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

#	count	content
1	n/a	"""An XML Reader is the SAX 2 name for an XML parser. XML Parsers
2	n/a	should be based on this code. """
3	n/a
4	n/a	from . import handler
5	n/a
6	n/a	from ._exceptions import SAXNotSupportedException, SAXNotRecognizedException
7	n/a
8	n/a
9	n/a	# ===== XMLREADER =====
10	n/a
11	n/a	class XMLReader:
12	n/a	"""Interface for reading an XML document using callbacks.
13	n/a
14	n/a	XMLReader is the interface that an XML parser's SAX2 driver must
15	n/a	implement. This interface allows an application to set and query
16	n/a	features and properties in the parser, to register event handlers
17	n/a	for document processing, and to initiate a document parse.
18	n/a
19	n/a	All SAX interfaces are assumed to be synchronous: the parse
20	n/a	methods must not return until parsing is complete, and readers
21	n/a	must wait for an event-handler callback to return before reporting
22	n/a	the next event."""
23	n/a
24	n/a	def __init__(self):
25	n/a	self._cont_handler = handler.ContentHandler()
26	n/a	self._dtd_handler = handler.DTDHandler()
27	n/a	self._ent_handler = handler.EntityResolver()
28	n/a	self._err_handler = handler.ErrorHandler()
29	n/a
30	n/a	def parse(self, source):
31	n/a	"Parse an XML document from a system identifier or an InputSource."
32	n/a	raise NotImplementedError("This method must be implemented!")
33	n/a
34	n/a	def getContentHandler(self):
35	n/a	"Returns the current ContentHandler."
36	n/a	return self._cont_handler
37	n/a
38	n/a	def setContentHandler(self, handler):
39	n/a	"Registers a new object to receive document content events."
40	n/a	self._cont_handler = handler
41	n/a
42	n/a	def getDTDHandler(self):
43	n/a	"Returns the current DTD handler."
44	n/a	return self._dtd_handler
45	n/a
46	n/a	def setDTDHandler(self, handler):
47	n/a	"Register an object to receive basic DTD-related events."
48	n/a	self._dtd_handler = handler
49	n/a
50	n/a	def getEntityResolver(self):
51	n/a	"Returns the current EntityResolver."
52	n/a	return self._ent_handler
53	n/a
54	n/a	def setEntityResolver(self, resolver):
55	n/a	"Register an object to resolve external entities."
56	n/a	self._ent_handler = resolver
57	n/a
58	n/a	def getErrorHandler(self):
59	n/a	"Returns the current ErrorHandler."
60	n/a	return self._err_handler
61	n/a
62	n/a	def setErrorHandler(self, handler):
63	n/a	"Register an object to receive error-message events."
64	n/a	self._err_handler = handler
65	n/a
66	n/a	def setLocale(self, locale):
67	n/a	"""Allow an application to set the locale for errors and warnings.
68	n/a
69	n/a	SAX parsers are not required to provide localization for errors
70	n/a	and warnings; if they cannot support the requested locale,
71	n/a	however, they must raise a SAX exception. Applications may
72	n/a	request a locale change in the middle of a parse."""
73	n/a	raise SAXNotSupportedException("Locale support not implemented")
74	n/a
75	n/a	def getFeature(self, name):
76	n/a	"Looks up and returns the state of a SAX2 feature."
77	n/a	raise SAXNotRecognizedException("Feature '%s' not recognized" % name)
78	n/a
79	n/a	def setFeature(self, name, state):
80	n/a	"Sets the state of a SAX2 feature."
81	n/a	raise SAXNotRecognizedException("Feature '%s' not recognized" % name)
82	n/a
83	n/a	def getProperty(self, name):
84	n/a	"Looks up and returns the value of a SAX2 property."
85	n/a	raise SAXNotRecognizedException("Property '%s' not recognized" % name)
86	n/a
87	n/a	def setProperty(self, name, value):
88	n/a	"Sets the value of a SAX2 property."
89	n/a	raise SAXNotRecognizedException("Property '%s' not recognized" % name)
90	n/a
91	n/a	class IncrementalParser(XMLReader):
92	n/a	"""This interface adds three extra methods to the XMLReader
93	n/a	interface that allow XML parsers to support incremental
94	n/a	parsing. Support for this interface is optional, since not all
95	n/a	underlying XML parsers support this functionality.
96	n/a
97	n/a	When the parser is instantiated it is ready to begin accepting
98	n/a	data from the feed method immediately. After parsing has been
99	n/a	finished with a call to close the reset method must be called to
100	n/a	make the parser ready to accept new data, either from feed or
101	n/a	using the parse method.
102	n/a
103	n/a	Note that these methods must _not_ be called during parsing, that
104	n/a	is, after parse has been called and before it returns.
105	n/a
106	n/a	By default, the class also implements the parse method of the XMLReader
107	n/a	interface using the feed, close and reset methods of the
108	n/a	IncrementalParser interface as a convenience to SAX 2.0 driver
109	n/a	writers."""
110	n/a
111	n/a	def __init__(self, bufsize=2**16):
112	n/a	self._bufsize = bufsize
113	n/a	XMLReader.__init__(self)
114	n/a
115	n/a	def parse(self, source):
116	n/a	from . import saxutils
117	n/a	source = saxutils.prepare_input_source(source)
118	n/a
119	n/a	self.prepareParser(source)
120	n/a	file = source.getCharacterStream()
121	n/a	if file is None:
122	n/a	file = source.getByteStream()
123	n/a	buffer = file.read(self._bufsize)
124	n/a	while buffer:
125	n/a	self.feed(buffer)
126	n/a	buffer = file.read(self._bufsize)
127	n/a	self.close()
128	n/a
129	n/a	def feed(self, data):
130	n/a	"""This method gives the raw XML data in the data parameter to
131	n/a	the parser and makes it parse the data, emitting the
132	n/a	corresponding events. It is allowed for XML constructs to be
133	n/a	split across several calls to feed.
134	n/a
135	n/a	feed may raise SAXException."""
136	n/a	raise NotImplementedError("This method must be implemented!")
137	n/a
138	n/a	def prepareParser(self, source):
139	n/a	"""This method is called by the parse implementation to allow
140	n/a	the SAX 2.0 driver to prepare itself for parsing."""
141	n/a	raise NotImplementedError("prepareParser must be overridden!")
142	n/a
143	n/a	def close(self):
144	n/a	"""This method is called when the entire XML document has been
145	n/a	passed to the parser through the feed method, to notify the
146	n/a	parser that there are no more data. This allows the parser to
147	n/a	do the final checks on the document and empty the internal
148	n/a	data buffer.
149	n/a
150	n/a	The parser will not be ready to parse another document until
151	n/a	the reset method has been called.
152	n/a
153	n/a	close may raise SAXException."""
154	n/a	raise NotImplementedError("This method must be implemented!")
155	n/a
156	n/a	def reset(self):
157	n/a	"""This method is called after close has been called to reset
158	n/a	the parser so that it is ready to parse new documents. The
159	n/a	results of calling parse or feed after close without calling
160	n/a	reset are undefined."""
161	n/a	raise NotImplementedError("This method must be implemented!")
162	n/a
163	n/a	# ===== LOCATOR =====
164	n/a
165	n/a	class Locator:
166	n/a	"""Interface for associating a SAX event with a document
167	n/a	location. A locator object will return valid results only during
168	n/a	calls to DocumentHandler methods; at any other time, the
169	n/a	results are unpredictable."""
170	n/a
171	n/a	def getColumnNumber(self):
172	n/a	"Return the column number where the current event ends."
173	n/a	return -1
174	n/a
175	n/a	def getLineNumber(self):
176	n/a	"Return the line number where the current event ends."
177	n/a	return -1
178	n/a
179	n/a	def getPublicId(self):
180	n/a	"Return the public identifier for the current event."
181	n/a	return None
182	n/a
183	n/a	def getSystemId(self):
184	n/a	"Return the system identifier for the current event."
185	n/a	return None
186	n/a
187	n/a	# ===== INPUTSOURCE =====
188	n/a
189	n/a	class InputSource:
190	n/a	"""Encapsulation of the information needed by the XMLReader to
191	n/a	read entities.
192	n/a
193	n/a	This class may include information about the public identifier,
194	n/a	system identifier, byte stream (possibly with character encoding
195	n/a	information) and/or the character stream of an entity.
196	n/a
197	n/a	Applications will create objects of this class for use in the
198	n/a	XMLReader.parse method and for returning from
199	n/a	EntityResolver.resolveEntity.
200	n/a
201	n/a	An InputSource belongs to the application, the XMLReader is not
202	n/a	allowed to modify InputSource objects passed to it from the
203	n/a	application, although it may make copies and modify those."""
204	n/a
205	n/a	def __init__(self, system_id = None):
206	n/a	self.__system_id = system_id
207	n/a	self.__public_id = None
208	n/a	self.__encoding = None
209	n/a	self.__bytefile = None
210	n/a	self.__charfile = None
211	n/a
212	n/a	def setPublicId(self, public_id):
213	n/a	"Sets the public identifier of this InputSource."
214	n/a	self.__public_id = public_id
215	n/a
216	n/a	def getPublicId(self):
217	n/a	"Returns the public identifier of this InputSource."
218	n/a	return self.__public_id
219	n/a
220	n/a	def setSystemId(self, system_id):
221	n/a	"Sets the system identifier of this InputSource."
222	n/a	self.__system_id = system_id
223	n/a
224	n/a	def getSystemId(self):
225	n/a	"Returns the system identifier of this InputSource."
226	n/a	return self.__system_id
227	n/a
228	n/a	def setEncoding(self, encoding):
229	n/a	"""Sets the character encoding of this InputSource.
230	n/a
231	n/a	The encoding must be a string acceptable for an XML encoding
232	n/a	declaration (see section 4.3.3 of the XML recommendation).
233	n/a
234	n/a	The encoding attribute of the InputSource is ignored if the
235	n/a	InputSource also contains a character stream."""
236	n/a	self.__encoding = encoding
237	n/a
238	n/a	def getEncoding(self):
239	n/a	"Get the character encoding of this InputSource."
240	n/a	return self.__encoding
241	n/a
242	n/a	def setByteStream(self, bytefile):
243	n/a	"""Set the byte stream (a Python file-like object which does
244	n/a	not perform byte-to-character conversion) for this input
245	n/a	source.
246	n/a
247	n/a	The SAX parser will ignore this if there is also a character
248	n/a	stream specified, but it will use a byte stream in preference
249	n/a	to opening a URI connection itself.
250	n/a
251	n/a	If the application knows the character encoding of the byte
252	n/a	stream, it should set it with the setEncoding method."""
253	n/a	self.__bytefile = bytefile
254	n/a
255	n/a	def getByteStream(self):
256	n/a	"""Get the byte stream for this input source.
257	n/a
258	n/a	The getEncoding method will return the character encoding for
259	n/a	this byte stream, or None if unknown."""
260	n/a	return self.__bytefile
261	n/a
262	n/a	def setCharacterStream(self, charfile):
263	n/a	"""Set the character stream for this input source. (The stream
264	n/a	must be a Python 2.0 Unicode-wrapped file-like that performs
265	n/a	conversion to Unicode strings.)
266	n/a
267	n/a	If there is a character stream specified, the SAX parser will
268	n/a	ignore any byte stream and will not attempt to open a URI
269	n/a	connection to the system identifier."""
270	n/a	self.__charfile = charfile
271	n/a
272	n/a	def getCharacterStream(self):
273	n/a	"Get the character stream for this input source."
274	n/a	return self.__charfile
275	n/a
276	n/a	# ===== ATTRIBUTESIMPL =====
277	n/a
278	n/a	class AttributesImpl:
279	n/a
280	n/a	def __init__(self, attrs):
281	n/a	"""Non-NS-aware implementation.
282	n/a
283	n/a	attrs should be of the form {name : value}."""
284	n/a	self._attrs = attrs
285	n/a
286	n/a	def getLength(self):
287	n/a	return len(self._attrs)
288	n/a
289	n/a	def getType(self, name):
290	n/a	return "CDATA"
291	n/a
292	n/a	def getValue(self, name):
293	n/a	return self._attrs[name]
294	n/a
295	n/a	def getValueByQName(self, name):
296	n/a	return self._attrs[name]
297	n/a
298	n/a	def getNameByQName(self, name):
299	n/a	if name not in self._attrs:
300	n/a	raise KeyError(name)
301	n/a	return name
302	n/a
303	n/a	def getQNameByName(self, name):
304	n/a	if name not in self._attrs:
305	n/a	raise KeyError(name)
306	n/a	return name
307	n/a
308	n/a	def getNames(self):
309	n/a	return list(self._attrs.keys())
310	n/a
311	n/a	def getQNames(self):
312	n/a	return list(self._attrs.keys())
313	n/a
314	n/a	def __len__(self):
315	n/a	return len(self._attrs)
316	n/a
317	n/a	def __getitem__(self, name):
318	n/a	return self._attrs[name]
319	n/a
320	n/a	def keys(self):
321	n/a	return list(self._attrs.keys())
322	n/a
323	n/a	def __contains__(self, name):
324	n/a	return name in self._attrs
325	n/a
326	n/a	def get(self, name, alternative=None):
327	n/a	return self._attrs.get(name, alternative)
328	n/a
329	n/a	def copy(self):
330	n/a	return self.__class__(self._attrs)
331	n/a
332	n/a	def items(self):
333	n/a	return list(self._attrs.items())
334	n/a
335	n/a	def values(self):
336	n/a	return list(self._attrs.values())
337	n/a
338	n/a	# ===== ATTRIBUTESNSIMPL =====
339	n/a
340	n/a	class AttributesNSImpl(AttributesImpl):
341	n/a
342	n/a	def __init__(self, attrs, qnames):
343	n/a	"""NS-aware implementation.
344	n/a
345	n/a	attrs should be of the form {(ns_uri, lname): value, ...}.
346	n/a	qnames of the form {(ns_uri, lname): qname, ...}."""
347	n/a	self._attrs = attrs
348	n/a	self._qnames = qnames
349	n/a
350	n/a	def getValueByQName(self, name):
351	n/a	for (nsname, qname) in self._qnames.items():
352	n/a	if qname == name:
353	n/a	return self._attrs[nsname]
354	n/a
355	n/a	raise KeyError(name)
356	n/a
357	n/a	def getNameByQName(self, name):
358	n/a	for (nsname, qname) in self._qnames.items():
359	n/a	if qname == name:
360	n/a	return nsname
361	n/a
362	n/a	raise KeyError(name)
363	n/a
364	n/a	def getQNameByName(self, name):
365	n/a	return self._qnames[name]
366	n/a
367	n/a	def getQNames(self):
368	n/a	return list(self._qnames.values())
369	n/a
370	n/a	def copy(self):
371	n/a	return self.__class__(self._attrs, self._qnames)
372	n/a
373	n/a
374	n/a	def _test():
375	n/a	XMLReader()
376	n/a	IncrementalParser()
377	n/a	Locator()
378	n/a
379	n/a	if __name__ == "__main__":
380	n/a	_test()