blob: 2280058fac00716fa3adb226da2c86e073b7d910 [file] [log] [blame]
Antoine Pitroub27ddc72010-10-27 18:58:04 +00001# regression test for SAX 2.0 -*- coding: utf-8 -*-
Lars Gustäbel96753b32000-09-24 12:24:24 +00002# $Id$
3
Fred Drakefbdeaad2006-07-29 16:56:15 +00004from xml.sax import make_parser, ContentHandler, \
5 SAXException, SAXReaderNotAvailable, SAXParseException
Martin v. Löwis962c9e72000-10-06 17:41:52 +00006try:
7 make_parser()
Martin v. Löwis80670bc2000-10-06 21:13:23 +00008except SAXReaderNotAvailable:
Martin v. Löwis962c9e72000-10-06 17:41:52 +00009 # don't try to test this module if we cannot create a parser
10 raise ImportError("no XML parsers available")
Fred Drakefbdeaad2006-07-29 16:56:15 +000011from xml.sax.saxutils import XMLGenerator, escape, unescape, quoteattr, \
12 XMLFilterBase
13from xml.sax.expatreader import create_parser
Antoine Pitrou7f081022010-10-27 18:43:21 +000014from xml.sax.handler import feature_namespaces
Fred Drakefbdeaad2006-07-29 16:56:15 +000015from xml.sax.xmlreader import InputSource, AttributesImpl, AttributesNSImpl
Lars Gustäbel96753b32000-09-24 12:24:24 +000016from cStringIO import StringIO
Collin Winterd28fcbc2007-03-28 23:34:06 +000017from test.test_support import findfile, run_unittest
18import unittest
Lars Gustäbel96753b32000-09-24 12:24:24 +000019
Florent Xicluna1b51c3d2010-03-13 12:41:48 +000020TEST_XMLFILE = findfile("test.xml", subdir="xmltestdata")
21TEST_XMLFILE_OUT = findfile("test.xml.out", subdir="xmltestdata")
Florent Xicluna13ba1a12010-03-13 11:18:49 +000022
Collin Winterd28fcbc2007-03-28 23:34:06 +000023ns_uri = "http://www.python.org/xml-ns/saxtest/"
Lars Gustäbel96753b32000-09-24 12:24:24 +000024
Collin Winterd28fcbc2007-03-28 23:34:06 +000025class XmlTestBase(unittest.TestCase):
26 def verify_empty_attrs(self, attrs):
27 self.assertRaises(KeyError, attrs.getValue, "attr")
28 self.assertRaises(KeyError, attrs.getValueByQName, "attr")
29 self.assertRaises(KeyError, attrs.getNameByQName, "attr")
30 self.assertRaises(KeyError, attrs.getQNameByName, "attr")
31 self.assertRaises(KeyError, attrs.__getitem__, "attr")
32 self.assertEquals(attrs.getLength(), 0)
33 self.assertEquals(attrs.getNames(), [])
34 self.assertEquals(attrs.getQNames(), [])
35 self.assertEquals(len(attrs), 0)
36 self.assertFalse(attrs.has_key("attr"))
37 self.assertEquals(attrs.keys(), [])
38 self.assertEquals(attrs.get("attrs"), None)
39 self.assertEquals(attrs.get("attrs", 25), 25)
40 self.assertEquals(attrs.items(), [])
41 self.assertEquals(attrs.values(), [])
Neal Norwitz0d4c06e2007-04-25 06:30:05 +000042
Collin Winterd28fcbc2007-03-28 23:34:06 +000043 def verify_empty_nsattrs(self, attrs):
44 self.assertRaises(KeyError, attrs.getValue, (ns_uri, "attr"))
45 self.assertRaises(KeyError, attrs.getValueByQName, "ns:attr")
46 self.assertRaises(KeyError, attrs.getNameByQName, "ns:attr")
47 self.assertRaises(KeyError, attrs.getQNameByName, (ns_uri, "attr"))
48 self.assertRaises(KeyError, attrs.__getitem__, (ns_uri, "attr"))
49 self.assertEquals(attrs.getLength(), 0)
50 self.assertEquals(attrs.getNames(), [])
51 self.assertEquals(attrs.getQNames(), [])
52 self.assertEquals(len(attrs), 0)
53 self.assertFalse(attrs.has_key((ns_uri, "attr")))
54 self.assertEquals(attrs.keys(), [])
55 self.assertEquals(attrs.get((ns_uri, "attr")), None)
56 self.assertEquals(attrs.get((ns_uri, "attr"), 25), 25)
57 self.assertEquals(attrs.items(), [])
58 self.assertEquals(attrs.values(), [])
Lars Gustäbel96753b32000-09-24 12:24:24 +000059
Collin Winterd28fcbc2007-03-28 23:34:06 +000060 def verify_attrs_wattr(self, attrs):
61 self.assertEquals(attrs.getLength(), 1)
62 self.assertEquals(attrs.getNames(), ["attr"])
63 self.assertEquals(attrs.getQNames(), ["attr"])
64 self.assertEquals(len(attrs), 1)
65 self.assertTrue(attrs.has_key("attr"))
66 self.assertEquals(attrs.keys(), ["attr"])
67 self.assertEquals(attrs.get("attr"), "val")
68 self.assertEquals(attrs.get("attr", 25), "val")
69 self.assertEquals(attrs.items(), [("attr", "val")])
70 self.assertEquals(attrs.values(), ["val"])
71 self.assertEquals(attrs.getValue("attr"), "val")
72 self.assertEquals(attrs.getValueByQName("attr"), "val")
73 self.assertEquals(attrs.getNameByQName("attr"), "attr")
74 self.assertEquals(attrs["attr"], "val")
75 self.assertEquals(attrs.getQNameByName("attr"), "attr")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +000076
Collin Winterd28fcbc2007-03-28 23:34:06 +000077class MakeParserTest(unittest.TestCase):
78 def test_make_parser2(self):
Lars Gustäbel2fc52942000-10-24 15:35:07 +000079 # Creating parsers several times in a row should succeed.
80 # Testing this because there have been failures of this kind
81 # before.
Fred Drakefbdeaad2006-07-29 16:56:15 +000082 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000083 p = make_parser()
Fred Drakefbdeaad2006-07-29 16:56:15 +000084 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000085 p = make_parser()
Fred Drakefbdeaad2006-07-29 16:56:15 +000086 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000087 p = make_parser()
Fred Drakefbdeaad2006-07-29 16:56:15 +000088 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000089 p = make_parser()
Fred Drakefbdeaad2006-07-29 16:56:15 +000090 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000091 p = make_parser()
Fred Drakefbdeaad2006-07-29 16:56:15 +000092 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000093 p = make_parser()
Tim Petersd2bf3b72001-01-18 02:22:22 +000094
95
Lars Gustäbel96753b32000-09-24 12:24:24 +000096# ===========================================================================
97#
98# saxutils tests
99#
100# ===========================================================================
101
Collin Winterd28fcbc2007-03-28 23:34:06 +0000102class SaxutilsTest(unittest.TestCase):
103 # ===== escape
104 def test_escape_basic(self):
105 self.assertEquals(escape("Donald Duck & Co"), "Donald Duck & Co")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000106
Collin Winterd28fcbc2007-03-28 23:34:06 +0000107 def test_escape_all(self):
108 self.assertEquals(escape("<Donald Duck & Co>"),
109 "&lt;Donald Duck &amp; Co&gt;")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000110
Collin Winterd28fcbc2007-03-28 23:34:06 +0000111 def test_escape_extra(self):
Antoine Pitroub27ddc72010-10-27 18:58:04 +0000112 self.assertEquals(escape("Hei på deg", {"å" : "&aring;"}),
Collin Winterd28fcbc2007-03-28 23:34:06 +0000113 "Hei p&aring; deg")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000114
Collin Winterd28fcbc2007-03-28 23:34:06 +0000115 # ===== unescape
116 def test_unescape_basic(self):
117 self.assertEquals(unescape("Donald Duck &amp; Co"), "Donald Duck & Co")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000118
Collin Winterd28fcbc2007-03-28 23:34:06 +0000119 def test_unescape_all(self):
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000120 self.assertEquals(unescape("&lt;Donald Duck &amp; Co&gt;"),
Collin Winterd28fcbc2007-03-28 23:34:06 +0000121 "<Donald Duck & Co>")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000122
Collin Winterd28fcbc2007-03-28 23:34:06 +0000123 def test_unescape_extra(self):
Antoine Pitroub27ddc72010-10-27 18:58:04 +0000124 self.assertEquals(unescape("Hei på deg", {"å" : "&aring;"}),
Collin Winterd28fcbc2007-03-28 23:34:06 +0000125 "Hei p&aring; deg")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000126
Collin Winterd28fcbc2007-03-28 23:34:06 +0000127 def test_unescape_amp_extra(self):
128 self.assertEquals(unescape("&amp;foo;", {"&foo;": "splat"}), "&foo;")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000129
Collin Winterd28fcbc2007-03-28 23:34:06 +0000130 # ===== quoteattr
131 def test_quoteattr_basic(self):
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000132 self.assertEquals(quoteattr("Donald Duck & Co"),
Collin Winterd28fcbc2007-03-28 23:34:06 +0000133 '"Donald Duck &amp; Co"')
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000134
Collin Winterd28fcbc2007-03-28 23:34:06 +0000135 def test_single_quoteattr(self):
136 self.assertEquals(quoteattr('Includes "double" quotes'),
137 '\'Includes "double" quotes\'')
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000138
Collin Winterd28fcbc2007-03-28 23:34:06 +0000139 def test_double_quoteattr(self):
140 self.assertEquals(quoteattr("Includes 'single' quotes"),
141 "\"Includes 'single' quotes\"")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000142
Collin Winterd28fcbc2007-03-28 23:34:06 +0000143 def test_single_double_quoteattr(self):
144 self.assertEquals(quoteattr("Includes 'single' and \"double\" quotes"),
145 "\"Includes 'single' and &quot;double&quot; quotes\"")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000146
Collin Winterd28fcbc2007-03-28 23:34:06 +0000147 # ===== make_parser
148 def test_make_parser(self):
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000149 # Creating a parser should succeed - it should fall back
150 # to the expatreader
Fred Drakefbdeaad2006-07-29 16:56:15 +0000151 p = make_parser(['xml.parsers.no_such_parser'])
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000152
153
Lars Gustäbel96753b32000-09-24 12:24:24 +0000154# ===== XMLGenerator
155
156start = '<?xml version="1.0" encoding="iso-8859-1"?>\n'
157
Collin Winterd28fcbc2007-03-28 23:34:06 +0000158class XmlgenTest(unittest.TestCase):
159 def test_xmlgen_basic(self):
160 result = StringIO()
161 gen = XMLGenerator(result)
162 gen.startDocument()
163 gen.startElement("doc", {})
164 gen.endElement("doc")
165 gen.endDocument()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000166
Collin Winterd28fcbc2007-03-28 23:34:06 +0000167 self.assertEquals(result.getvalue(), start + "<doc></doc>")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000168
Collin Winterd28fcbc2007-03-28 23:34:06 +0000169 def test_xmlgen_content(self):
170 result = StringIO()
171 gen = XMLGenerator(result)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000172
Collin Winterd28fcbc2007-03-28 23:34:06 +0000173 gen.startDocument()
174 gen.startElement("doc", {})
175 gen.characters("huhei")
176 gen.endElement("doc")
177 gen.endDocument()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000178
Collin Winterd28fcbc2007-03-28 23:34:06 +0000179 self.assertEquals(result.getvalue(), start + "<doc>huhei</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000180
Collin Winterd28fcbc2007-03-28 23:34:06 +0000181 def test_xmlgen_pi(self):
182 result = StringIO()
183 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000184
Collin Winterd28fcbc2007-03-28 23:34:06 +0000185 gen.startDocument()
186 gen.processingInstruction("test", "data")
187 gen.startElement("doc", {})
188 gen.endElement("doc")
189 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000190
Collin Winterd28fcbc2007-03-28 23:34:06 +0000191 self.assertEquals(result.getvalue(), start + "<?test data?><doc></doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000192
Collin Winterd28fcbc2007-03-28 23:34:06 +0000193 def test_xmlgen_content_escape(self):
194 result = StringIO()
195 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000196
Collin Winterd28fcbc2007-03-28 23:34:06 +0000197 gen.startDocument()
198 gen.startElement("doc", {})
199 gen.characters("<huhei&")
200 gen.endElement("doc")
201 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000202
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000203 self.assertEquals(result.getvalue(),
Collin Winterd28fcbc2007-03-28 23:34:06 +0000204 start + "<doc>&lt;huhei&amp;</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000205
Collin Winterd28fcbc2007-03-28 23:34:06 +0000206 def test_xmlgen_attr_escape(self):
207 result = StringIO()
208 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000209
Collin Winterd28fcbc2007-03-28 23:34:06 +0000210 gen.startDocument()
211 gen.startElement("doc", {"a": '"'})
212 gen.startElement("e", {"a": "'"})
213 gen.endElement("e")
214 gen.startElement("e", {"a": "'\""})
215 gen.endElement("e")
216 gen.startElement("e", {"a": "\n\r\t"})
217 gen.endElement("e")
218 gen.endElement("doc")
219 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000220
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000221 self.assertEquals(result.getvalue(), start +
Collin Winterd28fcbc2007-03-28 23:34:06 +0000222 ("<doc a='\"'><e a=\"'\"></e>"
223 "<e a=\"'&quot;\"></e>"
224 "<e a=\"&#10;&#13;&#9;\"></e></doc>"))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000225
Collin Winterd28fcbc2007-03-28 23:34:06 +0000226 def test_xmlgen_ignorable(self):
227 result = StringIO()
228 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000229
Collin Winterd28fcbc2007-03-28 23:34:06 +0000230 gen.startDocument()
231 gen.startElement("doc", {})
232 gen.ignorableWhitespace(" ")
233 gen.endElement("doc")
234 gen.endDocument()
Fred Drakec9fadf92001-08-07 19:17:06 +0000235
Collin Winterd28fcbc2007-03-28 23:34:06 +0000236 self.assertEquals(result.getvalue(), start + "<doc> </doc>")
Fred Drakec9fadf92001-08-07 19:17:06 +0000237
Collin Winterd28fcbc2007-03-28 23:34:06 +0000238 def test_xmlgen_ns(self):
239 result = StringIO()
240 gen = XMLGenerator(result)
Fred Drakec9fadf92001-08-07 19:17:06 +0000241
Collin Winterd28fcbc2007-03-28 23:34:06 +0000242 gen.startDocument()
243 gen.startPrefixMapping("ns1", ns_uri)
244 gen.startElementNS((ns_uri, "doc"), "ns1:doc", {})
245 # add an unqualified name
246 gen.startElementNS((None, "udoc"), None, {})
247 gen.endElementNS((None, "udoc"), None)
248 gen.endElementNS((ns_uri, "doc"), "ns1:doc")
249 gen.endPrefixMapping("ns1")
250 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000251
Collin Winterd28fcbc2007-03-28 23:34:06 +0000252 self.assertEquals(result.getvalue(), start + \
Martin v. Löwiscf0a1cc2000-10-03 22:35:29 +0000253 ('<ns1:doc xmlns:ns1="%s"><udoc></udoc></ns1:doc>' %
Collin Winterd28fcbc2007-03-28 23:34:06 +0000254 ns_uri))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000255
Collin Winterd28fcbc2007-03-28 23:34:06 +0000256 def test_1463026_1(self):
257 result = StringIO()
258 gen = XMLGenerator(result)
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000259
Collin Winterd28fcbc2007-03-28 23:34:06 +0000260 gen.startDocument()
261 gen.startElementNS((None, 'a'), 'a', {(None, 'b'):'c'})
262 gen.endElementNS((None, 'a'), 'a')
263 gen.endDocument()
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000264
Collin Winterd28fcbc2007-03-28 23:34:06 +0000265 self.assertEquals(result.getvalue(), start+'<a b="c"></a>')
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000266
Collin Winterd28fcbc2007-03-28 23:34:06 +0000267 def test_1463026_2(self):
268 result = StringIO()
269 gen = XMLGenerator(result)
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000270
Collin Winterd28fcbc2007-03-28 23:34:06 +0000271 gen.startDocument()
272 gen.startPrefixMapping(None, 'qux')
273 gen.startElementNS(('qux', 'a'), 'a', {})
274 gen.endElementNS(('qux', 'a'), 'a')
275 gen.endPrefixMapping(None)
276 gen.endDocument()
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000277
Collin Winterd28fcbc2007-03-28 23:34:06 +0000278 self.assertEquals(result.getvalue(), start+'<a xmlns="qux"></a>')
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000279
Collin Winterd28fcbc2007-03-28 23:34:06 +0000280 def test_1463026_3(self):
281 result = StringIO()
282 gen = XMLGenerator(result)
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000283
Collin Winterd28fcbc2007-03-28 23:34:06 +0000284 gen.startDocument()
285 gen.startPrefixMapping('my', 'qux')
286 gen.startElementNS(('qux', 'a'), 'a', {(None, 'b'):'c'})
287 gen.endElementNS(('qux', 'a'), 'a')
288 gen.endPrefixMapping('my')
289 gen.endDocument()
Martin v. Löwis2bad58f2007-02-12 12:21:10 +0000290
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000291 self.assertEquals(result.getvalue(),
Collin Winterd28fcbc2007-03-28 23:34:06 +0000292 start+'<my:a xmlns:my="qux" b="c"></my:a>')
Tim Petersea5962f2007-03-12 18:07:52 +0000293
Antoine Pitrou7f081022010-10-27 18:43:21 +0000294 def test_5027_1(self):
295 # The xml prefix (as in xml:lang below) is reserved and bound by
296 # definition to http://www.w3.org/XML/1998/namespace. XMLGenerator had
297 # a bug whereby a KeyError is thrown because this namespace is missing
298 # from a dictionary.
299 #
300 # This test demonstrates the bug by parsing a document.
301 test_xml = StringIO(
302 '<?xml version="1.0"?>'
303 '<a:g1 xmlns:a="http://example.com/ns">'
304 '<a:g2 xml:lang="en">Hello</a:g2>'
305 '</a:g1>')
306
307 parser = make_parser()
308 parser.setFeature(feature_namespaces, True)
309 result = StringIO()
310 gen = XMLGenerator(result)
311 parser.setContentHandler(gen)
312 parser.parse(test_xml)
313
314 self.assertEquals(result.getvalue(),
315 start + (
316 '<a:g1 xmlns:a="http://example.com/ns">'
317 '<a:g2 xml:lang="en">Hello</a:g2>'
318 '</a:g1>'))
319
320 def test_5027_2(self):
321 # The xml prefix (as in xml:lang below) is reserved and bound by
322 # definition to http://www.w3.org/XML/1998/namespace. XMLGenerator had
323 # a bug whereby a KeyError is thrown because this namespace is missing
324 # from a dictionary.
325 #
326 # This test demonstrates the bug by direct manipulation of the
327 # XMLGenerator.
328 result = StringIO()
329 gen = XMLGenerator(result)
330
331 gen.startDocument()
332 gen.startPrefixMapping('a', 'http://example.com/ns')
333 gen.startElementNS(('http://example.com/ns', 'g1'), 'g1', {})
334 lang_attr = {('http://www.w3.org/XML/1998/namespace', 'lang'): 'en'}
335 gen.startElementNS(('http://example.com/ns', 'g2'), 'g2', lang_attr)
336 gen.characters('Hello')
337 gen.endElementNS(('http://example.com/ns', 'g2'), 'g2')
338 gen.endElementNS(('http://example.com/ns', 'g1'), 'g1')
339 gen.endPrefixMapping('a')
340 gen.endDocument()
341
342 self.assertEquals(result.getvalue(),
343 start + (
344 '<a:g1 xmlns:a="http://example.com/ns">'
345 '<a:g2 xml:lang="en">Hello</a:g2>'
346 '</a:g1>'))
347
Lars Gustäbel96753b32000-09-24 12:24:24 +0000348
Collin Winterd28fcbc2007-03-28 23:34:06 +0000349class XMLFilterBaseTest(unittest.TestCase):
350 def test_filter_basic(self):
351 result = StringIO()
352 gen = XMLGenerator(result)
353 filter = XMLFilterBase()
354 filter.setContentHandler(gen)
Fred Drake004d5e62000-10-23 17:22:08 +0000355
Collin Winterd28fcbc2007-03-28 23:34:06 +0000356 filter.startDocument()
357 filter.startElement("doc", {})
358 filter.characters("content")
359 filter.ignorableWhitespace(" ")
360 filter.endElement("doc")
361 filter.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000362
Collin Winterd28fcbc2007-03-28 23:34:06 +0000363 self.assertEquals(result.getvalue(), start + "<doc>content </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000364
365# ===========================================================================
366#
367# expatreader tests
368#
369# ===========================================================================
370
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000371xml_test_out = open(TEST_XMLFILE_OUT).read()
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000372
Collin Winterd28fcbc2007-03-28 23:34:06 +0000373class ExpatReaderTest(XmlTestBase):
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000374
Collin Winterd28fcbc2007-03-28 23:34:06 +0000375 # ===== XMLReader support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000376
Collin Winterd28fcbc2007-03-28 23:34:06 +0000377 def test_expat_file(self):
378 parser = create_parser()
379 result = StringIO()
380 xmlgen = XMLGenerator(result)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000381
Collin Winterd28fcbc2007-03-28 23:34:06 +0000382 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000383 parser.parse(open(TEST_XMLFILE))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000384
Collin Winterd28fcbc2007-03-28 23:34:06 +0000385 self.assertEquals(result.getvalue(), xml_test_out)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000386
Collin Winterd28fcbc2007-03-28 23:34:06 +0000387 # ===== DTDHandler support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000388
Collin Winterd28fcbc2007-03-28 23:34:06 +0000389 class TestDTDHandler:
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000390
Collin Winterd28fcbc2007-03-28 23:34:06 +0000391 def __init__(self):
392 self._notations = []
393 self._entities = []
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000394
Collin Winterd28fcbc2007-03-28 23:34:06 +0000395 def notationDecl(self, name, publicId, systemId):
396 self._notations.append((name, publicId, systemId))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000397
Collin Winterd28fcbc2007-03-28 23:34:06 +0000398 def unparsedEntityDecl(self, name, publicId, systemId, ndata):
399 self._entities.append((name, publicId, systemId, ndata))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000400
Collin Winterd28fcbc2007-03-28 23:34:06 +0000401 def test_expat_dtdhandler(self):
402 parser = create_parser()
403 handler = self.TestDTDHandler()
404 parser.setDTDHandler(handler)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000405
Collin Winterd28fcbc2007-03-28 23:34:06 +0000406 parser.feed('<!DOCTYPE doc [\n')
407 parser.feed(' <!ENTITY img SYSTEM "expat.gif" NDATA GIF>\n')
408 parser.feed(' <!NOTATION GIF PUBLIC "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN">\n')
409 parser.feed(']>\n')
410 parser.feed('<doc></doc>')
411 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000412
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000413 self.assertEquals(handler._notations,
Collin Winterd28fcbc2007-03-28 23:34:06 +0000414 [("GIF", "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN", None)])
415 self.assertEquals(handler._entities, [("img", None, "expat.gif", "GIF")])
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000416
Collin Winterd28fcbc2007-03-28 23:34:06 +0000417 # ===== EntityResolver support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000418
Collin Winterd28fcbc2007-03-28 23:34:06 +0000419 class TestEntityResolver:
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000420
Collin Winterd28fcbc2007-03-28 23:34:06 +0000421 def resolveEntity(self, publicId, systemId):
422 inpsrc = InputSource()
423 inpsrc.setByteStream(StringIO("<entity/>"))
424 return inpsrc
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000425
Collin Winterd28fcbc2007-03-28 23:34:06 +0000426 def test_expat_entityresolver(self):
427 parser = create_parser()
428 parser.setEntityResolver(self.TestEntityResolver())
429 result = StringIO()
430 parser.setContentHandler(XMLGenerator(result))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000431
Collin Winterd28fcbc2007-03-28 23:34:06 +0000432 parser.feed('<!DOCTYPE doc [\n')
433 parser.feed(' <!ENTITY test SYSTEM "whatever">\n')
434 parser.feed(']>\n')
435 parser.feed('<doc>&test;</doc>')
436 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000437
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000438 self.assertEquals(result.getvalue(), start +
Collin Winterd28fcbc2007-03-28 23:34:06 +0000439 "<doc><entity></entity></doc>")
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000440
Collin Winterd28fcbc2007-03-28 23:34:06 +0000441 # ===== Attributes support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000442
Collin Winterd28fcbc2007-03-28 23:34:06 +0000443 class AttrGatherer(ContentHandler):
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000444
Collin Winterd28fcbc2007-03-28 23:34:06 +0000445 def startElement(self, name, attrs):
446 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000447
Collin Winterd28fcbc2007-03-28 23:34:06 +0000448 def startElementNS(self, name, qname, attrs):
449 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000450
Collin Winterd28fcbc2007-03-28 23:34:06 +0000451 def test_expat_attrs_empty(self):
452 parser = create_parser()
453 gather = self.AttrGatherer()
454 parser.setContentHandler(gather)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000455
Collin Winterd28fcbc2007-03-28 23:34:06 +0000456 parser.feed("<doc/>")
457 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000458
Collin Winterd28fcbc2007-03-28 23:34:06 +0000459 self.verify_empty_attrs(gather._attrs)
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000460
Collin Winterd28fcbc2007-03-28 23:34:06 +0000461 def test_expat_attrs_wattr(self):
462 parser = create_parser()
463 gather = self.AttrGatherer()
464 parser.setContentHandler(gather)
465
466 parser.feed("<doc attr='val'/>")
467 parser.close()
468
469 self.verify_attrs_wattr(gather._attrs)
470
471 def test_expat_nsattrs_empty(self):
472 parser = create_parser(1)
473 gather = self.AttrGatherer()
474 parser.setContentHandler(gather)
475
476 parser.feed("<doc/>")
477 parser.close()
478
479 self.verify_empty_nsattrs(gather._attrs)
480
481 def test_expat_nsattrs_wattr(self):
482 parser = create_parser(1)
483 gather = self.AttrGatherer()
484 parser.setContentHandler(gather)
485
486 parser.feed("<doc xmlns:ns='%s' ns:attr='val'/>" % ns_uri)
487 parser.close()
488
489 attrs = gather._attrs
490
491 self.assertEquals(attrs.getLength(), 1)
492 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000493 self.assertTrue((attrs.getQNames() == [] or
Collin Winterd28fcbc2007-03-28 23:34:06 +0000494 attrs.getQNames() == ["ns:attr"]))
495 self.assertEquals(len(attrs), 1)
496 self.assertTrue(attrs.has_key((ns_uri, "attr")))
497 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
498 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
499 self.assertEquals(attrs.items(), [((ns_uri, "attr"), "val")])
500 self.assertEquals(attrs.values(), ["val"])
501 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
502 self.assertEquals(attrs[(ns_uri, "attr")], "val")
503
504 # ===== InputSource support
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000505
Collin Winterd28fcbc2007-03-28 23:34:06 +0000506 def test_expat_inpsource_filename(self):
507 parser = create_parser()
508 result = StringIO()
509 xmlgen = XMLGenerator(result)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000510
Collin Winterd28fcbc2007-03-28 23:34:06 +0000511 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000512 parser.parse(TEST_XMLFILE)
Collin Winterd28fcbc2007-03-28 23:34:06 +0000513
514 self.assertEquals(result.getvalue(), xml_test_out)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000515
Collin Winterd28fcbc2007-03-28 23:34:06 +0000516 def test_expat_inpsource_sysid(self):
517 parser = create_parser()
518 result = StringIO()
519 xmlgen = XMLGenerator(result)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000520
Collin Winterd28fcbc2007-03-28 23:34:06 +0000521 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000522 parser.parse(InputSource(TEST_XMLFILE))
Collin Winterd28fcbc2007-03-28 23:34:06 +0000523
524 self.assertEquals(result.getvalue(), xml_test_out)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000525
Collin Winterd28fcbc2007-03-28 23:34:06 +0000526 def test_expat_inpsource_stream(self):
527 parser = create_parser()
528 result = StringIO()
529 xmlgen = XMLGenerator(result)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000530
Collin Winterd28fcbc2007-03-28 23:34:06 +0000531 parser.setContentHandler(xmlgen)
532 inpsrc = InputSource()
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000533 inpsrc.setByteStream(open(TEST_XMLFILE))
Collin Winterd28fcbc2007-03-28 23:34:06 +0000534 parser.parse(inpsrc)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000535
Collin Winterd28fcbc2007-03-28 23:34:06 +0000536 self.assertEquals(result.getvalue(), xml_test_out)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000537
Collin Winterd28fcbc2007-03-28 23:34:06 +0000538 # ===== IncrementalParser support
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000539
Collin Winterd28fcbc2007-03-28 23:34:06 +0000540 def test_expat_incremental(self):
541 result = StringIO()
542 xmlgen = XMLGenerator(result)
543 parser = create_parser()
544 parser.setContentHandler(xmlgen)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000545
Collin Winterd28fcbc2007-03-28 23:34:06 +0000546 parser.feed("<doc>")
547 parser.feed("</doc>")
548 parser.close()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000549
Collin Winterd28fcbc2007-03-28 23:34:06 +0000550 self.assertEquals(result.getvalue(), start + "<doc></doc>")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000551
Collin Winterd28fcbc2007-03-28 23:34:06 +0000552 def test_expat_incremental_reset(self):
553 result = StringIO()
554 xmlgen = XMLGenerator(result)
555 parser = create_parser()
556 parser.setContentHandler(xmlgen)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000557
Collin Winterd28fcbc2007-03-28 23:34:06 +0000558 parser.feed("<doc>")
559 parser.feed("text")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000560
Collin Winterd28fcbc2007-03-28 23:34:06 +0000561 result = StringIO()
562 xmlgen = XMLGenerator(result)
563 parser.setContentHandler(xmlgen)
564 parser.reset()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000565
Collin Winterd28fcbc2007-03-28 23:34:06 +0000566 parser.feed("<doc>")
567 parser.feed("text")
568 parser.feed("</doc>")
569 parser.close()
570
571 self.assertEquals(result.getvalue(), start + "<doc>text</doc>")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000572
Collin Winterd28fcbc2007-03-28 23:34:06 +0000573 # ===== Locator support
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000574
Collin Winterd28fcbc2007-03-28 23:34:06 +0000575 def test_expat_locator_noinfo(self):
576 result = StringIO()
577 xmlgen = XMLGenerator(result)
578 parser = create_parser()
579 parser.setContentHandler(xmlgen)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000580
Collin Winterd28fcbc2007-03-28 23:34:06 +0000581 parser.feed("<doc>")
582 parser.feed("</doc>")
583 parser.close()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000584
Collin Winterd28fcbc2007-03-28 23:34:06 +0000585 self.assertEquals(parser.getSystemId(), None)
586 self.assertEquals(parser.getPublicId(), None)
587 self.assertEquals(parser.getLineNumber(), 1)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000588
Collin Winterd28fcbc2007-03-28 23:34:06 +0000589 def test_expat_locator_withinfo(self):
590 result = StringIO()
591 xmlgen = XMLGenerator(result)
592 parser = create_parser()
593 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000594 parser.parse(TEST_XMLFILE)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000595
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000596 self.assertEquals(parser.getSystemId(), TEST_XMLFILE)
Collin Winterd28fcbc2007-03-28 23:34:06 +0000597 self.assertEquals(parser.getPublicId(), None)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000598
599
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000600# ===========================================================================
601#
602# error reporting
603#
604# ===========================================================================
605
Collin Winterd28fcbc2007-03-28 23:34:06 +0000606class ErrorReportingTest(unittest.TestCase):
607 def test_expat_inpsource_location(self):
608 parser = create_parser()
609 parser.setContentHandler(ContentHandler()) # do nothing
610 source = InputSource()
611 source.setByteStream(StringIO("<foo bar foobar>")) #ill-formed
612 name = "a file name"
613 source.setSystemId(name)
614 try:
615 parser.parse(source)
616 self.fail()
617 except SAXException, e:
618 self.assertEquals(e.getSystemId(), name)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000619
Collin Winterd28fcbc2007-03-28 23:34:06 +0000620 def test_expat_incomplete(self):
621 parser = create_parser()
622 parser.setContentHandler(ContentHandler()) # do nothing
623 self.assertRaises(SAXParseException, parser.parse, StringIO("<foo>"))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000624
Collin Winterd28fcbc2007-03-28 23:34:06 +0000625 def test_sax_parse_exception_str(self):
626 # pass various values from a locator to the SAXParseException to
627 # make sure that the __str__() doesn't fall apart when None is
628 # passed instead of an integer line and column number
629 #
630 # use "normal" values for the locator:
631 str(SAXParseException("message", None,
632 self.DummyLocator(1, 1)))
633 # use None for the line number:
634 str(SAXParseException("message", None,
635 self.DummyLocator(None, 1)))
636 # use None for the column number:
637 str(SAXParseException("message", None,
638 self.DummyLocator(1, None)))
639 # use None for both:
640 str(SAXParseException("message", None,
641 self.DummyLocator(None, None)))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000642
Collin Winterd28fcbc2007-03-28 23:34:06 +0000643 class DummyLocator:
644 def __init__(self, lineno, colno):
645 self._lineno = lineno
646 self._colno = colno
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000647
Collin Winterd28fcbc2007-03-28 23:34:06 +0000648 def getPublicId(self):
649 return "pubid"
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000650
Collin Winterd28fcbc2007-03-28 23:34:06 +0000651 def getSystemId(self):
652 return "sysid"
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000653
Collin Winterd28fcbc2007-03-28 23:34:06 +0000654 def getLineNumber(self):
655 return self._lineno
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000656
Collin Winterd28fcbc2007-03-28 23:34:06 +0000657 def getColumnNumber(self):
658 return self._colno
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000659
Lars Gustäbelab647872000-09-24 18:40:52 +0000660# ===========================================================================
661#
662# xmlreader tests
663#
664# ===========================================================================
665
Collin Winterd28fcbc2007-03-28 23:34:06 +0000666class XmlReaderTest(XmlTestBase):
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000667
Collin Winterd28fcbc2007-03-28 23:34:06 +0000668 # ===== AttributesImpl
669 def test_attrs_empty(self):
670 self.verify_empty_attrs(AttributesImpl({}))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000671
Collin Winterd28fcbc2007-03-28 23:34:06 +0000672 def test_attrs_wattr(self):
673 self.verify_attrs_wattr(AttributesImpl({"attr" : "val"}))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000674
Collin Winterd28fcbc2007-03-28 23:34:06 +0000675 def test_nsattrs_empty(self):
676 self.verify_empty_nsattrs(AttributesNSImpl({}, {}))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000677
Collin Winterd28fcbc2007-03-28 23:34:06 +0000678 def test_nsattrs_wattr(self):
679 attrs = AttributesNSImpl({(ns_uri, "attr") : "val"},
680 {(ns_uri, "attr") : "ns:attr"})
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000681
Collin Winterd28fcbc2007-03-28 23:34:06 +0000682 self.assertEquals(attrs.getLength(), 1)
683 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
684 self.assertEquals(attrs.getQNames(), ["ns:attr"])
685 self.assertEquals(len(attrs), 1)
686 self.assertTrue(attrs.has_key((ns_uri, "attr")))
687 self.assertEquals(attrs.keys(), [(ns_uri, "attr")])
688 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
689 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
690 self.assertEquals(attrs.items(), [((ns_uri, "attr"), "val")])
691 self.assertEquals(attrs.values(), ["val"])
692 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
693 self.assertEquals(attrs.getValueByQName("ns:attr"), "val")
694 self.assertEquals(attrs.getNameByQName("ns:attr"), (ns_uri, "attr"))
695 self.assertEquals(attrs[(ns_uri, "attr")], "val")
696 self.assertEquals(attrs.getQNameByName((ns_uri, "attr")), "ns:attr")
Fred Drake004d5e62000-10-23 17:22:08 +0000697
Lars Gustäbelab647872000-09-24 18:40:52 +0000698
Collin Winterd28fcbc2007-03-28 23:34:06 +0000699 # During the development of Python 2.5, an attempt to move the "xml"
700 # package implementation to a new package ("xmlcore") proved painful.
701 # The goal of this change was to allow applications to be able to
702 # obtain and rely on behavior in the standard library implementation
703 # of the XML support without needing to be concerned about the
704 # availability of the PyXML implementation.
705 #
706 # While the existing import hackery in Lib/xml/__init__.py can cause
707 # PyXML's _xmlpus package to supplant the "xml" package, that only
708 # works because either implementation uses the "xml" package name for
709 # imports.
710 #
711 # The move resulted in a number of problems related to the fact that
712 # the import machinery's "package context" is based on the name that's
713 # being imported rather than the __name__ of the actual package
714 # containment; it wasn't possible for the "xml" package to be replaced
715 # by a simple module that indirected imports to the "xmlcore" package.
716 #
717 # The following two tests exercised bugs that were introduced in that
718 # attempt. Keeping these tests around will help detect problems with
719 # other attempts to provide reliable access to the standard library's
720 # implementation of the XML support.
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000721
Collin Winterd28fcbc2007-03-28 23:34:06 +0000722 def test_sf_1511497(self):
723 # Bug report: http://www.python.org/sf/1511497
724 import sys
725 old_modules = sys.modules.copy()
726 for modname in sys.modules.keys():
727 if modname.startswith("xml."):
728 del sys.modules[modname]
729 try:
730 import xml.sax.expatreader
731 module = xml.sax.expatreader
732 self.assertEquals(module.__name__, "xml.sax.expatreader")
733 finally:
734 sys.modules.update(old_modules)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000735
Collin Winterd28fcbc2007-03-28 23:34:06 +0000736 def test_sf_1513611(self):
737 # Bug report: http://www.python.org/sf/1513611
738 sio = StringIO("invalid")
739 parser = make_parser()
740 from xml.sax import SAXParseException
741 self.assertRaises(SAXParseException, parser.parse, sio)
Fred Drakefbdeaad2006-07-29 16:56:15 +0000742
Fred Drakefbdeaad2006-07-29 16:56:15 +0000743
Neal Norwitzab364c42008-03-28 07:36:31 +0000744def test_main():
Collin Winterd28fcbc2007-03-28 23:34:06 +0000745 run_unittest(MakeParserTest,
746 SaxutilsTest,
747 XmlgenTest,
748 ExpatReaderTest,
749 ErrorReportingTest,
750 XmlReaderTest)
Fred Drakefbdeaad2006-07-29 16:56:15 +0000751
Collin Winterd28fcbc2007-03-28 23:34:06 +0000752if __name__ == "__main__":
Neal Norwitzab364c42008-03-28 07:36:31 +0000753 test_main()