blob: 911e7634c24133e7c1ee4395e58ca99646c2ab66 [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
Thomas Wouters0e3f5912006-08-11 14:57:12 +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")
Thomas Wouters0e3f5912006-08-11 14:57:12 +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
Guido van Rossum34d19282007-08-09 01:03:29 +000015from io import StringIO
Benjamin Petersonee8712c2008-05-20 21:35:26 +000016from test.support import findfile, run_unittest
Guido van Rossumd8faa362007-04-27 19:54:29 +000017import unittest
Florent Xiclunaf15351d2010-03-13 23:24:31 +000018
19TEST_XMLFILE = findfile("test.xml", subdir="xmltestdata")
20TEST_XMLFILE_OUT = findfile("test.xml.out", subdir="xmltestdata")
Victor Stinner6c6f8512010-08-07 10:09:35 +000021try:
22 TEST_XMLFILE.encode("utf8")
23 TEST_XMLFILE_OUT.encode("utf8")
24except UnicodeEncodeError:
25 raise unittest.SkipTest("filename is not encodable to utf8")
Lars Gustäbel96753b32000-09-24 12:24:24 +000026
Guido van Rossumd8faa362007-04-27 19:54:29 +000027ns_uri = "http://www.python.org/xml-ns/saxtest/"
Lars Gustäbel96753b32000-09-24 12:24:24 +000028
Guido van Rossumd8faa362007-04-27 19:54:29 +000029class XmlTestBase(unittest.TestCase):
30 def verify_empty_attrs(self, attrs):
31 self.assertRaises(KeyError, attrs.getValue, "attr")
32 self.assertRaises(KeyError, attrs.getValueByQName, "attr")
33 self.assertRaises(KeyError, attrs.getNameByQName, "attr")
34 self.assertRaises(KeyError, attrs.getQNameByName, "attr")
35 self.assertRaises(KeyError, attrs.__getitem__, "attr")
36 self.assertEquals(attrs.getLength(), 0)
37 self.assertEquals(attrs.getNames(), [])
38 self.assertEquals(attrs.getQNames(), [])
39 self.assertEquals(len(attrs), 0)
Ezio Melottib58e0bd2010-01-23 15:40:09 +000040 self.assertNotIn("attr", attrs)
Guido van Rossumd8faa362007-04-27 19:54:29 +000041 self.assertEquals(list(attrs.keys()), [])
42 self.assertEquals(attrs.get("attrs"), None)
43 self.assertEquals(attrs.get("attrs", 25), 25)
44 self.assertEquals(list(attrs.items()), [])
45 self.assertEquals(list(attrs.values()), [])
Lars Gustäbel96753b32000-09-24 12:24:24 +000046
Guido van Rossumd8faa362007-04-27 19:54:29 +000047 def verify_empty_nsattrs(self, attrs):
48 self.assertRaises(KeyError, attrs.getValue, (ns_uri, "attr"))
49 self.assertRaises(KeyError, attrs.getValueByQName, "ns:attr")
50 self.assertRaises(KeyError, attrs.getNameByQName, "ns:attr")
51 self.assertRaises(KeyError, attrs.getQNameByName, (ns_uri, "attr"))
52 self.assertRaises(KeyError, attrs.__getitem__, (ns_uri, "attr"))
53 self.assertEquals(attrs.getLength(), 0)
54 self.assertEquals(attrs.getNames(), [])
55 self.assertEquals(attrs.getQNames(), [])
56 self.assertEquals(len(attrs), 0)
Ezio Melottib58e0bd2010-01-23 15:40:09 +000057 self.assertNotIn((ns_uri, "attr"), attrs)
Guido van Rossumd8faa362007-04-27 19:54:29 +000058 self.assertEquals(list(attrs.keys()), [])
59 self.assertEquals(attrs.get((ns_uri, "attr")), None)
60 self.assertEquals(attrs.get((ns_uri, "attr"), 25), 25)
61 self.assertEquals(list(attrs.items()), [])
62 self.assertEquals(list(attrs.values()), [])
Lars Gustäbel96753b32000-09-24 12:24:24 +000063
Guido van Rossumd8faa362007-04-27 19:54:29 +000064 def verify_attrs_wattr(self, attrs):
65 self.assertEquals(attrs.getLength(), 1)
66 self.assertEquals(attrs.getNames(), ["attr"])
67 self.assertEquals(attrs.getQNames(), ["attr"])
68 self.assertEquals(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +000069 self.assertIn("attr", attrs)
Guido van Rossumd8faa362007-04-27 19:54:29 +000070 self.assertEquals(list(attrs.keys()), ["attr"])
71 self.assertEquals(attrs.get("attr"), "val")
72 self.assertEquals(attrs.get("attr", 25), "val")
73 self.assertEquals(list(attrs.items()), [("attr", "val")])
74 self.assertEquals(list(attrs.values()), ["val"])
75 self.assertEquals(attrs.getValue("attr"), "val")
76 self.assertEquals(attrs.getValueByQName("attr"), "val")
77 self.assertEquals(attrs.getNameByQName("attr"), "attr")
78 self.assertEquals(attrs["attr"], "val")
79 self.assertEquals(attrs.getQNameByName("attr"), "attr")
Lars Gustäbel96753b32000-09-24 12:24:24 +000080
Guido van Rossumd8faa362007-04-27 19:54:29 +000081class MakeParserTest(unittest.TestCase):
82 def test_make_parser2(self):
Lars Gustäbel2fc52942000-10-24 15:35:07 +000083 # Creating parsers several times in a row should succeed.
84 # Testing this because there have been failures of this kind
85 # before.
Thomas Wouters0e3f5912006-08-11 14:57:12 +000086 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000087 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000088 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000089 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000090 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000091 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000092 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000093 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000094 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000095 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000096 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000097 p = make_parser()
Tim Petersd2bf3b72001-01-18 02:22:22 +000098
99
Lars Gustäbel96753b32000-09-24 12:24:24 +0000100# ===========================================================================
101#
102# saxutils tests
103#
104# ===========================================================================
105
Guido van Rossumd8faa362007-04-27 19:54:29 +0000106class SaxutilsTest(unittest.TestCase):
107 # ===== escape
108 def test_escape_basic(self):
109 self.assertEquals(escape("Donald Duck & Co"), "Donald Duck & Co")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000110
Guido van Rossumd8faa362007-04-27 19:54:29 +0000111 def test_escape_all(self):
112 self.assertEquals(escape("<Donald Duck & Co>"),
113 "&lt;Donald Duck &amp; Co&gt;")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000114
Guido van Rossumd8faa362007-04-27 19:54:29 +0000115 def test_escape_extra(self):
116 self.assertEquals(escape("Hei på deg", {"å" : "&aring;"}),
117 "Hei p&aring; deg")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000118
Guido van Rossumd8faa362007-04-27 19:54:29 +0000119 # ===== unescape
120 def test_unescape_basic(self):
121 self.assertEquals(unescape("Donald Duck &amp; Co"), "Donald Duck & Co")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000122
Guido van Rossumd8faa362007-04-27 19:54:29 +0000123 def test_unescape_all(self):
124 self.assertEquals(unescape("&lt;Donald Duck &amp; Co&gt;"),
125 "<Donald Duck & Co>")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000126
Guido van Rossumd8faa362007-04-27 19:54:29 +0000127 def test_unescape_extra(self):
128 self.assertEquals(unescape("Hei på deg", {"å" : "&aring;"}),
129 "Hei p&aring; deg")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000130
Guido van Rossumd8faa362007-04-27 19:54:29 +0000131 def test_unescape_amp_extra(self):
132 self.assertEquals(unescape("&amp;foo;", {"&foo;": "splat"}), "&foo;")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000133
Guido van Rossumd8faa362007-04-27 19:54:29 +0000134 # ===== quoteattr
135 def test_quoteattr_basic(self):
136 self.assertEquals(quoteattr("Donald Duck & Co"),
137 '"Donald Duck &amp; Co"')
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000138
Guido van Rossumd8faa362007-04-27 19:54:29 +0000139 def test_single_quoteattr(self):
140 self.assertEquals(quoteattr('Includes "double" quotes'),
141 '\'Includes "double" quotes\'')
Fred Drake32f3add2002-10-28 17:58:48 +0000142
Guido van Rossumd8faa362007-04-27 19:54:29 +0000143 def test_double_quoteattr(self):
144 self.assertEquals(quoteattr("Includes 'single' quotes"),
145 "\"Includes 'single' quotes\"")
Fred Drakeacd32d32001-07-19 16:10:15 +0000146
Guido van Rossumd8faa362007-04-27 19:54:29 +0000147 def test_single_double_quoteattr(self):
148 self.assertEquals(quoteattr("Includes 'single' and \"double\" quotes"),
149 "\"Includes 'single' and &quot;double&quot; quotes\"")
Fred Drakeacd32d32001-07-19 16:10:15 +0000150
Guido van Rossumd8faa362007-04-27 19:54:29 +0000151 # ===== make_parser
152 def test_make_parser(self):
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000153 # Creating a parser should succeed - it should fall back
154 # to the expatreader
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000155 p = make_parser(['xml.parsers.no_such_parser'])
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000156
157
Lars Gustäbel96753b32000-09-24 12:24:24 +0000158# ===== XMLGenerator
159
160start = '<?xml version="1.0" encoding="iso-8859-1"?>\n'
161
Guido van Rossumd8faa362007-04-27 19:54:29 +0000162class XmlgenTest(unittest.TestCase):
163 def test_xmlgen_basic(self):
164 result = StringIO()
165 gen = XMLGenerator(result)
166 gen.startDocument()
167 gen.startElement("doc", {})
168 gen.endElement("doc")
169 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000170
Guido van Rossumd8faa362007-04-27 19:54:29 +0000171 self.assertEquals(result.getvalue(), start + "<doc></doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000172
Guido van Rossumd8faa362007-04-27 19:54:29 +0000173 def test_xmlgen_content(self):
174 result = StringIO()
175 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000176
Guido van Rossumd8faa362007-04-27 19:54:29 +0000177 gen.startDocument()
178 gen.startElement("doc", {})
179 gen.characters("huhei")
180 gen.endElement("doc")
181 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000182
Guido van Rossumd8faa362007-04-27 19:54:29 +0000183 self.assertEquals(result.getvalue(), start + "<doc>huhei</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000184
Guido van Rossumd8faa362007-04-27 19:54:29 +0000185 def test_xmlgen_pi(self):
186 result = StringIO()
187 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000188
Guido van Rossumd8faa362007-04-27 19:54:29 +0000189 gen.startDocument()
190 gen.processingInstruction("test", "data")
191 gen.startElement("doc", {})
192 gen.endElement("doc")
193 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000194
Guido van Rossumd8faa362007-04-27 19:54:29 +0000195 self.assertEquals(result.getvalue(), start + "<?test data?><doc></doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000196
Guido van Rossumd8faa362007-04-27 19:54:29 +0000197 def test_xmlgen_content_escape(self):
198 result = StringIO()
199 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000200
Guido van Rossumd8faa362007-04-27 19:54:29 +0000201 gen.startDocument()
202 gen.startElement("doc", {})
203 gen.characters("<huhei&")
204 gen.endElement("doc")
205 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000206
Guido van Rossumd8faa362007-04-27 19:54:29 +0000207 self.assertEquals(result.getvalue(),
208 start + "<doc>&lt;huhei&amp;</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000209
Guido van Rossumd8faa362007-04-27 19:54:29 +0000210 def test_xmlgen_attr_escape(self):
211 result = StringIO()
212 gen = XMLGenerator(result)
Fred Drakec9fadf92001-08-07 19:17:06 +0000213
Guido van Rossumd8faa362007-04-27 19:54:29 +0000214 gen.startDocument()
215 gen.startElement("doc", {"a": '"'})
216 gen.startElement("e", {"a": "'"})
217 gen.endElement("e")
218 gen.startElement("e", {"a": "'\""})
219 gen.endElement("e")
220 gen.startElement("e", {"a": "\n\r\t"})
221 gen.endElement("e")
222 gen.endElement("doc")
223 gen.endDocument()
Fred Drakec9fadf92001-08-07 19:17:06 +0000224
Guido van Rossumd8faa362007-04-27 19:54:29 +0000225 self.assertEquals(result.getvalue(), start +
226 ("<doc a='\"'><e a=\"'\"></e>"
227 "<e a=\"'&quot;\"></e>"
228 "<e a=\"&#10;&#13;&#9;\"></e></doc>"))
Fred Drakec9fadf92001-08-07 19:17:06 +0000229
Guido van Rossumd8faa362007-04-27 19:54:29 +0000230 def test_xmlgen_ignorable(self):
231 result = StringIO()
232 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000233
Guido van Rossumd8faa362007-04-27 19:54:29 +0000234 gen.startDocument()
235 gen.startElement("doc", {})
236 gen.ignorableWhitespace(" ")
237 gen.endElement("doc")
238 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000239
Guido van Rossumd8faa362007-04-27 19:54:29 +0000240 self.assertEquals(result.getvalue(), start + "<doc> </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000241
Guido van Rossumd8faa362007-04-27 19:54:29 +0000242 def test_xmlgen_ns(self):
243 result = StringIO()
244 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000245
Guido van Rossumd8faa362007-04-27 19:54:29 +0000246 gen.startDocument()
247 gen.startPrefixMapping("ns1", ns_uri)
248 gen.startElementNS((ns_uri, "doc"), "ns1:doc", {})
249 # add an unqualified name
250 gen.startElementNS((None, "udoc"), None, {})
251 gen.endElementNS((None, "udoc"), None)
252 gen.endElementNS((ns_uri, "doc"), "ns1:doc")
253 gen.endPrefixMapping("ns1")
254 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000255
Guido van Rossumd8faa362007-04-27 19:54:29 +0000256 self.assertEquals(result.getvalue(), start + \
Martin v. Löwiscf0a1cc2000-10-03 22:35:29 +0000257 ('<ns1:doc xmlns:ns1="%s"><udoc></udoc></ns1:doc>' %
Guido van Rossumd8faa362007-04-27 19:54:29 +0000258 ns_uri))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000259
Guido van Rossumd8faa362007-04-27 19:54:29 +0000260 def test_1463026_1(self):
261 result = StringIO()
262 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000263
Guido van Rossumd8faa362007-04-27 19:54:29 +0000264 gen.startDocument()
265 gen.startElementNS((None, 'a'), 'a', {(None, 'b'):'c'})
266 gen.endElementNS((None, 'a'), 'a')
267 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000268
Guido van Rossumd8faa362007-04-27 19:54:29 +0000269 self.assertEquals(result.getvalue(), start+'<a b="c"></a>')
Thomas Wouterscf297e42007-02-23 15:07:44 +0000270
Guido van Rossumd8faa362007-04-27 19:54:29 +0000271 def test_1463026_2(self):
272 result = StringIO()
273 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000274
Guido van Rossumd8faa362007-04-27 19:54:29 +0000275 gen.startDocument()
276 gen.startPrefixMapping(None, 'qux')
277 gen.startElementNS(('qux', 'a'), 'a', {})
278 gen.endElementNS(('qux', 'a'), 'a')
279 gen.endPrefixMapping(None)
280 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000281
Guido van Rossumd8faa362007-04-27 19:54:29 +0000282 self.assertEquals(result.getvalue(), start+'<a xmlns="qux"></a>')
Thomas Wouterscf297e42007-02-23 15:07:44 +0000283
Guido van Rossumd8faa362007-04-27 19:54:29 +0000284 def test_1463026_3(self):
285 result = StringIO()
286 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000287
Guido van Rossumd8faa362007-04-27 19:54:29 +0000288 gen.startDocument()
289 gen.startPrefixMapping('my', 'qux')
290 gen.startElementNS(('qux', 'a'), 'a', {(None, 'b'):'c'})
291 gen.endElementNS(('qux', 'a'), 'a')
292 gen.endPrefixMapping('my')
293 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000294
Guido van Rossumd8faa362007-04-27 19:54:29 +0000295 self.assertEquals(result.getvalue(),
296 start+'<my:a xmlns:my="qux" b="c"></my:a>')
Lars Gustäbel96753b32000-09-24 12:24:24 +0000297
Fred Drake004d5e62000-10-23 17:22:08 +0000298
Guido van Rossumd8faa362007-04-27 19:54:29 +0000299class XMLFilterBaseTest(unittest.TestCase):
300 def test_filter_basic(self):
301 result = StringIO()
302 gen = XMLGenerator(result)
303 filter = XMLFilterBase()
304 filter.setContentHandler(gen)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000305
Guido van Rossumd8faa362007-04-27 19:54:29 +0000306 filter.startDocument()
307 filter.startElement("doc", {})
308 filter.characters("content")
309 filter.ignorableWhitespace(" ")
310 filter.endElement("doc")
311 filter.endDocument()
312
313 self.assertEquals(result.getvalue(), start + "<doc>content </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000314
315# ===========================================================================
316#
317# expatreader tests
318#
319# ===========================================================================
320
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000321xml_test_out = open(TEST_XMLFILE_OUT).read()
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000322
Guido van Rossumd8faa362007-04-27 19:54:29 +0000323class ExpatReaderTest(XmlTestBase):
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000324
Guido van Rossumd8faa362007-04-27 19:54:29 +0000325 # ===== XMLReader support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000326
Guido van Rossumd8faa362007-04-27 19:54:29 +0000327 def test_expat_file(self):
328 parser = create_parser()
329 result = StringIO()
330 xmlgen = XMLGenerator(result)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000331
Guido van Rossumd8faa362007-04-27 19:54:29 +0000332 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000333 parser.parse(open(TEST_XMLFILE))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000334
Guido van Rossumd8faa362007-04-27 19:54:29 +0000335 self.assertEquals(result.getvalue(), xml_test_out)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000336
Guido van Rossumd8faa362007-04-27 19:54:29 +0000337 # ===== DTDHandler support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000338
Guido van Rossumd8faa362007-04-27 19:54:29 +0000339 class TestDTDHandler:
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000340
Guido van Rossumd8faa362007-04-27 19:54:29 +0000341 def __init__(self):
342 self._notations = []
343 self._entities = []
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000344
Guido van Rossumd8faa362007-04-27 19:54:29 +0000345 def notationDecl(self, name, publicId, systemId):
346 self._notations.append((name, publicId, systemId))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000347
Guido van Rossumd8faa362007-04-27 19:54:29 +0000348 def unparsedEntityDecl(self, name, publicId, systemId, ndata):
349 self._entities.append((name, publicId, systemId, ndata))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000350
Guido van Rossumd8faa362007-04-27 19:54:29 +0000351 def test_expat_dtdhandler(self):
352 parser = create_parser()
353 handler = self.TestDTDHandler()
354 parser.setDTDHandler(handler)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000355
Guido van Rossumd8faa362007-04-27 19:54:29 +0000356 parser.feed('<!DOCTYPE doc [\n')
357 parser.feed(' <!ENTITY img SYSTEM "expat.gif" NDATA GIF>\n')
358 parser.feed(' <!NOTATION GIF PUBLIC "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN">\n')
359 parser.feed(']>\n')
360 parser.feed('<doc></doc>')
361 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000362
Guido van Rossumd8faa362007-04-27 19:54:29 +0000363 self.assertEquals(handler._notations,
364 [("GIF", "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN", None)])
365 self.assertEquals(handler._entities, [("img", None, "expat.gif", "GIF")])
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000366
Guido van Rossumd8faa362007-04-27 19:54:29 +0000367 # ===== EntityResolver support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000368
Guido van Rossumd8faa362007-04-27 19:54:29 +0000369 class TestEntityResolver:
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000370
Guido van Rossumd8faa362007-04-27 19:54:29 +0000371 def resolveEntity(self, publicId, systemId):
372 inpsrc = InputSource()
373 inpsrc.setByteStream(StringIO("<entity/>"))
374 return inpsrc
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000375
Guido van Rossumd8faa362007-04-27 19:54:29 +0000376 def test_expat_entityresolver(self):
377 parser = create_parser()
378 parser.setEntityResolver(self.TestEntityResolver())
379 result = StringIO()
380 parser.setContentHandler(XMLGenerator(result))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000381
Guido van Rossumd8faa362007-04-27 19:54:29 +0000382 parser.feed('<!DOCTYPE doc [\n')
383 parser.feed(' <!ENTITY test SYSTEM "whatever">\n')
384 parser.feed(']>\n')
385 parser.feed('<doc>&test;</doc>')
386 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000387
Guido van Rossumd8faa362007-04-27 19:54:29 +0000388 self.assertEquals(result.getvalue(), start +
389 "<doc><entity></entity></doc>")
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000390
Guido van Rossumd8faa362007-04-27 19:54:29 +0000391 # ===== Attributes support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000392
Guido van Rossumd8faa362007-04-27 19:54:29 +0000393 class AttrGatherer(ContentHandler):
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000394
Guido van Rossumd8faa362007-04-27 19:54:29 +0000395 def startElement(self, name, attrs):
396 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000397
Guido van Rossumd8faa362007-04-27 19:54:29 +0000398 def startElementNS(self, name, qname, attrs):
399 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000400
Guido van Rossumd8faa362007-04-27 19:54:29 +0000401 def test_expat_attrs_empty(self):
402 parser = create_parser()
403 gather = self.AttrGatherer()
404 parser.setContentHandler(gather)
405
406 parser.feed("<doc/>")
407 parser.close()
408
409 self.verify_empty_attrs(gather._attrs)
410
411 def test_expat_attrs_wattr(self):
412 parser = create_parser()
413 gather = self.AttrGatherer()
414 parser.setContentHandler(gather)
415
416 parser.feed("<doc attr='val'/>")
417 parser.close()
418
419 self.verify_attrs_wattr(gather._attrs)
420
421 def test_expat_nsattrs_empty(self):
422 parser = create_parser(1)
423 gather = self.AttrGatherer()
424 parser.setContentHandler(gather)
425
426 parser.feed("<doc/>")
427 parser.close()
428
429 self.verify_empty_nsattrs(gather._attrs)
430
431 def test_expat_nsattrs_wattr(self):
432 parser = create_parser(1)
433 gather = self.AttrGatherer()
434 parser.setContentHandler(gather)
435
436 parser.feed("<doc xmlns:ns='%s' ns:attr='val'/>" % ns_uri)
437 parser.close()
438
439 attrs = gather._attrs
440
441 self.assertEquals(attrs.getLength(), 1)
442 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
443 self.assertTrue((attrs.getQNames() == [] or
444 attrs.getQNames() == ["ns:attr"]))
445 self.assertEquals(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +0000446 self.assertIn((ns_uri, "attr"), attrs)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000447 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
448 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
449 self.assertEquals(list(attrs.items()), [((ns_uri, "attr"), "val")])
450 self.assertEquals(list(attrs.values()), ["val"])
451 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
452 self.assertEquals(attrs[(ns_uri, "attr")], "val")
453
454 # ===== InputSource support
455
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000456 def test_expat_inpsource_filename(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000457 parser = create_parser()
458 result = StringIO()
459 xmlgen = XMLGenerator(result)
460
461 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000462 parser.parse(TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000463
464 self.assertEquals(result.getvalue(), xml_test_out)
465
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000466 def test_expat_inpsource_sysid(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000467 parser = create_parser()
468 result = StringIO()
469 xmlgen = XMLGenerator(result)
470
471 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000472 parser.parse(InputSource(TEST_XMLFILE))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000473
474 self.assertEquals(result.getvalue(), xml_test_out)
475
476 def test_expat_inpsource_stream(self):
477 parser = create_parser()
478 result = StringIO()
479 xmlgen = XMLGenerator(result)
480
481 parser.setContentHandler(xmlgen)
482 inpsrc = InputSource()
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000483 inpsrc.setByteStream(open(TEST_XMLFILE))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000484 parser.parse(inpsrc)
485
486 self.assertEquals(result.getvalue(), xml_test_out)
487
488 # ===== IncrementalParser support
489
490 def test_expat_incremental(self):
491 result = StringIO()
492 xmlgen = XMLGenerator(result)
493 parser = create_parser()
494 parser.setContentHandler(xmlgen)
495
496 parser.feed("<doc>")
497 parser.feed("</doc>")
498 parser.close()
499
500 self.assertEquals(result.getvalue(), start + "<doc></doc>")
501
502 def test_expat_incremental_reset(self):
503 result = StringIO()
504 xmlgen = XMLGenerator(result)
505 parser = create_parser()
506 parser.setContentHandler(xmlgen)
507
508 parser.feed("<doc>")
509 parser.feed("text")
510
511 result = StringIO()
512 xmlgen = XMLGenerator(result)
513 parser.setContentHandler(xmlgen)
514 parser.reset()
515
516 parser.feed("<doc>")
517 parser.feed("text")
518 parser.feed("</doc>")
519 parser.close()
520
521 self.assertEquals(result.getvalue(), start + "<doc>text</doc>")
522
523 # ===== Locator support
524
525 def test_expat_locator_noinfo(self):
526 result = StringIO()
527 xmlgen = XMLGenerator(result)
528 parser = create_parser()
529 parser.setContentHandler(xmlgen)
530
531 parser.feed("<doc>")
532 parser.feed("</doc>")
533 parser.close()
534
535 self.assertEquals(parser.getSystemId(), None)
536 self.assertEquals(parser.getPublicId(), None)
537 self.assertEquals(parser.getLineNumber(), 1)
538
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000539 def test_expat_locator_withinfo(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000540 result = StringIO()
541 xmlgen = XMLGenerator(result)
542 parser = create_parser()
543 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000544 parser.parse(TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000545
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000546 self.assertEquals(parser.getSystemId(), TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000547 self.assertEquals(parser.getPublicId(), None)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000548
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000549
550# ===========================================================================
551#
552# error reporting
553#
554# ===========================================================================
555
Guido van Rossumd8faa362007-04-27 19:54:29 +0000556class ErrorReportingTest(unittest.TestCase):
557 def test_expat_inpsource_location(self):
558 parser = create_parser()
559 parser.setContentHandler(ContentHandler()) # do nothing
560 source = InputSource()
561 source.setByteStream(StringIO("<foo bar foobar>")) #ill-formed
562 name = "a file name"
563 source.setSystemId(name)
564 try:
565 parser.parse(source)
566 self.fail()
567 except SAXException as e:
568 self.assertEquals(e.getSystemId(), name)
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000569
Guido van Rossumd8faa362007-04-27 19:54:29 +0000570 def test_expat_incomplete(self):
571 parser = create_parser()
572 parser.setContentHandler(ContentHandler()) # do nothing
573 self.assertRaises(SAXParseException, parser.parse, StringIO("<foo>"))
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000574
Guido van Rossumd8faa362007-04-27 19:54:29 +0000575 def test_sax_parse_exception_str(self):
576 # pass various values from a locator to the SAXParseException to
577 # make sure that the __str__() doesn't fall apart when None is
578 # passed instead of an integer line and column number
579 #
580 # use "normal" values for the locator:
581 str(SAXParseException("message", None,
582 self.DummyLocator(1, 1)))
583 # use None for the line number:
584 str(SAXParseException("message", None,
585 self.DummyLocator(None, 1)))
586 # use None for the column number:
587 str(SAXParseException("message", None,
588 self.DummyLocator(1, None)))
589 # use None for both:
590 str(SAXParseException("message", None,
591 self.DummyLocator(None, None)))
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000592
Guido van Rossumd8faa362007-04-27 19:54:29 +0000593 class DummyLocator:
594 def __init__(self, lineno, colno):
595 self._lineno = lineno
596 self._colno = colno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000597
Guido van Rossumd8faa362007-04-27 19:54:29 +0000598 def getPublicId(self):
599 return "pubid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000600
Guido van Rossumd8faa362007-04-27 19:54:29 +0000601 def getSystemId(self):
602 return "sysid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000603
Guido van Rossumd8faa362007-04-27 19:54:29 +0000604 def getLineNumber(self):
605 return self._lineno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000606
Guido van Rossumd8faa362007-04-27 19:54:29 +0000607 def getColumnNumber(self):
608 return self._colno
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000609
Lars Gustäbelab647872000-09-24 18:40:52 +0000610# ===========================================================================
611#
612# xmlreader tests
613#
614# ===========================================================================
615
Guido van Rossumd8faa362007-04-27 19:54:29 +0000616class XmlReaderTest(XmlTestBase):
Lars Gustäbelab647872000-09-24 18:40:52 +0000617
Guido van Rossumd8faa362007-04-27 19:54:29 +0000618 # ===== AttributesImpl
619 def test_attrs_empty(self):
620 self.verify_empty_attrs(AttributesImpl({}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000621
Guido van Rossumd8faa362007-04-27 19:54:29 +0000622 def test_attrs_wattr(self):
623 self.verify_attrs_wattr(AttributesImpl({"attr" : "val"}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000624
Guido van Rossumd8faa362007-04-27 19:54:29 +0000625 def test_nsattrs_empty(self):
626 self.verify_empty_nsattrs(AttributesNSImpl({}, {}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000627
Guido van Rossumd8faa362007-04-27 19:54:29 +0000628 def test_nsattrs_wattr(self):
629 attrs = AttributesNSImpl({(ns_uri, "attr") : "val"},
630 {(ns_uri, "attr") : "ns:attr"})
Fred Drake004d5e62000-10-23 17:22:08 +0000631
Guido van Rossumd8faa362007-04-27 19:54:29 +0000632 self.assertEquals(attrs.getLength(), 1)
633 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
634 self.assertEquals(attrs.getQNames(), ["ns:attr"])
635 self.assertEquals(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +0000636 self.assertIn((ns_uri, "attr"), attrs)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000637 self.assertEquals(list(attrs.keys()), [(ns_uri, "attr")])
638 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
639 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
640 self.assertEquals(list(attrs.items()), [((ns_uri, "attr"), "val")])
641 self.assertEquals(list(attrs.values()), ["val"])
642 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
643 self.assertEquals(attrs.getValueByQName("ns:attr"), "val")
644 self.assertEquals(attrs.getNameByQName("ns:attr"), (ns_uri, "attr"))
645 self.assertEquals(attrs[(ns_uri, "attr")], "val")
646 self.assertEquals(attrs.getQNameByName((ns_uri, "attr")), "ns:attr")
Fred Drake004d5e62000-10-23 17:22:08 +0000647
Lars Gustäbelab647872000-09-24 18:40:52 +0000648
Guido van Rossumd8faa362007-04-27 19:54:29 +0000649 # During the development of Python 2.5, an attempt to move the "xml"
650 # package implementation to a new package ("xmlcore") proved painful.
651 # The goal of this change was to allow applications to be able to
652 # obtain and rely on behavior in the standard library implementation
653 # of the XML support without needing to be concerned about the
654 # availability of the PyXML implementation.
655 #
656 # While the existing import hackery in Lib/xml/__init__.py can cause
657 # PyXML's _xmlpus package to supplant the "xml" package, that only
658 # works because either implementation uses the "xml" package name for
659 # imports.
660 #
661 # The move resulted in a number of problems related to the fact that
662 # the import machinery's "package context" is based on the name that's
663 # being imported rather than the __name__ of the actual package
664 # containment; it wasn't possible for the "xml" package to be replaced
665 # by a simple module that indirected imports to the "xmlcore" package.
666 #
667 # The following two tests exercised bugs that were introduced in that
668 # attempt. Keeping these tests around will help detect problems with
669 # other attempts to provide reliable access to the standard library's
670 # implementation of the XML support.
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000671
Guido van Rossumd8faa362007-04-27 19:54:29 +0000672 def test_sf_1511497(self):
673 # Bug report: http://www.python.org/sf/1511497
674 import sys
675 old_modules = sys.modules.copy()
676 for modname in list(sys.modules.keys()):
677 if modname.startswith("xml."):
678 del sys.modules[modname]
679 try:
680 import xml.sax.expatreader
681 module = xml.sax.expatreader
682 self.assertEquals(module.__name__, "xml.sax.expatreader")
683 finally:
684 sys.modules.update(old_modules)
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000685
Guido van Rossumd8faa362007-04-27 19:54:29 +0000686 def test_sf_1513611(self):
687 # Bug report: http://www.python.org/sf/1513611
688 sio = StringIO("invalid")
689 parser = make_parser()
690 from xml.sax import SAXParseException
691 self.assertRaises(SAXParseException, parser.parse, sio)
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000692
Lars Gustäbel96753b32000-09-24 12:24:24 +0000693
Christian Heimesbbe741d2008-03-28 10:53:29 +0000694def test_main():
Guido van Rossumd8faa362007-04-27 19:54:29 +0000695 run_unittest(MakeParserTest,
696 SaxutilsTest,
697 XmlgenTest,
698 ExpatReaderTest,
699 ErrorReportingTest,
700 XmlReaderTest)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000701
Guido van Rossumd8faa362007-04-27 19:54:29 +0000702if __name__ == "__main__":
Christian Heimesbbe741d2008-03-28 10:53:29 +0000703 test_main()