blob: 96687f359604ca4002cf386e7f8fb98fb94f1a51 [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. Löwis962c9e72000-10-06 17:41:52 +00009try:
10 from xml.parsers import expat
11except ImportError:
Fred Drake16f63292000-10-23 18:09:50 +000012 raise SAXReaderNotAvailable("expat not supported",None)
Martin v. Löwis2066fa02000-09-24 21:17:39 +000013from xml.sax import xmlreader, saxutils, handler
Fred Drake45cd9de2000-06-29 19:34:54 +000014
Lars Gustäbel32bf12e2000-09-24 18:39:23 +000015AttributesImpl = xmlreader.AttributesImpl
16AttributesNSImpl = xmlreader.AttributesNSImpl
17
Martin v. Löwis3f0969f2000-09-29 19:00:40 +000018import string
19
Fred Drake45cd9de2000-06-29 19:34:54 +000020# --- ExpatParser
21
Fred Drakeddb48672000-09-23 05:32:26 +000022class ExpatParser(xmlreader.IncrementalParser, xmlreader.Locator):
Fred Drake45cd9de2000-06-29 19:34:54 +000023 "SAX driver for the Pyexpat C module."
24
25 def __init__(self, namespaceHandling=0, bufsize=2**16-20):
26 xmlreader.IncrementalParser.__init__(self, bufsize)
Lars Gustäbele292a242000-09-24 20:19:45 +000027 self._source = xmlreader.InputSource()
Fred Drake45cd9de2000-06-29 19:34:54 +000028 self._parser = None
29 self._namespaces = namespaceHandling
Martin v. Löwis05917252001-01-27 08:56:24 +000030 self._lex_handler_prop = None
Fred Drake45cd9de2000-06-29 19:34:54 +000031 self._parsing = 0
Lars Gustäbele292a242000-09-24 20:19:45 +000032 self._entity_stack = []
Fred Drake45cd9de2000-06-29 19:34:54 +000033
34 # XMLReader methods
35
Lars Gustäbel523b0a62000-09-24 18:54:49 +000036 def parse(self, source):
Lars Gustäbelbb757132000-09-24 20:38:18 +000037 "Parse an XML document from a URL or an InputSource."
Lars Gustäbel523b0a62000-09-24 18:54:49 +000038 source = saxutils.prepare_input_source(source)
39
40 self._source = source
Fred Drake45cd9de2000-06-29 19:34:54 +000041 self.reset()
42 self._cont_handler.setDocumentLocator(self)
Fred Drake16f63292000-10-23 18:09:50 +000043 xmlreader.IncrementalParser.parse(self, source)
Fred Drake45cd9de2000-06-29 19:34:54 +000044
Lars Gustäbel523b0a62000-09-24 18:54:49 +000045 def prepareParser(self, source):
46 if source.getSystemId() != None:
47 self._parser.SetBase(source.getSystemId())
Fred Drake16f63292000-10-23 18:09:50 +000048
Fred Drake45cd9de2000-06-29 19:34:54 +000049 def getFeature(self, name):
Martin v. Löwis2066fa02000-09-24 21:17:39 +000050 if name == handler.feature_namespaces:
Lars Gustäbelf43cf312000-09-24 18:29:24 +000051 return self._namespaces
Fred Drake45cd9de2000-06-29 19:34:54 +000052 raise SAXNotRecognizedException("Feature '%s' not recognized" % name)
53
54 def setFeature(self, name, state):
Lars Gustäbelf43cf312000-09-24 18:29:24 +000055 if self._parsing:
56 raise SAXNotSupportedException("Cannot set features while parsing")
Martin v. Löwis2066fa02000-09-24 21:17:39 +000057 if name == handler.feature_namespaces:
Lars Gustäbelf43cf312000-09-24 18:29:24 +000058 self._namespaces = state
59 else:
60 raise SAXNotRecognizedException("Feature '%s' not recognized" %
61 name)
Fred Drake45cd9de2000-06-29 19:34:54 +000062
63 def getProperty(self, name):
Martin v. Löwis05917252001-01-27 08:56:24 +000064 if name == handler.property_lexical_handler:
65 return self._lex_handler_prop
Fred Drake45cd9de2000-06-29 19:34:54 +000066 raise SAXNotRecognizedException("Property '%s' not recognized" % name)
67
68 def setProperty(self, name, value):
Martin v. Löwis05917252001-01-27 08:56:24 +000069 if name == handler.property_lexical_handler:
70 self._lex_handler_prop = value
71 else:
72 raise SAXNotRecognizedException("Property '%s' not recognized" % name)
Fred Drake45cd9de2000-06-29 19:34:54 +000073
74 # IncrementalParser methods
75
Martin v. Löwisee1dc152000-10-06 21:08:59 +000076 def feed(self, data, isFinal = 0):
Fred Drake45cd9de2000-06-29 19:34:54 +000077 if not self._parsing:
Fred Drake45cd9de2000-06-29 19:34:54 +000078 self.reset()
Lars Gustäbel55b4efd2000-10-14 10:28:01 +000079 self._parsing = 1
Fred Drake45cd9de2000-06-29 19:34:54 +000080 self._cont_handler.startDocument()
Lars Gustäbelf43cf312000-09-24 18:29:24 +000081
Martin v. Löwisee1dc152000-10-06 21:08:59 +000082 try:
83 # The isFinal parameter is internal to the expat reader.
84 # If it is set to true, expat will check validity of the entire
85 # document. When feeding chunks, they are not normally final -
86 # except when invoked from close.
87 self._parser.Parse(data, isFinal)
88 except expat.error:
89 error_code = self._parser.ErrorCode
Martin v. Löwis04f49432000-10-09 16:45:54 +000090 exc = SAXParseException(expat.ErrorString(error_code), None, self)
Martin v. Löwis05917252001-01-27 08:56:24 +000091 # FIXME: when to invoke error()?
Martin v. Löwis04f49432000-10-09 16:45:54 +000092 self._err_handler.fatalError(exc)
Fred Drake45cd9de2000-06-29 19:34:54 +000093
94 def close(self):
Martin v. Löwisee1dc152000-10-06 21:08:59 +000095 if self._entity_stack:
96 # If we are completing an external entity, do nothing here
97 return
98 self.feed("", isFinal = 1)
99 self._cont_handler.endDocument()
100 self._parsing = 0
Martin v. Löwis05917252001-01-27 08:56:24 +0000101 # break cycle created by expat handlers pointing to our methods
102 self._parser = None
Fred Drake16f63292000-10-23 18:09:50 +0000103
Fred Drake45cd9de2000-06-29 19:34:54 +0000104 def reset(self):
105 if self._namespaces:
Fred Drake96ea1962000-09-23 04:49:30 +0000106 self._parser = expat.ParserCreate(None, " ")
Fred Drake45cd9de2000-06-29 19:34:54 +0000107 self._parser.StartElementHandler = self.start_element_ns
108 self._parser.EndElementHandler = self.end_element_ns
109 else:
Fred Drake96ea1962000-09-23 04:49:30 +0000110 self._parser = expat.ParserCreate()
Paul Prescod6c4753f2000-07-04 03:39:33 +0000111 self._parser.StartElementHandler = self.start_element
112 self._parser.EndElementHandler = self.end_element
Fred Drake45cd9de2000-06-29 19:34:54 +0000113
114 self._parser.ProcessingInstructionHandler = \
115 self._cont_handler.processingInstruction
116 self._parser.CharacterDataHandler = self._cont_handler.characters
117 self._parser.UnparsedEntityDeclHandler = self.unparsed_entity_decl
118 self._parser.NotationDeclHandler = self.notation_decl
119 self._parser.StartNamespaceDeclHandler = self.start_namespace_decl
120 self._parser.EndNamespaceDeclHandler = self.end_namespace_decl
Martin v. Löwis05917252001-01-27 08:56:24 +0000121
122 self._decl_handler_prop = None
123 if self._lex_handler_prop:
124 self._parser.CommentHandler = self._lex_handler_prop.comment
125 self._parser.StartCdataSectionHandler = self._lex_handler_prop.startCDATA
126 self._parser.EndCdataSectionHandler = self._lex_handler_prop.endCDATA
127# self._parser.DefaultHandler =
128# self._parser.DefaultHandlerExpand =
129# self._parser.NotStandaloneHandler =
Fred Drake45cd9de2000-06-29 19:34:54 +0000130 self._parser.ExternalEntityRefHandler = self.external_entity_ref
Lars Gustäbelbb757132000-09-24 20:38:18 +0000131
Lars Gustäbel55b4efd2000-10-14 10:28:01 +0000132 self._parsing = 0
Lars Gustäbelbb757132000-09-24 20:38:18 +0000133 self._entity_stack = []
Fred Drake16f63292000-10-23 18:09:50 +0000134
Fred Drake45cd9de2000-06-29 19:34:54 +0000135 # Locator methods
136
137 def getColumnNumber(self):
Martin v. Löwis05917252001-01-27 08:56:24 +0000138 if self._parser is None:
139 return None
Fred Drake45cd9de2000-06-29 19:34:54 +0000140 return self._parser.ErrorColumnNumber
141
142 def getLineNumber(self):
Martin v. Löwis05917252001-01-27 08:56:24 +0000143 if self._parser is None:
144 return 1
Fred Drake45cd9de2000-06-29 19:34:54 +0000145 return self._parser.ErrorLineNumber
146
147 def getPublicId(self):
148 return self._source.getPublicId()
149
150 def getSystemId(self):
Martin v. Löwisee1dc152000-10-06 21:08:59 +0000151 return self._source.getSystemId()
Fred Drake16f63292000-10-23 18:09:50 +0000152
Fred Drake45cd9de2000-06-29 19:34:54 +0000153 # event handlers
Fred Drake45cd9de2000-06-29 19:34:54 +0000154 def start_element(self, name, attrs):
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000155 self._cont_handler.startElement(name, AttributesImpl(attrs))
Fred Drake45cd9de2000-06-29 19:34:54 +0000156
157 def end_element(self, name):
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000158 self._cont_handler.endElement(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000159
160 def start_element_ns(self, name, attrs):
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000161 pair = string.split(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000162 if len(pair) == 1:
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000163 pair = (None, name)
Lars Gustäbeld2f5a9a2000-10-19 07:36:29 +0000164 else:
165 pair = tuple(pair)
Fred Drake45cd9de2000-06-29 19:34:54 +0000166
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000167 newattrs = {}
168 for (aname, value) in attrs.items():
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000169 apair = string.split(aname)
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000170 if len(apair) == 1:
171 apair = (None, aname)
172 else:
173 apair = tuple(apair)
174
175 newattrs[apair] = value
176
Fred Drake16f63292000-10-23 18:09:50 +0000177 self._cont_handler.startElementNS(pair, None,
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000178 AttributesNSImpl(newattrs, {}))
Fred Drake45cd9de2000-06-29 19:34:54 +0000179
180 def end_element_ns(self, name):
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000181 pair = string.split(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000182 if len(pair) == 1:
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000183 pair = (None, name)
Martin v. Löwis05917252001-01-27 08:56:24 +0000184 else:
185 pair = tuple(pair)
Fred Drake16f63292000-10-23 18:09:50 +0000186
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000187 self._cont_handler.endElementNS(pair, None)
Fred Drake45cd9de2000-06-29 19:34:54 +0000188
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000189 # this is not used (call directly to ContentHandler)
Fred Drake45cd9de2000-06-29 19:34:54 +0000190 def processing_instruction(self, target, data):
191 self._cont_handler.processingInstruction(target, data)
192
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000193 # this is not used (call directly to ContentHandler)
Fred Drake45cd9de2000-06-29 19:34:54 +0000194 def character_data(self, data):
195 self._cont_handler.characters(data)
196
197 def start_namespace_decl(self, prefix, uri):
198 self._cont_handler.startPrefixMapping(prefix, uri)
199
200 def end_namespace_decl(self, prefix):
201 self._cont_handler.endPrefixMapping(prefix)
Fred Drake16f63292000-10-23 18:09:50 +0000202
Fred Drake45cd9de2000-06-29 19:34:54 +0000203 def unparsed_entity_decl(self, name, base, sysid, pubid, notation_name):
204 self._dtd_handler.unparsedEntityDecl(name, pubid, sysid, notation_name)
205
206 def notation_decl(self, name, base, sysid, pubid):
207 self._dtd_handler.notationDecl(name, pubid, sysid)
208
209 def external_entity_ref(self, context, base, sysid, pubid):
Fred Drake45cd9de2000-06-29 19:34:54 +0000210 source = self._ent_handler.resolveEntity(pubid, sysid)
Lars Gustäbele292a242000-09-24 20:19:45 +0000211 source = saxutils.prepare_input_source(source,
212 self._source.getSystemId() or
213 "")
Fred Drake16f63292000-10-23 18:09:50 +0000214
Lars Gustäbele292a242000-09-24 20:19:45 +0000215 self._entity_stack.append((self._parser, self._source))
216 self._parser = self._parser.ExternalEntityParserCreate(context)
217 self._source = source
218
219 try:
220 xmlreader.IncrementalParser.parse(self, source)
Lars Gustäbele292a242000-09-24 20:19:45 +0000221 except:
222 return 0 # FIXME: save error info here?
223
224 (self._parser, self._source) = self._entity_stack[-1]
225 del self._entity_stack[-1]
Fred Drake45cd9de2000-06-29 19:34:54 +0000226 return 1
Fred Drake16f63292000-10-23 18:09:50 +0000227
Fred Drake45cd9de2000-06-29 19:34:54 +0000228# ---
Fred Drake16f63292000-10-23 18:09:50 +0000229
Fred Drake45cd9de2000-06-29 19:34:54 +0000230def create_parser(*args, **kwargs):
Fred Drakeddb48672000-09-23 05:32:26 +0000231 return apply(ExpatParser, args, kwargs)
Fred Drake16f63292000-10-23 18:09:50 +0000232
Fred Drake45cd9de2000-06-29 19:34:54 +0000233# ---
234
235if __name__ == "__main__":
236 import xml.sax
237 p = create_parser()
238 p.setContentHandler(xml.sax.XMLGenerator())
239 p.setErrorHandler(xml.sax.ErrorHandler())
240 p.parse("../../../hamlet.xml")