blob: a6238b278617071e9535cd276357e6ec9b55a84b [file] [log] [blame]
Antoine Pitroud72402e2010-10-27 18:52:48 +00001# regression test for SAX 2.0
Lars Gustäbel96753b32000-09-24 12:24:24 +00002# $Id$
3
Thomas Wouters0e3f5912006-08-11 14:57:12 +00004from xml.sax import make_parser, ContentHandler, \
5 SAXException, SAXReaderNotAvailable, SAXParseException
R David Murraya846f5a2013-03-18 00:18:12 -04006import unittest
Martin v. Löwis962c9e72000-10-06 17:41:52 +00007try:
8 make_parser()
Martin v. Löwis80670bc2000-10-06 21:13:23 +00009except SAXReaderNotAvailable:
Martin v. Löwis962c9e72000-10-06 17:41:52 +000010 # don't try to test this module if we cannot create a parser
R David Murraya846f5a2013-03-18 00:18:12 -040011 raise unittest.SkipTest("no XML parsers available")
Thomas Wouters0e3f5912006-08-11 14:57:12 +000012from xml.sax.saxutils import XMLGenerator, escape, unescape, quoteattr, \
13 XMLFilterBase
14from xml.sax.expatreader import create_parser
Antoine Pitrou6b03ee62010-10-27 18:33:30 +000015from xml.sax.handler import feature_namespaces
Thomas Wouters0e3f5912006-08-11 14:57:12 +000016from xml.sax.xmlreader import InputSource, AttributesImpl, AttributesNSImpl
Serhiy Storchaka88efc522013-02-10 14:29:52 +020017from io import BytesIO, StringIO
Georg Brandlc502df42013-05-12 11:41:12 +020018import codecs
Serhiy Storchaka1a4ed4c2013-02-02 12:17:05 +020019import os.path
Serhiy Storchakad5202392013-02-02 10:31:17 +020020import shutil
21from test import support
Benjamin Petersonee8712c2008-05-20 21:35:26 +000022from test.support import findfile, run_unittest
Florent Xiclunaf15351d2010-03-13 23:24:31 +000023
24TEST_XMLFILE = findfile("test.xml", subdir="xmltestdata")
25TEST_XMLFILE_OUT = findfile("test.xml.out", subdir="xmltestdata")
Victor Stinner6c6f8512010-08-07 10:09:35 +000026try:
Marc-André Lemburg8f36af72011-02-25 15:42:01 +000027 TEST_XMLFILE.encode("utf-8")
28 TEST_XMLFILE_OUT.encode("utf-8")
Victor Stinner6c6f8512010-08-07 10:09:35 +000029except UnicodeEncodeError:
30 raise unittest.SkipTest("filename is not encodable to utf8")
Lars Gustäbel96753b32000-09-24 12:24:24 +000031
Serhiy Storchaka1a4ed4c2013-02-02 12:17:05 +020032supports_nonascii_filenames = True
33if not os.path.supports_unicode_filenames:
34 try:
35 support.TESTFN_UNICODE.encode(support.TESTFN_ENCODING)
36 except (UnicodeError, TypeError):
37 # Either the file system encoding is None, or the file name
38 # cannot be encoded in the file system encoding.
39 supports_nonascii_filenames = False
40requires_nonascii_filenames = unittest.skipUnless(
41 supports_nonascii_filenames,
42 'Requires non-ascii filenames support')
43
Guido van Rossumd8faa362007-04-27 19:54:29 +000044ns_uri = "http://www.python.org/xml-ns/saxtest/"
Lars Gustäbel96753b32000-09-24 12:24:24 +000045
Guido van Rossumd8faa362007-04-27 19:54:29 +000046class XmlTestBase(unittest.TestCase):
47 def verify_empty_attrs(self, attrs):
48 self.assertRaises(KeyError, attrs.getValue, "attr")
49 self.assertRaises(KeyError, attrs.getValueByQName, "attr")
50 self.assertRaises(KeyError, attrs.getNameByQName, "attr")
51 self.assertRaises(KeyError, attrs.getQNameByName, "attr")
52 self.assertRaises(KeyError, attrs.__getitem__, "attr")
Ezio Melottib3aedd42010-11-20 19:04:17 +000053 self.assertEqual(attrs.getLength(), 0)
54 self.assertEqual(attrs.getNames(), [])
55 self.assertEqual(attrs.getQNames(), [])
56 self.assertEqual(len(attrs), 0)
Ezio Melottib58e0bd2010-01-23 15:40:09 +000057 self.assertNotIn("attr", attrs)
Ezio Melottib3aedd42010-11-20 19:04:17 +000058 self.assertEqual(list(attrs.keys()), [])
59 self.assertEqual(attrs.get("attrs"), None)
60 self.assertEqual(attrs.get("attrs", 25), 25)
61 self.assertEqual(list(attrs.items()), [])
62 self.assertEqual(list(attrs.values()), [])
Lars Gustäbel96753b32000-09-24 12:24:24 +000063
Guido van Rossumd8faa362007-04-27 19:54:29 +000064 def verify_empty_nsattrs(self, attrs):
65 self.assertRaises(KeyError, attrs.getValue, (ns_uri, "attr"))
66 self.assertRaises(KeyError, attrs.getValueByQName, "ns:attr")
67 self.assertRaises(KeyError, attrs.getNameByQName, "ns:attr")
68 self.assertRaises(KeyError, attrs.getQNameByName, (ns_uri, "attr"))
69 self.assertRaises(KeyError, attrs.__getitem__, (ns_uri, "attr"))
Ezio Melottib3aedd42010-11-20 19:04:17 +000070 self.assertEqual(attrs.getLength(), 0)
71 self.assertEqual(attrs.getNames(), [])
72 self.assertEqual(attrs.getQNames(), [])
73 self.assertEqual(len(attrs), 0)
Ezio Melottib58e0bd2010-01-23 15:40:09 +000074 self.assertNotIn((ns_uri, "attr"), attrs)
Ezio Melottib3aedd42010-11-20 19:04:17 +000075 self.assertEqual(list(attrs.keys()), [])
76 self.assertEqual(attrs.get((ns_uri, "attr")), None)
77 self.assertEqual(attrs.get((ns_uri, "attr"), 25), 25)
78 self.assertEqual(list(attrs.items()), [])
79 self.assertEqual(list(attrs.values()), [])
Lars Gustäbel96753b32000-09-24 12:24:24 +000080
Guido van Rossumd8faa362007-04-27 19:54:29 +000081 def verify_attrs_wattr(self, attrs):
Ezio Melottib3aedd42010-11-20 19:04:17 +000082 self.assertEqual(attrs.getLength(), 1)
83 self.assertEqual(attrs.getNames(), ["attr"])
84 self.assertEqual(attrs.getQNames(), ["attr"])
85 self.assertEqual(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +000086 self.assertIn("attr", attrs)
Ezio Melottib3aedd42010-11-20 19:04:17 +000087 self.assertEqual(list(attrs.keys()), ["attr"])
88 self.assertEqual(attrs.get("attr"), "val")
89 self.assertEqual(attrs.get("attr", 25), "val")
90 self.assertEqual(list(attrs.items()), [("attr", "val")])
91 self.assertEqual(list(attrs.values()), ["val"])
92 self.assertEqual(attrs.getValue("attr"), "val")
93 self.assertEqual(attrs.getValueByQName("attr"), "val")
94 self.assertEqual(attrs.getNameByQName("attr"), "attr")
95 self.assertEqual(attrs["attr"], "val")
96 self.assertEqual(attrs.getQNameByName("attr"), "attr")
Lars Gustäbel96753b32000-09-24 12:24:24 +000097
Guido van Rossumd8faa362007-04-27 19:54:29 +000098class MakeParserTest(unittest.TestCase):
99 def test_make_parser2(self):
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000100 # Creating parsers several times in a row should succeed.
101 # Testing this because there have been failures of this kind
102 # before.
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000103 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000104 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000105 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000106 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000107 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000108 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000109 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000110 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000111 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000112 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000113 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000114 p = make_parser()
Tim Petersd2bf3b72001-01-18 02:22:22 +0000115
116
Lars Gustäbel96753b32000-09-24 12:24:24 +0000117# ===========================================================================
118#
119# saxutils tests
120#
121# ===========================================================================
122
Guido van Rossumd8faa362007-04-27 19:54:29 +0000123class SaxutilsTest(unittest.TestCase):
124 # ===== escape
125 def test_escape_basic(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000126 self.assertEqual(escape("Donald Duck & Co"), "Donald Duck & Co")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000127
Guido van Rossumd8faa362007-04-27 19:54:29 +0000128 def test_escape_all(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000129 self.assertEqual(escape("<Donald Duck & Co>"),
130 "&lt;Donald Duck &amp; Co&gt;")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000131
Guido van Rossumd8faa362007-04-27 19:54:29 +0000132 def test_escape_extra(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000133 self.assertEqual(escape("Hei på deg", {"å" : "&aring;"}),
134 "Hei p&aring; deg")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000135
Guido van Rossumd8faa362007-04-27 19:54:29 +0000136 # ===== unescape
137 def test_unescape_basic(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000138 self.assertEqual(unescape("Donald Duck &amp; Co"), "Donald Duck & Co")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000139
Guido van Rossumd8faa362007-04-27 19:54:29 +0000140 def test_unescape_all(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000141 self.assertEqual(unescape("&lt;Donald Duck &amp; Co&gt;"),
142 "<Donald Duck & Co>")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000143
Guido van Rossumd8faa362007-04-27 19:54:29 +0000144 def test_unescape_extra(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000145 self.assertEqual(unescape("Hei på deg", {"å" : "&aring;"}),
146 "Hei p&aring; deg")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000147
Guido van Rossumd8faa362007-04-27 19:54:29 +0000148 def test_unescape_amp_extra(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000149 self.assertEqual(unescape("&amp;foo;", {"&foo;": "splat"}), "&foo;")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000150
Guido van Rossumd8faa362007-04-27 19:54:29 +0000151 # ===== quoteattr
152 def test_quoteattr_basic(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000153 self.assertEqual(quoteattr("Donald Duck & Co"),
154 '"Donald Duck &amp; Co"')
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000155
Guido van Rossumd8faa362007-04-27 19:54:29 +0000156 def test_single_quoteattr(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000157 self.assertEqual(quoteattr('Includes "double" quotes'),
158 '\'Includes "double" quotes\'')
Fred Drake32f3add2002-10-28 17:58:48 +0000159
Guido van Rossumd8faa362007-04-27 19:54:29 +0000160 def test_double_quoteattr(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000161 self.assertEqual(quoteattr("Includes 'single' quotes"),
162 "\"Includes 'single' quotes\"")
Fred Drakeacd32d32001-07-19 16:10:15 +0000163
Guido van Rossumd8faa362007-04-27 19:54:29 +0000164 def test_single_double_quoteattr(self):
Ezio Melottib3aedd42010-11-20 19:04:17 +0000165 self.assertEqual(quoteattr("Includes 'single' and \"double\" quotes"),
166 "\"Includes 'single' and &quot;double&quot; quotes\"")
Fred Drakeacd32d32001-07-19 16:10:15 +0000167
Guido van Rossumd8faa362007-04-27 19:54:29 +0000168 # ===== make_parser
169 def test_make_parser(self):
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000170 # Creating a parser should succeed - it should fall back
171 # to the expatreader
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000172 p = make_parser(['xml.parsers.no_such_parser'])
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000173
174
Lars Gustäbel96753b32000-09-24 12:24:24 +0000175# ===== XMLGenerator
176
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200177class XmlgenTest:
Guido van Rossumd8faa362007-04-27 19:54:29 +0000178 def test_xmlgen_basic(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200179 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000180 gen = XMLGenerator(result)
181 gen.startDocument()
182 gen.startElement("doc", {})
183 gen.endElement("doc")
184 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000185
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200186 self.assertEqual(result.getvalue(), self.xml("<doc></doc>"))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000187
R. David Murraya90032a2010-10-17 22:46:45 +0000188 def test_xmlgen_basic_empty(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200189 result = self.ioclass()
R. David Murraya90032a2010-10-17 22:46:45 +0000190 gen = XMLGenerator(result, short_empty_elements=True)
191 gen.startDocument()
192 gen.startElement("doc", {})
193 gen.endElement("doc")
194 gen.endDocument()
195
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200196 self.assertEqual(result.getvalue(), self.xml("<doc/>"))
R. David Murraya90032a2010-10-17 22:46:45 +0000197
Guido van Rossumd8faa362007-04-27 19:54:29 +0000198 def test_xmlgen_content(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200199 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000200 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000201
Guido van Rossumd8faa362007-04-27 19:54:29 +0000202 gen.startDocument()
203 gen.startElement("doc", {})
204 gen.characters("huhei")
205 gen.endElement("doc")
206 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000207
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200208 self.assertEqual(result.getvalue(), self.xml("<doc>huhei</doc>"))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000209
R. David Murraya90032a2010-10-17 22:46:45 +0000210 def test_xmlgen_content_empty(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200211 result = self.ioclass()
R. David Murraya90032a2010-10-17 22:46:45 +0000212 gen = XMLGenerator(result, short_empty_elements=True)
213
214 gen.startDocument()
215 gen.startElement("doc", {})
216 gen.characters("huhei")
217 gen.endElement("doc")
218 gen.endDocument()
219
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200220 self.assertEqual(result.getvalue(), self.xml("<doc>huhei</doc>"))
R. David Murraya90032a2010-10-17 22:46:45 +0000221
Guido van Rossumd8faa362007-04-27 19:54:29 +0000222 def test_xmlgen_pi(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200223 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000224 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000225
Guido van Rossumd8faa362007-04-27 19:54:29 +0000226 gen.startDocument()
227 gen.processingInstruction("test", "data")
228 gen.startElement("doc", {})
229 gen.endElement("doc")
230 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000231
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200232 self.assertEqual(result.getvalue(),
233 self.xml("<?test data?><doc></doc>"))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000234
Guido van Rossumd8faa362007-04-27 19:54:29 +0000235 def test_xmlgen_content_escape(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200236 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000237 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000238
Guido van Rossumd8faa362007-04-27 19:54:29 +0000239 gen.startDocument()
240 gen.startElement("doc", {})
241 gen.characters("<huhei&")
242 gen.endElement("doc")
243 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000244
Ezio Melottib3aedd42010-11-20 19:04:17 +0000245 self.assertEqual(result.getvalue(),
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200246 self.xml("<doc>&lt;huhei&amp;</doc>"))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000247
Guido van Rossumd8faa362007-04-27 19:54:29 +0000248 def test_xmlgen_attr_escape(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200249 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000250 gen = XMLGenerator(result)
Fred Drakec9fadf92001-08-07 19:17:06 +0000251
Guido van Rossumd8faa362007-04-27 19:54:29 +0000252 gen.startDocument()
253 gen.startElement("doc", {"a": '"'})
254 gen.startElement("e", {"a": "'"})
255 gen.endElement("e")
256 gen.startElement("e", {"a": "'\""})
257 gen.endElement("e")
258 gen.startElement("e", {"a": "\n\r\t"})
259 gen.endElement("e")
260 gen.endElement("doc")
261 gen.endDocument()
Fred Drakec9fadf92001-08-07 19:17:06 +0000262
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200263 self.assertEqual(result.getvalue(), self.xml(
264 "<doc a='\"'><e a=\"'\"></e>"
265 "<e a=\"'&quot;\"></e>"
266 "<e a=\"&#10;&#13;&#9;\"></e></doc>"))
267
268 def test_xmlgen_encoding(self):
269 encodings = ('iso-8859-15', 'utf-8', 'utf-8-sig',
270 'utf-16', 'utf-16be', 'utf-16le',
271 'utf-32', 'utf-32be', 'utf-32le')
272 for encoding in encodings:
273 result = self.ioclass()
274 gen = XMLGenerator(result, encoding=encoding)
275
276 gen.startDocument()
277 gen.startElement("doc", {"a": '\u20ac'})
278 gen.characters("\u20ac")
279 gen.endElement("doc")
280 gen.endDocument()
281
282 self.assertEqual(result.getvalue(),
283 self.xml('<doc a="\u20ac">\u20ac</doc>', encoding=encoding))
284
285 def test_xmlgen_unencodable(self):
286 result = self.ioclass()
287 gen = XMLGenerator(result, encoding='ascii')
288
289 gen.startDocument()
290 gen.startElement("doc", {"a": '\u20ac'})
291 gen.characters("\u20ac")
292 gen.endElement("doc")
293 gen.endDocument()
294
295 self.assertEqual(result.getvalue(),
296 self.xml('<doc a="&#8364;">&#8364;</doc>', encoding='ascii'))
Fred Drakec9fadf92001-08-07 19:17:06 +0000297
Guido van Rossumd8faa362007-04-27 19:54:29 +0000298 def test_xmlgen_ignorable(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200299 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000300 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000301
Guido van Rossumd8faa362007-04-27 19:54:29 +0000302 gen.startDocument()
303 gen.startElement("doc", {})
304 gen.ignorableWhitespace(" ")
305 gen.endElement("doc")
306 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000307
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200308 self.assertEqual(result.getvalue(), self.xml("<doc> </doc>"))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000309
R. David Murraya90032a2010-10-17 22:46:45 +0000310 def test_xmlgen_ignorable_empty(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200311 result = self.ioclass()
R. David Murraya90032a2010-10-17 22:46:45 +0000312 gen = XMLGenerator(result, short_empty_elements=True)
313
314 gen.startDocument()
315 gen.startElement("doc", {})
316 gen.ignorableWhitespace(" ")
317 gen.endElement("doc")
318 gen.endDocument()
319
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200320 self.assertEqual(result.getvalue(), self.xml("<doc> </doc>"))
R. David Murraya90032a2010-10-17 22:46:45 +0000321
Serhiy Storchaka3eab6b32013-05-12 17:31:16 +0300322 def test_xmlgen_encoding_bytes(self):
323 encodings = ('iso-8859-15', 'utf-8', 'utf-8-sig',
324 'utf-16', 'utf-16be', 'utf-16le',
325 'utf-32', 'utf-32be', 'utf-32le')
326 for encoding in encodings:
327 result = self.ioclass()
328 gen = XMLGenerator(result, encoding=encoding)
329
330 gen.startDocument()
331 gen.startElement("doc", {"a": '\u20ac'})
332 gen.characters("\u20ac".encode(encoding))
333 gen.ignorableWhitespace(" ".encode(encoding))
334 gen.endElement("doc")
335 gen.endDocument()
336
337 self.assertEqual(result.getvalue(),
338 self.xml('<doc a="\u20ac">\u20ac </doc>', encoding=encoding))
339
Guido van Rossumd8faa362007-04-27 19:54:29 +0000340 def test_xmlgen_ns(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200341 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000342 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000343
Guido van Rossumd8faa362007-04-27 19:54:29 +0000344 gen.startDocument()
345 gen.startPrefixMapping("ns1", ns_uri)
346 gen.startElementNS((ns_uri, "doc"), "ns1:doc", {})
347 # add an unqualified name
348 gen.startElementNS((None, "udoc"), None, {})
349 gen.endElementNS((None, "udoc"), None)
350 gen.endElementNS((ns_uri, "doc"), "ns1:doc")
351 gen.endPrefixMapping("ns1")
352 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000353
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200354 self.assertEqual(result.getvalue(), self.xml(
355 '<ns1:doc xmlns:ns1="%s"><udoc></udoc></ns1:doc>' %
Guido van Rossumd8faa362007-04-27 19:54:29 +0000356 ns_uri))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000357
R. David Murraya90032a2010-10-17 22:46:45 +0000358 def test_xmlgen_ns_empty(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200359 result = self.ioclass()
R. David Murraya90032a2010-10-17 22:46:45 +0000360 gen = XMLGenerator(result, short_empty_elements=True)
361
362 gen.startDocument()
363 gen.startPrefixMapping("ns1", ns_uri)
364 gen.startElementNS((ns_uri, "doc"), "ns1:doc", {})
365 # add an unqualified name
366 gen.startElementNS((None, "udoc"), None, {})
367 gen.endElementNS((None, "udoc"), None)
368 gen.endElementNS((ns_uri, "doc"), "ns1:doc")
369 gen.endPrefixMapping("ns1")
370 gen.endDocument()
371
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200372 self.assertEqual(result.getvalue(), self.xml(
373 '<ns1:doc xmlns:ns1="%s"><udoc/></ns1:doc>' %
R. David Murraya90032a2010-10-17 22:46:45 +0000374 ns_uri))
375
Guido van Rossumd8faa362007-04-27 19:54:29 +0000376 def test_1463026_1(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200377 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000378 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000379
Guido van Rossumd8faa362007-04-27 19:54:29 +0000380 gen.startDocument()
381 gen.startElementNS((None, 'a'), 'a', {(None, 'b'):'c'})
382 gen.endElementNS((None, 'a'), 'a')
383 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000384
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200385 self.assertEqual(result.getvalue(), self.xml('<a b="c"></a>'))
Thomas Wouterscf297e42007-02-23 15:07:44 +0000386
R. David Murraya90032a2010-10-17 22:46:45 +0000387 def test_1463026_1_empty(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200388 result = self.ioclass()
R. David Murraya90032a2010-10-17 22:46:45 +0000389 gen = XMLGenerator(result, short_empty_elements=True)
390
391 gen.startDocument()
392 gen.startElementNS((None, 'a'), 'a', {(None, 'b'):'c'})
393 gen.endElementNS((None, 'a'), 'a')
394 gen.endDocument()
395
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200396 self.assertEqual(result.getvalue(), self.xml('<a b="c"/>'))
R. David Murraya90032a2010-10-17 22:46:45 +0000397
Guido van Rossumd8faa362007-04-27 19:54:29 +0000398 def test_1463026_2(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200399 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000400 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000401
Guido van Rossumd8faa362007-04-27 19:54:29 +0000402 gen.startDocument()
403 gen.startPrefixMapping(None, 'qux')
404 gen.startElementNS(('qux', 'a'), 'a', {})
405 gen.endElementNS(('qux', 'a'), 'a')
406 gen.endPrefixMapping(None)
407 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000408
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200409 self.assertEqual(result.getvalue(), self.xml('<a xmlns="qux"></a>'))
Thomas Wouterscf297e42007-02-23 15:07:44 +0000410
R. David Murraya90032a2010-10-17 22:46:45 +0000411 def test_1463026_2_empty(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200412 result = self.ioclass()
R. David Murraya90032a2010-10-17 22:46:45 +0000413 gen = XMLGenerator(result, short_empty_elements=True)
414
415 gen.startDocument()
416 gen.startPrefixMapping(None, 'qux')
417 gen.startElementNS(('qux', 'a'), 'a', {})
418 gen.endElementNS(('qux', 'a'), 'a')
419 gen.endPrefixMapping(None)
420 gen.endDocument()
421
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200422 self.assertEqual(result.getvalue(), self.xml('<a xmlns="qux"/>'))
R. David Murraya90032a2010-10-17 22:46:45 +0000423
Guido van Rossumd8faa362007-04-27 19:54:29 +0000424 def test_1463026_3(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200425 result = self.ioclass()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000426 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000427
Guido van Rossumd8faa362007-04-27 19:54:29 +0000428 gen.startDocument()
429 gen.startPrefixMapping('my', 'qux')
430 gen.startElementNS(('qux', 'a'), 'a', {(None, 'b'):'c'})
431 gen.endElementNS(('qux', 'a'), 'a')
432 gen.endPrefixMapping('my')
433 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000434
Ezio Melottib3aedd42010-11-20 19:04:17 +0000435 self.assertEqual(result.getvalue(),
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200436 self.xml('<my:a xmlns:my="qux" b="c"></my:a>'))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000437
R. David Murraya90032a2010-10-17 22:46:45 +0000438 def test_1463026_3_empty(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200439 result = self.ioclass()
R. David Murraya90032a2010-10-17 22:46:45 +0000440 gen = XMLGenerator(result, short_empty_elements=True)
441
442 gen.startDocument()
443 gen.startPrefixMapping('my', 'qux')
444 gen.startElementNS(('qux', 'a'), 'a', {(None, 'b'):'c'})
445 gen.endElementNS(('qux', 'a'), 'a')
446 gen.endPrefixMapping('my')
447 gen.endDocument()
448
Ezio Melottib3aedd42010-11-20 19:04:17 +0000449 self.assertEqual(result.getvalue(),
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200450 self.xml('<my:a xmlns:my="qux" b="c"/>'))
R. David Murraya90032a2010-10-17 22:46:45 +0000451
Antoine Pitrou6b03ee62010-10-27 18:33:30 +0000452 def test_5027_1(self):
453 # The xml prefix (as in xml:lang below) is reserved and bound by
454 # definition to http://www.w3.org/XML/1998/namespace. XMLGenerator had
Andrew Svetlov737fb892012-12-18 21:14:22 +0200455 # a bug whereby a KeyError is raised because this namespace is missing
Antoine Pitrou6b03ee62010-10-27 18:33:30 +0000456 # from a dictionary.
457 #
458 # This test demonstrates the bug by parsing a document.
459 test_xml = StringIO(
460 '<?xml version="1.0"?>'
461 '<a:g1 xmlns:a="http://example.com/ns">'
462 '<a:g2 xml:lang="en">Hello</a:g2>'
463 '</a:g1>')
464
465 parser = make_parser()
466 parser.setFeature(feature_namespaces, True)
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200467 result = self.ioclass()
Antoine Pitrou6b03ee62010-10-27 18:33:30 +0000468 gen = XMLGenerator(result)
469 parser.setContentHandler(gen)
470 parser.parse(test_xml)
471
Ezio Melottib3aedd42010-11-20 19:04:17 +0000472 self.assertEqual(result.getvalue(),
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200473 self.xml(
Ezio Melottib3aedd42010-11-20 19:04:17 +0000474 '<a:g1 xmlns:a="http://example.com/ns">'
475 '<a:g2 xml:lang="en">Hello</a:g2>'
476 '</a:g1>'))
Antoine Pitrou6b03ee62010-10-27 18:33:30 +0000477
478 def test_5027_2(self):
479 # The xml prefix (as in xml:lang below) is reserved and bound by
480 # definition to http://www.w3.org/XML/1998/namespace. XMLGenerator had
Andrew Svetlov737fb892012-12-18 21:14:22 +0200481 # a bug whereby a KeyError is raised because this namespace is missing
Antoine Pitrou6b03ee62010-10-27 18:33:30 +0000482 # from a dictionary.
483 #
484 # This test demonstrates the bug by direct manipulation of the
485 # XMLGenerator.
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200486 result = self.ioclass()
Antoine Pitrou6b03ee62010-10-27 18:33:30 +0000487 gen = XMLGenerator(result)
488
489 gen.startDocument()
490 gen.startPrefixMapping('a', 'http://example.com/ns')
491 gen.startElementNS(('http://example.com/ns', 'g1'), 'g1', {})
492 lang_attr = {('http://www.w3.org/XML/1998/namespace', 'lang'): 'en'}
493 gen.startElementNS(('http://example.com/ns', 'g2'), 'g2', lang_attr)
494 gen.characters('Hello')
495 gen.endElementNS(('http://example.com/ns', 'g2'), 'g2')
496 gen.endElementNS(('http://example.com/ns', 'g1'), 'g1')
497 gen.endPrefixMapping('a')
498 gen.endDocument()
499
Ezio Melottib3aedd42010-11-20 19:04:17 +0000500 self.assertEqual(result.getvalue(),
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200501 self.xml(
Ezio Melottib3aedd42010-11-20 19:04:17 +0000502 '<a:g1 xmlns:a="http://example.com/ns">'
503 '<a:g2 xml:lang="en">Hello</a:g2>'
504 '</a:g1>'))
Antoine Pitrou6b03ee62010-10-27 18:33:30 +0000505
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200506 def test_no_close_file(self):
507 result = self.ioclass()
508 def func(out):
509 gen = XMLGenerator(out)
510 gen.startDocument()
511 gen.startElement("doc", {})
512 func(result)
513 self.assertFalse(result.closed)
514
Serhiy Storchakaa5f13d22013-02-25 13:46:10 +0200515 def test_xmlgen_fragment(self):
516 result = self.ioclass()
517 gen = XMLGenerator(result)
518
519 # Don't call gen.startDocument()
520 gen.startElement("foo", {"a": "1.0"})
521 gen.characters("Hello")
522 gen.endElement("foo")
523 gen.startElement("bar", {"b": "2.0"})
524 gen.endElement("bar")
525 # Don't call gen.endDocument()
526
527 self.assertEqual(result.getvalue(),
528 self.xml('<foo a="1.0">Hello</foo><bar b="2.0"></bar>')[len(self.xml('')):])
529
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200530class StringXmlgenTest(XmlgenTest, unittest.TestCase):
531 ioclass = StringIO
532
533 def xml(self, doc, encoding='iso-8859-1'):
534 return '<?xml version="1.0" encoding="%s"?>\n%s' % (encoding, doc)
535
536 test_xmlgen_unencodable = None
537
538class BytesXmlgenTest(XmlgenTest, unittest.TestCase):
539 ioclass = BytesIO
540
541 def xml(self, doc, encoding='iso-8859-1'):
542 return ('<?xml version="1.0" encoding="%s"?>\n%s' %
543 (encoding, doc)).encode(encoding, 'xmlcharrefreplace')
544
545class WriterXmlgenTest(BytesXmlgenTest):
546 class ioclass(list):
547 write = list.append
548 closed = False
549
550 def seekable(self):
551 return True
552
553 def tell(self):
554 # return 0 at start and not 0 after start
555 return len(self)
556
557 def getvalue(self):
558 return b''.join(self)
559
Georg Brandlc502df42013-05-12 11:41:12 +0200560class StreamWriterXmlgenTest(XmlgenTest, unittest.TestCase):
561 def ioclass(self):
562 raw = BytesIO()
563 writer = codecs.getwriter('ascii')(raw, 'xmlcharrefreplace')
564 writer.getvalue = raw.getvalue
565 return writer
566
567 def xml(self, doc, encoding='iso-8859-1'):
568 return ('<?xml version="1.0" encoding="%s"?>\n%s' %
569 (encoding, doc)).encode('ascii', 'xmlcharrefreplace')
570
571class StreamReaderWriterXmlgenTest(XmlgenTest, unittest.TestCase):
572 fname = support.TESTFN + '-codecs'
573
574 def ioclass(self):
575 writer = codecs.open(self.fname, 'w', encoding='ascii',
576 errors='xmlcharrefreplace', buffering=0)
Antoine Pitrou2adb6fe2013-05-13 22:34:21 +0200577 def cleanup():
578 writer.close()
579 support.unlink(self.fname)
580 self.addCleanup(cleanup)
Richard Oudkerk90a24272013-05-18 18:11:30 +0100581 def getvalue():
582 # Windows will not let use reopen without first closing
583 writer.close()
584 with open(writer.name, 'rb') as f:
585 return f.read()
586 writer.getvalue = getvalue
Georg Brandlc502df42013-05-12 11:41:12 +0200587 return writer
588
Georg Brandlc502df42013-05-12 11:41:12 +0200589 def xml(self, doc, encoding='iso-8859-1'):
590 return ('<?xml version="1.0" encoding="%s"?>\n%s' %
591 (encoding, doc)).encode('ascii', 'xmlcharrefreplace')
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200592
593start = b'<?xml version="1.0" encoding="iso-8859-1"?>\n'
594
Fred Drake004d5e62000-10-23 17:22:08 +0000595
Guido van Rossumd8faa362007-04-27 19:54:29 +0000596class XMLFilterBaseTest(unittest.TestCase):
597 def test_filter_basic(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200598 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000599 gen = XMLGenerator(result)
600 filter = XMLFilterBase()
601 filter.setContentHandler(gen)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000602
Guido van Rossumd8faa362007-04-27 19:54:29 +0000603 filter.startDocument()
604 filter.startElement("doc", {})
605 filter.characters("content")
606 filter.ignorableWhitespace(" ")
607 filter.endElement("doc")
608 filter.endDocument()
609
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200610 self.assertEqual(result.getvalue(), start + b"<doc>content </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000611
612# ===========================================================================
613#
614# expatreader tests
615#
616# ===========================================================================
617
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200618with open(TEST_XMLFILE_OUT, 'rb') as f:
Benjamin Petersond5df36d2010-10-31 18:23:23 +0000619 xml_test_out = f.read()
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000620
Guido van Rossumd8faa362007-04-27 19:54:29 +0000621class ExpatReaderTest(XmlTestBase):
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000622
Guido van Rossumd8faa362007-04-27 19:54:29 +0000623 # ===== XMLReader support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000624
Guido van Rossumd8faa362007-04-27 19:54:29 +0000625 def test_expat_file(self):
626 parser = create_parser()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200627 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000628 xmlgen = XMLGenerator(result)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000629
Guido van Rossumd8faa362007-04-27 19:54:29 +0000630 parser.setContentHandler(xmlgen)
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200631 with open(TEST_XMLFILE, 'rb') as f:
Benjamin Petersond5df36d2010-10-31 18:23:23 +0000632 parser.parse(f)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000633
Ezio Melottib3aedd42010-11-20 19:04:17 +0000634 self.assertEqual(result.getvalue(), xml_test_out)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000635
Serhiy Storchaka1a4ed4c2013-02-02 12:17:05 +0200636 @requires_nonascii_filenames
Serhiy Storchakad5202392013-02-02 10:31:17 +0200637 def test_expat_file_nonascii(self):
638 fname = support.TESTFN_UNICODE
639 shutil.copyfile(TEST_XMLFILE, fname)
640 self.addCleanup(support.unlink, fname)
641
642 parser = create_parser()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200643 result = BytesIO()
Serhiy Storchakad5202392013-02-02 10:31:17 +0200644 xmlgen = XMLGenerator(result)
645
646 parser.setContentHandler(xmlgen)
647 parser.parse(open(fname))
648
649 self.assertEqual(result.getvalue(), xml_test_out)
650
Serhiy Storchakafc8e9b02014-11-27 22:13:16 +0200651 def test_expat_binary_file_bytes_name(self):
652 fname = os.fsencode(TEST_XMLFILE)
653 parser = create_parser()
654 result = BytesIO()
655 xmlgen = XMLGenerator(result)
656
657 parser.setContentHandler(xmlgen)
658 with open(fname, 'rb') as f:
659 parser.parse(f)
660
661 self.assertEqual(result.getvalue(), xml_test_out)
662
663 def test_expat_binary_file_int_name(self):
664 parser = create_parser()
665 result = BytesIO()
666 xmlgen = XMLGenerator(result)
667
668 parser.setContentHandler(xmlgen)
669 with open(TEST_XMLFILE, 'rb') as f:
670 with open(f.fileno(), 'rb', closefd=False) as f2:
671 parser.parse(f2)
672
673 self.assertEqual(result.getvalue(), xml_test_out)
674
Guido van Rossumd8faa362007-04-27 19:54:29 +0000675 # ===== DTDHandler support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000676
Guido van Rossumd8faa362007-04-27 19:54:29 +0000677 class TestDTDHandler:
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000678
Guido van Rossumd8faa362007-04-27 19:54:29 +0000679 def __init__(self):
680 self._notations = []
681 self._entities = []
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000682
Guido van Rossumd8faa362007-04-27 19:54:29 +0000683 def notationDecl(self, name, publicId, systemId):
684 self._notations.append((name, publicId, systemId))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000685
Guido van Rossumd8faa362007-04-27 19:54:29 +0000686 def unparsedEntityDecl(self, name, publicId, systemId, ndata):
687 self._entities.append((name, publicId, systemId, ndata))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000688
Guido van Rossumd8faa362007-04-27 19:54:29 +0000689 def test_expat_dtdhandler(self):
690 parser = create_parser()
691 handler = self.TestDTDHandler()
692 parser.setDTDHandler(handler)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000693
Guido van Rossumd8faa362007-04-27 19:54:29 +0000694 parser.feed('<!DOCTYPE doc [\n')
695 parser.feed(' <!ENTITY img SYSTEM "expat.gif" NDATA GIF>\n')
696 parser.feed(' <!NOTATION GIF PUBLIC "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN">\n')
697 parser.feed(']>\n')
698 parser.feed('<doc></doc>')
699 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000700
Ezio Melottib3aedd42010-11-20 19:04:17 +0000701 self.assertEqual(handler._notations,
Guido van Rossumd8faa362007-04-27 19:54:29 +0000702 [("GIF", "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN", None)])
Ezio Melottib3aedd42010-11-20 19:04:17 +0000703 self.assertEqual(handler._entities, [("img", None, "expat.gif", "GIF")])
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000704
Guido van Rossumd8faa362007-04-27 19:54:29 +0000705 # ===== EntityResolver support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000706
Guido van Rossumd8faa362007-04-27 19:54:29 +0000707 class TestEntityResolver:
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000708
Guido van Rossumd8faa362007-04-27 19:54:29 +0000709 def resolveEntity(self, publicId, systemId):
710 inpsrc = InputSource()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200711 inpsrc.setByteStream(BytesIO(b"<entity/>"))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000712 return inpsrc
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000713
Guido van Rossumd8faa362007-04-27 19:54:29 +0000714 def test_expat_entityresolver(self):
715 parser = create_parser()
716 parser.setEntityResolver(self.TestEntityResolver())
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200717 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000718 parser.setContentHandler(XMLGenerator(result))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000719
Guido van Rossumd8faa362007-04-27 19:54:29 +0000720 parser.feed('<!DOCTYPE doc [\n')
721 parser.feed(' <!ENTITY test SYSTEM "whatever">\n')
722 parser.feed(']>\n')
723 parser.feed('<doc>&test;</doc>')
724 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000725
Ezio Melottib3aedd42010-11-20 19:04:17 +0000726 self.assertEqual(result.getvalue(), start +
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200727 b"<doc><entity></entity></doc>")
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000728
Guido van Rossumd8faa362007-04-27 19:54:29 +0000729 # ===== Attributes support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000730
Guido van Rossumd8faa362007-04-27 19:54:29 +0000731 class AttrGatherer(ContentHandler):
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000732
Guido van Rossumd8faa362007-04-27 19:54:29 +0000733 def startElement(self, name, attrs):
734 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000735
Guido van Rossumd8faa362007-04-27 19:54:29 +0000736 def startElementNS(self, name, qname, attrs):
737 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000738
Guido van Rossumd8faa362007-04-27 19:54:29 +0000739 def test_expat_attrs_empty(self):
740 parser = create_parser()
741 gather = self.AttrGatherer()
742 parser.setContentHandler(gather)
743
744 parser.feed("<doc/>")
745 parser.close()
746
747 self.verify_empty_attrs(gather._attrs)
748
749 def test_expat_attrs_wattr(self):
750 parser = create_parser()
751 gather = self.AttrGatherer()
752 parser.setContentHandler(gather)
753
754 parser.feed("<doc attr='val'/>")
755 parser.close()
756
757 self.verify_attrs_wattr(gather._attrs)
758
759 def test_expat_nsattrs_empty(self):
760 parser = create_parser(1)
761 gather = self.AttrGatherer()
762 parser.setContentHandler(gather)
763
764 parser.feed("<doc/>")
765 parser.close()
766
767 self.verify_empty_nsattrs(gather._attrs)
768
769 def test_expat_nsattrs_wattr(self):
770 parser = create_parser(1)
771 gather = self.AttrGatherer()
772 parser.setContentHandler(gather)
773
774 parser.feed("<doc xmlns:ns='%s' ns:attr='val'/>" % ns_uri)
775 parser.close()
776
777 attrs = gather._attrs
778
Ezio Melottib3aedd42010-11-20 19:04:17 +0000779 self.assertEqual(attrs.getLength(), 1)
780 self.assertEqual(attrs.getNames(), [(ns_uri, "attr")])
Guido van Rossumd8faa362007-04-27 19:54:29 +0000781 self.assertTrue((attrs.getQNames() == [] or
782 attrs.getQNames() == ["ns:attr"]))
Ezio Melottib3aedd42010-11-20 19:04:17 +0000783 self.assertEqual(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +0000784 self.assertIn((ns_uri, "attr"), attrs)
Ezio Melottib3aedd42010-11-20 19:04:17 +0000785 self.assertEqual(attrs.get((ns_uri, "attr")), "val")
786 self.assertEqual(attrs.get((ns_uri, "attr"), 25), "val")
787 self.assertEqual(list(attrs.items()), [((ns_uri, "attr"), "val")])
788 self.assertEqual(list(attrs.values()), ["val"])
789 self.assertEqual(attrs.getValue((ns_uri, "attr")), "val")
790 self.assertEqual(attrs[(ns_uri, "attr")], "val")
Guido van Rossumd8faa362007-04-27 19:54:29 +0000791
792 # ===== InputSource support
793
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000794 def test_expat_inpsource_filename(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000795 parser = create_parser()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200796 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000797 xmlgen = XMLGenerator(result)
798
799 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000800 parser.parse(TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000801
Ezio Melottib3aedd42010-11-20 19:04:17 +0000802 self.assertEqual(result.getvalue(), xml_test_out)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000803
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000804 def test_expat_inpsource_sysid(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000805 parser = create_parser()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200806 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000807 xmlgen = XMLGenerator(result)
808
809 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000810 parser.parse(InputSource(TEST_XMLFILE))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000811
Ezio Melottib3aedd42010-11-20 19:04:17 +0000812 self.assertEqual(result.getvalue(), xml_test_out)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000813
Serhiy Storchaka1a4ed4c2013-02-02 12:17:05 +0200814 @requires_nonascii_filenames
Serhiy Storchakad5202392013-02-02 10:31:17 +0200815 def test_expat_inpsource_sysid_nonascii(self):
816 fname = support.TESTFN_UNICODE
817 shutil.copyfile(TEST_XMLFILE, fname)
818 self.addCleanup(support.unlink, fname)
819
820 parser = create_parser()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200821 result = BytesIO()
Serhiy Storchakad5202392013-02-02 10:31:17 +0200822 xmlgen = XMLGenerator(result)
823
824 parser.setContentHandler(xmlgen)
825 parser.parse(InputSource(fname))
826
827 self.assertEqual(result.getvalue(), xml_test_out)
828
Guido van Rossumd8faa362007-04-27 19:54:29 +0000829 def test_expat_inpsource_stream(self):
830 parser = create_parser()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200831 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000832 xmlgen = XMLGenerator(result)
833
834 parser.setContentHandler(xmlgen)
835 inpsrc = InputSource()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200836 with open(TEST_XMLFILE, 'rb') as f:
Benjamin Petersond5df36d2010-10-31 18:23:23 +0000837 inpsrc.setByteStream(f)
838 parser.parse(inpsrc)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000839
Ezio Melottib3aedd42010-11-20 19:04:17 +0000840 self.assertEqual(result.getvalue(), xml_test_out)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000841
842 # ===== IncrementalParser support
843
844 def test_expat_incremental(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200845 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000846 xmlgen = XMLGenerator(result)
847 parser = create_parser()
848 parser.setContentHandler(xmlgen)
849
850 parser.feed("<doc>")
851 parser.feed("</doc>")
852 parser.close()
853
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200854 self.assertEqual(result.getvalue(), start + b"<doc></doc>")
Guido van Rossumd8faa362007-04-27 19:54:29 +0000855
856 def test_expat_incremental_reset(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200857 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000858 xmlgen = XMLGenerator(result)
859 parser = create_parser()
860 parser.setContentHandler(xmlgen)
861
862 parser.feed("<doc>")
863 parser.feed("text")
864
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200865 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000866 xmlgen = XMLGenerator(result)
867 parser.setContentHandler(xmlgen)
868 parser.reset()
869
870 parser.feed("<doc>")
871 parser.feed("text")
872 parser.feed("</doc>")
873 parser.close()
874
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200875 self.assertEqual(result.getvalue(), start + b"<doc>text</doc>")
Guido van Rossumd8faa362007-04-27 19:54:29 +0000876
877 # ===== Locator support
878
879 def test_expat_locator_noinfo(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200880 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000881 xmlgen = XMLGenerator(result)
882 parser = create_parser()
883 parser.setContentHandler(xmlgen)
884
885 parser.feed("<doc>")
886 parser.feed("</doc>")
887 parser.close()
888
Ezio Melottib3aedd42010-11-20 19:04:17 +0000889 self.assertEqual(parser.getSystemId(), None)
890 self.assertEqual(parser.getPublicId(), None)
891 self.assertEqual(parser.getLineNumber(), 1)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000892
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000893 def test_expat_locator_withinfo(self):
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200894 result = BytesIO()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000895 xmlgen = XMLGenerator(result)
896 parser = create_parser()
897 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000898 parser.parse(TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000899
Ezio Melottib3aedd42010-11-20 19:04:17 +0000900 self.assertEqual(parser.getSystemId(), TEST_XMLFILE)
901 self.assertEqual(parser.getPublicId(), None)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000902
Serhiy Storchaka1a4ed4c2013-02-02 12:17:05 +0200903 @requires_nonascii_filenames
Serhiy Storchakad5202392013-02-02 10:31:17 +0200904 def test_expat_locator_withinfo_nonascii(self):
905 fname = support.TESTFN_UNICODE
906 shutil.copyfile(TEST_XMLFILE, fname)
907 self.addCleanup(support.unlink, fname)
908
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200909 result = BytesIO()
Serhiy Storchakad5202392013-02-02 10:31:17 +0200910 xmlgen = XMLGenerator(result)
911 parser = create_parser()
912 parser.setContentHandler(xmlgen)
913 parser.parse(fname)
914
915 self.assertEqual(parser.getSystemId(), fname)
916 self.assertEqual(parser.getPublicId(), None)
917
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000918
919# ===========================================================================
920#
921# error reporting
922#
923# ===========================================================================
924
Guido van Rossumd8faa362007-04-27 19:54:29 +0000925class ErrorReportingTest(unittest.TestCase):
926 def test_expat_inpsource_location(self):
927 parser = create_parser()
928 parser.setContentHandler(ContentHandler()) # do nothing
929 source = InputSource()
Serhiy Storchaka88efc522013-02-10 14:29:52 +0200930 source.setByteStream(BytesIO(b"<foo bar foobar>")) #ill-formed
Guido van Rossumd8faa362007-04-27 19:54:29 +0000931 name = "a file name"
932 source.setSystemId(name)
933 try:
934 parser.parse(source)
935 self.fail()
936 except SAXException as e:
Ezio Melottib3aedd42010-11-20 19:04:17 +0000937 self.assertEqual(e.getSystemId(), name)
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000938
Guido van Rossumd8faa362007-04-27 19:54:29 +0000939 def test_expat_incomplete(self):
940 parser = create_parser()
941 parser.setContentHandler(ContentHandler()) # do nothing
942 self.assertRaises(SAXParseException, parser.parse, StringIO("<foo>"))
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000943
Guido van Rossumd8faa362007-04-27 19:54:29 +0000944 def test_sax_parse_exception_str(self):
945 # pass various values from a locator to the SAXParseException to
946 # make sure that the __str__() doesn't fall apart when None is
947 # passed instead of an integer line and column number
948 #
949 # use "normal" values for the locator:
950 str(SAXParseException("message", None,
951 self.DummyLocator(1, 1)))
952 # use None for the line number:
953 str(SAXParseException("message", None,
954 self.DummyLocator(None, 1)))
955 # use None for the column number:
956 str(SAXParseException("message", None,
957 self.DummyLocator(1, None)))
958 # use None for both:
959 str(SAXParseException("message", None,
960 self.DummyLocator(None, None)))
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000961
Guido van Rossumd8faa362007-04-27 19:54:29 +0000962 class DummyLocator:
963 def __init__(self, lineno, colno):
964 self._lineno = lineno
965 self._colno = colno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000966
Guido van Rossumd8faa362007-04-27 19:54:29 +0000967 def getPublicId(self):
968 return "pubid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000969
Guido van Rossumd8faa362007-04-27 19:54:29 +0000970 def getSystemId(self):
971 return "sysid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000972
Guido van Rossumd8faa362007-04-27 19:54:29 +0000973 def getLineNumber(self):
974 return self._lineno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000975
Guido van Rossumd8faa362007-04-27 19:54:29 +0000976 def getColumnNumber(self):
977 return self._colno
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000978
Lars Gustäbelab647872000-09-24 18:40:52 +0000979# ===========================================================================
980#
981# xmlreader tests
982#
983# ===========================================================================
984
Guido van Rossumd8faa362007-04-27 19:54:29 +0000985class XmlReaderTest(XmlTestBase):
Lars Gustäbelab647872000-09-24 18:40:52 +0000986
Guido van Rossumd8faa362007-04-27 19:54:29 +0000987 # ===== AttributesImpl
988 def test_attrs_empty(self):
989 self.verify_empty_attrs(AttributesImpl({}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000990
Guido van Rossumd8faa362007-04-27 19:54:29 +0000991 def test_attrs_wattr(self):
992 self.verify_attrs_wattr(AttributesImpl({"attr" : "val"}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000993
Guido van Rossumd8faa362007-04-27 19:54:29 +0000994 def test_nsattrs_empty(self):
995 self.verify_empty_nsattrs(AttributesNSImpl({}, {}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000996
Guido van Rossumd8faa362007-04-27 19:54:29 +0000997 def test_nsattrs_wattr(self):
998 attrs = AttributesNSImpl({(ns_uri, "attr") : "val"},
999 {(ns_uri, "attr") : "ns:attr"})
Fred Drake004d5e62000-10-23 17:22:08 +00001000
Ezio Melottib3aedd42010-11-20 19:04:17 +00001001 self.assertEqual(attrs.getLength(), 1)
1002 self.assertEqual(attrs.getNames(), [(ns_uri, "attr")])
1003 self.assertEqual(attrs.getQNames(), ["ns:attr"])
1004 self.assertEqual(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +00001005 self.assertIn((ns_uri, "attr"), attrs)
Ezio Melottib3aedd42010-11-20 19:04:17 +00001006 self.assertEqual(list(attrs.keys()), [(ns_uri, "attr")])
1007 self.assertEqual(attrs.get((ns_uri, "attr")), "val")
1008 self.assertEqual(attrs.get((ns_uri, "attr"), 25), "val")
1009 self.assertEqual(list(attrs.items()), [((ns_uri, "attr"), "val")])
1010 self.assertEqual(list(attrs.values()), ["val"])
1011 self.assertEqual(attrs.getValue((ns_uri, "attr")), "val")
1012 self.assertEqual(attrs.getValueByQName("ns:attr"), "val")
1013 self.assertEqual(attrs.getNameByQName("ns:attr"), (ns_uri, "attr"))
1014 self.assertEqual(attrs[(ns_uri, "attr")], "val")
1015 self.assertEqual(attrs.getQNameByName((ns_uri, "attr")), "ns:attr")
Fred Drake004d5e62000-10-23 17:22:08 +00001016
Lars Gustäbelab647872000-09-24 18:40:52 +00001017
Christian Heimesbbe741d2008-03-28 10:53:29 +00001018def test_main():
Guido van Rossumd8faa362007-04-27 19:54:29 +00001019 run_unittest(MakeParserTest,
1020 SaxutilsTest,
Serhiy Storchaka88efc522013-02-10 14:29:52 +02001021 StringXmlgenTest,
1022 BytesXmlgenTest,
1023 WriterXmlgenTest,
Georg Brandlc502df42013-05-12 11:41:12 +02001024 StreamWriterXmlgenTest,
1025 StreamReaderWriterXmlgenTest,
Guido van Rossumd8faa362007-04-27 19:54:29 +00001026 ExpatReaderTest,
1027 ErrorReportingTest,
1028 XmlReaderTest)
Lars Gustäbelb7536d52000-09-24 18:53:56 +00001029
Guido van Rossumd8faa362007-04-27 19:54:29 +00001030if __name__ == "__main__":
Christian Heimesbbe741d2008-03-28 10:53:29 +00001031 test_main()