blob: 22048e6f7a0a36cca2edd064760f51f9520cdbd7 [file] [log] [blame]
Martin v. Löwisa729daf2002-08-04 17:28:33 +00001# regression test for SAX 2.0 -*- coding: iso-8859-1 -*-
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
14from xml.sax.xmlreader import InputSource, AttributesImpl, AttributesNSImpl
Lars Gustäbel96753b32000-09-24 12:24:24 +000015from cStringIO import StringIO
Collin Winterd28fcbc2007-03-28 23:34:06 +000016from test.test_support import findfile, run_unittest
17import unittest
Guido van Rossume2ae77b2001-10-24 20:42:55 +000018import os
Lars Gustäbel96753b32000-09-24 12:24:24 +000019
Florent Xicluna13ba1a12010-03-13 11:18:49 +000020TEST_XMLFILE = findfile(os.path.join("xmltestdata", "test.xml"))
21TEST_XMLFILE_OUT = findfile(os.path.join("xmltestdata", "test.xml.out"))
22
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):
112 self.assertEquals(escape("Hei på deg", {"å" : "&aring;"}),
113 "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):
Neal Norwitz0d4c06e2007-04-25 06:30:05 +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
Lars Gustäbel96753b32000-09-24 12:24:24 +0000294
Collin Winterd28fcbc2007-03-28 23:34:06 +0000295class XMLFilterBaseTest(unittest.TestCase):
296 def test_filter_basic(self):
297 result = StringIO()
298 gen = XMLGenerator(result)
299 filter = XMLFilterBase()
300 filter.setContentHandler(gen)
Fred Drake004d5e62000-10-23 17:22:08 +0000301
Collin Winterd28fcbc2007-03-28 23:34:06 +0000302 filter.startDocument()
303 filter.startElement("doc", {})
304 filter.characters("content")
305 filter.ignorableWhitespace(" ")
306 filter.endElement("doc")
307 filter.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000308
Collin Winterd28fcbc2007-03-28 23:34:06 +0000309 self.assertEquals(result.getvalue(), start + "<doc>content </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000310
311# ===========================================================================
312#
313# expatreader tests
314#
315# ===========================================================================
316
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000317xml_test_out = open(TEST_XMLFILE_OUT).read()
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000318
Collin Winterd28fcbc2007-03-28 23:34:06 +0000319class ExpatReaderTest(XmlTestBase):
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000320
Collin Winterd28fcbc2007-03-28 23:34:06 +0000321 # ===== XMLReader support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000322
Collin Winterd28fcbc2007-03-28 23:34:06 +0000323 def test_expat_file(self):
324 parser = create_parser()
325 result = StringIO()
326 xmlgen = XMLGenerator(result)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000327
Collin Winterd28fcbc2007-03-28 23:34:06 +0000328 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000329 parser.parse(open(TEST_XMLFILE))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000330
Collin Winterd28fcbc2007-03-28 23:34:06 +0000331 self.assertEquals(result.getvalue(), xml_test_out)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000332
Collin Winterd28fcbc2007-03-28 23:34:06 +0000333 # ===== DTDHandler support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000334
Collin Winterd28fcbc2007-03-28 23:34:06 +0000335 class TestDTDHandler:
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000336
Collin Winterd28fcbc2007-03-28 23:34:06 +0000337 def __init__(self):
338 self._notations = []
339 self._entities = []
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000340
Collin Winterd28fcbc2007-03-28 23:34:06 +0000341 def notationDecl(self, name, publicId, systemId):
342 self._notations.append((name, publicId, systemId))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000343
Collin Winterd28fcbc2007-03-28 23:34:06 +0000344 def unparsedEntityDecl(self, name, publicId, systemId, ndata):
345 self._entities.append((name, publicId, systemId, ndata))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000346
Collin Winterd28fcbc2007-03-28 23:34:06 +0000347 def test_expat_dtdhandler(self):
348 parser = create_parser()
349 handler = self.TestDTDHandler()
350 parser.setDTDHandler(handler)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000351
Collin Winterd28fcbc2007-03-28 23:34:06 +0000352 parser.feed('<!DOCTYPE doc [\n')
353 parser.feed(' <!ENTITY img SYSTEM "expat.gif" NDATA GIF>\n')
354 parser.feed(' <!NOTATION GIF PUBLIC "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN">\n')
355 parser.feed(']>\n')
356 parser.feed('<doc></doc>')
357 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000358
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000359 self.assertEquals(handler._notations,
Collin Winterd28fcbc2007-03-28 23:34:06 +0000360 [("GIF", "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN", None)])
361 self.assertEquals(handler._entities, [("img", None, "expat.gif", "GIF")])
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000362
Collin Winterd28fcbc2007-03-28 23:34:06 +0000363 # ===== EntityResolver support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000364
Collin Winterd28fcbc2007-03-28 23:34:06 +0000365 class TestEntityResolver:
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000366
Collin Winterd28fcbc2007-03-28 23:34:06 +0000367 def resolveEntity(self, publicId, systemId):
368 inpsrc = InputSource()
369 inpsrc.setByteStream(StringIO("<entity/>"))
370 return inpsrc
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000371
Collin Winterd28fcbc2007-03-28 23:34:06 +0000372 def test_expat_entityresolver(self):
373 parser = create_parser()
374 parser.setEntityResolver(self.TestEntityResolver())
375 result = StringIO()
376 parser.setContentHandler(XMLGenerator(result))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000377
Collin Winterd28fcbc2007-03-28 23:34:06 +0000378 parser.feed('<!DOCTYPE doc [\n')
379 parser.feed(' <!ENTITY test SYSTEM "whatever">\n')
380 parser.feed(']>\n')
381 parser.feed('<doc>&test;</doc>')
382 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000383
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000384 self.assertEquals(result.getvalue(), start +
Collin Winterd28fcbc2007-03-28 23:34:06 +0000385 "<doc><entity></entity></doc>")
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000386
Collin Winterd28fcbc2007-03-28 23:34:06 +0000387 # ===== Attributes support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000388
Collin Winterd28fcbc2007-03-28 23:34:06 +0000389 class AttrGatherer(ContentHandler):
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000390
Collin Winterd28fcbc2007-03-28 23:34:06 +0000391 def startElement(self, name, attrs):
392 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000393
Collin Winterd28fcbc2007-03-28 23:34:06 +0000394 def startElementNS(self, name, qname, attrs):
395 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000396
Collin Winterd28fcbc2007-03-28 23:34:06 +0000397 def test_expat_attrs_empty(self):
398 parser = create_parser()
399 gather = self.AttrGatherer()
400 parser.setContentHandler(gather)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000401
Collin Winterd28fcbc2007-03-28 23:34:06 +0000402 parser.feed("<doc/>")
403 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000404
Collin Winterd28fcbc2007-03-28 23:34:06 +0000405 self.verify_empty_attrs(gather._attrs)
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000406
Collin Winterd28fcbc2007-03-28 23:34:06 +0000407 def test_expat_attrs_wattr(self):
408 parser = create_parser()
409 gather = self.AttrGatherer()
410 parser.setContentHandler(gather)
411
412 parser.feed("<doc attr='val'/>")
413 parser.close()
414
415 self.verify_attrs_wattr(gather._attrs)
416
417 def test_expat_nsattrs_empty(self):
418 parser = create_parser(1)
419 gather = self.AttrGatherer()
420 parser.setContentHandler(gather)
421
422 parser.feed("<doc/>")
423 parser.close()
424
425 self.verify_empty_nsattrs(gather._attrs)
426
427 def test_expat_nsattrs_wattr(self):
428 parser = create_parser(1)
429 gather = self.AttrGatherer()
430 parser.setContentHandler(gather)
431
432 parser.feed("<doc xmlns:ns='%s' ns:attr='val'/>" % ns_uri)
433 parser.close()
434
435 attrs = gather._attrs
436
437 self.assertEquals(attrs.getLength(), 1)
438 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000439 self.assertTrue((attrs.getQNames() == [] or
Collin Winterd28fcbc2007-03-28 23:34:06 +0000440 attrs.getQNames() == ["ns:attr"]))
441 self.assertEquals(len(attrs), 1)
442 self.assertTrue(attrs.has_key((ns_uri, "attr")))
443 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
444 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
445 self.assertEquals(attrs.items(), [((ns_uri, "attr"), "val")])
446 self.assertEquals(attrs.values(), ["val"])
447 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
448 self.assertEquals(attrs[(ns_uri, "attr")], "val")
449
450 # ===== InputSource support
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000451
Collin Winterd28fcbc2007-03-28 23:34:06 +0000452 def test_expat_inpsource_filename(self):
453 parser = create_parser()
454 result = StringIO()
455 xmlgen = XMLGenerator(result)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000456
Collin Winterd28fcbc2007-03-28 23:34:06 +0000457 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000458 parser.parse(TEST_XMLFILE)
Collin Winterd28fcbc2007-03-28 23:34:06 +0000459
460 self.assertEquals(result.getvalue(), xml_test_out)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000461
Collin Winterd28fcbc2007-03-28 23:34:06 +0000462 def test_expat_inpsource_sysid(self):
463 parser = create_parser()
464 result = StringIO()
465 xmlgen = XMLGenerator(result)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000466
Collin Winterd28fcbc2007-03-28 23:34:06 +0000467 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000468 parser.parse(InputSource(TEST_XMLFILE))
Collin Winterd28fcbc2007-03-28 23:34:06 +0000469
470 self.assertEquals(result.getvalue(), xml_test_out)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000471
Collin Winterd28fcbc2007-03-28 23:34:06 +0000472 def test_expat_inpsource_stream(self):
473 parser = create_parser()
474 result = StringIO()
475 xmlgen = XMLGenerator(result)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000476
Collin Winterd28fcbc2007-03-28 23:34:06 +0000477 parser.setContentHandler(xmlgen)
478 inpsrc = InputSource()
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000479 inpsrc.setByteStream(open(TEST_XMLFILE))
Collin Winterd28fcbc2007-03-28 23:34:06 +0000480 parser.parse(inpsrc)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000481
Collin Winterd28fcbc2007-03-28 23:34:06 +0000482 self.assertEquals(result.getvalue(), xml_test_out)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000483
Collin Winterd28fcbc2007-03-28 23:34:06 +0000484 # ===== IncrementalParser support
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000485
Collin Winterd28fcbc2007-03-28 23:34:06 +0000486 def test_expat_incremental(self):
487 result = StringIO()
488 xmlgen = XMLGenerator(result)
489 parser = create_parser()
490 parser.setContentHandler(xmlgen)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000491
Collin Winterd28fcbc2007-03-28 23:34:06 +0000492 parser.feed("<doc>")
493 parser.feed("</doc>")
494 parser.close()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000495
Collin Winterd28fcbc2007-03-28 23:34:06 +0000496 self.assertEquals(result.getvalue(), start + "<doc></doc>")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000497
Collin Winterd28fcbc2007-03-28 23:34:06 +0000498 def test_expat_incremental_reset(self):
499 result = StringIO()
500 xmlgen = XMLGenerator(result)
501 parser = create_parser()
502 parser.setContentHandler(xmlgen)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000503
Collin Winterd28fcbc2007-03-28 23:34:06 +0000504 parser.feed("<doc>")
505 parser.feed("text")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000506
Collin Winterd28fcbc2007-03-28 23:34:06 +0000507 result = StringIO()
508 xmlgen = XMLGenerator(result)
509 parser.setContentHandler(xmlgen)
510 parser.reset()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000511
Collin Winterd28fcbc2007-03-28 23:34:06 +0000512 parser.feed("<doc>")
513 parser.feed("text")
514 parser.feed("</doc>")
515 parser.close()
516
517 self.assertEquals(result.getvalue(), start + "<doc>text</doc>")
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000518
Collin Winterd28fcbc2007-03-28 23:34:06 +0000519 # ===== Locator support
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000520
Collin Winterd28fcbc2007-03-28 23:34:06 +0000521 def test_expat_locator_noinfo(self):
522 result = StringIO()
523 xmlgen = XMLGenerator(result)
524 parser = create_parser()
525 parser.setContentHandler(xmlgen)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000526
Collin Winterd28fcbc2007-03-28 23:34:06 +0000527 parser.feed("<doc>")
528 parser.feed("</doc>")
529 parser.close()
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000530
Collin Winterd28fcbc2007-03-28 23:34:06 +0000531 self.assertEquals(parser.getSystemId(), None)
532 self.assertEquals(parser.getPublicId(), None)
533 self.assertEquals(parser.getLineNumber(), 1)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000534
Collin Winterd28fcbc2007-03-28 23:34:06 +0000535 def test_expat_locator_withinfo(self):
536 result = StringIO()
537 xmlgen = XMLGenerator(result)
538 parser = create_parser()
539 parser.setContentHandler(xmlgen)
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000540 parser.parse(TEST_XMLFILE)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000541
Florent Xicluna13ba1a12010-03-13 11:18:49 +0000542 self.assertEquals(parser.getSystemId(), TEST_XMLFILE)
Collin Winterd28fcbc2007-03-28 23:34:06 +0000543 self.assertEquals(parser.getPublicId(), None)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000544
545
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000546# ===========================================================================
547#
548# error reporting
549#
550# ===========================================================================
551
Collin Winterd28fcbc2007-03-28 23:34:06 +0000552class ErrorReportingTest(unittest.TestCase):
553 def test_expat_inpsource_location(self):
554 parser = create_parser()
555 parser.setContentHandler(ContentHandler()) # do nothing
556 source = InputSource()
557 source.setByteStream(StringIO("<foo bar foobar>")) #ill-formed
558 name = "a file name"
559 source.setSystemId(name)
560 try:
561 parser.parse(source)
562 self.fail()
563 except SAXException, e:
564 self.assertEquals(e.getSystemId(), name)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000565
Collin Winterd28fcbc2007-03-28 23:34:06 +0000566 def test_expat_incomplete(self):
567 parser = create_parser()
568 parser.setContentHandler(ContentHandler()) # do nothing
569 self.assertRaises(SAXParseException, parser.parse, StringIO("<foo>"))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000570
Collin Winterd28fcbc2007-03-28 23:34:06 +0000571 def test_sax_parse_exception_str(self):
572 # pass various values from a locator to the SAXParseException to
573 # make sure that the __str__() doesn't fall apart when None is
574 # passed instead of an integer line and column number
575 #
576 # use "normal" values for the locator:
577 str(SAXParseException("message", None,
578 self.DummyLocator(1, 1)))
579 # use None for the line number:
580 str(SAXParseException("message", None,
581 self.DummyLocator(None, 1)))
582 # use None for the column number:
583 str(SAXParseException("message", None,
584 self.DummyLocator(1, None)))
585 # use None for both:
586 str(SAXParseException("message", None,
587 self.DummyLocator(None, None)))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000588
Collin Winterd28fcbc2007-03-28 23:34:06 +0000589 class DummyLocator:
590 def __init__(self, lineno, colno):
591 self._lineno = lineno
592 self._colno = colno
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000593
Collin Winterd28fcbc2007-03-28 23:34:06 +0000594 def getPublicId(self):
595 return "pubid"
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000596
Collin Winterd28fcbc2007-03-28 23:34:06 +0000597 def getSystemId(self):
598 return "sysid"
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000599
Collin Winterd28fcbc2007-03-28 23:34:06 +0000600 def getLineNumber(self):
601 return self._lineno
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000602
Collin Winterd28fcbc2007-03-28 23:34:06 +0000603 def getColumnNumber(self):
604 return self._colno
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000605
Lars Gustäbelab647872000-09-24 18:40:52 +0000606# ===========================================================================
607#
608# xmlreader tests
609#
610# ===========================================================================
611
Collin Winterd28fcbc2007-03-28 23:34:06 +0000612class XmlReaderTest(XmlTestBase):
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000613
Collin Winterd28fcbc2007-03-28 23:34:06 +0000614 # ===== AttributesImpl
615 def test_attrs_empty(self):
616 self.verify_empty_attrs(AttributesImpl({}))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000617
Collin Winterd28fcbc2007-03-28 23:34:06 +0000618 def test_attrs_wattr(self):
619 self.verify_attrs_wattr(AttributesImpl({"attr" : "val"}))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000620
Collin Winterd28fcbc2007-03-28 23:34:06 +0000621 def test_nsattrs_empty(self):
622 self.verify_empty_nsattrs(AttributesNSImpl({}, {}))
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000623
Collin Winterd28fcbc2007-03-28 23:34:06 +0000624 def test_nsattrs_wattr(self):
625 attrs = AttributesNSImpl({(ns_uri, "attr") : "val"},
626 {(ns_uri, "attr") : "ns:attr"})
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000627
Collin Winterd28fcbc2007-03-28 23:34:06 +0000628 self.assertEquals(attrs.getLength(), 1)
629 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
630 self.assertEquals(attrs.getQNames(), ["ns:attr"])
631 self.assertEquals(len(attrs), 1)
632 self.assertTrue(attrs.has_key((ns_uri, "attr")))
633 self.assertEquals(attrs.keys(), [(ns_uri, "attr")])
634 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
635 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
636 self.assertEquals(attrs.items(), [((ns_uri, "attr"), "val")])
637 self.assertEquals(attrs.values(), ["val"])
638 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
639 self.assertEquals(attrs.getValueByQName("ns:attr"), "val")
640 self.assertEquals(attrs.getNameByQName("ns:attr"), (ns_uri, "attr"))
641 self.assertEquals(attrs[(ns_uri, "attr")], "val")
642 self.assertEquals(attrs.getQNameByName((ns_uri, "attr")), "ns:attr")
Fred Drake004d5e62000-10-23 17:22:08 +0000643
Lars Gustäbelab647872000-09-24 18:40:52 +0000644
Collin Winterd28fcbc2007-03-28 23:34:06 +0000645 # During the development of Python 2.5, an attempt to move the "xml"
646 # package implementation to a new package ("xmlcore") proved painful.
647 # The goal of this change was to allow applications to be able to
648 # obtain and rely on behavior in the standard library implementation
649 # of the XML support without needing to be concerned about the
650 # availability of the PyXML implementation.
651 #
652 # While the existing import hackery in Lib/xml/__init__.py can cause
653 # PyXML's _xmlpus package to supplant the "xml" package, that only
654 # works because either implementation uses the "xml" package name for
655 # imports.
656 #
657 # The move resulted in a number of problems related to the fact that
658 # the import machinery's "package context" is based on the name that's
659 # being imported rather than the __name__ of the actual package
660 # containment; it wasn't possible for the "xml" package to be replaced
661 # by a simple module that indirected imports to the "xmlcore" package.
662 #
663 # The following two tests exercised bugs that were introduced in that
664 # attempt. Keeping these tests around will help detect problems with
665 # other attempts to provide reliable access to the standard library's
666 # implementation of the XML support.
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000667
Collin Winterd28fcbc2007-03-28 23:34:06 +0000668 def test_sf_1511497(self):
669 # Bug report: http://www.python.org/sf/1511497
670 import sys
671 old_modules = sys.modules.copy()
672 for modname in sys.modules.keys():
673 if modname.startswith("xml."):
674 del sys.modules[modname]
675 try:
676 import xml.sax.expatreader
677 module = xml.sax.expatreader
678 self.assertEquals(module.__name__, "xml.sax.expatreader")
679 finally:
680 sys.modules.update(old_modules)
Neal Norwitz0d4c06e2007-04-25 06:30:05 +0000681
Collin Winterd28fcbc2007-03-28 23:34:06 +0000682 def test_sf_1513611(self):
683 # Bug report: http://www.python.org/sf/1513611
684 sio = StringIO("invalid")
685 parser = make_parser()
686 from xml.sax import SAXParseException
687 self.assertRaises(SAXParseException, parser.parse, sio)
Fred Drakefbdeaad2006-07-29 16:56:15 +0000688
Fred Drakefbdeaad2006-07-29 16:56:15 +0000689
Neal Norwitzab364c42008-03-28 07:36:31 +0000690def test_main():
Collin Winterd28fcbc2007-03-28 23:34:06 +0000691 run_unittest(MakeParserTest,
692 SaxutilsTest,
693 XmlgenTest,
694 ExpatReaderTest,
695 ErrorReportingTest,
696 XmlReaderTest)
Fred Drakefbdeaad2006-07-29 16:56:15 +0000697
Collin Winterd28fcbc2007-03-28 23:34:06 +0000698if __name__ == "__main__":
Neal Norwitzab364c42008-03-28 07:36:31 +0000699 test_main()