blob: d3c0da081d1fe32ff92e7e7f15ee82bcfc9a6d6b [file] [log] [blame]
Eli Bendersky865756a2012-03-09 13:38:15 +02001# IMPORTANT: the same tests are run from "test_xml_etree_c" in order
2# to ensure consistency between the C implementation and the Python
3# implementation.
Florent Xiclunaf15351d2010-03-13 23:24:31 +00004#
5# For this purpose, the module-level "ET" symbol is temporarily
6# monkey-patched when running the "test_xml_etree_c" test suite.
Florent Xiclunaf15351d2010-03-13 23:24:31 +00007
Georg Brandl1f7fffb2010-10-15 15:57:45 +00008import html
Eli Benderskyf996e772012-03-16 05:53:30 +02009import io
Eli Bendersky698bdb22013-01-10 06:01:06 -080010import operator
Eli Bendersky7ec45f72012-12-30 06:17:49 -080011import pickle
Eli Bendersky0192ba32012-03-30 16:38:33 +030012import sys
Eli Benderskye26fa1b2013-05-19 17:49:54 -070013import types
Victor Stinner6c6f8512010-08-07 10:09:35 +000014import unittest
Eli Benderskya5e82202012-03-31 13:55:38 +030015import weakref
Armin Rigo9ed73062005-12-14 18:10:45 +000016
Eli Bendersky698bdb22013-01-10 06:01:06 -080017from itertools import product
Benjamin Petersonee8712c2008-05-20 21:35:26 +000018from test import support
Eli Bendersky23687042013-02-26 05:53:23 -080019from test.support import TESTFN, findfile, import_fresh_module, gc_collect
Armin Rigo9ed73062005-12-14 18:10:45 +000020
Eli Bendersky698bdb22013-01-10 06:01:06 -080021# pyET is the pure-Python implementation.
Eli Bendersky458c0d52013-01-10 06:07:00 -080022#
Eli Bendersky698bdb22013-01-10 06:01:06 -080023# ET is pyET in test_xml_etree and is the C accelerated version in
24# test_xml_etree_c.
Eli Bendersky64d11e62012-06-15 07:42:50 +030025pyET = None
26ET = None
Florent Xiclunaf15351d2010-03-13 23:24:31 +000027
28SIMPLE_XMLFILE = findfile("simple.xml", subdir="xmltestdata")
Victor Stinner6c6f8512010-08-07 10:09:35 +000029try:
Marc-André Lemburg8f36af72011-02-25 15:42:01 +000030 SIMPLE_XMLFILE.encode("utf-8")
Victor Stinner6c6f8512010-08-07 10:09:35 +000031except UnicodeEncodeError:
32 raise unittest.SkipTest("filename is not encodable to utf8")
Florent Xiclunaf15351d2010-03-13 23:24:31 +000033SIMPLE_NS_XMLFILE = findfile("simple-ns.xml", subdir="xmltestdata")
34
35SAMPLE_XML = """\
Armin Rigo9ed73062005-12-14 18:10:45 +000036<body>
Florent Xiclunaf15351d2010-03-13 23:24:31 +000037 <tag class='a'>text</tag>
38 <tag class='b' />
Armin Rigo9ed73062005-12-14 18:10:45 +000039 <section>
Florent Xiclunaf15351d2010-03-13 23:24:31 +000040 <tag class='b' id='inner'>subtext</tag>
Armin Rigo9ed73062005-12-14 18:10:45 +000041 </section>
42</body>
43"""
44
Florent Xiclunaf15351d2010-03-13 23:24:31 +000045SAMPLE_SECTION = """\
46<section>
47 <tag class='b' id='inner'>subtext</tag>
48 <nexttag />
49 <nextsection>
50 <tag />
51 </nextsection>
52</section>
53"""
54
Armin Rigo9ed73062005-12-14 18:10:45 +000055SAMPLE_XML_NS = """
56<body xmlns="http://effbot.org/ns">
57 <tag>text</tag>
58 <tag />
59 <section>
60 <tag>subtext</tag>
61 </section>
62</body>
63"""
64
Eli Bendersky737b1732012-05-29 06:02:56 +030065SAMPLE_XML_NS_ELEMS = """
66<root>
67<h:table xmlns:h="hello">
68 <h:tr>
69 <h:td>Apples</h:td>
70 <h:td>Bananas</h:td>
71 </h:tr>
72</h:table>
73
74<f:table xmlns:f="foo">
75 <f:name>African Coffee Table</f:name>
76 <f:width>80</f:width>
77 <f:length>120</f:length>
78</f:table>
79</root>
80"""
Florent Xiclunaf15351d2010-03-13 23:24:31 +000081
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +020082ENTITY_XML = """\
83<!DOCTYPE points [
84<!ENTITY % user-entities SYSTEM 'user-entities.xml'>
85%user-entities;
86]>
87<document>&entity;</document>
88"""
Armin Rigo9ed73062005-12-14 18:10:45 +000089
Armin Rigo9ed73062005-12-14 18:10:45 +000090
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +020091class ModuleTest(unittest.TestCase):
Eli Bendersky23687042013-02-26 05:53:23 -080092 # TODO: this should be removed once we get rid of the global module vars
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +020093
94 def test_sanity(self):
95 # Import sanity.
96
97 from xml.etree import ElementTree
98 from xml.etree import ElementInclude
99 from xml.etree import ElementPath
100
Armin Rigo9ed73062005-12-14 18:10:45 +0000101
Florent Xiclunac17f1722010-08-08 19:48:29 +0000102def serialize(elem, to_string=True, encoding='unicode', **options):
Florent Xiclunac17f1722010-08-08 19:48:29 +0000103 if encoding != 'unicode':
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000104 file = io.BytesIO()
105 else:
106 file = io.StringIO()
Armin Rigo9ed73062005-12-14 18:10:45 +0000107 tree = ET.ElementTree(elem)
Florent Xiclunac17f1722010-08-08 19:48:29 +0000108 tree.write(file, encoding=encoding, **options)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000109 if to_string:
110 return file.getvalue()
111 else:
112 file.seek(0)
113 return file
Armin Rigo9ed73062005-12-14 18:10:45 +0000114
Armin Rigo9ed73062005-12-14 18:10:45 +0000115def summarize_list(seq):
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200116 return [elem.tag for elem in seq]
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000117
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000118
Eli Bendersky698bdb22013-01-10 06:01:06 -0800119class ElementTestCase:
120 @classmethod
121 def setUpClass(cls):
122 cls.modules = {pyET, ET}
123
Serhiy Storchakabad12572014-12-15 14:03:42 +0200124 def pickleRoundTrip(self, obj, name, dumper, loader, proto):
Eli Bendersky698bdb22013-01-10 06:01:06 -0800125 save_m = sys.modules[name]
126 try:
127 sys.modules[name] = dumper
Serhiy Storchakabad12572014-12-15 14:03:42 +0200128 temp = pickle.dumps(obj, proto)
Eli Bendersky698bdb22013-01-10 06:01:06 -0800129 sys.modules[name] = loader
130 result = pickle.loads(temp)
131 except pickle.PicklingError as pe:
132 # pyET must be second, because pyET may be (equal to) ET.
133 human = dict([(ET, "cET"), (pyET, "pyET")])
134 raise support.TestFailed("Failed to round-trip %r from %r to %r"
135 % (obj,
136 human.get(dumper, dumper),
137 human.get(loader, loader))) from pe
138 finally:
139 sys.modules[name] = save_m
140 return result
141
142 def assertEqualElements(self, alice, bob):
143 self.assertIsInstance(alice, (ET.Element, pyET.Element))
144 self.assertIsInstance(bob, (ET.Element, pyET.Element))
145 self.assertEqual(len(list(alice)), len(list(bob)))
146 for x, y in zip(alice, bob):
147 self.assertEqualElements(x, y)
148 properties = operator.attrgetter('tag', 'tail', 'text', 'attrib')
149 self.assertEqual(properties(alice), properties(bob))
150
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000151# --------------------------------------------------------------------
152# element tree tests
Armin Rigo9ed73062005-12-14 18:10:45 +0000153
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200154class ElementTreeTest(unittest.TestCase):
Armin Rigo9ed73062005-12-14 18:10:45 +0000155
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200156 def serialize_check(self, elem, expected):
157 self.assertEqual(serialize(elem), expected)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000158
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200159 def test_interface(self):
160 # Test element tree interface.
Armin Rigo9ed73062005-12-14 18:10:45 +0000161
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200162 def check_string(string):
163 len(string)
164 for char in string:
165 self.assertEqual(len(char), 1,
166 msg="expected one-character string, got %r" % char)
167 new_string = string + ""
168 new_string = string + " "
169 string[:0]
Armin Rigo9ed73062005-12-14 18:10:45 +0000170
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200171 def check_mapping(mapping):
172 len(mapping)
173 keys = mapping.keys()
174 items = mapping.items()
175 for key in keys:
176 item = mapping[key]
177 mapping["key"] = "value"
178 self.assertEqual(mapping["key"], "value",
179 msg="expected value string, got %r" % mapping["key"])
Armin Rigo9ed73062005-12-14 18:10:45 +0000180
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200181 def check_element(element):
182 self.assertTrue(ET.iselement(element), msg="not an element")
183 self.assertTrue(hasattr(element, "tag"), msg="no tag member")
184 self.assertTrue(hasattr(element, "attrib"), msg="no attrib member")
185 self.assertTrue(hasattr(element, "text"), msg="no text member")
186 self.assertTrue(hasattr(element, "tail"), msg="no tail member")
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000187
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200188 check_string(element.tag)
189 check_mapping(element.attrib)
190 if element.text is not None:
191 check_string(element.text)
192 if element.tail is not None:
193 check_string(element.tail)
194 for elem in element:
195 check_element(elem)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000196
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200197 element = ET.Element("tag")
198 check_element(element)
199 tree = ET.ElementTree(element)
200 check_element(tree.getroot())
201 element = ET.Element("t\xe4g", key="value")
202 tree = ET.ElementTree(element)
203 self.assertRegex(repr(element), r"^<Element 't\xe4g' at 0x.*>$")
204 element = ET.Element("tag", key="value")
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000205
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200206 # Make sure all standard element methods exist.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000207
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200208 def check_method(method):
209 self.assertTrue(hasattr(method, '__call__'),
210 msg="%s not callable" % method)
Armin Rigo9ed73062005-12-14 18:10:45 +0000211
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200212 check_method(element.append)
213 check_method(element.extend)
214 check_method(element.insert)
215 check_method(element.remove)
216 check_method(element.getchildren)
217 check_method(element.find)
218 check_method(element.iterfind)
219 check_method(element.findall)
220 check_method(element.findtext)
221 check_method(element.clear)
222 check_method(element.get)
223 check_method(element.set)
224 check_method(element.keys)
225 check_method(element.items)
226 check_method(element.iter)
227 check_method(element.itertext)
228 check_method(element.getiterator)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000229
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200230 # These methods return an iterable. See bug 6472.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000231
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200232 def check_iter(it):
233 check_method(it.__next__)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000234
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200235 check_iter(element.iterfind("tag"))
236 check_iter(element.iterfind("*"))
237 check_iter(tree.iterfind("tag"))
238 check_iter(tree.iterfind("*"))
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000239
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200240 # These aliases are provided:
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000241
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200242 self.assertEqual(ET.XML, ET.fromstring)
243 self.assertEqual(ET.PI, ET.ProcessingInstruction)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000244
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200245 def test_simpleops(self):
246 # Basic method sanity checks.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000247
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200248 elem = ET.XML("<body><tag/></body>")
249 self.serialize_check(elem, '<body><tag /></body>')
250 e = ET.Element("tag2")
251 elem.append(e)
252 self.serialize_check(elem, '<body><tag /><tag2 /></body>')
253 elem.remove(e)
254 self.serialize_check(elem, '<body><tag /></body>')
255 elem.insert(0, e)
256 self.serialize_check(elem, '<body><tag2 /><tag /></body>')
257 elem.remove(e)
258 elem.extend([e])
259 self.serialize_check(elem, '<body><tag /><tag2 /></body>')
260 elem.remove(e)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000261
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200262 element = ET.Element("tag", key="value")
263 self.serialize_check(element, '<tag key="value" />') # 1
264 subelement = ET.Element("subtag")
265 element.append(subelement)
266 self.serialize_check(element, '<tag key="value"><subtag /></tag>') # 2
267 element.insert(0, subelement)
268 self.serialize_check(element,
269 '<tag key="value"><subtag /><subtag /></tag>') # 3
270 element.remove(subelement)
271 self.serialize_check(element, '<tag key="value"><subtag /></tag>') # 4
272 element.remove(subelement)
273 self.serialize_check(element, '<tag key="value" />') # 5
274 with self.assertRaises(ValueError) as cm:
275 element.remove(subelement)
276 self.assertEqual(str(cm.exception), 'list.remove(x): x not in list')
277 self.serialize_check(element, '<tag key="value" />') # 6
278 element[0:0] = [subelement, subelement, subelement]
279 self.serialize_check(element[1], '<subtag />')
280 self.assertEqual(element[1:9], [element[1], element[2]])
281 self.assertEqual(element[:9:2], [element[0], element[2]])
282 del element[1:2]
283 self.serialize_check(element,
284 '<tag key="value"><subtag /><subtag /></tag>')
Florent Xiclunaa72a98f2012-02-13 11:03:30 +0100285
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200286 def test_cdata(self):
287 # Test CDATA handling (etc).
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000288
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200289 self.serialize_check(ET.XML("<tag>hello</tag>"),
290 '<tag>hello</tag>')
291 self.serialize_check(ET.XML("<tag>&#104;&#101;&#108;&#108;&#111;</tag>"),
292 '<tag>hello</tag>')
293 self.serialize_check(ET.XML("<tag><![CDATA[hello]]></tag>"),
294 '<tag>hello</tag>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000295
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200296 def test_file_init(self):
297 stringfile = io.BytesIO(SAMPLE_XML.encode("utf-8"))
298 tree = ET.ElementTree(file=stringfile)
299 self.assertEqual(tree.find("tag").tag, 'tag')
300 self.assertEqual(tree.find("section/tag").tag, 'tag')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000301
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200302 tree = ET.ElementTree(file=SIMPLE_XMLFILE)
303 self.assertEqual(tree.find("element").tag, 'element')
304 self.assertEqual(tree.find("element/../empty-element").tag,
305 'empty-element')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000306
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200307 def test_path_cache(self):
308 # Check that the path cache behaves sanely.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000309
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200310 from xml.etree import ElementPath
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000311
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200312 elem = ET.XML(SAMPLE_XML)
313 for i in range(10): ET.ElementTree(elem).find('./'+str(i))
314 cache_len_10 = len(ElementPath._cache)
315 for i in range(10): ET.ElementTree(elem).find('./'+str(i))
316 self.assertEqual(len(ElementPath._cache), cache_len_10)
317 for i in range(20): ET.ElementTree(elem).find('./'+str(i))
318 self.assertGreater(len(ElementPath._cache), cache_len_10)
319 for i in range(600): ET.ElementTree(elem).find('./'+str(i))
320 self.assertLess(len(ElementPath._cache), 500)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000321
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200322 def test_copy(self):
323 # Test copy handling (etc).
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000324
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200325 import copy
326 e1 = ET.XML("<tag>hello<foo/></tag>")
327 e2 = copy.copy(e1)
328 e3 = copy.deepcopy(e1)
329 e1.find("foo").tag = "bar"
330 self.serialize_check(e1, '<tag>hello<bar /></tag>')
331 self.serialize_check(e2, '<tag>hello<bar /></tag>')
332 self.serialize_check(e3, '<tag>hello<foo /></tag>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000333
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200334 def test_attrib(self):
335 # Test attribute handling.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000336
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200337 elem = ET.Element("tag")
338 elem.get("key") # 1.1
339 self.assertEqual(elem.get("key", "default"), 'default') # 1.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000340
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200341 elem.set("key", "value")
342 self.assertEqual(elem.get("key"), 'value') # 1.3
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000343
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200344 elem = ET.Element("tag", key="value")
345 self.assertEqual(elem.get("key"), 'value') # 2.1
346 self.assertEqual(elem.attrib, {'key': 'value'}) # 2.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000347
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200348 attrib = {"key": "value"}
349 elem = ET.Element("tag", attrib)
350 attrib.clear() # check for aliasing issues
351 self.assertEqual(elem.get("key"), 'value') # 3.1
352 self.assertEqual(elem.attrib, {'key': 'value'}) # 3.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000353
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200354 attrib = {"key": "value"}
355 elem = ET.Element("tag", **attrib)
356 attrib.clear() # check for aliasing issues
357 self.assertEqual(elem.get("key"), 'value') # 4.1
358 self.assertEqual(elem.attrib, {'key': 'value'}) # 4.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000359
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200360 elem = ET.Element("tag", {"key": "other"}, key="value")
361 self.assertEqual(elem.get("key"), 'value') # 5.1
362 self.assertEqual(elem.attrib, {'key': 'value'}) # 5.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000363
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200364 elem = ET.Element('test')
365 elem.text = "aa"
366 elem.set('testa', 'testval')
367 elem.set('testb', 'test2')
368 self.assertEqual(ET.tostring(elem),
369 b'<test testa="testval" testb="test2">aa</test>')
370 self.assertEqual(sorted(elem.keys()), ['testa', 'testb'])
371 self.assertEqual(sorted(elem.items()),
372 [('testa', 'testval'), ('testb', 'test2')])
373 self.assertEqual(elem.attrib['testb'], 'test2')
374 elem.attrib['testb'] = 'test1'
375 elem.attrib['testc'] = 'test2'
376 self.assertEqual(ET.tostring(elem),
377 b'<test testa="testval" testb="test1" testc="test2">aa</test>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000378
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200379 def test_makeelement(self):
380 # Test makeelement handling.
Antoine Pitroub86680e2010-10-14 21:15:17 +0000381
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200382 elem = ET.Element("tag")
383 attrib = {"key": "value"}
384 subelem = elem.makeelement("subtag", attrib)
385 self.assertIsNot(subelem.attrib, attrib, msg="attrib aliasing")
386 elem.append(subelem)
387 self.serialize_check(elem, '<tag><subtag key="value" /></tag>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000388
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200389 elem.clear()
390 self.serialize_check(elem, '<tag />')
391 elem.append(subelem)
392 self.serialize_check(elem, '<tag><subtag key="value" /></tag>')
393 elem.extend([subelem, subelem])
394 self.serialize_check(elem,
395 '<tag><subtag key="value" /><subtag key="value" /><subtag key="value" /></tag>')
396 elem[:] = [subelem]
397 self.serialize_check(elem, '<tag><subtag key="value" /></tag>')
398 elem[:] = tuple([subelem])
399 self.serialize_check(elem, '<tag><subtag key="value" /></tag>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000400
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200401 def test_parsefile(self):
402 # Test parsing from file.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000403
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200404 tree = ET.parse(SIMPLE_XMLFILE)
405 stream = io.StringIO()
406 tree.write(stream, encoding='unicode')
407 self.assertEqual(stream.getvalue(),
408 '<root>\n'
409 ' <element key="value">text</element>\n'
410 ' <element>text</element>tail\n'
411 ' <empty-element />\n'
412 '</root>')
413 tree = ET.parse(SIMPLE_NS_XMLFILE)
414 stream = io.StringIO()
415 tree.write(stream, encoding='unicode')
416 self.assertEqual(stream.getvalue(),
417 '<ns0:root xmlns:ns0="namespace">\n'
418 ' <ns0:element key="value">text</ns0:element>\n'
419 ' <ns0:element>text</ns0:element>tail\n'
420 ' <ns0:empty-element />\n'
421 '</ns0:root>')
Armin Rigo9ed73062005-12-14 18:10:45 +0000422
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200423 with open(SIMPLE_XMLFILE) as f:
424 data = f.read()
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000425
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200426 parser = ET.XMLParser()
427 self.assertRegex(parser.version, r'^Expat ')
428 parser.feed(data)
429 self.serialize_check(parser.close(),
430 '<root>\n'
431 ' <element key="value">text</element>\n'
432 ' <element>text</element>tail\n'
433 ' <empty-element />\n'
434 '</root>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000435
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200436 target = ET.TreeBuilder()
437 parser = ET.XMLParser(target=target)
438 parser.feed(data)
439 self.serialize_check(parser.close(),
440 '<root>\n'
441 ' <element key="value">text</element>\n'
442 ' <element>text</element>tail\n'
443 ' <empty-element />\n'
444 '</root>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000445
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200446 def test_parseliteral(self):
447 element = ET.XML("<html><body>text</body></html>")
448 self.assertEqual(ET.tostring(element, encoding='unicode'),
449 '<html><body>text</body></html>')
450 element = ET.fromstring("<html><body>text</body></html>")
451 self.assertEqual(ET.tostring(element, encoding='unicode'),
452 '<html><body>text</body></html>')
453 sequence = ["<html><body>", "text</bo", "dy></html>"]
454 element = ET.fromstringlist(sequence)
455 self.assertEqual(ET.tostring(element),
456 b'<html><body>text</body></html>')
457 self.assertEqual(b"".join(ET.tostringlist(element)),
458 b'<html><body>text</body></html>')
459 self.assertEqual(ET.tostring(element, "ascii"),
460 b"<?xml version='1.0' encoding='ascii'?>\n"
461 b"<html><body>text</body></html>")
462 _, ids = ET.XMLID("<html><body>text</body></html>")
463 self.assertEqual(len(ids), 0)
464 _, ids = ET.XMLID("<html><body id='body'>text</body></html>")
465 self.assertEqual(len(ids), 1)
466 self.assertEqual(ids["body"].tag, 'body')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000467
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200468 def test_iterparse(self):
469 # Test iterparse interface.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000470
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200471 iterparse = ET.iterparse
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000472
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200473 context = iterparse(SIMPLE_XMLFILE)
474 action, elem = next(context)
475 self.assertEqual((action, elem.tag), ('end', 'element'))
476 self.assertEqual([(action, elem.tag) for action, elem in context], [
477 ('end', 'element'),
478 ('end', 'empty-element'),
479 ('end', 'root'),
480 ])
481 self.assertEqual(context.root.tag, 'root')
482
483 context = iterparse(SIMPLE_NS_XMLFILE)
484 self.assertEqual([(action, elem.tag) for action, elem in context], [
485 ('end', '{namespace}element'),
486 ('end', '{namespace}element'),
487 ('end', '{namespace}empty-element'),
488 ('end', '{namespace}root'),
489 ])
490
491 events = ()
492 context = iterparse(SIMPLE_XMLFILE, events)
493 self.assertEqual([(action, elem.tag) for action, elem in context], [])
494
495 events = ()
496 context = iterparse(SIMPLE_XMLFILE, events=events)
497 self.assertEqual([(action, elem.tag) for action, elem in context], [])
498
499 events = ("start", "end")
500 context = iterparse(SIMPLE_XMLFILE, events)
501 self.assertEqual([(action, elem.tag) for action, elem in context], [
502 ('start', 'root'),
503 ('start', 'element'),
504 ('end', 'element'),
505 ('start', 'element'),
506 ('end', 'element'),
507 ('start', 'empty-element'),
508 ('end', 'empty-element'),
509 ('end', 'root'),
510 ])
511
512 events = ("start", "end", "start-ns", "end-ns")
513 context = iterparse(SIMPLE_NS_XMLFILE, events)
Eli Bendersky23687042013-02-26 05:53:23 -0800514 self.assertEqual([(action, elem.tag) if action in ("start", "end")
515 else (action, elem)
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200516 for action, elem in context], [
517 ('start-ns', ('', 'namespace')),
518 ('start', '{namespace}root'),
519 ('start', '{namespace}element'),
520 ('end', '{namespace}element'),
521 ('start', '{namespace}element'),
522 ('end', '{namespace}element'),
523 ('start', '{namespace}empty-element'),
524 ('end', '{namespace}empty-element'),
525 ('end', '{namespace}root'),
526 ('end-ns', None),
527 ])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000528
Eli Bendersky5dd40e52013-11-28 06:31:58 -0800529 events = ('start-ns', 'end-ns')
530 context = iterparse(io.StringIO(r"<root xmlns=''/>"), events)
531 res = [action for action, elem in context]
532 self.assertEqual(res, ['start-ns', 'end-ns'])
533
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200534 events = ("start", "end", "bogus")
535 with self.assertRaises(ValueError) as cm:
536 with open(SIMPLE_XMLFILE, "rb") as f:
537 iterparse(f, events)
538 self.assertEqual(str(cm.exception), "unknown event 'bogus'")
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000539
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200540 source = io.BytesIO(
541 b"<?xml version='1.0' encoding='iso-8859-1'?>\n"
542 b"<body xmlns='http://&#233;ffbot.org/ns'\n"
543 b" xmlns:cl\xe9='http://effbot.org/ns'>text</body>\n")
544 events = ("start-ns",)
545 context = iterparse(source, events)
546 self.assertEqual([(action, elem) for action, elem in context], [
547 ('start-ns', ('', 'http://\xe9ffbot.org/ns')),
548 ('start-ns', ('cl\xe9', 'http://effbot.org/ns')),
549 ])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000550
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200551 source = io.StringIO("<document />junk")
552 it = iterparse(source)
553 action, elem = next(it)
554 self.assertEqual((action, elem.tag), ('end', 'document'))
555 with self.assertRaises(ET.ParseError) as cm:
556 next(it)
557 self.assertEqual(str(cm.exception),
558 'junk after document element: line 1, column 12')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000559
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200560 def test_writefile(self):
561 elem = ET.Element("tag")
562 elem.text = "text"
563 self.serialize_check(elem, '<tag>text</tag>')
564 ET.SubElement(elem, "subtag").text = "subtext"
565 self.serialize_check(elem, '<tag>text<subtag>subtext</subtag></tag>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000566
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200567 # Test tag suppression
568 elem.tag = None
569 self.serialize_check(elem, 'text<subtag>subtext</subtag>')
570 elem.insert(0, ET.Comment("comment"))
571 self.serialize_check(elem,
572 'text<!--comment--><subtag>subtext</subtag>') # assumes 1.3
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000573
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200574 elem[0] = ET.PI("key", "value")
575 self.serialize_check(elem, 'text<?key value?><subtag>subtext</subtag>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000576
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200577 def test_custom_builder(self):
578 # Test parser w. custom builder.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000579
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200580 with open(SIMPLE_XMLFILE) as f:
581 data = f.read()
582 class Builder(list):
583 def start(self, tag, attrib):
584 self.append(("start", tag))
585 def end(self, tag):
586 self.append(("end", tag))
587 def data(self, text):
588 pass
589 builder = Builder()
590 parser = ET.XMLParser(target=builder)
591 parser.feed(data)
592 self.assertEqual(builder, [
593 ('start', 'root'),
594 ('start', 'element'),
595 ('end', 'element'),
596 ('start', 'element'),
597 ('end', 'element'),
598 ('start', 'empty-element'),
599 ('end', 'empty-element'),
600 ('end', 'root'),
601 ])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000602
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200603 with open(SIMPLE_NS_XMLFILE) as f:
604 data = f.read()
605 class Builder(list):
606 def start(self, tag, attrib):
607 self.append(("start", tag))
608 def end(self, tag):
609 self.append(("end", tag))
610 def data(self, text):
611 pass
612 def pi(self, target, data):
613 self.append(("pi", target, data))
614 def comment(self, data):
615 self.append(("comment", data))
616 builder = Builder()
617 parser = ET.XMLParser(target=builder)
618 parser.feed(data)
619 self.assertEqual(builder, [
620 ('pi', 'pi', 'data'),
621 ('comment', ' comment '),
622 ('start', '{namespace}root'),
623 ('start', '{namespace}element'),
624 ('end', '{namespace}element'),
625 ('start', '{namespace}element'),
626 ('end', '{namespace}element'),
627 ('start', '{namespace}empty-element'),
628 ('end', '{namespace}empty-element'),
629 ('end', '{namespace}root'),
630 ])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000631
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000632
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200633 def test_getchildren(self):
634 # Test Element.getchildren()
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000635
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200636 with open(SIMPLE_XMLFILE, "rb") as f:
637 tree = ET.parse(f)
638 self.assertEqual([summarize_list(elem.getchildren())
639 for elem in tree.getroot().iter()], [
640 ['element', 'element', 'empty-element'],
641 [],
642 [],
643 [],
644 ])
645 self.assertEqual([summarize_list(elem.getchildren())
646 for elem in tree.getiterator()], [
647 ['element', 'element', 'empty-element'],
648 [],
649 [],
650 [],
651 ])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000652
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200653 elem = ET.XML(SAMPLE_XML)
654 self.assertEqual(len(elem.getchildren()), 3)
655 self.assertEqual(len(elem[2].getchildren()), 1)
656 self.assertEqual(elem[:], elem.getchildren())
657 child1 = elem[0]
658 child2 = elem[2]
659 del elem[1:2]
660 self.assertEqual(len(elem.getchildren()), 2)
661 self.assertEqual(child1, elem[0])
662 self.assertEqual(child2, elem[1])
663 elem[0:2] = [child2, child1]
664 self.assertEqual(child2, elem[0])
665 self.assertEqual(child1, elem[1])
666 self.assertNotEqual(child1, elem[0])
667 elem.clear()
668 self.assertEqual(elem.getchildren(), [])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000669
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200670 def test_writestring(self):
671 elem = ET.XML("<html><body>text</body></html>")
672 self.assertEqual(ET.tostring(elem), b'<html><body>text</body></html>')
673 elem = ET.fromstring("<html><body>text</body></html>")
674 self.assertEqual(ET.tostring(elem), b'<html><body>text</body></html>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000675
Serhiy Storchaka66d53fa2013-05-22 17:07:51 +0300676 def test_encoding(self):
677 def check(encoding, body=''):
678 xml = ("<?xml version='1.0' encoding='%s'?><xml>%s</xml>" %
679 (encoding, body))
680 self.assertEqual(ET.XML(xml.encode(encoding)).text, body)
681 self.assertEqual(ET.XML(xml).text, body)
682 check("ascii", 'a')
683 check("us-ascii", 'a')
684 check("iso-8859-1", '\xbd')
685 check("iso-8859-15", '\u20ac')
686 check("cp437", '\u221a')
687 check("mac-roman", '\u02da')
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000688
Eli Bendersky6dc32b32013-05-25 05:25:48 -0700689 def xml(encoding):
690 return "<?xml version='1.0' encoding='%s'?><xml />" % encoding
691 def bxml(encoding):
692 return xml(encoding).encode(encoding)
693 supported_encodings = [
694 'ascii', 'utf-8', 'utf-8-sig', 'utf-16', 'utf-16be', 'utf-16le',
695 'iso8859-1', 'iso8859-2', 'iso8859-3', 'iso8859-4', 'iso8859-5',
696 'iso8859-6', 'iso8859-7', 'iso8859-8', 'iso8859-9', 'iso8859-10',
697 'iso8859-13', 'iso8859-14', 'iso8859-15', 'iso8859-16',
698 'cp437', 'cp720', 'cp737', 'cp775', 'cp850', 'cp852',
699 'cp855', 'cp856', 'cp857', 'cp858', 'cp860', 'cp861', 'cp862',
Serhiy Storchakabe0c3252013-11-23 18:52:23 +0200700 'cp863', 'cp865', 'cp866', 'cp869', 'cp874', 'cp1006', 'cp1125',
701 'cp1250', 'cp1251', 'cp1252', 'cp1253', 'cp1254', 'cp1255',
702 'cp1256', 'cp1257', 'cp1258',
Eli Bendersky6dc32b32013-05-25 05:25:48 -0700703 'mac-cyrillic', 'mac-greek', 'mac-iceland', 'mac-latin2',
704 'mac-roman', 'mac-turkish',
705 'iso2022-jp', 'iso2022-jp-1', 'iso2022-jp-2', 'iso2022-jp-2004',
706 'iso2022-jp-3', 'iso2022-jp-ext',
707 'koi8-r', 'koi8-u',
708 'hz', 'ptcp154',
709 ]
710 for encoding in supported_encodings:
711 self.assertEqual(ET.tostring(ET.XML(bxml(encoding))), b'<xml />')
712
713 unsupported_ascii_compatible_encodings = [
714 'big5', 'big5hkscs',
715 'cp932', 'cp949', 'cp950',
716 'euc-jp', 'euc-jis-2004', 'euc-jisx0213', 'euc-kr',
717 'gb2312', 'gbk', 'gb18030',
718 'iso2022-kr', 'johab',
719 'shift-jis', 'shift-jis-2004', 'shift-jisx0213',
720 'utf-7',
721 ]
722 for encoding in unsupported_ascii_compatible_encodings:
723 self.assertRaises(ValueError, ET.XML, bxml(encoding))
724
725 unsupported_ascii_incompatible_encodings = [
726 'cp037', 'cp424', 'cp500', 'cp864', 'cp875', 'cp1026', 'cp1140',
727 'utf_32', 'utf_32_be', 'utf_32_le',
728 ]
729 for encoding in unsupported_ascii_incompatible_encodings:
730 self.assertRaises(ET.ParseError, ET.XML, bxml(encoding))
731
732 self.assertRaises(ValueError, ET.XML, xml('undefined').encode('ascii'))
733 self.assertRaises(LookupError, ET.XML, xml('xxx').encode('ascii'))
734
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200735 def test_methods(self):
736 # Test serialization methods.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000737
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200738 e = ET.XML("<html><link/><script>1 &lt; 2</script></html>")
739 e.tail = "\n"
740 self.assertEqual(serialize(e),
741 '<html><link /><script>1 &lt; 2</script></html>\n')
742 self.assertEqual(serialize(e, method=None),
743 '<html><link /><script>1 &lt; 2</script></html>\n')
744 self.assertEqual(serialize(e, method="xml"),
745 '<html><link /><script>1 &lt; 2</script></html>\n')
746 self.assertEqual(serialize(e, method="html"),
747 '<html><link><script>1 < 2</script></html>\n')
748 self.assertEqual(serialize(e, method="text"), '1 < 2\n')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000749
Christian Heimes54ad7e32013-07-05 01:39:49 +0200750 def test_issue18347(self):
751 e = ET.XML('<html><CamelCase>text</CamelCase></html>')
752 self.assertEqual(serialize(e),
753 '<html><CamelCase>text</CamelCase></html>')
754 self.assertEqual(serialize(e, method="html"),
755 '<html><CamelCase>text</CamelCase></html>')
756
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200757 def test_entity(self):
758 # Test entity handling.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000759
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200760 # 1) good entities
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000761
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200762 e = ET.XML("<document title='&#x8230;'>test</document>")
763 self.assertEqual(serialize(e, encoding="us-ascii"),
764 b'<document title="&#33328;">test</document>')
765 self.serialize_check(e, '<document title="\u8230">test</document>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000766
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200767 # 2) bad entities
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000768
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200769 with self.assertRaises(ET.ParseError) as cm:
770 ET.XML("<document>&entity;</document>")
771 self.assertEqual(str(cm.exception),
772 'undefined entity: line 1, column 10')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000773
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200774 with self.assertRaises(ET.ParseError) as cm:
775 ET.XML(ENTITY_XML)
776 self.assertEqual(str(cm.exception),
777 'undefined entity &entity;: line 5, column 10')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000778
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200779 # 3) custom entity
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000780
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200781 parser = ET.XMLParser()
782 parser.entity["entity"] = "text"
783 parser.feed(ENTITY_XML)
784 root = parser.close()
785 self.serialize_check(root, '<document>text</document>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000786
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200787 def test_namespace(self):
788 # Test namespace issues.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000789
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200790 # 1) xml namespace
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000791
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200792 elem = ET.XML("<tag xml:lang='en' />")
793 self.serialize_check(elem, '<tag xml:lang="en" />') # 1.1
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000794
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200795 # 2) other "well-known" namespaces
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000796
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200797 elem = ET.XML("<rdf:RDF xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#' />")
798 self.serialize_check(elem,
799 '<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" />') # 2.1
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000800
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200801 elem = ET.XML("<html:html xmlns:html='http://www.w3.org/1999/xhtml' />")
802 self.serialize_check(elem,
803 '<html:html xmlns:html="http://www.w3.org/1999/xhtml" />') # 2.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000804
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200805 elem = ET.XML("<soap:Envelope xmlns:soap='http://schemas.xmlsoap.org/soap/envelope' />")
806 self.serialize_check(elem,
807 '<ns0:Envelope xmlns:ns0="http://schemas.xmlsoap.org/soap/envelope" />') # 2.3
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000808
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200809 # 3) unknown namespaces
810 elem = ET.XML(SAMPLE_XML_NS)
811 self.serialize_check(elem,
812 '<ns0:body xmlns:ns0="http://effbot.org/ns">\n'
813 ' <ns0:tag>text</ns0:tag>\n'
814 ' <ns0:tag />\n'
815 ' <ns0:section>\n'
816 ' <ns0:tag>subtext</ns0:tag>\n'
817 ' </ns0:section>\n'
818 '</ns0:body>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000819
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200820 def test_qname(self):
821 # Test QName handling.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000822
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200823 # 1) decorated tags
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000824
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200825 elem = ET.Element("{uri}tag")
826 self.serialize_check(elem, '<ns0:tag xmlns:ns0="uri" />') # 1.1
827 elem = ET.Element(ET.QName("{uri}tag"))
828 self.serialize_check(elem, '<ns0:tag xmlns:ns0="uri" />') # 1.2
829 elem = ET.Element(ET.QName("uri", "tag"))
830 self.serialize_check(elem, '<ns0:tag xmlns:ns0="uri" />') # 1.3
831 elem = ET.Element(ET.QName("uri", "tag"))
832 subelem = ET.SubElement(elem, ET.QName("uri", "tag1"))
833 subelem = ET.SubElement(elem, ET.QName("uri", "tag2"))
834 self.serialize_check(elem,
835 '<ns0:tag xmlns:ns0="uri"><ns0:tag1 /><ns0:tag2 /></ns0:tag>') # 1.4
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000836
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200837 # 2) decorated attributes
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000838
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200839 elem.clear()
840 elem.attrib["{uri}key"] = "value"
841 self.serialize_check(elem,
842 '<ns0:tag xmlns:ns0="uri" ns0:key="value" />') # 2.1
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000843
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200844 elem.clear()
845 elem.attrib[ET.QName("{uri}key")] = "value"
846 self.serialize_check(elem,
847 '<ns0:tag xmlns:ns0="uri" ns0:key="value" />') # 2.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000848
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200849 # 3) decorated values are not converted by default, but the
850 # QName wrapper can be used for values
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000851
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200852 elem.clear()
853 elem.attrib["{uri}key"] = "{uri}value"
854 self.serialize_check(elem,
855 '<ns0:tag xmlns:ns0="uri" ns0:key="{uri}value" />') # 3.1
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000856
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200857 elem.clear()
858 elem.attrib["{uri}key"] = ET.QName("{uri}value")
859 self.serialize_check(elem,
860 '<ns0:tag xmlns:ns0="uri" ns0:key="ns0:value" />') # 3.2
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000861
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200862 elem.clear()
863 subelem = ET.Element("tag")
864 subelem.attrib["{uri1}key"] = ET.QName("{uri2}value")
865 elem.append(subelem)
866 elem.append(subelem)
867 self.serialize_check(elem,
868 '<ns0:tag xmlns:ns0="uri" xmlns:ns1="uri1" xmlns:ns2="uri2">'
869 '<tag ns1:key="ns2:value" />'
870 '<tag ns1:key="ns2:value" />'
871 '</ns0:tag>') # 3.3
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000872
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200873 # 4) Direct QName tests
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000874
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200875 self.assertEqual(str(ET.QName('ns', 'tag')), '{ns}tag')
876 self.assertEqual(str(ET.QName('{ns}tag')), '{ns}tag')
877 q1 = ET.QName('ns', 'tag')
878 q2 = ET.QName('ns', 'tag')
879 self.assertEqual(q1, q2)
880 q2 = ET.QName('ns', 'other-tag')
881 self.assertNotEqual(q1, q2)
882 self.assertNotEqual(q1, 'ns:tag')
883 self.assertEqual(q1, '{ns}tag')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000884
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200885 def test_doctype_public(self):
886 # Test PUBLIC doctype.
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000887
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200888 elem = ET.XML('<!DOCTYPE html PUBLIC'
889 ' "-//W3C//DTD XHTML 1.0 Transitional//EN"'
890 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">'
891 '<html>text</html>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000892
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200893 def test_xpath_tokenizer(self):
894 # Test the XPath tokenizer.
895 from xml.etree import ElementPath
896 def check(p, expected):
897 self.assertEqual([op or tag
898 for op, tag in ElementPath.xpath_tokenizer(p)],
899 expected)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000900
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200901 # tests from the xml specification
902 check("*", ['*'])
903 check("text()", ['text', '()'])
904 check("@name", ['@', 'name'])
905 check("@*", ['@', '*'])
906 check("para[1]", ['para', '[', '1', ']'])
907 check("para[last()]", ['para', '[', 'last', '()', ']'])
908 check("*/para", ['*', '/', 'para'])
909 check("/doc/chapter[5]/section[2]",
910 ['/', 'doc', '/', 'chapter', '[', '5', ']',
911 '/', 'section', '[', '2', ']'])
912 check("chapter//para", ['chapter', '//', 'para'])
913 check("//para", ['//', 'para'])
914 check("//olist/item", ['//', 'olist', '/', 'item'])
915 check(".", ['.'])
916 check(".//para", ['.', '//', 'para'])
917 check("..", ['..'])
918 check("../@lang", ['..', '/', '@', 'lang'])
919 check("chapter[title]", ['chapter', '[', 'title', ']'])
920 check("employee[@secretary and @assistant]", ['employee',
921 '[', '@', 'secretary', '', 'and', '', '@', 'assistant', ']'])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000922
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200923 # additional tests
924 check("{http://spam}egg", ['{http://spam}egg'])
925 check("./spam.egg", ['.', '/', 'spam.egg'])
926 check(".//{http://spam}egg", ['.', '//', '{http://spam}egg'])
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000927
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200928 def test_processinginstruction(self):
929 # Test ProcessingInstruction directly
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000930
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200931 self.assertEqual(ET.tostring(ET.ProcessingInstruction('test', 'instruction')),
932 b'<?test instruction?>')
933 self.assertEqual(ET.tostring(ET.PI('test', 'instruction')),
934 b'<?test instruction?>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000935
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200936 # Issue #2746
Antoine Pitrou99f69ee2010-02-09 17:25:47 +0000937
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200938 self.assertEqual(ET.tostring(ET.PI('test', '<testing&>')),
939 b'<?test <testing&>?>')
940 self.assertEqual(ET.tostring(ET.PI('test', '<testing&>\xe3'), 'latin-1'),
941 b"<?xml version='1.0' encoding='latin-1'?>\n"
942 b"<?test <testing&>\xe3?>")
Antoine Pitrou99f69ee2010-02-09 17:25:47 +0000943
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +0200944 def test_html_empty_elems_serialization(self):
945 # issue 15970
946 # from http://www.w3.org/TR/html401/index/elements.html
947 for element in ['AREA', 'BASE', 'BASEFONT', 'BR', 'COL', 'FRAME', 'HR',
948 'IMG', 'INPUT', 'ISINDEX', 'LINK', 'META', 'PARAM']:
949 for elem in [element, element.lower()]:
950 expected = '<%s>' % elem
951 serialized = serialize(ET.XML('<%s />' % elem), method='html')
952 self.assertEqual(serialized, expected)
953 serialized = serialize(ET.XML('<%s></%s>' % (elem,elem)),
954 method='html')
955 self.assertEqual(serialized, expected)
Antoine Pitrou99f69ee2010-02-09 17:25:47 +0000956
Fredrik Lundh8911ca3d2005-12-16 22:07:17 +0000957
Eli Benderskyb5869342013-08-30 05:51:20 -0700958class XMLPullParserTest(unittest.TestCase):
Antoine Pitrou5b235d02013-04-18 19:37:06 +0200959
960 def _feed(self, parser, data, chunk_size=None):
961 if chunk_size is None:
Eli Benderskyb5869342013-08-30 05:51:20 -0700962 parser.feed(data)
Antoine Pitrou5b235d02013-04-18 19:37:06 +0200963 else:
964 for i in range(0, len(data), chunk_size):
Eli Benderskyb5869342013-08-30 05:51:20 -0700965 parser.feed(data[i:i+chunk_size])
Antoine Pitrou5b235d02013-04-18 19:37:06 +0200966
967 def assert_event_tags(self, parser, expected):
Eli Benderskyb5869342013-08-30 05:51:20 -0700968 events = parser.read_events()
Antoine Pitrou5b235d02013-04-18 19:37:06 +0200969 self.assertEqual([(action, elem.tag) for action, elem in events],
970 expected)
971
972 def test_simple_xml(self):
973 for chunk_size in (None, 1, 5):
974 with self.subTest(chunk_size=chunk_size):
Eli Benderskyb5869342013-08-30 05:51:20 -0700975 parser = ET.XMLPullParser()
Antoine Pitrou5b235d02013-04-18 19:37:06 +0200976 self.assert_event_tags(parser, [])
977 self._feed(parser, "<!-- comment -->\n", chunk_size)
978 self.assert_event_tags(parser, [])
979 self._feed(parser,
980 "<root>\n <element key='value'>text</element",
981 chunk_size)
982 self.assert_event_tags(parser, [])
983 self._feed(parser, ">\n", chunk_size)
984 self.assert_event_tags(parser, [('end', 'element')])
985 self._feed(parser, "<element>text</element>tail\n", chunk_size)
986 self._feed(parser, "<empty-element/>\n", chunk_size)
987 self.assert_event_tags(parser, [
988 ('end', 'element'),
989 ('end', 'empty-element'),
990 ])
991 self._feed(parser, "</root>\n", chunk_size)
992 self.assert_event_tags(parser, [('end', 'root')])
Nick Coghlan4cc2afa2013-09-28 23:50:35 +1000993 self.assertIsNone(parser.close())
Antoine Pitrou5b235d02013-04-18 19:37:06 +0200994
Eli Benderskyb5869342013-08-30 05:51:20 -0700995 def test_feed_while_iterating(self):
996 parser = ET.XMLPullParser()
997 it = parser.read_events()
Antoine Pitrou5b235d02013-04-18 19:37:06 +0200998 self._feed(parser, "<root>\n <element key='value'>text</element>\n")
999 action, elem = next(it)
1000 self.assertEqual((action, elem.tag), ('end', 'element'))
1001 self._feed(parser, "</root>\n")
1002 action, elem = next(it)
1003 self.assertEqual((action, elem.tag), ('end', 'root'))
1004 with self.assertRaises(StopIteration):
1005 next(it)
1006
1007 def test_simple_xml_with_ns(self):
Eli Benderskyb5869342013-08-30 05:51:20 -07001008 parser = ET.XMLPullParser()
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001009 self.assert_event_tags(parser, [])
1010 self._feed(parser, "<!-- comment -->\n")
1011 self.assert_event_tags(parser, [])
1012 self._feed(parser, "<root xmlns='namespace'>\n")
1013 self.assert_event_tags(parser, [])
1014 self._feed(parser, "<element key='value'>text</element")
1015 self.assert_event_tags(parser, [])
1016 self._feed(parser, ">\n")
1017 self.assert_event_tags(parser, [('end', '{namespace}element')])
1018 self._feed(parser, "<element>text</element>tail\n")
1019 self._feed(parser, "<empty-element/>\n")
1020 self.assert_event_tags(parser, [
1021 ('end', '{namespace}element'),
1022 ('end', '{namespace}empty-element'),
1023 ])
1024 self._feed(parser, "</root>\n")
1025 self.assert_event_tags(parser, [('end', '{namespace}root')])
Nick Coghlan4cc2afa2013-09-28 23:50:35 +10001026 self.assertIsNone(parser.close())
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001027
Eli Bendersky3a4fbd82013-05-19 09:01:49 -07001028 def test_ns_events(self):
Eli Benderskyb5869342013-08-30 05:51:20 -07001029 parser = ET.XMLPullParser(events=('start-ns', 'end-ns'))
Eli Bendersky3a4fbd82013-05-19 09:01:49 -07001030 self._feed(parser, "<!-- comment -->\n")
1031 self._feed(parser, "<root xmlns='namespace'>\n")
1032 self.assertEqual(
Eli Benderskyb5869342013-08-30 05:51:20 -07001033 list(parser.read_events()),
Eli Bendersky3a4fbd82013-05-19 09:01:49 -07001034 [('start-ns', ('', 'namespace'))])
1035 self._feed(parser, "<element key='value'>text</element")
1036 self._feed(parser, ">\n")
1037 self._feed(parser, "<element>text</element>tail\n")
1038 self._feed(parser, "<empty-element/>\n")
1039 self._feed(parser, "</root>\n")
Eli Benderskyb5869342013-08-30 05:51:20 -07001040 self.assertEqual(list(parser.read_events()), [('end-ns', None)])
Nick Coghlan4cc2afa2013-09-28 23:50:35 +10001041 self.assertIsNone(parser.close())
Eli Bendersky3a4fbd82013-05-19 09:01:49 -07001042
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001043 def test_events(self):
Eli Benderskyb5869342013-08-30 05:51:20 -07001044 parser = ET.XMLPullParser(events=())
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001045 self._feed(parser, "<root/>\n")
1046 self.assert_event_tags(parser, [])
1047
Eli Benderskyb5869342013-08-30 05:51:20 -07001048 parser = ET.XMLPullParser(events=('start', 'end'))
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001049 self._feed(parser, "<!-- comment -->\n")
1050 self.assert_event_tags(parser, [])
1051 self._feed(parser, "<root>\n")
1052 self.assert_event_tags(parser, [('start', 'root')])
1053 self._feed(parser, "<element key='value'>text</element")
1054 self.assert_event_tags(parser, [('start', 'element')])
1055 self._feed(parser, ">\n")
1056 self.assert_event_tags(parser, [('end', 'element')])
1057 self._feed(parser,
1058 "<element xmlns='foo'>text<empty-element/></element>tail\n")
1059 self.assert_event_tags(parser, [
1060 ('start', '{foo}element'),
1061 ('start', '{foo}empty-element'),
1062 ('end', '{foo}empty-element'),
1063 ('end', '{foo}element'),
1064 ])
1065 self._feed(parser, "</root>")
Nick Coghlan4cc2afa2013-09-28 23:50:35 +10001066 self.assertIsNone(parser.close())
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001067 self.assert_event_tags(parser, [('end', 'root')])
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001068
Eli Benderskyb5869342013-08-30 05:51:20 -07001069 parser = ET.XMLPullParser(events=('start',))
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001070 self._feed(parser, "<!-- comment -->\n")
1071 self.assert_event_tags(parser, [])
1072 self._feed(parser, "<root>\n")
1073 self.assert_event_tags(parser, [('start', 'root')])
1074 self._feed(parser, "<element key='value'>text</element")
1075 self.assert_event_tags(parser, [('start', 'element')])
1076 self._feed(parser, ">\n")
1077 self.assert_event_tags(parser, [])
1078 self._feed(parser,
1079 "<element xmlns='foo'>text<empty-element/></element>tail\n")
1080 self.assert_event_tags(parser, [
1081 ('start', '{foo}element'),
1082 ('start', '{foo}empty-element'),
1083 ])
1084 self._feed(parser, "</root>")
Nick Coghlan4cc2afa2013-09-28 23:50:35 +10001085 self.assertIsNone(parser.close())
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001086
Eli Bendersky3a4fbd82013-05-19 09:01:49 -07001087 def test_events_sequence(self):
1088 # Test that events can be some sequence that's not just a tuple or list
1089 eventset = {'end', 'start'}
Eli Benderskyb5869342013-08-30 05:51:20 -07001090 parser = ET.XMLPullParser(events=eventset)
Eli Bendersky3a4fbd82013-05-19 09:01:49 -07001091 self._feed(parser, "<foo>bar</foo>")
1092 self.assert_event_tags(parser, [('start', 'foo'), ('end', 'foo')])
1093
1094 class DummyIter:
1095 def __init__(self):
1096 self.events = iter(['start', 'end', 'start-ns'])
1097 def __iter__(self):
1098 return self
1099 def __next__(self):
1100 return next(self.events)
1101
Eli Benderskyb5869342013-08-30 05:51:20 -07001102 parser = ET.XMLPullParser(events=DummyIter())
Eli Bendersky3a4fbd82013-05-19 09:01:49 -07001103 self._feed(parser, "<foo>bar</foo>")
1104 self.assert_event_tags(parser, [('start', 'foo'), ('end', 'foo')])
1105
1106
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001107 def test_unknown_event(self):
1108 with self.assertRaises(ValueError):
Eli Benderskyb5869342013-08-30 05:51:20 -07001109 ET.XMLPullParser(events=('start', 'end', 'bogus'))
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001110
1111
Armin Rigo9ed73062005-12-14 18:10:45 +00001112#
1113# xinclude tests (samples from appendix C of the xinclude specification)
1114
1115XINCLUDE = {}
1116
1117XINCLUDE["C1.xml"] = """\
1118<?xml version='1.0'?>
1119<document xmlns:xi="http://www.w3.org/2001/XInclude">
1120 <p>120 Mz is adequate for an average home user.</p>
1121 <xi:include href="disclaimer.xml"/>
1122</document>
1123"""
1124
1125XINCLUDE["disclaimer.xml"] = """\
1126<?xml version='1.0'?>
1127<disclaimer>
1128 <p>The opinions represented herein represent those of the individual
1129 and should not be interpreted as official policy endorsed by this
1130 organization.</p>
1131</disclaimer>
1132"""
1133
1134XINCLUDE["C2.xml"] = """\
1135<?xml version='1.0'?>
1136<document xmlns:xi="http://www.w3.org/2001/XInclude">
1137 <p>This document has been accessed
1138 <xi:include href="count.txt" parse="text"/> times.</p>
1139</document>
1140"""
1141
1142XINCLUDE["count.txt"] = "324387"
1143
Florent Xiclunaba8a9862010-08-08 23:08:41 +00001144XINCLUDE["C2b.xml"] = """\
1145<?xml version='1.0'?>
1146<document xmlns:xi="http://www.w3.org/2001/XInclude">
1147 <p>This document has been <em>accessed</em>
1148 <xi:include href="count.txt" parse="text"/> times.</p>
1149</document>
1150"""
1151
Armin Rigo9ed73062005-12-14 18:10:45 +00001152XINCLUDE["C3.xml"] = """\
1153<?xml version='1.0'?>
1154<document xmlns:xi="http://www.w3.org/2001/XInclude">
1155 <p>The following is the source of the "data.xml" resource:</p>
1156 <example><xi:include href="data.xml" parse="text"/></example>
1157</document>
1158"""
1159
1160XINCLUDE["data.xml"] = """\
1161<?xml version='1.0'?>
1162<data>
1163 <item><![CDATA[Brooks & Shields]]></item>
1164</data>
1165"""
1166
1167XINCLUDE["C5.xml"] = """\
1168<?xml version='1.0'?>
1169<div xmlns:xi="http://www.w3.org/2001/XInclude">
1170 <xi:include href="example.txt" parse="text">
1171 <xi:fallback>
1172 <xi:include href="fallback-example.txt" parse="text">
1173 <xi:fallback><a href="mailto:bob@example.org">Report error</a></xi:fallback>
1174 </xi:include>
1175 </xi:fallback>
1176 </xi:include>
1177</div>
1178"""
1179
1180XINCLUDE["default.xml"] = """\
1181<?xml version='1.0'?>
1182<document xmlns:xi="http://www.w3.org/2001/XInclude">
1183 <p>Example.</p>
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001184 <xi:include href="{}"/>
Armin Rigo9ed73062005-12-14 18:10:45 +00001185</document>
Georg Brandl1f7fffb2010-10-15 15:57:45 +00001186""".format(html.escape(SIMPLE_XMLFILE, True))
Armin Rigo9ed73062005-12-14 18:10:45 +00001187
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001188#
1189# badly formatted xi:include tags
1190
1191XINCLUDE_BAD = {}
1192
1193XINCLUDE_BAD["B1.xml"] = """\
1194<?xml version='1.0'?>
1195<document xmlns:xi="http://www.w3.org/2001/XInclude">
1196 <p>120 Mz is adequate for an average home user.</p>
1197 <xi:include href="disclaimer.xml" parse="BAD_TYPE"/>
1198</document>
1199"""
1200
1201XINCLUDE_BAD["B2.xml"] = """\
1202<?xml version='1.0'?>
1203<div xmlns:xi="http://www.w3.org/2001/XInclude">
1204 <xi:fallback></xi:fallback>
1205</div>
1206"""
1207
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001208class XIncludeTest(unittest.TestCase):
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001209
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001210 def xinclude_loader(self, href, parse="xml", encoding=None):
1211 try:
1212 data = XINCLUDE[href]
1213 except KeyError:
1214 raise OSError("resource not found")
1215 if parse == "xml":
1216 data = ET.XML(data)
1217 return data
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001218
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001219 def none_loader(self, href, parser, encoding=None):
1220 return None
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001221
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001222 def _my_loader(self, href, parse):
1223 # Used to avoid a test-dependency problem where the default loader
1224 # of ElementInclude uses the pyET parser for cET tests.
1225 if parse == 'xml':
1226 with open(href, 'rb') as f:
1227 return ET.parse(f).getroot()
1228 else:
1229 return None
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001230
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001231 def test_xinclude_default(self):
1232 from xml.etree import ElementInclude
1233 doc = self.xinclude_loader('default.xml')
1234 ElementInclude.include(doc, self._my_loader)
1235 self.assertEqual(serialize(doc),
1236 '<document>\n'
1237 ' <p>Example.</p>\n'
1238 ' <root>\n'
1239 ' <element key="value">text</element>\n'
1240 ' <element>text</element>tail\n'
1241 ' <empty-element />\n'
1242 '</root>\n'
1243 '</document>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001244
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001245 def test_xinclude(self):
1246 from xml.etree import ElementInclude
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001247
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001248 # Basic inclusion example (XInclude C.1)
1249 document = self.xinclude_loader("C1.xml")
1250 ElementInclude.include(document, self.xinclude_loader)
1251 self.assertEqual(serialize(document),
1252 '<document>\n'
1253 ' <p>120 Mz is adequate for an average home user.</p>\n'
1254 ' <disclaimer>\n'
1255 ' <p>The opinions represented herein represent those of the individual\n'
1256 ' and should not be interpreted as official policy endorsed by this\n'
1257 ' organization.</p>\n'
1258 '</disclaimer>\n'
1259 '</document>') # C1
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001260
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001261 # Textual inclusion example (XInclude C.2)
1262 document = self.xinclude_loader("C2.xml")
1263 ElementInclude.include(document, self.xinclude_loader)
1264 self.assertEqual(serialize(document),
1265 '<document>\n'
1266 ' <p>This document has been accessed\n'
1267 ' 324387 times.</p>\n'
1268 '</document>') # C2
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001269
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001270 # Textual inclusion after sibling element (based on modified XInclude C.2)
1271 document = self.xinclude_loader("C2b.xml")
1272 ElementInclude.include(document, self.xinclude_loader)
1273 self.assertEqual(serialize(document),
1274 '<document>\n'
1275 ' <p>This document has been <em>accessed</em>\n'
1276 ' 324387 times.</p>\n'
1277 '</document>') # C2b
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001278
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001279 # Textual inclusion of XML example (XInclude C.3)
1280 document = self.xinclude_loader("C3.xml")
1281 ElementInclude.include(document, self.xinclude_loader)
1282 self.assertEqual(serialize(document),
1283 '<document>\n'
1284 ' <p>The following is the source of the "data.xml" resource:</p>\n'
1285 " <example>&lt;?xml version='1.0'?&gt;\n"
1286 '&lt;data&gt;\n'
1287 ' &lt;item&gt;&lt;![CDATA[Brooks &amp; Shields]]&gt;&lt;/item&gt;\n'
1288 '&lt;/data&gt;\n'
1289 '</example>\n'
1290 '</document>') # C3
1291
1292 # Fallback example (XInclude C.5)
1293 # Note! Fallback support is not yet implemented
1294 document = self.xinclude_loader("C5.xml")
1295 with self.assertRaises(OSError) as cm:
1296 ElementInclude.include(document, self.xinclude_loader)
1297 self.assertEqual(str(cm.exception), 'resource not found')
1298 self.assertEqual(serialize(document),
1299 '<div xmlns:ns0="http://www.w3.org/2001/XInclude">\n'
1300 ' <ns0:include href="example.txt" parse="text">\n'
1301 ' <ns0:fallback>\n'
1302 ' <ns0:include href="fallback-example.txt" parse="text">\n'
1303 ' <ns0:fallback><a href="mailto:bob@example.org">Report error</a></ns0:fallback>\n'
1304 ' </ns0:include>\n'
1305 ' </ns0:fallback>\n'
1306 ' </ns0:include>\n'
1307 '</div>') # C5
1308
1309 def test_xinclude_failures(self):
1310 from xml.etree import ElementInclude
1311
1312 # Test failure to locate included XML file.
1313 document = ET.XML(XINCLUDE["C1.xml"])
1314 with self.assertRaises(ElementInclude.FatalIncludeError) as cm:
1315 ElementInclude.include(document, loader=self.none_loader)
1316 self.assertEqual(str(cm.exception),
1317 "cannot load 'disclaimer.xml' as 'xml'")
1318
1319 # Test failure to locate included text file.
1320 document = ET.XML(XINCLUDE["C2.xml"])
1321 with self.assertRaises(ElementInclude.FatalIncludeError) as cm:
1322 ElementInclude.include(document, loader=self.none_loader)
1323 self.assertEqual(str(cm.exception),
1324 "cannot load 'count.txt' as 'text'")
1325
1326 # Test bad parse type.
1327 document = ET.XML(XINCLUDE_BAD["B1.xml"])
1328 with self.assertRaises(ElementInclude.FatalIncludeError) as cm:
1329 ElementInclude.include(document, loader=self.none_loader)
1330 self.assertEqual(str(cm.exception),
1331 "unknown parse type in xi:include tag ('BAD_TYPE')")
1332
1333 # Test xi:fallback outside xi:include.
1334 document = ET.XML(XINCLUDE_BAD["B2.xml"])
1335 with self.assertRaises(ElementInclude.FatalIncludeError) as cm:
1336 ElementInclude.include(document, loader=self.none_loader)
1337 self.assertEqual(str(cm.exception),
1338 "xi:fallback tag must be child of xi:include "
1339 "('{http://www.w3.org/2001/XInclude}fallback')")
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001340
1341# --------------------------------------------------------------------
1342# reported bugs
1343
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001344class BugsTest(unittest.TestCase):
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001345
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001346 def test_bug_xmltoolkit21(self):
1347 # marshaller gives obscure errors for non-string values
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001348
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001349 def check(elem):
1350 with self.assertRaises(TypeError) as cm:
1351 serialize(elem)
1352 self.assertEqual(str(cm.exception),
1353 'cannot serialize 123 (type int)')
Armin Rigo9ed73062005-12-14 18:10:45 +00001354
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001355 elem = ET.Element(123)
1356 check(elem) # tag
Armin Rigo9ed73062005-12-14 18:10:45 +00001357
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001358 elem = ET.Element("elem")
1359 elem.text = 123
1360 check(elem) # text
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001361
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001362 elem = ET.Element("elem")
1363 elem.tail = 123
1364 check(elem) # tail
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001365
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001366 elem = ET.Element("elem")
1367 elem.set(123, "123")
1368 check(elem) # attribute key
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001369
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001370 elem = ET.Element("elem")
1371 elem.set("123", 123)
1372 check(elem) # attribute value
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001373
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001374 def test_bug_xmltoolkit25(self):
1375 # typo in ElementTree.findtext
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001376
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001377 elem = ET.XML(SAMPLE_XML)
1378 tree = ET.ElementTree(elem)
1379 self.assertEqual(tree.findtext("tag"), 'text')
1380 self.assertEqual(tree.findtext("section/tag"), 'subtext')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001381
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001382 def test_bug_xmltoolkit28(self):
1383 # .//tag causes exceptions
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001384
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001385 tree = ET.XML("<doc><table><tbody/></table></doc>")
1386 self.assertEqual(summarize_list(tree.findall(".//thead")), [])
1387 self.assertEqual(summarize_list(tree.findall(".//tbody")), ['tbody'])
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001388
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001389 def test_bug_xmltoolkitX1(self):
1390 # dump() doesn't flush the output buffer
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001391
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001392 tree = ET.XML("<doc><table><tbody/></table></doc>")
1393 with support.captured_stdout() as stdout:
1394 ET.dump(tree)
1395 self.assertEqual(stdout.getvalue(), '<doc><table><tbody /></table></doc>\n')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001396
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001397 def test_bug_xmltoolkit39(self):
1398 # non-ascii element and attribute names doesn't work
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001399
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001400 tree = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?><t\xe4g />")
1401 self.assertEqual(ET.tostring(tree, "utf-8"), b'<t\xc3\xa4g />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001402
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001403 tree = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?>"
1404 b"<tag \xe4ttr='v&#228;lue' />")
1405 self.assertEqual(tree.attrib, {'\xe4ttr': 'v\xe4lue'})
1406 self.assertEqual(ET.tostring(tree, "utf-8"),
1407 b'<tag \xc3\xa4ttr="v\xc3\xa4lue" />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001408
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001409 tree = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?>"
1410 b'<t\xe4g>text</t\xe4g>')
1411 self.assertEqual(ET.tostring(tree, "utf-8"),
1412 b'<t\xc3\xa4g>text</t\xc3\xa4g>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001413
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001414 tree = ET.Element("t\u00e4g")
1415 self.assertEqual(ET.tostring(tree, "utf-8"), b'<t\xc3\xa4g />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001416
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001417 tree = ET.Element("tag")
1418 tree.set("\u00e4ttr", "v\u00e4lue")
1419 self.assertEqual(ET.tostring(tree, "utf-8"),
1420 b'<tag \xc3\xa4ttr="v\xc3\xa4lue" />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001421
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001422 def test_bug_xmltoolkit54(self):
1423 # problems handling internally defined entities
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001424
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001425 e = ET.XML("<!DOCTYPE doc [<!ENTITY ldots '&#x8230;'>]>"
1426 '<doc>&ldots;</doc>')
1427 self.assertEqual(serialize(e, encoding="us-ascii"),
1428 b'<doc>&#33328;</doc>')
1429 self.assertEqual(serialize(e), '<doc>\u8230</doc>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001430
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001431 def test_bug_xmltoolkit55(self):
1432 # make sure we're reporting the first error, not the last
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001433
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001434 with self.assertRaises(ET.ParseError) as cm:
1435 ET.XML(b"<!DOCTYPE doc SYSTEM 'doc.dtd'>"
1436 b'<doc>&ldots;&ndots;&rdots;</doc>')
1437 self.assertEqual(str(cm.exception),
1438 'undefined entity &ldots;: line 1, column 36')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001439
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001440 def test_bug_xmltoolkit60(self):
1441 # Handle crash in stream source.
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001442
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001443 class ExceptionFile:
1444 def read(self, x):
1445 raise OSError
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001446
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001447 self.assertRaises(OSError, ET.parse, ExceptionFile())
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001448
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001449 def test_bug_xmltoolkit62(self):
1450 # Don't crash when using custom entities.
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001451
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001452 ENTITIES = {'rsquo': '\u2019', 'lsquo': '\u2018'}
Eli Benderskyc4e98a62013-05-19 09:24:43 -07001453 parser = ET.XMLParser()
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001454 parser.entity.update(ENTITIES)
1455 parser.feed("""<?xml version="1.0" encoding="UTF-8"?>
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001456<!DOCTYPE patent-application-publication SYSTEM "pap-v15-2001-01-31.dtd" []>
1457<patent-application-publication>
1458<subdoc-abstract>
1459<paragraph id="A-0001" lvl="0">A new cultivar of Begonia plant named &lsquo;BCT9801BEG&rsquo;.</paragraph>
1460</subdoc-abstract>
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001461</patent-application-publication>""")
1462 t = parser.close()
1463 self.assertEqual(t.find('.//paragraph').text,
1464 'A new cultivar of Begonia plant named \u2018BCT9801BEG\u2019.')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001465
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001466 def test_bug_xmltoolkit63(self):
1467 # Check reference leak.
1468 def xmltoolkit63():
1469 tree = ET.TreeBuilder()
1470 tree.start("tag", {})
1471 tree.data("text")
1472 tree.end("tag")
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001473
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001474 xmltoolkit63()
1475 count = sys.getrefcount(None)
1476 for i in range(1000):
1477 xmltoolkit63()
1478 self.assertEqual(sys.getrefcount(None), count)
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001479
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001480 def test_bug_200708_newline(self):
1481 # Preserve newlines in attributes.
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001482
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001483 e = ET.Element('SomeTag', text="def _f():\n return 3\n")
1484 self.assertEqual(ET.tostring(e),
1485 b'<SomeTag text="def _f():&#10; return 3&#10;" />')
1486 self.assertEqual(ET.XML(ET.tostring(e)).get("text"),
1487 'def _f():\n return 3\n')
1488 self.assertEqual(ET.tostring(ET.XML(ET.tostring(e))),
1489 b'<SomeTag text="def _f():&#10; return 3&#10;" />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001490
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001491 def test_bug_200708_close(self):
1492 # Test default builder.
1493 parser = ET.XMLParser() # default
1494 parser.feed("<element>some text</element>")
1495 self.assertEqual(parser.close().tag, 'element')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001496
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001497 # Test custom builder.
1498 class EchoTarget:
1499 def close(self):
1500 return ET.Element("element") # simulate root
1501 parser = ET.XMLParser(EchoTarget())
1502 parser.feed("<element>some text</element>")
1503 self.assertEqual(parser.close().tag, 'element')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001504
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001505 def test_bug_200709_default_namespace(self):
1506 e = ET.Element("{default}elem")
1507 s = ET.SubElement(e, "{default}elem")
1508 self.assertEqual(serialize(e, default_namespace="default"), # 1
1509 '<elem xmlns="default"><elem /></elem>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001510
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001511 e = ET.Element("{default}elem")
1512 s = ET.SubElement(e, "{default}elem")
1513 s = ET.SubElement(e, "{not-default}elem")
1514 self.assertEqual(serialize(e, default_namespace="default"), # 2
1515 '<elem xmlns="default" xmlns:ns1="not-default">'
1516 '<elem />'
1517 '<ns1:elem />'
1518 '</elem>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001519
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001520 e = ET.Element("{default}elem")
1521 s = ET.SubElement(e, "{default}elem")
1522 s = ET.SubElement(e, "elem") # unprefixed name
1523 with self.assertRaises(ValueError) as cm:
1524 serialize(e, default_namespace="default") # 3
1525 self.assertEqual(str(cm.exception),
1526 'cannot use non-qualified names with default_namespace option')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001527
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001528 def test_bug_200709_register_namespace(self):
1529 e = ET.Element("{http://namespace.invalid/does/not/exist/}title")
1530 self.assertEqual(ET.tostring(e),
1531 b'<ns0:title xmlns:ns0="http://namespace.invalid/does/not/exist/" />')
1532 ET.register_namespace("foo", "http://namespace.invalid/does/not/exist/")
1533 e = ET.Element("{http://namespace.invalid/does/not/exist/}title")
1534 self.assertEqual(ET.tostring(e),
1535 b'<foo:title xmlns:foo="http://namespace.invalid/does/not/exist/" />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001536
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001537 # And the Dublin Core namespace is in the default list:
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001538
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001539 e = ET.Element("{http://purl.org/dc/elements/1.1/}title")
1540 self.assertEqual(ET.tostring(e),
1541 b'<dc:title xmlns:dc="http://purl.org/dc/elements/1.1/" />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001542
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001543 def test_bug_200709_element_comment(self):
1544 # Not sure if this can be fixed, really (since the serializer needs
1545 # ET.Comment, not cET.comment).
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001546
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001547 a = ET.Element('a')
1548 a.append(ET.Comment('foo'))
1549 self.assertEqual(a[0].tag, ET.Comment)
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001550
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001551 a = ET.Element('a')
1552 a.append(ET.PI('foo'))
1553 self.assertEqual(a[0].tag, ET.PI)
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001554
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001555 def test_bug_200709_element_insert(self):
1556 a = ET.Element('a')
1557 b = ET.SubElement(a, 'b')
1558 c = ET.SubElement(a, 'c')
1559 d = ET.Element('d')
1560 a.insert(0, d)
1561 self.assertEqual(summarize_list(a), ['d', 'b', 'c'])
1562 a.insert(-1, d)
1563 self.assertEqual(summarize_list(a), ['d', 'b', 'd', 'c'])
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001564
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001565 def test_bug_200709_iter_comment(self):
1566 a = ET.Element('a')
1567 b = ET.SubElement(a, 'b')
1568 comment_b = ET.Comment("TEST-b")
1569 b.append(comment_b)
1570 self.assertEqual(summarize_list(a.iter(ET.Comment)), [ET.Comment])
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001571
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001572 # --------------------------------------------------------------------
1573 # reported on bugs.python.org
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001574
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001575 def test_bug_1534630(self):
1576 bob = ET.TreeBuilder()
1577 e = bob.data("data")
1578 e = bob.start("tag", {})
1579 e = bob.end("tag")
1580 e = bob.close()
1581 self.assertEqual(serialize(e), '<tag />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001582
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001583 def test_issue6233(self):
1584 e = ET.XML(b"<?xml version='1.0' encoding='utf-8'?>"
1585 b'<body>t\xc3\xa3g</body>')
1586 self.assertEqual(ET.tostring(e, 'ascii'),
1587 b"<?xml version='1.0' encoding='ascii'?>\n"
1588 b'<body>t&#227;g</body>')
1589 e = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?>"
1590 b'<body>t\xe3g</body>')
1591 self.assertEqual(ET.tostring(e, 'ascii'),
1592 b"<?xml version='1.0' encoding='ascii'?>\n"
1593 b'<body>t&#227;g</body>')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001594
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001595 def test_issue3151(self):
1596 e = ET.XML('<prefix:localname xmlns:prefix="${stuff}"/>')
1597 self.assertEqual(e.tag, '{${stuff}}localname')
1598 t = ET.ElementTree(e)
1599 self.assertEqual(ET.tostring(e), b'<ns0:localname xmlns:ns0="${stuff}" />')
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001600
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001601 def test_issue6565(self):
1602 elem = ET.XML("<body><tag/></body>")
1603 self.assertEqual(summarize_list(elem), ['tag'])
1604 newelem = ET.XML(SAMPLE_XML)
1605 elem[:] = newelem[:]
1606 self.assertEqual(summarize_list(elem), ['tag', 'tag', 'section'])
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001607
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001608 def test_issue10777(self):
1609 # Registering a namespace twice caused a "dictionary changed size during
1610 # iteration" bug.
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001611
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001612 ET.register_namespace('test10777', 'http://myuri/')
1613 ET.register_namespace('test10777', 'http://myuri/')
Georg Brandl90b20672010-12-28 10:38:33 +00001614
Antoine Pitrou5b235d02013-04-18 19:37:06 +02001615
Florent Xiclunaf15351d2010-03-13 23:24:31 +00001616# --------------------------------------------------------------------
1617
1618
Eli Bendersky698bdb22013-01-10 06:01:06 -08001619class BasicElementTest(ElementTestCase, unittest.TestCase):
Eli Bendersky396e8fc2012-03-23 14:24:20 +02001620 def test_augmentation_type_errors(self):
1621 e = ET.Element('joe')
1622 self.assertRaises(TypeError, e.append, 'b')
1623 self.assertRaises(TypeError, e.extend, [ET.Element('bar'), 'foo'])
1624 self.assertRaises(TypeError, e.insert, 0, 'foo')
Florent Xicluna41fe6152010-04-02 18:52:12 +00001625
Eli Bendersky0192ba32012-03-30 16:38:33 +03001626 def test_cyclic_gc(self):
Eli Benderskya5e82202012-03-31 13:55:38 +03001627 class Dummy:
1628 pass
Eli Bendersky0192ba32012-03-30 16:38:33 +03001629
Eli Benderskya5e82202012-03-31 13:55:38 +03001630 # Test the shortest cycle: d->element->d
1631 d = Dummy()
1632 d.dummyref = ET.Element('joe', attr=d)
1633 wref = weakref.ref(d)
1634 del d
1635 gc_collect()
1636 self.assertIsNone(wref())
Eli Bendersky0192ba32012-03-30 16:38:33 +03001637
Eli Benderskyebf37a22012-04-03 22:02:37 +03001638 # A longer cycle: d->e->e2->d
1639 e = ET.Element('joe')
1640 d = Dummy()
1641 d.dummyref = e
1642 wref = weakref.ref(d)
1643 e2 = ET.SubElement(e, 'foo', attr=d)
1644 del d, e, e2
1645 gc_collect()
1646 self.assertIsNone(wref())
1647
1648 # A cycle between Element objects as children of one another
1649 # e1->e2->e3->e1
1650 e1 = ET.Element('e1')
1651 e2 = ET.Element('e2')
1652 e3 = ET.Element('e3')
1653 e1.append(e2)
1654 e2.append(e2)
1655 e3.append(e1)
1656 wref = weakref.ref(e1)
1657 del e1, e2, e3
1658 gc_collect()
1659 self.assertIsNone(wref())
1660
1661 def test_weakref(self):
1662 flag = False
1663 def wref_cb(w):
1664 nonlocal flag
1665 flag = True
1666 e = ET.Element('e')
1667 wref = weakref.ref(e, wref_cb)
1668 self.assertEqual(wref().tag, 'e')
1669 del e
1670 self.assertEqual(flag, True)
1671 self.assertEqual(wref(), None)
1672
Eli Benderskya8736902013-01-05 06:26:39 -08001673 def test_get_keyword_args(self):
1674 e1 = ET.Element('foo' , x=1, y=2, z=3)
1675 self.assertEqual(e1.get('x', default=7), 1)
1676 self.assertEqual(e1.get('w', default=7), 7)
1677
Eli Bendersky7ec45f72012-12-30 06:17:49 -08001678 def test_pickle(self):
Eli Bendersky698bdb22013-01-10 06:01:06 -08001679 # issue #16076: the C implementation wasn't pickleable.
Serhiy Storchakabad12572014-12-15 14:03:42 +02001680 for proto in range(2, pickle.HIGHEST_PROTOCOL + 1):
1681 for dumper, loader in product(self.modules, repeat=2):
1682 e = dumper.Element('foo', bar=42)
1683 e.text = "text goes here"
1684 e.tail = "opposite of head"
1685 dumper.SubElement(e, 'child').append(dumper.Element('grandchild'))
1686 e.append(dumper.Element('child'))
1687 e.findall('.//grandchild')[0].set('attr', 'other value')
Eli Bendersky7ec45f72012-12-30 06:17:49 -08001688
Serhiy Storchakabad12572014-12-15 14:03:42 +02001689 e2 = self.pickleRoundTrip(e, 'xml.etree.ElementTree',
1690 dumper, loader, proto)
Eli Bendersky698bdb22013-01-10 06:01:06 -08001691
Serhiy Storchakabad12572014-12-15 14:03:42 +02001692 self.assertEqual(e2.tag, 'foo')
1693 self.assertEqual(e2.attrib['bar'], 42)
1694 self.assertEqual(len(e2), 2)
1695 self.assertEqualElements(e, e2)
Eli Bendersky396e8fc2012-03-23 14:24:20 +02001696
Eli Benderskydd3661e2013-09-13 06:24:25 -07001697 def test_pickle_issue18997(self):
Serhiy Storchakabad12572014-12-15 14:03:42 +02001698 for proto in range(2, pickle.HIGHEST_PROTOCOL + 1):
1699 for dumper, loader in product(self.modules, repeat=2):
1700 XMLTEXT = """<?xml version="1.0"?>
1701 <group><dogs>4</dogs>
1702 </group>"""
1703 e1 = dumper.fromstring(XMLTEXT)
1704 if hasattr(e1, '__getstate__'):
1705 self.assertEqual(e1.__getstate__()['tag'], 'group')
1706 e2 = self.pickleRoundTrip(e1, 'xml.etree.ElementTree',
1707 dumper, loader, proto)
1708 self.assertEqual(e2.tag, 'group')
1709 self.assertEqual(e2[0].tag, 'dogs')
Eli Benderskydd3661e2013-09-13 06:24:25 -07001710
Eli Bendersky23687042013-02-26 05:53:23 -08001711
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02001712class ElementTreeTypeTest(unittest.TestCase):
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01001713 def test_istype(self):
1714 self.assertIsInstance(ET.ParseError, type)
1715 self.assertIsInstance(ET.QName, type)
1716 self.assertIsInstance(ET.ElementTree, type)
Eli Bendersky092af1f2012-03-04 07:14:03 +02001717 self.assertIsInstance(ET.Element, type)
Eli Bendersky64d11e62012-06-15 07:42:50 +03001718 self.assertIsInstance(ET.TreeBuilder, type)
1719 self.assertIsInstance(ET.XMLParser, type)
Eli Bendersky092af1f2012-03-04 07:14:03 +02001720
1721 def test_Element_subclass_trivial(self):
1722 class MyElement(ET.Element):
1723 pass
1724
1725 mye = MyElement('foo')
1726 self.assertIsInstance(mye, ET.Element)
1727 self.assertIsInstance(mye, MyElement)
1728 self.assertEqual(mye.tag, 'foo')
1729
Eli Benderskyb20df952012-05-20 06:33:29 +03001730 # test that attribute assignment works (issue 14849)
1731 mye.text = "joe"
1732 self.assertEqual(mye.text, "joe")
1733
Eli Bendersky092af1f2012-03-04 07:14:03 +02001734 def test_Element_subclass_constructor(self):
1735 class MyElement(ET.Element):
1736 def __init__(self, tag, attrib={}, **extra):
1737 super(MyElement, self).__init__(tag + '__', attrib, **extra)
1738
1739 mye = MyElement('foo', {'a': 1, 'b': 2}, c=3, d=4)
1740 self.assertEqual(mye.tag, 'foo__')
1741 self.assertEqual(sorted(mye.items()),
1742 [('a', 1), ('b', 2), ('c', 3), ('d', 4)])
1743
1744 def test_Element_subclass_new_method(self):
1745 class MyElement(ET.Element):
1746 def newmethod(self):
1747 return self.tag
1748
1749 mye = MyElement('joe')
1750 self.assertEqual(mye.newmethod(), 'joe')
Eli Benderskyda578192012-02-16 06:52:39 +02001751
Eli Benderskyceab1a92013-01-12 07:42:46 -08001752
1753class ElementFindTest(unittest.TestCase):
1754 def test_find_simple(self):
1755 e = ET.XML(SAMPLE_XML)
1756 self.assertEqual(e.find('tag').tag, 'tag')
1757 self.assertEqual(e.find('section/tag').tag, 'tag')
1758 self.assertEqual(e.find('./tag').tag, 'tag')
1759
1760 e[2] = ET.XML(SAMPLE_SECTION)
1761 self.assertEqual(e.find('section/nexttag').tag, 'nexttag')
1762
1763 self.assertEqual(e.findtext('./tag'), 'text')
1764 self.assertEqual(e.findtext('section/tag'), 'subtext')
1765
1766 # section/nexttag is found but has no text
1767 self.assertEqual(e.findtext('section/nexttag'), '')
1768 self.assertEqual(e.findtext('section/nexttag', 'default'), '')
1769
1770 # tog doesn't exist and 'default' kicks in
1771 self.assertIsNone(e.findtext('tog'))
1772 self.assertEqual(e.findtext('tog', 'default'), 'default')
1773
Eli Bendersky25771b32013-01-13 05:26:07 -08001774 # Issue #16922
1775 self.assertEqual(ET.XML('<tag><empty /></tag>').findtext('empty'), '')
1776
Eli Benderskya80f7612013-01-22 06:12:54 -08001777 def test_find_xpath(self):
1778 LINEAR_XML = '''
1779 <body>
1780 <tag class='a'/>
1781 <tag class='b'/>
1782 <tag class='c'/>
1783 <tag class='d'/>
1784 </body>'''
1785 e = ET.XML(LINEAR_XML)
1786
1787 # Test for numeric indexing and last()
1788 self.assertEqual(e.find('./tag[1]').attrib['class'], 'a')
1789 self.assertEqual(e.find('./tag[2]').attrib['class'], 'b')
1790 self.assertEqual(e.find('./tag[last()]').attrib['class'], 'd')
1791 self.assertEqual(e.find('./tag[last()-1]').attrib['class'], 'c')
1792 self.assertEqual(e.find('./tag[last()-2]').attrib['class'], 'b')
1793
Eli Bendersky5c6198b2013-01-24 06:29:26 -08001794 self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[0]')
1795 self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[-1]')
1796 self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[last()-0]')
1797 self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[last()+1]')
1798
Eli Benderskyceab1a92013-01-12 07:42:46 -08001799 def test_findall(self):
1800 e = ET.XML(SAMPLE_XML)
1801 e[2] = ET.XML(SAMPLE_SECTION)
1802 self.assertEqual(summarize_list(e.findall('.')), ['body'])
1803 self.assertEqual(summarize_list(e.findall('tag')), ['tag', 'tag'])
1804 self.assertEqual(summarize_list(e.findall('tog')), [])
1805 self.assertEqual(summarize_list(e.findall('tog/foo')), [])
1806 self.assertEqual(summarize_list(e.findall('*')),
1807 ['tag', 'tag', 'section'])
1808 self.assertEqual(summarize_list(e.findall('.//tag')),
1809 ['tag'] * 4)
1810 self.assertEqual(summarize_list(e.findall('section/tag')), ['tag'])
1811 self.assertEqual(summarize_list(e.findall('section//tag')), ['tag'] * 2)
1812 self.assertEqual(summarize_list(e.findall('section/*')),
1813 ['tag', 'nexttag', 'nextsection'])
1814 self.assertEqual(summarize_list(e.findall('section//*')),
1815 ['tag', 'nexttag', 'nextsection', 'tag'])
1816 self.assertEqual(summarize_list(e.findall('section/.//*')),
1817 ['tag', 'nexttag', 'nextsection', 'tag'])
1818 self.assertEqual(summarize_list(e.findall('*/*')),
1819 ['tag', 'nexttag', 'nextsection'])
1820 self.assertEqual(summarize_list(e.findall('*//*')),
1821 ['tag', 'nexttag', 'nextsection', 'tag'])
1822 self.assertEqual(summarize_list(e.findall('*/tag')), ['tag'])
1823 self.assertEqual(summarize_list(e.findall('*/./tag')), ['tag'])
1824 self.assertEqual(summarize_list(e.findall('./tag')), ['tag'] * 2)
1825 self.assertEqual(summarize_list(e.findall('././tag')), ['tag'] * 2)
1826
1827 self.assertEqual(summarize_list(e.findall('.//tag[@class]')),
1828 ['tag'] * 3)
1829 self.assertEqual(summarize_list(e.findall('.//tag[@class="a"]')),
1830 ['tag'])
1831 self.assertEqual(summarize_list(e.findall('.//tag[@class="b"]')),
1832 ['tag'] * 2)
1833 self.assertEqual(summarize_list(e.findall('.//tag[@id]')),
1834 ['tag'])
1835 self.assertEqual(summarize_list(e.findall('.//section[tag]')),
1836 ['section'])
1837 self.assertEqual(summarize_list(e.findall('.//section[element]')), [])
1838 self.assertEqual(summarize_list(e.findall('../tag')), [])
1839 self.assertEqual(summarize_list(e.findall('section/../tag')),
1840 ['tag'] * 2)
1841 self.assertEqual(e.findall('section//'), e.findall('section//*'))
1842
1843 def test_test_find_with_ns(self):
1844 e = ET.XML(SAMPLE_XML_NS)
1845 self.assertEqual(summarize_list(e.findall('tag')), [])
1846 self.assertEqual(
1847 summarize_list(e.findall("{http://effbot.org/ns}tag")),
1848 ['{http://effbot.org/ns}tag'] * 2)
1849 self.assertEqual(
1850 summarize_list(e.findall(".//{http://effbot.org/ns}tag")),
1851 ['{http://effbot.org/ns}tag'] * 3)
1852
Eli Bendersky2acc5252013-08-03 17:47:47 -07001853 def test_findall_different_nsmaps(self):
1854 root = ET.XML('''
1855 <a xmlns:x="X" xmlns:y="Y">
1856 <x:b><c/></x:b>
1857 <b/>
1858 <c><x:b/><b/></c><y:b/>
1859 </a>''')
1860 nsmap = {'xx': 'X'}
1861 self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 2)
1862 self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2)
1863 nsmap = {'xx': 'Y'}
1864 self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 1)
1865 self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2)
1866
Eli Benderskyceab1a92013-01-12 07:42:46 -08001867 def test_bad_find(self):
1868 e = ET.XML(SAMPLE_XML)
1869 with self.assertRaisesRegex(SyntaxError, 'cannot use absolute path'):
1870 e.findall('/tag')
Eli Benderskyc31f7732013-01-12 07:44:32 -08001871
Eli Benderskyceab1a92013-01-12 07:42:46 -08001872 def test_find_through_ElementTree(self):
1873 e = ET.XML(SAMPLE_XML)
1874 self.assertEqual(ET.ElementTree(e).find('tag').tag, 'tag')
1875 self.assertEqual(ET.ElementTree(e).findtext('tag'), 'text')
1876 self.assertEqual(summarize_list(ET.ElementTree(e).findall('tag')),
1877 ['tag'] * 2)
1878 # this produces a warning
1879 self.assertEqual(summarize_list(ET.ElementTree(e).findall('//tag')),
1880 ['tag'] * 3)
Eli Benderskyc31f7732013-01-12 07:44:32 -08001881
Eli Benderskyceab1a92013-01-12 07:42:46 -08001882
Eli Bendersky64d11e62012-06-15 07:42:50 +03001883class ElementIterTest(unittest.TestCase):
1884 def _ilist(self, elem, tag=None):
1885 return summarize_list(elem.iter(tag))
1886
1887 def test_basic(self):
1888 doc = ET.XML("<html><body>this is a <i>paragraph</i>.</body>..</html>")
1889 self.assertEqual(self._ilist(doc), ['html', 'body', 'i'])
1890 self.assertEqual(self._ilist(doc.find('body')), ['body', 'i'])
1891 self.assertEqual(next(doc.iter()).tag, 'html')
1892 self.assertEqual(''.join(doc.itertext()), 'this is a paragraph...')
1893 self.assertEqual(''.join(doc.find('body').itertext()),
1894 'this is a paragraph.')
1895 self.assertEqual(next(doc.itertext()), 'this is a ')
1896
1897 # iterparse should return an iterator
1898 sourcefile = serialize(doc, to_string=False)
1899 self.assertEqual(next(ET.iterparse(sourcefile))[0], 'end')
1900
Eli Benderskyaaa97802013-01-24 07:15:19 -08001901 # With an explitit parser too (issue #9708)
1902 sourcefile = serialize(doc, to_string=False)
1903 parser = ET.XMLParser(target=ET.TreeBuilder())
1904 self.assertEqual(next(ET.iterparse(sourcefile, parser=parser))[0],
1905 'end')
1906
Eli Bendersky64d11e62012-06-15 07:42:50 +03001907 tree = ET.ElementTree(None)
1908 self.assertRaises(AttributeError, tree.iter)
1909
Eli Benderskye6174ca2013-01-10 06:27:53 -08001910 # Issue #16913
1911 doc = ET.XML("<root>a&amp;<sub>b&amp;</sub>c&amp;</root>")
1912 self.assertEqual(''.join(doc.itertext()), 'a&b&c&')
1913
Eli Bendersky64d11e62012-06-15 07:42:50 +03001914 def test_corners(self):
1915 # single root, no subelements
1916 a = ET.Element('a')
1917 self.assertEqual(self._ilist(a), ['a'])
1918
1919 # one child
1920 b = ET.SubElement(a, 'b')
1921 self.assertEqual(self._ilist(a), ['a', 'b'])
1922
1923 # one child and one grandchild
1924 c = ET.SubElement(b, 'c')
1925 self.assertEqual(self._ilist(a), ['a', 'b', 'c'])
1926
1927 # two children, only first with grandchild
1928 d = ET.SubElement(a, 'd')
1929 self.assertEqual(self._ilist(a), ['a', 'b', 'c', 'd'])
1930
1931 # replace first child by second
1932 a[0] = a[1]
1933 del a[1]
1934 self.assertEqual(self._ilist(a), ['a', 'd'])
1935
1936 def test_iter_by_tag(self):
1937 doc = ET.XML('''
1938 <document>
1939 <house>
1940 <room>bedroom1</room>
1941 <room>bedroom2</room>
1942 </house>
1943 <shed>nothing here
1944 </shed>
1945 <house>
1946 <room>bedroom8</room>
1947 </house>
1948 </document>''')
1949
1950 self.assertEqual(self._ilist(doc, 'room'), ['room'] * 3)
1951 self.assertEqual(self._ilist(doc, 'house'), ['house'] * 2)
1952
Eli Benderskya8736902013-01-05 06:26:39 -08001953 # test that iter also accepts 'tag' as a keyword arg
1954 self.assertEqual(
1955 summarize_list(doc.iter(tag='room')),
1956 ['room'] * 3)
1957
Eli Bendersky64d11e62012-06-15 07:42:50 +03001958 # make sure both tag=None and tag='*' return all tags
1959 all_tags = ['document', 'house', 'room', 'room',
1960 'shed', 'house', 'room']
1961 self.assertEqual(self._ilist(doc), all_tags)
1962 self.assertEqual(self._ilist(doc, '*'), all_tags)
1963
1964
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01001965class TreeBuilderTest(unittest.TestCase):
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01001966 sample1 = ('<!DOCTYPE html PUBLIC'
1967 ' "-//W3C//DTD XHTML 1.0 Transitional//EN"'
1968 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">'
Antoine Pitrouee329312012-10-04 19:53:29 +02001969 '<html>text<div>subtext</div>tail</html>')
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01001970
Eli Bendersky48d358b2012-05-30 17:57:50 +03001971 sample2 = '''<toplevel>sometext</toplevel>'''
1972
Antoine Pitrouee329312012-10-04 19:53:29 +02001973 def _check_sample1_element(self, e):
1974 self.assertEqual(e.tag, 'html')
1975 self.assertEqual(e.text, 'text')
1976 self.assertEqual(e.tail, None)
1977 self.assertEqual(e.attrib, {})
1978 children = list(e)
1979 self.assertEqual(len(children), 1)
1980 child = children[0]
1981 self.assertEqual(child.tag, 'div')
1982 self.assertEqual(child.text, 'subtext')
1983 self.assertEqual(child.tail, 'tail')
1984 self.assertEqual(child.attrib, {})
1985
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01001986 def test_dummy_builder(self):
1987 class BaseDummyBuilder:
1988 def close(self):
1989 return 42
1990
1991 class DummyBuilder(BaseDummyBuilder):
1992 data = start = end = lambda *a: None
1993
1994 parser = ET.XMLParser(target=DummyBuilder())
1995 parser.feed(self.sample1)
1996 self.assertEqual(parser.close(), 42)
1997
1998 parser = ET.XMLParser(target=BaseDummyBuilder())
1999 parser.feed(self.sample1)
2000 self.assertEqual(parser.close(), 42)
2001
2002 parser = ET.XMLParser(target=object())
2003 parser.feed(self.sample1)
2004 self.assertIsNone(parser.close())
2005
Eli Bendersky08231a92013-05-18 15:47:16 -07002006 def test_treebuilder_elementfactory_none(self):
2007 parser = ET.XMLParser(target=ET.TreeBuilder(element_factory=None))
2008 parser.feed(self.sample1)
2009 e = parser.close()
2010 self._check_sample1_element(e)
2011
Eli Bendersky58d548d2012-05-29 15:45:16 +03002012 def test_subclass(self):
2013 class MyTreeBuilder(ET.TreeBuilder):
2014 def foobar(self, x):
2015 return x * 2
2016
2017 tb = MyTreeBuilder()
2018 self.assertEqual(tb.foobar(10), 20)
2019
2020 parser = ET.XMLParser(target=tb)
2021 parser.feed(self.sample1)
2022
2023 e = parser.close()
Antoine Pitrouee329312012-10-04 19:53:29 +02002024 self._check_sample1_element(e)
Eli Bendersky58d548d2012-05-29 15:45:16 +03002025
Eli Bendersky2b711402012-03-16 15:29:50 +02002026 def test_element_factory(self):
Eli Bendersky48d358b2012-05-30 17:57:50 +03002027 lst = []
2028 def myfactory(tag, attrib):
2029 nonlocal lst
2030 lst.append(tag)
2031 return ET.Element(tag, attrib)
2032
2033 tb = ET.TreeBuilder(element_factory=myfactory)
2034 parser = ET.XMLParser(target=tb)
2035 parser.feed(self.sample2)
2036 parser.close()
2037
2038 self.assertEqual(lst, ['toplevel'])
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01002039
Antoine Pitrouee329312012-10-04 19:53:29 +02002040 def _check_element_factory_class(self, cls):
2041 tb = ET.TreeBuilder(element_factory=cls)
2042
2043 parser = ET.XMLParser(target=tb)
2044 parser.feed(self.sample1)
2045 e = parser.close()
2046 self.assertIsInstance(e, cls)
2047 self._check_sample1_element(e)
2048
2049 def test_element_factory_subclass(self):
2050 class MyElement(ET.Element):
2051 pass
2052 self._check_element_factory_class(MyElement)
2053
2054 def test_element_factory_pure_python_subclass(self):
2055 # Mimick SimpleTAL's behaviour (issue #16089): both versions of
2056 # TreeBuilder should be able to cope with a subclass of the
2057 # pure Python Element class.
Eli Bendersky46955b22013-05-19 09:20:50 -07002058 base = ET._Element_Py
Antoine Pitrouee329312012-10-04 19:53:29 +02002059 # Not from a C extension
2060 self.assertEqual(base.__module__, 'xml.etree.ElementTree')
2061 # Force some multiple inheritance with a C class to make things
2062 # more interesting.
2063 class MyElement(base, ValueError):
2064 pass
2065 self._check_element_factory_class(MyElement)
2066
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01002067 def test_doctype(self):
2068 class DoctypeParser:
2069 _doctype = None
2070
2071 def doctype(self, name, pubid, system):
2072 self._doctype = (name, pubid, system)
2073
2074 def close(self):
2075 return self._doctype
2076
2077 parser = ET.XMLParser(target=DoctypeParser())
2078 parser.feed(self.sample1)
2079
2080 self.assertEqual(parser.close(),
2081 ('html', '-//W3C//DTD XHTML 1.0 Transitional//EN',
2082 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'))
2083
Eli Bendersky175fada2012-06-15 08:37:08 +03002084
Eli Bendersky52467b12012-06-01 07:13:08 +03002085class XMLParserTest(unittest.TestCase):
Serhiy Storchaka66d53fa2013-05-22 17:07:51 +03002086 sample1 = b'<file><line>22</line></file>'
2087 sample2 = (b'<!DOCTYPE html PUBLIC'
2088 b' "-//W3C//DTD XHTML 1.0 Transitional//EN"'
2089 b' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">'
2090 b'<html>text</html>')
2091 sample3 = ('<?xml version="1.0" encoding="iso-8859-1"?>\n'
2092 '<money value="$\xa3\u20ac\U0001017b">$\xa3\u20ac\U0001017b</money>')
Eli Bendersky52467b12012-06-01 07:13:08 +03002093
2094 def _check_sample_element(self, e):
2095 self.assertEqual(e.tag, 'file')
2096 self.assertEqual(e[0].tag, 'line')
2097 self.assertEqual(e[0].text, '22')
2098
2099 def test_constructor_args(self):
2100 # Positional args. The first (html) is not supported, but should be
2101 # nevertheless correctly accepted.
2102 parser = ET.XMLParser(None, ET.TreeBuilder(), 'utf-8')
2103 parser.feed(self.sample1)
2104 self._check_sample_element(parser.close())
2105
2106 # Now as keyword args.
Eli Bendersky23687042013-02-26 05:53:23 -08002107 parser2 = ET.XMLParser(encoding='utf-8',
2108 html=[{}],
2109 target=ET.TreeBuilder())
Eli Bendersky52467b12012-06-01 07:13:08 +03002110 parser2.feed(self.sample1)
2111 self._check_sample_element(parser2.close())
2112
2113 def test_subclass(self):
2114 class MyParser(ET.XMLParser):
2115 pass
2116 parser = MyParser()
2117 parser.feed(self.sample1)
2118 self._check_sample_element(parser.close())
2119
Eli Bendersky2b6b73e2012-06-01 11:32:34 +03002120 def test_subclass_doctype(self):
2121 _doctype = None
2122 class MyParserWithDoctype(ET.XMLParser):
2123 def doctype(self, name, pubid, system):
2124 nonlocal _doctype
2125 _doctype = (name, pubid, system)
2126
2127 parser = MyParserWithDoctype()
Serhiy Storchaka66d53fa2013-05-22 17:07:51 +03002128 with self.assertWarns(DeprecationWarning):
2129 parser.feed(self.sample2)
Eli Bendersky2b6b73e2012-06-01 11:32:34 +03002130 parser.close()
2131 self.assertEqual(_doctype,
2132 ('html', '-//W3C//DTD XHTML 1.0 Transitional//EN',
2133 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'))
2134
Serhiy Storchaka66d53fa2013-05-22 17:07:51 +03002135 def test_parse_string(self):
2136 parser = ET.XMLParser(target=ET.TreeBuilder())
2137 parser.feed(self.sample3)
2138 e = parser.close()
2139 self.assertEqual(e.tag, 'money')
2140 self.assertEqual(e.attrib['value'], '$\xa3\u20ac\U0001017b')
2141 self.assertEqual(e.text, '$\xa3\u20ac\U0001017b')
2142
Eli Bendersky52467b12012-06-01 07:13:08 +03002143
Eli Bendersky737b1732012-05-29 06:02:56 +03002144class NamespaceParseTest(unittest.TestCase):
2145 def test_find_with_namespace(self):
2146 nsmap = {'h': 'hello', 'f': 'foo'}
2147 doc = ET.fromstring(SAMPLE_XML_NS_ELEMS)
2148
2149 self.assertEqual(len(doc.findall('{hello}table', nsmap)), 1)
2150 self.assertEqual(len(doc.findall('.//{hello}td', nsmap)), 2)
2151 self.assertEqual(len(doc.findall('.//{foo}name', nsmap)), 1)
2152
2153
Eli Bendersky865756a2012-03-09 13:38:15 +02002154class ElementSlicingTest(unittest.TestCase):
2155 def _elem_tags(self, elemlist):
2156 return [e.tag for e in elemlist]
2157
2158 def _subelem_tags(self, elem):
2159 return self._elem_tags(list(elem))
2160
2161 def _make_elem_with_children(self, numchildren):
2162 """Create an Element with a tag 'a', with the given amount of children
2163 named 'a0', 'a1' ... and so on.
2164
2165 """
2166 e = ET.Element('a')
2167 for i in range(numchildren):
2168 ET.SubElement(e, 'a%s' % i)
2169 return e
2170
2171 def test_getslice_single_index(self):
2172 e = self._make_elem_with_children(10)
2173
2174 self.assertEqual(e[1].tag, 'a1')
2175 self.assertEqual(e[-2].tag, 'a8')
2176
2177 self.assertRaises(IndexError, lambda: e[12])
2178
2179 def test_getslice_range(self):
2180 e = self._make_elem_with_children(6)
2181
2182 self.assertEqual(self._elem_tags(e[3:]), ['a3', 'a4', 'a5'])
2183 self.assertEqual(self._elem_tags(e[3:6]), ['a3', 'a4', 'a5'])
2184 self.assertEqual(self._elem_tags(e[3:16]), ['a3', 'a4', 'a5'])
2185 self.assertEqual(self._elem_tags(e[3:5]), ['a3', 'a4'])
2186 self.assertEqual(self._elem_tags(e[3:-1]), ['a3', 'a4'])
2187 self.assertEqual(self._elem_tags(e[:2]), ['a0', 'a1'])
2188
2189 def test_getslice_steps(self):
2190 e = self._make_elem_with_children(10)
2191
2192 self.assertEqual(self._elem_tags(e[8:10:1]), ['a8', 'a9'])
2193 self.assertEqual(self._elem_tags(e[::3]), ['a0', 'a3', 'a6', 'a9'])
2194 self.assertEqual(self._elem_tags(e[::8]), ['a0', 'a8'])
2195 self.assertEqual(self._elem_tags(e[1::8]), ['a1', 'a9'])
2196
2197 def test_getslice_negative_steps(self):
2198 e = self._make_elem_with_children(4)
2199
2200 self.assertEqual(self._elem_tags(e[::-1]), ['a3', 'a2', 'a1', 'a0'])
2201 self.assertEqual(self._elem_tags(e[::-2]), ['a3', 'a1'])
2202
2203 def test_delslice(self):
2204 e = self._make_elem_with_children(4)
2205 del e[0:2]
2206 self.assertEqual(self._subelem_tags(e), ['a2', 'a3'])
2207
2208 e = self._make_elem_with_children(4)
2209 del e[0:]
2210 self.assertEqual(self._subelem_tags(e), [])
2211
2212 e = self._make_elem_with_children(4)
2213 del e[::-1]
2214 self.assertEqual(self._subelem_tags(e), [])
2215
2216 e = self._make_elem_with_children(4)
2217 del e[::-2]
2218 self.assertEqual(self._subelem_tags(e), ['a0', 'a2'])
2219
2220 e = self._make_elem_with_children(4)
2221 del e[1::2]
2222 self.assertEqual(self._subelem_tags(e), ['a0', 'a2'])
2223
2224 e = self._make_elem_with_children(2)
2225 del e[::2]
2226 self.assertEqual(self._subelem_tags(e), ['a1'])
2227
Eli Benderskyf996e772012-03-16 05:53:30 +02002228
Eli Bendersky00f402b2012-07-15 06:02:22 +03002229class IOTest(unittest.TestCase):
2230 def tearDown(self):
Eli Bendersky23687042013-02-26 05:53:23 -08002231 support.unlink(TESTFN)
Eli Bendersky00f402b2012-07-15 06:02:22 +03002232
2233 def test_encoding(self):
2234 # Test encoding issues.
2235 elem = ET.Element("tag")
2236 elem.text = "abc"
2237 self.assertEqual(serialize(elem), '<tag>abc</tag>')
2238 self.assertEqual(serialize(elem, encoding="utf-8"),
2239 b'<tag>abc</tag>')
2240 self.assertEqual(serialize(elem, encoding="us-ascii"),
2241 b'<tag>abc</tag>')
2242 for enc in ("iso-8859-1", "utf-16", "utf-32"):
2243 self.assertEqual(serialize(elem, encoding=enc),
2244 ("<?xml version='1.0' encoding='%s'?>\n"
2245 "<tag>abc</tag>" % enc).encode(enc))
2246
2247 elem = ET.Element("tag")
2248 elem.text = "<&\"\'>"
2249 self.assertEqual(serialize(elem), '<tag>&lt;&amp;"\'&gt;</tag>')
2250 self.assertEqual(serialize(elem, encoding="utf-8"),
2251 b'<tag>&lt;&amp;"\'&gt;</tag>')
2252 self.assertEqual(serialize(elem, encoding="us-ascii"),
2253 b'<tag>&lt;&amp;"\'&gt;</tag>')
2254 for enc in ("iso-8859-1", "utf-16", "utf-32"):
2255 self.assertEqual(serialize(elem, encoding=enc),
2256 ("<?xml version='1.0' encoding='%s'?>\n"
2257 "<tag>&lt;&amp;\"'&gt;</tag>" % enc).encode(enc))
2258
2259 elem = ET.Element("tag")
2260 elem.attrib["key"] = "<&\"\'>"
2261 self.assertEqual(serialize(elem), '<tag key="&lt;&amp;&quot;\'&gt;" />')
2262 self.assertEqual(serialize(elem, encoding="utf-8"),
2263 b'<tag key="&lt;&amp;&quot;\'&gt;" />')
2264 self.assertEqual(serialize(elem, encoding="us-ascii"),
2265 b'<tag key="&lt;&amp;&quot;\'&gt;" />')
2266 for enc in ("iso-8859-1", "utf-16", "utf-32"):
2267 self.assertEqual(serialize(elem, encoding=enc),
2268 ("<?xml version='1.0' encoding='%s'?>\n"
2269 "<tag key=\"&lt;&amp;&quot;'&gt;\" />" % enc).encode(enc))
2270
2271 elem = ET.Element("tag")
2272 elem.text = '\xe5\xf6\xf6<>'
2273 self.assertEqual(serialize(elem), '<tag>\xe5\xf6\xf6&lt;&gt;</tag>')
2274 self.assertEqual(serialize(elem, encoding="utf-8"),
2275 b'<tag>\xc3\xa5\xc3\xb6\xc3\xb6&lt;&gt;</tag>')
2276 self.assertEqual(serialize(elem, encoding="us-ascii"),
2277 b'<tag>&#229;&#246;&#246;&lt;&gt;</tag>')
2278 for enc in ("iso-8859-1", "utf-16", "utf-32"):
2279 self.assertEqual(serialize(elem, encoding=enc),
2280 ("<?xml version='1.0' encoding='%s'?>\n"
2281 "<tag>Ă¥Ă¶Ă¶&lt;&gt;</tag>" % enc).encode(enc))
2282
2283 elem = ET.Element("tag")
2284 elem.attrib["key"] = '\xe5\xf6\xf6<>'
2285 self.assertEqual(serialize(elem), '<tag key="\xe5\xf6\xf6&lt;&gt;" />')
2286 self.assertEqual(serialize(elem, encoding="utf-8"),
2287 b'<tag key="\xc3\xa5\xc3\xb6\xc3\xb6&lt;&gt;" />')
2288 self.assertEqual(serialize(elem, encoding="us-ascii"),
2289 b'<tag key="&#229;&#246;&#246;&lt;&gt;" />')
2290 for enc in ("iso-8859-1", "utf-16", "utf-16le", "utf-16be", "utf-32"):
2291 self.assertEqual(serialize(elem, encoding=enc),
2292 ("<?xml version='1.0' encoding='%s'?>\n"
2293 "<tag key=\"Ă¥Ă¶Ă¶&lt;&gt;\" />" % enc).encode(enc))
2294
2295 def test_write_to_filename(self):
2296 tree = ET.ElementTree(ET.XML('''<site />'''))
2297 tree.write(TESTFN)
2298 with open(TESTFN, 'rb') as f:
2299 self.assertEqual(f.read(), b'''<site />''')
2300
2301 def test_write_to_text_file(self):
2302 tree = ET.ElementTree(ET.XML('''<site />'''))
2303 with open(TESTFN, 'w', encoding='utf-8') as f:
2304 tree.write(f, encoding='unicode')
2305 self.assertFalse(f.closed)
2306 with open(TESTFN, 'rb') as f:
2307 self.assertEqual(f.read(), b'''<site />''')
2308
2309 def test_write_to_binary_file(self):
2310 tree = ET.ElementTree(ET.XML('''<site />'''))
2311 with open(TESTFN, 'wb') as f:
2312 tree.write(f)
2313 self.assertFalse(f.closed)
2314 with open(TESTFN, 'rb') as f:
2315 self.assertEqual(f.read(), b'''<site />''')
2316
2317 def test_write_to_binary_file_with_bom(self):
2318 tree = ET.ElementTree(ET.XML('''<site />'''))
2319 # test BOM writing to buffered file
2320 with open(TESTFN, 'wb') as f:
2321 tree.write(f, encoding='utf-16')
2322 self.assertFalse(f.closed)
2323 with open(TESTFN, 'rb') as f:
2324 self.assertEqual(f.read(),
2325 '''<?xml version='1.0' encoding='utf-16'?>\n'''
2326 '''<site />'''.encode("utf-16"))
2327 # test BOM writing to non-buffered file
2328 with open(TESTFN, 'wb', buffering=0) as f:
2329 tree.write(f, encoding='utf-16')
2330 self.assertFalse(f.closed)
2331 with open(TESTFN, 'rb') as f:
2332 self.assertEqual(f.read(),
2333 '''<?xml version='1.0' encoding='utf-16'?>\n'''
2334 '''<site />'''.encode("utf-16"))
2335
Eli Benderskyf996e772012-03-16 05:53:30 +02002336 def test_read_from_stringio(self):
2337 tree = ET.ElementTree()
Eli Bendersky00f402b2012-07-15 06:02:22 +03002338 stream = io.StringIO('''<?xml version="1.0"?><site></site>''')
Eli Benderskyf996e772012-03-16 05:53:30 +02002339 tree.parse(stream)
Eli Benderskyf996e772012-03-16 05:53:30 +02002340 self.assertEqual(tree.getroot().tag, 'site')
2341
Eli Bendersky00f402b2012-07-15 06:02:22 +03002342 def test_write_to_stringio(self):
2343 tree = ET.ElementTree(ET.XML('''<site />'''))
2344 stream = io.StringIO()
2345 tree.write(stream, encoding='unicode')
2346 self.assertEqual(stream.getvalue(), '''<site />''')
2347
2348 def test_read_from_bytesio(self):
2349 tree = ET.ElementTree()
2350 raw = io.BytesIO(b'''<?xml version="1.0"?><site></site>''')
2351 tree.parse(raw)
2352 self.assertEqual(tree.getroot().tag, 'site')
2353
2354 def test_write_to_bytesio(self):
2355 tree = ET.ElementTree(ET.XML('''<site />'''))
2356 raw = io.BytesIO()
2357 tree.write(raw)
2358 self.assertEqual(raw.getvalue(), b'''<site />''')
2359
2360 class dummy:
2361 pass
2362
2363 def test_read_from_user_text_reader(self):
2364 stream = io.StringIO('''<?xml version="1.0"?><site></site>''')
2365 reader = self.dummy()
2366 reader.read = stream.read
2367 tree = ET.ElementTree()
2368 tree.parse(reader)
2369 self.assertEqual(tree.getroot().tag, 'site')
2370
2371 def test_write_to_user_text_writer(self):
2372 tree = ET.ElementTree(ET.XML('''<site />'''))
2373 stream = io.StringIO()
2374 writer = self.dummy()
2375 writer.write = stream.write
2376 tree.write(writer, encoding='unicode')
2377 self.assertEqual(stream.getvalue(), '''<site />''')
2378
2379 def test_read_from_user_binary_reader(self):
2380 raw = io.BytesIO(b'''<?xml version="1.0"?><site></site>''')
2381 reader = self.dummy()
2382 reader.read = raw.read
2383 tree = ET.ElementTree()
2384 tree.parse(reader)
2385 self.assertEqual(tree.getroot().tag, 'site')
2386 tree = ET.ElementTree()
2387
2388 def test_write_to_user_binary_writer(self):
2389 tree = ET.ElementTree(ET.XML('''<site />'''))
2390 raw = io.BytesIO()
2391 writer = self.dummy()
2392 writer.write = raw.write
2393 tree.write(writer)
2394 self.assertEqual(raw.getvalue(), b'''<site />''')
2395
2396 def test_write_to_user_binary_writer_with_bom(self):
2397 tree = ET.ElementTree(ET.XML('''<site />'''))
2398 raw = io.BytesIO()
2399 writer = self.dummy()
2400 writer.write = raw.write
2401 writer.seekable = lambda: True
2402 writer.tell = raw.tell
2403 tree.write(writer, encoding="utf-16")
2404 self.assertEqual(raw.getvalue(),
2405 '''<?xml version='1.0' encoding='utf-16'?>\n'''
2406 '''<site />'''.encode("utf-16"))
2407
Eli Bendersky426e2482012-07-17 05:45:11 +03002408 def test_tostringlist_invariant(self):
2409 root = ET.fromstring('<tag>foo</tag>')
2410 self.assertEqual(
2411 ET.tostring(root, 'unicode'),
2412 ''.join(ET.tostringlist(root, 'unicode')))
2413 self.assertEqual(
2414 ET.tostring(root, 'utf-16'),
2415 b''.join(ET.tostringlist(root, 'utf-16')))
2416
Eli Benderskya9a2ef52013-01-13 06:04:43 -08002417 def test_short_empty_elements(self):
2418 root = ET.fromstring('<tag>a<x />b<y></y>c</tag>')
2419 self.assertEqual(
2420 ET.tostring(root, 'unicode'),
2421 '<tag>a<x />b<y />c</tag>')
2422 self.assertEqual(
2423 ET.tostring(root, 'unicode', short_empty_elements=True),
2424 '<tag>a<x />b<y />c</tag>')
2425 self.assertEqual(
2426 ET.tostring(root, 'unicode', short_empty_elements=False),
2427 '<tag>a<x></x>b<y></y>c</tag>')
2428
Eli Benderskyf996e772012-03-16 05:53:30 +02002429
Eli Bendersky5b77d812012-03-16 08:20:05 +02002430class ParseErrorTest(unittest.TestCase):
2431 def test_subclass(self):
2432 self.assertIsInstance(ET.ParseError(), SyntaxError)
2433
2434 def _get_error(self, s):
2435 try:
2436 ET.fromstring(s)
2437 except ET.ParseError as e:
2438 return e
2439
2440 def test_error_position(self):
2441 self.assertEqual(self._get_error('foo').position, (1, 0))
2442 self.assertEqual(self._get_error('<tag>&foo;</tag>').position, (1, 5))
2443 self.assertEqual(self._get_error('foobar<').position, (1, 6))
2444
2445 def test_error_code(self):
2446 import xml.parsers.expat.errors as ERRORS
2447 self.assertEqual(self._get_error('foo').code,
2448 ERRORS.codes[ERRORS.XML_ERROR_SYNTAX])
2449
2450
Eli Bendersky737b1732012-05-29 06:02:56 +03002451class KeywordArgsTest(unittest.TestCase):
2452 # Test various issues with keyword arguments passed to ET.Element
2453 # constructor and methods
2454 def test_issue14818(self):
2455 x = ET.XML("<a>foo</a>")
2456 self.assertEqual(x.find('a', None),
2457 x.find(path='a', namespaces=None))
2458 self.assertEqual(x.findtext('a', None, None),
2459 x.findtext(path='a', default=None, namespaces=None))
2460 self.assertEqual(x.findall('a', None),
2461 x.findall(path='a', namespaces=None))
2462 self.assertEqual(list(x.iterfind('a', None)),
2463 list(x.iterfind(path='a', namespaces=None)))
2464
2465 self.assertEqual(ET.Element('a').attrib, {})
2466 elements = [
2467 ET.Element('a', dict(href="#", id="foo")),
2468 ET.Element('a', attrib=dict(href="#", id="foo")),
2469 ET.Element('a', dict(href="#"), id="foo"),
2470 ET.Element('a', href="#", id="foo"),
2471 ET.Element('a', dict(href="#", id="foo"), href="#", id="foo"),
2472 ]
2473 for e in elements:
2474 self.assertEqual(e.tag, 'a')
2475 self.assertEqual(e.attrib, dict(href="#", id="foo"))
2476
2477 e2 = ET.SubElement(elements[0], 'foobar', attrib={'key1': 'value1'})
2478 self.assertEqual(e2.attrib['key1'], 'value1')
2479
2480 with self.assertRaisesRegex(TypeError, 'must be dict, not str'):
2481 ET.Element('a', "I'm not a dict")
2482 with self.assertRaisesRegex(TypeError, 'must be dict, not str'):
2483 ET.Element('a', attrib="I'm not a dict")
2484
Eli Bendersky64d11e62012-06-15 07:42:50 +03002485# --------------------------------------------------------------------
2486
Eli Bendersky64d11e62012-06-15 07:42:50 +03002487class NoAcceleratorTest(unittest.TestCase):
Eli Bendersky52280c42012-12-30 06:27:56 -08002488 def setUp(self):
2489 if not pyET:
Eli Bendersky698bdb22013-01-10 06:01:06 -08002490 raise unittest.SkipTest('only for the Python version')
Eli Bendersky52280c42012-12-30 06:27:56 -08002491
Eli Bendersky64d11e62012-06-15 07:42:50 +03002492 # Test that the C accelerator was not imported for pyET
2493 def test_correct_import_pyET(self):
Eli Benderskye26fa1b2013-05-19 17:49:54 -07002494 # The type of methods defined in Python code is types.FunctionType,
2495 # while the type of methods defined inside _elementtree is
2496 # <class 'wrapper_descriptor'>
2497 self.assertIsInstance(pyET.Element.__init__, types.FunctionType)
2498 self.assertIsInstance(pyET.XMLParser.__init__, types.FunctionType)
Eli Bendersky64d11e62012-06-15 07:42:50 +03002499
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01002500# --------------------------------------------------------------------
2501
2502
2503class CleanContext(object):
2504 """Provide default namespace mapping and path cache."""
2505 checkwarnings = None
2506
2507 def __init__(self, quiet=False):
2508 if sys.flags.optimize >= 2:
2509 # under -OO, doctests cannot be run and therefore not all warnings
2510 # will be emitted
2511 quiet = True
2512 deprecations = (
2513 # Search behaviour is broken if search path starts with "/".
2514 ("This search is broken in 1.3 and earlier, and will be fixed "
2515 "in a future version. If you rely on the current behaviour, "
2516 "change it to '.+'", FutureWarning),
2517 # Element.getchildren() and Element.getiterator() are deprecated.
2518 ("This method will be removed in future versions. "
2519 "Use .+ instead.", DeprecationWarning),
2520 ("This method will be removed in future versions. "
2521 "Use .+ instead.", PendingDeprecationWarning))
2522 self.checkwarnings = support.check_warnings(*deprecations, quiet=quiet)
2523
2524 def __enter__(self):
2525 from xml.etree import ElementPath
2526 self._nsmap = ET.register_namespace._namespace_map
2527 # Copy the default namespace mapping
2528 self._nsmap_copy = self._nsmap.copy()
2529 # Copy the path cache (should be empty)
2530 self._path_cache = ElementPath._cache
2531 ElementPath._cache = self._path_cache.copy()
2532 self.checkwarnings.__enter__()
2533
2534 def __exit__(self, *args):
2535 from xml.etree import ElementPath
2536 # Restore mapping and path cache
2537 self._nsmap.clear()
2538 self._nsmap.update(self._nsmap_copy)
2539 ElementPath._cache = self._path_cache
2540 self.checkwarnings.__exit__(*args)
2541
2542
Eli Bendersky64d11e62012-06-15 07:42:50 +03002543def test_main(module=None):
2544 # When invoked without a module, runs the Python ET tests by loading pyET.
2545 # Otherwise, uses the given module as the ET.
Eli Bendersky698bdb22013-01-10 06:01:06 -08002546 global pyET
2547 pyET = import_fresh_module('xml.etree.ElementTree',
2548 blocked=['_elementtree'])
Eli Bendersky64d11e62012-06-15 07:42:50 +03002549 if module is None:
Eli Bendersky64d11e62012-06-15 07:42:50 +03002550 module = pyET
Florent Xicluna41fe6152010-04-02 18:52:12 +00002551
Eli Bendersky64d11e62012-06-15 07:42:50 +03002552 global ET
2553 ET = module
Florent Xiclunaf15351d2010-03-13 23:24:31 +00002554
Eli Bendersky865756a2012-03-09 13:38:15 +02002555 test_classes = [
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02002556 ModuleTest,
Eli Bendersky865756a2012-03-09 13:38:15 +02002557 ElementSlicingTest,
Eli Bendersky396e8fc2012-03-23 14:24:20 +02002558 BasicElementTest,
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02002559 ElementTreeTest,
Eli Bendersky00f402b2012-07-15 06:02:22 +03002560 IOTest,
Eli Bendersky5b77d812012-03-16 08:20:05 +02002561 ParseErrorTest,
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02002562 XIncludeTest,
2563 ElementTreeTypeTest,
Eli Benderskyceab1a92013-01-12 07:42:46 -08002564 ElementFindTest,
Eli Bendersky64d11e62012-06-15 07:42:50 +03002565 ElementIterTest,
Eli Bendersky737b1732012-05-29 06:02:56 +03002566 TreeBuilderTest,
Serhiy Storchaka66d53fa2013-05-22 17:07:51 +03002567 XMLParserTest,
Eli Benderskyb5869342013-08-30 05:51:20 -07002568 XMLPullParserTest,
Serhiy Storchakaf8cf59e2013-02-25 17:20:59 +02002569 BugsTest,
Eli Bendersky64d11e62012-06-15 07:42:50 +03002570 ]
2571
2572 # These tests will only run for the pure-Python version that doesn't import
2573 # _elementtree. We can't use skipUnless here, because pyET is filled in only
2574 # after the module is loaded.
Eli Bendersky698bdb22013-01-10 06:01:06 -08002575 if pyET is not ET:
Eli Bendersky64d11e62012-06-15 07:42:50 +03002576 test_classes.extend([
2577 NoAcceleratorTest,
Eli Bendersky64d11e62012-06-15 07:42:50 +03002578 ])
Florent Xicluna75b5e7e2012-03-05 10:42:19 +01002579
Eli Bendersky6319e0f2012-06-16 06:47:44 +03002580 try:
Eli Bendersky6319e0f2012-06-16 06:47:44 +03002581 # XXX the C module should give the same warnings as the Python module
Eli Bendersky698bdb22013-01-10 06:01:06 -08002582 with CleanContext(quiet=(pyET is not ET)):
Eli Benderskyceab1a92013-01-12 07:42:46 -08002583 support.run_unittest(*test_classes)
Eli Bendersky6319e0f2012-06-16 06:47:44 +03002584 finally:
2585 # don't interfere with subsequent tests
2586 ET = pyET = None
2587
Florent Xiclunaf15351d2010-03-13 23:24:31 +00002588
Armin Rigo9ed73062005-12-14 18:10:45 +00002589if __name__ == '__main__':
2590 test_main()