Python code coverage for Lib/xml/sax/expatreader.py

#	count	content
1	n/a	"""
2	n/a	SAX driver for the pyexpat C module. This driver works with
3	n/a	pyexpat.__version__ == '2.22'.
4	n/a	"""
5	n/a
6	n/a	version = "0.20"
7	n/a
8	n/a	from xml.sax._exceptions import *
9	n/a	from xml.sax.handler import feature_validation, feature_namespaces
10	n/a	from xml.sax.handler import feature_namespace_prefixes
11	n/a	from xml.sax.handler import feature_external_ges, feature_external_pes
12	n/a	from xml.sax.handler import feature_string_interning
13	n/a	from xml.sax.handler import property_xml_string, property_interning_dict
14	n/a
15	n/a	# xml.parsers.expat does not raise ImportError in Jython
16	n/a	import sys
17	n/a	if sys.platform[:4] == "java":
18	n/a	raise SAXReaderNotAvailable("expat not available in Java", None)
19	n/a	del sys
20	n/a
21	n/a	try:
22	n/a	from xml.parsers import expat
23	n/a	except ImportError:
24	n/a	raise SAXReaderNotAvailable("expat not supported", None)
25	n/a	else:
26	n/a	if not hasattr(expat, "ParserCreate"):
27	n/a	raise SAXReaderNotAvailable("expat not supported", None)
28	n/a	from xml.sax import xmlreader, saxutils, handler
29	n/a
30	n/a	AttributesImpl = xmlreader.AttributesImpl
31	n/a	AttributesNSImpl = xmlreader.AttributesNSImpl
32	n/a
33	n/a	# If we're using a sufficiently recent version of Python, we can use
34	n/a	# weak references to avoid cycles between the parser and content
35	n/a	# handler, otherwise we'll just have to pretend.
36	n/a	try:
37	n/a	import _weakref
38	n/a	except ImportError:
39	n/a	def _mkproxy(o):
40	n/a	return o
41	n/a	else:
42	n/a	import weakref
43	n/a	_mkproxy = weakref.proxy
44	n/a	del weakref, _weakref
45	n/a
46	n/a	class _ClosedParser:
47	n/a	pass
48	n/a
49	n/a	# --- ExpatLocator
50	n/a
51	n/a	class ExpatLocator(xmlreader.Locator):
52	n/a	"""Locator for use with the ExpatParser class.
53	n/a
54	n/a	This uses a weak reference to the parser object to avoid creating
55	n/a	a circular reference between the parser and the content handler.
56	n/a	"""
57	n/a	def __init__(self, parser):
58	n/a	self._ref = _mkproxy(parser)
59	n/a
60	n/a	def getColumnNumber(self):
61	n/a	parser = self._ref
62	n/a	if parser._parser is None:
63	n/a	return None
64	n/a	return parser._parser.ErrorColumnNumber
65	n/a
66	n/a	def getLineNumber(self):
67	n/a	parser = self._ref
68	n/a	if parser._parser is None:
69	n/a	return 1
70	n/a	return parser._parser.ErrorLineNumber
71	n/a
72	n/a	def getPublicId(self):
73	n/a	parser = self._ref
74	n/a	if parser is None:
75	n/a	return None
76	n/a	return parser._source.getPublicId()
77	n/a
78	n/a	def getSystemId(self):
79	n/a	parser = self._ref
80	n/a	if parser is None:
81	n/a	return None
82	n/a	return parser._source.getSystemId()
83	n/a
84	n/a
85	n/a	# --- ExpatParser
86	n/a
87	n/a	class ExpatParser(xmlreader.IncrementalParser, xmlreader.Locator):
88	n/a	"""SAX driver for the pyexpat C module."""
89	n/a
90	n/a	def __init__(self, namespaceHandling=0, bufsize=2**16-20):
91	n/a	xmlreader.IncrementalParser.__init__(self, bufsize)
92	n/a	self._source = xmlreader.InputSource()
93	n/a	self._parser = None
94	n/a	self._namespaces = namespaceHandling
95	n/a	self._lex_handler_prop = None
96	n/a	self._parsing = 0
97	n/a	self._entity_stack = []
98	n/a	self._external_ges = 1
99	n/a	self._interning = None
100	n/a
101	n/a	# XMLReader methods
102	n/a
103	n/a	def parse(self, source):
104	n/a	"Parse an XML document from a URL or an InputSource."
105	n/a	source = saxutils.prepare_input_source(source)
106	n/a
107	n/a	self._source = source
108	n/a	self.reset()
109	n/a	self._cont_handler.setDocumentLocator(ExpatLocator(self))
110	n/a	xmlreader.IncrementalParser.parse(self, source)
111	n/a
112	n/a	def prepareParser(self, source):
113	n/a	if source.getSystemId() is not None:
114	n/a	self._parser.SetBase(source.getSystemId())
115	n/a
116	n/a	# Redefined setContentHandler to allow changing handlers during parsing
117	n/a
118	n/a	def setContentHandler(self, handler):
119	n/a	xmlreader.IncrementalParser.setContentHandler(self, handler)
120	n/a	if self._parsing:
121	n/a	self._reset_cont_handler()
122	n/a
123	n/a	def getFeature(self, name):
124	n/a	if name == feature_namespaces:
125	n/a	return self._namespaces
126	n/a	elif name == feature_string_interning:
127	n/a	return self._interning is not None
128	n/a	elif name in (feature_validation, feature_external_pes,
129	n/a	feature_namespace_prefixes):
130	n/a	return 0
131	n/a	elif name == feature_external_ges:
132	n/a	return self._external_ges
133	n/a	raise SAXNotRecognizedException("Feature '%s' not recognized" % name)
134	n/a
135	n/a	def setFeature(self, name, state):
136	n/a	if self._parsing:
137	n/a	raise SAXNotSupportedException("Cannot set features while parsing")
138	n/a
139	n/a	if name == feature_namespaces:
140	n/a	self._namespaces = state
141	n/a	elif name == feature_external_ges:
142	n/a	self._external_ges = state
143	n/a	elif name == feature_string_interning:
144	n/a	if state:
145	n/a	if self._interning is None:
146	n/a	self._interning = {}
147	n/a	else:
148	n/a	self._interning = None
149	n/a	elif name == feature_validation:
150	n/a	if state:
151	n/a	raise SAXNotSupportedException(
152	n/a	"expat does not support validation")
153	n/a	elif name == feature_external_pes:
154	n/a	if state:
155	n/a	raise SAXNotSupportedException(
156	n/a	"expat does not read external parameter entities")
157	n/a	elif name == feature_namespace_prefixes:
158	n/a	if state:
159	n/a	raise SAXNotSupportedException(
160	n/a	"expat does not report namespace prefixes")
161	n/a	else:
162	n/a	raise SAXNotRecognizedException(
163	n/a	"Feature '%s' not recognized" % name)
164	n/a
165	n/a	def getProperty(self, name):
166	n/a	if name == handler.property_lexical_handler:
167	n/a	return self._lex_handler_prop
168	n/a	elif name == property_interning_dict:
169	n/a	return self._interning
170	n/a	elif name == property_xml_string:
171	n/a	if self._parser:
172	n/a	if hasattr(self._parser, "GetInputContext"):
173	n/a	return self._parser.GetInputContext()
174	n/a	else:
175	n/a	raise SAXNotRecognizedException(
176	n/a	"This version of expat does not support getting"
177	n/a	" the XML string")
178	n/a	else:
179	n/a	raise SAXNotSupportedException(
180	n/a	"XML string cannot be returned when not parsing")
181	n/a	raise SAXNotRecognizedException("Property '%s' not recognized" % name)
182	n/a
183	n/a	def setProperty(self, name, value):
184	n/a	if name == handler.property_lexical_handler:
185	n/a	self._lex_handler_prop = value
186	n/a	if self._parsing:
187	n/a	self._reset_lex_handler_prop()
188	n/a	elif name == property_interning_dict:
189	n/a	self._interning = value
190	n/a	elif name == property_xml_string:
191	n/a	raise SAXNotSupportedException("Property '%s' cannot be set" %
192	n/a	name)
193	n/a	else:
194	n/a	raise SAXNotRecognizedException("Property '%s' not recognized" %
195	n/a	name)
196	n/a
197	n/a	# IncrementalParser methods
198	n/a
199	n/a	def feed(self, data, isFinal = 0):
200	n/a	if not self._parsing:
201	n/a	self.reset()
202	n/a	self._parsing = 1
203	n/a	self._cont_handler.startDocument()
204	n/a
205	n/a	try:
206	n/a	# The isFinal parameter is internal to the expat reader.
207	n/a	# If it is set to true, expat will check validity of the entire
208	n/a	# document. When feeding chunks, they are not normally final -
209	n/a	# except when invoked from close.
210	n/a	self._parser.Parse(data, isFinal)
211	n/a	except expat.error as e:
212	n/a	exc = SAXParseException(expat.ErrorString(e.code), e, self)
213	n/a	# FIXME: when to invoke error()?
214	n/a	self._err_handler.fatalError(exc)
215	n/a
216	n/a	def close(self):
217	n/a	if (self._entity_stack or self._parser is None or
218	n/a	isinstance(self._parser, _ClosedParser)):
219	n/a	# If we are completing an external entity, do nothing here
220	n/a	return
221	n/a	try:
222	n/a	self.feed("", isFinal = 1)
223	n/a	self._cont_handler.endDocument()
224	n/a	self._parsing = 0
225	n/a	# break cycle created by expat handlers pointing to our methods
226	n/a	self._parser = None
227	n/a	finally:
228	n/a	self._parsing = 0
229	n/a	if self._parser is not None:
230	n/a	# Keep ErrorColumnNumber and ErrorLineNumber after closing.
231	n/a	parser = _ClosedParser()
232	n/a	parser.ErrorColumnNumber = self._parser.ErrorColumnNumber
233	n/a	parser.ErrorLineNumber = self._parser.ErrorLineNumber
234	n/a	self._parser = parser
235	n/a	try:
236	n/a	file = self._source.getCharacterStream()
237	n/a	if file is not None:
238	n/a	file.close()
239	n/a	finally:
240	n/a	file = self._source.getByteStream()
241	n/a	if file is not None:
242	n/a	file.close()
243	n/a
244	n/a	def _reset_cont_handler(self):
245	n/a	self._parser.ProcessingInstructionHandler = \
246	n/a	self._cont_handler.processingInstruction
247	n/a	self._parser.CharacterDataHandler = self._cont_handler.characters
248	n/a
249	n/a	def _reset_lex_handler_prop(self):
250	n/a	lex = self._lex_handler_prop
251	n/a	parser = self._parser
252	n/a	if lex is None:
253	n/a	parser.CommentHandler = None
254	n/a	parser.StartCdataSectionHandler = None
255	n/a	parser.EndCdataSectionHandler = None
256	n/a	parser.StartDoctypeDeclHandler = None
257	n/a	parser.EndDoctypeDeclHandler = None
258	n/a	else:
259	n/a	parser.CommentHandler = lex.comment
260	n/a	parser.StartCdataSectionHandler = lex.startCDATA
261	n/a	parser.EndCdataSectionHandler = lex.endCDATA
262	n/a	parser.StartDoctypeDeclHandler = self.start_doctype_decl
263	n/a	parser.EndDoctypeDeclHandler = lex.endDTD
264	n/a
265	n/a	def reset(self):
266	n/a	if self._namespaces:
267	n/a	self._parser = expat.ParserCreate(self._source.getEncoding(), " ",
268	n/a	intern=self._interning)
269	n/a	self._parser.namespace_prefixes = 1
270	n/a	self._parser.StartElementHandler = self.start_element_ns
271	n/a	self._parser.EndElementHandler = self.end_element_ns
272	n/a	else:
273	n/a	self._parser = expat.ParserCreate(self._source.getEncoding(),
274	n/a	intern = self._interning)
275	n/a	self._parser.StartElementHandler = self.start_element
276	n/a	self._parser.EndElementHandler = self.end_element
277	n/a
278	n/a	self._reset_cont_handler()
279	n/a	self._parser.UnparsedEntityDeclHandler = self.unparsed_entity_decl
280	n/a	self._parser.NotationDeclHandler = self.notation_decl
281	n/a	self._parser.StartNamespaceDeclHandler = self.start_namespace_decl
282	n/a	self._parser.EndNamespaceDeclHandler = self.end_namespace_decl
283	n/a
284	n/a	self._decl_handler_prop = None
285	n/a	if self._lex_handler_prop:
286	n/a	self._reset_lex_handler_prop()
287	n/a	# self._parser.DefaultHandler =
288	n/a	# self._parser.DefaultHandlerExpand =
289	n/a	# self._parser.NotStandaloneHandler =
290	n/a	self._parser.ExternalEntityRefHandler = self.external_entity_ref
291	n/a	try:
292	n/a	self._parser.SkippedEntityHandler = self.skipped_entity_handler
293	n/a	except AttributeError:
294	n/a	# This pyexpat does not support SkippedEntity
295	n/a	pass
296	n/a	self._parser.SetParamEntityParsing(
297	n/a	expat.XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE)
298	n/a
299	n/a	self._parsing = 0
300	n/a	self._entity_stack = []
301	n/a
302	n/a	# Locator methods
303	n/a
304	n/a	def getColumnNumber(self):
305	n/a	if self._parser is None:
306	n/a	return None
307	n/a	return self._parser.ErrorColumnNumber
308	n/a
309	n/a	def getLineNumber(self):
310	n/a	if self._parser is None:
311	n/a	return 1
312	n/a	return self._parser.ErrorLineNumber
313	n/a
314	n/a	def getPublicId(self):
315	n/a	return self._source.getPublicId()
316	n/a
317	n/a	def getSystemId(self):
318	n/a	return self._source.getSystemId()
319	n/a
320	n/a	# event handlers
321	n/a	def start_element(self, name, attrs):
322	n/a	self._cont_handler.startElement(name, AttributesImpl(attrs))
323	n/a
324	n/a	def end_element(self, name):
325	n/a	self._cont_handler.endElement(name)
326	n/a
327	n/a	def start_element_ns(self, name, attrs):
328	n/a	pair = name.split()
329	n/a	if len(pair) == 1:
330	n/a	# no namespace
331	n/a	pair = (None, name)
332	n/a	elif len(pair) == 3:
333	n/a	pair = pair[0], pair[1]
334	n/a	else:
335	n/a	# default namespace
336	n/a	pair = tuple(pair)
337	n/a
338	n/a	newattrs = {}
339	n/a	qnames = {}
340	n/a	for (aname, value) in attrs.items():
341	n/a	parts = aname.split()
342	n/a	length = len(parts)
343	n/a	if length == 1:
344	n/a	# no namespace
345	n/a	qname = aname
346	n/a	apair = (None, aname)
347	n/a	elif length == 3:
348	n/a	qname = "%s:%s" % (parts[2], parts[1])
349	n/a	apair = parts[0], parts[1]
350	n/a	else:
351	n/a	# default namespace
352	n/a	qname = parts[1]
353	n/a	apair = tuple(parts)
354	n/a
355	n/a	newattrs[apair] = value
356	n/a	qnames[apair] = qname
357	n/a
358	n/a	self._cont_handler.startElementNS(pair, None,
359	n/a	AttributesNSImpl(newattrs, qnames))
360	n/a
361	n/a	def end_element_ns(self, name):
362	n/a	pair = name.split()
363	n/a	if len(pair) == 1:
364	n/a	pair = (None, name)
365	n/a	elif len(pair) == 3:
366	n/a	pair = pair[0], pair[1]
367	n/a	else:
368	n/a	pair = tuple(pair)
369	n/a
370	n/a	self._cont_handler.endElementNS(pair, None)
371	n/a
372	n/a	# this is not used (call directly to ContentHandler)
373	n/a	def processing_instruction(self, target, data):
374	n/a	self._cont_handler.processingInstruction(target, data)
375	n/a
376	n/a	# this is not used (call directly to ContentHandler)
377	n/a	def character_data(self, data):
378	n/a	self._cont_handler.characters(data)
379	n/a
380	n/a	def start_namespace_decl(self, prefix, uri):
381	n/a	self._cont_handler.startPrefixMapping(prefix, uri)
382	n/a
383	n/a	def end_namespace_decl(self, prefix):
384	n/a	self._cont_handler.endPrefixMapping(prefix)
385	n/a
386	n/a	def start_doctype_decl(self, name, sysid, pubid, has_internal_subset):
387	n/a	self._lex_handler_prop.startDTD(name, pubid, sysid)
388	n/a
389	n/a	def unparsed_entity_decl(self, name, base, sysid, pubid, notation_name):
390	n/a	self._dtd_handler.unparsedEntityDecl(name, pubid, sysid, notation_name)
391	n/a
392	n/a	def notation_decl(self, name, base, sysid, pubid):
393	n/a	self._dtd_handler.notationDecl(name, pubid, sysid)
394	n/a
395	n/a	def external_entity_ref(self, context, base, sysid, pubid):
396	n/a	if not self._external_ges:
397	n/a	return 1
398	n/a
399	n/a	source = self._ent_handler.resolveEntity(pubid, sysid)
400	n/a	source = saxutils.prepare_input_source(source,
401	n/a	self._source.getSystemId() or
402	n/a	"")
403	n/a
404	n/a	self._entity_stack.append((self._parser, self._source))
405	n/a	self._parser = self._parser.ExternalEntityParserCreate(context)
406	n/a	self._source = source
407	n/a
408	n/a	try:
409	n/a	xmlreader.IncrementalParser.parse(self, source)
410	n/a	except:
411	n/a	return 0 # FIXME: save error info here?
412	n/a
413	n/a	(self._parser, self._source) = self._entity_stack[-1]
414	n/a	del self._entity_stack[-1]
415	n/a	return 1
416	n/a
417	n/a	def skipped_entity_handler(self, name, is_pe):
418	n/a	if is_pe:
419	n/a	# The SAX spec requires to report skipped PEs with a '%'
420	n/a	name = '%'+name
421	n/a	self._cont_handler.skippedEntity(name)
422	n/a
423	n/a	# ---
424	n/a
425	n/a	def create_parser(args, *kwargs):
426	n/a	return ExpatParser(args, *kwargs)
427	n/a
428	n/a	# ---
429	n/a
430	n/a	if __name__ == "__main__":
431	n/a	import xml.sax.saxutils
432	n/a	p = create_parser()
433	n/a	p.setContentHandler(xml.sax.saxutils.XMLGenerator())
434	n/a	p.setErrorHandler(xml.sax.ErrorHandler())
435	n/a	p.parse("http://www.ibiblio.org/xml/examples/shakespeare/hamlet.xml")