| Home | Trees | Indices | Help |
|
|---|
|
|
1 # -*- coding: utf-8 -*-
2
3 """
4 Tests specific to the extended etree API
5
6 Tests that apply to the general ElementTree API should go into
7 test_elementtree
8 """
9
10 import os.path
11 import unittest
12 import copy
13 import sys
14 import re
15 import gc
16 import operator
17 import tempfile
18 import zlib
19 import gzip
20
21 this_dir = os.path.dirname(__file__)
22 if this_dir not in sys.path:
23 sys.path.insert(0, this_dir) # needed for Py3
24
25 from common_imports import etree, StringIO, BytesIO, HelperTestCase, fileInTestDir, read_file
26 from common_imports import SillyFileLike, LargeFileLikeUnicode, doctest, make_doctest
27 from common_imports import canonicalize, sorted, _str, _bytes
28
29 print("")
30 print("TESTED VERSION: %s" % etree.__version__)
31 print(" Python: " + repr(sys.version_info))
32 print(" lxml.etree: " + repr(etree.LXML_VERSION))
33 print(" libxml used: " + repr(etree.LIBXML_VERSION))
34 print(" libxml compiled: " + repr(etree.LIBXML_COMPILED_VERSION))
35 print(" libxslt used: " + repr(etree.LIBXSLT_VERSION))
36 print(" libxslt compiled: " + repr(etree.LIBXSLT_COMPILED_VERSION))
37 print("")
38
39 try:
40 _unicode = unicode
41 except NameError:
42 # Python 3
43 _unicode = str
44
46 """Tests only for etree, not ElementTree"""
47 etree = etree
48
50 self.assertTrue(isinstance(etree.__version__, _unicode))
51 self.assertTrue(isinstance(etree.LXML_VERSION, tuple))
52 self.assertEqual(len(etree.LXML_VERSION), 4)
53 self.assertTrue(isinstance(etree.LXML_VERSION[0], int))
54 self.assertTrue(isinstance(etree.LXML_VERSION[1], int))
55 self.assertTrue(isinstance(etree.LXML_VERSION[2], int))
56 self.assertTrue(isinstance(etree.LXML_VERSION[3], int))
57 self.assertTrue(etree.__version__.startswith(
58 str(etree.LXML_VERSION[0])))
59
61 if hasattr(self.etree, '__pyx_capi__'):
62 # newer Pyrex compatible C-API
63 self.assertTrue(isinstance(self.etree.__pyx_capi__, dict))
64 self.assertTrue(len(self.etree.__pyx_capi__) > 0)
65 else:
66 # older C-API mechanism
67 self.assertTrue(hasattr(self.etree, '_import_c_api'))
68
70 Element = self.etree.Element
71 el = Element('name')
72 self.assertEqual(el.tag, 'name')
73 el = Element('{}name')
74 self.assertEqual(el.tag, 'name')
75
77 Element = self.etree.Element
78 el = Element('name')
79 self.assertRaises(ValueError, Element, '{}')
80 self.assertRaises(ValueError, setattr, el, 'tag', '{}')
81
82 self.assertRaises(ValueError, Element, '{test}')
83 self.assertRaises(ValueError, setattr, el, 'tag', '{test}')
84
86 Element = self.etree.Element
87 self.assertRaises(ValueError, Element, 'p:name')
88 self.assertRaises(ValueError, Element, '{test}p:name')
89
90 el = Element('name')
91 self.assertRaises(ValueError, setattr, el, 'tag', 'p:name')
92
94 Element = self.etree.Element
95 self.assertRaises(ValueError, Element, "p'name")
96 self.assertRaises(ValueError, Element, 'p"name')
97
98 self.assertRaises(ValueError, Element, "{test}p'name")
99 self.assertRaises(ValueError, Element, '{test}p"name')
100
101 el = Element('name')
102 self.assertRaises(ValueError, setattr, el, 'tag', "p'name")
103 self.assertRaises(ValueError, setattr, el, 'tag', 'p"name')
104
106 Element = self.etree.Element
107 self.assertRaises(ValueError, Element, ' name ')
108 self.assertRaises(ValueError, Element, 'na me')
109 self.assertRaises(ValueError, Element, '{test} name')
110
111 el = Element('name')
112 self.assertRaises(ValueError, setattr, el, 'tag', ' name ')
113
115 Element = self.etree.Element
116 SubElement = self.etree.SubElement
117
118 el = Element('name')
119 self.assertRaises(ValueError, SubElement, el, '{}')
120 self.assertRaises(ValueError, SubElement, el, '{test}')
121
123 Element = self.etree.Element
124 SubElement = self.etree.SubElement
125
126 el = Element('name')
127 self.assertRaises(ValueError, SubElement, el, 'p:name')
128 self.assertRaises(ValueError, SubElement, el, '{test}p:name')
129
131 Element = self.etree.Element
132 SubElement = self.etree.SubElement
133
134 el = Element('name')
135 self.assertRaises(ValueError, SubElement, el, "p'name")
136 self.assertRaises(ValueError, SubElement, el, "{test}p'name")
137
138 self.assertRaises(ValueError, SubElement, el, 'p"name')
139 self.assertRaises(ValueError, SubElement, el, '{test}p"name')
140
142 Element = self.etree.Element
143 SubElement = self.etree.SubElement
144
145 el = Element('name')
146 self.assertRaises(ValueError, SubElement, el, ' name ')
147 self.assertRaises(ValueError, SubElement, el, 'na me')
148 self.assertRaises(ValueError, SubElement, el, '{test} name')
149
151 Element = self.etree.Element
152 SubElement = self.etree.SubElement
153
154 el = Element('name')
155 self.assertRaises(ValueError, SubElement, el, 'name', {'a b c' : 'abc'})
156 self.assertRaises(ValueError, SubElement, el, 'name', {'a' : 'a\0\n'})
157 self.assertEqual(0, len(el))
158
160 QName = self.etree.QName
161 self.assertRaises(ValueError, QName, '')
162 self.assertRaises(ValueError, QName, 'test', '')
163
165 QName = self.etree.QName
166 self.assertRaises(ValueError, QName, 'p:name')
167 self.assertRaises(ValueError, QName, 'test', 'p:name')
168
170 QName = self.etree.QName
171 self.assertRaises(ValueError, QName, ' name ')
172 self.assertRaises(ValueError, QName, 'na me')
173 self.assertRaises(ValueError, QName, 'test', ' name')
174
176 # ET doesn't have namespace/localname properties on QNames
177 QName = self.etree.QName
178 namespace, localname = 'http://myns', 'a'
179 qname = QName(namespace, localname)
180 self.assertEqual(namespace, qname.namespace)
181 self.assertEqual(localname, qname.localname)
182
184 # ET doesn't have namespace/localname properties on QNames
185 QName = self.etree.QName
186 qname1 = QName('http://myns', 'a')
187 a = self.etree.Element(qname1, nsmap={'p' : 'http://myns'})
188
189 qname2 = QName(a)
190 self.assertEqual(a.tag, qname1.text)
191 self.assertEqual(qname1.text, qname2.text)
192 self.assertEqual(qname1, qname2)
193
195 # ET doesn't resove QNames as text values
196 etree = self.etree
197 qname = etree.QName('http://myns', 'a')
198 a = etree.Element(qname, nsmap={'p' : 'http://myns'})
199 a.text = qname
200
201 self.assertEqual("p:a", a.text)
202
204 etree = self.etree
205 self.assertRaises(ValueError,
206 etree.Element, "root", nsmap={'"' : 'testns'})
207 self.assertRaises(ValueError,
208 etree.Element, "root", nsmap={'&' : 'testns'})
209 self.assertRaises(ValueError,
210 etree.Element, "root", nsmap={'a:b' : 'testns'})
211
213 # ET in Py 3.x has no "attrib.has_key()" method
214 XML = self.etree.XML
215
216 root = XML(_bytes('<foo bar="Bar" xmlns:ns="http://ns.codespeak.net/test" ns:baz="Baz" />'))
217 self.assertEqual(
218 True, root.attrib.has_key('bar'))
219 self.assertEqual(
220 False, root.attrib.has_key('baz'))
221 self.assertEqual(
222 False, root.attrib.has_key('hah'))
223 self.assertEqual(
224 True,
225 root.attrib.has_key('{http://ns.codespeak.net/test}baz'))
226
228 Element = self.etree.Element
229 root = Element("root")
230 root.set("attr", "TEST")
231 self.assertEqual("TEST", root.get("attr"))
232
234 # ElementTree accepts arbitrary attribute values
235 # lxml.etree allows only strings
236 Element = self.etree.Element
237 root = Element("root")
238 self.assertRaises(TypeError, root.set, "newattr", 5)
239 self.assertRaises(TypeError, root.set, "newattr", None)
240
242 XML = self.etree.XML
243 xml = _bytes('<test a="5" b="10" c="20"><x a="4" b="2"/></test>')
244
245 root = XML(xml)
246 self.etree.strip_attributes(root, 'a')
247 self.assertEqual(_bytes('<test b="10" c="20"><x b="2"></x></test>'),
248 self._writeElement(root))
249
250 root = XML(xml)
251 self.etree.strip_attributes(root, 'b', 'c')
252 self.assertEqual(_bytes('<test a="5"><x a="4"></x></test>'),
253 self._writeElement(root))
254
256 XML = self.etree.XML
257 xml = _bytes('<test xmlns:n="http://test/ns" a="6" b="10" c="20" n:a="5"><x a="4" n:b="2"/></test>')
258
259 root = XML(xml)
260 self.etree.strip_attributes(root, 'a')
261 self.assertEqual(
262 _bytes('<test xmlns:n="http://test/ns" b="10" c="20" n:a="5"><x n:b="2"></x></test>'),
263 self._writeElement(root))
264
265 root = XML(xml)
266 self.etree.strip_attributes(root, '{http://test/ns}a', 'c')
267 self.assertEqual(
268 _bytes('<test xmlns:n="http://test/ns" a="6" b="10"><x a="4" n:b="2"></x></test>'),
269 self._writeElement(root))
270
271 root = XML(xml)
272 self.etree.strip_attributes(root, '{http://test/ns}*')
273 self.assertEqual(
274 _bytes('<test xmlns:n="http://test/ns" a="6" b="10" c="20"><x a="4"></x></test>'),
275 self._writeElement(root))
276
278 XML = self.etree.XML
279 xml = _bytes('<test><a><b><c/></b></a><x><a><b/><c/></a></x></test>')
280
281 root = XML(xml)
282 self.etree.strip_elements(root, 'a')
283 self.assertEqual(_bytes('<test><x></x></test>'),
284 self._writeElement(root))
285
286 root = XML(xml)
287 self.etree.strip_elements(root, 'b', 'c', 'X', 'Y', 'Z')
288 self.assertEqual(_bytes('<test><a></a><x><a></a></x></test>'),
289 self._writeElement(root))
290
291 root = XML(xml)
292 self.etree.strip_elements(root, 'c')
293 self.assertEqual(_bytes('<test><a><b></b></a><x><a><b></b></a></x></test>'),
294 self._writeElement(root))
295
297 XML = self.etree.XML
298 xml = _bytes('<test>TEST<n:a xmlns:n="urn:a">A<b>B<c xmlns="urn:c"/>C</b>BT</n:a>AT<x>X<a>A<b xmlns="urn:a"/>BT<c xmlns="urn:x"/>CT</a>AT</x>XT</test>')
299
300 root = XML(xml)
301 self.etree.strip_elements(root, 'a')
302 self.assertEqual(_bytes('<test>TEST<n:a xmlns:n="urn:a">A<b>B<c xmlns="urn:c"></c>C</b>BT</n:a>AT<x>X</x>XT</test>'),
303 self._writeElement(root))
304
305 root = XML(xml)
306 self.etree.strip_elements(root, '{urn:a}b', 'c')
307 self.assertEqual(_bytes('<test>TEST<n:a xmlns:n="urn:a">A<b>B<c xmlns="urn:c"></c>C</b>BT</n:a>AT<x>X<a>A<c xmlns="urn:x"></c>CT</a>AT</x>XT</test>'),
308 self._writeElement(root))
309
310 root = XML(xml)
311 self.etree.strip_elements(root, '{urn:a}*', 'c')
312 self.assertEqual(_bytes('<test>TEST<x>X<a>A<c xmlns="urn:x"></c>CT</a>AT</x>XT</test>'),
313 self._writeElement(root))
314
315 root = XML(xml)
316 self.etree.strip_elements(root, '{urn:a}*', 'c', with_tail=False)
317 self.assertEqual(_bytes('<test>TESTAT<x>X<a>ABT<c xmlns="urn:x"></c>CT</a>AT</x>XT</test>'),
318 self._writeElement(root))
319
338
364
391
418
437
450
452 # lxml.etree separates target and text
453 Element = self.etree.Element
454 SubElement = self.etree.SubElement
455 ProcessingInstruction = self.etree.ProcessingInstruction
456
457 a = Element('a')
458 a.append(ProcessingInstruction('foo', 'some more text'))
459 self.assertEqual(a[0].target, 'foo')
460 self.assertEqual(a[0].text, 'some more text')
461
463 XML = self.etree.XML
464 root = XML(_bytes("<test><?mypi my test ?></test>"))
465 self.assertEqual(root[0].target, "mypi")
466 self.assertEqual(root[0].text, "my test ")
467
469 XML = self.etree.XML
470 root = XML(_bytes("<test><?mypi my='1' test=\" abc \" quotes=\"' '\" only names ?></test>"))
471 self.assertEqual(root[0].target, "mypi")
472 self.assertEqual(root[0].get('my'), "1")
473 self.assertEqual(root[0].get('test'), " abc ")
474 self.assertEqual(root[0].get('quotes'), "' '")
475 self.assertEqual(root[0].get('only'), None)
476 self.assertEqual(root[0].get('names'), None)
477 self.assertEqual(root[0].get('nope'), None)
478
480 XML = self.etree.XML
481 root = XML(_bytes("<test><?mypi my='1' test=\" abc \" quotes=\"' '\" only names ?></test>"))
482 self.assertEqual(root[0].target, "mypi")
483 self.assertEqual(root[0].attrib['my'], "1")
484 self.assertEqual(root[0].attrib['test'], " abc ")
485 self.assertEqual(root[0].attrib['quotes'], "' '")
486 self.assertRaises(KeyError, root[0].attrib.__getitem__, 'only')
487 self.assertRaises(KeyError, root[0].attrib.__getitem__, 'names')
488 self.assertRaises(KeyError, root[0].attrib.__getitem__, 'nope')
489
491 # previously caused a crash
492 ProcessingInstruction = self.etree.ProcessingInstruction
493
494 a = ProcessingInstruction("PI", "ONE")
495 b = copy.deepcopy(a)
496 b.text = "ANOTHER"
497
498 self.assertEqual('ONE', a.text)
499 self.assertEqual('ANOTHER', b.text)
500
502 XML = self.etree.XML
503 tostring = self.etree.tostring
504 root = XML(_bytes("<?mypi my test ?><test/><!--comment -->"))
505 tree1 = self.etree.ElementTree(root)
506 self.assertEqual(_bytes("<?mypi my test ?><test/><!--comment -->"),
507 tostring(tree1))
508
509 tree2 = copy.deepcopy(tree1)
510 self.assertEqual(_bytes("<?mypi my test ?><test/><!--comment -->"),
511 tostring(tree2))
512
513 root2 = copy.deepcopy(tree1.getroot())
514 self.assertEqual(_bytes("<test/>"),
515 tostring(root2))
516
518 XML = self.etree.XML
519 tostring = self.etree.tostring
520 xml = _bytes('<!DOCTYPE test [\n<!ENTITY entity "tasty">\n]>\n<test/>')
521 root = XML(xml)
522 tree1 = self.etree.ElementTree(root)
523 self.assertEqual(xml, tostring(tree1))
524
525 tree2 = copy.deepcopy(tree1)
526 self.assertEqual(xml, tostring(tree2))
527
528 root2 = copy.deepcopy(tree1.getroot())
529 self.assertEqual(_bytes("<test/>"),
530 tostring(root2))
531
533 # ElementTree accepts arbitrary attribute values
534 # lxml.etree allows only strings
535 Element = self.etree.Element
536
537 root = Element("root")
538 root.set("attr", "TEST")
539 self.assertEqual("TEST", root.get("attr"))
540 self.assertRaises(TypeError, root.set, "newattr", 5)
541
543 fromstring = self.etree.fromstring
544 tostring = self.etree.tostring
545 XMLParser = self.etree.XMLParser
546
547 xml = _bytes('<a><!--A--><b><!-- B --><c/></b><!--C--></a>')
548 parser = XMLParser(remove_comments=True)
549 root = fromstring(xml, parser)
550 self.assertEqual(
551 _bytes('<a><b><c/></b></a>'),
552 tostring(root))
553
555 parse = self.etree.parse
556 tostring = self.etree.tostring
557 XMLParser = self.etree.XMLParser
558
559 xml = _bytes('<?test?><a><?A?><b><?B?><c/></b><?C?></a><?tail?>')
560
561 f = BytesIO(xml)
562 tree = parse(f)
563 self.assertEqual(
564 xml,
565 tostring(tree))
566
567 parser = XMLParser(remove_pis=True)
568 tree = parse(f, parser)
569 self.assertEqual(
570 _bytes('<a><b><c/></b></a>'),
571 tostring(tree))
572
574 # ET raises IOError only
575 parse = self.etree.parse
576 self.assertRaises(TypeError, parse, 'notthere.xml', object())
577
579 # ET removes comments
580 iterparse = self.etree.iterparse
581 tostring = self.etree.tostring
582
583 f = BytesIO('<a><!--A--><b><!-- B --><c/></b><!--C--></a>')
584 events = list(iterparse(f))
585 root = events[-1][1]
586 self.assertEqual(3, len(events))
587 self.assertEqual(
588 _bytes('<a><!--A--><b><!-- B --><c/></b><!--C--></a>'),
589 tostring(root))
590
592 # ET removes comments
593 iterparse = self.etree.iterparse
594 tostring = self.etree.tostring
595
596 def name(event, el):
597 if event == 'comment':
598 return el.text
599 else:
600 return el.tag
601
602 f = BytesIO('<a><!--A--><b><!-- B --><c/></b><!--C--></a>')
603 events = list(iterparse(f, events=('end', 'comment')))
604 root = events[-1][1]
605 self.assertEqual(6, len(events))
606 self.assertEqual(['A', ' B ', 'c', 'b', 'C', 'a'],
607 [ name(*item) for item in events ])
608 self.assertEqual(
609 _bytes('<a><!--A--><b><!-- B --><c/></b><!--C--></a>'),
610 tostring(root))
611
613 # ET removes pis
614 iterparse = self.etree.iterparse
615 tostring = self.etree.tostring
616 ElementTree = self.etree.ElementTree
617
618 def name(event, el):
619 if event == 'pi':
620 return (el.target, el.text)
621 else:
622 return el.tag
623
624 f = BytesIO('<?pia a?><a><?pib b?><b><?pic c?><c/></b><?pid d?></a><?pie e?>')
625 events = list(iterparse(f, events=('end', 'pi')))
626 root = events[-2][1]
627 self.assertEqual(8, len(events))
628 self.assertEqual([('pia','a'), ('pib','b'), ('pic','c'), 'c', 'b',
629 ('pid','d'), 'a', ('pie','e')],
630 [ name(*item) for item in events ])
631 self.assertEqual(
632 _bytes('<?pia a?><a><?pib b?><b><?pic c?><c/></b><?pid d?></a><?pie e?>'),
633 tostring(ElementTree(root)))
634
636 iterparse = self.etree.iterparse
637 tostring = self.etree.tostring
638
639 f = BytesIO('<a><!--A--><b><!-- B --><c/></b><!--C--></a>')
640 events = list(iterparse(f, remove_comments=True,
641 events=('end', 'comment')))
642 root = events[-1][1]
643 self.assertEqual(3, len(events))
644 self.assertEqual(['c', 'b', 'a'],
645 [ el.tag for (event, el) in events ])
646 self.assertEqual(
647 _bytes('<a><b><c/></b></a>'),
648 tostring(root))
649
651 iterparse = self.etree.iterparse
652 f = BytesIO('<a><b><c/></a>')
653 # ET raises ExpatError, lxml raises XMLSyntaxError
654 self.assertRaises(self.etree.XMLSyntaxError, list, iterparse(f))
655
657 iterparse = self.etree.iterparse
658 f = BytesIO("""
659 <a> \n \n <b> b test </b> \n
660
661 \n\t <c> \n </c> </a> \n """)
662 iterator = iterparse(f, remove_blank_text=True)
663 text = [ (element.text, element.tail)
664 for event, element in iterator ]
665 self.assertEqual(
666 [(" b test ", None), (" \n ", None), (None, None)],
667 text)
668
670 iterparse = self.etree.iterparse
671 f = BytesIO('<a><b><d/></b><c/></a>')
672
673 iterator = iterparse(f, tag="b", events=('start', 'end'))
674 events = list(iterator)
675 root = iterator.root
676 self.assertEqual(
677 [('start', root[0]), ('end', root[0])],
678 events)
679
681 iterparse = self.etree.iterparse
682 f = BytesIO('<a><b><d/></b><c/></a>')
683
684 iterator = iterparse(f, tag="*", events=('start', 'end'))
685 events = list(iterator)
686 self.assertEqual(
687 8,
688 len(events))
689
691 iterparse = self.etree.iterparse
692 f = BytesIO('<a xmlns="urn:test:1"><b><d/></b><c/></a>')
693
694 iterator = iterparse(f, tag="{urn:test:1}b", events=('start', 'end'))
695 events = list(iterator)
696 root = iterator.root
697 self.assertEqual(
698 [('start', root[0]), ('end', root[0])],
699 events)
700
702 iterparse = self.etree.iterparse
703 f = BytesIO('<a><b><d/></b><c/></a>')
704 iterator = iterparse(f, tag="{}b", events=('start', 'end'))
705 events = list(iterator)
706 root = iterator.root
707 self.assertEqual(
708 [('start', root[0]), ('end', root[0])],
709 events)
710
711 f = BytesIO('<a xmlns="urn:test:1"><b><d/></b><c/></a>')
712 iterator = iterparse(f, tag="{}b", events=('start', 'end'))
713 events = list(iterator)
714 root = iterator.root
715 self.assertEqual([], events)
716
718 iterparse = self.etree.iterparse
719 f = BytesIO('<a xmlns="urn:test:1"><b><d/></b><c/></a>')
720 iterator = iterparse(f, tag="{urn:test:1}*", events=('start', 'end'))
721 events = list(iterator)
722 self.assertEqual(8, len(events))
723
725 iterparse = self.etree.iterparse
726 f = BytesIO('<a xmlns="urn:test:1"><b><d/></b><c/></a>')
727 iterator = iterparse(f, tag="{}*", events=('start', 'end'))
728 events = list(iterator)
729 self.assertEqual([], events)
730
731 f = BytesIO('<a><b><d/></b><c/></a>')
732 iterator = iterparse(f, tag="{}*", events=('start', 'end'))
733 events = list(iterator)
734 self.assertEqual(8, len(events))
735
737 text = _str('Søk på nettet')
738 wrong_declaration = "<?xml version='1.0' encoding='UTF-8'?>"
739 xml_latin1 = (_str('%s<a>%s</a>') % (wrong_declaration, text)
740 ).encode('iso-8859-1')
741
742 self.assertRaises(self.etree.ParseError,
743 list, self.etree.iterparse(BytesIO(xml_latin1)))
744
746 text = _str('Søk på nettet', encoding="UTF-8")
747 wrong_declaration = "<?xml version='1.0' encoding='UTF-8'?>"
748 xml_latin1 = (_str('%s<a>%s</a>') % (wrong_declaration, text)
749 ).encode('iso-8859-1')
750
751 iterator = self.etree.iterparse(BytesIO(xml_latin1),
752 encoding="iso-8859-1")
753 self.assertEqual(1, len(list(iterator)))
754
755 a = iterator.root
756 self.assertEqual(a.text, text)
757
759 tostring = self.etree.tostring
760 f = BytesIO('<root><![CDATA[test]]></root>')
761 context = self.etree.iterparse(f, strip_cdata=False)
762 content = [ el.text for event,el in context ]
763
764 self.assertEqual(['test'], content)
765 self.assertEqual(_bytes('<root><![CDATA[test]]></root>'),
766 tostring(context.root))
767
771
773 self.etree.XMLParser(encoding="ascii")
774 self.etree.XMLParser(encoding="utf-8")
775 self.etree.XMLParser(encoding="iso-8859-1")
776
778 parser = self.etree.XMLParser(recover=True)
779
780 parser.feed('<?xml version=')
781 parser.feed('"1.0"?><ro')
782 parser.feed('ot><')
783 parser.feed('a test="works"')
784 parser.feed('><othertag/></root') # <a> not closed!
785 parser.feed('>')
786
787 root = parser.close()
788
789 self.assertEqual(root.tag, "root")
790 self.assertEqual(len(root), 1)
791 self.assertEqual(root[0].tag, "a")
792 self.assertEqual(root[0].get("test"), "works")
793 self.assertEqual(len(root[0]), 1)
794 self.assertEqual(root[0][0].tag, "othertag")
795 # FIXME: would be nice to get some errors logged ...
796 #self.assertTrue(len(parser.error_log) > 0, "error log is empty")
797
799 assertEqual = self.assertEqual
800 assertFalse = self.assertFalse
801
802 events = []
803 class Target(object):
804 def start(self, tag, attrib):
805 events.append("start")
806 assertFalse(attrib)
807 assertEqual("TAG", tag)
808 def end(self, tag):
809 events.append("end")
810 assertEqual("TAG", tag)
811 def close(self):
812 return "DONE" # no Element!
813
814 parser = self.etree.XMLParser(target=Target())
815 tree = self.etree.ElementTree()
816
817 self.assertRaises(TypeError,
818 tree.parse, BytesIO("<TAG/>"), parser=parser)
819 self.assertEqual(["start", "end"], events)
820
822 # ET doesn't call .close() on errors
823 events = []
824 class Target(object):
825 def start(self, tag, attrib):
826 events.append("start-" + tag)
827 def end(self, tag):
828 events.append("end-" + tag)
829 if tag == 'a':
830 raise ValueError("dead and gone")
831 def data(self, data):
832 events.append("data-" + data)
833 def close(self):
834 events.append("close")
835 return "DONE"
836
837 parser = self.etree.XMLParser(target=Target())
838
839 try:
840 parser.feed(_bytes('<root>A<a>ca</a>B</root>'))
841 done = parser.close()
842 self.fail("error expected, but parsing succeeded")
843 except ValueError:
844 done = 'value error received as expected'
845
846 self.assertEqual(["start-root", "data-A", "start-a",
847 "data-ca", "end-a", "close"],
848 events)
849
851 # ET doesn't call .close() on errors
852 events = []
853 class Target(object):
854 def start(self, tag, attrib):
855 events.append("start-" + tag)
856 def end(self, tag):
857 events.append("end-" + tag)
858 if tag == 'a':
859 raise ValueError("dead and gone")
860 def data(self, data):
861 events.append("data-" + data)
862 def close(self):
863 events.append("close")
864 return "DONE"
865
866 parser = self.etree.XMLParser(target=Target())
867
868 try:
869 done = self.etree.fromstring(_bytes('<root>A<a>ca</a>B</root>'),
870 parser=parser)
871 self.fail("error expected, but parsing succeeded")
872 except ValueError:
873 done = 'value error received as expected'
874
875 self.assertEqual(["start-root", "data-A", "start-a",
876 "data-ca", "end-a", "close"],
877 events)
878
880 events = []
881 class Target(object):
882 def start(self, tag, attrib):
883 events.append("start-" + tag)
884 def end(self, tag):
885 events.append("end-" + tag)
886 def data(self, data):
887 events.append("data-" + data)
888 def comment(self, text):
889 events.append("comment-" + text)
890 def close(self):
891 return "DONE"
892
893 parser = self.etree.XMLParser(target=Target())
894
895 parser.feed(_bytes('<!--a--><root>A<!--b--><sub/><!--c-->B</root><!--d-->'))
896 done = parser.close()
897
898 self.assertEqual("DONE", done)
899 self.assertEqual(["comment-a", "start-root", "data-A", "comment-b",
900 "start-sub", "end-sub", "comment-c", "data-B",
901 "end-root", "comment-d"],
902 events)
903
905 events = []
906 class Target(object):
907 def start(self, tag, attrib):
908 events.append("start-" + tag)
909 def end(self, tag):
910 events.append("end-" + tag)
911 def data(self, data):
912 events.append("data-" + data)
913 def pi(self, target, data):
914 events.append("pi-" + target + "-" + data)
915 def close(self):
916 return "DONE"
917
918 parser = self.etree.XMLParser(target=Target())
919
920 parser.feed(_bytes('<?test a?><root>A<?test b?>B</root><?test c?>'))
921 done = parser.close()
922
923 self.assertEqual("DONE", done)
924 self.assertEqual(["pi-test-a", "start-root", "data-A", "pi-test-b",
925 "data-B", "end-root", "pi-test-c"],
926 events)
927
929 events = []
930 class Target(object):
931 def start(self, tag, attrib):
932 events.append("start-" + tag)
933 def end(self, tag):
934 events.append("end-" + tag)
935 def data(self, data):
936 events.append("data-" + data)
937 def close(self):
938 return "DONE"
939
940 parser = self.etree.XMLParser(target=Target(),
941 strip_cdata=False)
942
943 parser.feed(_bytes('<root>A<a><![CDATA[ca]]></a>B</root>'))
944 done = parser.close()
945
946 self.assertEqual("DONE", done)
947 self.assertEqual(["start-root", "data-A", "start-a",
948 "data-ca", "end-a", "data-B", "end-root"],
949 events)
950
952 events = []
953 class Target(object):
954 def start(self, tag, attrib):
955 events.append("start-" + tag)
956 def end(self, tag):
957 events.append("end-" + tag)
958 def data(self, data):
959 events.append("data-" + data)
960 def close(self):
961 events.append("close")
962 return "DONE"
963
964 parser = self.etree.XMLParser(target=Target(),
965 recover=True)
966
967 parser.feed(_bytes('<root>A<a>ca</a>B</not-root>'))
968 done = parser.close()
969
970 self.assertEqual("DONE", done)
971 self.assertEqual(["start-root", "data-A", "start-a",
972 "data-ca", "end-a", "data-B",
973 "end-root", "close"],
974 events)
975
977 iterwalk = self.etree.iterwalk
978 root = self.etree.XML(_bytes('<a><b><d/></b><c/></a>'))
979
980 iterator = iterwalk(root, tag="b", events=('start', 'end'))
981 events = list(iterator)
982 self.assertEqual(
983 [('start', root[0]), ('end', root[0])],
984 events)
985
987 iterwalk = self.etree.iterwalk
988 root = self.etree.XML(_bytes('<a><b><d/></b><c/></a>'))
989
990 iterator = iterwalk(root, tag="*", events=('start', 'end'))
991 events = list(iterator)
992 self.assertEqual(
993 8,
994 len(events))
995
997 iterwalk = self.etree.iterwalk
998 root = self.etree.XML(_bytes('<a><b></b><c/></a>'))
999
1000 events = list(iterwalk(root))
1001 self.assertEqual(
1002 [('end', root[0]), ('end', root[1]), ('end', root)],
1003 events)
1004
1006 iterwalk = self.etree.iterwalk
1007 root = self.etree.XML(_bytes('<a><b></b><c/></a>'))
1008
1009 iterator = iterwalk(root, events=('start',))
1010 events = list(iterator)
1011 self.assertEqual(
1012 [('start', root), ('start', root[0]), ('start', root[1])],
1013 events)
1014
1016 iterwalk = self.etree.iterwalk
1017 root = self.etree.XML(_bytes('<a><b></b><c/></a>'))
1018
1019 iterator = iterwalk(root, events=('start','end'))
1020 events = list(iterator)
1021 self.assertEqual(
1022 [('start', root), ('start', root[0]), ('end', root[0]),
1023 ('start', root[1]), ('end', root[1]), ('end', root)],
1024 events)
1025
1027 iterwalk = self.etree.iterwalk
1028 root = self.etree.XML(_bytes('<a><b></b><c/></a>'))
1029
1030 iterator = iterwalk(root)
1031 for event, elem in iterator:
1032 elem.clear()
1033
1034 self.assertEqual(0,
1035 len(root))
1036
1038 iterwalk = self.etree.iterwalk
1039 root = self.etree.XML(_bytes('<a xmlns="ns1"><b><c xmlns="ns2"/></b></a>'))
1040
1041 attr_name = '{testns}bla'
1042 events = []
1043 iterator = iterwalk(root, events=('start','end','start-ns','end-ns'))
1044 for event, elem in iterator:
1045 events.append(event)
1046 if event == 'start':
1047 if elem.tag != '{ns1}a':
1048 elem.set(attr_name, 'value')
1049
1050 self.assertEqual(
1051 ['start-ns', 'start', 'start', 'start-ns', 'start',
1052 'end', 'end-ns', 'end', 'end', 'end-ns'],
1053 events)
1054
1055 self.assertEqual(
1056 None,
1057 root.get(attr_name))
1058 self.assertEqual(
1059 'value',
1060 root[0].get(attr_name))
1061
1063 iterwalk = self.etree.iterwalk
1064 root = self.etree.XML(_bytes('<a><b><d/></b><c/></a>'))
1065
1066 counts = []
1067 for event, elem in iterwalk(root):
1068 counts.append(len(list(elem.getiterator())))
1069 self.assertEqual(
1070 [1,2,1,4],
1071 counts)
1072
1074 parse = self.etree.parse
1075 parser = self.etree.XMLParser(dtd_validation=True)
1076 assertEqual = self.assertEqual
1077 test_url = _str("__nosuch.dtd")
1078
1079 class MyResolver(self.etree.Resolver):
1080 def resolve(self, url, id, context):
1081 assertEqual(url, test_url)
1082 return self.resolve_string(
1083 _str('''<!ENTITY myentity "%s">
1084 <!ELEMENT doc ANY>''') % url, context)
1085
1086 parser.resolvers.add(MyResolver())
1087
1088 xml = _str('<!DOCTYPE doc SYSTEM "%s"><doc>&myentity;</doc>') % test_url
1089 tree = parse(StringIO(xml), parser)
1090 root = tree.getroot()
1091 self.assertEqual(root.text, test_url)
1092
1094 parse = self.etree.parse
1095 parser = self.etree.XMLParser(dtd_validation=True)
1096 assertEqual = self.assertEqual
1097 test_url = _str("__nosuch.dtd")
1098
1099 class MyResolver(self.etree.Resolver):
1100 def resolve(self, url, id, context):
1101 assertEqual(url, test_url)
1102 return self.resolve_string(
1103 (_str('''<!ENTITY myentity "%s">
1104 <!ELEMENT doc ANY>''') % url).encode('utf-8'),
1105 context)
1106
1107 parser.resolvers.add(MyResolver())
1108
1109 xml = _str('<!DOCTYPE doc SYSTEM "%s"><doc>&myentity;</doc>') % test_url
1110 tree = parse(StringIO(xml), parser)
1111 root = tree.getroot()
1112 self.assertEqual(root.text, test_url)
1113
1115 parse = self.etree.parse
1116 parser = self.etree.XMLParser(dtd_validation=True)
1117 assertEqual = self.assertEqual
1118 test_url = _str("__nosuch.dtd")
1119
1120 class MyResolver(self.etree.Resolver):
1121 def resolve(self, url, id, context):
1122 assertEqual(url, test_url)
1123 return self.resolve_file(
1124 SillyFileLike(
1125 _str('''<!ENTITY myentity "%s">
1126 <!ELEMENT doc ANY>''') % url), context)
1127
1128 parser.resolvers.add(MyResolver())
1129
1130 xml = _str('<!DOCTYPE doc SYSTEM "%s"><doc>&myentity;</doc>') % test_url
1131 tree = parse(StringIO(xml), parser)
1132 root = tree.getroot()
1133 self.assertEqual(root.text, test_url)
1134
1136 parse = self.etree.parse
1137 parser = self.etree.XMLParser(attribute_defaults=True)
1138 assertEqual = self.assertEqual
1139 test_url = _str("__nosuch.dtd")
1140
1141 class MyResolver(self.etree.Resolver):
1142 def resolve(self, url, id, context):
1143 assertEqual(url, test_url)
1144 return self.resolve_filename(
1145 fileInTestDir('test.dtd'), context)
1146
1147 parser.resolvers.add(MyResolver())
1148
1149 xml = _str('<!DOCTYPE a SYSTEM "%s"><a><b/></a>') % test_url
1150 tree = parse(StringIO(xml), parser)
1151 root = tree.getroot()
1152 self.assertEqual(
1153 root.attrib, {'default': 'valueA'})
1154 self.assertEqual(
1155 root[0].attrib, {'default': 'valueB'})
1156
1158 parse = self.etree.parse
1159 parser = self.etree.XMLParser(attribute_defaults=True)
1160 assertEqual = self.assertEqual
1161 test_url = _str("__nosuch.dtd")
1162
1163 class MyResolver(self.etree.Resolver):
1164 def resolve(self, url, id, context):
1165 assertEqual(url, fileInTestDir(test_url))
1166 return self.resolve_filename(
1167 fileInTestDir('test.dtd'), context)
1168
1169 parser.resolvers.add(MyResolver())
1170
1171 xml = _str('<!DOCTYPE a SYSTEM "%s"><a><b/></a>') % test_url
1172 tree = parse(StringIO(xml), parser,
1173 base_url=fileInTestDir('__test.xml'))
1174 root = tree.getroot()
1175 self.assertEqual(
1176 root.attrib, {'default': 'valueA'})
1177 self.assertEqual(
1178 root[0].attrib, {'default': 'valueB'})
1179
1181 parse = self.etree.parse
1182 parser = self.etree.XMLParser(attribute_defaults=True)
1183 assertEqual = self.assertEqual
1184 test_url = _str("__nosuch.dtd")
1185
1186 class MyResolver(self.etree.Resolver):
1187 def resolve(self, url, id, context):
1188 assertEqual(url, test_url)
1189 return self.resolve_file(
1190 open(fileInTestDir('test.dtd'), 'rb'), context)
1191
1192 parser.resolvers.add(MyResolver())
1193
1194 xml = _str('<!DOCTYPE a SYSTEM "%s"><a><b/></a>') % test_url
1195 tree = parse(StringIO(xml), parser)
1196 root = tree.getroot()
1197 self.assertEqual(
1198 root.attrib, {'default': 'valueA'})
1199 self.assertEqual(
1200 root[0].attrib, {'default': 'valueB'})
1201
1203 parse = self.etree.parse
1204 parser = self.etree.XMLParser(load_dtd=True)
1205 assertEqual = self.assertEqual
1206 test_url = _str("__nosuch.dtd")
1207
1208 class check(object):
1209 resolved = False
1210
1211 class MyResolver(self.etree.Resolver):
1212 def resolve(self, url, id, context):
1213 assertEqual(url, test_url)
1214 check.resolved = True
1215 return self.resolve_empty(context)
1216
1217 parser.resolvers.add(MyResolver())
1218
1219 xml = _str('<!DOCTYPE doc SYSTEM "%s"><doc>&myentity;</doc>') % test_url
1220 self.assertRaises(etree.XMLSyntaxError, parse, StringIO(xml), parser)
1221 self.assertTrue(check.resolved)
1222
1224 parse = self.etree.parse
1225 parser = self.etree.XMLParser(dtd_validation=True)
1226
1227 class _LocalException(Exception):
1228 pass
1229
1230 class MyResolver(self.etree.Resolver):
1231 def resolve(self, url, id, context):
1232 raise _LocalException
1233
1234 parser.resolvers.add(MyResolver())
1235
1236 xml = '<!DOCTYPE doc SYSTEM "test"><doc>&myentity;</doc>'
1237 self.assertRaises(_LocalException, parse, BytesIO(xml), parser)
1238
1239 if etree.LIBXML_VERSION > (2,6,20):
1241 parse = self.etree.parse
1242 tostring = self.etree.tostring
1243 parser = self.etree.XMLParser(resolve_entities=False)
1244 Entity = self.etree.Entity
1245
1246 xml = _bytes('<!DOCTYPE doc SYSTEM "test"><doc>&myentity;</doc>')
1247 tree = parse(BytesIO(xml), parser)
1248 root = tree.getroot()
1249 self.assertEqual(root[0].tag, Entity)
1250 self.assertEqual(root[0].text, "&myentity;")
1251 self.assertEqual(root[0].tail, None)
1252 self.assertEqual(root[0].name, "myentity")
1253
1254 self.assertEqual(_bytes('<doc>&myentity;</doc>'),
1255 tostring(root))
1256
1258 xml = _bytes('''<!DOCTYPE root [ <!ENTITY nbsp " "> ]>
1259 <root>
1260 <child1/>
1261 <child2/>
1262 <child3> </child3>
1263 </root>''')
1264
1265 parser = self.etree.XMLParser(resolve_entities=False)
1266 root = etree.fromstring(xml, parser)
1267 self.assertEqual([ el.tag for el in root ],
1268 ['child1', 'child2', 'child3'])
1269
1270 root[0] = root[-1]
1271 self.assertEqual([ el.tag for el in root ],
1272 ['child3', 'child2'])
1273 self.assertEqual(root[0][0].text, ' ')
1274 self.assertEqual(root[0][0].name, 'nbsp')
1275
1277 Entity = self.etree.Entity
1278 Element = self.etree.Element
1279 tostring = self.etree.tostring
1280
1281 root = Element("root")
1282 root.append( Entity("test") )
1283
1284 self.assertEqual(root[0].tag, Entity)
1285 self.assertEqual(root[0].text, "&test;")
1286 self.assertEqual(root[0].tail, None)
1287 self.assertEqual(root[0].name, "test")
1288
1289 self.assertEqual(_bytes('<root>&test;</root>'),
1290 tostring(root))
1291
1293 Entity = self.etree.Entity
1294 self.assertEqual(Entity("test").text, '&test;')
1295 self.assertEqual(Entity("#17683").text, '䔓')
1296 self.assertEqual(Entity("#x1768").text, 'ᝨ')
1297 self.assertEqual(Entity("#x98AF").text, '颯')
1298
1300 Entity = self.etree.Entity
1301 self.assertRaises(ValueError, Entity, 'a b c')
1302 self.assertRaises(ValueError, Entity, 'a,b')
1303 self.assertRaises(ValueError, Entity, 'a\0b')
1304 self.assertRaises(ValueError, Entity, '#abc')
1305 self.assertRaises(ValueError, Entity, '#xxyz')
1306
1308 CDATA = self.etree.CDATA
1309 Element = self.etree.Element
1310 tostring = self.etree.tostring
1311
1312 root = Element("root")
1313 root.text = CDATA('test')
1314
1315 self.assertEqual('test',
1316 root.text)
1317 self.assertEqual(_bytes('<root><![CDATA[test]]></root>'),
1318 tostring(root))
1319
1321 CDATA = self.etree.CDATA
1322 Element = self.etree.Element
1323 root = Element("root")
1324
1325 root.text = CDATA("test")
1326 self.assertEqual('test', root.text)
1327
1328 root.text = CDATA(_str("test"))
1329 self.assertEqual('test', root.text)
1330
1331 self.assertRaises(TypeError, CDATA, 1)
1332
1334 CDATA = self.etree.CDATA
1335 Element = self.etree.Element
1336
1337 root = Element("root")
1338 cdata = CDATA('test')
1339
1340 self.assertRaises(TypeError,
1341 setattr, root, 'tail', cdata)
1342 self.assertRaises(TypeError,
1343 root.set, 'attr', cdata)
1344 self.assertRaises(TypeError,
1345 operator.setitem, root.attrib, 'attr', cdata)
1346
1348 tostring = self.etree.tostring
1349 parser = self.etree.XMLParser(strip_cdata=False)
1350 root = self.etree.XML(_bytes('<root><![CDATA[test]]></root>'), parser)
1351
1352 self.assertEqual('test', root.text)
1353 self.assertEqual(_bytes('<root><![CDATA[test]]></root>'),
1354 tostring(root))
1355
1357 tostring = self.etree.tostring
1358 parser = self.etree.XMLParser(strip_cdata=False)
1359 root = self.etree.XML(_bytes('<root><![CDATA[test]]></root>'), parser)
1360 self.assertEqual(_bytes('<root><![CDATA[test]]></root>'),
1361 tostring(root))
1362
1363 self.assertEqual(['test'], root.xpath('//text()'))
1364
1365 # TypeError in etree, AssertionError in ElementTree;
1367 Element = self.etree.Element
1368 SubElement = self.etree.SubElement
1369
1370 a = Element('a')
1371 b = SubElement(a, 'b')
1372
1373 self.assertRaises(TypeError,
1374 a.__setitem__, 0, 'foo')
1375
1377 Element = self.etree.Element
1378 root = Element('root')
1379 # raises AssertionError in ElementTree
1380 self.assertRaises(TypeError, root.append, None)
1381 self.assertRaises(TypeError, root.extend, [None])
1382 self.assertRaises(TypeError, root.extend, [Element('one'), None])
1383 self.assertEqual('one', root[0].tag)
1384
1386 Element = self.etree.Element
1387 SubElement = self.etree.SubElement
1388 root = Element('root')
1389 self.assertRaises(ValueError, root.append, root)
1390 child = SubElement(root, 'child')
1391 self.assertRaises(ValueError, child.append, root)
1392 child2 = SubElement(child, 'child2')
1393 self.assertRaises(ValueError, child2.append, root)
1394 self.assertRaises(ValueError, child2.append, child)
1395 self.assertEqual('child2', root[0][0].tag)
1396
1398 Element = self.etree.Element
1399 SubElement = self.etree.SubElement
1400 root = Element('root')
1401 SubElement(root, 'a')
1402 SubElement(root, 'b')
1403
1404 self.assertEqual(['a', 'b'],
1405 [c.tag for c in root])
1406 root[1].addnext(root[0])
1407 self.assertEqual(['b', 'a'],
1408 [c.tag for c in root])
1409
1411 Element = self.etree.Element
1412 SubElement = self.etree.SubElement
1413 root = Element('root')
1414 SubElement(root, 'a')
1415 SubElement(root, 'b')
1416
1417 self.assertEqual(['a', 'b'],
1418 [c.tag for c in root])
1419 root[0].addprevious(root[1])
1420 self.assertEqual(['b', 'a'],
1421 [c.tag for c in root])
1422
1424 Element = self.etree.Element
1425 SubElement = self.etree.SubElement
1426 root = Element('root')
1427 a = SubElement(root, 'a')
1428 b = SubElement(root, 'b')
1429 a.addprevious(a)
1430 self.assertEqual('a', root[0].tag)
1431 self.assertEqual('b', root[1].tag)
1432 b.addprevious(b)
1433 self.assertEqual('a', root[0].tag)
1434 self.assertEqual('b', root[1].tag)
1435 b.addprevious(a)
1436 self.assertEqual('a', root[0].tag)
1437 self.assertEqual('b', root[1].tag)
1438
1440 Element = self.etree.Element
1441 SubElement = self.etree.SubElement
1442 root = Element('root')
1443 a = SubElement(root, 'a')
1444 b = SubElement(root, 'b')
1445 a.addnext(a)
1446 self.assertEqual('a', root[0].tag)
1447 self.assertEqual('b', root[1].tag)
1448 b.addnext(b)
1449 self.assertEqual('a', root[0].tag)
1450 self.assertEqual('b', root[1].tag)
1451 a.addnext(b)
1452 self.assertEqual('a', root[0].tag)
1453 self.assertEqual('b', root[1].tag)
1454
1456 Element = self.etree.Element
1457 a = Element('a')
1458 b = Element('b')
1459 self.assertRaises(TypeError, a.addnext, b)
1460
1462 Element = self.etree.Element
1463 SubElement = self.etree.SubElement
1464 PI = self.etree.PI
1465 root = Element('root')
1466 SubElement(root, 'a')
1467 pi = PI('TARGET', 'TEXT')
1468 pi.tail = "TAIL"
1469
1470 self.assertEqual(_bytes('<root><a></a></root>'),
1471 self._writeElement(root))
1472 root[0].addprevious(pi)
1473 self.assertEqual(_bytes('<root><?TARGET TEXT?>TAIL<a></a></root>'),
1474 self._writeElement(root))
1475
1477 Element = self.etree.Element
1478 PI = self.etree.PI
1479 root = Element('root')
1480 pi = PI('TARGET', 'TEXT')
1481 pi.tail = "TAIL"
1482
1483 self.assertEqual(_bytes('<root></root>'),
1484 self._writeElement(root))
1485 root.addprevious(pi)
1486 self.assertEqual(_bytes('<?TARGET TEXT?>\n<root></root>'),
1487 self._writeElement(root))
1488
1490 Element = self.etree.Element
1491 SubElement = self.etree.SubElement
1492 PI = self.etree.PI
1493 root = Element('root')
1494 SubElement(root, 'a')
1495 pi = PI('TARGET', 'TEXT')
1496 pi.tail = "TAIL"
1497
1498 self.assertEqual(_bytes('<root><a></a></root>'),
1499 self._writeElement(root))
1500 root[0].addnext(pi)
1501 self.assertEqual(_bytes('<root><a></a><?TARGET TEXT?>TAIL</root>'),
1502 self._writeElement(root))
1503
1505 Element = self.etree.Element
1506 PI = self.etree.PI
1507 root = Element('root')
1508 pi = PI('TARGET', 'TEXT')
1509 pi.tail = "TAIL"
1510
1511 self.assertEqual(_bytes('<root></root>'),
1512 self._writeElement(root))
1513 root.addnext(pi)
1514 self.assertEqual(_bytes('<root></root>\n<?TARGET TEXT?>'),
1515 self._writeElement(root))
1516
1518 Element = self.etree.Element
1519 SubElement = self.etree.SubElement
1520 Comment = self.etree.Comment
1521 root = Element('root')
1522 SubElement(root, 'a')
1523 comment = Comment('TEXT ')
1524 comment.tail = "TAIL"
1525
1526 self.assertEqual(_bytes('<root><a></a></root>'),
1527 self._writeElement(root))
1528 root[0].addnext(comment)
1529 self.assertEqual(_bytes('<root><a></a><!--TEXT -->TAIL</root>'),
1530 self._writeElement(root))
1531
1533 Element = self.etree.Element
1534 Comment = self.etree.Comment
1535 root = Element('root')
1536 comment = Comment('TEXT ')
1537 comment.tail = "TAIL"
1538
1539 self.assertEqual(_bytes('<root></root>'),
1540 self._writeElement(root))
1541 root.addnext(comment)
1542 self.assertEqual(_bytes('<root></root>\n<!--TEXT -->'),
1543 self._writeElement(root))
1544
1546 Element = self.etree.Element
1547 SubElement = self.etree.SubElement
1548 Comment = self.etree.Comment
1549 root = Element('root')
1550 SubElement(root, 'a')
1551 comment = Comment('TEXT ')
1552 comment.tail = "TAIL"
1553
1554 self.assertEqual(_bytes('<root><a></a></root>'),
1555 self._writeElement(root))
1556 root[0].addprevious(comment)
1557 self.assertEqual(_bytes('<root><!--TEXT -->TAIL<a></a></root>'),
1558 self._writeElement(root))
1559
1561 Element = self.etree.Element
1562 Comment = self.etree.Comment
1563 root = Element('root')
1564 comment = Comment('TEXT ')
1565 comment.tail = "TAIL"
1566
1567 self.assertEqual(_bytes('<root></root>'),
1568 self._writeElement(root))
1569 root.addprevious(comment)
1570 self.assertEqual(_bytes('<!--TEXT -->\n<root></root>'),
1571 self._writeElement(root))
1572
1573 # ET's Elements have items() and key(), but not values()
1575 XML = self.etree.XML
1576
1577 root = XML(_bytes('<doc alpha="Alpha" beta="Beta" gamma="Gamma"/>'))
1578 values = root.values()
1579 values.sort()
1580 self.assertEqual(['Alpha', 'Beta', 'Gamma'], values)
1581
1582 # gives error in ElementTree
1584 Element = self.etree.Element
1585 Comment = self.etree.Comment
1586
1587 a = Element('a')
1588 a.append(Comment())
1589 self.assertEqual(
1590 _bytes('<a><!----></a>'),
1591 self._writeElement(a))
1592
1593 # ElementTree ignores comments
1595 ElementTree = self.etree.ElementTree
1596 tostring = self.etree.tostring
1597
1598 xml = _bytes('<a><b/><!----><c/></a>')
1599 f = BytesIO(xml)
1600 doc = ElementTree(file=f)
1601 a = doc.getroot()
1602 self.assertEqual(
1603 '',
1604 a[1].text)
1605 self.assertEqual(
1606 xml,
1607 tostring(a))
1608
1609 # ElementTree ignores comments
1611 ElementTree = self.etree.ElementTree
1612
1613 f = BytesIO('<a><b></b><!-- hoi --><c></c></a>')
1614 doc = ElementTree(file=f)
1615 a = doc.getroot()
1616 self.assertEqual(
1617 ' hoi ',
1618 a[1].text)
1619
1620 # does not raise an exception in ElementTree
1622 Element = self.etree.Element
1623 Comment = self.etree.Comment
1624
1625 c = Comment()
1626 el = Element('myel')
1627
1628 self.assertRaises(TypeError, c.append, el)
1629 self.assertRaises(TypeError, c.insert, 0, el)
1630 self.assertRaises(TypeError, c.set, "myattr", "test")
1631
1632 # test passing 'None' to dump
1635
1637 ElementTree = self.etree.ElementTree
1638
1639 f = BytesIO('<a xmlns:foo="http://www.infrae.com/ns/1"><foo:b/></a>')
1640 doc = ElementTree(file=f)
1641 a = doc.getroot()
1642 self.assertEqual(
1643 None,
1644 a.prefix)
1645 self.assertEqual(
1646 'foo',
1647 a[0].prefix)
1648
1650 ElementTree = self.etree.ElementTree
1651
1652 f = BytesIO('<a xmlns="http://www.infrae.com/ns/1"><b/></a>')
1653 doc = ElementTree(file=f)
1654 a = doc.getroot()
1655 self.assertEqual(
1656 None,
1657 a.prefix)
1658 self.assertEqual(
1659 None,
1660 a[0].prefix)
1661
1663 Element = self.etree.Element
1664 SubElement = self.etree.SubElement
1665
1666 a = Element('a')
1667 b = SubElement(a, 'b')
1668 c = SubElement(a, 'c')
1669 d = SubElement(b, 'd')
1670 self.assertEqual(
1671 None,
1672 a.getparent())
1673 self.assertEqual(
1674 a,
1675 b.getparent())
1676 self.assertEqual(
1677 b.getparent(),
1678 c.getparent())
1679 self.assertEqual(
1680 b,
1681 d.getparent())
1682
1684 XML = self.etree.XML
1685
1686 root = XML(_bytes('<doc><one/><two>Two</two>Hm<three/></doc>'))
1687 result = []
1688 for el in root.iterchildren():
1689 result.append(el.tag)
1690 self.assertEqual(['one', 'two', 'three'], result)
1691
1693 XML = self.etree.XML
1694
1695 root = XML(_bytes('<doc><one/><two>Two</two>Hm<three/></doc>'))
1696 result = []
1697 for el in root.iterchildren(reversed=True):
1698 result.append(el.tag)
1699 self.assertEqual(['three', 'two', 'one'], result)
1700
1702 XML = self.etree.XML
1703
1704 root = XML(_bytes('<doc><one/><two>Two</two>Hm<two>Bla</two></doc>'))
1705 result = []
1706 for el in root.iterchildren(tag='two'):
1707 result.append(el.text)
1708 self.assertEqual(['Two', 'Bla'], result)
1709
1711 XML = self.etree.XML
1712
1713 root = XML(_bytes('<doc><one/><two>Two</two>Hm<two>Bla</two></doc>'))
1714 result = []
1715 for el in root.iterchildren('two'):
1716 result.append(el.text)
1717 self.assertEqual(['Two', 'Bla'], result)
1718
1720 XML = self.etree.XML
1721
1722 root = XML(_bytes('<doc><one/><two>Two</two>Hm<two>Bla</two></doc>'))
1723 result = []
1724 for el in root.iterchildren(reversed=True, tag='two'):
1725 result.append(el.text)
1726 self.assertEqual(['Bla', 'Two'], result)
1727
1729 XML = self.etree.XML
1730
1731 root = XML(_bytes('<doc><one/><two>Two</two>Hm<two>Bla</two><three/></doc>'))
1732 result = []
1733 for el in root.iterchildren(tag=['two', 'three']):
1734 result.append(el.text)
1735 self.assertEqual(['Two', 'Bla', None], result)
1736
1738 XML = self.etree.XML
1739
1740 root = XML(_bytes('<doc><one/><two>Two</two>Hm<two>Bla</two><three/></doc>'))
1741 result = []
1742 for el in root.iterchildren('two', 'three'):
1743 result.append(el.text)
1744 self.assertEqual(['Two', 'Bla', None], result)
1745
1747 XML = self.etree.XML
1748
1749 root = XML(_bytes('<doc><one/><two>Two</two>Hm<two>Bla</two><three/></doc>'))
1750 result = []
1751 for el in root.iterchildren(reversed=True, tag=['two', 'three']):
1752 result.append(el.text)
1753 self.assertEqual([None, 'Bla', 'Two'], result)
1754
1756 Element = self.etree.Element
1757 SubElement = self.etree.SubElement
1758
1759 a = Element('a')
1760 b = SubElement(a, 'b')
1761 c = SubElement(a, 'c')
1762 d = SubElement(b, 'd')
1763 self.assertEqual(
1764 [],
1765 list(a.iterancestors()))
1766 self.assertEqual(
1767 [a],
1768 list(b.iterancestors()))
1769 self.assertEqual(
1770 [a],
1771 list(c.iterancestors()))
1772 self.assertEqual(
1773 [b, a],
1774 list(d.iterancestors()))
1775
1777 Element = self.etree.Element
1778 SubElement = self.etree.SubElement
1779
1780 a = Element('a')
1781 b = SubElement(a, 'b')
1782 c = SubElement(a, 'c')
1783 d = SubElement(b, 'd')
1784 self.assertEqual(
1785 [a],
1786 list(d.iterancestors('a')))
1787 self.assertEqual(
1788 [a],
1789 list(d.iterancestors(tag='a')))
1790
1791 self.assertEqual(
1792 [b, a],
1793 list(d.iterancestors('*')))
1794 self.assertEqual(
1795 [b, a],
1796 list(d.iterancestors(tag='*')))
1797
1799 Element = self.etree.Element
1800 SubElement = self.etree.SubElement
1801
1802 a = Element('a')
1803 b = SubElement(a, 'b')
1804 c = SubElement(a, 'c')
1805 d = SubElement(b, 'd')
1806 self.assertEqual(
1807 [b, a],
1808 list(d.iterancestors(tag=('a', 'b'))))
1809 self.assertEqual(
1810 [b, a],
1811 list(d.iterancestors('a', 'b')))
1812
1813 self.assertEqual(
1814 [],
1815 list(d.iterancestors(tag=('w', 'x', 'y', 'z'))))
1816 self.assertEqual(
1817 [],
1818 list(d.iterancestors('w', 'x', 'y', 'z')))
1819
1820 self.assertEqual(
1821 [],
1822 list(d.iterancestors(tag=('d', 'x'))))
1823 self.assertEqual(
1824 [],
1825 list(d.iterancestors('d', 'x')))
1826
1827 self.assertEqual(
1828 [b, a],
1829 list(d.iterancestors(tag=('b', '*'))))
1830 self.assertEqual(
1831 [b, a],
1832 list(d.iterancestors('b', '*')))
1833
1834 self.assertEqual(
1835 [b],
1836 list(d.iterancestors(tag=('b', 'c'))))
1837 self.assertEqual(
1838 [b],
1839 list(d.iterancestors('b', 'c')))
1840
1842 Element = self.etree.Element
1843 SubElement = self.etree.SubElement
1844
1845 a = Element('a')
1846 b = SubElement(a, 'b')
1847 c = SubElement(a, 'c')
1848 d = SubElement(b, 'd')
1849 e = SubElement(c, 'e')
1850
1851 self.assertEqual(
1852 [b, d, c, e],
1853 list(a.iterdescendants()))
1854 self.assertEqual(
1855 [],
1856 list(d.iterdescendants()))
1857
1859 Element = self.etree.Element
1860 SubElement = self.etree.SubElement
1861
1862 a = Element('a')
1863 b = SubElement(a, 'b')
1864 c = SubElement(a, 'c')
1865 d = SubElement(b, 'd')
1866 e = SubElement(c, 'e')
1867
1868 self.assertEqual(
1869 [],
1870 list(a.iterdescendants('a')))
1871 self.assertEqual(
1872 [],
1873 list(a.iterdescendants(tag='a')))
1874
1875 a2 = SubElement(e, 'a')
1876 self.assertEqual(
1877 [a2],
1878 list(a.iterdescendants('a')))
1879
1880 self.assertEqual(
1881 [a2],
1882 list(c.iterdescendants('a')))
1883 self.assertEqual(
1884 [a2],
1885 list(c.iterdescendants(tag='a')))
1886
1888 Element = self.etree.Element
1889 SubElement = self.etree.SubElement
1890
1891 a = Element('a')
1892 b = SubElement(a, 'b')
1893 c = SubElement(a, 'c')
1894 d = SubElement(b, 'd')
1895 e = SubElement(c, 'e')
1896
1897 self.assertEqual(
1898 [b, e],
1899 list(a.iterdescendants(tag=('a', 'b', 'e'))))
1900 self.assertEqual(
1901 [b, e],
1902 list(a.iterdescendants('a', 'b', 'e')))
1903
1904 a2 = SubElement(e, 'a')
1905 self.assertEqual(
1906 [b, a2],
1907 list(a.iterdescendants(tag=('a', 'b'))))
1908 self.assertEqual(
1909 [b, a2],
1910 list(a.iterdescendants('a', 'b')))
1911
1912 self.assertEqual(
1913 [],
1914 list(c.iterdescendants(tag=('x', 'y', 'z'))))
1915 self.assertEqual(
1916 [],
1917 list(c.iterdescendants('x', 'y', 'z')))
1918
1919 self.assertEqual(
1920 [b, d, c, e, a2],
1921 list(a.iterdescendants(tag=('x', 'y', 'z', '*'))))
1922 self.assertEqual(
1923 [b, d, c, e, a2],
1924 list(a.iterdescendants('x', 'y', 'z', '*')))
1925
1927 Element = self.etree.Element
1928 SubElement = self.etree.SubElement
1929
1930 a = Element('a')
1931 b = SubElement(a, 'b')
1932 c = SubElement(a, 'c')
1933 d = SubElement(b, 'd')
1934 self.assertEqual(
1935 a,
1936 a.getroottree().getroot())
1937 self.assertEqual(
1938 a,
1939 b.getroottree().getroot())
1940 self.assertEqual(
1941 a,
1942 d.getroottree().getroot())
1943
1945 Element = self.etree.Element
1946 SubElement = self.etree.SubElement
1947
1948 a = Element('a')
1949 b = SubElement(a, 'b')
1950 c = SubElement(a, 'c')
1951 self.assertEqual(
1952 None,
1953 a.getnext())
1954 self.assertEqual(
1955 c,
1956 b.getnext())
1957 self.assertEqual(
1958 None,
1959 c.getnext())
1960
1962 Element = self.etree.Element
1963 SubElement = self.etree.SubElement
1964
1965 a = Element('a')
1966 b = SubElement(a, 'b')
1967 c = SubElement(a, 'c')
1968 d = SubElement(b, 'd')
1969 self.assertEqual(
1970 None,
1971 a.getprevious())
1972 self.assertEqual(
1973 b,
1974 c.getprevious())
1975 self.assertEqual(
1976 None,
1977 b.getprevious())
1978
1980 Element = self.etree.Element
1981 SubElement = self.etree.SubElement
1982
1983 a = Element('a')
1984 b = SubElement(a, 'b')
1985 c = SubElement(a, 'c')
1986 d = SubElement(b, 'd')
1987 self.assertEqual(
1988 [],
1989 list(a.itersiblings()))
1990 self.assertEqual(
1991 [c],
1992 list(b.itersiblings()))
1993 self.assertEqual(
1994 [],
1995 list(c.itersiblings()))
1996 self.assertEqual(
1997 [b],
1998 list(c.itersiblings(preceding=True)))
1999 self.assertEqual(
2000 [],
2001 list(b.itersiblings(preceding=True)))
2002
2004 Element = self.etree.Element
2005 SubElement = self.etree.SubElement
2006
2007 a = Element('a')
2008 b = SubElement(a, 'b')
2009 c = SubElement(a, 'c')
2010 d = SubElement(b, 'd')
2011 self.assertEqual(
2012 [],
2013 list(a.itersiblings(tag='XXX')))
2014 self.assertEqual(
2015 [c],
2016 list(b.itersiblings(tag='c')))
2017 self.assertEqual(
2018 [c],
2019 list(b.itersiblings(tag='*')))
2020 self.assertEqual(
2021 [b],
2022 list(c.itersiblings(preceding=True, tag='b')))
2023 self.assertEqual(
2024 [],
2025 list(c.itersiblings(preceding=True, tag='c')))
2026
2028 Element = self.etree.Element
2029 SubElement = self.etree.SubElement
2030
2031 a = Element('a')
2032 b = SubElement(a, 'b')
2033 c = SubElement(a, 'c')
2034 d = SubElement(b, 'd')
2035 e = SubElement(a, 'e')
2036 self.assertEqual(
2037 [],
2038 list(a.itersiblings(tag=('XXX', 'YYY'))))
2039 self.assertEqual(
2040 [c, e],
2041 list(b.itersiblings(tag=('c', 'd', 'e'))))
2042 self.assertEqual(
2043 [b],
2044 list(c.itersiblings(preceding=True, tag=('b', 'b', 'c', 'd'))))
2045 self.assertEqual(
2046 [c, b],
2047 list(e.itersiblings(preceding=True, tag=('c', '*'))))
2048
2050 parseid = self.etree.parseid
2051 XML = self.etree.XML
2052 xml_text = _bytes('''
2053 <!DOCTYPE document [
2054 <!ELEMENT document (h1,p)*>
2055 <!ELEMENT h1 (#PCDATA)>
2056 <!ATTLIST h1 myid ID #REQUIRED>
2057 <!ELEMENT p (#PCDATA)>
2058 <!ATTLIST p someid ID #REQUIRED>
2059 ]>
2060 <document>
2061 <h1 myid="chapter1">...</h1>
2062 <p id="note1" class="note">...</p>
2063 <p>Regular paragraph.</p>
2064 <p xml:id="xmlid">XML:ID paragraph.</p>
2065 <p someid="warn1" class="warning">...</p>
2066 </document>
2067 ''')
2068
2069 tree, dic = parseid(BytesIO(xml_text))
2070 root = tree.getroot()
2071 root2 = XML(xml_text)
2072 self.assertEqual(self._writeElement(root),
2073 self._writeElement(root2))
2074 expected = {
2075 "chapter1" : root[0],
2076 "xmlid" : root[3],
2077 "warn1" : root[4]
2078 }
2079 self.assertTrue("chapter1" in dic)
2080 self.assertTrue("warn1" in dic)
2081 self.assertTrue("xmlid" in dic)
2082 self._checkIDDict(dic, expected)
2083
2085 XMLDTDID = self.etree.XMLDTDID
2086 XML = self.etree.XML
2087 xml_text = _bytes('''
2088 <!DOCTYPE document [
2089 <!ELEMENT document (h1,p)*>
2090 <!ELEMENT h1 (#PCDATA)>
2091 <!ATTLIST h1 myid ID #REQUIRED>
2092 <!ELEMENT p (#PCDATA)>
2093 <!ATTLIST p someid ID #REQUIRED>
2094 ]>
2095 <document>
2096 <h1 myid="chapter1">...</h1>
2097 <p id="note1" class="note">...</p>
2098 <p>Regular paragraph.</p>
2099 <p xml:id="xmlid">XML:ID paragraph.</p>
2100 <p someid="warn1" class="warning">...</p>
2101 </document>
2102 ''')
2103
2104 root, dic = XMLDTDID(xml_text)
2105 root2 = XML(xml_text)
2106 self.assertEqual(self._writeElement(root),
2107 self._writeElement(root2))
2108 expected = {
2109 "chapter1" : root[0],
2110 "xmlid" : root[3],
2111 "warn1" : root[4]
2112 }
2113 self.assertTrue("chapter1" in dic)
2114 self.assertTrue("warn1" in dic)
2115 self.assertTrue("xmlid" in dic)
2116 self._checkIDDict(dic, expected)
2117
2119 XMLDTDID = self.etree.XMLDTDID
2120 XML = self.etree.XML
2121 xml_text = _bytes('''
2122 <document>
2123 <h1 myid="chapter1">...</h1>
2124 <p id="note1" class="note">...</p>
2125 <p>Regular paragraph.</p>
2126 <p someid="warn1" class="warning">...</p>
2127 </document>
2128 ''')
2129
2130 root, dic = XMLDTDID(xml_text)
2131 root2 = XML(xml_text)
2132 self.assertEqual(self._writeElement(root),
2133 self._writeElement(root2))
2134 expected = {}
2135 self._checkIDDict(dic, expected)
2136
2138 self.assertEqual(len(dic),
2139 len(expected))
2140 self.assertEqual(sorted(dic.items()),
2141 sorted(expected.items()))
2142 if sys.version_info < (3,):
2143 self.assertEqual(sorted(dic.iteritems()),
2144 sorted(expected.iteritems()))
2145 self.assertEqual(sorted(dic.keys()),
2146 sorted(expected.keys()))
2147 if sys.version_info < (3,):
2148 self.assertEqual(sorted(dic.iterkeys()),
2149 sorted(expected.iterkeys()))
2150 if sys.version_info < (3,):
2151 self.assertEqual(sorted(dic.values()),
2152 sorted(expected.values()))
2153 self.assertEqual(sorted(dic.itervalues()),
2154 sorted(expected.itervalues()))
2155
2157 etree = self.etree
2158
2159 r = {'foo': 'http://ns.infrae.com/foo'}
2160 e = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=r)
2161 self.assertEqual(
2162 'foo',
2163 e.prefix)
2164 self.assertEqual(
2165 _bytes('<foo:bar xmlns:foo="http://ns.infrae.com/foo"></foo:bar>'),
2166 self._writeElement(e))
2167
2169 etree = self.etree
2170
2171 r = {None: 'http://ns.infrae.com/foo'}
2172 e = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=r)
2173 self.assertEqual(
2174 None,
2175 e.prefix)
2176 self.assertEqual(
2177 '{http://ns.infrae.com/foo}bar',
2178 e.tag)
2179 self.assertEqual(
2180 _bytes('<bar xmlns="http://ns.infrae.com/foo"></bar>'),
2181 self._writeElement(e))
2182
2184 etree = self.etree
2185
2186 r = {None: 'http://ns.infrae.com/foo',
2187 'hoi': 'http://ns.infrae.com/hoi'}
2188 e = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=r)
2189 e.set('{http://ns.infrae.com/hoi}test', 'value')
2190 self.assertEqual(
2191 _bytes('<bar xmlns="http://ns.infrae.com/foo" xmlns:hoi="http://ns.infrae.com/hoi" hoi:test="value"></bar>'),
2192 self._writeElement(e))
2193
2195 etree = self.etree
2196
2197 root = etree.Element('{http://test/ns}root',
2198 nsmap={None: 'http://test/ns'})
2199 sub = etree.Element('{http://test/ns}sub',
2200 nsmap={'test': 'http://test/ns'})
2201
2202 sub.attrib['{http://test/ns}attr'] = 'value'
2203 self.assertEqual(sub.attrib['{http://test/ns}attr'], 'value')
2204 self.assertEqual(
2205 _bytes('<test:sub xmlns:test="http://test/ns" test:attr="value"/>'),
2206 etree.tostring(sub))
2207
2208 root.append(sub)
2209 self.assertEqual(
2210 _bytes('<root xmlns="http://test/ns">'
2211 '<sub xmlns:test="http://test/ns" test:attr="value"/>'
2212 '</root>'),
2213 etree.tostring(root))
2214
2216 etree = self.etree
2217
2218 root = etree.Element('root')
2219 sub = etree.Element('{http://test/ns}sub',
2220 nsmap={'test': 'http://test/ns'})
2221
2222 sub.attrib['{http://test/ns}attr'] = 'value'
2223 self.assertEqual(sub.attrib['{http://test/ns}attr'], 'value')
2224 self.assertEqual(
2225 _bytes('<test:sub xmlns:test="http://test/ns" test:attr="value"/>'),
2226 etree.tostring(sub))
2227
2228 root.append(sub)
2229 self.assertEqual(
2230 _bytes('<root>'
2231 '<test:sub xmlns:test="http://test/ns" test:attr="value"/>'
2232 '</root>'),
2233 etree.tostring(root))
2234
2236 etree = self.etree
2237
2238 root = etree.Element('root')
2239 sub = etree.Element('{http://test/ns}sub',
2240 nsmap={None: 'http://test/ns'})
2241
2242 sub.attrib['{http://test/ns}attr'] = 'value'
2243 self.assertEqual(sub.attrib['{http://test/ns}attr'], 'value')
2244 self.assertEqual(
2245 _bytes('<sub xmlns="http://test/ns" '
2246 'xmlns:ns0="http://test/ns" ns0:attr="value"/>'),
2247 etree.tostring(sub))
2248
2249 root.append(sub)
2250 self.assertEqual(
2251 _bytes('<root>'
2252 '<sub xmlns="http://test/ns"'
2253 ' xmlns:ns0="http://test/ns" ns0:attr="value"/>'
2254 '</root>'),
2255 etree.tostring(root))
2256
2258 etree = self.etree
2259
2260 root = etree.Element('{http://test/ns}root',
2261 nsmap={'test': 'http://test/ns',
2262 None: 'http://test/ns'})
2263 sub = etree.Element('{http://test/ns}sub',
2264 nsmap={None: 'http://test/ns'})
2265
2266 sub.attrib['{http://test/ns}attr'] = 'value'
2267 self.assertEqual(sub.attrib['{http://test/ns}attr'], 'value')
2268 self.assertEqual(
2269 _bytes('<sub xmlns="http://test/ns" '
2270 'xmlns:ns0="http://test/ns" ns0:attr="value"/>'),
2271 etree.tostring(sub))
2272
2273 root.append(sub)
2274 self.assertEqual(
2275 _bytes('<test:root xmlns:test="http://test/ns" xmlns="http://test/ns">'
2276 '<test:sub test:attr="value"/>'
2277 '</test:root>'),
2278 etree.tostring(root))
2279
2281 etree = self.etree
2282 r = {None: 'http://ns.infrae.com/foo',
2283 'hoi': 'http://ns.infrae.com/hoi'}
2284 e = etree.Element('{http://ns.infrae.com/foo}z', nsmap=r)
2285 tree = etree.ElementTree(element=e)
2286 etree.SubElement(e, '{http://ns.infrae.com/hoi}x')
2287 self.assertEqual(
2288 _bytes('<z xmlns="http://ns.infrae.com/foo" xmlns:hoi="http://ns.infrae.com/hoi"><hoi:x></hoi:x></z>'),
2289 self._writeElement(e))
2290
2292 etree = self.etree
2293
2294 r = {None: 'http://ns.infrae.com/foo'}
2295 e1 = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=r)
2296 e2 = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=r)
2297
2298 e1.append(e2)
2299
2300 self.assertEqual(
2301 None,
2302 e1.prefix)
2303 self.assertEqual(
2304 None,
2305 e1[0].prefix)
2306 self.assertEqual(
2307 '{http://ns.infrae.com/foo}bar',
2308 e1.tag)
2309 self.assertEqual(
2310 '{http://ns.infrae.com/foo}bar',
2311 e1[0].tag)
2312
2314 etree = self.etree
2315
2316 r = {None: 'http://ns.infrae.com/BAR'}
2317 e1 = etree.Element('{http://ns.infrae.com/BAR}bar', nsmap=r)
2318 e2 = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=r)
2319
2320 e1.append(e2)
2321
2322 self.assertEqual(
2323 None,
2324 e1.prefix)
2325 self.assertNotEqual(
2326 None,
2327 e2.prefix)
2328 self.assertEqual(
2329 '{http://ns.infrae.com/BAR}bar',
2330 e1.tag)
2331 self.assertEqual(
2332 '{http://ns.infrae.com/foo}bar',
2333 e2.tag)
2334
2336 ns_href = "http://a.b.c"
2337 one = self.etree.fromstring(
2338 _bytes('<foo><bar xmlns:ns="%s"><ns:baz/></bar></foo>' % ns_href))
2339 baz = one[0][0]
2340
2341 two = self.etree.fromstring(
2342 _bytes('<root xmlns:ns="%s"/>' % ns_href))
2343 two.append(baz)
2344 del one # make sure the source document is deallocated
2345
2346 self.assertEqual('{%s}baz' % ns_href, baz.tag)
2347 self.assertEqual(
2348 _bytes('<root xmlns:ns="%s"><ns:baz/></root>' % ns_href),
2349 self.etree.tostring(two))
2350
2352 xml = _bytes('<foo xmlns="F" xmlns:x="x"><bar xmlns:ns="NS" xmlns:b="b" xmlns="B"><ns:baz/></bar></foo>')
2353 root = self.etree.fromstring(xml)
2354 self.assertEqual(xml,
2355 self.etree.tostring(root))
2356 self.etree.cleanup_namespaces(root)
2357 self.assertEqual(
2358 _bytes('<foo xmlns="F"><bar xmlns:ns="NS" xmlns="B"><ns:baz/></bar></foo>'),
2359 self.etree.tostring(root))
2360
2362 etree = self.etree
2363
2364 r = {None: 'http://ns.infrae.com/foo',
2365 'hoi': 'http://ns.infrae.com/hoi'}
2366 e = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=r)
2367 self.assertEqual(
2368 r,
2369 e.nsmap)
2370
2372 etree = self.etree
2373
2374 re = {None: 'http://ns.infrae.com/foo',
2375 'hoi': 'http://ns.infrae.com/hoi'}
2376 e = etree.Element('{http://ns.infrae.com/foo}bar', nsmap=re)
2377
2378 rs = {None: 'http://ns.infrae.com/honk',
2379 'top': 'http://ns.infrae.com/top'}
2380 s = etree.SubElement(e, '{http://ns.infrae.com/honk}bar', nsmap=rs)
2381
2382 r = re.copy()
2383 r.update(rs)
2384 self.assertEqual(re, e.nsmap)
2385 self.assertEqual(r, s.nsmap)
2386
2388 etree = self.etree
2389 el = etree.HTML('<hha:page-description>aa</hha:page-description>').find('.//page-description')
2390 self.assertEqual({'hha': None}, el.nsmap)
2391
2393 Element = self.etree.Element
2394 SubElement = self.etree.SubElement
2395
2396 a = Element('a')
2397 b = SubElement(a, 'b')
2398 c = SubElement(a, 'c')
2399 d = SubElement(b, 'd')
2400 e = SubElement(c, 'e')
2401 f = SubElement(c, 'f')
2402
2403 self.assertEqual(
2404 [a, b],
2405 list(a.getiterator('a', 'b')))
2406 self.assertEqual(
2407 [],
2408 list(a.getiterator('x', 'y')))
2409 self.assertEqual(
2410 [a, f],
2411 list(a.getiterator('f', 'a')))
2412 self.assertEqual(
2413 [c, e, f],
2414 list(c.getiterator('c', '*', 'a')))
2415 self.assertEqual(
2416 [],
2417 list(a.getiterator( (), () )))
2418
2420 Element = self.etree.Element
2421 SubElement = self.etree.SubElement
2422
2423 a = Element('a')
2424 b = SubElement(a, 'b')
2425 c = SubElement(a, 'c')
2426 d = SubElement(b, 'd')
2427 e = SubElement(c, 'e')
2428 f = SubElement(c, 'f')
2429
2430 self.assertEqual(
2431 [a, b],
2432 list(a.getiterator( ('a', 'b') )))
2433 self.assertEqual(
2434 [],
2435 list(a.getiterator( ('x', 'y') )))
2436 self.assertEqual(
2437 [a, f],
2438 list(a.getiterator( ('f', 'a') )))
2439 self.assertEqual(
2440 [c, e, f],
2441 list(c.getiterator( ('c', '*', 'a') )))
2442 self.assertEqual(
2443 [],
2444 list(a.getiterator( () )))
2445
2447 Element = self.etree.Element
2448 SubElement = self.etree.SubElement
2449
2450 a = Element('{a}a')
2451 b = SubElement(a, '{a}b')
2452 c = SubElement(a, '{a}c')
2453 d = SubElement(b, '{b}d')
2454 e = SubElement(c, '{a}e')
2455 f = SubElement(c, '{b}f')
2456 g = SubElement(c, 'g')
2457
2458 self.assertEqual(
2459 [a],
2460 list(a.getiterator('{a}a')))
2461 self.assertEqual(
2462 [],
2463 list(a.getiterator('{b}a')))
2464 self.assertEqual(
2465 [],
2466 list(a.getiterator('a')))
2467 self.assertEqual(
2468 [a,b,d,c,e,f,g],
2469 list(a.getiterator('*')))
2470 self.assertEqual(
2471 [f],
2472 list(c.getiterator('{b}*')))
2473 self.assertEqual(
2474 [d, f],
2475 list(a.getiterator('{b}*')))
2476 self.assertEqual(
2477 [g],
2478 list(a.getiterator('g')))
2479 self.assertEqual(
2480 [g],
2481 list(a.getiterator('{}g')))
2482 self.assertEqual(
2483 [g],
2484 list(a.getiterator('{}*')))
2485
2487 Element = self.etree.Element
2488 SubElement = self.etree.SubElement
2489
2490 a = Element('{a}a')
2491 b = SubElement(a, '{nsA}b')
2492 c = SubElement(b, '{nsB}b')
2493 d = SubElement(a, 'b')
2494 e = SubElement(a, '{nsA}e')
2495 f = SubElement(e, '{nsB}e')
2496 g = SubElement(e, 'e')
2497
2498 self.assertEqual(
2499 [b, c, d],
2500 list(a.getiterator('{*}b')))
2501 self.assertEqual(
2502 [e, f, g],
2503 list(a.getiterator('{*}e')))
2504 self.assertEqual(
2505 [a, b, c, d, e, f, g],
2506 list(a.getiterator('{*}*')))
2507
2509 Element = self.etree.Element
2510 Entity = self.etree.Entity
2511 SubElement = self.etree.SubElement
2512
2513 a = Element('a')
2514 b = SubElement(a, 'b')
2515 entity_b = Entity("TEST-b")
2516 b.append(entity_b)
2517
2518 self.assertEqual(
2519 [entity_b],
2520 list(a.getiterator(Entity)))
2521
2522 entity_a = Entity("TEST-a")
2523 a.append(entity_a)
2524
2525 self.assertEqual(
2526 [entity_b, entity_a],
2527 list(a.getiterator(Entity)))
2528
2529 self.assertEqual(
2530 [entity_b],
2531 list(b.getiterator(Entity)))
2532
2534 Element = self.etree.Element
2535 Comment = self.etree.Comment
2536 PI = self.etree.PI
2537 SubElement = self.etree.SubElement
2538
2539 a = Element('a')
2540 b = SubElement(a, 'b')
2541 a.append(Comment("test"))
2542 a.append(PI("pi", "content"))
2543 c = SubElement(a, 'c')
2544
2545 self.assertEqual(
2546 [a, b, c],
2547 list(a.getiterator(Element)))
2548
2550 # ElementTree iterates over everything here
2551 Element = self.etree.Element
2552 Comment = self.etree.Comment
2553 PI = self.etree.PI
2554 SubElement = self.etree.SubElement
2555
2556 a = Element('a')
2557 b = SubElement(a, 'b')
2558 a.append(Comment("test"))
2559 a.append(PI("pi", "content"))
2560 c = SubElement(a, 'c')
2561
2562 self.assertEqual(
2563 [a, b, c],
2564 list(a.getiterator('*')))
2565
2567 XML = self.etree.XML
2568 ElementTree = self.etree.ElementTree
2569 QName = self.etree.QName
2570 tree = ElementTree(XML(_bytes('<a><b><c/></b><b/><c><b/></c></a>')))
2571 self.assertEqual(tree.find(QName("c")), tree.getroot()[2])
2572
2574 XML = self.etree.XML
2575 ElementTree = self.etree.ElementTree
2576 QName = self.etree.QName
2577 tree = ElementTree(XML(_bytes('<a><b><c/></b><b/><c><b/></c></a>')))
2578 self.assertEqual(len(list(tree.findall(QName("c")))), 1)
2579
2581 XML = self.etree.XML
2582 ElementTree = self.etree.ElementTree
2583 QName = self.etree.QName
2584 tree = ElementTree(XML(
2585 _bytes('<a xmlns:x="X" xmlns:y="Y"><x:b><c/></x:b><b/><c><x:b/><b/></c><b/></a>')))
2586 self.assertEqual(len(list(tree.findall(QName("b")))), 2)
2587 self.assertEqual(len(list(tree.findall(QName("X", "b")))), 1)
2588
2590 XML = self.etree.XML
2591 root = XML(_bytes('<a xmlns:x="X" xmlns:y="Y"><x:b><c/></x:b><b/><c><x:b/><b/></c><b/></a>'))
2592 self.assertEqual(len(root.findall(".//{X}b")), 2)
2593 self.assertEqual(len(root.findall(".//{X}*")), 2)
2594 self.assertEqual(len(root.findall(".//b")), 3)
2595
2597 XML = self.etree.XML
2598 root = XML(_bytes('<a xmlns:x="X" xmlns:y="Y"><x:b><c/></x:b><b/><c><x:b/><b/></c><y:b/></a>'))
2599 nsmap = {'xx': 'X'}
2600 self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 2)
2601 self.assertEqual(len(root.findall(".//xx:*", namespaces=nsmap)), 2)
2602 self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2)
2603 nsmap = {'xx': 'Y'}
2604 self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 1)
2605 self.assertEqual(len(root.findall(".//xx:*", namespaces=nsmap)), 1)
2606 self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2)
2607
2609 XML = self.etree.XML
2610 root = XML(_bytes('<a xmlns:x="X" xmlns:y="Y"><x:b><c/></x:b><b/><c><x:b/><b/></c><y:b/></a>'))
2611 nsmap = {'xx': 'X'}
2612 self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 2)
2613 self.assertEqual(len(root.findall(".//xx:*", namespaces=nsmap)), 2)
2614 self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2)
2615 nsmap = {'xx': 'Y'}
2616 self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 1)
2617 self.assertEqual(len(root.findall(".//xx:*", namespaces=nsmap)), 1)
2618 self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2)
2619
2621 XML = self.etree.XML
2622 root = XML(_bytes('<a><b><c/></b><b/><c><b/><b/></c><b/></a>'))
2623 self.assertRaises(SyntaxError, root.findall, '')
2624 self.assertRaises(SyntaxError, root.findall, '//') # absolute path on Element
2625 self.assertRaises(SyntaxError, root.findall, './//')
2626
2628 etree = self.etree
2629 e = etree.Element('foo')
2630 for i in range(10):
2631 etree.SubElement(e, 'a%s' % i)
2632 for i in range(10):
2633 self.assertEqual(
2634 i,
2635 e.index(e[i]))
2636 self.assertEqual(
2637 3, e.index(e[3], 3))
2638 self.assertRaises(
2639 ValueError, e.index, e[3], 4)
2640 self.assertRaises(
2641 ValueError, e.index, e[3], 0, 2)
2642 self.assertRaises(
2643 ValueError, e.index, e[8], 0, -3)
2644 self.assertRaises(
2645 ValueError, e.index, e[8], -5, -3)
2646 self.assertEqual(
2647 8, e.index(e[8], 0, -1))
2648 self.assertEqual(
2649 8, e.index(e[8], -12, -1))
2650 self.assertEqual(
2651 0, e.index(e[0], -12, -1))
2652
2654 etree = self.etree
2655 e = etree.Element('foo')
2656 for i in range(10):
2657 el = etree.SubElement(e, 'a%s' % i)
2658 el.text = "text%d" % i
2659 el.tail = "tail%d" % i
2660
2661 child0 = e[0]
2662 child1 = e[1]
2663 child2 = e[2]
2664
2665 e.replace(e[0], e[1])
2666 self.assertEqual(
2667 9, len(e))
2668 self.assertEqual(
2669 child1, e[0])
2670 self.assertEqual(
2671 child1.text, "text1")
2672 self.assertEqual(
2673 child1.tail, "tail1")
2674 self.assertEqual(
2675 child0.tail, "tail0")
2676 self.assertEqual(
2677 child2, e[1])
2678
2679 e.replace(e[-1], e[0])
2680 self.assertEqual(
2681 child1, e[-1])
2682 self.assertEqual(
2683 child1.text, "text1")
2684 self.assertEqual(
2685 child1.tail, "tail1")
2686 self.assertEqual(
2687 child2, e[0])
2688
2690 etree = self.etree
2691 e = etree.Element('foo')
2692 for i in range(10):
2693 etree.SubElement(e, 'a%s' % i)
2694
2695 new_element = etree.Element("test")
2696 new_element.text = "TESTTEXT"
2697 new_element.tail = "TESTTAIL"
2698 child1 = e[1]
2699 e.replace(e[0], new_element)
2700 self.assertEqual(
2701 new_element, e[0])
2702 self.assertEqual(
2703 "TESTTEXT",
2704 e[0].text)
2705 self.assertEqual(
2706 "TESTTAIL",
2707 e[0].tail)
2708 self.assertEqual(
2709 child1, e[1])
2710
2712 Element = self.etree.Element
2713 SubElement = self.etree.SubElement
2714
2715 a = Element('a')
2716
2717 e = Element('e')
2718 f = Element('f')
2719 g = Element('g')
2720
2721 s = [e, f, g]
2722 a[::-1] = s
2723 self.assertEqual(
2724 [g, f, e],
2725 list(a))
2726
2728 Element = self.etree.Element
2729 SubElement = self.etree.SubElement
2730
2731 a = Element('a')
2732 b = SubElement(a, 'b')
2733 c = SubElement(a, 'c')
2734 d = SubElement(a, 'd')
2735 e = SubElement(a, 'e')
2736
2737 x = Element('x')
2738 y = Element('y')
2739
2740 a[1::2] = [x, y]
2741 self.assertEqual(
2742 [b, x, d, y],
2743 list(a))
2744
2746 Element = self.etree.Element
2747 SubElement = self.etree.SubElement
2748
2749 a = Element('a')
2750 b = SubElement(a, 'b')
2751 c = SubElement(a, 'c')
2752 d = SubElement(a, 'd')
2753 e = SubElement(a, 'e')
2754
2755 x = Element('x')
2756 y = Element('y')
2757
2758 a[1::-1] = [x, y]
2759 self.assertEqual(
2760 [y, x, d, e],
2761 list(a))
2762
2764 Element = self.etree.Element
2765 SubElement = self.etree.SubElement
2766
2767 a = Element('a')
2768 b = SubElement(a, 'b')
2769 c = SubElement(a, 'c')
2770 d = SubElement(a, 'd')
2771 e = SubElement(a, 'e')
2772
2773 x = Element('x')
2774 y = Element('y')
2775
2776 a[::-2] = [x, y]
2777 self.assertEqual(
2778 [b, y, d, x],
2779 list(a))
2780
2782 Element = self.etree.Element
2783 SubElement = self.etree.SubElement
2784 try:
2785 slice
2786 except NameError:
2787 print("slice() not found")
2788 return
2789
2790 a = Element('a')
2791 b = SubElement(a, 'b')
2792 c = SubElement(a, 'c')
2793 d = SubElement(a, 'd')
2794 e = SubElement(a, 'e')
2795
2796 x = Element('x')
2797 y = Element('y')
2798 z = Element('z')
2799
2800 self.assertRaises(
2801 ValueError,
2802 operator.setitem, a, slice(1,None,2), [x, y, z])
2803
2804 self.assertEqual(
2805 [b, c, d, e],
2806 list(a))
2807
2809 XML = self.etree.XML
2810 root = XML(_bytes('''<?xml version="1.0"?>
2811 <root><test>
2812
2813 <bla/></test>
2814 </root>
2815 '''))
2816
2817 self.assertEqual(
2818 [2, 2, 4],
2819 [ el.sourceline for el in root.getiterator() ])
2820
2822 parse = self.etree.parse
2823 tree = parse(fileInTestDir('include/test_xinclude.xml'))
2824
2825 self.assertEqual(
2826 [1, 2, 3],
2827 [ el.sourceline for el in tree.getiterator() ])
2828
2830 iterparse = self.etree.iterparse
2831 lines = [ el.sourceline for (event, el) in
2832 iterparse(fileInTestDir('include/test_xinclude.xml')) ]
2833
2834 self.assertEqual(
2835 [2, 3, 1],
2836 lines)
2837
2839 iterparse = self.etree.iterparse
2840 lines = [ el.sourceline for (event, el) in
2841 iterparse(fileInTestDir('include/test_xinclude.xml'),
2842 events=("start",)) ]
2843
2844 self.assertEqual(
2845 [1, 2, 3],
2846 lines)
2847
2849 Element = self.etree.Element
2850 SubElement = self.etree.SubElement
2851 el = Element("test")
2852 self.assertEqual(None, el.sourceline)
2853
2854 child = SubElement(el, "test")
2855 self.assertEqual(None, el.sourceline)
2856 self.assertEqual(None, child.sourceline)
2857
2859 etree = self.etree
2860 root = etree.XML(_bytes("<root/>"), base_url="http://no/such/url")
2861 docinfo = root.getroottree().docinfo
2862 self.assertEqual(docinfo.URL, "http://no/such/url")
2863
2865 etree = self.etree
2866 root = etree.XML(_bytes("<root/>"), base_url="http://no/such/url")
2867 docinfo = root.getroottree().docinfo
2868 self.assertEqual(docinfo.URL, "http://no/such/url")
2869 docinfo.URL = "https://secret/url"
2870 self.assertEqual(docinfo.URL, "https://secret/url")
2871
2873 etree = self.etree
2874 tree = etree.parse(BytesIO("<root/>"), base_url="http://no/such/url")
2875 docinfo = tree.docinfo
2876 self.assertEqual(docinfo.URL, "http://no/such/url")
2877
2879 etree = self.etree
2880 tree = etree.parse(fileInTestDir('include/test_xinclude.xml'),
2881 base_url="http://no/such/url")
2882 docinfo = tree.docinfo
2883 self.assertEqual(docinfo.URL, "http://no/such/url")
2884
2886 etree = self.etree
2887 root = etree.HTML(_bytes("<html/>"), base_url="http://no/such/url")
2888 docinfo = root.getroottree().docinfo
2889 self.assertEqual(docinfo.URL, "http://no/such/url")
2890
2892 etree = self.etree
2893 xml_header = '<?xml version="1.0" encoding="ascii"?>'
2894 pub_id = "-//W3C//DTD XHTML 1.0 Transitional//EN"
2895 sys_id = "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"
2896 doctype_string = '<!DOCTYPE html PUBLIC "%s" "%s">' % (pub_id, sys_id)
2897
2898 xml = _bytes(xml_header + doctype_string + '<html><body></body></html>')
2899
2900 tree = etree.parse(BytesIO(xml))
2901 docinfo = tree.docinfo
2902 self.assertEqual(docinfo.encoding, "ascii")
2903 self.assertEqual(docinfo.xml_version, "1.0")
2904 self.assertEqual(docinfo.public_id, pub_id)
2905 self.assertEqual(docinfo.system_url, sys_id)
2906 self.assertEqual(docinfo.root_name, 'html')
2907 self.assertEqual(docinfo.doctype, doctype_string)
2908
2910 etree = self.etree
2911 xml_header = '<?xml version="1.0" encoding="UTF-8"?>'
2912 sys_id = "some.dtd"
2913 doctype_string = '<!DOCTYPE html SYSTEM "%s">' % sys_id
2914 xml = _bytes(xml_header + doctype_string + '<html><body></body></html>')
2915
2916 tree = etree.parse(BytesIO(xml))
2917 docinfo = tree.docinfo
2918 self.assertEqual(docinfo.encoding, "UTF-8")
2919 self.assertEqual(docinfo.xml_version, "1.0")
2920 self.assertEqual(docinfo.public_id, None)
2921 self.assertEqual(docinfo.system_url, sys_id)
2922 self.assertEqual(docinfo.root_name, 'html')
2923 self.assertEqual(docinfo.doctype, doctype_string)
2924
2926 etree = self.etree
2927 xml = _bytes('<html><body></body></html>')
2928 tree = etree.parse(BytesIO(xml))
2929 docinfo = tree.docinfo
2930 self.assertEqual(docinfo.encoding, "UTF-8")
2931 self.assertEqual(docinfo.xml_version, "1.0")
2932 self.assertEqual(docinfo.public_id, None)
2933 self.assertEqual(docinfo.system_url, None)
2934 self.assertEqual(docinfo.root_name, 'html')
2935 self.assertEqual(docinfo.doctype, '')
2936
2938 etree = self.etree
2939 xml = _bytes('<!DOCTYPE root><root></root>')
2940 tree = etree.parse(BytesIO(xml))
2941 docinfo = tree.docinfo
2942 self.assertEqual(docinfo.encoding, "UTF-8")
2943 self.assertEqual(docinfo.xml_version, "1.0")
2944 self.assertEqual(docinfo.public_id, None)
2945 self.assertEqual(docinfo.system_url, None)
2946 self.assertEqual(docinfo.root_name, 'root')
2947 self.assertEqual(docinfo.doctype, '<!DOCTYPE root>')
2948
2950 etree = self.etree
2951 xml = _bytes('<!DOCTYPE root>\n<root/>')
2952 tree = etree.parse(BytesIO(xml))
2953 self.assertEqual(xml, etree.tostring(tree))
2954
2956 etree = self.etree
2957 pub_id = "-//W3C//DTD XHTML 1.0 Transitional//EN"
2958 sys_id = "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"
2959 doctype_string = _bytes('<!DOCTYPE html PUBLIC "%s" "%s">' % (pub_id, sys_id))
2960
2961 xml = _bytes('<!DOCTYPE root>\n<root/>')
2962 tree = etree.parse(BytesIO(xml))
2963 self.assertEqual(xml.replace(_bytes('<!DOCTYPE root>'), doctype_string),
2964 etree.tostring(tree, doctype=doctype_string))
2965
2967 etree = self.etree
2968 root = etree.XML(_bytes("<root/>"), base_url="http://no/such/url")
2969 self.assertEqual(root.base, "http://no/such/url")
2970 self.assertEqual(
2971 root.get('{http://www.w3.org/XML/1998/namespace}base'), None)
2972 root.base = "https://secret/url"
2973 self.assertEqual(root.base, "https://secret/url")
2974 self.assertEqual(
2975 root.get('{http://www.w3.org/XML/1998/namespace}base'),
2976 "https://secret/url")
2977
2979 etree = self.etree
2980 root = etree.XML(_bytes("<root/>"), base_url="http://no/such/url")
2981 self.assertEqual(root.base, "http://no/such/url")
2982 self.assertEqual(
2983 root.get('{http://www.w3.org/XML/1998/namespace}base'), None)
2984 root.set('{http://www.w3.org/XML/1998/namespace}base',
2985 "https://secret/url")
2986 self.assertEqual(root.base, "https://secret/url")
2987 self.assertEqual(
2988 root.get('{http://www.w3.org/XML/1998/namespace}base'),
2989 "https://secret/url")
2990
2992 etree = self.etree
2993 root = etree.HTML(_bytes("<html><body></body></html>"),
2994 base_url="http://no/such/url")
2995 self.assertEqual(root.base, "http://no/such/url")
2996
2998 etree = self.etree
2999 root = etree.HTML(_bytes('<html><head><base href="http://no/such/url"></head></html>'))
3000 self.assertEqual(root.base, "http://no/such/url")
3001
3003 # parse from a file object that returns unicode strings
3004 f = LargeFileLikeUnicode()
3005 tree = self.etree.parse(f)
3006 root = tree.getroot()
3007 self.assertTrue(root.tag.endswith('root'))
3008
3010 # check that DTDs that go in also go back out
3011 xml = _bytes('''\
3012 <!DOCTYPE test SYSTEM "test.dtd" [
3013 <!ENTITY entity "tasty">
3014 <!ELEMENT test (a)>
3015 <!ELEMENT a (#PCDATA)>
3016 ]>
3017 <test><a>test-test</a></test>\
3018 ''')
3019 tree = self.etree.parse(BytesIO(xml))
3020 self.assertEqual(self.etree.tostring(tree).replace(_bytes(" "), _bytes("")),
3021 xml.replace(_bytes(" "), _bytes("")))
3022
3024 Element = self.etree.Element
3025
3026 a = Element('a')
3027 self.assertRaises(ValueError, setattr, a, "text", 'ha\0ho')
3028 self.assertRaises(ValueError, setattr, a, "tail", 'ha\0ho')
3029
3030 self.assertRaises(ValueError, Element, 'ha\0ho')
3031
3033 Element = self.etree.Element
3034
3035 a = Element('a')
3036 self.assertRaises(ValueError, setattr, a, "text",
3037 _str('ha\0ho'))
3038 self.assertRaises(ValueError, setattr, a, "tail",
3039 _str('ha\0ho'))
3040
3041 self.assertRaises(ValueError, Element,
3042 _str('ha\0ho'))
3043
3045 Element = self.etree.Element
3046
3047 a = Element('a')
3048 self.assertRaises(ValueError, setattr, a, "text", 'ha\x07ho')
3049 self.assertRaises(ValueError, setattr, a, "text", 'ha\x02ho')
3050
3051 self.assertRaises(ValueError, setattr, a, "tail", 'ha\x07ho')
3052 self.assertRaises(ValueError, setattr, a, "tail", 'ha\x02ho')
3053
3054 self.assertRaises(ValueError, Element, 'ha\x07ho')
3055 self.assertRaises(ValueError, Element, 'ha\x02ho')
3056
3058 Element = self.etree.Element
3059
3060 a = Element('a')
3061 self.assertRaises(ValueError, setattr, a, "text",
3062 _str('ha\x07ho'))
3063 self.assertRaises(ValueError, setattr, a, "text",
3064 _str('ha\x02ho'))
3065
3066 self.assertRaises(ValueError, setattr, a, "tail",
3067 _str('ha\x07ho'))
3068 self.assertRaises(ValueError, setattr, a, "tail",
3069 _str('ha\x02ho'))
3070
3071 self.assertRaises(ValueError, Element,
3072 _str('ha\x07ho'))
3073 self.assertRaises(ValueError, Element,
3074 _str('ha\x02ho'))
3075
3077 Element = self.etree.Element
3078
3079 a = Element('a')
3080 self.assertRaises(ValueError, setattr, a, "text",
3081 _str('ha\u1234\x07ho'))
3082 self.assertRaises(ValueError, setattr, a, "text",
3083 _str('ha\u1234\x02ho'))
3084
3085 self.assertRaises(ValueError, setattr, a, "tail",
3086 _str('ha\u1234\x07ho'))
3087 self.assertRaises(ValueError, setattr, a, "tail",
3088 _str('ha\u1234\x02ho'))
3089
3090 self.assertRaises(ValueError, Element,
3091 _str('ha\u1234\x07ho'))
3092 self.assertRaises(ValueError, Element,
3093 _str('ha\u1234\x02ho'))
3094
3096 # ElementTree fails to serialize this
3097 tostring = self.etree.tostring
3098 Element = self.etree.Element
3099 SubElement = self.etree.SubElement
3100
3101 a = Element('a')
3102 b = SubElement(a, 'b')
3103 c = SubElement(a, 'c')
3104
3105 result = tostring(a, encoding='UTF-16')
3106 self.assertEqual(_bytes('<a><b></b><c></c></a>'),
3107 canonicalize(result))
3108
3110 # ElementTree raises an AssertionError here
3111 tostring = self.etree.tostring
3112 self.assertRaises(TypeError, self.etree.tostring, None)
3113
3115 tostring = self.etree.tostring
3116 Element = self.etree.Element
3117 SubElement = self.etree.SubElement
3118
3119 a = Element('a')
3120 b = SubElement(a, 'b')
3121 c = SubElement(a, 'c')
3122
3123 result = tostring(a)
3124 self.assertEqual(result, _bytes("<a><b/><c/></a>"))
3125
3126 result = tostring(a, pretty_print=False)
3127 self.assertEqual(result, _bytes("<a><b/><c/></a>"))
3128
3129 result = tostring(a, pretty_print=True)
3130 self.assertEqual(result, _bytes("<a>\n <b/>\n <c/>\n</a>\n"))
3131
3133 tostring = self.etree.tostring
3134 Element = self.etree.Element
3135 SubElement = self.etree.SubElement
3136
3137 a = Element('a')
3138 a.tail = "aTAIL"
3139 b = SubElement(a, 'b')
3140 b.tail = "bTAIL"
3141 c = SubElement(a, 'c')
3142
3143 result = tostring(a)
3144 self.assertEqual(result, _bytes("<a><b/>bTAIL<c/></a>aTAIL"))
3145
3146 result = tostring(a, with_tail=False)
3147 self.assertEqual(result, _bytes("<a><b/>bTAIL<c/></a>"))
3148
3149 result = tostring(a, with_tail=True)
3150 self.assertEqual(result, _bytes("<a><b/>bTAIL<c/></a>aTAIL"))
3151
3153 tostring = self.etree.tostring
3154 XML = self.etree.XML
3155 ElementTree = self.etree.ElementTree
3156 Element = self.etree.Element
3157
3158 tree = Element("root").getroottree()
3159 self.assertEqual(None, tree.docinfo.standalone)
3160
3161 tree = XML(_bytes("<root/>")).getroottree()
3162 self.assertEqual(None, tree.docinfo.standalone)
3163
3164 tree = XML(_bytes(
3165 "<?xml version='1.0' encoding='ASCII' standalone='yes'?>\n<root/>"
3166 )).getroottree()
3167 self.assertEqual(True, tree.docinfo.standalone)
3168
3169 tree = XML(_bytes(
3170 "<?xml version='1.0' encoding='ASCII' standalone='no'?>\n<root/>"
3171 )).getroottree()
3172 self.assertEqual(False, tree.docinfo.standalone)
3173
3175 tostring = self.etree.tostring
3176 XML = self.etree.XML
3177 ElementTree = self.etree.ElementTree
3178
3179 root = XML(_bytes("<root/>"))
3180
3181 tree = ElementTree(root)
3182 self.assertEqual(None, tree.docinfo.standalone)
3183
3184 result = tostring(root, xml_declaration=True, encoding="ASCII")
3185 self.assertEqual(result, _bytes(
3186 "<?xml version='1.0' encoding='ASCII'?>\n<root/>"))
3187
3188 result = tostring(root, xml_declaration=True, encoding="ASCII",
3189 standalone=True)
3190 self.assertEqual(result, _bytes(
3191 "<?xml version='1.0' encoding='ASCII' standalone='yes'?>\n<root/>"))
3192
3193 tree = ElementTree(XML(result))
3194 self.assertEqual(True, tree.docinfo.standalone)
3195
3196 result = tostring(root, xml_declaration=True, encoding="ASCII",
3197 standalone=False)
3198 self.assertEqual(result, _bytes(
3199 "<?xml version='1.0' encoding='ASCII' standalone='no'?>\n<root/>"))
3200
3201 tree = ElementTree(XML(result))
3202 self.assertEqual(False, tree.docinfo.standalone)
3203
3205 tostring = self.etree.tostring
3206 XML = self.etree.XML
3207 ElementTree = self.etree.ElementTree
3208
3209 root = XML(_bytes(
3210 "<?xml version='1.0' encoding='UTF-8' standalone='yes'?>\n<root/>"))
3211
3212 tree = ElementTree(root)
3213 self.assertEqual(True, tree.docinfo.standalone)
3214
3215 result = tostring(root, xml_declaration=True, encoding="ASCII")
3216 self.assertEqual(result, _bytes(
3217 "<?xml version='1.0' encoding='ASCII'?>\n<root/>"))
3218
3219 result = tostring(root, xml_declaration=True, encoding="ASCII",
3220 standalone=True)
3221 self.assertEqual(result, _bytes(
3222 "<?xml version='1.0' encoding='ASCII' standalone='yes'?>\n<root/>"))
3223
3225 tostring = self.etree.tostring
3226 Element = self.etree.Element
3227 SubElement = self.etree.SubElement
3228
3229 a = Element('a')
3230 a.text = "A"
3231 a.tail = "tail"
3232 b = SubElement(a, 'b')
3233 b.text = "B"
3234 b.tail = _str("Søk på nettet")
3235 c = SubElement(a, 'c')
3236 c.text = "C"
3237
3238 result = tostring(a, method="text", encoding="UTF-16")
3239
3240 self.assertEqual(_str('ABSøk på nettetCtail').encode("UTF-16"),
3241 result)
3242
3244 tostring = self.etree.tostring
3245 Element = self.etree.Element
3246 SubElement = self.etree.SubElement
3247
3248 a = Element('a')
3249 a.text = _str('Søk på nettetA')
3250 a.tail = "tail"
3251 b = SubElement(a, 'b')
3252 b.text = "B"
3253 b.tail = _str('Søk på nettetB')
3254 c = SubElement(a, 'c')
3255 c.text = "C"
3256
3257 self.assertRaises(UnicodeEncodeError,
3258 tostring, a, method="text")
3259
3260 self.assertEqual(
3261 _str('Søk på nettetABSøk på nettetBCtail').encode('utf-8'),
3262 tostring(a, encoding="UTF-8", method="text"))
3263
3265 tounicode = self.etree.tounicode
3266 Element = self.etree.Element
3267 SubElement = self.etree.SubElement
3268
3269 a = Element('a')
3270 b = SubElement(a, 'b')
3271 c = SubElement(a, 'c')
3272
3273 self.assertTrue(isinstance(tounicode(a), _unicode))
3274 self.assertEqual(_bytes('<a><b></b><c></c></a>'),
3275 canonicalize(tounicode(a)))
3276
3278 tounicode = self.etree.tounicode
3279 Element = self.etree.Element
3280 SubElement = self.etree.SubElement
3281
3282 a = Element('a')
3283 b = SubElement(a, 'b')
3284 c = SubElement(a, 'c')
3285 d = SubElement(c, 'd')
3286 self.assertTrue(isinstance(tounicode(b), _unicode))
3287 self.assertTrue(isinstance(tounicode(c), _unicode))
3288 self.assertEqual(_bytes('<b></b>'),
3289 canonicalize(tounicode(b)))
3290 self.assertEqual(_bytes('<c><d></d></c>'),
3291 canonicalize(tounicode(c)))
3292
3296
3298 tounicode = self.etree.tounicode
3299 Element = self.etree.Element
3300 SubElement = self.etree.SubElement
3301
3302 a = Element('a')
3303 b = SubElement(a, 'b')
3304 c = SubElement(a, 'c')
3305 d = SubElement(c, 'd')
3306 b.tail = 'Foo'
3307
3308 self.assertTrue(isinstance(tounicode(b), _unicode))
3309 self.assertTrue(tounicode(b) == '<b/>Foo' or
3310 tounicode(b) == '<b />Foo')
3311
3313 tounicode = self.etree.tounicode
3314 Element = self.etree.Element
3315 SubElement = self.etree.SubElement
3316
3317 a = Element('a')
3318 b = SubElement(a, 'b')
3319 c = SubElement(a, 'c')
3320
3321 result = tounicode(a)
3322 self.assertEqual(result, "<a><b/><c/></a>")
3323
3324 result = tounicode(a, pretty_print=False)
3325 self.assertEqual(result, "<a><b/><c/></a>")
3326
3327 result = tounicode(a, pretty_print=True)
3328 self.assertEqual(result, "<a>\n <b/>\n <c/>\n</a>\n")
3329
3331 tostring = self.etree.tostring
3332 Element = self.etree.Element
3333 SubElement = self.etree.SubElement
3334
3335 a = Element('a')
3336 b = SubElement(a, 'b')
3337 c = SubElement(a, 'c')
3338
3339 self.assertTrue(isinstance(tostring(a, encoding=_unicode), _unicode))
3340 self.assertEqual(_bytes('<a><b></b><c></c></a>'),
3341 canonicalize(tostring(a, encoding=_unicode)))
3342
3344 tostring = self.etree.tostring
3345 Element = self.etree.Element
3346 SubElement = self.etree.SubElement
3347
3348 a = Element('a')
3349 b = SubElement(a, 'b')
3350 c = SubElement(a, 'c')
3351 d = SubElement(c, 'd')
3352 self.assertTrue(isinstance(tostring(b, encoding=_unicode), _unicode))
3353 self.assertTrue(isinstance(tostring(c, encoding=_unicode), _unicode))
3354 self.assertEqual(_bytes('<b></b>'),
3355 canonicalize(tostring(b, encoding=_unicode)))
3356 self.assertEqual(_bytes('<c><d></d></c>'),
3357 canonicalize(tostring(c, encoding=_unicode)))
3358
3360 tostring = self.etree.tostring
3361 self.assertRaises(TypeError, self.etree.tostring,
3362 None, encoding=_unicode)
3363
3365 tostring = self.etree.tostring
3366 Element = self.etree.Element
3367 SubElement = self.etree.SubElement
3368
3369 a = Element('a')
3370 b = SubElement(a, 'b')
3371 c = SubElement(a, 'c')
3372 d = SubElement(c, 'd')
3373 b.tail = 'Foo'
3374
3375 self.assertTrue(isinstance(tostring(b, encoding=_unicode), _unicode))
3376 self.assertTrue(tostring(b, encoding=_unicode) == '<b/>Foo' or
3377 tostring(b, encoding=_unicode) == '<b />Foo')
3378
3380 tostring = self.etree.tostring
3381 Element = self.etree.Element
3382 SubElement = self.etree.SubElement
3383
3384 a = Element('a')
3385 b = SubElement(a, 'b')
3386 c = SubElement(a, 'c')
3387
3388 result = tostring(a, encoding=_unicode)
3389 self.assertEqual(result, "<a><b/><c/></a>")
3390
3391 result = tostring(a, encoding=_unicode, pretty_print=False)
3392 self.assertEqual(result, "<a><b/><c/></a>")
3393
3394 result = tostring(a, encoding=_unicode, pretty_print=True)
3395 self.assertEqual(result, "<a>\n <b/>\n <c/>\n</a>\n")
3396
3398 root = etree.Element('parent')
3399 etree.SubElement(root, 'child')
3400
3401 self.assertEqual(len(root), 1)
3402 self.assertEqual(root[0].tag, 'child')
3403
3404 # in PyPy, GC used to kill the Python proxy instance without cleanup
3405 gc.collect()
3406 self.assertEqual(len(root), 1)
3407 self.assertEqual(root[0].tag, 'child')
3408
3412
3413 el1 = SubEl()
3414 el2 = SubEl()
3415 self.assertEqual('SubEl', el1.tag)
3416 self.assertEqual('SubEl', el2.tag)
3417 el1.other = el2
3418 el2.other = el1
3419
3420 del el1, el2
3421 gc.collect()
3422 # not really testing anything here, but it shouldn't crash
3423
3424 # helper methods
3425
3427 """Write out element for comparison.
3428 """
3429 ElementTree = self.etree.ElementTree
3430 f = BytesIO()
3431 tree = ElementTree(element=element)
3432 tree.write(f, encoding=encoding, compression=compression)
3433 data = f.getvalue()
3434 if compression:
3435 data = zlib.decompress(data)
3436 return canonicalize(data)
3437
3438
3441 filename = fileInTestDir('test_broken.xml')
3442 root = etree.XML(_bytes('''\
3443 <doc xmlns:xi="http://www.w3.org/2001/XInclude">
3444 <xi:include href="%s" parse="text"/>
3445 </doc>
3446 ''' % filename))
3447 old_text = root.text
3448 content = read_file(filename)
3449 old_tail = root[0].tail
3450
3451 self.include( etree.ElementTree(root) )
3452 self.assertEqual(old_text + content + old_tail,
3453 root.text)
3454
3456 tree = etree.parse(fileInTestDir('include/test_xinclude.xml'))
3457 self.assertNotEqual(
3458 'a',
3459 tree.getroot()[1].tag)
3460 # process xincludes
3461 self.include( tree )
3462 # check whether we find it replaced with included data
3463 self.assertEqual(
3464 'a',
3465 tree.getroot()[1].tag)
3466
3468 class res(etree.Resolver):
3469 include_text = read_file(fileInTestDir('test.xml'))
3470 called = {}
3471 def resolve(self, url, id, context):
3472 if url.endswith(".dtd"):
3473 self.called["dtd"] = True
3474 return self.resolve_filename(
3475 fileInTestDir('test.dtd'), context)
3476 elif url.endswith("test_xinclude.xml"):
3477 self.called["input"] = True
3478 return None # delegate to default resolver
3479 else:
3480 self.called["include"] = True
3481 return self.resolve_string(self.include_text, context)
3482
3483 res_instance = res()
3484 parser = etree.XMLParser(load_dtd = True)
3485 parser.resolvers.add(res_instance)
3486
3487 tree = etree.parse(fileInTestDir('include/test_xinclude.xml'),
3488 parser = parser)
3489
3490 self.include(tree)
3491
3492 called = list(res_instance.called.items())
3493 called.sort()
3494 self.assertEqual(
3495 [("dtd", True), ("include", True), ("input", True)],
3496 called)
3497
3501
3502
3507
3508
3511 tree = self.parse(_bytes('<a><b/></a>'))
3512 f = BytesIO()
3513 tree.write_c14n(f)
3514 s = f.getvalue()
3515 self.assertEqual(_bytes('<a><b></b></a>'),
3516 s)
3517
3519 tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
3520 f = BytesIO()
3521 tree.write_c14n(f, compression=9)
3522 gzfile = gzip.GzipFile(fileobj=BytesIO(f.getvalue()))
3523 try:
3524 s = gzfile.read()
3525 finally:
3526 gzfile.close()
3527 self.assertEqual(_bytes('<a>'+'<b></b>'*200+'</a>'),
3528 s)
3529
3531 tree = self.parse(_bytes('<a><b/></a>'))
3532 handle, filename = tempfile.mkstemp()
3533 try:
3534 tree.write_c14n(filename)
3535 data = read_file(filename, 'rb')
3536 finally:
3537 os.close(handle)
3538 os.remove(filename)
3539 self.assertEqual(_bytes('<a><b></b></a>'),
3540 data)
3541
3543 tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
3544 handle, filename = tempfile.mkstemp()
3545 try:
3546 tree.write_c14n(filename, compression=9)
3547 f = gzip.open(filename, 'rb')
3548 try:
3549 data = f.read()
3550 finally:
3551 f.close()
3552 finally:
3553 os.close(handle)
3554 os.remove(filename)
3555 self.assertEqual(_bytes('<a>'+'<b></b>'*200+'</a>'),
3556 data)
3557
3559 tree = self.parse(_bytes('<!--hi--><a><!--ho--><b/></a><!--hu-->'))
3560 f = BytesIO()
3561 tree.write_c14n(f)
3562 s = f.getvalue()
3563 self.assertEqual(_bytes('<!--hi-->\n<a><!--ho--><b></b></a>\n<!--hu-->'),
3564 s)
3565 f = BytesIO()
3566 tree.write_c14n(f, with_comments=True)
3567 s = f.getvalue()
3568 self.assertEqual(_bytes('<!--hi-->\n<a><!--ho--><b></b></a>\n<!--hu-->'),
3569 s)
3570 f = BytesIO()
3571 tree.write_c14n(f, with_comments=False)
3572 s = f.getvalue()
3573 self.assertEqual(_bytes('<a><b></b></a>'),
3574 s)
3575
3577 tree = self.parse(_bytes('<!--hi--><a><!--ho--><b/></a><!--hu-->'))
3578 s = etree.tostring(tree, method='c14n')
3579 self.assertEqual(_bytes('<!--hi-->\n<a><!--ho--><b></b></a>\n<!--hu-->'),
3580 s)
3581 s = etree.tostring(tree, method='c14n', with_comments=True)
3582 self.assertEqual(_bytes('<!--hi-->\n<a><!--ho--><b></b></a>\n<!--hu-->'),
3583 s)
3584 s = etree.tostring(tree, method='c14n', with_comments=False)
3585 self.assertEqual(_bytes('<a><b></b></a>'),
3586 s)
3587
3589 tree = self.parse(_bytes('<!--hi--><a><!--ho--><b/></a><!--hu-->'))
3590 s = etree.tostring(tree.getroot(), method='c14n')
3591 self.assertEqual(_bytes('<a><!--ho--><b></b></a>'),
3592 s)
3593 s = etree.tostring(tree.getroot(), method='c14n', with_comments=True)
3594 self.assertEqual(_bytes('<a><!--ho--><b></b></a>'),
3595 s)
3596 s = etree.tostring(tree.getroot(), method='c14n', with_comments=False)
3597 self.assertEqual(_bytes('<a><b></b></a>'),
3598 s)
3599
3601 tree = self.parse(_bytes(
3602 '<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b/></a>'))
3603 f = BytesIO()
3604 tree.write_c14n(f)
3605 s = f.getvalue()
3606 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b></z:b></a>'),
3607 s)
3608 f = BytesIO()
3609 tree.write_c14n(f, exclusive=False)
3610 s = f.getvalue()
3611 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b></z:b></a>'),
3612 s)
3613 f = BytesIO()
3614 tree.write_c14n(f, exclusive=True)
3615 s = f.getvalue()
3616 self.assertEqual(_bytes('<a xmlns="http://abc"><z:b xmlns:z="http://cde"></z:b></a>'),
3617 s)
3618
3619 f = BytesIO()
3620 tree.write_c14n(f, exclusive=True, inclusive_ns_prefixes=['z'])
3621 s = f.getvalue()
3622 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:z="http://cde"><z:b></z:b></a>'),
3623 s)
3624
3626 tree = self.parse(_bytes(
3627 '<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b/></a>'))
3628 s = etree.tostring(tree, method='c14n')
3629 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b></z:b></a>'),
3630 s)
3631 s = etree.tostring(tree, method='c14n', exclusive=False)
3632 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b></z:b></a>'),
3633 s)
3634 s = etree.tostring(tree, method='c14n', exclusive=True)
3635 self.assertEqual(_bytes('<a xmlns="http://abc"><z:b xmlns:z="http://cde"></z:b></a>'),
3636 s)
3637
3638 s = etree.tostring(tree, method='c14n', exclusive=True, inclusive_ns_prefixes=['y'])
3639 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:y="http://bcd"><z:b xmlns:z="http://cde"></z:b></a>'),
3640 s)
3641
3643 tree = self.parse(_bytes(
3644 '<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b/></a>'))
3645 s = etree.tostring(tree.getroot(), method='c14n')
3646 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b></z:b></a>'),
3647 s)
3648 s = etree.tostring(tree.getroot(), method='c14n', exclusive=False)
3649 self.assertEqual(_bytes('<a xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b></z:b></a>'),
3650 s)
3651 s = etree.tostring(tree.getroot(), method='c14n', exclusive=True)
3652 self.assertEqual(_bytes('<a xmlns="http://abc"><z:b xmlns:z="http://cde"></z:b></a>'),
3653 s)
3654
3655 s = etree.tostring(tree.getroot()[0], method='c14n', exclusive=False)
3656 self.assertEqual(_bytes('<z:b xmlns="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"></z:b>'),
3657 s)
3658 s = etree.tostring(tree.getroot()[0], method='c14n', exclusive=True)
3659 self.assertEqual(_bytes('<z:b xmlns:z="http://cde"></z:b>'),
3660 s)
3661
3662 s = etree.tostring(tree.getroot()[0], method='c14n', exclusive=True, inclusive_ns_prefixes=['y'])
3663 self.assertEqual(_bytes('<z:b xmlns:y="http://bcd" xmlns:z="http://cde"></z:b>'),
3664 s)
3665
3667 """ Regression test to fix memory allocation issues (use 3+ inclusive NS spaces)"""
3668 tree = self.parse(_bytes(
3669 '<a xmlns:x="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b/></a>'))
3670
3671 s = etree.tostring(tree, method='c14n', exclusive=True, inclusive_ns_prefixes=['x', 'y', 'z'])
3672 self.assertEqual(_bytes('<a xmlns:x="http://abc" xmlns:y="http://bcd" xmlns:z="http://cde"><z:b></z:b></a>'),
3673 s)
3674
3675
3678 tree = self.parse(_bytes('<a><b/></a>'))
3679 f = BytesIO()
3680 tree.write(f)
3681 s = f.getvalue()
3682 self.assertEqual(_bytes('<a><b/></a>'),
3683 s)
3684
3686 tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
3687 f = BytesIO()
3688 tree.write(f, compression=9)
3689 gzfile = gzip.GzipFile(fileobj=BytesIO(f.getvalue()))
3690 try:
3691 s = gzfile.read()
3692 finally:
3693 gzfile.close()
3694 self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
3695 s)
3696
3698 tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
3699 f = BytesIO()
3700 tree.write(f, compression=0)
3701 s0 = f.getvalue()
3702
3703 f = BytesIO()
3704 tree.write(f)
3705 self.assertEqual(f.getvalue(), s0)
3706
3707 f = BytesIO()
3708 tree.write(f, compression=1)
3709 s = f.getvalue()
3710 self.assertTrue(len(s) <= len(s0))
3711 gzfile = gzip.GzipFile(fileobj=BytesIO(s))
3712 try:
3713 s1 = gzfile.read()
3714 finally:
3715 gzfile.close()
3716
3717 f = BytesIO()
3718 tree.write(f, compression=9)
3719 s = f.getvalue()
3720 self.assertTrue(len(s) <= len(s0))
3721 gzfile = gzip.GzipFile(fileobj=BytesIO(s))
3722 try:
3723 s9 = gzfile.read()
3724 finally:
3725 gzfile.close()
3726
3727 self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
3728 s0)
3729 self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
3730 s1)
3731 self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
3732 s9)
3733
3735 tree = self.parse(_bytes('<a><b/></a>'))
3736 handle, filename = tempfile.mkstemp()
3737 try:
3738 tree.write(filename)
3739 data = read_file(filename, 'rb')
3740 finally:
3741 os.close(handle)
3742 os.remove(filename)
3743 self.assertEqual(_bytes('<a><b/></a>'),
3744 data)
3745
3747 tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
3748 handle, filename = tempfile.mkstemp()
3749 try:
3750 tree.write(filename, compression=9)
3751 f = gzip.open(filename, 'rb')
3752 try:
3753 data = f.read()
3754 finally:
3755 f.close()
3756 finally:
3757 os.close(handle)
3758 os.remove(filename)
3759 self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
3760 data)
3761
3763 tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
3764 handle, filename = tempfile.mkstemp()
3765 try:
3766 tree.write(filename, compression=9)
3767 data = etree.tostring(etree.parse(filename))
3768 finally:
3769 os.close(handle)
3770 os.remove(filename)
3771 self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
3772 data)
3773
3775 tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
3776 handle, filename = tempfile.mkstemp()
3777 try:
3778 tree.write(filename, compression=9)
3779 data = etree.tostring(etree.parse(
3780 gzip.GzipFile(filename)))
3781 finally:
3782 os.close(handle)
3783 os.remove(filename)
3784 self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
3785 data)
3786
3788 etree = etree
3789
3791 parse = self.etree.parse
3792 f = BytesIO('<a><b></c></b></a>')
3793 self.etree.clear_error_log()
3794 try:
3795 parse(f)
3796 logs = None
3797 except SyntaxError:
3798 e = sys.exc_info()[1]
3799 logs = e.error_log
3800 f.close()
3801 self.assertTrue([ log for log in logs
3802 if 'mismatch' in log.message ])
3803 self.assertTrue([ log for log in logs
3804 if 'PARSER' in log.domain_name])
3805 self.assertTrue([ log for log in logs
3806 if 'ERR_TAG_NAME_MISMATCH' in log.type_name ])
3807 self.assertTrue([ log for log in logs
3808 if 1 == log.line ])
3809 self.assertTrue([ log for log in logs
3810 if 15 == log.column ])
3811
3822
3823 self.etree.use_global_python_log(Logger())
3824 f = BytesIO('<a><b></c></b></a>')
3825 try:
3826 parse(f)
3827 except SyntaxError:
3828 pass
3829 f.close()
3830
3831 self.assertTrue([ message for message in messages
3832 if 'mismatch' in message ])
3833 self.assertTrue([ message for message in messages
3834 if ':PARSER:' in message])
3835 self.assertTrue([ message for message in messages
3836 if ':ERR_TAG_NAME_MISMATCH:' in message ])
3837 self.assertTrue([ message for message in messages
3838 if ':1:15:' in message ])
3839
3840
3854 def close(self):
3855 return 'close()'
3856
3857 parser = self.etree.XMLPullParser(target=Target())
3858 events = parser.read_events()
3859
3860 parser.feed('<root><element>')
3861 self.assertFalse(list(events))
3862 self.assertFalse(list(events))
3863 parser.feed('</element><child>')
3864 self.assertEqual([('end', 'end(element)')], list(events))
3865 parser.feed('</child>')
3866 self.assertEqual([('end', 'end(child)')], list(events))
3867 parser.feed('</root>')
3868 self.assertEqual([('end', 'end(root)')], list(events))
3869 self.assertFalse(list(events))
3870 self.assertEqual('close()', parser.close())
3871
3876 def end(self, tag):
3877 return 'end(%s)' % tag
3878 def close(self):
3879 return 'close()'
3880
3881 parser = self.etree.XMLPullParser(
3882 ['start', 'end'], target=Target())
3883 events = parser.read_events()
3884
3885 parser.feed('<root><element>')
3886 self.assertEqual(
3887 [('start', 'start(root)'), ('start', 'start(element)')],
3888 list(events))
3889 self.assertFalse(list(events))
3890 parser.feed('</element><child>')
3891 self.assertEqual(
3892 [('end', 'end(element)'), ('start', 'start(child)')],
3893 list(events))
3894 parser.feed('</child>')
3895 self.assertEqual(
3896 [('end', 'end(child)')],
3897 list(events))
3898 parser.feed('</root>')
3899 self.assertEqual(
3900 [('end', 'end(root)')],
3901 list(events))
3902 self.assertFalse(list(events))
3903 self.assertEqual('close()', parser.close())
3904
3906 parser = self.etree.XMLPullParser(
3907 ['start', 'end'], target=etree.TreeBuilder())
3908 events = parser.read_events()
3909
3910 parser.feed('<root><element>')
3911 self.assert_event_tags(
3912 events, [('start', 'root'), ('start', 'element')])
3913 self.assertFalse(list(events))
3914 parser.feed('</element><child>')
3915 self.assert_event_tags(
3916 events, [('end', 'element'), ('start', 'child')])
3917 parser.feed('</child>')
3918 self.assert_event_tags(
3919 events, [('end', 'child')])
3920 parser.feed('</root>')
3921 self.assert_event_tags(
3922 events, [('end', 'root')])
3923 self.assertFalse(list(events))
3924 root = parser.close()
3925 self.assertEqual('root', root.tag)
3926
3928 class Target(etree.TreeBuilder):
3929 def end(self, tag):
3930 el = super(Target, self).end(tag)
3931 el.tag += '-huhu'
3932 return el
3933
3934 parser = self.etree.XMLPullParser(
3935 ['start', 'end'], target=Target())
3936 events = parser.read_events()
3937
3938 parser.feed('<root><element>')
3939 self.assert_event_tags(
3940 events, [('start', 'root'), ('start', 'element')])
3941 self.assertFalse(list(events))
3942 parser.feed('</element><child>')
3943 self.assert_event_tags(
3944 events, [('end', 'element-huhu'), ('start', 'child')])
3945 parser.feed('</child>')
3946 self.assert_event_tags(
3947 events, [('end', 'child-huhu')])
3948 parser.feed('</root>')
3949 self.assert_event_tags(
3950 events, [('end', 'root-huhu')])
3951 self.assertFalse(list(events))
3952 root = parser.close()
3953 self.assertEqual('root-huhu', root.tag)
3954
3955
3957 suite = unittest.TestSuite()
3958 suite.addTests([unittest.makeSuite(ETreeOnlyTestCase)])
3959 suite.addTests([unittest.makeSuite(ETreeXIncludeTestCase)])
3960 suite.addTests([unittest.makeSuite(ElementIncludeTestCase)])
3961 suite.addTests([unittest.makeSuite(ETreeC14NTestCase)])
3962 suite.addTests([unittest.makeSuite(ETreeWriteTestCase)])
3963 suite.addTests([unittest.makeSuite(ETreeErrorLogTest)])
3964 suite.addTests([unittest.makeSuite(XMLPullParserTest)])
3965 suite.addTests(
3966 [make_doctest('../../../doc/tutorial.txt')])
3967 if sys.version_info >= (2,6):
3968 # now requires the 'with' statement
3969 suite.addTests(
3970 [make_doctest('../../../doc/api.txt')])
3971 suite.addTests(
3972 [make_doctest('../../../doc/FAQ.txt')])
3973 suite.addTests(
3974 [make_doctest('../../../doc/parsing.txt')])
3975 suite.addTests(
3976 [make_doctest('../../../doc/resolvers.txt')])
3977 return suite
3978
3979 if __name__ == '__main__':
3980 print('to test use test.py %s' % __file__)
3981
| Home | Trees | Indices | Help |
|
|---|
| Generated by Epydoc 3.0.1 on Thu Dec 12 20:23:41 2013 | http://epydoc.sourceforge.net |