. Löwis05917252001-01-27 08:56:24 +0000168 if self._parser is None:
169 return 1
Fred Drake45cd9de2000-06-29 19:34:54 +0000170 return self._parser.ErrorLineNumber
171
172 def getPublicId(self):
173 return self._source.getPublicId()
174
175 def getSystemId(self):
Martin v. Löwisee1dc152000-10-06 21:08:59 +0000176 return self._source.getSystemId()
Fred Drake16f63292000-10-23 18:09:50 +0000177
Fred Drake45cd9de2000-06-29 19:34:54 +0000178 # event handlers
Fred Drake45cd9de2000-06-29 19:34:54 +0000179 def start_element(self, name, attrs):
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000180 self._cont_handler.startElement(name, AttributesImpl(attrs))
Fred Drake45cd9de2000-06-29 19:34:54 +0000181
182 def end_element(self, name):
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000183 self._cont_handler.endElement(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000184
185 def start_element_ns(self, name, attrs):
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000186 pair = string.split(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000187 if len(pair) == 1:
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000188 pair = (None, name)
Lars Gustäbeld2f5a9a2000-10-19 07:36:29 +0000189 else:
190 pair = tuple(pair)
Fred Drake45cd9de2000-06-29 19:34:54 +0000191
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000192 newattrs = {}
193 for (aname, value) in attrs.items():
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000194 apair = string.split(aname)
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000195 if len(apair) == 1:
196 apair = (None, aname)
197 else:
198 apair = tuple(apair)
199
200 newattrs[apair] = value
201
Fred Drake16f63292000-10-23 18:09:50 +0000202 self._cont_handler.startElementNS(pair, None,
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000203 AttributesNSImpl(newattrs, {}))
Fred Drake45cd9de2000-06-29 19:34:54 +0000204
205 def end_element_ns(self, name):
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000206 pair = string.split(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000207 if len(pair) == 1:
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000208 pair = (None, name)
Martin v. Löwis05917252001-01-27 08:56:24 +0000209 else:
210 pair = tuple(pair)
Fred Drake16f63292000-10-23 18:09:50 +0000211
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000212 self._cont_handler.endElementNS(pair, None)
Fred Drake45cd9de2000-06-29 19:34:54 +0000213
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000214 # this is not used (call directly to ContentHandler)
Fred Drake45cd9de2000-06-29 19:34:54 +0000215 def processing_instruction(self, target, data):
216 self._cont_handler.processingInstruction(target, data)
217
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000218 # this is not used (call directly to ContentHandler)
Fred Drake45cd9de2000-06-29 19:34:54 +0000219 def character_data(self, data):
220 self._cont_handler.characters(data)
221
222 def start_namespace_decl(self, prefix, uri):
223 self._cont_handler.startPrefixMapping(prefix, uri)
224
225 def end_namespace_decl(self, prefix):
226 self._cont_handler.endPrefixMapping(prefix)
Fred Drake16f63292000-10-23 18:09:50 +0000227
Fred Drake45cd9de2000-06-29 19:34:54 +0000228 def unparsed_entity_decl(self, name, base, sysid, pubid, notation_name):
229 self._dtd_handler.unparsedEntityDecl(name, pubid, sysid, notation_name)
230
231 def notation_decl(self, name, base, sysid, pubid):
232 self._dtd_handler.notationDecl(name, pubid, sysid)
233
234 def external_entity_ref(self, context, base, sysid, pubid):
Fred Drake45cd9de2000-06-29 19:34:54 +0000235 source = self._ent_handler.resolveEntity(pubid, sysid)
Lars Gustäbele292a242000-09-24 20:19:45 +0000236 source = saxutils.prepare_input_source(source,
237 self._source.getSystemId() or
238 "")
Fred Drake16f63292000-10-23 18:09:50 +0000239
Lars Gustäbele292a242000-09-24 20:19:45 +0000240 self._entity_stack.append((self._parser, self._source))
241 self._parser = self._parser.ExternalEntityParserCreate(context)
242 self._source = source
243
244 try:
245 xmlreader.IncrementalParser.parse(self, source)
Lars Gustäbele292a242000-09-24 20:19:45 +0000246 except:
247 return 0 # FIXME: save error info here?
248
249 (self._parser, self._source) = self._entity_stack[-1]
250 del self._entity_stack[-1]
Fred Drake45cd9de2000-06-29 19:34:54 +0000251 return 1
Fred Drake16f63292000-10-23 18:09:50 +0000252
Fred Drake45cd9de2000-06-29 19:34:54 +0000253# ---
Fred Drake16f63292000-10-23 18:09:50 +0000254
Fred Drake45cd9de2000-06-29 19:34:54 +0000255def create_parser(*args, **kwargs):
Fred Drakeddb48672000-09-23 05:32:26 +0000256 return apply(ExpatParser, args, kwargs)
Fred Drake16f63292000-10-23 18:09:50 +0000257
Fred Drake45cd9de2000-06-29 19:34:54 +0000258# ---
259
260if __name__ == "__main__":
261 import xml.sax
262 p = create_parser()
263 p.setContentHandler(xml.sax.XMLGenerator())
264 p.setErrorHandler(xml.sax.ErrorHandler())
265 p.parse("../../../hamlet.xml")