1
# XXX TypeErrors on calling handlers, or on bad return values from a
2
# handler, are obscure and unhelpful.
9
from xml.parsers import expat
11
from test.support import sortdict, run_unittest
14
class SetAttributeTest(unittest.TestCase):
16
self.parser = expat.ParserCreate(namespace_separator='!')
17
self.set_get_pairs = [
24
def test_ordered_attributes(self):
25
for x, y in self.set_get_pairs:
26
self.parser.ordered_attributes = x
27
self.assertEquals(self.parser.ordered_attributes, y)
29
def test_specified_attributes(self):
30
for x, y in self.set_get_pairs:
31
self.parser.specified_attributes = x
32
self.assertEquals(self.parser.specified_attributes, y)
36
<?xml version="1.0" encoding="iso-8859-1" standalone="no"?>
37
<?xml-stylesheet href="stylesheet.css"?>
39
<!DOCTYPE quotations SYSTEM "quotations.dtd" [
41
<!NOTATION notation SYSTEM "notation.jpeg">
42
<!ENTITY acirc "â">
43
<!ENTITY external_entity SYSTEM "entity.file">
44
<!ENTITY unparsed_entity SYSTEM "entity.file" NDATA notation>
48
<root attr1="value1" attr2="value2ὀ">
49
<myns:subelement xmlns:myns="http://www.python.org/namespace">
50
Contents of subelements
52
<sub2><![CDATA[contents of CDATA section]]></sub2>
58
# Produce UTF-8 output
59
class ParseTest(unittest.TestCase):
64
def StartElementHandler(self, name, attrs):
65
self.out.append('Start element: ' + repr(name) + ' ' +
68
def EndElementHandler(self, name):
69
self.out.append('End element: ' + repr(name))
71
def CharacterDataHandler(self, data):
74
self.out.append('Character data: ' + repr(data))
76
def ProcessingInstructionHandler(self, target, data):
77
self.out.append('PI: ' + repr(target) + ' ' + repr(data))
79
def StartNamespaceDeclHandler(self, prefix, uri):
80
self.out.append('NS decl: ' + repr(prefix) + ' ' + repr(uri))
82
def EndNamespaceDeclHandler(self, prefix):
83
self.out.append('End of NS decl: ' + repr(prefix))
85
def StartCdataSectionHandler(self):
86
self.out.append('Start of CDATA section')
88
def EndCdataSectionHandler(self):
89
self.out.append('End of CDATA section')
91
def CommentHandler(self, text):
92
self.out.append('Comment: ' + repr(text))
94
def NotationDeclHandler(self, *args):
95
name, base, sysid, pubid = args
96
self.out.append('Notation declared: %s' %(args,))
98
def UnparsedEntityDeclHandler(self, *args):
99
entityName, base, systemId, publicId, notationName = args
100
self.out.append('Unparsed entity decl: %s' %(args,))
102
def NotStandaloneHandler(self, userData):
103
self.out.append('Not standalone')
106
def ExternalEntityRefHandler(self, *args):
107
context, base, sysId, pubId = args
108
self.out.append('External entity ref: %s' %(args[1:],))
111
def DefaultHandler(self, userData):
114
def DefaultHandlerExpand(self, userData):
118
'StartElementHandler', 'EndElementHandler',
119
'CharacterDataHandler', 'ProcessingInstructionHandler',
120
'UnparsedEntityDeclHandler', 'NotationDeclHandler',
121
'StartNamespaceDeclHandler', 'EndNamespaceDeclHandler',
122
'CommentHandler', 'StartCdataSectionHandler',
123
'EndCdataSectionHandler',
124
'DefaultHandler', 'DefaultHandlerExpand',
125
#'NotStandaloneHandler',
126
'ExternalEntityRefHandler'
129
def _verify_parse_output(self, op):
130
self.assertEquals(op[0], 'PI: \'xml-stylesheet\' \'href="stylesheet.css"\'')
131
self.assertEquals(op[1], "Comment: ' comment data '")
132
self.assertEquals(op[2], "Notation declared: ('notation', None, 'notation.jpeg', None)")
133
self.assertEquals(op[3], "Unparsed entity decl: ('unparsed_entity', None, 'entity.file', None, 'notation')")
134
self.assertEquals(op[4], "Start element: 'root' {'attr1': 'value1', 'attr2': 'value2\u1f40'}")
135
self.assertEquals(op[5], "NS decl: 'myns' 'http://www.python.org/namespace'")
136
self.assertEquals(op[6], "Start element: 'http://www.python.org/namespace!subelement' {}")
137
self.assertEquals(op[7], "Character data: 'Contents of subelements'")
138
self.assertEquals(op[8], "End element: 'http://www.python.org/namespace!subelement'")
139
self.assertEquals(op[9], "End of NS decl: 'myns'")
140
self.assertEquals(op[10], "Start element: 'sub2' {}")
141
self.assertEquals(op[11], 'Start of CDATA section')
142
self.assertEquals(op[12], "Character data: 'contents of CDATA section'")
143
self.assertEquals(op[13], 'End of CDATA section')
144
self.assertEquals(op[14], "End element: 'sub2'")
145
self.assertEquals(op[15], "External entity ref: (None, 'entity.file', None)")
146
self.assertEquals(op[16], "End element: 'root'")
149
def test_unicode(self):
150
# Try the parse again, this time producing Unicode output
151
out = self.Outputter()
152
parser = expat.ParserCreate(namespace_separator='!')
153
for name in self.handler_names:
154
setattr(parser, name, getattr(out, name))
156
parser.Parse(data, 1)
159
self._verify_parse_output(op)
161
def test_parse_file(self):
163
out = self.Outputter()
164
parser = expat.ParserCreate(namespace_separator='!')
165
for name in self.handler_names:
166
setattr(parser, name, getattr(out, name))
169
parser.ParseFile(file)
172
self._verify_parse_output(op)
174
class NamespaceSeparatorTest(unittest.TestCase):
175
def test_legal(self):
176
# Tests that make sure we get errors when the namespace_separator value
177
# is illegal, and that we don't for good values:
179
expat.ParserCreate(namespace_separator=None)
180
expat.ParserCreate(namespace_separator=' ')
182
def test_illegal(self):
184
expat.ParserCreate(namespace_separator=42)
186
except TypeError as e:
187
self.assertEquals(str(e),
188
'ParserCreate() argument 2 must be string or None, not int')
191
expat.ParserCreate(namespace_separator='too long')
193
except ValueError as e:
194
self.assertEquals(str(e),
195
'namespace_separator must be at most one character, omitted, or None')
197
def test_zero_length(self):
198
# ParserCreate() needs to accept a namespace_separator of zero length
199
# to satisfy the requirements of RDF applications that are required
200
# to simply glue together the namespace URI and the localname. Though
201
# considered a wart of the RDF specifications, it needs to be supported.
203
# See XML-SIG mailing list thread starting with
204
# http://mail.python.org/pipermail/xml-sig/2001-April/005202.html
206
expat.ParserCreate(namespace_separator='') # too short
209
class InterningTest(unittest.TestCase):
211
# Test the interning machinery.
212
p = expat.ParserCreate()
214
def collector(name, *args):
216
p.StartElementHandler = collector
217
p.EndElementHandler = collector
218
p.Parse("<e> <e/> <e></e> </e>", 1)
220
self.assertEquals(len(L), 6)
222
# L should have the same string repeated over and over.
223
self.assertTrue(tag is entry)
226
class BufferTextTest(unittest.TestCase):
229
self.parser = expat.ParserCreate()
230
self.parser.buffer_text = 1
231
self.parser.CharacterDataHandler = self.CharacterDataHandler
233
def check(self, expected, label):
234
self.assertEquals(self.stuff, expected,
235
"%s\nstuff = %r\nexpected = %r"
236
% (label, self.stuff, map(str, expected)))
238
def CharacterDataHandler(self, text):
239
self.stuff.append(text)
241
def StartElementHandler(self, name, attrs):
242
self.stuff.append("<%s>" % name)
243
bt = attrs.get("buffer-text")
245
self.parser.buffer_text = 1
247
self.parser.buffer_text = 0
249
def EndElementHandler(self, name):
250
self.stuff.append("</%s>" % name)
252
def CommentHandler(self, data):
253
self.stuff.append("<!--%s-->" % data)
255
def setHandlers(self, handlers=[]):
256
for name in handlers:
257
setattr(self.parser, name, getattr(self, name))
259
def test_default_to_disabled(self):
260
parser = expat.ParserCreate()
261
self.assertFalse(parser.buffer_text)
263
def test_buffering_enabled(self):
264
# Make sure buffering is turned on
265
self.assertTrue(self.parser.buffer_text)
266
self.parser.Parse("<a>1<b/>2<c/>3</a>", 1)
267
self.assertEquals(self.stuff, ['123'],
268
"buffered text not properly collapsed")
271
# XXX This test exposes more detail of Expat's text chunking than we
272
# XXX like, but it tests what we need to concisely.
273
self.setHandlers(["StartElementHandler"])
274
self.parser.Parse("<a>1<b buffer-text='no'/>2\n3<c buffer-text='yes'/>4\n5</a>", 1)
275
self.assertEquals(self.stuff,
276
["<a>", "1", "<b>", "2", "\n", "3", "<c>", "4\n5"],
277
"buffering control not reacting as expected")
280
self.parser.Parse("<a>1<b/><2><c/> \n 3</a>", 1)
281
self.assertEquals(self.stuff, ["1<2> \n 3"],
282
"buffered text not properly collapsed")
285
self.setHandlers(["StartElementHandler"])
286
self.parser.Parse("<a>1<b/>2<c/>3</a>", 1)
287
self.assertEquals(self.stuff, ["<a>", "1", "<b>", "2", "<c>", "3"],
288
"buffered text not properly split")
291
self.setHandlers(["StartElementHandler", "EndElementHandler"])
292
self.parser.CharacterDataHandler = None
293
self.parser.Parse("<a>1<b/>2<c/>3</a>", 1)
294
self.assertEquals(self.stuff,
295
["<a>", "<b>", "</b>", "<c>", "</c>", "</a>"])
298
self.setHandlers(["StartElementHandler", "EndElementHandler"])
299
self.parser.Parse("<a>1<b></b>2<c/>3</a>", 1)
300
self.assertEquals(self.stuff,
301
["<a>", "1", "<b>", "</b>", "2", "<c>", "</c>", "3", "</a>"])
304
self.setHandlers(["CommentHandler", "EndElementHandler",
305
"StartElementHandler"])
306
self.parser.Parse("<a>1<b/>2<c></c>345</a> ", 1)
307
self.assertEquals(self.stuff,
308
["<a>", "1", "<b>", "</b>", "2", "<c>", "</c>", "345", "</a>"],
309
"buffered text not properly split")
312
self.setHandlers(["CommentHandler", "EndElementHandler",
313
"StartElementHandler"])
314
self.parser.Parse("<a>1<b/>2<c></c>3<!--abc-->4<!--def-->5</a> ", 1)
315
self.assertEquals(self.stuff,
316
["<a>", "1", "<b>", "</b>", "2", "<c>", "</c>", "3",
317
"<!--abc-->", "4", "<!--def-->", "5", "</a>"],
318
"buffered text not properly split")
321
# Test handling of exception from callback:
322
class HandlerExceptionTest(unittest.TestCase):
323
def StartElementHandler(self, name, attrs):
324
raise RuntimeError(name)
327
parser = expat.ParserCreate()
328
parser.StartElementHandler = self.StartElementHandler
330
parser.Parse("<a><b><c/></b></a>", 1)
332
except RuntimeError as e:
333
self.assertEquals(e.args[0], 'a',
334
"Expected RuntimeError for element 'a', but" + \
335
" found %r" % e.args[0])
338
# Test Current* members:
339
class PositionTest(unittest.TestCase):
340
def StartElementHandler(self, name, attrs):
343
def EndElementHandler(self, name):
346
def check_pos(self, event):
348
self.parser.CurrentByteIndex,
349
self.parser.CurrentLineNumber,
350
self.parser.CurrentColumnNumber)
351
self.assertTrue(self.upto < len(self.expected_list),
352
'too many parser events')
353
expected = self.expected_list[self.upto]
354
self.assertEquals(pos, expected,
355
'Expected position %s, got position %s' %(pos, expected))
359
self.parser = expat.ParserCreate()
360
self.parser.StartElementHandler = self.StartElementHandler
361
self.parser.EndElementHandler = self.EndElementHandler
363
self.expected_list = [('s', 0, 1, 0), ('s', 5, 2, 1), ('s', 11, 3, 2),
364
('e', 15, 3, 6), ('e', 17, 4, 1), ('e', 22, 5, 0)]
366
xml = '<a>\n <b>\n <c/>\n </b>\n</a>'
367
self.parser.Parse(xml, 1)
370
class sf1296433Test(unittest.TestCase):
371
def test_parse_only_xml_data(self):
372
# http://python.org/sf/1296433
374
xml = "<?xml version='1.0' encoding='iso8859'?><s>%s</s>" % ('a' * 1025)
375
# this one doesn't crash
376
#xml = "<?xml version='1.0'?><s>%s</s>" % ('a' * 10000)
378
class SpecificException(Exception):
382
raise SpecificException
384
parser = expat.ParserCreate()
385
parser.CharacterDataHandler = handler
387
self.assertRaises(Exception, parser.Parse, xml)
389
class ChardataBufferTest(unittest.TestCase):
391
test setting of chardata buffer size
394
def test_1025_bytes(self):
395
self.assertEquals(self.small_buffer_test(1025), 2)
397
def test_1000_bytes(self):
398
self.assertEquals(self.small_buffer_test(1000), 1)
400
def test_wrong_size(self):
401
parser = expat.ParserCreate()
402
parser.buffer_text = 1
404
parser.buffer_size = size
406
self.assertRaises(ValueError, f, -1)
407
self.assertRaises(ValueError, f, 0)
409
def test_unchanged_size(self):
410
xml1 = ("<?xml version='1.0' encoding='iso8859'?><s>%s" % ('a' * 512))
411
xml2 = 'a'*512 + '</s>'
412
parser = expat.ParserCreate()
413
parser.CharacterDataHandler = self.counting_handler
414
parser.buffer_size = 512
415
parser.buffer_text = 1
417
# Feed 512 bytes of character data: the handler should be called
421
self.assertEquals(self.n, 1)
423
# Reassign to buffer_size, but assign the same size.
424
parser.buffer_size = parser.buffer_size
425
self.assertEquals(self.n, 1)
427
# Try parsing rest of the document
429
self.assertEquals(self.n, 2)
432
def test_disabling_buffer(self):
433
xml1 = "<?xml version='1.0' encoding='iso8859'?><a>%s" % ('a' * 512)
435
xml3 = "%s</a>" % ('c' * 1024)
436
parser = expat.ParserCreate()
437
parser.CharacterDataHandler = self.counting_handler
438
parser.buffer_text = 1
439
parser.buffer_size = 1024
440
self.assertEquals(parser.buffer_size, 1024)
442
# Parse one chunk of XML
444
parser.Parse(xml1, 0)
445
self.assertEquals(parser.buffer_size, 1024)
446
self.assertEquals(self.n, 1)
448
# Turn off buffering and parse the next chunk.
449
parser.buffer_text = 0
450
self.assertFalse(parser.buffer_text)
451
self.assertEquals(parser.buffer_size, 1024)
453
parser.Parse(xml2, 0)
454
self.assertEquals(self.n, 11)
456
parser.buffer_text = 1
457
self.assertTrue(parser.buffer_text)
458
self.assertEquals(parser.buffer_size, 1024)
459
parser.Parse(xml3, 1)
460
self.assertEquals(self.n, 12)
464
def make_document(self, bytes):
465
return ("<?xml version='1.0'?><tag>" + bytes * 'a' + '</tag>')
467
def counting_handler(self, text):
470
def small_buffer_test(self, buffer_len):
471
xml = "<?xml version='1.0' encoding='iso8859'?><s>%s</s>" % ('a' * buffer_len)
472
parser = expat.ParserCreate()
473
parser.CharacterDataHandler = self.counting_handler
474
parser.buffer_size = 1024
475
parser.buffer_text = 1
481
def test_change_size_1(self):
482
xml1 = "<?xml version='1.0' encoding='iso8859'?><a><s>%s" % ('a' * 1024)
483
xml2 = "aaa</s><s>%s</s></a>" % ('a' * 1025)
484
parser = expat.ParserCreate()
485
parser.CharacterDataHandler = self.counting_handler
486
parser.buffer_text = 1
487
parser.buffer_size = 1024
488
self.assertEquals(parser.buffer_size, 1024)
491
parser.Parse(xml1, 0)
492
parser.buffer_size *= 2
493
self.assertEquals(parser.buffer_size, 2048)
494
parser.Parse(xml2, 1)
495
self.assertEquals(self.n, 2)
497
def test_change_size_2(self):
498
xml1 = "<?xml version='1.0' encoding='iso8859'?><a>a<s>%s" % ('a' * 1023)
499
xml2 = "aaa</s><s>%s</s></a>" % ('a' * 1025)
500
parser = expat.ParserCreate()
501
parser.CharacterDataHandler = self.counting_handler
502
parser.buffer_text = 1
503
parser.buffer_size = 2048
504
self.assertEquals(parser.buffer_size, 2048)
507
parser.Parse(xml1, 0)
508
parser.buffer_size = parser.buffer_size // 2
509
self.assertEquals(parser.buffer_size, 1024)
510
parser.Parse(xml2, 1)
511
self.assertEquals(self.n, 4)
515
run_unittest(SetAttributeTest,
517
NamespaceSeparatorTest,
520
HandlerExceptionTest,
525
if __name__ == "__main__":