blob: eb46bcd10052fec778a749c68007978caaa92fb4 [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:
12 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
30 self._parsing = 0
Lars Gustäbele292a242000-09-24 20:19:45 +000031 self._entity_stack = []
Fred Drake45cd9de2000-06-29 19:34:54 +000032
33 # XMLReader methods
34
Lars Gustäbel523b0a62000-09-24 18:54:49 +000035 def parse(self, source):
Lars Gustäbelbb757132000-09-24 20:38:18 +000036 "Parse an XML document from a URL or an InputSource."
Lars Gustäbel523b0a62000-09-24 18:54:49 +000037 source = saxutils.prepare_input_source(source)
38
39 self._source = source
Fred Drake45cd9de2000-06-29 19:34:54 +000040 self.reset()
41 self._cont_handler.setDocumentLocator(self)
42 try:
Lars Gustäbel523b0a62000-09-24 18:54:49 +000043 xmlreader.IncrementalParser.parse(self, source)
Fred Drake96ea1962000-09-23 04:49:30 +000044 except expat.error:
Fred Drake45cd9de2000-06-29 19:34:54 +000045 error_code = self._parser.ErrorCode
Lars Gustäbelf43cf312000-09-24 18:29:24 +000046 raise SAXParseException(expat.ErrorString(error_code), None, self)
Fred Drake45cd9de2000-06-29 19:34:54 +000047
Lars Gustäbelbb757132000-09-24 20:38:18 +000048 self._cont_handler.endDocument()
Fred Drake45cd9de2000-06-29 19:34:54 +000049
Lars Gustäbel523b0a62000-09-24 18:54:49 +000050 def prepareParser(self, source):
51 if source.getSystemId() != None:
52 self._parser.SetBase(source.getSystemId())
Fred Drake45cd9de2000-06-29 19:34:54 +000053
54 def getFeature(self, name):
Martin v. Löwis2066fa02000-09-24 21:17:39 +000055 if name == handler.feature_namespaces:
Lars Gustäbelf43cf312000-09-24 18:29:24 +000056 return self._namespaces
Fred Drake45cd9de2000-06-29 19:34:54 +000057 raise SAXNotRecognizedException("Feature '%s' not recognized" % name)
58
59 def setFeature(self, name, state):
Lars Gustäbelf43cf312000-09-24 18:29:24 +000060 if self._parsing:
61 raise SAXNotSupportedException("Cannot set features while parsing")
Martin v. Löwis2066fa02000-09-24 21:17:39 +000062 if name == handler.feature_namespaces:
Lars Gustäbelf43cf312000-09-24 18:29:24 +000063 self._namespaces = state
64 else:
65 raise SAXNotRecognizedException("Feature '%s' not recognized" %
66 name)
Fred Drake45cd9de2000-06-29 19:34:54 +000067
68 def getProperty(self, name):
Fred Drake45cd9de2000-06-29 19:34:54 +000069 raise SAXNotRecognizedException("Property '%s' not recognized" % name)
70
71 def setProperty(self, name, value):
Fred Drake45cd9de2000-06-29 19:34:54 +000072 raise SAXNotRecognizedException("Property '%s' not recognized" % name)
73
74 # IncrementalParser methods
75
76 def feed(self, data):
77 if not self._parsing:
Fred Drakeddb48672000-09-23 05:32:26 +000078 self._parsing = 1
Fred Drake45cd9de2000-06-29 19:34:54 +000079 self.reset()
80 self._cont_handler.startDocument()
Lars Gustäbelf43cf312000-09-24 18:29:24 +000081
82 if not self._parser.Parse(data, 0):
83 msg = pyexpat.ErrorString(self._parser.ErrorCode)
84 raise SAXParseException(msg, None, self)
Fred Drake45cd9de2000-06-29 19:34:54 +000085
86 def close(self):
87 if self._parsing:
88 self._cont_handler.endDocument()
Fred Drakeddb48672000-09-23 05:32:26 +000089 self._parsing = 0
Fred Drake45cd9de2000-06-29 19:34:54 +000090 self._parser.Parse("", 1)
91
92 def reset(self):
93 if self._namespaces:
Fred Drake96ea1962000-09-23 04:49:30 +000094 self._parser = expat.ParserCreate(None, " ")
Fred Drake45cd9de2000-06-29 19:34:54 +000095 self._parser.StartElementHandler = self.start_element_ns
96 self._parser.EndElementHandler = self.end_element_ns
97 else:
Fred Drake96ea1962000-09-23 04:49:30 +000098 self._parser = expat.ParserCreate()
Paul Prescod6c4753f2000-07-04 03:39:33 +000099 self._parser.StartElementHandler = self.start_element
100 self._parser.EndElementHandler = self.end_element
Fred Drake45cd9de2000-06-29 19:34:54 +0000101
102 self._parser.ProcessingInstructionHandler = \
103 self._cont_handler.processingInstruction
104 self._parser.CharacterDataHandler = self._cont_handler.characters
105 self._parser.UnparsedEntityDeclHandler = self.unparsed_entity_decl
106 self._parser.NotationDeclHandler = self.notation_decl
107 self._parser.StartNamespaceDeclHandler = self.start_namespace_decl
108 self._parser.EndNamespaceDeclHandler = self.end_namespace_decl
109# self._parser.CommentHandler =
110# self._parser.StartCdataSectionHandler =
111# self._parser.EndCdataSectionHandler =
112# self._parser.DefaultHandler =
113# self._parser.DefaultHandlerExpand =
114# self._parser.NotStandaloneHandler =
115 self._parser.ExternalEntityRefHandler = self.external_entity_ref
Lars Gustäbelbb757132000-09-24 20:38:18 +0000116
Lars Gustäbelbb757132000-09-24 20:38:18 +0000117 self._entity_stack = []
118
Fred Drake45cd9de2000-06-29 19:34:54 +0000119 # Locator methods
120
121 def getColumnNumber(self):
122 return self._parser.ErrorColumnNumber
123
124 def getLineNumber(self):
125 return self._parser.ErrorLineNumber
126
127 def getPublicId(self):
128 return self._source.getPublicId()
129
130 def getSystemId(self):
131 return self._parser.GetBase()
132
Fred Drake45cd9de2000-06-29 19:34:54 +0000133 # event handlers
Fred Drake45cd9de2000-06-29 19:34:54 +0000134 def start_element(self, name, attrs):
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000135 self._cont_handler.startElement(name, AttributesImpl(attrs))
Fred Drake45cd9de2000-06-29 19:34:54 +0000136
137 def end_element(self, name):
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000138 self._cont_handler.endElement(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000139
140 def start_element_ns(self, name, attrs):
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000141 pair = string.split(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000142 if len(pair) == 1:
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000143 pair = (None, name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000144
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000145 newattrs = {}
146 for (aname, value) in attrs.items():
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000147 apair = string.split(aname)
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000148 if len(apair) == 1:
149 apair = (None, aname)
150 else:
151 apair = tuple(apair)
152
153 newattrs[apair] = value
154
155 self._cont_handler.startElementNS(pair, None,
156 AttributesNSImpl(newattrs, {}))
Fred Drake45cd9de2000-06-29 19:34:54 +0000157
158 def end_element_ns(self, name):
Martin v. Löwis3f0969f2000-09-29 19:00:40 +0000159 pair = string.split(name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000160 if len(pair) == 1:
Lars Gustäbel32bf12e2000-09-24 18:39:23 +0000161 pair = (None, name)
Fred Drake45cd9de2000-06-29 19:34:54 +0000162
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000163 self._cont_handler.endElementNS(pair, None)
Fred Drake45cd9de2000-06-29 19:34:54 +0000164
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000165 # this is not used (call directly to ContentHandler)
Fred Drake45cd9de2000-06-29 19:34:54 +0000166 def processing_instruction(self, target, data):
167 self._cont_handler.processingInstruction(target, data)
168
Lars Gustäbelf43cf312000-09-24 18:29:24 +0000169 # this is not used (call directly to ContentHandler)
Fred Drake45cd9de2000-06-29 19:34:54 +0000170 def character_data(self, data):
171 self._cont_handler.characters(data)
172
173 def start_namespace_decl(self, prefix, uri):
174 self._cont_handler.startPrefixMapping(prefix, uri)
175
176 def end_namespace_decl(self, prefix):
177 self._cont_handler.endPrefixMapping(prefix)
178
179 def unparsed_entity_decl(self, name, base, sysid, pubid, notation_name):
180 self._dtd_handler.unparsedEntityDecl(name, pubid, sysid, notation_name)
181
182 def notation_decl(self, name, base, sysid, pubid):
183 self._dtd_handler.notationDecl(name, pubid, sysid)
184
185 def external_entity_ref(self, context, base, sysid, pubid):
Fred Drake45cd9de2000-06-29 19:34:54 +0000186 source = self._ent_handler.resolveEntity(pubid, sysid)
Lars Gustäbele292a242000-09-24 20:19:45 +0000187 source = saxutils.prepare_input_source(source,
188 self._source.getSystemId() or
189 "")
190
191 self._entity_stack.append((self._parser, self._source))
192 self._parser = self._parser.ExternalEntityParserCreate(context)
193 self._source = source
194
195 try:
196 xmlreader.IncrementalParser.parse(self, source)
197 self.close()
198 except:
199 return 0 # FIXME: save error info here?
200
201 (self._parser, self._source) = self._entity_stack[-1]
202 del self._entity_stack[-1]
Fred Drake45cd9de2000-06-29 19:34:54 +0000203 return 1
204
205# ---
206
207def create_parser(*args, **kwargs):
Fred Drakeddb48672000-09-23 05:32:26 +0000208 return apply(ExpatParser, args, kwargs)
Fred Drake45cd9de2000-06-29 19:34:54 +0000209
210# ---
211
212if __name__ == "__main__":
213 import xml.sax
214 p = create_parser()
215 p.setContentHandler(xml.sax.XMLGenerator())
216 p.setErrorHandler(xml.sax.ErrorHandler())
217 p.parse("../../../hamlet.xml")