blob: a45d1f322e00e3f651b4b3a54977dd13ba3cb0a1 [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
Guido van Rossume2ae77b2001-10-24 20:42:55 +000018import os
Lars Gustäbel96753b32000-09-24 12:24:24 +000019
Guido van Rossumd8faa362007-04-27 19:54:29 +000020ns_uri = "http://www.python.org/xml-ns/saxtest/"
Lars Gustäbel96753b32000-09-24 12:24:24 +000021
Guido van Rossumd8faa362007-04-27 19:54:29 +000022class XmlTestBase(unittest.TestCase):
23 def verify_empty_attrs(self, attrs):
24 self.assertRaises(KeyError, attrs.getValue, "attr")
25 self.assertRaises(KeyError, attrs.getValueByQName, "attr")
26 self.assertRaises(KeyError, attrs.getNameByQName, "attr")
27 self.assertRaises(KeyError, attrs.getQNameByName, "attr")
28 self.assertRaises(KeyError, attrs.__getitem__, "attr")
29 self.assertEquals(attrs.getLength(), 0)
30 self.assertEquals(attrs.getNames(), [])
31 self.assertEquals(attrs.getQNames(), [])
32 self.assertEquals(len(attrs), 0)
33 self.assertFalse("attr" in attrs)
34 self.assertEquals(list(attrs.keys()), [])
35 self.assertEquals(attrs.get("attrs"), None)
36 self.assertEquals(attrs.get("attrs", 25), 25)
37 self.assertEquals(list(attrs.items()), [])
38 self.assertEquals(list(attrs.values()), [])
Lars Gustäbel96753b32000-09-24 12:24:24 +000039
Guido van Rossumd8faa362007-04-27 19:54:29 +000040 def verify_empty_nsattrs(self, attrs):
41 self.assertRaises(KeyError, attrs.getValue, (ns_uri, "attr"))
42 self.assertRaises(KeyError, attrs.getValueByQName, "ns:attr")
43 self.assertRaises(KeyError, attrs.getNameByQName, "ns:attr")
44 self.assertRaises(KeyError, attrs.getQNameByName, (ns_uri, "attr"))
45 self.assertRaises(KeyError, attrs.__getitem__, (ns_uri, "attr"))
46 self.assertEquals(attrs.getLength(), 0)
47 self.assertEquals(attrs.getNames(), [])
48 self.assertEquals(attrs.getQNames(), [])
49 self.assertEquals(len(attrs), 0)
50 self.assertFalse((ns_uri, "attr") in attrs)
51 self.assertEquals(list(attrs.keys()), [])
52 self.assertEquals(attrs.get((ns_uri, "attr")), None)
53 self.assertEquals(attrs.get((ns_uri, "attr"), 25), 25)
54 self.assertEquals(list(attrs.items()), [])
55 self.assertEquals(list(attrs.values()), [])
Lars Gustäbel96753b32000-09-24 12:24:24 +000056
Guido van Rossumd8faa362007-04-27 19:54:29 +000057 def verify_attrs_wattr(self, attrs):
58 self.assertEquals(attrs.getLength(), 1)
59 self.assertEquals(attrs.getNames(), ["attr"])
60 self.assertEquals(attrs.getQNames(), ["attr"])
61 self.assertEquals(len(attrs), 1)
62 self.assertTrue("attr" in attrs)
63 self.assertEquals(list(attrs.keys()), ["attr"])
64 self.assertEquals(attrs.get("attr"), "val")
65 self.assertEquals(attrs.get("attr", 25), "val")
66 self.assertEquals(list(attrs.items()), [("attr", "val")])
67 self.assertEquals(list(attrs.values()), ["val"])
68 self.assertEquals(attrs.getValue("attr"), "val")
69 self.assertEquals(attrs.getValueByQName("attr"), "val")
70 self.assertEquals(attrs.getNameByQName("attr"), "attr")
71 self.assertEquals(attrs["attr"], "val")
72 self.assertEquals(attrs.getQNameByName("attr"), "attr")
Lars Gustäbel96753b32000-09-24 12:24:24 +000073
Guido van Rossumd8faa362007-04-27 19:54:29 +000074class MakeParserTest(unittest.TestCase):
75 def test_make_parser2(self):
Lars Gustäbel2fc52942000-10-24 15:35:07 +000076 # Creating parsers several times in a row should succeed.
77 # Testing this because there have been failures of this kind
78 # before.
Thomas Wouters0e3f5912006-08-11 14:57:12 +000079 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000080 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000081 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000082 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000083 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000084 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000085 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000086 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000087 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000088 p = make_parser()
Thomas Wouters0e3f5912006-08-11 14:57:12 +000089 from xml.sax import make_parser
Lars Gustäbel2fc52942000-10-24 15:35:07 +000090 p = make_parser()
Tim Petersd2bf3b72001-01-18 02:22:22 +000091
92
Lars Gustäbel96753b32000-09-24 12:24:24 +000093# ===========================================================================
94#
95# saxutils tests
96#
97# ===========================================================================
98
Guido van Rossumd8faa362007-04-27 19:54:29 +000099class SaxutilsTest(unittest.TestCase):
100 # ===== escape
101 def test_escape_basic(self):
102 self.assertEquals(escape("Donald Duck & Co"), "Donald Duck & Co")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000103
Guido van Rossumd8faa362007-04-27 19:54:29 +0000104 def test_escape_all(self):
105 self.assertEquals(escape("<Donald Duck & Co>"),
106 "&lt;Donald Duck &amp; Co&gt;")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000107
Guido van Rossumd8faa362007-04-27 19:54:29 +0000108 def test_escape_extra(self):
109 self.assertEquals(escape("Hei på deg", {"å" : "&aring;"}),
110 "Hei p&aring; deg")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000111
Guido van Rossumd8faa362007-04-27 19:54:29 +0000112 # ===== unescape
113 def test_unescape_basic(self):
114 self.assertEquals(unescape("Donald Duck &amp; Co"), "Donald Duck & Co")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000115
Guido van Rossumd8faa362007-04-27 19:54:29 +0000116 def test_unescape_all(self):
117 self.assertEquals(unescape("&lt;Donald Duck &amp; Co&gt;"),
118 "<Donald Duck & Co>")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000119
Guido van Rossumd8faa362007-04-27 19:54:29 +0000120 def test_unescape_extra(self):
121 self.assertEquals(unescape("Hei på deg", {"å" : "&aring;"}),
122 "Hei p&aring; deg")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000123
Guido van Rossumd8faa362007-04-27 19:54:29 +0000124 def test_unescape_amp_extra(self):
125 self.assertEquals(unescape("&amp;foo;", {"&foo;": "splat"}), "&foo;")
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000126
Guido van Rossumd8faa362007-04-27 19:54:29 +0000127 # ===== quoteattr
128 def test_quoteattr_basic(self):
129 self.assertEquals(quoteattr("Donald Duck & Co"),
130 '"Donald Duck &amp; Co"')
Martin v. Löwis74b51ac2002-10-26 14:50:45 +0000131
Guido van Rossumd8faa362007-04-27 19:54:29 +0000132 def test_single_quoteattr(self):
133 self.assertEquals(quoteattr('Includes "double" quotes'),
134 '\'Includes "double" quotes\'')
Fred Drake32f3add2002-10-28 17:58:48 +0000135
Guido van Rossumd8faa362007-04-27 19:54:29 +0000136 def test_double_quoteattr(self):
137 self.assertEquals(quoteattr("Includes 'single' quotes"),
138 "\"Includes 'single' quotes\"")
Fred Drakeacd32d32001-07-19 16:10:15 +0000139
Guido van Rossumd8faa362007-04-27 19:54:29 +0000140 def test_single_double_quoteattr(self):
141 self.assertEquals(quoteattr("Includes 'single' and \"double\" quotes"),
142 "\"Includes 'single' and &quot;double&quot; quotes\"")
Fred Drakeacd32d32001-07-19 16:10:15 +0000143
Guido van Rossumd8faa362007-04-27 19:54:29 +0000144 # ===== make_parser
145 def test_make_parser(self):
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000146 # Creating a parser should succeed - it should fall back
147 # to the expatreader
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000148 p = make_parser(['xml.parsers.no_such_parser'])
Martin v. Löwis962c9e72000-10-06 17:41:52 +0000149
150
Lars Gustäbel96753b32000-09-24 12:24:24 +0000151# ===== XMLGenerator
152
153start = '<?xml version="1.0" encoding="iso-8859-1"?>\n'
154
Guido van Rossumd8faa362007-04-27 19:54:29 +0000155class XmlgenTest(unittest.TestCase):
156 def test_xmlgen_basic(self):
157 result = StringIO()
158 gen = XMLGenerator(result)
159 gen.startDocument()
160 gen.startElement("doc", {})
161 gen.endElement("doc")
162 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000163
Guido van Rossumd8faa362007-04-27 19:54:29 +0000164 self.assertEquals(result.getvalue(), start + "<doc></doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000165
Guido van Rossumd8faa362007-04-27 19:54:29 +0000166 def test_xmlgen_content(self):
167 result = StringIO()
168 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000169
Guido van Rossumd8faa362007-04-27 19:54:29 +0000170 gen.startDocument()
171 gen.startElement("doc", {})
172 gen.characters("huhei")
173 gen.endElement("doc")
174 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000175
Guido van Rossumd8faa362007-04-27 19:54:29 +0000176 self.assertEquals(result.getvalue(), start + "<doc>huhei</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000177
Guido van Rossumd8faa362007-04-27 19:54:29 +0000178 def test_xmlgen_pi(self):
179 result = StringIO()
180 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000181
Guido van Rossumd8faa362007-04-27 19:54:29 +0000182 gen.startDocument()
183 gen.processingInstruction("test", "data")
184 gen.startElement("doc", {})
185 gen.endElement("doc")
186 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000187
Guido van Rossumd8faa362007-04-27 19:54:29 +0000188 self.assertEquals(result.getvalue(), start + "<?test data?><doc></doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000189
Guido van Rossumd8faa362007-04-27 19:54:29 +0000190 def test_xmlgen_content_escape(self):
191 result = StringIO()
192 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000193
Guido van Rossumd8faa362007-04-27 19:54:29 +0000194 gen.startDocument()
195 gen.startElement("doc", {})
196 gen.characters("<huhei&")
197 gen.endElement("doc")
198 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000199
Guido van Rossumd8faa362007-04-27 19:54:29 +0000200 self.assertEquals(result.getvalue(),
201 start + "<doc>&lt;huhei&amp;</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000202
Guido van Rossumd8faa362007-04-27 19:54:29 +0000203 def test_xmlgen_attr_escape(self):
204 result = StringIO()
205 gen = XMLGenerator(result)
Fred Drakec9fadf92001-08-07 19:17:06 +0000206
Guido van Rossumd8faa362007-04-27 19:54:29 +0000207 gen.startDocument()
208 gen.startElement("doc", {"a": '"'})
209 gen.startElement("e", {"a": "'"})
210 gen.endElement("e")
211 gen.startElement("e", {"a": "'\""})
212 gen.endElement("e")
213 gen.startElement("e", {"a": "\n\r\t"})
214 gen.endElement("e")
215 gen.endElement("doc")
216 gen.endDocument()
Fred Drakec9fadf92001-08-07 19:17:06 +0000217
Guido van Rossumd8faa362007-04-27 19:54:29 +0000218 self.assertEquals(result.getvalue(), start +
219 ("<doc a='\"'><e a=\"'\"></e>"
220 "<e a=\"'&quot;\"></e>"
221 "<e a=\"&#10;&#13;&#9;\"></e></doc>"))
Fred Drakec9fadf92001-08-07 19:17:06 +0000222
Guido van Rossumd8faa362007-04-27 19:54:29 +0000223 def test_xmlgen_ignorable(self):
224 result = StringIO()
225 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000226
Guido van Rossumd8faa362007-04-27 19:54:29 +0000227 gen.startDocument()
228 gen.startElement("doc", {})
229 gen.ignorableWhitespace(" ")
230 gen.endElement("doc")
231 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000232
Guido van Rossumd8faa362007-04-27 19:54:29 +0000233 self.assertEquals(result.getvalue(), start + "<doc> </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000234
Guido van Rossumd8faa362007-04-27 19:54:29 +0000235 def test_xmlgen_ns(self):
236 result = StringIO()
237 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000238
Guido van Rossumd8faa362007-04-27 19:54:29 +0000239 gen.startDocument()
240 gen.startPrefixMapping("ns1", ns_uri)
241 gen.startElementNS((ns_uri, "doc"), "ns1:doc", {})
242 # add an unqualified name
243 gen.startElementNS((None, "udoc"), None, {})
244 gen.endElementNS((None, "udoc"), None)
245 gen.endElementNS((ns_uri, "doc"), "ns1:doc")
246 gen.endPrefixMapping("ns1")
247 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000248
Guido van Rossumd8faa362007-04-27 19:54:29 +0000249 self.assertEquals(result.getvalue(), start + \
Martin v. Löwiscf0a1cc2000-10-03 22:35:29 +0000250 ('<ns1:doc xmlns:ns1="%s"><udoc></udoc></ns1:doc>' %
Guido van Rossumd8faa362007-04-27 19:54:29 +0000251 ns_uri))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000252
Guido van Rossumd8faa362007-04-27 19:54:29 +0000253 def test_1463026_1(self):
254 result = StringIO()
255 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000256
Guido van Rossumd8faa362007-04-27 19:54:29 +0000257 gen.startDocument()
258 gen.startElementNS((None, 'a'), 'a', {(None, 'b'):'c'})
259 gen.endElementNS((None, 'a'), 'a')
260 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000261
Guido van Rossumd8faa362007-04-27 19:54:29 +0000262 self.assertEquals(result.getvalue(), start+'<a b="c"></a>')
Thomas Wouterscf297e42007-02-23 15:07:44 +0000263
Guido van Rossumd8faa362007-04-27 19:54:29 +0000264 def test_1463026_2(self):
265 result = StringIO()
266 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000267
Guido van Rossumd8faa362007-04-27 19:54:29 +0000268 gen.startDocument()
269 gen.startPrefixMapping(None, 'qux')
270 gen.startElementNS(('qux', 'a'), 'a', {})
271 gen.endElementNS(('qux', 'a'), 'a')
272 gen.endPrefixMapping(None)
273 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000274
Guido van Rossumd8faa362007-04-27 19:54:29 +0000275 self.assertEquals(result.getvalue(), start+'<a xmlns="qux"></a>')
Thomas Wouterscf297e42007-02-23 15:07:44 +0000276
Guido van Rossumd8faa362007-04-27 19:54:29 +0000277 def test_1463026_3(self):
278 result = StringIO()
279 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000280
Guido van Rossumd8faa362007-04-27 19:54:29 +0000281 gen.startDocument()
282 gen.startPrefixMapping('my', 'qux')
283 gen.startElementNS(('qux', 'a'), 'a', {(None, 'b'):'c'})
284 gen.endElementNS(('qux', 'a'), 'a')
285 gen.endPrefixMapping('my')
286 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000287
Guido van Rossumd8faa362007-04-27 19:54:29 +0000288 self.assertEquals(result.getvalue(),
289 start+'<my:a xmlns:my="qux" b="c"></my:a>')
Lars Gustäbel96753b32000-09-24 12:24:24 +0000290
Fred Drake004d5e62000-10-23 17:22:08 +0000291
Guido van Rossumd8faa362007-04-27 19:54:29 +0000292class XMLFilterBaseTest(unittest.TestCase):
293 def test_filter_basic(self):
294 result = StringIO()
295 gen = XMLGenerator(result)
296 filter = XMLFilterBase()
297 filter.setContentHandler(gen)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000298
Guido van Rossumd8faa362007-04-27 19:54:29 +0000299 filter.startDocument()
300 filter.startElement("doc", {})
301 filter.characters("content")
302 filter.ignorableWhitespace(" ")
303 filter.endElement("doc")
304 filter.endDocument()
305
306 self.assertEquals(result.getvalue(), start + "<doc>content </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000307
308# ===========================================================================
309#
310# expatreader tests
311#
312# ===========================================================================
313
Skip Montanaro7a98be22007-08-16 14:35:24 +0000314xml_test_out = open(findfile("test.xml.out")).read()
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000315
Guido van Rossumd8faa362007-04-27 19:54:29 +0000316class ExpatReaderTest(XmlTestBase):
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000317
Guido van Rossumd8faa362007-04-27 19:54:29 +0000318 # ===== XMLReader support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000319
Guido van Rossumd8faa362007-04-27 19:54:29 +0000320 def test_expat_file(self):
321 parser = create_parser()
322 result = StringIO()
323 xmlgen = XMLGenerator(result)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000324
Guido van Rossumd8faa362007-04-27 19:54:29 +0000325 parser.setContentHandler(xmlgen)
Skip Montanaro7a98be22007-08-16 14:35:24 +0000326 parser.parse(open(findfile("test.xml")))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000327
Guido van Rossumd8faa362007-04-27 19:54:29 +0000328 self.assertEquals(result.getvalue(), xml_test_out)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000329
Guido van Rossumd8faa362007-04-27 19:54:29 +0000330 # ===== DTDHandler support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000331
Guido van Rossumd8faa362007-04-27 19:54:29 +0000332 class TestDTDHandler:
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000333
Guido van Rossumd8faa362007-04-27 19:54:29 +0000334 def __init__(self):
335 self._notations = []
336 self._entities = []
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000337
Guido van Rossumd8faa362007-04-27 19:54:29 +0000338 def notationDecl(self, name, publicId, systemId):
339 self._notations.append((name, publicId, systemId))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000340
Guido van Rossumd8faa362007-04-27 19:54:29 +0000341 def unparsedEntityDecl(self, name, publicId, systemId, ndata):
342 self._entities.append((name, publicId, systemId, ndata))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000343
Guido van Rossumd8faa362007-04-27 19:54:29 +0000344 def test_expat_dtdhandler(self):
345 parser = create_parser()
346 handler = self.TestDTDHandler()
347 parser.setDTDHandler(handler)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000348
Guido van Rossumd8faa362007-04-27 19:54:29 +0000349 parser.feed('<!DOCTYPE doc [\n')
350 parser.feed(' <!ENTITY img SYSTEM "expat.gif" NDATA GIF>\n')
351 parser.feed(' <!NOTATION GIF PUBLIC "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN">\n')
352 parser.feed(']>\n')
353 parser.feed('<doc></doc>')
354 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000355
Guido van Rossumd8faa362007-04-27 19:54:29 +0000356 self.assertEquals(handler._notations,
357 [("GIF", "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN", None)])
358 self.assertEquals(handler._entities, [("img", None, "expat.gif", "GIF")])
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000359
Guido van Rossumd8faa362007-04-27 19:54:29 +0000360 # ===== EntityResolver support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000361
Guido van Rossumd8faa362007-04-27 19:54:29 +0000362 class TestEntityResolver:
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000363
Guido van Rossumd8faa362007-04-27 19:54:29 +0000364 def resolveEntity(self, publicId, systemId):
365 inpsrc = InputSource()
366 inpsrc.setByteStream(StringIO("<entity/>"))
367 return inpsrc
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000368
Guido van Rossumd8faa362007-04-27 19:54:29 +0000369 def test_expat_entityresolver(self):
370 parser = create_parser()
371 parser.setEntityResolver(self.TestEntityResolver())
372 result = StringIO()
373 parser.setContentHandler(XMLGenerator(result))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000374
Guido van Rossumd8faa362007-04-27 19:54:29 +0000375 parser.feed('<!DOCTYPE doc [\n')
376 parser.feed(' <!ENTITY test SYSTEM "whatever">\n')
377 parser.feed(']>\n')
378 parser.feed('<doc>&test;</doc>')
379 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000380
Guido van Rossumd8faa362007-04-27 19:54:29 +0000381 self.assertEquals(result.getvalue(), start +
382 "<doc><entity></entity></doc>")
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000383
Guido van Rossumd8faa362007-04-27 19:54:29 +0000384 # ===== Attributes support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000385
Guido van Rossumd8faa362007-04-27 19:54:29 +0000386 class AttrGatherer(ContentHandler):
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000387
Guido van Rossumd8faa362007-04-27 19:54:29 +0000388 def startElement(self, name, attrs):
389 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000390
Guido van Rossumd8faa362007-04-27 19:54:29 +0000391 def startElementNS(self, name, qname, attrs):
392 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000393
Guido van Rossumd8faa362007-04-27 19:54:29 +0000394 def test_expat_attrs_empty(self):
395 parser = create_parser()
396 gather = self.AttrGatherer()
397 parser.setContentHandler(gather)
398
399 parser.feed("<doc/>")
400 parser.close()
401
402 self.verify_empty_attrs(gather._attrs)
403
404 def test_expat_attrs_wattr(self):
405 parser = create_parser()
406 gather = self.AttrGatherer()
407 parser.setContentHandler(gather)
408
409 parser.feed("<doc attr='val'/>")
410 parser.close()
411
412 self.verify_attrs_wattr(gather._attrs)
413
414 def test_expat_nsattrs_empty(self):
415 parser = create_parser(1)
416 gather = self.AttrGatherer()
417 parser.setContentHandler(gather)
418
419 parser.feed("<doc/>")
420 parser.close()
421
422 self.verify_empty_nsattrs(gather._attrs)
423
424 def test_expat_nsattrs_wattr(self):
425 parser = create_parser(1)
426 gather = self.AttrGatherer()
427 parser.setContentHandler(gather)
428
429 parser.feed("<doc xmlns:ns='%s' ns:attr='val'/>" % ns_uri)
430 parser.close()
431
432 attrs = gather._attrs
433
434 self.assertEquals(attrs.getLength(), 1)
435 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
436 self.assertTrue((attrs.getQNames() == [] or
437 attrs.getQNames() == ["ns:attr"]))
438 self.assertEquals(len(attrs), 1)
439 self.assertTrue((ns_uri, "attr") in attrs)
440 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
441 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
442 self.assertEquals(list(attrs.items()), [((ns_uri, "attr"), "val")])
443 self.assertEquals(list(attrs.values()), ["val"])
444 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
445 self.assertEquals(attrs[(ns_uri, "attr")], "val")
446
447 # ===== InputSource support
448
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000449 def test_expat_inpsource_filename(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000450 parser = create_parser()
451 result = StringIO()
452 xmlgen = XMLGenerator(result)
453
454 parser.setContentHandler(xmlgen)
Skip Montanaro7a98be22007-08-16 14:35:24 +0000455 parser.parse(findfile("test.xml"))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000456
457 self.assertEquals(result.getvalue(), xml_test_out)
458
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000459 def test_expat_inpsource_sysid(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000460 parser = create_parser()
461 result = StringIO()
462 xmlgen = XMLGenerator(result)
463
464 parser.setContentHandler(xmlgen)
Skip Montanaro7a98be22007-08-16 14:35:24 +0000465 parser.parse(InputSource(findfile("test.xml")))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000466
467 self.assertEquals(result.getvalue(), xml_test_out)
468
469 def test_expat_inpsource_stream(self):
470 parser = create_parser()
471 result = StringIO()
472 xmlgen = XMLGenerator(result)
473
474 parser.setContentHandler(xmlgen)
475 inpsrc = InputSource()
Skip Montanaro7a98be22007-08-16 14:35:24 +0000476 inpsrc.setByteStream(open(findfile("test.xml")))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000477 parser.parse(inpsrc)
478
479 self.assertEquals(result.getvalue(), xml_test_out)
480
481 # ===== IncrementalParser support
482
483 def test_expat_incremental(self):
484 result = StringIO()
485 xmlgen = XMLGenerator(result)
486 parser = create_parser()
487 parser.setContentHandler(xmlgen)
488
489 parser.feed("<doc>")
490 parser.feed("</doc>")
491 parser.close()
492
493 self.assertEquals(result.getvalue(), start + "<doc></doc>")
494
495 def test_expat_incremental_reset(self):
496 result = StringIO()
497 xmlgen = XMLGenerator(result)
498 parser = create_parser()
499 parser.setContentHandler(xmlgen)
500
501 parser.feed("<doc>")
502 parser.feed("text")
503
504 result = StringIO()
505 xmlgen = XMLGenerator(result)
506 parser.setContentHandler(xmlgen)
507 parser.reset()
508
509 parser.feed("<doc>")
510 parser.feed("text")
511 parser.feed("</doc>")
512 parser.close()
513
514 self.assertEquals(result.getvalue(), start + "<doc>text</doc>")
515
516 # ===== Locator support
517
518 def test_expat_locator_noinfo(self):
519 result = StringIO()
520 xmlgen = XMLGenerator(result)
521 parser = create_parser()
522 parser.setContentHandler(xmlgen)
523
524 parser.feed("<doc>")
525 parser.feed("</doc>")
526 parser.close()
527
528 self.assertEquals(parser.getSystemId(), None)
529 self.assertEquals(parser.getPublicId(), None)
530 self.assertEquals(parser.getLineNumber(), 1)
531
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000532 def test_expat_locator_withinfo(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000533 result = StringIO()
534 xmlgen = XMLGenerator(result)
535 parser = create_parser()
536 parser.setContentHandler(xmlgen)
537 parser.parse(findfile("test.xml"))
538
539 self.assertEquals(parser.getSystemId(), findfile("test.xml"))
540 self.assertEquals(parser.getPublicId(), None)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000541
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000542
543# ===========================================================================
544#
545# error reporting
546#
547# ===========================================================================
548
Guido van Rossumd8faa362007-04-27 19:54:29 +0000549class ErrorReportingTest(unittest.TestCase):
550 def test_expat_inpsource_location(self):
551 parser = create_parser()
552 parser.setContentHandler(ContentHandler()) # do nothing
553 source = InputSource()
554 source.setByteStream(StringIO("<foo bar foobar>")) #ill-formed
555 name = "a file name"
556 source.setSystemId(name)
557 try:
558 parser.parse(source)
559 self.fail()
560 except SAXException as e:
561 self.assertEquals(e.getSystemId(), name)
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000562
Guido van Rossumd8faa362007-04-27 19:54:29 +0000563 def test_expat_incomplete(self):
564 parser = create_parser()
565 parser.setContentHandler(ContentHandler()) # do nothing
566 self.assertRaises(SAXParseException, parser.parse, StringIO("<foo>"))
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000567
Guido van Rossumd8faa362007-04-27 19:54:29 +0000568 def test_sax_parse_exception_str(self):
569 # pass various values from a locator to the SAXParseException to
570 # make sure that the __str__() doesn't fall apart when None is
571 # passed instead of an integer line and column number
572 #
573 # use "normal" values for the locator:
574 str(SAXParseException("message", None,
575 self.DummyLocator(1, 1)))
576 # use None for the line number:
577 str(SAXParseException("message", None,
578 self.DummyLocator(None, 1)))
579 # use None for the column number:
580 str(SAXParseException("message", None,
581 self.DummyLocator(1, None)))
582 # use None for both:
583 str(SAXParseException("message", None,
584 self.DummyLocator(None, None)))
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000585
Guido van Rossumd8faa362007-04-27 19:54:29 +0000586 class DummyLocator:
587 def __init__(self, lineno, colno):
588 self._lineno = lineno
589 self._colno = colno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000590
Guido van Rossumd8faa362007-04-27 19:54:29 +0000591 def getPublicId(self):
592 return "pubid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000593
Guido van Rossumd8faa362007-04-27 19:54:29 +0000594 def getSystemId(self):
595 return "sysid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000596
Guido van Rossumd8faa362007-04-27 19:54:29 +0000597 def getLineNumber(self):
598 return self._lineno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000599
Guido van Rossumd8faa362007-04-27 19:54:29 +0000600 def getColumnNumber(self):
601 return self._colno
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000602
Lars Gustäbelab647872000-09-24 18:40:52 +0000603# ===========================================================================
604#
605# xmlreader tests
606#
607# ===========================================================================
608
Guido van Rossumd8faa362007-04-27 19:54:29 +0000609class XmlReaderTest(XmlTestBase):
Lars Gustäbelab647872000-09-24 18:40:52 +0000610
Guido van Rossumd8faa362007-04-27 19:54:29 +0000611 # ===== AttributesImpl
612 def test_attrs_empty(self):
613 self.verify_empty_attrs(AttributesImpl({}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000614
Guido van Rossumd8faa362007-04-27 19:54:29 +0000615 def test_attrs_wattr(self):
616 self.verify_attrs_wattr(AttributesImpl({"attr" : "val"}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000617
Guido van Rossumd8faa362007-04-27 19:54:29 +0000618 def test_nsattrs_empty(self):
619 self.verify_empty_nsattrs(AttributesNSImpl({}, {}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000620
Guido van Rossumd8faa362007-04-27 19:54:29 +0000621 def test_nsattrs_wattr(self):
622 attrs = AttributesNSImpl({(ns_uri, "attr") : "val"},
623 {(ns_uri, "attr") : "ns:attr"})
Fred Drake004d5e62000-10-23 17:22:08 +0000624
Guido van Rossumd8faa362007-04-27 19:54:29 +0000625 self.assertEquals(attrs.getLength(), 1)
626 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
627 self.assertEquals(attrs.getQNames(), ["ns:attr"])
628 self.assertEquals(len(attrs), 1)
629 self.assertTrue((ns_uri, "attr") in attrs)
630 self.assertEquals(list(attrs.keys()), [(ns_uri, "attr")])
631 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
632 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
633 self.assertEquals(list(attrs.items()), [((ns_uri, "attr"), "val")])
634 self.assertEquals(list(attrs.values()), ["val"])
635 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
636 self.assertEquals(attrs.getValueByQName("ns:attr"), "val")
637 self.assertEquals(attrs.getNameByQName("ns:attr"), (ns_uri, "attr"))
638 self.assertEquals(attrs[(ns_uri, "attr")], "val")
639 self.assertEquals(attrs.getQNameByName((ns_uri, "attr")), "ns:attr")
Fred Drake004d5e62000-10-23 17:22:08 +0000640
Lars Gustäbelab647872000-09-24 18:40:52 +0000641
Guido van Rossumd8faa362007-04-27 19:54:29 +0000642 # During the development of Python 2.5, an attempt to move the "xml"
643 # package implementation to a new package ("xmlcore") proved painful.
644 # The goal of this change was to allow applications to be able to
645 # obtain and rely on behavior in the standard library implementation
646 # of the XML support without needing to be concerned about the
647 # availability of the PyXML implementation.
648 #
649 # While the existing import hackery in Lib/xml/__init__.py can cause
650 # PyXML's _xmlpus package to supplant the "xml" package, that only
651 # works because either implementation uses the "xml" package name for
652 # imports.
653 #
654 # The move resulted in a number of problems related to the fact that
655 # the import machinery's "package context" is based on the name that's
656 # being imported rather than the __name__ of the actual package
657 # containment; it wasn't possible for the "xml" package to be replaced
658 # by a simple module that indirected imports to the "xmlcore" package.
659 #
660 # The following two tests exercised bugs that were introduced in that
661 # attempt. Keeping these tests around will help detect problems with
662 # other attempts to provide reliable access to the standard library's
663 # implementation of the XML support.
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000664
Guido van Rossumd8faa362007-04-27 19:54:29 +0000665 def test_sf_1511497(self):
666 # Bug report: http://www.python.org/sf/1511497
667 import sys
668 old_modules = sys.modules.copy()
669 for modname in list(sys.modules.keys()):
670 if modname.startswith("xml."):
671 del sys.modules[modname]
672 try:
673 import xml.sax.expatreader
674 module = xml.sax.expatreader
675 self.assertEquals(module.__name__, "xml.sax.expatreader")
676 finally:
677 sys.modules.update(old_modules)
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000678
Guido van Rossumd8faa362007-04-27 19:54:29 +0000679 def test_sf_1513611(self):
680 # Bug report: http://www.python.org/sf/1513611
681 sio = StringIO("invalid")
682 parser = make_parser()
683 from xml.sax import SAXParseException
684 self.assertRaises(SAXParseException, parser.parse, sio)
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000685
Lars Gustäbel96753b32000-09-24 12:24:24 +0000686
Christian Heimesbbe741d2008-03-28 10:53:29 +0000687def test_main():
Guido van Rossumd8faa362007-04-27 19:54:29 +0000688 run_unittest(MakeParserTest,
689 SaxutilsTest,
690 XmlgenTest,
691 ExpatReaderTest,
692 ErrorReportingTest,
693 XmlReaderTest)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000694
Guido van Rossumd8faa362007-04-27 19:54:29 +0000695if __name__ == "__main__":
Christian Heimesbbe741d2008-03-28 10:53:29 +0000696 test_main()