Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 1 | # 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 Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 4 | # |
| 5 | # For this purpose, the module-level "ET" symbol is temporarily |
| 6 | # monkey-patched when running the "test_xml_etree_c" test suite. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 7 | |
Serhiy Storchaka | d7a4415 | 2015-11-12 11:23:04 +0200 | [diff] [blame^] | 8 | import copy |
Georg Brandl | 1f7fffb | 2010-10-15 15:57:45 +0000 | [diff] [blame] | 9 | import html |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 10 | import io |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 11 | import operator |
Eli Bendersky | 7ec45f7 | 2012-12-30 06:17:49 -0800 | [diff] [blame] | 12 | import pickle |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 13 | import sys |
Eli Bendersky | e26fa1b | 2013-05-19 17:49:54 -0700 | [diff] [blame] | 14 | import types |
Victor Stinner | 6c6f851 | 2010-08-07 10:09:35 +0000 | [diff] [blame] | 15 | import unittest |
Serhiy Storchaka | 05744ac | 2015-06-29 22:35:58 +0300 | [diff] [blame] | 16 | import warnings |
Eli Bendersky | a5e8220 | 2012-03-31 13:55:38 +0300 | [diff] [blame] | 17 | import weakref |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 18 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 19 | from itertools import product |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 20 | from test import support |
Eli Bendersky | 2368704 | 2013-02-26 05:53:23 -0800 | [diff] [blame] | 21 | from test.support import TESTFN, findfile, import_fresh_module, gc_collect |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 22 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 23 | # pyET is the pure-Python implementation. |
Eli Bendersky | 458c0d5a | 2013-01-10 06:07:00 -0800 | [diff] [blame] | 24 | # |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 25 | # ET is pyET in test_xml_etree and is the C accelerated version in |
| 26 | # test_xml_etree_c. |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 27 | pyET = None |
| 28 | ET = None |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 29 | |
| 30 | SIMPLE_XMLFILE = findfile("simple.xml", subdir="xmltestdata") |
Victor Stinner | 6c6f851 | 2010-08-07 10:09:35 +0000 | [diff] [blame] | 31 | try: |
Marc-André Lemburg | 8f36af7 | 2011-02-25 15:42:01 +0000 | [diff] [blame] | 32 | SIMPLE_XMLFILE.encode("utf-8") |
Victor Stinner | 6c6f851 | 2010-08-07 10:09:35 +0000 | [diff] [blame] | 33 | except UnicodeEncodeError: |
| 34 | raise unittest.SkipTest("filename is not encodable to utf8") |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 35 | SIMPLE_NS_XMLFILE = findfile("simple-ns.xml", subdir="xmltestdata") |
| 36 | |
| 37 | SAMPLE_XML = """\ |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 38 | <body> |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 39 | <tag class='a'>text</tag> |
| 40 | <tag class='b' /> |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 41 | <section> |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 42 | <tag class='b' id='inner'>subtext</tag> |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 43 | </section> |
| 44 | </body> |
| 45 | """ |
| 46 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 47 | SAMPLE_SECTION = """\ |
| 48 | <section> |
| 49 | <tag class='b' id='inner'>subtext</tag> |
| 50 | <nexttag /> |
| 51 | <nextsection> |
| 52 | <tag /> |
| 53 | </nextsection> |
| 54 | </section> |
| 55 | """ |
| 56 | |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 57 | SAMPLE_XML_NS = """ |
| 58 | <body xmlns="http://effbot.org/ns"> |
| 59 | <tag>text</tag> |
| 60 | <tag /> |
| 61 | <section> |
| 62 | <tag>subtext</tag> |
| 63 | </section> |
| 64 | </body> |
| 65 | """ |
| 66 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 67 | SAMPLE_XML_NS_ELEMS = """ |
| 68 | <root> |
| 69 | <h:table xmlns:h="hello"> |
| 70 | <h:tr> |
| 71 | <h:td>Apples</h:td> |
| 72 | <h:td>Bananas</h:td> |
| 73 | </h:tr> |
| 74 | </h:table> |
| 75 | |
| 76 | <f:table xmlns:f="foo"> |
| 77 | <f:name>African Coffee Table</f:name> |
| 78 | <f:width>80</f:width> |
| 79 | <f:length>120</f:length> |
| 80 | </f:table> |
| 81 | </root> |
| 82 | """ |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 83 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 84 | ENTITY_XML = """\ |
| 85 | <!DOCTYPE points [ |
| 86 | <!ENTITY % user-entities SYSTEM 'user-entities.xml'> |
| 87 | %user-entities; |
| 88 | ]> |
| 89 | <document>&entity;</document> |
| 90 | """ |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 91 | |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 92 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 93 | class ModuleTest(unittest.TestCase): |
Eli Bendersky | 2368704 | 2013-02-26 05:53:23 -0800 | [diff] [blame] | 94 | # TODO: this should be removed once we get rid of the global module vars |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 95 | |
| 96 | def test_sanity(self): |
| 97 | # Import sanity. |
| 98 | |
| 99 | from xml.etree import ElementTree |
| 100 | from xml.etree import ElementInclude |
| 101 | from xml.etree import ElementPath |
| 102 | |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 103 | |
Florent Xicluna | c17f172 | 2010-08-08 19:48:29 +0000 | [diff] [blame] | 104 | def serialize(elem, to_string=True, encoding='unicode', **options): |
Florent Xicluna | c17f172 | 2010-08-08 19:48:29 +0000 | [diff] [blame] | 105 | if encoding != 'unicode': |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 106 | file = io.BytesIO() |
| 107 | else: |
| 108 | file = io.StringIO() |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 109 | tree = ET.ElementTree(elem) |
Florent Xicluna | c17f172 | 2010-08-08 19:48:29 +0000 | [diff] [blame] | 110 | tree.write(file, encoding=encoding, **options) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 111 | if to_string: |
| 112 | return file.getvalue() |
| 113 | else: |
| 114 | file.seek(0) |
| 115 | return file |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 116 | |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 117 | def summarize_list(seq): |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 118 | return [elem.tag for elem in seq] |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 119 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 120 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 121 | class ElementTestCase: |
| 122 | @classmethod |
| 123 | def setUpClass(cls): |
| 124 | cls.modules = {pyET, ET} |
| 125 | |
Serhiy Storchaka | bad1257 | 2014-12-15 14:03:42 +0200 | [diff] [blame] | 126 | def pickleRoundTrip(self, obj, name, dumper, loader, proto): |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 127 | save_m = sys.modules[name] |
| 128 | try: |
| 129 | sys.modules[name] = dumper |
Serhiy Storchaka | bad1257 | 2014-12-15 14:03:42 +0200 | [diff] [blame] | 130 | temp = pickle.dumps(obj, proto) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 131 | sys.modules[name] = loader |
| 132 | result = pickle.loads(temp) |
| 133 | except pickle.PicklingError as pe: |
| 134 | # pyET must be second, because pyET may be (equal to) ET. |
| 135 | human = dict([(ET, "cET"), (pyET, "pyET")]) |
| 136 | raise support.TestFailed("Failed to round-trip %r from %r to %r" |
| 137 | % (obj, |
| 138 | human.get(dumper, dumper), |
| 139 | human.get(loader, loader))) from pe |
| 140 | finally: |
| 141 | sys.modules[name] = save_m |
| 142 | return result |
| 143 | |
| 144 | def assertEqualElements(self, alice, bob): |
| 145 | self.assertIsInstance(alice, (ET.Element, pyET.Element)) |
| 146 | self.assertIsInstance(bob, (ET.Element, pyET.Element)) |
| 147 | self.assertEqual(len(list(alice)), len(list(bob))) |
| 148 | for x, y in zip(alice, bob): |
| 149 | self.assertEqualElements(x, y) |
| 150 | properties = operator.attrgetter('tag', 'tail', 'text', 'attrib') |
| 151 | self.assertEqual(properties(alice), properties(bob)) |
| 152 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 153 | # -------------------------------------------------------------------- |
| 154 | # element tree tests |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 155 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 156 | class ElementTreeTest(unittest.TestCase): |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 157 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 158 | def serialize_check(self, elem, expected): |
| 159 | self.assertEqual(serialize(elem), expected) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 160 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 161 | def test_interface(self): |
| 162 | # Test element tree interface. |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 163 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 164 | def check_string(string): |
| 165 | len(string) |
| 166 | for char in string: |
| 167 | self.assertEqual(len(char), 1, |
| 168 | msg="expected one-character string, got %r" % char) |
| 169 | new_string = string + "" |
| 170 | new_string = string + " " |
| 171 | string[:0] |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 172 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 173 | def check_mapping(mapping): |
| 174 | len(mapping) |
| 175 | keys = mapping.keys() |
| 176 | items = mapping.items() |
| 177 | for key in keys: |
| 178 | item = mapping[key] |
| 179 | mapping["key"] = "value" |
| 180 | self.assertEqual(mapping["key"], "value", |
| 181 | msg="expected value string, got %r" % mapping["key"]) |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 182 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 183 | def check_element(element): |
| 184 | self.assertTrue(ET.iselement(element), msg="not an element") |
| 185 | self.assertTrue(hasattr(element, "tag"), msg="no tag member") |
| 186 | self.assertTrue(hasattr(element, "attrib"), msg="no attrib member") |
| 187 | self.assertTrue(hasattr(element, "text"), msg="no text member") |
| 188 | self.assertTrue(hasattr(element, "tail"), msg="no tail member") |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 189 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 190 | check_string(element.tag) |
| 191 | check_mapping(element.attrib) |
| 192 | if element.text is not None: |
| 193 | check_string(element.text) |
| 194 | if element.tail is not None: |
| 195 | check_string(element.tail) |
| 196 | for elem in element: |
| 197 | check_element(elem) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 198 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 199 | element = ET.Element("tag") |
| 200 | check_element(element) |
| 201 | tree = ET.ElementTree(element) |
| 202 | check_element(tree.getroot()) |
| 203 | element = ET.Element("t\xe4g", key="value") |
| 204 | tree = ET.ElementTree(element) |
| 205 | self.assertRegex(repr(element), r"^<Element 't\xe4g' at 0x.*>$") |
| 206 | element = ET.Element("tag", key="value") |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 207 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 208 | # Make sure all standard element methods exist. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 209 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 210 | def check_method(method): |
| 211 | self.assertTrue(hasattr(method, '__call__'), |
| 212 | msg="%s not callable" % method) |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 213 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 214 | check_method(element.append) |
| 215 | check_method(element.extend) |
| 216 | check_method(element.insert) |
| 217 | check_method(element.remove) |
| 218 | check_method(element.getchildren) |
| 219 | check_method(element.find) |
| 220 | check_method(element.iterfind) |
| 221 | check_method(element.findall) |
| 222 | check_method(element.findtext) |
| 223 | check_method(element.clear) |
| 224 | check_method(element.get) |
| 225 | check_method(element.set) |
| 226 | check_method(element.keys) |
| 227 | check_method(element.items) |
| 228 | check_method(element.iter) |
| 229 | check_method(element.itertext) |
| 230 | check_method(element.getiterator) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 231 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 232 | # These methods return an iterable. See bug 6472. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 233 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 234 | def check_iter(it): |
| 235 | check_method(it.__next__) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 236 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 237 | check_iter(element.iterfind("tag")) |
| 238 | check_iter(element.iterfind("*")) |
| 239 | check_iter(tree.iterfind("tag")) |
| 240 | check_iter(tree.iterfind("*")) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 241 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 242 | # These aliases are provided: |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 243 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 244 | self.assertEqual(ET.XML, ET.fromstring) |
| 245 | self.assertEqual(ET.PI, ET.ProcessingInstruction) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 246 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 247 | def test_simpleops(self): |
| 248 | # Basic method sanity checks. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 249 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 250 | elem = ET.XML("<body><tag/></body>") |
| 251 | self.serialize_check(elem, '<body><tag /></body>') |
| 252 | e = ET.Element("tag2") |
| 253 | elem.append(e) |
| 254 | self.serialize_check(elem, '<body><tag /><tag2 /></body>') |
| 255 | elem.remove(e) |
| 256 | self.serialize_check(elem, '<body><tag /></body>') |
| 257 | elem.insert(0, e) |
| 258 | self.serialize_check(elem, '<body><tag2 /><tag /></body>') |
| 259 | elem.remove(e) |
| 260 | elem.extend([e]) |
| 261 | self.serialize_check(elem, '<body><tag /><tag2 /></body>') |
| 262 | elem.remove(e) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 263 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 264 | element = ET.Element("tag", key="value") |
| 265 | self.serialize_check(element, '<tag key="value" />') # 1 |
| 266 | subelement = ET.Element("subtag") |
| 267 | element.append(subelement) |
| 268 | self.serialize_check(element, '<tag key="value"><subtag /></tag>') # 2 |
| 269 | element.insert(0, subelement) |
| 270 | self.serialize_check(element, |
| 271 | '<tag key="value"><subtag /><subtag /></tag>') # 3 |
| 272 | element.remove(subelement) |
| 273 | self.serialize_check(element, '<tag key="value"><subtag /></tag>') # 4 |
| 274 | element.remove(subelement) |
| 275 | self.serialize_check(element, '<tag key="value" />') # 5 |
| 276 | with self.assertRaises(ValueError) as cm: |
| 277 | element.remove(subelement) |
| 278 | self.assertEqual(str(cm.exception), 'list.remove(x): x not in list') |
| 279 | self.serialize_check(element, '<tag key="value" />') # 6 |
| 280 | element[0:0] = [subelement, subelement, subelement] |
| 281 | self.serialize_check(element[1], '<subtag />') |
| 282 | self.assertEqual(element[1:9], [element[1], element[2]]) |
| 283 | self.assertEqual(element[:9:2], [element[0], element[2]]) |
| 284 | del element[1:2] |
| 285 | self.serialize_check(element, |
| 286 | '<tag key="value"><subtag /><subtag /></tag>') |
Florent Xicluna | a72a98f | 2012-02-13 11:03:30 +0100 | [diff] [blame] | 287 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 288 | def test_cdata(self): |
| 289 | # Test CDATA handling (etc). |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 290 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 291 | self.serialize_check(ET.XML("<tag>hello</tag>"), |
| 292 | '<tag>hello</tag>') |
| 293 | self.serialize_check(ET.XML("<tag>hello</tag>"), |
| 294 | '<tag>hello</tag>') |
| 295 | self.serialize_check(ET.XML("<tag><![CDATA[hello]]></tag>"), |
| 296 | '<tag>hello</tag>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 297 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 298 | def test_file_init(self): |
| 299 | stringfile = io.BytesIO(SAMPLE_XML.encode("utf-8")) |
| 300 | tree = ET.ElementTree(file=stringfile) |
| 301 | self.assertEqual(tree.find("tag").tag, 'tag') |
| 302 | self.assertEqual(tree.find("section/tag").tag, 'tag') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 303 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 304 | tree = ET.ElementTree(file=SIMPLE_XMLFILE) |
| 305 | self.assertEqual(tree.find("element").tag, 'element') |
| 306 | self.assertEqual(tree.find("element/../empty-element").tag, |
| 307 | 'empty-element') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 308 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 309 | def test_path_cache(self): |
| 310 | # Check that the path cache behaves sanely. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 311 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 312 | from xml.etree import ElementPath |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 313 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 314 | elem = ET.XML(SAMPLE_XML) |
| 315 | for i in range(10): ET.ElementTree(elem).find('./'+str(i)) |
| 316 | cache_len_10 = len(ElementPath._cache) |
| 317 | for i in range(10): ET.ElementTree(elem).find('./'+str(i)) |
| 318 | self.assertEqual(len(ElementPath._cache), cache_len_10) |
| 319 | for i in range(20): ET.ElementTree(elem).find('./'+str(i)) |
| 320 | self.assertGreater(len(ElementPath._cache), cache_len_10) |
| 321 | for i in range(600): ET.ElementTree(elem).find('./'+str(i)) |
| 322 | self.assertLess(len(ElementPath._cache), 500) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 323 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 324 | def test_copy(self): |
| 325 | # Test copy handling (etc). |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 326 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 327 | import copy |
| 328 | e1 = ET.XML("<tag>hello<foo/></tag>") |
| 329 | e2 = copy.copy(e1) |
| 330 | e3 = copy.deepcopy(e1) |
| 331 | e1.find("foo").tag = "bar" |
| 332 | self.serialize_check(e1, '<tag>hello<bar /></tag>') |
| 333 | self.serialize_check(e2, '<tag>hello<bar /></tag>') |
| 334 | self.serialize_check(e3, '<tag>hello<foo /></tag>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 335 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 336 | def test_attrib(self): |
| 337 | # Test attribute handling. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 338 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 339 | elem = ET.Element("tag") |
| 340 | elem.get("key") # 1.1 |
| 341 | self.assertEqual(elem.get("key", "default"), 'default') # 1.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 342 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 343 | elem.set("key", "value") |
| 344 | self.assertEqual(elem.get("key"), 'value') # 1.3 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 345 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 346 | elem = ET.Element("tag", key="value") |
| 347 | self.assertEqual(elem.get("key"), 'value') # 2.1 |
| 348 | self.assertEqual(elem.attrib, {'key': 'value'}) # 2.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 349 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 350 | attrib = {"key": "value"} |
| 351 | elem = ET.Element("tag", attrib) |
| 352 | attrib.clear() # check for aliasing issues |
| 353 | self.assertEqual(elem.get("key"), 'value') # 3.1 |
| 354 | self.assertEqual(elem.attrib, {'key': 'value'}) # 3.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 355 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 356 | attrib = {"key": "value"} |
| 357 | elem = ET.Element("tag", **attrib) |
| 358 | attrib.clear() # check for aliasing issues |
| 359 | self.assertEqual(elem.get("key"), 'value') # 4.1 |
| 360 | self.assertEqual(elem.attrib, {'key': 'value'}) # 4.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 361 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 362 | elem = ET.Element("tag", {"key": "other"}, key="value") |
| 363 | self.assertEqual(elem.get("key"), 'value') # 5.1 |
| 364 | self.assertEqual(elem.attrib, {'key': 'value'}) # 5.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 365 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 366 | elem = ET.Element('test') |
| 367 | elem.text = "aa" |
| 368 | elem.set('testa', 'testval') |
| 369 | elem.set('testb', 'test2') |
| 370 | self.assertEqual(ET.tostring(elem), |
| 371 | b'<test testa="testval" testb="test2">aa</test>') |
| 372 | self.assertEqual(sorted(elem.keys()), ['testa', 'testb']) |
| 373 | self.assertEqual(sorted(elem.items()), |
| 374 | [('testa', 'testval'), ('testb', 'test2')]) |
| 375 | self.assertEqual(elem.attrib['testb'], 'test2') |
| 376 | elem.attrib['testb'] = 'test1' |
| 377 | elem.attrib['testc'] = 'test2' |
| 378 | self.assertEqual(ET.tostring(elem), |
| 379 | b'<test testa="testval" testb="test1" testc="test2">aa</test>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 380 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 381 | def test_makeelement(self): |
| 382 | # Test makeelement handling. |
Antoine Pitrou | b86680e | 2010-10-14 21:15:17 +0000 | [diff] [blame] | 383 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 384 | elem = ET.Element("tag") |
| 385 | attrib = {"key": "value"} |
| 386 | subelem = elem.makeelement("subtag", attrib) |
| 387 | self.assertIsNot(subelem.attrib, attrib, msg="attrib aliasing") |
| 388 | elem.append(subelem) |
| 389 | self.serialize_check(elem, '<tag><subtag key="value" /></tag>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 390 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 391 | elem.clear() |
| 392 | self.serialize_check(elem, '<tag />') |
| 393 | elem.append(subelem) |
| 394 | self.serialize_check(elem, '<tag><subtag key="value" /></tag>') |
| 395 | elem.extend([subelem, subelem]) |
| 396 | self.serialize_check(elem, |
| 397 | '<tag><subtag key="value" /><subtag key="value" /><subtag key="value" /></tag>') |
| 398 | elem[:] = [subelem] |
| 399 | self.serialize_check(elem, '<tag><subtag key="value" /></tag>') |
| 400 | elem[:] = tuple([subelem]) |
| 401 | self.serialize_check(elem, '<tag><subtag key="value" /></tag>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 402 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 403 | def test_parsefile(self): |
| 404 | # Test parsing from file. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 405 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 406 | tree = ET.parse(SIMPLE_XMLFILE) |
| 407 | stream = io.StringIO() |
| 408 | tree.write(stream, encoding='unicode') |
| 409 | self.assertEqual(stream.getvalue(), |
| 410 | '<root>\n' |
| 411 | ' <element key="value">text</element>\n' |
| 412 | ' <element>text</element>tail\n' |
| 413 | ' <empty-element />\n' |
| 414 | '</root>') |
| 415 | tree = ET.parse(SIMPLE_NS_XMLFILE) |
| 416 | stream = io.StringIO() |
| 417 | tree.write(stream, encoding='unicode') |
| 418 | self.assertEqual(stream.getvalue(), |
| 419 | '<ns0:root xmlns:ns0="namespace">\n' |
| 420 | ' <ns0:element key="value">text</ns0:element>\n' |
| 421 | ' <ns0:element>text</ns0:element>tail\n' |
| 422 | ' <ns0:empty-element />\n' |
| 423 | '</ns0:root>') |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 424 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 425 | with open(SIMPLE_XMLFILE) as f: |
| 426 | data = f.read() |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 427 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 428 | parser = ET.XMLParser() |
| 429 | self.assertRegex(parser.version, r'^Expat ') |
| 430 | parser.feed(data) |
| 431 | self.serialize_check(parser.close(), |
| 432 | '<root>\n' |
| 433 | ' <element key="value">text</element>\n' |
| 434 | ' <element>text</element>tail\n' |
| 435 | ' <empty-element />\n' |
| 436 | '</root>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 437 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 438 | target = ET.TreeBuilder() |
| 439 | parser = ET.XMLParser(target=target) |
| 440 | parser.feed(data) |
| 441 | self.serialize_check(parser.close(), |
| 442 | '<root>\n' |
| 443 | ' <element key="value">text</element>\n' |
| 444 | ' <element>text</element>tail\n' |
| 445 | ' <empty-element />\n' |
| 446 | '</root>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 447 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 448 | def test_parseliteral(self): |
| 449 | element = ET.XML("<html><body>text</body></html>") |
| 450 | self.assertEqual(ET.tostring(element, encoding='unicode'), |
| 451 | '<html><body>text</body></html>') |
| 452 | element = ET.fromstring("<html><body>text</body></html>") |
| 453 | self.assertEqual(ET.tostring(element, encoding='unicode'), |
| 454 | '<html><body>text</body></html>') |
| 455 | sequence = ["<html><body>", "text</bo", "dy></html>"] |
| 456 | element = ET.fromstringlist(sequence) |
| 457 | self.assertEqual(ET.tostring(element), |
| 458 | b'<html><body>text</body></html>') |
| 459 | self.assertEqual(b"".join(ET.tostringlist(element)), |
| 460 | b'<html><body>text</body></html>') |
| 461 | self.assertEqual(ET.tostring(element, "ascii"), |
| 462 | b"<?xml version='1.0' encoding='ascii'?>\n" |
| 463 | b"<html><body>text</body></html>") |
| 464 | _, ids = ET.XMLID("<html><body>text</body></html>") |
| 465 | self.assertEqual(len(ids), 0) |
| 466 | _, ids = ET.XMLID("<html><body id='body'>text</body></html>") |
| 467 | self.assertEqual(len(ids), 1) |
| 468 | self.assertEqual(ids["body"].tag, 'body') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 469 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 470 | def test_iterparse(self): |
| 471 | # Test iterparse interface. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 472 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 473 | iterparse = ET.iterparse |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 474 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 475 | context = iterparse(SIMPLE_XMLFILE) |
| 476 | action, elem = next(context) |
| 477 | self.assertEqual((action, elem.tag), ('end', 'element')) |
| 478 | self.assertEqual([(action, elem.tag) for action, elem in context], [ |
| 479 | ('end', 'element'), |
| 480 | ('end', 'empty-element'), |
| 481 | ('end', 'root'), |
| 482 | ]) |
| 483 | self.assertEqual(context.root.tag, 'root') |
| 484 | |
| 485 | context = iterparse(SIMPLE_NS_XMLFILE) |
| 486 | self.assertEqual([(action, elem.tag) for action, elem in context], [ |
| 487 | ('end', '{namespace}element'), |
| 488 | ('end', '{namespace}element'), |
| 489 | ('end', '{namespace}empty-element'), |
| 490 | ('end', '{namespace}root'), |
| 491 | ]) |
| 492 | |
| 493 | events = () |
| 494 | context = iterparse(SIMPLE_XMLFILE, events) |
| 495 | self.assertEqual([(action, elem.tag) for action, elem in context], []) |
| 496 | |
| 497 | events = () |
| 498 | context = iterparse(SIMPLE_XMLFILE, events=events) |
| 499 | self.assertEqual([(action, elem.tag) for action, elem in context], []) |
| 500 | |
| 501 | events = ("start", "end") |
| 502 | context = iterparse(SIMPLE_XMLFILE, events) |
| 503 | self.assertEqual([(action, elem.tag) for action, elem in context], [ |
| 504 | ('start', 'root'), |
| 505 | ('start', 'element'), |
| 506 | ('end', 'element'), |
| 507 | ('start', 'element'), |
| 508 | ('end', 'element'), |
| 509 | ('start', 'empty-element'), |
| 510 | ('end', 'empty-element'), |
| 511 | ('end', 'root'), |
| 512 | ]) |
| 513 | |
| 514 | events = ("start", "end", "start-ns", "end-ns") |
| 515 | context = iterparse(SIMPLE_NS_XMLFILE, events) |
Eli Bendersky | 2368704 | 2013-02-26 05:53:23 -0800 | [diff] [blame] | 516 | self.assertEqual([(action, elem.tag) if action in ("start", "end") |
| 517 | else (action, elem) |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 518 | for action, elem in context], [ |
| 519 | ('start-ns', ('', 'namespace')), |
| 520 | ('start', '{namespace}root'), |
| 521 | ('start', '{namespace}element'), |
| 522 | ('end', '{namespace}element'), |
| 523 | ('start', '{namespace}element'), |
| 524 | ('end', '{namespace}element'), |
| 525 | ('start', '{namespace}empty-element'), |
| 526 | ('end', '{namespace}empty-element'), |
| 527 | ('end', '{namespace}root'), |
| 528 | ('end-ns', None), |
| 529 | ]) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 530 | |
Eli Bendersky | 5dd40e5 | 2013-11-28 06:31:58 -0800 | [diff] [blame] | 531 | events = ('start-ns', 'end-ns') |
| 532 | context = iterparse(io.StringIO(r"<root xmlns=''/>"), events) |
| 533 | res = [action for action, elem in context] |
| 534 | self.assertEqual(res, ['start-ns', 'end-ns']) |
| 535 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 536 | events = ("start", "end", "bogus") |
| 537 | with self.assertRaises(ValueError) as cm: |
| 538 | with open(SIMPLE_XMLFILE, "rb") as f: |
| 539 | iterparse(f, events) |
| 540 | self.assertEqual(str(cm.exception), "unknown event 'bogus'") |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 541 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 542 | source = io.BytesIO( |
| 543 | b"<?xml version='1.0' encoding='iso-8859-1'?>\n" |
| 544 | b"<body xmlns='http://éffbot.org/ns'\n" |
| 545 | b" xmlns:cl\xe9='http://effbot.org/ns'>text</body>\n") |
| 546 | events = ("start-ns",) |
| 547 | context = iterparse(source, events) |
| 548 | self.assertEqual([(action, elem) for action, elem in context], [ |
| 549 | ('start-ns', ('', 'http://\xe9ffbot.org/ns')), |
| 550 | ('start-ns', ('cl\xe9', 'http://effbot.org/ns')), |
| 551 | ]) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 552 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 553 | source = io.StringIO("<document />junk") |
| 554 | it = iterparse(source) |
| 555 | action, elem = next(it) |
| 556 | self.assertEqual((action, elem.tag), ('end', 'document')) |
| 557 | with self.assertRaises(ET.ParseError) as cm: |
| 558 | next(it) |
| 559 | self.assertEqual(str(cm.exception), |
| 560 | 'junk after document element: line 1, column 12') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 561 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 562 | def test_writefile(self): |
| 563 | elem = ET.Element("tag") |
| 564 | elem.text = "text" |
| 565 | self.serialize_check(elem, '<tag>text</tag>') |
| 566 | ET.SubElement(elem, "subtag").text = "subtext" |
| 567 | self.serialize_check(elem, '<tag>text<subtag>subtext</subtag></tag>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 568 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 569 | # Test tag suppression |
| 570 | elem.tag = None |
| 571 | self.serialize_check(elem, 'text<subtag>subtext</subtag>') |
| 572 | elem.insert(0, ET.Comment("comment")) |
| 573 | self.serialize_check(elem, |
| 574 | 'text<!--comment--><subtag>subtext</subtag>') # assumes 1.3 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 575 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 576 | elem[0] = ET.PI("key", "value") |
| 577 | self.serialize_check(elem, 'text<?key value?><subtag>subtext</subtag>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 578 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 579 | def test_custom_builder(self): |
| 580 | # Test parser w. custom builder. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 581 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 582 | with open(SIMPLE_XMLFILE) as f: |
| 583 | data = f.read() |
| 584 | class Builder(list): |
| 585 | def start(self, tag, attrib): |
| 586 | self.append(("start", tag)) |
| 587 | def end(self, tag): |
| 588 | self.append(("end", tag)) |
| 589 | def data(self, text): |
| 590 | pass |
| 591 | builder = Builder() |
| 592 | parser = ET.XMLParser(target=builder) |
| 593 | parser.feed(data) |
| 594 | self.assertEqual(builder, [ |
| 595 | ('start', 'root'), |
| 596 | ('start', 'element'), |
| 597 | ('end', 'element'), |
| 598 | ('start', 'element'), |
| 599 | ('end', 'element'), |
| 600 | ('start', 'empty-element'), |
| 601 | ('end', 'empty-element'), |
| 602 | ('end', 'root'), |
| 603 | ]) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 604 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 605 | with open(SIMPLE_NS_XMLFILE) as f: |
| 606 | data = f.read() |
| 607 | class Builder(list): |
| 608 | def start(self, tag, attrib): |
| 609 | self.append(("start", tag)) |
| 610 | def end(self, tag): |
| 611 | self.append(("end", tag)) |
| 612 | def data(self, text): |
| 613 | pass |
| 614 | def pi(self, target, data): |
| 615 | self.append(("pi", target, data)) |
| 616 | def comment(self, data): |
| 617 | self.append(("comment", data)) |
| 618 | builder = Builder() |
| 619 | parser = ET.XMLParser(target=builder) |
| 620 | parser.feed(data) |
| 621 | self.assertEqual(builder, [ |
| 622 | ('pi', 'pi', 'data'), |
| 623 | ('comment', ' comment '), |
| 624 | ('start', '{namespace}root'), |
| 625 | ('start', '{namespace}element'), |
| 626 | ('end', '{namespace}element'), |
| 627 | ('start', '{namespace}element'), |
| 628 | ('end', '{namespace}element'), |
| 629 | ('start', '{namespace}empty-element'), |
| 630 | ('end', '{namespace}empty-element'), |
| 631 | ('end', '{namespace}root'), |
| 632 | ]) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 633 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 634 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 635 | def test_getchildren(self): |
| 636 | # Test Element.getchildren() |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 637 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 638 | with open(SIMPLE_XMLFILE, "rb") as f: |
| 639 | tree = ET.parse(f) |
| 640 | self.assertEqual([summarize_list(elem.getchildren()) |
| 641 | for elem in tree.getroot().iter()], [ |
| 642 | ['element', 'element', 'empty-element'], |
| 643 | [], |
| 644 | [], |
| 645 | [], |
| 646 | ]) |
| 647 | self.assertEqual([summarize_list(elem.getchildren()) |
| 648 | for elem in tree.getiterator()], [ |
| 649 | ['element', 'element', 'empty-element'], |
| 650 | [], |
| 651 | [], |
| 652 | [], |
| 653 | ]) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 654 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 655 | elem = ET.XML(SAMPLE_XML) |
| 656 | self.assertEqual(len(elem.getchildren()), 3) |
| 657 | self.assertEqual(len(elem[2].getchildren()), 1) |
| 658 | self.assertEqual(elem[:], elem.getchildren()) |
| 659 | child1 = elem[0] |
| 660 | child2 = elem[2] |
| 661 | del elem[1:2] |
| 662 | self.assertEqual(len(elem.getchildren()), 2) |
| 663 | self.assertEqual(child1, elem[0]) |
| 664 | self.assertEqual(child2, elem[1]) |
| 665 | elem[0:2] = [child2, child1] |
| 666 | self.assertEqual(child2, elem[0]) |
| 667 | self.assertEqual(child1, elem[1]) |
| 668 | self.assertNotEqual(child1, elem[0]) |
| 669 | elem.clear() |
| 670 | self.assertEqual(elem.getchildren(), []) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 671 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 672 | def test_writestring(self): |
| 673 | elem = ET.XML("<html><body>text</body></html>") |
| 674 | self.assertEqual(ET.tostring(elem), b'<html><body>text</body></html>') |
| 675 | elem = ET.fromstring("<html><body>text</body></html>") |
| 676 | self.assertEqual(ET.tostring(elem), b'<html><body>text</body></html>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 677 | |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 678 | def test_encoding(self): |
| 679 | def check(encoding, body=''): |
| 680 | xml = ("<?xml version='1.0' encoding='%s'?><xml>%s</xml>" % |
| 681 | (encoding, body)) |
| 682 | self.assertEqual(ET.XML(xml.encode(encoding)).text, body) |
| 683 | self.assertEqual(ET.XML(xml).text, body) |
| 684 | check("ascii", 'a') |
| 685 | check("us-ascii", 'a') |
| 686 | check("iso-8859-1", '\xbd') |
| 687 | check("iso-8859-15", '\u20ac') |
| 688 | check("cp437", '\u221a') |
| 689 | check("mac-roman", '\u02da') |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 690 | |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 691 | def xml(encoding): |
| 692 | return "<?xml version='1.0' encoding='%s'?><xml />" % encoding |
| 693 | def bxml(encoding): |
| 694 | return xml(encoding).encode(encoding) |
| 695 | supported_encodings = [ |
| 696 | 'ascii', 'utf-8', 'utf-8-sig', 'utf-16', 'utf-16be', 'utf-16le', |
| 697 | 'iso8859-1', 'iso8859-2', 'iso8859-3', 'iso8859-4', 'iso8859-5', |
| 698 | 'iso8859-6', 'iso8859-7', 'iso8859-8', 'iso8859-9', 'iso8859-10', |
| 699 | 'iso8859-13', 'iso8859-14', 'iso8859-15', 'iso8859-16', |
| 700 | 'cp437', 'cp720', 'cp737', 'cp775', 'cp850', 'cp852', |
| 701 | 'cp855', 'cp856', 'cp857', 'cp858', 'cp860', 'cp861', 'cp862', |
Serhiy Storchaka | be0c325 | 2013-11-23 18:52:23 +0200 | [diff] [blame] | 702 | 'cp863', 'cp865', 'cp866', 'cp869', 'cp874', 'cp1006', 'cp1125', |
| 703 | 'cp1250', 'cp1251', 'cp1252', 'cp1253', 'cp1254', 'cp1255', |
| 704 | 'cp1256', 'cp1257', 'cp1258', |
Eli Bendersky | 6dc32b3 | 2013-05-25 05:25:48 -0700 | [diff] [blame] | 705 | 'mac-cyrillic', 'mac-greek', 'mac-iceland', 'mac-latin2', |
| 706 | 'mac-roman', 'mac-turkish', |
| 707 | 'iso2022-jp', 'iso2022-jp-1', 'iso2022-jp-2', 'iso2022-jp-2004', |
| 708 | 'iso2022-jp-3', 'iso2022-jp-ext', |
| 709 | 'koi8-r', 'koi8-u', |
| 710 | 'hz', 'ptcp154', |
| 711 | ] |
| 712 | for encoding in supported_encodings: |
| 713 | self.assertEqual(ET.tostring(ET.XML(bxml(encoding))), b'<xml />') |
| 714 | |
| 715 | unsupported_ascii_compatible_encodings = [ |
| 716 | 'big5', 'big5hkscs', |
| 717 | 'cp932', 'cp949', 'cp950', |
| 718 | 'euc-jp', 'euc-jis-2004', 'euc-jisx0213', 'euc-kr', |
| 719 | 'gb2312', 'gbk', 'gb18030', |
| 720 | 'iso2022-kr', 'johab', |
| 721 | 'shift-jis', 'shift-jis-2004', 'shift-jisx0213', |
| 722 | 'utf-7', |
| 723 | ] |
| 724 | for encoding in unsupported_ascii_compatible_encodings: |
| 725 | self.assertRaises(ValueError, ET.XML, bxml(encoding)) |
| 726 | |
| 727 | unsupported_ascii_incompatible_encodings = [ |
| 728 | 'cp037', 'cp424', 'cp500', 'cp864', 'cp875', 'cp1026', 'cp1140', |
| 729 | 'utf_32', 'utf_32_be', 'utf_32_le', |
| 730 | ] |
| 731 | for encoding in unsupported_ascii_incompatible_encodings: |
| 732 | self.assertRaises(ET.ParseError, ET.XML, bxml(encoding)) |
| 733 | |
| 734 | self.assertRaises(ValueError, ET.XML, xml('undefined').encode('ascii')) |
| 735 | self.assertRaises(LookupError, ET.XML, xml('xxx').encode('ascii')) |
| 736 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 737 | def test_methods(self): |
| 738 | # Test serialization methods. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 739 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 740 | e = ET.XML("<html><link/><script>1 < 2</script></html>") |
| 741 | e.tail = "\n" |
| 742 | self.assertEqual(serialize(e), |
| 743 | '<html><link /><script>1 < 2</script></html>\n') |
| 744 | self.assertEqual(serialize(e, method=None), |
| 745 | '<html><link /><script>1 < 2</script></html>\n') |
| 746 | self.assertEqual(serialize(e, method="xml"), |
| 747 | '<html><link /><script>1 < 2</script></html>\n') |
| 748 | self.assertEqual(serialize(e, method="html"), |
| 749 | '<html><link><script>1 < 2</script></html>\n') |
| 750 | self.assertEqual(serialize(e, method="text"), '1 < 2\n') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 751 | |
Christian Heimes | 54ad7e3 | 2013-07-05 01:39:49 +0200 | [diff] [blame] | 752 | def test_issue18347(self): |
| 753 | e = ET.XML('<html><CamelCase>text</CamelCase></html>') |
| 754 | self.assertEqual(serialize(e), |
| 755 | '<html><CamelCase>text</CamelCase></html>') |
| 756 | self.assertEqual(serialize(e, method="html"), |
| 757 | '<html><CamelCase>text</CamelCase></html>') |
| 758 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 759 | def test_entity(self): |
| 760 | # Test entity handling. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 761 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 762 | # 1) good entities |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 763 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 764 | e = ET.XML("<document title='舰'>test</document>") |
| 765 | self.assertEqual(serialize(e, encoding="us-ascii"), |
| 766 | b'<document title="舰">test</document>') |
| 767 | self.serialize_check(e, '<document title="\u8230">test</document>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 768 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 769 | # 2) bad entities |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 770 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 771 | with self.assertRaises(ET.ParseError) as cm: |
| 772 | ET.XML("<document>&entity;</document>") |
| 773 | self.assertEqual(str(cm.exception), |
| 774 | 'undefined entity: line 1, column 10') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 775 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 776 | with self.assertRaises(ET.ParseError) as cm: |
| 777 | ET.XML(ENTITY_XML) |
| 778 | self.assertEqual(str(cm.exception), |
| 779 | 'undefined entity &entity;: line 5, column 10') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 780 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 781 | # 3) custom entity |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 782 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 783 | parser = ET.XMLParser() |
| 784 | parser.entity["entity"] = "text" |
| 785 | parser.feed(ENTITY_XML) |
| 786 | root = parser.close() |
| 787 | self.serialize_check(root, '<document>text</document>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 788 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 789 | def test_namespace(self): |
| 790 | # Test namespace issues. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 791 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 792 | # 1) xml namespace |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 793 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 794 | elem = ET.XML("<tag xml:lang='en' />") |
| 795 | self.serialize_check(elem, '<tag xml:lang="en" />') # 1.1 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 796 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 797 | # 2) other "well-known" namespaces |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 798 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 799 | elem = ET.XML("<rdf:RDF xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#' />") |
| 800 | self.serialize_check(elem, |
| 801 | '<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" />') # 2.1 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 802 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 803 | elem = ET.XML("<html:html xmlns:html='http://www.w3.org/1999/xhtml' />") |
| 804 | self.serialize_check(elem, |
| 805 | '<html:html xmlns:html="http://www.w3.org/1999/xhtml" />') # 2.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 806 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 807 | elem = ET.XML("<soap:Envelope xmlns:soap='http://schemas.xmlsoap.org/soap/envelope' />") |
| 808 | self.serialize_check(elem, |
| 809 | '<ns0:Envelope xmlns:ns0="http://schemas.xmlsoap.org/soap/envelope" />') # 2.3 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 810 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 811 | # 3) unknown namespaces |
| 812 | elem = ET.XML(SAMPLE_XML_NS) |
| 813 | self.serialize_check(elem, |
| 814 | '<ns0:body xmlns:ns0="http://effbot.org/ns">\n' |
| 815 | ' <ns0:tag>text</ns0:tag>\n' |
| 816 | ' <ns0:tag />\n' |
| 817 | ' <ns0:section>\n' |
| 818 | ' <ns0:tag>subtext</ns0:tag>\n' |
| 819 | ' </ns0:section>\n' |
| 820 | '</ns0:body>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 821 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 822 | def test_qname(self): |
| 823 | # Test QName handling. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 824 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 825 | # 1) decorated tags |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 826 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 827 | elem = ET.Element("{uri}tag") |
| 828 | self.serialize_check(elem, '<ns0:tag xmlns:ns0="uri" />') # 1.1 |
| 829 | elem = ET.Element(ET.QName("{uri}tag")) |
| 830 | self.serialize_check(elem, '<ns0:tag xmlns:ns0="uri" />') # 1.2 |
| 831 | elem = ET.Element(ET.QName("uri", "tag")) |
| 832 | self.serialize_check(elem, '<ns0:tag xmlns:ns0="uri" />') # 1.3 |
| 833 | elem = ET.Element(ET.QName("uri", "tag")) |
| 834 | subelem = ET.SubElement(elem, ET.QName("uri", "tag1")) |
| 835 | subelem = ET.SubElement(elem, ET.QName("uri", "tag2")) |
| 836 | self.serialize_check(elem, |
| 837 | '<ns0:tag xmlns:ns0="uri"><ns0:tag1 /><ns0:tag2 /></ns0:tag>') # 1.4 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 838 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 839 | # 2) decorated attributes |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 840 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 841 | elem.clear() |
| 842 | elem.attrib["{uri}key"] = "value" |
| 843 | self.serialize_check(elem, |
| 844 | '<ns0:tag xmlns:ns0="uri" ns0:key="value" />') # 2.1 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 845 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 846 | elem.clear() |
| 847 | elem.attrib[ET.QName("{uri}key")] = "value" |
| 848 | self.serialize_check(elem, |
| 849 | '<ns0:tag xmlns:ns0="uri" ns0:key="value" />') # 2.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 850 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 851 | # 3) decorated values are not converted by default, but the |
| 852 | # QName wrapper can be used for values |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 853 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 854 | elem.clear() |
| 855 | elem.attrib["{uri}key"] = "{uri}value" |
| 856 | self.serialize_check(elem, |
| 857 | '<ns0:tag xmlns:ns0="uri" ns0:key="{uri}value" />') # 3.1 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 858 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 859 | elem.clear() |
| 860 | elem.attrib["{uri}key"] = ET.QName("{uri}value") |
| 861 | self.serialize_check(elem, |
| 862 | '<ns0:tag xmlns:ns0="uri" ns0:key="ns0:value" />') # 3.2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 863 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 864 | elem.clear() |
| 865 | subelem = ET.Element("tag") |
| 866 | subelem.attrib["{uri1}key"] = ET.QName("{uri2}value") |
| 867 | elem.append(subelem) |
| 868 | elem.append(subelem) |
| 869 | self.serialize_check(elem, |
| 870 | '<ns0:tag xmlns:ns0="uri" xmlns:ns1="uri1" xmlns:ns2="uri2">' |
| 871 | '<tag ns1:key="ns2:value" />' |
| 872 | '<tag ns1:key="ns2:value" />' |
| 873 | '</ns0:tag>') # 3.3 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 874 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 875 | # 4) Direct QName tests |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 876 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 877 | self.assertEqual(str(ET.QName('ns', 'tag')), '{ns}tag') |
| 878 | self.assertEqual(str(ET.QName('{ns}tag')), '{ns}tag') |
| 879 | q1 = ET.QName('ns', 'tag') |
| 880 | q2 = ET.QName('ns', 'tag') |
| 881 | self.assertEqual(q1, q2) |
| 882 | q2 = ET.QName('ns', 'other-tag') |
| 883 | self.assertNotEqual(q1, q2) |
| 884 | self.assertNotEqual(q1, 'ns:tag') |
| 885 | self.assertEqual(q1, '{ns}tag') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 886 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 887 | def test_doctype_public(self): |
| 888 | # Test PUBLIC doctype. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 889 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 890 | elem = ET.XML('<!DOCTYPE html PUBLIC' |
| 891 | ' "-//W3C//DTD XHTML 1.0 Transitional//EN"' |
| 892 | ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">' |
| 893 | '<html>text</html>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 894 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 895 | def test_xpath_tokenizer(self): |
| 896 | # Test the XPath tokenizer. |
| 897 | from xml.etree import ElementPath |
| 898 | def check(p, expected): |
| 899 | self.assertEqual([op or tag |
| 900 | for op, tag in ElementPath.xpath_tokenizer(p)], |
| 901 | expected) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 902 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 903 | # tests from the xml specification |
| 904 | check("*", ['*']) |
| 905 | check("text()", ['text', '()']) |
| 906 | check("@name", ['@', 'name']) |
| 907 | check("@*", ['@', '*']) |
| 908 | check("para[1]", ['para', '[', '1', ']']) |
| 909 | check("para[last()]", ['para', '[', 'last', '()', ']']) |
| 910 | check("*/para", ['*', '/', 'para']) |
| 911 | check("/doc/chapter[5]/section[2]", |
| 912 | ['/', 'doc', '/', 'chapter', '[', '5', ']', |
| 913 | '/', 'section', '[', '2', ']']) |
| 914 | check("chapter//para", ['chapter', '//', 'para']) |
| 915 | check("//para", ['//', 'para']) |
| 916 | check("//olist/item", ['//', 'olist', '/', 'item']) |
| 917 | check(".", ['.']) |
| 918 | check(".//para", ['.', '//', 'para']) |
| 919 | check("..", ['..']) |
| 920 | check("../@lang", ['..', '/', '@', 'lang']) |
| 921 | check("chapter[title]", ['chapter', '[', 'title', ']']) |
| 922 | check("employee[@secretary and @assistant]", ['employee', |
| 923 | '[', '@', 'secretary', '', 'and', '', '@', 'assistant', ']']) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 924 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 925 | # additional tests |
| 926 | check("{http://spam}egg", ['{http://spam}egg']) |
| 927 | check("./spam.egg", ['.', '/', 'spam.egg']) |
| 928 | check(".//{http://spam}egg", ['.', '//', '{http://spam}egg']) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 929 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 930 | def test_processinginstruction(self): |
| 931 | # Test ProcessingInstruction directly |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 932 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 933 | self.assertEqual(ET.tostring(ET.ProcessingInstruction('test', 'instruction')), |
| 934 | b'<?test instruction?>') |
| 935 | self.assertEqual(ET.tostring(ET.PI('test', 'instruction')), |
| 936 | b'<?test instruction?>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 937 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 938 | # Issue #2746 |
Antoine Pitrou | 99f69ee | 2010-02-09 17:25:47 +0000 | [diff] [blame] | 939 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 940 | self.assertEqual(ET.tostring(ET.PI('test', '<testing&>')), |
| 941 | b'<?test <testing&>?>') |
| 942 | self.assertEqual(ET.tostring(ET.PI('test', '<testing&>\xe3'), 'latin-1'), |
| 943 | b"<?xml version='1.0' encoding='latin-1'?>\n" |
| 944 | b"<?test <testing&>\xe3?>") |
Antoine Pitrou | 99f69ee | 2010-02-09 17:25:47 +0000 | [diff] [blame] | 945 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 946 | def test_html_empty_elems_serialization(self): |
| 947 | # issue 15970 |
| 948 | # from http://www.w3.org/TR/html401/index/elements.html |
| 949 | for element in ['AREA', 'BASE', 'BASEFONT', 'BR', 'COL', 'FRAME', 'HR', |
| 950 | 'IMG', 'INPUT', 'ISINDEX', 'LINK', 'META', 'PARAM']: |
| 951 | for elem in [element, element.lower()]: |
| 952 | expected = '<%s>' % elem |
| 953 | serialized = serialize(ET.XML('<%s />' % elem), method='html') |
| 954 | self.assertEqual(serialized, expected) |
| 955 | serialized = serialize(ET.XML('<%s></%s>' % (elem,elem)), |
| 956 | method='html') |
| 957 | self.assertEqual(serialized, expected) |
Antoine Pitrou | 99f69ee | 2010-02-09 17:25:47 +0000 | [diff] [blame] | 958 | |
Fredrik Lundh | 8911ca3d | 2005-12-16 22:07:17 +0000 | [diff] [blame] | 959 | |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 960 | class XMLPullParserTest(unittest.TestCase): |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 961 | |
| 962 | def _feed(self, parser, data, chunk_size=None): |
| 963 | if chunk_size is None: |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 964 | parser.feed(data) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 965 | else: |
| 966 | for i in range(0, len(data), chunk_size): |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 967 | parser.feed(data[i:i+chunk_size]) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 968 | |
| 969 | def assert_event_tags(self, parser, expected): |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 970 | events = parser.read_events() |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 971 | self.assertEqual([(action, elem.tag) for action, elem in events], |
| 972 | expected) |
| 973 | |
| 974 | def test_simple_xml(self): |
| 975 | for chunk_size in (None, 1, 5): |
| 976 | with self.subTest(chunk_size=chunk_size): |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 977 | parser = ET.XMLPullParser() |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 978 | self.assert_event_tags(parser, []) |
| 979 | self._feed(parser, "<!-- comment -->\n", chunk_size) |
| 980 | self.assert_event_tags(parser, []) |
| 981 | self._feed(parser, |
| 982 | "<root>\n <element key='value'>text</element", |
| 983 | chunk_size) |
| 984 | self.assert_event_tags(parser, []) |
| 985 | self._feed(parser, ">\n", chunk_size) |
| 986 | self.assert_event_tags(parser, [('end', 'element')]) |
| 987 | self._feed(parser, "<element>text</element>tail\n", chunk_size) |
| 988 | self._feed(parser, "<empty-element/>\n", chunk_size) |
| 989 | self.assert_event_tags(parser, [ |
| 990 | ('end', 'element'), |
| 991 | ('end', 'empty-element'), |
| 992 | ]) |
| 993 | self._feed(parser, "</root>\n", chunk_size) |
| 994 | self.assert_event_tags(parser, [('end', 'root')]) |
Nick Coghlan | 4cc2afa | 2013-09-28 23:50:35 +1000 | [diff] [blame] | 995 | self.assertIsNone(parser.close()) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 996 | |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 997 | def test_feed_while_iterating(self): |
| 998 | parser = ET.XMLPullParser() |
| 999 | it = parser.read_events() |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1000 | self._feed(parser, "<root>\n <element key='value'>text</element>\n") |
| 1001 | action, elem = next(it) |
| 1002 | self.assertEqual((action, elem.tag), ('end', 'element')) |
| 1003 | self._feed(parser, "</root>\n") |
| 1004 | action, elem = next(it) |
| 1005 | self.assertEqual((action, elem.tag), ('end', 'root')) |
| 1006 | with self.assertRaises(StopIteration): |
| 1007 | next(it) |
| 1008 | |
| 1009 | def test_simple_xml_with_ns(self): |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1010 | parser = ET.XMLPullParser() |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1011 | self.assert_event_tags(parser, []) |
| 1012 | self._feed(parser, "<!-- comment -->\n") |
| 1013 | self.assert_event_tags(parser, []) |
| 1014 | self._feed(parser, "<root xmlns='namespace'>\n") |
| 1015 | self.assert_event_tags(parser, []) |
| 1016 | self._feed(parser, "<element key='value'>text</element") |
| 1017 | self.assert_event_tags(parser, []) |
| 1018 | self._feed(parser, ">\n") |
| 1019 | self.assert_event_tags(parser, [('end', '{namespace}element')]) |
| 1020 | self._feed(parser, "<element>text</element>tail\n") |
| 1021 | self._feed(parser, "<empty-element/>\n") |
| 1022 | self.assert_event_tags(parser, [ |
| 1023 | ('end', '{namespace}element'), |
| 1024 | ('end', '{namespace}empty-element'), |
| 1025 | ]) |
| 1026 | self._feed(parser, "</root>\n") |
| 1027 | self.assert_event_tags(parser, [('end', '{namespace}root')]) |
Nick Coghlan | 4cc2afa | 2013-09-28 23:50:35 +1000 | [diff] [blame] | 1028 | self.assertIsNone(parser.close()) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1029 | |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 1030 | def test_ns_events(self): |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1031 | parser = ET.XMLPullParser(events=('start-ns', 'end-ns')) |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 1032 | self._feed(parser, "<!-- comment -->\n") |
| 1033 | self._feed(parser, "<root xmlns='namespace'>\n") |
| 1034 | self.assertEqual( |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1035 | list(parser.read_events()), |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 1036 | [('start-ns', ('', 'namespace'))]) |
| 1037 | self._feed(parser, "<element key='value'>text</element") |
| 1038 | self._feed(parser, ">\n") |
| 1039 | self._feed(parser, "<element>text</element>tail\n") |
| 1040 | self._feed(parser, "<empty-element/>\n") |
| 1041 | self._feed(parser, "</root>\n") |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1042 | self.assertEqual(list(parser.read_events()), [('end-ns', None)]) |
Nick Coghlan | 4cc2afa | 2013-09-28 23:50:35 +1000 | [diff] [blame] | 1043 | self.assertIsNone(parser.close()) |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 1044 | |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1045 | def test_events(self): |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1046 | parser = ET.XMLPullParser(events=()) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1047 | self._feed(parser, "<root/>\n") |
| 1048 | self.assert_event_tags(parser, []) |
| 1049 | |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1050 | parser = ET.XMLPullParser(events=('start', 'end')) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1051 | self._feed(parser, "<!-- comment -->\n") |
| 1052 | self.assert_event_tags(parser, []) |
| 1053 | self._feed(parser, "<root>\n") |
| 1054 | self.assert_event_tags(parser, [('start', 'root')]) |
| 1055 | self._feed(parser, "<element key='value'>text</element") |
| 1056 | self.assert_event_tags(parser, [('start', 'element')]) |
| 1057 | self._feed(parser, ">\n") |
| 1058 | self.assert_event_tags(parser, [('end', 'element')]) |
| 1059 | self._feed(parser, |
| 1060 | "<element xmlns='foo'>text<empty-element/></element>tail\n") |
| 1061 | self.assert_event_tags(parser, [ |
| 1062 | ('start', '{foo}element'), |
| 1063 | ('start', '{foo}empty-element'), |
| 1064 | ('end', '{foo}empty-element'), |
| 1065 | ('end', '{foo}element'), |
| 1066 | ]) |
| 1067 | self._feed(parser, "</root>") |
Nick Coghlan | 4cc2afa | 2013-09-28 23:50:35 +1000 | [diff] [blame] | 1068 | self.assertIsNone(parser.close()) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1069 | self.assert_event_tags(parser, [('end', 'root')]) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1070 | |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1071 | parser = ET.XMLPullParser(events=('start',)) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1072 | self._feed(parser, "<!-- comment -->\n") |
| 1073 | self.assert_event_tags(parser, []) |
| 1074 | self._feed(parser, "<root>\n") |
| 1075 | self.assert_event_tags(parser, [('start', 'root')]) |
| 1076 | self._feed(parser, "<element key='value'>text</element") |
| 1077 | self.assert_event_tags(parser, [('start', 'element')]) |
| 1078 | self._feed(parser, ">\n") |
| 1079 | self.assert_event_tags(parser, []) |
| 1080 | self._feed(parser, |
| 1081 | "<element xmlns='foo'>text<empty-element/></element>tail\n") |
| 1082 | self.assert_event_tags(parser, [ |
| 1083 | ('start', '{foo}element'), |
| 1084 | ('start', '{foo}empty-element'), |
| 1085 | ]) |
| 1086 | self._feed(parser, "</root>") |
Nick Coghlan | 4cc2afa | 2013-09-28 23:50:35 +1000 | [diff] [blame] | 1087 | self.assertIsNone(parser.close()) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1088 | |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 1089 | def test_events_sequence(self): |
| 1090 | # Test that events can be some sequence that's not just a tuple or list |
| 1091 | eventset = {'end', 'start'} |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1092 | parser = ET.XMLPullParser(events=eventset) |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 1093 | self._feed(parser, "<foo>bar</foo>") |
| 1094 | self.assert_event_tags(parser, [('start', 'foo'), ('end', 'foo')]) |
| 1095 | |
| 1096 | class DummyIter: |
| 1097 | def __init__(self): |
| 1098 | self.events = iter(['start', 'end', 'start-ns']) |
| 1099 | def __iter__(self): |
| 1100 | return self |
| 1101 | def __next__(self): |
| 1102 | return next(self.events) |
| 1103 | |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1104 | parser = ET.XMLPullParser(events=DummyIter()) |
Eli Bendersky | 3a4fbd8 | 2013-05-19 09:01:49 -0700 | [diff] [blame] | 1105 | self._feed(parser, "<foo>bar</foo>") |
| 1106 | self.assert_event_tags(parser, [('start', 'foo'), ('end', 'foo')]) |
| 1107 | |
| 1108 | |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1109 | def test_unknown_event(self): |
| 1110 | with self.assertRaises(ValueError): |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 1111 | ET.XMLPullParser(events=('start', 'end', 'bogus')) |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1112 | |
| 1113 | |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 1114 | # |
| 1115 | # xinclude tests (samples from appendix C of the xinclude specification) |
| 1116 | |
| 1117 | XINCLUDE = {} |
| 1118 | |
| 1119 | XINCLUDE["C1.xml"] = """\ |
| 1120 | <?xml version='1.0'?> |
| 1121 | <document xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1122 | <p>120 Mz is adequate for an average home user.</p> |
| 1123 | <xi:include href="disclaimer.xml"/> |
| 1124 | </document> |
| 1125 | """ |
| 1126 | |
| 1127 | XINCLUDE["disclaimer.xml"] = """\ |
| 1128 | <?xml version='1.0'?> |
| 1129 | <disclaimer> |
| 1130 | <p>The opinions represented herein represent those of the individual |
| 1131 | and should not be interpreted as official policy endorsed by this |
| 1132 | organization.</p> |
| 1133 | </disclaimer> |
| 1134 | """ |
| 1135 | |
| 1136 | XINCLUDE["C2.xml"] = """\ |
| 1137 | <?xml version='1.0'?> |
| 1138 | <document xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1139 | <p>This document has been accessed |
| 1140 | <xi:include href="count.txt" parse="text"/> times.</p> |
| 1141 | </document> |
| 1142 | """ |
| 1143 | |
| 1144 | XINCLUDE["count.txt"] = "324387" |
| 1145 | |
Florent Xicluna | ba8a986 | 2010-08-08 23:08:41 +0000 | [diff] [blame] | 1146 | XINCLUDE["C2b.xml"] = """\ |
| 1147 | <?xml version='1.0'?> |
| 1148 | <document xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1149 | <p>This document has been <em>accessed</em> |
| 1150 | <xi:include href="count.txt" parse="text"/> times.</p> |
| 1151 | </document> |
| 1152 | """ |
| 1153 | |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 1154 | XINCLUDE["C3.xml"] = """\ |
| 1155 | <?xml version='1.0'?> |
| 1156 | <document xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1157 | <p>The following is the source of the "data.xml" resource:</p> |
| 1158 | <example><xi:include href="data.xml" parse="text"/></example> |
| 1159 | </document> |
| 1160 | """ |
| 1161 | |
| 1162 | XINCLUDE["data.xml"] = """\ |
| 1163 | <?xml version='1.0'?> |
| 1164 | <data> |
| 1165 | <item><![CDATA[Brooks & Shields]]></item> |
| 1166 | </data> |
| 1167 | """ |
| 1168 | |
| 1169 | XINCLUDE["C5.xml"] = """\ |
| 1170 | <?xml version='1.0'?> |
| 1171 | <div xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1172 | <xi:include href="example.txt" parse="text"> |
| 1173 | <xi:fallback> |
| 1174 | <xi:include href="fallback-example.txt" parse="text"> |
| 1175 | <xi:fallback><a href="mailto:bob@example.org">Report error</a></xi:fallback> |
| 1176 | </xi:include> |
| 1177 | </xi:fallback> |
| 1178 | </xi:include> |
| 1179 | </div> |
| 1180 | """ |
| 1181 | |
| 1182 | XINCLUDE["default.xml"] = """\ |
| 1183 | <?xml version='1.0'?> |
| 1184 | <document xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1185 | <p>Example.</p> |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1186 | <xi:include href="{}"/> |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 1187 | </document> |
Georg Brandl | 1f7fffb | 2010-10-15 15:57:45 +0000 | [diff] [blame] | 1188 | """.format(html.escape(SIMPLE_XMLFILE, True)) |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 1189 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1190 | # |
| 1191 | # badly formatted xi:include tags |
| 1192 | |
| 1193 | XINCLUDE_BAD = {} |
| 1194 | |
| 1195 | XINCLUDE_BAD["B1.xml"] = """\ |
| 1196 | <?xml version='1.0'?> |
| 1197 | <document xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1198 | <p>120 Mz is adequate for an average home user.</p> |
| 1199 | <xi:include href="disclaimer.xml" parse="BAD_TYPE"/> |
| 1200 | </document> |
| 1201 | """ |
| 1202 | |
| 1203 | XINCLUDE_BAD["B2.xml"] = """\ |
| 1204 | <?xml version='1.0'?> |
| 1205 | <div xmlns:xi="http://www.w3.org/2001/XInclude"> |
| 1206 | <xi:fallback></xi:fallback> |
| 1207 | </div> |
| 1208 | """ |
| 1209 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1210 | class XIncludeTest(unittest.TestCase): |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1211 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1212 | def xinclude_loader(self, href, parse="xml", encoding=None): |
| 1213 | try: |
| 1214 | data = XINCLUDE[href] |
| 1215 | except KeyError: |
| 1216 | raise OSError("resource not found") |
| 1217 | if parse == "xml": |
| 1218 | data = ET.XML(data) |
| 1219 | return data |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1220 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1221 | def none_loader(self, href, parser, encoding=None): |
| 1222 | return None |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1223 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1224 | def _my_loader(self, href, parse): |
| 1225 | # Used to avoid a test-dependency problem where the default loader |
| 1226 | # of ElementInclude uses the pyET parser for cET tests. |
| 1227 | if parse == 'xml': |
| 1228 | with open(href, 'rb') as f: |
| 1229 | return ET.parse(f).getroot() |
| 1230 | else: |
| 1231 | return None |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1232 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1233 | def test_xinclude_default(self): |
| 1234 | from xml.etree import ElementInclude |
| 1235 | doc = self.xinclude_loader('default.xml') |
| 1236 | ElementInclude.include(doc, self._my_loader) |
| 1237 | self.assertEqual(serialize(doc), |
| 1238 | '<document>\n' |
| 1239 | ' <p>Example.</p>\n' |
| 1240 | ' <root>\n' |
| 1241 | ' <element key="value">text</element>\n' |
| 1242 | ' <element>text</element>tail\n' |
| 1243 | ' <empty-element />\n' |
| 1244 | '</root>\n' |
| 1245 | '</document>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1246 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1247 | def test_xinclude(self): |
| 1248 | from xml.etree import ElementInclude |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1249 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1250 | # Basic inclusion example (XInclude C.1) |
| 1251 | document = self.xinclude_loader("C1.xml") |
| 1252 | ElementInclude.include(document, self.xinclude_loader) |
| 1253 | self.assertEqual(serialize(document), |
| 1254 | '<document>\n' |
| 1255 | ' <p>120 Mz is adequate for an average home user.</p>\n' |
| 1256 | ' <disclaimer>\n' |
| 1257 | ' <p>The opinions represented herein represent those of the individual\n' |
| 1258 | ' and should not be interpreted as official policy endorsed by this\n' |
| 1259 | ' organization.</p>\n' |
| 1260 | '</disclaimer>\n' |
| 1261 | '</document>') # C1 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1262 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1263 | # Textual inclusion example (XInclude C.2) |
| 1264 | document = self.xinclude_loader("C2.xml") |
| 1265 | ElementInclude.include(document, self.xinclude_loader) |
| 1266 | self.assertEqual(serialize(document), |
| 1267 | '<document>\n' |
| 1268 | ' <p>This document has been accessed\n' |
| 1269 | ' 324387 times.</p>\n' |
| 1270 | '</document>') # C2 |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1271 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1272 | # Textual inclusion after sibling element (based on modified XInclude C.2) |
| 1273 | document = self.xinclude_loader("C2b.xml") |
| 1274 | ElementInclude.include(document, self.xinclude_loader) |
| 1275 | self.assertEqual(serialize(document), |
| 1276 | '<document>\n' |
| 1277 | ' <p>This document has been <em>accessed</em>\n' |
| 1278 | ' 324387 times.</p>\n' |
| 1279 | '</document>') # C2b |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1280 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1281 | # Textual inclusion of XML example (XInclude C.3) |
| 1282 | document = self.xinclude_loader("C3.xml") |
| 1283 | ElementInclude.include(document, self.xinclude_loader) |
| 1284 | self.assertEqual(serialize(document), |
| 1285 | '<document>\n' |
| 1286 | ' <p>The following is the source of the "data.xml" resource:</p>\n' |
| 1287 | " <example><?xml version='1.0'?>\n" |
| 1288 | '<data>\n' |
| 1289 | ' <item><![CDATA[Brooks & Shields]]></item>\n' |
| 1290 | '</data>\n' |
| 1291 | '</example>\n' |
| 1292 | '</document>') # C3 |
| 1293 | |
| 1294 | # Fallback example (XInclude C.5) |
| 1295 | # Note! Fallback support is not yet implemented |
| 1296 | document = self.xinclude_loader("C5.xml") |
| 1297 | with self.assertRaises(OSError) as cm: |
| 1298 | ElementInclude.include(document, self.xinclude_loader) |
| 1299 | self.assertEqual(str(cm.exception), 'resource not found') |
| 1300 | self.assertEqual(serialize(document), |
| 1301 | '<div xmlns:ns0="http://www.w3.org/2001/XInclude">\n' |
| 1302 | ' <ns0:include href="example.txt" parse="text">\n' |
| 1303 | ' <ns0:fallback>\n' |
| 1304 | ' <ns0:include href="fallback-example.txt" parse="text">\n' |
| 1305 | ' <ns0:fallback><a href="mailto:bob@example.org">Report error</a></ns0:fallback>\n' |
| 1306 | ' </ns0:include>\n' |
| 1307 | ' </ns0:fallback>\n' |
| 1308 | ' </ns0:include>\n' |
| 1309 | '</div>') # C5 |
| 1310 | |
| 1311 | def test_xinclude_failures(self): |
| 1312 | from xml.etree import ElementInclude |
| 1313 | |
| 1314 | # Test failure to locate included XML file. |
| 1315 | document = ET.XML(XINCLUDE["C1.xml"]) |
| 1316 | with self.assertRaises(ElementInclude.FatalIncludeError) as cm: |
| 1317 | ElementInclude.include(document, loader=self.none_loader) |
| 1318 | self.assertEqual(str(cm.exception), |
| 1319 | "cannot load 'disclaimer.xml' as 'xml'") |
| 1320 | |
| 1321 | # Test failure to locate included text file. |
| 1322 | document = ET.XML(XINCLUDE["C2.xml"]) |
| 1323 | with self.assertRaises(ElementInclude.FatalIncludeError) as cm: |
| 1324 | ElementInclude.include(document, loader=self.none_loader) |
| 1325 | self.assertEqual(str(cm.exception), |
| 1326 | "cannot load 'count.txt' as 'text'") |
| 1327 | |
| 1328 | # Test bad parse type. |
| 1329 | document = ET.XML(XINCLUDE_BAD["B1.xml"]) |
| 1330 | with self.assertRaises(ElementInclude.FatalIncludeError) as cm: |
| 1331 | ElementInclude.include(document, loader=self.none_loader) |
| 1332 | self.assertEqual(str(cm.exception), |
| 1333 | "unknown parse type in xi:include tag ('BAD_TYPE')") |
| 1334 | |
| 1335 | # Test xi:fallback outside xi:include. |
| 1336 | document = ET.XML(XINCLUDE_BAD["B2.xml"]) |
| 1337 | with self.assertRaises(ElementInclude.FatalIncludeError) as cm: |
| 1338 | ElementInclude.include(document, loader=self.none_loader) |
| 1339 | self.assertEqual(str(cm.exception), |
| 1340 | "xi:fallback tag must be child of xi:include " |
| 1341 | "('{http://www.w3.org/2001/XInclude}fallback')") |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1342 | |
| 1343 | # -------------------------------------------------------------------- |
| 1344 | # reported bugs |
| 1345 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1346 | class BugsTest(unittest.TestCase): |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1347 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1348 | def test_bug_xmltoolkit21(self): |
| 1349 | # marshaller gives obscure errors for non-string values |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1350 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1351 | def check(elem): |
| 1352 | with self.assertRaises(TypeError) as cm: |
| 1353 | serialize(elem) |
| 1354 | self.assertEqual(str(cm.exception), |
| 1355 | 'cannot serialize 123 (type int)') |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 1356 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1357 | elem = ET.Element(123) |
| 1358 | check(elem) # tag |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 1359 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1360 | elem = ET.Element("elem") |
| 1361 | elem.text = 123 |
| 1362 | check(elem) # text |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1363 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1364 | elem = ET.Element("elem") |
| 1365 | elem.tail = 123 |
| 1366 | check(elem) # tail |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1367 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1368 | elem = ET.Element("elem") |
| 1369 | elem.set(123, "123") |
| 1370 | check(elem) # attribute key |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1371 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1372 | elem = ET.Element("elem") |
| 1373 | elem.set("123", 123) |
| 1374 | check(elem) # attribute value |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1375 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1376 | def test_bug_xmltoolkit25(self): |
| 1377 | # typo in ElementTree.findtext |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1378 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1379 | elem = ET.XML(SAMPLE_XML) |
| 1380 | tree = ET.ElementTree(elem) |
| 1381 | self.assertEqual(tree.findtext("tag"), 'text') |
| 1382 | self.assertEqual(tree.findtext("section/tag"), 'subtext') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1383 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1384 | def test_bug_xmltoolkit28(self): |
| 1385 | # .//tag causes exceptions |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1386 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1387 | tree = ET.XML("<doc><table><tbody/></table></doc>") |
| 1388 | self.assertEqual(summarize_list(tree.findall(".//thead")), []) |
| 1389 | self.assertEqual(summarize_list(tree.findall(".//tbody")), ['tbody']) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1390 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1391 | def test_bug_xmltoolkitX1(self): |
| 1392 | # dump() doesn't flush the output buffer |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1393 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1394 | tree = ET.XML("<doc><table><tbody/></table></doc>") |
| 1395 | with support.captured_stdout() as stdout: |
| 1396 | ET.dump(tree) |
| 1397 | self.assertEqual(stdout.getvalue(), '<doc><table><tbody /></table></doc>\n') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1398 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1399 | def test_bug_xmltoolkit39(self): |
| 1400 | # non-ascii element and attribute names doesn't work |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1401 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1402 | tree = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?><t\xe4g />") |
| 1403 | self.assertEqual(ET.tostring(tree, "utf-8"), b'<t\xc3\xa4g />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1404 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1405 | tree = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?>" |
| 1406 | b"<tag \xe4ttr='välue' />") |
| 1407 | self.assertEqual(tree.attrib, {'\xe4ttr': 'v\xe4lue'}) |
| 1408 | self.assertEqual(ET.tostring(tree, "utf-8"), |
| 1409 | b'<tag \xc3\xa4ttr="v\xc3\xa4lue" />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1410 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1411 | tree = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?>" |
| 1412 | b'<t\xe4g>text</t\xe4g>') |
| 1413 | self.assertEqual(ET.tostring(tree, "utf-8"), |
| 1414 | b'<t\xc3\xa4g>text</t\xc3\xa4g>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1415 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1416 | tree = ET.Element("t\u00e4g") |
| 1417 | self.assertEqual(ET.tostring(tree, "utf-8"), b'<t\xc3\xa4g />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1418 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1419 | tree = ET.Element("tag") |
| 1420 | tree.set("\u00e4ttr", "v\u00e4lue") |
| 1421 | self.assertEqual(ET.tostring(tree, "utf-8"), |
| 1422 | b'<tag \xc3\xa4ttr="v\xc3\xa4lue" />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1423 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1424 | def test_bug_xmltoolkit54(self): |
| 1425 | # problems handling internally defined entities |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1426 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1427 | e = ET.XML("<!DOCTYPE doc [<!ENTITY ldots '舰'>]>" |
| 1428 | '<doc>&ldots;</doc>') |
| 1429 | self.assertEqual(serialize(e, encoding="us-ascii"), |
| 1430 | b'<doc>舰</doc>') |
| 1431 | self.assertEqual(serialize(e), '<doc>\u8230</doc>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1432 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1433 | def test_bug_xmltoolkit55(self): |
| 1434 | # make sure we're reporting the first error, not the last |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1435 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1436 | with self.assertRaises(ET.ParseError) as cm: |
| 1437 | ET.XML(b"<!DOCTYPE doc SYSTEM 'doc.dtd'>" |
| 1438 | b'<doc>&ldots;&ndots;&rdots;</doc>') |
| 1439 | self.assertEqual(str(cm.exception), |
| 1440 | 'undefined entity &ldots;: line 1, column 36') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1441 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1442 | def test_bug_xmltoolkit60(self): |
| 1443 | # Handle crash in stream source. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1444 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1445 | class ExceptionFile: |
| 1446 | def read(self, x): |
| 1447 | raise OSError |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1448 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1449 | self.assertRaises(OSError, ET.parse, ExceptionFile()) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1450 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1451 | def test_bug_xmltoolkit62(self): |
| 1452 | # Don't crash when using custom entities. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1453 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1454 | ENTITIES = {'rsquo': '\u2019', 'lsquo': '\u2018'} |
Eli Bendersky | c4e98a6 | 2013-05-19 09:24:43 -0700 | [diff] [blame] | 1455 | parser = ET.XMLParser() |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1456 | parser.entity.update(ENTITIES) |
| 1457 | parser.feed("""<?xml version="1.0" encoding="UTF-8"?> |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1458 | <!DOCTYPE patent-application-publication SYSTEM "pap-v15-2001-01-31.dtd" []> |
| 1459 | <patent-application-publication> |
| 1460 | <subdoc-abstract> |
| 1461 | <paragraph id="A-0001" lvl="0">A new cultivar of Begonia plant named ‘BCT9801BEG’.</paragraph> |
| 1462 | </subdoc-abstract> |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1463 | </patent-application-publication>""") |
| 1464 | t = parser.close() |
| 1465 | self.assertEqual(t.find('.//paragraph').text, |
| 1466 | 'A new cultivar of Begonia plant named \u2018BCT9801BEG\u2019.') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1467 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1468 | def test_bug_xmltoolkit63(self): |
| 1469 | # Check reference leak. |
| 1470 | def xmltoolkit63(): |
| 1471 | tree = ET.TreeBuilder() |
| 1472 | tree.start("tag", {}) |
| 1473 | tree.data("text") |
| 1474 | tree.end("tag") |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1475 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1476 | xmltoolkit63() |
| 1477 | count = sys.getrefcount(None) |
| 1478 | for i in range(1000): |
| 1479 | xmltoolkit63() |
| 1480 | self.assertEqual(sys.getrefcount(None), count) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1481 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1482 | def test_bug_200708_newline(self): |
| 1483 | # Preserve newlines in attributes. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1484 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1485 | e = ET.Element('SomeTag', text="def _f():\n return 3\n") |
| 1486 | self.assertEqual(ET.tostring(e), |
| 1487 | b'<SomeTag text="def _f(): return 3 " />') |
| 1488 | self.assertEqual(ET.XML(ET.tostring(e)).get("text"), |
| 1489 | 'def _f():\n return 3\n') |
| 1490 | self.assertEqual(ET.tostring(ET.XML(ET.tostring(e))), |
| 1491 | b'<SomeTag text="def _f(): return 3 " />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1492 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1493 | def test_bug_200708_close(self): |
| 1494 | # Test default builder. |
| 1495 | parser = ET.XMLParser() # default |
| 1496 | parser.feed("<element>some text</element>") |
| 1497 | self.assertEqual(parser.close().tag, 'element') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1498 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1499 | # Test custom builder. |
| 1500 | class EchoTarget: |
| 1501 | def close(self): |
| 1502 | return ET.Element("element") # simulate root |
| 1503 | parser = ET.XMLParser(EchoTarget()) |
| 1504 | parser.feed("<element>some text</element>") |
| 1505 | self.assertEqual(parser.close().tag, 'element') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1506 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1507 | def test_bug_200709_default_namespace(self): |
| 1508 | e = ET.Element("{default}elem") |
| 1509 | s = ET.SubElement(e, "{default}elem") |
| 1510 | self.assertEqual(serialize(e, default_namespace="default"), # 1 |
| 1511 | '<elem xmlns="default"><elem /></elem>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1512 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1513 | e = ET.Element("{default}elem") |
| 1514 | s = ET.SubElement(e, "{default}elem") |
| 1515 | s = ET.SubElement(e, "{not-default}elem") |
| 1516 | self.assertEqual(serialize(e, default_namespace="default"), # 2 |
| 1517 | '<elem xmlns="default" xmlns:ns1="not-default">' |
| 1518 | '<elem />' |
| 1519 | '<ns1:elem />' |
| 1520 | '</elem>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1521 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1522 | e = ET.Element("{default}elem") |
| 1523 | s = ET.SubElement(e, "{default}elem") |
| 1524 | s = ET.SubElement(e, "elem") # unprefixed name |
| 1525 | with self.assertRaises(ValueError) as cm: |
| 1526 | serialize(e, default_namespace="default") # 3 |
| 1527 | self.assertEqual(str(cm.exception), |
| 1528 | 'cannot use non-qualified names with default_namespace option') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1529 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1530 | def test_bug_200709_register_namespace(self): |
| 1531 | e = ET.Element("{http://namespace.invalid/does/not/exist/}title") |
| 1532 | self.assertEqual(ET.tostring(e), |
| 1533 | b'<ns0:title xmlns:ns0="http://namespace.invalid/does/not/exist/" />') |
| 1534 | ET.register_namespace("foo", "http://namespace.invalid/does/not/exist/") |
| 1535 | e = ET.Element("{http://namespace.invalid/does/not/exist/}title") |
| 1536 | self.assertEqual(ET.tostring(e), |
| 1537 | b'<foo:title xmlns:foo="http://namespace.invalid/does/not/exist/" />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1538 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1539 | # And the Dublin Core namespace is in the default list: |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1540 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1541 | e = ET.Element("{http://purl.org/dc/elements/1.1/}title") |
| 1542 | self.assertEqual(ET.tostring(e), |
| 1543 | b'<dc:title xmlns:dc="http://purl.org/dc/elements/1.1/" />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1544 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1545 | def test_bug_200709_element_comment(self): |
| 1546 | # Not sure if this can be fixed, really (since the serializer needs |
| 1547 | # ET.Comment, not cET.comment). |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1548 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1549 | a = ET.Element('a') |
| 1550 | a.append(ET.Comment('foo')) |
| 1551 | self.assertEqual(a[0].tag, ET.Comment) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1552 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1553 | a = ET.Element('a') |
| 1554 | a.append(ET.PI('foo')) |
| 1555 | self.assertEqual(a[0].tag, ET.PI) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1556 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1557 | def test_bug_200709_element_insert(self): |
| 1558 | a = ET.Element('a') |
| 1559 | b = ET.SubElement(a, 'b') |
| 1560 | c = ET.SubElement(a, 'c') |
| 1561 | d = ET.Element('d') |
| 1562 | a.insert(0, d) |
| 1563 | self.assertEqual(summarize_list(a), ['d', 'b', 'c']) |
| 1564 | a.insert(-1, d) |
| 1565 | self.assertEqual(summarize_list(a), ['d', 'b', 'd', 'c']) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1566 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1567 | def test_bug_200709_iter_comment(self): |
| 1568 | a = ET.Element('a') |
| 1569 | b = ET.SubElement(a, 'b') |
| 1570 | comment_b = ET.Comment("TEST-b") |
| 1571 | b.append(comment_b) |
| 1572 | self.assertEqual(summarize_list(a.iter(ET.Comment)), [ET.Comment]) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1573 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1574 | # -------------------------------------------------------------------- |
| 1575 | # reported on bugs.python.org |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1576 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1577 | def test_bug_1534630(self): |
| 1578 | bob = ET.TreeBuilder() |
| 1579 | e = bob.data("data") |
| 1580 | e = bob.start("tag", {}) |
| 1581 | e = bob.end("tag") |
| 1582 | e = bob.close() |
| 1583 | self.assertEqual(serialize(e), '<tag />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1584 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1585 | def test_issue6233(self): |
| 1586 | e = ET.XML(b"<?xml version='1.0' encoding='utf-8'?>" |
| 1587 | b'<body>t\xc3\xa3g</body>') |
| 1588 | self.assertEqual(ET.tostring(e, 'ascii'), |
| 1589 | b"<?xml version='1.0' encoding='ascii'?>\n" |
| 1590 | b'<body>tãg</body>') |
| 1591 | e = ET.XML(b"<?xml version='1.0' encoding='iso-8859-1'?>" |
| 1592 | b'<body>t\xe3g</body>') |
| 1593 | self.assertEqual(ET.tostring(e, 'ascii'), |
| 1594 | b"<?xml version='1.0' encoding='ascii'?>\n" |
| 1595 | b'<body>tãg</body>') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1596 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1597 | def test_issue3151(self): |
| 1598 | e = ET.XML('<prefix:localname xmlns:prefix="${stuff}"/>') |
| 1599 | self.assertEqual(e.tag, '{${stuff}}localname') |
| 1600 | t = ET.ElementTree(e) |
| 1601 | self.assertEqual(ET.tostring(e), b'<ns0:localname xmlns:ns0="${stuff}" />') |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1602 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1603 | def test_issue6565(self): |
| 1604 | elem = ET.XML("<body><tag/></body>") |
| 1605 | self.assertEqual(summarize_list(elem), ['tag']) |
| 1606 | newelem = ET.XML(SAMPLE_XML) |
| 1607 | elem[:] = newelem[:] |
| 1608 | self.assertEqual(summarize_list(elem), ['tag', 'tag', 'section']) |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1609 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1610 | def test_issue10777(self): |
| 1611 | # Registering a namespace twice caused a "dictionary changed size during |
| 1612 | # iteration" bug. |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1613 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1614 | ET.register_namespace('test10777', 'http://myuri/') |
| 1615 | ET.register_namespace('test10777', 'http://myuri/') |
Georg Brandl | 90b2067 | 2010-12-28 10:38:33 +0000 | [diff] [blame] | 1616 | |
Antoine Pitrou | 5b235d0 | 2013-04-18 19:37:06 +0200 | [diff] [blame] | 1617 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 1618 | # -------------------------------------------------------------------- |
| 1619 | |
| 1620 | |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1621 | class BasicElementTest(ElementTestCase, unittest.TestCase): |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 1622 | def test_augmentation_type_errors(self): |
| 1623 | e = ET.Element('joe') |
| 1624 | self.assertRaises(TypeError, e.append, 'b') |
| 1625 | self.assertRaises(TypeError, e.extend, [ET.Element('bar'), 'foo']) |
| 1626 | self.assertRaises(TypeError, e.insert, 0, 'foo') |
Florent Xicluna | 41fe615 | 2010-04-02 18:52:12 +0000 | [diff] [blame] | 1627 | |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1628 | def test_cyclic_gc(self): |
Eli Bendersky | a5e8220 | 2012-03-31 13:55:38 +0300 | [diff] [blame] | 1629 | class Dummy: |
| 1630 | pass |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1631 | |
Eli Bendersky | a5e8220 | 2012-03-31 13:55:38 +0300 | [diff] [blame] | 1632 | # Test the shortest cycle: d->element->d |
| 1633 | d = Dummy() |
| 1634 | d.dummyref = ET.Element('joe', attr=d) |
| 1635 | wref = weakref.ref(d) |
| 1636 | del d |
| 1637 | gc_collect() |
| 1638 | self.assertIsNone(wref()) |
Eli Bendersky | 0192ba3 | 2012-03-30 16:38:33 +0300 | [diff] [blame] | 1639 | |
Eli Bendersky | ebf37a2 | 2012-04-03 22:02:37 +0300 | [diff] [blame] | 1640 | # A longer cycle: d->e->e2->d |
| 1641 | e = ET.Element('joe') |
| 1642 | d = Dummy() |
| 1643 | d.dummyref = e |
| 1644 | wref = weakref.ref(d) |
| 1645 | e2 = ET.SubElement(e, 'foo', attr=d) |
| 1646 | del d, e, e2 |
| 1647 | gc_collect() |
| 1648 | self.assertIsNone(wref()) |
| 1649 | |
| 1650 | # A cycle between Element objects as children of one another |
| 1651 | # e1->e2->e3->e1 |
| 1652 | e1 = ET.Element('e1') |
| 1653 | e2 = ET.Element('e2') |
| 1654 | e3 = ET.Element('e3') |
| 1655 | e1.append(e2) |
| 1656 | e2.append(e2) |
| 1657 | e3.append(e1) |
| 1658 | wref = weakref.ref(e1) |
| 1659 | del e1, e2, e3 |
| 1660 | gc_collect() |
| 1661 | self.assertIsNone(wref()) |
| 1662 | |
| 1663 | def test_weakref(self): |
| 1664 | flag = False |
| 1665 | def wref_cb(w): |
| 1666 | nonlocal flag |
| 1667 | flag = True |
| 1668 | e = ET.Element('e') |
| 1669 | wref = weakref.ref(e, wref_cb) |
| 1670 | self.assertEqual(wref().tag, 'e') |
| 1671 | del e |
| 1672 | self.assertEqual(flag, True) |
| 1673 | self.assertEqual(wref(), None) |
| 1674 | |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 1675 | def test_get_keyword_args(self): |
| 1676 | e1 = ET.Element('foo' , x=1, y=2, z=3) |
| 1677 | self.assertEqual(e1.get('x', default=7), 1) |
| 1678 | self.assertEqual(e1.get('w', default=7), 7) |
| 1679 | |
Eli Bendersky | 7ec45f7 | 2012-12-30 06:17:49 -0800 | [diff] [blame] | 1680 | def test_pickle(self): |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1681 | # issue #16076: the C implementation wasn't pickleable. |
Serhiy Storchaka | bad1257 | 2014-12-15 14:03:42 +0200 | [diff] [blame] | 1682 | for proto in range(2, pickle.HIGHEST_PROTOCOL + 1): |
| 1683 | for dumper, loader in product(self.modules, repeat=2): |
| 1684 | e = dumper.Element('foo', bar=42) |
| 1685 | e.text = "text goes here" |
| 1686 | e.tail = "opposite of head" |
| 1687 | dumper.SubElement(e, 'child').append(dumper.Element('grandchild')) |
| 1688 | e.append(dumper.Element('child')) |
| 1689 | e.findall('.//grandchild')[0].set('attr', 'other value') |
Eli Bendersky | 7ec45f7 | 2012-12-30 06:17:49 -0800 | [diff] [blame] | 1690 | |
Serhiy Storchaka | bad1257 | 2014-12-15 14:03:42 +0200 | [diff] [blame] | 1691 | e2 = self.pickleRoundTrip(e, 'xml.etree.ElementTree', |
| 1692 | dumper, loader, proto) |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 1693 | |
Serhiy Storchaka | bad1257 | 2014-12-15 14:03:42 +0200 | [diff] [blame] | 1694 | self.assertEqual(e2.tag, 'foo') |
| 1695 | self.assertEqual(e2.attrib['bar'], 42) |
| 1696 | self.assertEqual(len(e2), 2) |
| 1697 | self.assertEqualElements(e, e2) |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 1698 | |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 1699 | def test_pickle_issue18997(self): |
Serhiy Storchaka | bad1257 | 2014-12-15 14:03:42 +0200 | [diff] [blame] | 1700 | for proto in range(2, pickle.HIGHEST_PROTOCOL + 1): |
| 1701 | for dumper, loader in product(self.modules, repeat=2): |
| 1702 | XMLTEXT = """<?xml version="1.0"?> |
| 1703 | <group><dogs>4</dogs> |
| 1704 | </group>""" |
| 1705 | e1 = dumper.fromstring(XMLTEXT) |
| 1706 | if hasattr(e1, '__getstate__'): |
| 1707 | self.assertEqual(e1.__getstate__()['tag'], 'group') |
| 1708 | e2 = self.pickleRoundTrip(e1, 'xml.etree.ElementTree', |
| 1709 | dumper, loader, proto) |
| 1710 | self.assertEqual(e2.tag, 'group') |
| 1711 | self.assertEqual(e2[0].tag, 'dogs') |
Eli Bendersky | dd3661e | 2013-09-13 06:24:25 -0700 | [diff] [blame] | 1712 | |
Eli Bendersky | 2368704 | 2013-02-26 05:53:23 -0800 | [diff] [blame] | 1713 | |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 1714 | class BadElementTest(ElementTestCase, unittest.TestCase): |
| 1715 | def test_extend_mutable_list(self): |
| 1716 | class X: |
| 1717 | @property |
| 1718 | def __class__(self): |
| 1719 | L[:] = [ET.Element('baz')] |
| 1720 | return ET.Element |
| 1721 | L = [X()] |
| 1722 | e = ET.Element('foo') |
| 1723 | try: |
| 1724 | e.extend(L) |
| 1725 | except TypeError: |
| 1726 | pass |
| 1727 | |
| 1728 | class Y(X, ET.Element): |
| 1729 | pass |
| 1730 | L = [Y('x')] |
| 1731 | e = ET.Element('foo') |
| 1732 | e.extend(L) |
| 1733 | |
| 1734 | def test_extend_mutable_list2(self): |
| 1735 | class X: |
| 1736 | @property |
| 1737 | def __class__(self): |
| 1738 | del L[:] |
| 1739 | return ET.Element |
| 1740 | L = [X(), ET.Element('baz')] |
| 1741 | e = ET.Element('foo') |
| 1742 | try: |
| 1743 | e.extend(L) |
| 1744 | except TypeError: |
| 1745 | pass |
| 1746 | |
| 1747 | class Y(X, ET.Element): |
| 1748 | pass |
| 1749 | L = [Y('bar'), ET.Element('baz')] |
| 1750 | e = ET.Element('foo') |
| 1751 | e.extend(L) |
| 1752 | |
| 1753 | def test_remove_with_mutating(self): |
| 1754 | class X(ET.Element): |
| 1755 | def __eq__(self, o): |
| 1756 | del e[:] |
| 1757 | return False |
| 1758 | e = ET.Element('foo') |
| 1759 | e.extend([X('bar')]) |
| 1760 | self.assertRaises(ValueError, e.remove, ET.Element('baz')) |
| 1761 | |
| 1762 | e = ET.Element('foo') |
| 1763 | e.extend([ET.Element('bar')]) |
| 1764 | self.assertRaises(ValueError, e.remove, X('baz')) |
| 1765 | |
| 1766 | |
| 1767 | class MutatingElementPath(str): |
| 1768 | def __new__(cls, elem, *args): |
| 1769 | self = str.__new__(cls, *args) |
| 1770 | self.elem = elem |
| 1771 | return self |
| 1772 | def __eq__(self, o): |
| 1773 | del self.elem[:] |
| 1774 | return True |
| 1775 | MutatingElementPath.__hash__ = str.__hash__ |
| 1776 | |
| 1777 | class BadElementPath(str): |
| 1778 | def __eq__(self, o): |
| 1779 | raise 1/0 |
| 1780 | BadElementPath.__hash__ = str.__hash__ |
| 1781 | |
| 1782 | class BadElementPathTest(ElementTestCase, unittest.TestCase): |
| 1783 | def setUp(self): |
| 1784 | super().setUp() |
| 1785 | from xml.etree import ElementPath |
| 1786 | self.path_cache = ElementPath._cache |
| 1787 | ElementPath._cache = {} |
| 1788 | |
| 1789 | def tearDown(self): |
| 1790 | from xml.etree import ElementPath |
| 1791 | ElementPath._cache = self.path_cache |
| 1792 | super().tearDown() |
| 1793 | |
| 1794 | def test_find_with_mutating(self): |
| 1795 | e = ET.Element('foo') |
| 1796 | e.extend([ET.Element('bar')]) |
| 1797 | e.find(MutatingElementPath(e, 'x')) |
| 1798 | |
| 1799 | def test_find_with_error(self): |
| 1800 | e = ET.Element('foo') |
| 1801 | e.extend([ET.Element('bar')]) |
| 1802 | try: |
| 1803 | e.find(BadElementPath('x')) |
| 1804 | except ZeroDivisionError: |
| 1805 | pass |
| 1806 | |
| 1807 | def test_findtext_with_mutating(self): |
| 1808 | e = ET.Element('foo') |
| 1809 | e.extend([ET.Element('bar')]) |
| 1810 | e.findtext(MutatingElementPath(e, 'x')) |
| 1811 | |
| 1812 | def test_findtext_with_error(self): |
| 1813 | e = ET.Element('foo') |
| 1814 | e.extend([ET.Element('bar')]) |
| 1815 | try: |
| 1816 | e.findtext(BadElementPath('x')) |
| 1817 | except ZeroDivisionError: |
| 1818 | pass |
| 1819 | |
| 1820 | def test_findall_with_mutating(self): |
| 1821 | e = ET.Element('foo') |
| 1822 | e.extend([ET.Element('bar')]) |
| 1823 | e.findall(MutatingElementPath(e, 'x')) |
| 1824 | |
| 1825 | def test_findall_with_error(self): |
| 1826 | e = ET.Element('foo') |
| 1827 | e.extend([ET.Element('bar')]) |
| 1828 | try: |
| 1829 | e.findall(BadElementPath('x')) |
| 1830 | except ZeroDivisionError: |
| 1831 | pass |
| 1832 | |
| 1833 | |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 1834 | class ElementTreeTypeTest(unittest.TestCase): |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 1835 | def test_istype(self): |
| 1836 | self.assertIsInstance(ET.ParseError, type) |
| 1837 | self.assertIsInstance(ET.QName, type) |
| 1838 | self.assertIsInstance(ET.ElementTree, type) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1839 | self.assertIsInstance(ET.Element, type) |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 1840 | self.assertIsInstance(ET.TreeBuilder, type) |
| 1841 | self.assertIsInstance(ET.XMLParser, type) |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1842 | |
| 1843 | def test_Element_subclass_trivial(self): |
| 1844 | class MyElement(ET.Element): |
| 1845 | pass |
| 1846 | |
| 1847 | mye = MyElement('foo') |
| 1848 | self.assertIsInstance(mye, ET.Element) |
| 1849 | self.assertIsInstance(mye, MyElement) |
| 1850 | self.assertEqual(mye.tag, 'foo') |
| 1851 | |
Eli Bendersky | b20df95 | 2012-05-20 06:33:29 +0300 | [diff] [blame] | 1852 | # test that attribute assignment works (issue 14849) |
| 1853 | mye.text = "joe" |
| 1854 | self.assertEqual(mye.text, "joe") |
| 1855 | |
Eli Bendersky | 092af1f | 2012-03-04 07:14:03 +0200 | [diff] [blame] | 1856 | def test_Element_subclass_constructor(self): |
| 1857 | class MyElement(ET.Element): |
| 1858 | def __init__(self, tag, attrib={}, **extra): |
| 1859 | super(MyElement, self).__init__(tag + '__', attrib, **extra) |
| 1860 | |
| 1861 | mye = MyElement('foo', {'a': 1, 'b': 2}, c=3, d=4) |
| 1862 | self.assertEqual(mye.tag, 'foo__') |
| 1863 | self.assertEqual(sorted(mye.items()), |
| 1864 | [('a', 1), ('b', 2), ('c', 3), ('d', 4)]) |
| 1865 | |
| 1866 | def test_Element_subclass_new_method(self): |
| 1867 | class MyElement(ET.Element): |
| 1868 | def newmethod(self): |
| 1869 | return self.tag |
| 1870 | |
| 1871 | mye = MyElement('joe') |
| 1872 | self.assertEqual(mye.newmethod(), 'joe') |
Eli Bendersky | da57819 | 2012-02-16 06:52:39 +0200 | [diff] [blame] | 1873 | |
Eli Bendersky | ceab1a9 | 2013-01-12 07:42:46 -0800 | [diff] [blame] | 1874 | |
| 1875 | class ElementFindTest(unittest.TestCase): |
| 1876 | def test_find_simple(self): |
| 1877 | e = ET.XML(SAMPLE_XML) |
| 1878 | self.assertEqual(e.find('tag').tag, 'tag') |
| 1879 | self.assertEqual(e.find('section/tag').tag, 'tag') |
| 1880 | self.assertEqual(e.find('./tag').tag, 'tag') |
| 1881 | |
| 1882 | e[2] = ET.XML(SAMPLE_SECTION) |
| 1883 | self.assertEqual(e.find('section/nexttag').tag, 'nexttag') |
| 1884 | |
| 1885 | self.assertEqual(e.findtext('./tag'), 'text') |
| 1886 | self.assertEqual(e.findtext('section/tag'), 'subtext') |
| 1887 | |
| 1888 | # section/nexttag is found but has no text |
| 1889 | self.assertEqual(e.findtext('section/nexttag'), '') |
| 1890 | self.assertEqual(e.findtext('section/nexttag', 'default'), '') |
| 1891 | |
| 1892 | # tog doesn't exist and 'default' kicks in |
| 1893 | self.assertIsNone(e.findtext('tog')) |
| 1894 | self.assertEqual(e.findtext('tog', 'default'), 'default') |
| 1895 | |
Eli Bendersky | 25771b3 | 2013-01-13 05:26:07 -0800 | [diff] [blame] | 1896 | # Issue #16922 |
| 1897 | self.assertEqual(ET.XML('<tag><empty /></tag>').findtext('empty'), '') |
| 1898 | |
Eli Bendersky | a80f761 | 2013-01-22 06:12:54 -0800 | [diff] [blame] | 1899 | def test_find_xpath(self): |
| 1900 | LINEAR_XML = ''' |
| 1901 | <body> |
| 1902 | <tag class='a'/> |
| 1903 | <tag class='b'/> |
| 1904 | <tag class='c'/> |
| 1905 | <tag class='d'/> |
| 1906 | </body>''' |
| 1907 | e = ET.XML(LINEAR_XML) |
| 1908 | |
| 1909 | # Test for numeric indexing and last() |
| 1910 | self.assertEqual(e.find('./tag[1]').attrib['class'], 'a') |
| 1911 | self.assertEqual(e.find('./tag[2]').attrib['class'], 'b') |
| 1912 | self.assertEqual(e.find('./tag[last()]').attrib['class'], 'd') |
| 1913 | self.assertEqual(e.find('./tag[last()-1]').attrib['class'], 'c') |
| 1914 | self.assertEqual(e.find('./tag[last()-2]').attrib['class'], 'b') |
| 1915 | |
Eli Bendersky | 5c6198b | 2013-01-24 06:29:26 -0800 | [diff] [blame] | 1916 | self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[0]') |
| 1917 | self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[-1]') |
| 1918 | self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[last()-0]') |
| 1919 | self.assertRaisesRegex(SyntaxError, 'XPath', e.find, './tag[last()+1]') |
| 1920 | |
Eli Bendersky | ceab1a9 | 2013-01-12 07:42:46 -0800 | [diff] [blame] | 1921 | def test_findall(self): |
| 1922 | e = ET.XML(SAMPLE_XML) |
| 1923 | e[2] = ET.XML(SAMPLE_SECTION) |
| 1924 | self.assertEqual(summarize_list(e.findall('.')), ['body']) |
| 1925 | self.assertEqual(summarize_list(e.findall('tag')), ['tag', 'tag']) |
| 1926 | self.assertEqual(summarize_list(e.findall('tog')), []) |
| 1927 | self.assertEqual(summarize_list(e.findall('tog/foo')), []) |
| 1928 | self.assertEqual(summarize_list(e.findall('*')), |
| 1929 | ['tag', 'tag', 'section']) |
| 1930 | self.assertEqual(summarize_list(e.findall('.//tag')), |
| 1931 | ['tag'] * 4) |
| 1932 | self.assertEqual(summarize_list(e.findall('section/tag')), ['tag']) |
| 1933 | self.assertEqual(summarize_list(e.findall('section//tag')), ['tag'] * 2) |
| 1934 | self.assertEqual(summarize_list(e.findall('section/*')), |
| 1935 | ['tag', 'nexttag', 'nextsection']) |
| 1936 | self.assertEqual(summarize_list(e.findall('section//*')), |
| 1937 | ['tag', 'nexttag', 'nextsection', 'tag']) |
| 1938 | self.assertEqual(summarize_list(e.findall('section/.//*')), |
| 1939 | ['tag', 'nexttag', 'nextsection', 'tag']) |
| 1940 | self.assertEqual(summarize_list(e.findall('*/*')), |
| 1941 | ['tag', 'nexttag', 'nextsection']) |
| 1942 | self.assertEqual(summarize_list(e.findall('*//*')), |
| 1943 | ['tag', 'nexttag', 'nextsection', 'tag']) |
| 1944 | self.assertEqual(summarize_list(e.findall('*/tag')), ['tag']) |
| 1945 | self.assertEqual(summarize_list(e.findall('*/./tag')), ['tag']) |
| 1946 | self.assertEqual(summarize_list(e.findall('./tag')), ['tag'] * 2) |
| 1947 | self.assertEqual(summarize_list(e.findall('././tag')), ['tag'] * 2) |
| 1948 | |
| 1949 | self.assertEqual(summarize_list(e.findall('.//tag[@class]')), |
| 1950 | ['tag'] * 3) |
| 1951 | self.assertEqual(summarize_list(e.findall('.//tag[@class="a"]')), |
| 1952 | ['tag']) |
| 1953 | self.assertEqual(summarize_list(e.findall('.//tag[@class="b"]')), |
| 1954 | ['tag'] * 2) |
| 1955 | self.assertEqual(summarize_list(e.findall('.//tag[@id]')), |
| 1956 | ['tag']) |
| 1957 | self.assertEqual(summarize_list(e.findall('.//section[tag]')), |
| 1958 | ['section']) |
| 1959 | self.assertEqual(summarize_list(e.findall('.//section[element]')), []) |
| 1960 | self.assertEqual(summarize_list(e.findall('../tag')), []) |
| 1961 | self.assertEqual(summarize_list(e.findall('section/../tag')), |
| 1962 | ['tag'] * 2) |
| 1963 | self.assertEqual(e.findall('section//'), e.findall('section//*')) |
| 1964 | |
| 1965 | def test_test_find_with_ns(self): |
| 1966 | e = ET.XML(SAMPLE_XML_NS) |
| 1967 | self.assertEqual(summarize_list(e.findall('tag')), []) |
| 1968 | self.assertEqual( |
| 1969 | summarize_list(e.findall("{http://effbot.org/ns}tag")), |
| 1970 | ['{http://effbot.org/ns}tag'] * 2) |
| 1971 | self.assertEqual( |
| 1972 | summarize_list(e.findall(".//{http://effbot.org/ns}tag")), |
| 1973 | ['{http://effbot.org/ns}tag'] * 3) |
| 1974 | |
Eli Bendersky | 2acc525 | 2013-08-03 17:47:47 -0700 | [diff] [blame] | 1975 | def test_findall_different_nsmaps(self): |
| 1976 | root = ET.XML(''' |
| 1977 | <a xmlns:x="X" xmlns:y="Y"> |
| 1978 | <x:b><c/></x:b> |
| 1979 | <b/> |
| 1980 | <c><x:b/><b/></c><y:b/> |
| 1981 | </a>''') |
| 1982 | nsmap = {'xx': 'X'} |
| 1983 | self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 2) |
| 1984 | self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2) |
| 1985 | nsmap = {'xx': 'Y'} |
| 1986 | self.assertEqual(len(root.findall(".//xx:b", namespaces=nsmap)), 1) |
| 1987 | self.assertEqual(len(root.findall(".//b", namespaces=nsmap)), 2) |
| 1988 | |
Eli Bendersky | ceab1a9 | 2013-01-12 07:42:46 -0800 | [diff] [blame] | 1989 | def test_bad_find(self): |
| 1990 | e = ET.XML(SAMPLE_XML) |
| 1991 | with self.assertRaisesRegex(SyntaxError, 'cannot use absolute path'): |
| 1992 | e.findall('/tag') |
Eli Bendersky | c31f773 | 2013-01-12 07:44:32 -0800 | [diff] [blame] | 1993 | |
Eli Bendersky | ceab1a9 | 2013-01-12 07:42:46 -0800 | [diff] [blame] | 1994 | def test_find_through_ElementTree(self): |
| 1995 | e = ET.XML(SAMPLE_XML) |
| 1996 | self.assertEqual(ET.ElementTree(e).find('tag').tag, 'tag') |
| 1997 | self.assertEqual(ET.ElementTree(e).findtext('tag'), 'text') |
| 1998 | self.assertEqual(summarize_list(ET.ElementTree(e).findall('tag')), |
| 1999 | ['tag'] * 2) |
| 2000 | # this produces a warning |
| 2001 | self.assertEqual(summarize_list(ET.ElementTree(e).findall('//tag')), |
| 2002 | ['tag'] * 3) |
Eli Bendersky | c31f773 | 2013-01-12 07:44:32 -0800 | [diff] [blame] | 2003 | |
Eli Bendersky | ceab1a9 | 2013-01-12 07:42:46 -0800 | [diff] [blame] | 2004 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2005 | class ElementIterTest(unittest.TestCase): |
| 2006 | def _ilist(self, elem, tag=None): |
| 2007 | return summarize_list(elem.iter(tag)) |
| 2008 | |
| 2009 | def test_basic(self): |
| 2010 | doc = ET.XML("<html><body>this is a <i>paragraph</i>.</body>..</html>") |
| 2011 | self.assertEqual(self._ilist(doc), ['html', 'body', 'i']) |
| 2012 | self.assertEqual(self._ilist(doc.find('body')), ['body', 'i']) |
| 2013 | self.assertEqual(next(doc.iter()).tag, 'html') |
| 2014 | self.assertEqual(''.join(doc.itertext()), 'this is a paragraph...') |
| 2015 | self.assertEqual(''.join(doc.find('body').itertext()), |
| 2016 | 'this is a paragraph.') |
| 2017 | self.assertEqual(next(doc.itertext()), 'this is a ') |
| 2018 | |
| 2019 | # iterparse should return an iterator |
| 2020 | sourcefile = serialize(doc, to_string=False) |
| 2021 | self.assertEqual(next(ET.iterparse(sourcefile))[0], 'end') |
| 2022 | |
Eli Bendersky | aaa9780 | 2013-01-24 07:15:19 -0800 | [diff] [blame] | 2023 | # With an explitit parser too (issue #9708) |
| 2024 | sourcefile = serialize(doc, to_string=False) |
| 2025 | parser = ET.XMLParser(target=ET.TreeBuilder()) |
| 2026 | self.assertEqual(next(ET.iterparse(sourcefile, parser=parser))[0], |
| 2027 | 'end') |
| 2028 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2029 | tree = ET.ElementTree(None) |
| 2030 | self.assertRaises(AttributeError, tree.iter) |
| 2031 | |
Eli Bendersky | e6174ca | 2013-01-10 06:27:53 -0800 | [diff] [blame] | 2032 | # Issue #16913 |
| 2033 | doc = ET.XML("<root>a&<sub>b&</sub>c&</root>") |
| 2034 | self.assertEqual(''.join(doc.itertext()), 'a&b&c&') |
| 2035 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2036 | def test_corners(self): |
| 2037 | # single root, no subelements |
| 2038 | a = ET.Element('a') |
| 2039 | self.assertEqual(self._ilist(a), ['a']) |
| 2040 | |
| 2041 | # one child |
| 2042 | b = ET.SubElement(a, 'b') |
| 2043 | self.assertEqual(self._ilist(a), ['a', 'b']) |
| 2044 | |
| 2045 | # one child and one grandchild |
| 2046 | c = ET.SubElement(b, 'c') |
| 2047 | self.assertEqual(self._ilist(a), ['a', 'b', 'c']) |
| 2048 | |
| 2049 | # two children, only first with grandchild |
| 2050 | d = ET.SubElement(a, 'd') |
| 2051 | self.assertEqual(self._ilist(a), ['a', 'b', 'c', 'd']) |
| 2052 | |
| 2053 | # replace first child by second |
| 2054 | a[0] = a[1] |
| 2055 | del a[1] |
| 2056 | self.assertEqual(self._ilist(a), ['a', 'd']) |
| 2057 | |
| 2058 | def test_iter_by_tag(self): |
| 2059 | doc = ET.XML(''' |
| 2060 | <document> |
| 2061 | <house> |
| 2062 | <room>bedroom1</room> |
| 2063 | <room>bedroom2</room> |
| 2064 | </house> |
| 2065 | <shed>nothing here |
| 2066 | </shed> |
| 2067 | <house> |
| 2068 | <room>bedroom8</room> |
| 2069 | </house> |
| 2070 | </document>''') |
| 2071 | |
| 2072 | self.assertEqual(self._ilist(doc, 'room'), ['room'] * 3) |
| 2073 | self.assertEqual(self._ilist(doc, 'house'), ['house'] * 2) |
| 2074 | |
Eli Bendersky | a873690 | 2013-01-05 06:26:39 -0800 | [diff] [blame] | 2075 | # test that iter also accepts 'tag' as a keyword arg |
| 2076 | self.assertEqual( |
| 2077 | summarize_list(doc.iter(tag='room')), |
| 2078 | ['room'] * 3) |
| 2079 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2080 | # make sure both tag=None and tag='*' return all tags |
| 2081 | all_tags = ['document', 'house', 'room', 'room', |
| 2082 | 'shed', 'house', 'room'] |
| 2083 | self.assertEqual(self._ilist(doc), all_tags) |
| 2084 | self.assertEqual(self._ilist(doc, '*'), all_tags) |
| 2085 | |
Serhiy Storchaka | d7a4415 | 2015-11-12 11:23:04 +0200 | [diff] [blame^] | 2086 | def test_copy(self): |
| 2087 | a = ET.Element('a') |
| 2088 | it = a.iter() |
| 2089 | with self.assertRaises(TypeError): |
| 2090 | copy.copy(it) |
| 2091 | |
| 2092 | def test_pickle(self): |
| 2093 | a = ET.Element('a') |
| 2094 | it = a.iter() |
| 2095 | for proto in range(pickle.HIGHEST_PROTOCOL + 1): |
| 2096 | with self.assertRaises((TypeError, pickle.PicklingError)): |
| 2097 | pickle.dumps(it, proto) |
| 2098 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2099 | |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2100 | class TreeBuilderTest(unittest.TestCase): |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2101 | sample1 = ('<!DOCTYPE html PUBLIC' |
| 2102 | ' "-//W3C//DTD XHTML 1.0 Transitional//EN"' |
| 2103 | ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">' |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2104 | '<html>text<div>subtext</div>tail</html>') |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2105 | |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2106 | sample2 = '''<toplevel>sometext</toplevel>''' |
| 2107 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2108 | def _check_sample1_element(self, e): |
| 2109 | self.assertEqual(e.tag, 'html') |
| 2110 | self.assertEqual(e.text, 'text') |
| 2111 | self.assertEqual(e.tail, None) |
| 2112 | self.assertEqual(e.attrib, {}) |
| 2113 | children = list(e) |
| 2114 | self.assertEqual(len(children), 1) |
| 2115 | child = children[0] |
| 2116 | self.assertEqual(child.tag, 'div') |
| 2117 | self.assertEqual(child.text, 'subtext') |
| 2118 | self.assertEqual(child.tail, 'tail') |
| 2119 | self.assertEqual(child.attrib, {}) |
| 2120 | |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2121 | def test_dummy_builder(self): |
| 2122 | class BaseDummyBuilder: |
| 2123 | def close(self): |
| 2124 | return 42 |
| 2125 | |
| 2126 | class DummyBuilder(BaseDummyBuilder): |
| 2127 | data = start = end = lambda *a: None |
| 2128 | |
| 2129 | parser = ET.XMLParser(target=DummyBuilder()) |
| 2130 | parser.feed(self.sample1) |
| 2131 | self.assertEqual(parser.close(), 42) |
| 2132 | |
| 2133 | parser = ET.XMLParser(target=BaseDummyBuilder()) |
| 2134 | parser.feed(self.sample1) |
| 2135 | self.assertEqual(parser.close(), 42) |
| 2136 | |
| 2137 | parser = ET.XMLParser(target=object()) |
| 2138 | parser.feed(self.sample1) |
| 2139 | self.assertIsNone(parser.close()) |
| 2140 | |
Eli Bendersky | 08231a9 | 2013-05-18 15:47:16 -0700 | [diff] [blame] | 2141 | def test_treebuilder_elementfactory_none(self): |
| 2142 | parser = ET.XMLParser(target=ET.TreeBuilder(element_factory=None)) |
| 2143 | parser.feed(self.sample1) |
| 2144 | e = parser.close() |
| 2145 | self._check_sample1_element(e) |
| 2146 | |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2147 | def test_subclass(self): |
| 2148 | class MyTreeBuilder(ET.TreeBuilder): |
| 2149 | def foobar(self, x): |
| 2150 | return x * 2 |
| 2151 | |
| 2152 | tb = MyTreeBuilder() |
| 2153 | self.assertEqual(tb.foobar(10), 20) |
| 2154 | |
| 2155 | parser = ET.XMLParser(target=tb) |
| 2156 | parser.feed(self.sample1) |
| 2157 | |
| 2158 | e = parser.close() |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2159 | self._check_sample1_element(e) |
Eli Bendersky | 58d548d | 2012-05-29 15:45:16 +0300 | [diff] [blame] | 2160 | |
Eli Bendersky | 2b71140 | 2012-03-16 15:29:50 +0200 | [diff] [blame] | 2161 | def test_element_factory(self): |
Eli Bendersky | 48d358b | 2012-05-30 17:57:50 +0300 | [diff] [blame] | 2162 | lst = [] |
| 2163 | def myfactory(tag, attrib): |
| 2164 | nonlocal lst |
| 2165 | lst.append(tag) |
| 2166 | return ET.Element(tag, attrib) |
| 2167 | |
| 2168 | tb = ET.TreeBuilder(element_factory=myfactory) |
| 2169 | parser = ET.XMLParser(target=tb) |
| 2170 | parser.feed(self.sample2) |
| 2171 | parser.close() |
| 2172 | |
| 2173 | self.assertEqual(lst, ['toplevel']) |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2174 | |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2175 | def _check_element_factory_class(self, cls): |
| 2176 | tb = ET.TreeBuilder(element_factory=cls) |
| 2177 | |
| 2178 | parser = ET.XMLParser(target=tb) |
| 2179 | parser.feed(self.sample1) |
| 2180 | e = parser.close() |
| 2181 | self.assertIsInstance(e, cls) |
| 2182 | self._check_sample1_element(e) |
| 2183 | |
| 2184 | def test_element_factory_subclass(self): |
| 2185 | class MyElement(ET.Element): |
| 2186 | pass |
| 2187 | self._check_element_factory_class(MyElement) |
| 2188 | |
| 2189 | def test_element_factory_pure_python_subclass(self): |
| 2190 | # Mimick SimpleTAL's behaviour (issue #16089): both versions of |
| 2191 | # TreeBuilder should be able to cope with a subclass of the |
| 2192 | # pure Python Element class. |
Eli Bendersky | 46955b2 | 2013-05-19 09:20:50 -0700 | [diff] [blame] | 2193 | base = ET._Element_Py |
Antoine Pitrou | ee32931 | 2012-10-04 19:53:29 +0200 | [diff] [blame] | 2194 | # Not from a C extension |
| 2195 | self.assertEqual(base.__module__, 'xml.etree.ElementTree') |
| 2196 | # Force some multiple inheritance with a C class to make things |
| 2197 | # more interesting. |
| 2198 | class MyElement(base, ValueError): |
| 2199 | pass |
| 2200 | self._check_element_factory_class(MyElement) |
| 2201 | |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2202 | def test_doctype(self): |
| 2203 | class DoctypeParser: |
| 2204 | _doctype = None |
| 2205 | |
| 2206 | def doctype(self, name, pubid, system): |
| 2207 | self._doctype = (name, pubid, system) |
| 2208 | |
| 2209 | def close(self): |
| 2210 | return self._doctype |
| 2211 | |
| 2212 | parser = ET.XMLParser(target=DoctypeParser()) |
| 2213 | parser.feed(self.sample1) |
| 2214 | |
| 2215 | self.assertEqual(parser.close(), |
| 2216 | ('html', '-//W3C//DTD XHTML 1.0 Transitional//EN', |
| 2217 | 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd')) |
| 2218 | |
Eli Bendersky | 175fada | 2012-06-15 08:37:08 +0300 | [diff] [blame] | 2219 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2220 | class XMLParserTest(unittest.TestCase): |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 2221 | sample1 = b'<file><line>22</line></file>' |
| 2222 | sample2 = (b'<!DOCTYPE html PUBLIC' |
| 2223 | b' "-//W3C//DTD XHTML 1.0 Transitional//EN"' |
| 2224 | b' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">' |
| 2225 | b'<html>text</html>') |
| 2226 | sample3 = ('<?xml version="1.0" encoding="iso-8859-1"?>\n' |
| 2227 | '<money value="$\xa3\u20ac\U0001017b">$\xa3\u20ac\U0001017b</money>') |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2228 | |
| 2229 | def _check_sample_element(self, e): |
| 2230 | self.assertEqual(e.tag, 'file') |
| 2231 | self.assertEqual(e[0].tag, 'line') |
| 2232 | self.assertEqual(e[0].text, '22') |
| 2233 | |
| 2234 | def test_constructor_args(self): |
| 2235 | # Positional args. The first (html) is not supported, but should be |
| 2236 | # nevertheless correctly accepted. |
| 2237 | parser = ET.XMLParser(None, ET.TreeBuilder(), 'utf-8') |
| 2238 | parser.feed(self.sample1) |
| 2239 | self._check_sample_element(parser.close()) |
| 2240 | |
| 2241 | # Now as keyword args. |
Eli Bendersky | 2368704 | 2013-02-26 05:53:23 -0800 | [diff] [blame] | 2242 | parser2 = ET.XMLParser(encoding='utf-8', |
| 2243 | html=[{}], |
| 2244 | target=ET.TreeBuilder()) |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2245 | parser2.feed(self.sample1) |
| 2246 | self._check_sample_element(parser2.close()) |
| 2247 | |
| 2248 | def test_subclass(self): |
| 2249 | class MyParser(ET.XMLParser): |
| 2250 | pass |
| 2251 | parser = MyParser() |
| 2252 | parser.feed(self.sample1) |
| 2253 | self._check_sample_element(parser.close()) |
| 2254 | |
Serhiy Storchaka | 05744ac | 2015-06-29 22:35:58 +0300 | [diff] [blame] | 2255 | def test_doctype_warning(self): |
| 2256 | parser = ET.XMLParser() |
| 2257 | with self.assertWarns(DeprecationWarning): |
| 2258 | parser.doctype('html', '-//W3C//DTD XHTML 1.0 Transitional//EN', |
| 2259 | 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd') |
| 2260 | parser.feed('<html/>') |
| 2261 | parser.close() |
| 2262 | |
| 2263 | with warnings.catch_warnings(): |
| 2264 | warnings.simplefilter('error', DeprecationWarning) |
| 2265 | parser = ET.XMLParser() |
| 2266 | parser.feed(self.sample2) |
| 2267 | parser.close() |
| 2268 | |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2269 | def test_subclass_doctype(self): |
| 2270 | _doctype = None |
| 2271 | class MyParserWithDoctype(ET.XMLParser): |
| 2272 | def doctype(self, name, pubid, system): |
| 2273 | nonlocal _doctype |
| 2274 | _doctype = (name, pubid, system) |
| 2275 | |
| 2276 | parser = MyParserWithDoctype() |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 2277 | with self.assertWarns(DeprecationWarning): |
| 2278 | parser.feed(self.sample2) |
Eli Bendersky | 2b6b73e | 2012-06-01 11:32:34 +0300 | [diff] [blame] | 2279 | parser.close() |
| 2280 | self.assertEqual(_doctype, |
| 2281 | ('html', '-//W3C//DTD XHTML 1.0 Transitional//EN', |
| 2282 | 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd')) |
| 2283 | |
Serhiy Storchaka | 05744ac | 2015-06-29 22:35:58 +0300 | [diff] [blame] | 2284 | _doctype = _doctype2 = None |
| 2285 | with warnings.catch_warnings(): |
| 2286 | warnings.simplefilter('error', DeprecationWarning) |
| 2287 | class DoctypeParser: |
| 2288 | def doctype(self, name, pubid, system): |
| 2289 | nonlocal _doctype2 |
| 2290 | _doctype2 = (name, pubid, system) |
| 2291 | |
| 2292 | parser = MyParserWithDoctype(target=DoctypeParser()) |
| 2293 | parser.feed(self.sample2) |
| 2294 | parser.close() |
| 2295 | self.assertIsNone(_doctype) |
| 2296 | self.assertEqual(_doctype2, |
| 2297 | ('html', '-//W3C//DTD XHTML 1.0 Transitional//EN', |
| 2298 | 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd')) |
| 2299 | |
| 2300 | def test_inherited_doctype(self): |
| 2301 | '''Ensure that ordinary usage is not deprecated (Issue 19176)''' |
| 2302 | with warnings.catch_warnings(): |
| 2303 | warnings.simplefilter('error', DeprecationWarning) |
| 2304 | class MyParserWithoutDoctype(ET.XMLParser): |
| 2305 | pass |
| 2306 | parser = MyParserWithoutDoctype() |
| 2307 | parser.feed(self.sample2) |
| 2308 | parser.close() |
| 2309 | |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 2310 | def test_parse_string(self): |
| 2311 | parser = ET.XMLParser(target=ET.TreeBuilder()) |
| 2312 | parser.feed(self.sample3) |
| 2313 | e = parser.close() |
| 2314 | self.assertEqual(e.tag, 'money') |
| 2315 | self.assertEqual(e.attrib['value'], '$\xa3\u20ac\U0001017b') |
| 2316 | self.assertEqual(e.text, '$\xa3\u20ac\U0001017b') |
| 2317 | |
Eli Bendersky | 52467b1 | 2012-06-01 07:13:08 +0300 | [diff] [blame] | 2318 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 2319 | class NamespaceParseTest(unittest.TestCase): |
| 2320 | def test_find_with_namespace(self): |
| 2321 | nsmap = {'h': 'hello', 'f': 'foo'} |
| 2322 | doc = ET.fromstring(SAMPLE_XML_NS_ELEMS) |
| 2323 | |
| 2324 | self.assertEqual(len(doc.findall('{hello}table', nsmap)), 1) |
| 2325 | self.assertEqual(len(doc.findall('.//{hello}td', nsmap)), 2) |
| 2326 | self.assertEqual(len(doc.findall('.//{foo}name', nsmap)), 1) |
| 2327 | |
| 2328 | |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 2329 | class ElementSlicingTest(unittest.TestCase): |
| 2330 | def _elem_tags(self, elemlist): |
| 2331 | return [e.tag for e in elemlist] |
| 2332 | |
| 2333 | def _subelem_tags(self, elem): |
| 2334 | return self._elem_tags(list(elem)) |
| 2335 | |
| 2336 | def _make_elem_with_children(self, numchildren): |
| 2337 | """Create an Element with a tag 'a', with the given amount of children |
| 2338 | named 'a0', 'a1' ... and so on. |
| 2339 | |
| 2340 | """ |
| 2341 | e = ET.Element('a') |
| 2342 | for i in range(numchildren): |
| 2343 | ET.SubElement(e, 'a%s' % i) |
| 2344 | return e |
| 2345 | |
| 2346 | def test_getslice_single_index(self): |
| 2347 | e = self._make_elem_with_children(10) |
| 2348 | |
| 2349 | self.assertEqual(e[1].tag, 'a1') |
| 2350 | self.assertEqual(e[-2].tag, 'a8') |
| 2351 | |
| 2352 | self.assertRaises(IndexError, lambda: e[12]) |
| 2353 | |
| 2354 | def test_getslice_range(self): |
| 2355 | e = self._make_elem_with_children(6) |
| 2356 | |
| 2357 | self.assertEqual(self._elem_tags(e[3:]), ['a3', 'a4', 'a5']) |
| 2358 | self.assertEqual(self._elem_tags(e[3:6]), ['a3', 'a4', 'a5']) |
| 2359 | self.assertEqual(self._elem_tags(e[3:16]), ['a3', 'a4', 'a5']) |
| 2360 | self.assertEqual(self._elem_tags(e[3:5]), ['a3', 'a4']) |
| 2361 | self.assertEqual(self._elem_tags(e[3:-1]), ['a3', 'a4']) |
| 2362 | self.assertEqual(self._elem_tags(e[:2]), ['a0', 'a1']) |
| 2363 | |
| 2364 | def test_getslice_steps(self): |
| 2365 | e = self._make_elem_with_children(10) |
| 2366 | |
| 2367 | self.assertEqual(self._elem_tags(e[8:10:1]), ['a8', 'a9']) |
| 2368 | self.assertEqual(self._elem_tags(e[::3]), ['a0', 'a3', 'a6', 'a9']) |
| 2369 | self.assertEqual(self._elem_tags(e[::8]), ['a0', 'a8']) |
| 2370 | self.assertEqual(self._elem_tags(e[1::8]), ['a1', 'a9']) |
| 2371 | |
| 2372 | def test_getslice_negative_steps(self): |
| 2373 | e = self._make_elem_with_children(4) |
| 2374 | |
| 2375 | self.assertEqual(self._elem_tags(e[::-1]), ['a3', 'a2', 'a1', 'a0']) |
| 2376 | self.assertEqual(self._elem_tags(e[::-2]), ['a3', 'a1']) |
| 2377 | |
| 2378 | def test_delslice(self): |
| 2379 | e = self._make_elem_with_children(4) |
| 2380 | del e[0:2] |
| 2381 | self.assertEqual(self._subelem_tags(e), ['a2', 'a3']) |
| 2382 | |
| 2383 | e = self._make_elem_with_children(4) |
| 2384 | del e[0:] |
| 2385 | self.assertEqual(self._subelem_tags(e), []) |
| 2386 | |
| 2387 | e = self._make_elem_with_children(4) |
| 2388 | del e[::-1] |
| 2389 | self.assertEqual(self._subelem_tags(e), []) |
| 2390 | |
| 2391 | e = self._make_elem_with_children(4) |
| 2392 | del e[::-2] |
| 2393 | self.assertEqual(self._subelem_tags(e), ['a0', 'a2']) |
| 2394 | |
| 2395 | e = self._make_elem_with_children(4) |
| 2396 | del e[1::2] |
| 2397 | self.assertEqual(self._subelem_tags(e), ['a0', 'a2']) |
| 2398 | |
| 2399 | e = self._make_elem_with_children(2) |
| 2400 | del e[::2] |
| 2401 | self.assertEqual(self._subelem_tags(e), ['a1']) |
| 2402 | |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2403 | |
Eli Bendersky | 00f402b | 2012-07-15 06:02:22 +0300 | [diff] [blame] | 2404 | class IOTest(unittest.TestCase): |
| 2405 | def tearDown(self): |
Eli Bendersky | 2368704 | 2013-02-26 05:53:23 -0800 | [diff] [blame] | 2406 | support.unlink(TESTFN) |
Eli Bendersky | 00f402b | 2012-07-15 06:02:22 +0300 | [diff] [blame] | 2407 | |
| 2408 | def test_encoding(self): |
| 2409 | # Test encoding issues. |
| 2410 | elem = ET.Element("tag") |
| 2411 | elem.text = "abc" |
| 2412 | self.assertEqual(serialize(elem), '<tag>abc</tag>') |
Martin Panter | 89f76d3 | 2015-09-23 01:14:35 +0000 | [diff] [blame] | 2413 | for enc in ("utf-8", "us-ascii"): |
| 2414 | with self.subTest(enc): |
| 2415 | self.assertEqual(serialize(elem, encoding=enc), |
| 2416 | b'<tag>abc</tag>') |
| 2417 | self.assertEqual(serialize(elem, encoding=enc.upper()), |
| 2418 | b'<tag>abc</tag>') |
Eli Bendersky | 00f402b | 2012-07-15 06:02:22 +0300 | [diff] [blame] | 2419 | for enc in ("iso-8859-1", "utf-16", "utf-32"): |
Martin Panter | 89f76d3 | 2015-09-23 01:14:35 +0000 | [diff] [blame] | 2420 | with self.subTest(enc): |
| 2421 | self.assertEqual(serialize(elem, encoding=enc), |
| 2422 | ("<?xml version='1.0' encoding='%s'?>\n" |
| 2423 | "<tag>abc</tag>" % enc).encode(enc)) |
| 2424 | upper = enc.upper() |
| 2425 | self.assertEqual(serialize(elem, encoding=upper), |
| 2426 | ("<?xml version='1.0' encoding='%s'?>\n" |
| 2427 | "<tag>abc</tag>" % upper).encode(enc)) |
Eli Bendersky | 00f402b | 2012-07-15 06:02:22 +0300 | [diff] [blame] | 2428 | |
| 2429 | elem = ET.Element("tag") |
| 2430 | elem.text = "<&\"\'>" |
| 2431 | self.assertEqual(serialize(elem), '<tag><&"\'></tag>') |
| 2432 | self.assertEqual(serialize(elem, encoding="utf-8"), |
| 2433 | b'<tag><&"\'></tag>') |
| 2434 | self.assertEqual(serialize(elem, encoding="us-ascii"), |
| 2435 | b'<tag><&"\'></tag>') |
| 2436 | for enc in ("iso-8859-1", "utf-16", "utf-32"): |
| 2437 | self.assertEqual(serialize(elem, encoding=enc), |
| 2438 | ("<?xml version='1.0' encoding='%s'?>\n" |
| 2439 | "<tag><&\"'></tag>" % enc).encode(enc)) |
| 2440 | |
| 2441 | elem = ET.Element("tag") |
| 2442 | elem.attrib["key"] = "<&\"\'>" |
| 2443 | self.assertEqual(serialize(elem), '<tag key="<&"\'>" />') |
| 2444 | self.assertEqual(serialize(elem, encoding="utf-8"), |
| 2445 | b'<tag key="<&"\'>" />') |
| 2446 | self.assertEqual(serialize(elem, encoding="us-ascii"), |
| 2447 | b'<tag key="<&"\'>" />') |
| 2448 | for enc in ("iso-8859-1", "utf-16", "utf-32"): |
| 2449 | self.assertEqual(serialize(elem, encoding=enc), |
| 2450 | ("<?xml version='1.0' encoding='%s'?>\n" |
| 2451 | "<tag key=\"<&"'>\" />" % enc).encode(enc)) |
| 2452 | |
| 2453 | elem = ET.Element("tag") |
| 2454 | elem.text = '\xe5\xf6\xf6<>' |
| 2455 | self.assertEqual(serialize(elem), '<tag>\xe5\xf6\xf6<></tag>') |
| 2456 | self.assertEqual(serialize(elem, encoding="utf-8"), |
| 2457 | b'<tag>\xc3\xa5\xc3\xb6\xc3\xb6<></tag>') |
| 2458 | self.assertEqual(serialize(elem, encoding="us-ascii"), |
| 2459 | b'<tag>åöö<></tag>') |
| 2460 | for enc in ("iso-8859-1", "utf-16", "utf-32"): |
| 2461 | self.assertEqual(serialize(elem, encoding=enc), |
| 2462 | ("<?xml version='1.0' encoding='%s'?>\n" |
| 2463 | "<tag>Ă¥Ă¶Ă¶<></tag>" % enc).encode(enc)) |
| 2464 | |
| 2465 | elem = ET.Element("tag") |
| 2466 | elem.attrib["key"] = '\xe5\xf6\xf6<>' |
| 2467 | self.assertEqual(serialize(elem), '<tag key="\xe5\xf6\xf6<>" />') |
| 2468 | self.assertEqual(serialize(elem, encoding="utf-8"), |
| 2469 | b'<tag key="\xc3\xa5\xc3\xb6\xc3\xb6<>" />') |
| 2470 | self.assertEqual(serialize(elem, encoding="us-ascii"), |
| 2471 | b'<tag key="åöö<>" />') |
| 2472 | for enc in ("iso-8859-1", "utf-16", "utf-16le", "utf-16be", "utf-32"): |
| 2473 | self.assertEqual(serialize(elem, encoding=enc), |
| 2474 | ("<?xml version='1.0' encoding='%s'?>\n" |
| 2475 | "<tag key=\"Ă¥Ă¶Ă¶<>\" />" % enc).encode(enc)) |
| 2476 | |
| 2477 | def test_write_to_filename(self): |
| 2478 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2479 | tree.write(TESTFN) |
| 2480 | with open(TESTFN, 'rb') as f: |
| 2481 | self.assertEqual(f.read(), b'''<site />''') |
| 2482 | |
| 2483 | def test_write_to_text_file(self): |
| 2484 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2485 | with open(TESTFN, 'w', encoding='utf-8') as f: |
| 2486 | tree.write(f, encoding='unicode') |
| 2487 | self.assertFalse(f.closed) |
| 2488 | with open(TESTFN, 'rb') as f: |
| 2489 | self.assertEqual(f.read(), b'''<site />''') |
| 2490 | |
| 2491 | def test_write_to_binary_file(self): |
| 2492 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2493 | with open(TESTFN, 'wb') as f: |
| 2494 | tree.write(f) |
| 2495 | self.assertFalse(f.closed) |
| 2496 | with open(TESTFN, 'rb') as f: |
| 2497 | self.assertEqual(f.read(), b'''<site />''') |
| 2498 | |
| 2499 | def test_write_to_binary_file_with_bom(self): |
| 2500 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2501 | # test BOM writing to buffered file |
| 2502 | with open(TESTFN, 'wb') as f: |
| 2503 | tree.write(f, encoding='utf-16') |
| 2504 | self.assertFalse(f.closed) |
| 2505 | with open(TESTFN, 'rb') as f: |
| 2506 | self.assertEqual(f.read(), |
| 2507 | '''<?xml version='1.0' encoding='utf-16'?>\n''' |
| 2508 | '''<site />'''.encode("utf-16")) |
| 2509 | # test BOM writing to non-buffered file |
| 2510 | with open(TESTFN, 'wb', buffering=0) as f: |
| 2511 | tree.write(f, encoding='utf-16') |
| 2512 | self.assertFalse(f.closed) |
| 2513 | with open(TESTFN, 'rb') as f: |
| 2514 | self.assertEqual(f.read(), |
| 2515 | '''<?xml version='1.0' encoding='utf-16'?>\n''' |
| 2516 | '''<site />'''.encode("utf-16")) |
| 2517 | |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2518 | def test_read_from_stringio(self): |
| 2519 | tree = ET.ElementTree() |
Eli Bendersky | 00f402b | 2012-07-15 06:02:22 +0300 | [diff] [blame] | 2520 | stream = io.StringIO('''<?xml version="1.0"?><site></site>''') |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2521 | tree.parse(stream) |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2522 | self.assertEqual(tree.getroot().tag, 'site') |
| 2523 | |
Eli Bendersky | 00f402b | 2012-07-15 06:02:22 +0300 | [diff] [blame] | 2524 | def test_write_to_stringio(self): |
| 2525 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2526 | stream = io.StringIO() |
| 2527 | tree.write(stream, encoding='unicode') |
| 2528 | self.assertEqual(stream.getvalue(), '''<site />''') |
| 2529 | |
| 2530 | def test_read_from_bytesio(self): |
| 2531 | tree = ET.ElementTree() |
| 2532 | raw = io.BytesIO(b'''<?xml version="1.0"?><site></site>''') |
| 2533 | tree.parse(raw) |
| 2534 | self.assertEqual(tree.getroot().tag, 'site') |
| 2535 | |
| 2536 | def test_write_to_bytesio(self): |
| 2537 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2538 | raw = io.BytesIO() |
| 2539 | tree.write(raw) |
| 2540 | self.assertEqual(raw.getvalue(), b'''<site />''') |
| 2541 | |
| 2542 | class dummy: |
| 2543 | pass |
| 2544 | |
| 2545 | def test_read_from_user_text_reader(self): |
| 2546 | stream = io.StringIO('''<?xml version="1.0"?><site></site>''') |
| 2547 | reader = self.dummy() |
| 2548 | reader.read = stream.read |
| 2549 | tree = ET.ElementTree() |
| 2550 | tree.parse(reader) |
| 2551 | self.assertEqual(tree.getroot().tag, 'site') |
| 2552 | |
| 2553 | def test_write_to_user_text_writer(self): |
| 2554 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2555 | stream = io.StringIO() |
| 2556 | writer = self.dummy() |
| 2557 | writer.write = stream.write |
| 2558 | tree.write(writer, encoding='unicode') |
| 2559 | self.assertEqual(stream.getvalue(), '''<site />''') |
| 2560 | |
| 2561 | def test_read_from_user_binary_reader(self): |
| 2562 | raw = io.BytesIO(b'''<?xml version="1.0"?><site></site>''') |
| 2563 | reader = self.dummy() |
| 2564 | reader.read = raw.read |
| 2565 | tree = ET.ElementTree() |
| 2566 | tree.parse(reader) |
| 2567 | self.assertEqual(tree.getroot().tag, 'site') |
| 2568 | tree = ET.ElementTree() |
| 2569 | |
| 2570 | def test_write_to_user_binary_writer(self): |
| 2571 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2572 | raw = io.BytesIO() |
| 2573 | writer = self.dummy() |
| 2574 | writer.write = raw.write |
| 2575 | tree.write(writer) |
| 2576 | self.assertEqual(raw.getvalue(), b'''<site />''') |
| 2577 | |
| 2578 | def test_write_to_user_binary_writer_with_bom(self): |
| 2579 | tree = ET.ElementTree(ET.XML('''<site />''')) |
| 2580 | raw = io.BytesIO() |
| 2581 | writer = self.dummy() |
| 2582 | writer.write = raw.write |
| 2583 | writer.seekable = lambda: True |
| 2584 | writer.tell = raw.tell |
| 2585 | tree.write(writer, encoding="utf-16") |
| 2586 | self.assertEqual(raw.getvalue(), |
| 2587 | '''<?xml version='1.0' encoding='utf-16'?>\n''' |
| 2588 | '''<site />'''.encode("utf-16")) |
| 2589 | |
Eli Bendersky | 426e248 | 2012-07-17 05:45:11 +0300 | [diff] [blame] | 2590 | def test_tostringlist_invariant(self): |
| 2591 | root = ET.fromstring('<tag>foo</tag>') |
| 2592 | self.assertEqual( |
| 2593 | ET.tostring(root, 'unicode'), |
| 2594 | ''.join(ET.tostringlist(root, 'unicode'))) |
| 2595 | self.assertEqual( |
| 2596 | ET.tostring(root, 'utf-16'), |
| 2597 | b''.join(ET.tostringlist(root, 'utf-16'))) |
| 2598 | |
Eli Bendersky | a9a2ef5 | 2013-01-13 06:04:43 -0800 | [diff] [blame] | 2599 | def test_short_empty_elements(self): |
| 2600 | root = ET.fromstring('<tag>a<x />b<y></y>c</tag>') |
| 2601 | self.assertEqual( |
| 2602 | ET.tostring(root, 'unicode'), |
| 2603 | '<tag>a<x />b<y />c</tag>') |
| 2604 | self.assertEqual( |
| 2605 | ET.tostring(root, 'unicode', short_empty_elements=True), |
| 2606 | '<tag>a<x />b<y />c</tag>') |
| 2607 | self.assertEqual( |
| 2608 | ET.tostring(root, 'unicode', short_empty_elements=False), |
| 2609 | '<tag>a<x></x>b<y></y>c</tag>') |
| 2610 | |
Eli Bendersky | f996e77 | 2012-03-16 05:53:30 +0200 | [diff] [blame] | 2611 | |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2612 | class ParseErrorTest(unittest.TestCase): |
| 2613 | def test_subclass(self): |
| 2614 | self.assertIsInstance(ET.ParseError(), SyntaxError) |
| 2615 | |
| 2616 | def _get_error(self, s): |
| 2617 | try: |
| 2618 | ET.fromstring(s) |
| 2619 | except ET.ParseError as e: |
| 2620 | return e |
| 2621 | |
| 2622 | def test_error_position(self): |
| 2623 | self.assertEqual(self._get_error('foo').position, (1, 0)) |
| 2624 | self.assertEqual(self._get_error('<tag>&foo;</tag>').position, (1, 5)) |
| 2625 | self.assertEqual(self._get_error('foobar<').position, (1, 6)) |
| 2626 | |
| 2627 | def test_error_code(self): |
| 2628 | import xml.parsers.expat.errors as ERRORS |
| 2629 | self.assertEqual(self._get_error('foo').code, |
| 2630 | ERRORS.codes[ERRORS.XML_ERROR_SYNTAX]) |
| 2631 | |
| 2632 | |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 2633 | class KeywordArgsTest(unittest.TestCase): |
| 2634 | # Test various issues with keyword arguments passed to ET.Element |
| 2635 | # constructor and methods |
| 2636 | def test_issue14818(self): |
| 2637 | x = ET.XML("<a>foo</a>") |
| 2638 | self.assertEqual(x.find('a', None), |
| 2639 | x.find(path='a', namespaces=None)) |
| 2640 | self.assertEqual(x.findtext('a', None, None), |
| 2641 | x.findtext(path='a', default=None, namespaces=None)) |
| 2642 | self.assertEqual(x.findall('a', None), |
| 2643 | x.findall(path='a', namespaces=None)) |
| 2644 | self.assertEqual(list(x.iterfind('a', None)), |
| 2645 | list(x.iterfind(path='a', namespaces=None))) |
| 2646 | |
| 2647 | self.assertEqual(ET.Element('a').attrib, {}) |
| 2648 | elements = [ |
| 2649 | ET.Element('a', dict(href="#", id="foo")), |
| 2650 | ET.Element('a', attrib=dict(href="#", id="foo")), |
| 2651 | ET.Element('a', dict(href="#"), id="foo"), |
| 2652 | ET.Element('a', href="#", id="foo"), |
| 2653 | ET.Element('a', dict(href="#", id="foo"), href="#", id="foo"), |
| 2654 | ] |
| 2655 | for e in elements: |
| 2656 | self.assertEqual(e.tag, 'a') |
| 2657 | self.assertEqual(e.attrib, dict(href="#", id="foo")) |
| 2658 | |
| 2659 | e2 = ET.SubElement(elements[0], 'foobar', attrib={'key1': 'value1'}) |
| 2660 | self.assertEqual(e2.attrib['key1'], 'value1') |
| 2661 | |
| 2662 | with self.assertRaisesRegex(TypeError, 'must be dict, not str'): |
| 2663 | ET.Element('a', "I'm not a dict") |
| 2664 | with self.assertRaisesRegex(TypeError, 'must be dict, not str'): |
| 2665 | ET.Element('a', attrib="I'm not a dict") |
| 2666 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2667 | # -------------------------------------------------------------------- |
| 2668 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2669 | class NoAcceleratorTest(unittest.TestCase): |
Eli Bendersky | 52280c4 | 2012-12-30 06:27:56 -0800 | [diff] [blame] | 2670 | def setUp(self): |
| 2671 | if not pyET: |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 2672 | raise unittest.SkipTest('only for the Python version') |
Eli Bendersky | 52280c4 | 2012-12-30 06:27:56 -0800 | [diff] [blame] | 2673 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2674 | # Test that the C accelerator was not imported for pyET |
| 2675 | def test_correct_import_pyET(self): |
Eli Bendersky | e26fa1b | 2013-05-19 17:49:54 -0700 | [diff] [blame] | 2676 | # The type of methods defined in Python code is types.FunctionType, |
| 2677 | # while the type of methods defined inside _elementtree is |
| 2678 | # <class 'wrapper_descriptor'> |
| 2679 | self.assertIsInstance(pyET.Element.__init__, types.FunctionType) |
| 2680 | self.assertIsInstance(pyET.XMLParser.__init__, types.FunctionType) |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2681 | |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2682 | # -------------------------------------------------------------------- |
| 2683 | |
| 2684 | |
| 2685 | class CleanContext(object): |
| 2686 | """Provide default namespace mapping and path cache.""" |
| 2687 | checkwarnings = None |
| 2688 | |
| 2689 | def __init__(self, quiet=False): |
| 2690 | if sys.flags.optimize >= 2: |
| 2691 | # under -OO, doctests cannot be run and therefore not all warnings |
| 2692 | # will be emitted |
| 2693 | quiet = True |
| 2694 | deprecations = ( |
| 2695 | # Search behaviour is broken if search path starts with "/". |
| 2696 | ("This search is broken in 1.3 and earlier, and will be fixed " |
| 2697 | "in a future version. If you rely on the current behaviour, " |
| 2698 | "change it to '.+'", FutureWarning), |
| 2699 | # Element.getchildren() and Element.getiterator() are deprecated. |
| 2700 | ("This method will be removed in future versions. " |
| 2701 | "Use .+ instead.", DeprecationWarning), |
| 2702 | ("This method will be removed in future versions. " |
| 2703 | "Use .+ instead.", PendingDeprecationWarning)) |
| 2704 | self.checkwarnings = support.check_warnings(*deprecations, quiet=quiet) |
| 2705 | |
| 2706 | def __enter__(self): |
| 2707 | from xml.etree import ElementPath |
| 2708 | self._nsmap = ET.register_namespace._namespace_map |
| 2709 | # Copy the default namespace mapping |
| 2710 | self._nsmap_copy = self._nsmap.copy() |
| 2711 | # Copy the path cache (should be empty) |
| 2712 | self._path_cache = ElementPath._cache |
| 2713 | ElementPath._cache = self._path_cache.copy() |
| 2714 | self.checkwarnings.__enter__() |
| 2715 | |
| 2716 | def __exit__(self, *args): |
| 2717 | from xml.etree import ElementPath |
| 2718 | # Restore mapping and path cache |
| 2719 | self._nsmap.clear() |
| 2720 | self._nsmap.update(self._nsmap_copy) |
| 2721 | ElementPath._cache = self._path_cache |
| 2722 | self.checkwarnings.__exit__(*args) |
| 2723 | |
| 2724 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2725 | def test_main(module=None): |
| 2726 | # When invoked without a module, runs the Python ET tests by loading pyET. |
| 2727 | # Otherwise, uses the given module as the ET. |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 2728 | global pyET |
| 2729 | pyET = import_fresh_module('xml.etree.ElementTree', |
| 2730 | blocked=['_elementtree']) |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2731 | if module is None: |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2732 | module = pyET |
Florent Xicluna | 41fe615 | 2010-04-02 18:52:12 +0000 | [diff] [blame] | 2733 | |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2734 | global ET |
| 2735 | ET = module |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2736 | |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 2737 | test_classes = [ |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 2738 | ModuleTest, |
Eli Bendersky | 865756a | 2012-03-09 13:38:15 +0200 | [diff] [blame] | 2739 | ElementSlicingTest, |
Eli Bendersky | 396e8fc | 2012-03-23 14:24:20 +0200 | [diff] [blame] | 2740 | BasicElementTest, |
Serhiy Storchaka | 5bf3120 | 2015-05-18 18:29:33 +0300 | [diff] [blame] | 2741 | BadElementTest, |
| 2742 | BadElementPathTest, |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 2743 | ElementTreeTest, |
Eli Bendersky | 00f402b | 2012-07-15 06:02:22 +0300 | [diff] [blame] | 2744 | IOTest, |
Eli Bendersky | 5b77d81 | 2012-03-16 08:20:05 +0200 | [diff] [blame] | 2745 | ParseErrorTest, |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 2746 | XIncludeTest, |
| 2747 | ElementTreeTypeTest, |
Eli Bendersky | ceab1a9 | 2013-01-12 07:42:46 -0800 | [diff] [blame] | 2748 | ElementFindTest, |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2749 | ElementIterTest, |
Eli Bendersky | 737b173 | 2012-05-29 06:02:56 +0300 | [diff] [blame] | 2750 | TreeBuilderTest, |
Serhiy Storchaka | 66d53fa | 2013-05-22 17:07:51 +0300 | [diff] [blame] | 2751 | XMLParserTest, |
Eli Bendersky | b586934 | 2013-08-30 05:51:20 -0700 | [diff] [blame] | 2752 | XMLPullParserTest, |
Serhiy Storchaka | f8cf59e | 2013-02-25 17:20:59 +0200 | [diff] [blame] | 2753 | BugsTest, |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2754 | ] |
| 2755 | |
| 2756 | # These tests will only run for the pure-Python version that doesn't import |
| 2757 | # _elementtree. We can't use skipUnless here, because pyET is filled in only |
| 2758 | # after the module is loaded. |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 2759 | if pyET is not ET: |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2760 | test_classes.extend([ |
| 2761 | NoAcceleratorTest, |
Eli Bendersky | 64d11e6 | 2012-06-15 07:42:50 +0300 | [diff] [blame] | 2762 | ]) |
Florent Xicluna | 75b5e7e | 2012-03-05 10:42:19 +0100 | [diff] [blame] | 2763 | |
Eli Bendersky | 6319e0f | 2012-06-16 06:47:44 +0300 | [diff] [blame] | 2764 | try: |
Eli Bendersky | 6319e0f | 2012-06-16 06:47:44 +0300 | [diff] [blame] | 2765 | # XXX the C module should give the same warnings as the Python module |
Eli Bendersky | 698bdb2 | 2013-01-10 06:01:06 -0800 | [diff] [blame] | 2766 | with CleanContext(quiet=(pyET is not ET)): |
Eli Bendersky | ceab1a9 | 2013-01-12 07:42:46 -0800 | [diff] [blame] | 2767 | support.run_unittest(*test_classes) |
Eli Bendersky | 6319e0f | 2012-06-16 06:47:44 +0300 | [diff] [blame] | 2768 | finally: |
| 2769 | # don't interfere with subsequent tests |
| 2770 | ET = pyET = None |
| 2771 | |
Florent Xicluna | f15351d | 2010-03-13 23:24:31 +0000 | [diff] [blame] | 2772 | |
Armin Rigo | 9ed7306 | 2005-12-14 18:10:45 +0000 | [diff] [blame] | 2773 | if __name__ == '__main__': |
| 2774 | test_main() |