blob: 143ddf2940d828207c6956ae40b91f2a7bcb1db3 [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
R. David Murraya90032a2010-10-17 22:46:45 +0000173 def test_xmlgen_basic_empty(self):
174 result = StringIO()
175 gen = XMLGenerator(result, short_empty_elements=True)
176 gen.startDocument()
177 gen.startElement("doc", {})
178 gen.endElement("doc")
179 gen.endDocument()
180
181 self.assertEquals(result.getvalue(), start + "<doc/>")
182
Guido van Rossumd8faa362007-04-27 19:54:29 +0000183 def test_xmlgen_content(self):
184 result = StringIO()
185 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000186
Guido van Rossumd8faa362007-04-27 19:54:29 +0000187 gen.startDocument()
188 gen.startElement("doc", {})
189 gen.characters("huhei")
190 gen.endElement("doc")
191 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000192
Guido van Rossumd8faa362007-04-27 19:54:29 +0000193 self.assertEquals(result.getvalue(), start + "<doc>huhei</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000194
R. David Murraya90032a2010-10-17 22:46:45 +0000195 def test_xmlgen_content_empty(self):
196 result = StringIO()
197 gen = XMLGenerator(result, short_empty_elements=True)
198
199 gen.startDocument()
200 gen.startElement("doc", {})
201 gen.characters("huhei")
202 gen.endElement("doc")
203 gen.endDocument()
204
205 self.assertEquals(result.getvalue(), start + "<doc>huhei</doc>")
206
Guido van Rossumd8faa362007-04-27 19:54:29 +0000207 def test_xmlgen_pi(self):
208 result = StringIO()
209 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000210
Guido van Rossumd8faa362007-04-27 19:54:29 +0000211 gen.startDocument()
212 gen.processingInstruction("test", "data")
213 gen.startElement("doc", {})
214 gen.endElement("doc")
215 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000216
Guido van Rossumd8faa362007-04-27 19:54:29 +0000217 self.assertEquals(result.getvalue(), start + "<?test data?><doc></doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000218
Guido van Rossumd8faa362007-04-27 19:54:29 +0000219 def test_xmlgen_content_escape(self):
220 result = StringIO()
221 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000222
Guido van Rossumd8faa362007-04-27 19:54:29 +0000223 gen.startDocument()
224 gen.startElement("doc", {})
225 gen.characters("<huhei&")
226 gen.endElement("doc")
227 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000228
Guido van Rossumd8faa362007-04-27 19:54:29 +0000229 self.assertEquals(result.getvalue(),
230 start + "<doc>&lt;huhei&amp;</doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000231
Guido van Rossumd8faa362007-04-27 19:54:29 +0000232 def test_xmlgen_attr_escape(self):
233 result = StringIO()
234 gen = XMLGenerator(result)
Fred Drakec9fadf92001-08-07 19:17:06 +0000235
Guido van Rossumd8faa362007-04-27 19:54:29 +0000236 gen.startDocument()
237 gen.startElement("doc", {"a": '"'})
238 gen.startElement("e", {"a": "'"})
239 gen.endElement("e")
240 gen.startElement("e", {"a": "'\""})
241 gen.endElement("e")
242 gen.startElement("e", {"a": "\n\r\t"})
243 gen.endElement("e")
244 gen.endElement("doc")
245 gen.endDocument()
Fred Drakec9fadf92001-08-07 19:17:06 +0000246
Guido van Rossumd8faa362007-04-27 19:54:29 +0000247 self.assertEquals(result.getvalue(), start +
248 ("<doc a='\"'><e a=\"'\"></e>"
249 "<e a=\"'&quot;\"></e>"
250 "<e a=\"&#10;&#13;&#9;\"></e></doc>"))
Fred Drakec9fadf92001-08-07 19:17:06 +0000251
Guido van Rossumd8faa362007-04-27 19:54:29 +0000252 def test_xmlgen_ignorable(self):
253 result = StringIO()
254 gen = XMLGenerator(result)
Fred Drake004d5e62000-10-23 17:22:08 +0000255
Guido van Rossumd8faa362007-04-27 19:54:29 +0000256 gen.startDocument()
257 gen.startElement("doc", {})
258 gen.ignorableWhitespace(" ")
259 gen.endElement("doc")
260 gen.endDocument()
Lars Gustäbel96753b32000-09-24 12:24:24 +0000261
Guido van Rossumd8faa362007-04-27 19:54:29 +0000262 self.assertEquals(result.getvalue(), start + "<doc> </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000263
R. David Murraya90032a2010-10-17 22:46:45 +0000264 def test_xmlgen_ignorable_empty(self):
265 result = StringIO()
266 gen = XMLGenerator(result, short_empty_elements=True)
267
268 gen.startDocument()
269 gen.startElement("doc", {})
270 gen.ignorableWhitespace(" ")
271 gen.endElement("doc")
272 gen.endDocument()
273
274 self.assertEquals(result.getvalue(), start + "<doc> </doc>")
275
Guido van Rossumd8faa362007-04-27 19:54:29 +0000276 def test_xmlgen_ns(self):
277 result = StringIO()
278 gen = XMLGenerator(result)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000279
Guido van Rossumd8faa362007-04-27 19:54:29 +0000280 gen.startDocument()
281 gen.startPrefixMapping("ns1", ns_uri)
282 gen.startElementNS((ns_uri, "doc"), "ns1:doc", {})
283 # add an unqualified name
284 gen.startElementNS((None, "udoc"), None, {})
285 gen.endElementNS((None, "udoc"), None)
286 gen.endElementNS((ns_uri, "doc"), "ns1:doc")
287 gen.endPrefixMapping("ns1")
288 gen.endDocument()
Fred Drake004d5e62000-10-23 17:22:08 +0000289
Guido van Rossumd8faa362007-04-27 19:54:29 +0000290 self.assertEquals(result.getvalue(), start + \
Martin v. Löwiscf0a1cc2000-10-03 22:35:29 +0000291 ('<ns1:doc xmlns:ns1="%s"><udoc></udoc></ns1:doc>' %
Guido van Rossumd8faa362007-04-27 19:54:29 +0000292 ns_uri))
Lars Gustäbel96753b32000-09-24 12:24:24 +0000293
R. David Murraya90032a2010-10-17 22:46:45 +0000294 def test_xmlgen_ns_empty(self):
295 result = StringIO()
296 gen = XMLGenerator(result, short_empty_elements=True)
297
298 gen.startDocument()
299 gen.startPrefixMapping("ns1", ns_uri)
300 gen.startElementNS((ns_uri, "doc"), "ns1:doc", {})
301 # add an unqualified name
302 gen.startElementNS((None, "udoc"), None, {})
303 gen.endElementNS((None, "udoc"), None)
304 gen.endElementNS((ns_uri, "doc"), "ns1:doc")
305 gen.endPrefixMapping("ns1")
306 gen.endDocument()
307
308 self.assertEquals(result.getvalue(), start + \
309 ('<ns1:doc xmlns:ns1="%s"><udoc/></ns1:doc>' %
310 ns_uri))
311
Guido van Rossumd8faa362007-04-27 19:54:29 +0000312 def test_1463026_1(self):
313 result = StringIO()
314 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000315
Guido van Rossumd8faa362007-04-27 19:54:29 +0000316 gen.startDocument()
317 gen.startElementNS((None, 'a'), 'a', {(None, 'b'):'c'})
318 gen.endElementNS((None, 'a'), 'a')
319 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000320
Guido van Rossumd8faa362007-04-27 19:54:29 +0000321 self.assertEquals(result.getvalue(), start+'<a b="c"></a>')
Thomas Wouterscf297e42007-02-23 15:07:44 +0000322
R. David Murraya90032a2010-10-17 22:46:45 +0000323 def test_1463026_1_empty(self):
324 result = StringIO()
325 gen = XMLGenerator(result, short_empty_elements=True)
326
327 gen.startDocument()
328 gen.startElementNS((None, 'a'), 'a', {(None, 'b'):'c'})
329 gen.endElementNS((None, 'a'), 'a')
330 gen.endDocument()
331
332 self.assertEquals(result.getvalue(), start+'<a b="c"/>')
333
Guido van Rossumd8faa362007-04-27 19:54:29 +0000334 def test_1463026_2(self):
335 result = StringIO()
336 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000337
Guido van Rossumd8faa362007-04-27 19:54:29 +0000338 gen.startDocument()
339 gen.startPrefixMapping(None, 'qux')
340 gen.startElementNS(('qux', 'a'), 'a', {})
341 gen.endElementNS(('qux', 'a'), 'a')
342 gen.endPrefixMapping(None)
343 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000344
Guido van Rossumd8faa362007-04-27 19:54:29 +0000345 self.assertEquals(result.getvalue(), start+'<a xmlns="qux"></a>')
Thomas Wouterscf297e42007-02-23 15:07:44 +0000346
R. David Murraya90032a2010-10-17 22:46:45 +0000347 def test_1463026_2_empty(self):
348 result = StringIO()
349 gen = XMLGenerator(result, short_empty_elements=True)
350
351 gen.startDocument()
352 gen.startPrefixMapping(None, 'qux')
353 gen.startElementNS(('qux', 'a'), 'a', {})
354 gen.endElementNS(('qux', 'a'), 'a')
355 gen.endPrefixMapping(None)
356 gen.endDocument()
357
358 self.assertEquals(result.getvalue(), start+'<a xmlns="qux"/>')
359
Guido van Rossumd8faa362007-04-27 19:54:29 +0000360 def test_1463026_3(self):
361 result = StringIO()
362 gen = XMLGenerator(result)
Thomas Wouterscf297e42007-02-23 15:07:44 +0000363
Guido van Rossumd8faa362007-04-27 19:54:29 +0000364 gen.startDocument()
365 gen.startPrefixMapping('my', 'qux')
366 gen.startElementNS(('qux', 'a'), 'a', {(None, 'b'):'c'})
367 gen.endElementNS(('qux', 'a'), 'a')
368 gen.endPrefixMapping('my')
369 gen.endDocument()
Thomas Wouterscf297e42007-02-23 15:07:44 +0000370
Guido van Rossumd8faa362007-04-27 19:54:29 +0000371 self.assertEquals(result.getvalue(),
372 start+'<my:a xmlns:my="qux" b="c"></my:a>')
Lars Gustäbel96753b32000-09-24 12:24:24 +0000373
R. David Murraya90032a2010-10-17 22:46:45 +0000374 def test_1463026_3_empty(self):
375 result = StringIO()
376 gen = XMLGenerator(result, short_empty_elements=True)
377
378 gen.startDocument()
379 gen.startPrefixMapping('my', 'qux')
380 gen.startElementNS(('qux', 'a'), 'a', {(None, 'b'):'c'})
381 gen.endElementNS(('qux', 'a'), 'a')
382 gen.endPrefixMapping('my')
383 gen.endDocument()
384
385 self.assertEquals(result.getvalue(),
386 start+'<my:a xmlns:my="qux" b="c"/>')
387
Fred Drake004d5e62000-10-23 17:22:08 +0000388
Guido van Rossumd8faa362007-04-27 19:54:29 +0000389class XMLFilterBaseTest(unittest.TestCase):
390 def test_filter_basic(self):
391 result = StringIO()
392 gen = XMLGenerator(result)
393 filter = XMLFilterBase()
394 filter.setContentHandler(gen)
Lars Gustäbel96753b32000-09-24 12:24:24 +0000395
Guido van Rossumd8faa362007-04-27 19:54:29 +0000396 filter.startDocument()
397 filter.startElement("doc", {})
398 filter.characters("content")
399 filter.ignorableWhitespace(" ")
400 filter.endElement("doc")
401 filter.endDocument()
402
403 self.assertEquals(result.getvalue(), start + "<doc>content </doc>")
Lars Gustäbel96753b32000-09-24 12:24:24 +0000404
405# ===========================================================================
406#
407# expatreader tests
408#
409# ===========================================================================
410
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000411xml_test_out = open(TEST_XMLFILE_OUT).read()
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000412
Guido van Rossumd8faa362007-04-27 19:54:29 +0000413class ExpatReaderTest(XmlTestBase):
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000414
Guido van Rossumd8faa362007-04-27 19:54:29 +0000415 # ===== XMLReader support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000416
Guido van Rossumd8faa362007-04-27 19:54:29 +0000417 def test_expat_file(self):
418 parser = create_parser()
419 result = StringIO()
420 xmlgen = XMLGenerator(result)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000421
Guido van Rossumd8faa362007-04-27 19:54:29 +0000422 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000423 parser.parse(open(TEST_XMLFILE))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000424
Guido van Rossumd8faa362007-04-27 19:54:29 +0000425 self.assertEquals(result.getvalue(), xml_test_out)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000426
Guido van Rossumd8faa362007-04-27 19:54:29 +0000427 # ===== DTDHandler support
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000428
Guido van Rossumd8faa362007-04-27 19:54:29 +0000429 class TestDTDHandler:
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000430
Guido van Rossumd8faa362007-04-27 19:54:29 +0000431 def __init__(self):
432 self._notations = []
433 self._entities = []
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000434
Guido van Rossumd8faa362007-04-27 19:54:29 +0000435 def notationDecl(self, name, publicId, systemId):
436 self._notations.append((name, publicId, systemId))
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000437
Guido van Rossumd8faa362007-04-27 19:54:29 +0000438 def unparsedEntityDecl(self, name, publicId, systemId, ndata):
439 self._entities.append((name, publicId, systemId, ndata))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000440
Guido van Rossumd8faa362007-04-27 19:54:29 +0000441 def test_expat_dtdhandler(self):
442 parser = create_parser()
443 handler = self.TestDTDHandler()
444 parser.setDTDHandler(handler)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000445
Guido van Rossumd8faa362007-04-27 19:54:29 +0000446 parser.feed('<!DOCTYPE doc [\n')
447 parser.feed(' <!ENTITY img SYSTEM "expat.gif" NDATA GIF>\n')
448 parser.feed(' <!NOTATION GIF PUBLIC "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN">\n')
449 parser.feed(']>\n')
450 parser.feed('<doc></doc>')
451 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000452
Guido van Rossumd8faa362007-04-27 19:54:29 +0000453 self.assertEquals(handler._notations,
454 [("GIF", "-//CompuServe//NOTATION Graphics Interchange Format 89a//EN", None)])
455 self.assertEquals(handler._entities, [("img", None, "expat.gif", "GIF")])
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000456
Guido van Rossumd8faa362007-04-27 19:54:29 +0000457 # ===== EntityResolver support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000458
Guido van Rossumd8faa362007-04-27 19:54:29 +0000459 class TestEntityResolver:
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000460
Guido van Rossumd8faa362007-04-27 19:54:29 +0000461 def resolveEntity(self, publicId, systemId):
462 inpsrc = InputSource()
463 inpsrc.setByteStream(StringIO("<entity/>"))
464 return inpsrc
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000465
Guido van Rossumd8faa362007-04-27 19:54:29 +0000466 def test_expat_entityresolver(self):
467 parser = create_parser()
468 parser.setEntityResolver(self.TestEntityResolver())
469 result = StringIO()
470 parser.setContentHandler(XMLGenerator(result))
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000471
Guido van Rossumd8faa362007-04-27 19:54:29 +0000472 parser.feed('<!DOCTYPE doc [\n')
473 parser.feed(' <!ENTITY test SYSTEM "whatever">\n')
474 parser.feed(']>\n')
475 parser.feed('<doc>&test;</doc>')
476 parser.close()
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000477
Guido van Rossumd8faa362007-04-27 19:54:29 +0000478 self.assertEquals(result.getvalue(), start +
479 "<doc><entity></entity></doc>")
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000480
Guido van Rossumd8faa362007-04-27 19:54:29 +0000481 # ===== Attributes support
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000482
Guido van Rossumd8faa362007-04-27 19:54:29 +0000483 class AttrGatherer(ContentHandler):
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000484
Guido van Rossumd8faa362007-04-27 19:54:29 +0000485 def startElement(self, name, attrs):
486 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000487
Guido van Rossumd8faa362007-04-27 19:54:29 +0000488 def startElementNS(self, name, qname, attrs):
489 self._attrs = attrs
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000490
Guido van Rossumd8faa362007-04-27 19:54:29 +0000491 def test_expat_attrs_empty(self):
492 parser = create_parser()
493 gather = self.AttrGatherer()
494 parser.setContentHandler(gather)
495
496 parser.feed("<doc/>")
497 parser.close()
498
499 self.verify_empty_attrs(gather._attrs)
500
501 def test_expat_attrs_wattr(self):
502 parser = create_parser()
503 gather = self.AttrGatherer()
504 parser.setContentHandler(gather)
505
506 parser.feed("<doc attr='val'/>")
507 parser.close()
508
509 self.verify_attrs_wattr(gather._attrs)
510
511 def test_expat_nsattrs_empty(self):
512 parser = create_parser(1)
513 gather = self.AttrGatherer()
514 parser.setContentHandler(gather)
515
516 parser.feed("<doc/>")
517 parser.close()
518
519 self.verify_empty_nsattrs(gather._attrs)
520
521 def test_expat_nsattrs_wattr(self):
522 parser = create_parser(1)
523 gather = self.AttrGatherer()
524 parser.setContentHandler(gather)
525
526 parser.feed("<doc xmlns:ns='%s' ns:attr='val'/>" % ns_uri)
527 parser.close()
528
529 attrs = gather._attrs
530
531 self.assertEquals(attrs.getLength(), 1)
532 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
533 self.assertTrue((attrs.getQNames() == [] or
534 attrs.getQNames() == ["ns:attr"]))
535 self.assertEquals(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +0000536 self.assertIn((ns_uri, "attr"), attrs)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000537 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
538 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
539 self.assertEquals(list(attrs.items()), [((ns_uri, "attr"), "val")])
540 self.assertEquals(list(attrs.values()), ["val"])
541 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
542 self.assertEquals(attrs[(ns_uri, "attr")], "val")
543
544 # ===== InputSource support
545
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000546 def test_expat_inpsource_filename(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000547 parser = create_parser()
548 result = StringIO()
549 xmlgen = XMLGenerator(result)
550
551 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000552 parser.parse(TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000553
554 self.assertEquals(result.getvalue(), xml_test_out)
555
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000556 def test_expat_inpsource_sysid(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000557 parser = create_parser()
558 result = StringIO()
559 xmlgen = XMLGenerator(result)
560
561 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000562 parser.parse(InputSource(TEST_XMLFILE))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000563
564 self.assertEquals(result.getvalue(), xml_test_out)
565
566 def test_expat_inpsource_stream(self):
567 parser = create_parser()
568 result = StringIO()
569 xmlgen = XMLGenerator(result)
570
571 parser.setContentHandler(xmlgen)
572 inpsrc = InputSource()
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000573 inpsrc.setByteStream(open(TEST_XMLFILE))
Guido van Rossumd8faa362007-04-27 19:54:29 +0000574 parser.parse(inpsrc)
575
576 self.assertEquals(result.getvalue(), xml_test_out)
577
578 # ===== IncrementalParser support
579
580 def test_expat_incremental(self):
581 result = StringIO()
582 xmlgen = XMLGenerator(result)
583 parser = create_parser()
584 parser.setContentHandler(xmlgen)
585
586 parser.feed("<doc>")
587 parser.feed("</doc>")
588 parser.close()
589
590 self.assertEquals(result.getvalue(), start + "<doc></doc>")
591
592 def test_expat_incremental_reset(self):
593 result = StringIO()
594 xmlgen = XMLGenerator(result)
595 parser = create_parser()
596 parser.setContentHandler(xmlgen)
597
598 parser.feed("<doc>")
599 parser.feed("text")
600
601 result = StringIO()
602 xmlgen = XMLGenerator(result)
603 parser.setContentHandler(xmlgen)
604 parser.reset()
605
606 parser.feed("<doc>")
607 parser.feed("text")
608 parser.feed("</doc>")
609 parser.close()
610
611 self.assertEquals(result.getvalue(), start + "<doc>text</doc>")
612
613 # ===== Locator support
614
615 def test_expat_locator_noinfo(self):
616 result = StringIO()
617 xmlgen = XMLGenerator(result)
618 parser = create_parser()
619 parser.setContentHandler(xmlgen)
620
621 parser.feed("<doc>")
622 parser.feed("</doc>")
623 parser.close()
624
625 self.assertEquals(parser.getSystemId(), None)
626 self.assertEquals(parser.getPublicId(), None)
627 self.assertEquals(parser.getLineNumber(), 1)
628
Benjamin Petersona7f4f5a2008-09-04 02:22:52 +0000629 def test_expat_locator_withinfo(self):
Guido van Rossumd8faa362007-04-27 19:54:29 +0000630 result = StringIO()
631 xmlgen = XMLGenerator(result)
632 parser = create_parser()
633 parser.setContentHandler(xmlgen)
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000634 parser.parse(TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000635
Florent Xiclunaf15351d2010-03-13 23:24:31 +0000636 self.assertEquals(parser.getSystemId(), TEST_XMLFILE)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000637 self.assertEquals(parser.getPublicId(), None)
Lars Gustäbel2fc52942000-10-24 15:35:07 +0000638
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000639
640# ===========================================================================
641#
642# error reporting
643#
644# ===========================================================================
645
Guido van Rossumd8faa362007-04-27 19:54:29 +0000646class ErrorReportingTest(unittest.TestCase):
647 def test_expat_inpsource_location(self):
648 parser = create_parser()
649 parser.setContentHandler(ContentHandler()) # do nothing
650 source = InputSource()
651 source.setByteStream(StringIO("<foo bar foobar>")) #ill-formed
652 name = "a file name"
653 source.setSystemId(name)
654 try:
655 parser.parse(source)
656 self.fail()
657 except SAXException as e:
658 self.assertEquals(e.getSystemId(), name)
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000659
Guido van Rossumd8faa362007-04-27 19:54:29 +0000660 def test_expat_incomplete(self):
661 parser = create_parser()
662 parser.setContentHandler(ContentHandler()) # do nothing
663 self.assertRaises(SAXParseException, parser.parse, StringIO("<foo>"))
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000664
Guido van Rossumd8faa362007-04-27 19:54:29 +0000665 def test_sax_parse_exception_str(self):
666 # pass various values from a locator to the SAXParseException to
667 # make sure that the __str__() doesn't fall apart when None is
668 # passed instead of an integer line and column number
669 #
670 # use "normal" values for the locator:
671 str(SAXParseException("message", None,
672 self.DummyLocator(1, 1)))
673 # use None for the line number:
674 str(SAXParseException("message", None,
675 self.DummyLocator(None, 1)))
676 # use None for the column number:
677 str(SAXParseException("message", None,
678 self.DummyLocator(1, None)))
679 # use None for both:
680 str(SAXParseException("message", None,
681 self.DummyLocator(None, None)))
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000682
Guido van Rossumd8faa362007-04-27 19:54:29 +0000683 class DummyLocator:
684 def __init__(self, lineno, colno):
685 self._lineno = lineno
686 self._colno = colno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000687
Guido van Rossumd8faa362007-04-27 19:54:29 +0000688 def getPublicId(self):
689 return "pubid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000690
Guido van Rossumd8faa362007-04-27 19:54:29 +0000691 def getSystemId(self):
692 return "sysid"
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000693
Guido van Rossumd8faa362007-04-27 19:54:29 +0000694 def getLineNumber(self):
695 return self._lineno
Fred Drake6fd0b0d2004-03-20 08:15:30 +0000696
Guido van Rossumd8faa362007-04-27 19:54:29 +0000697 def getColumnNumber(self):
698 return self._colno
Martin v. Löwis80670bc2000-10-06 21:13:23 +0000699
Lars Gustäbelab647872000-09-24 18:40:52 +0000700# ===========================================================================
701#
702# xmlreader tests
703#
704# ===========================================================================
705
Guido van Rossumd8faa362007-04-27 19:54:29 +0000706class XmlReaderTest(XmlTestBase):
Lars Gustäbelab647872000-09-24 18:40:52 +0000707
Guido van Rossumd8faa362007-04-27 19:54:29 +0000708 # ===== AttributesImpl
709 def test_attrs_empty(self):
710 self.verify_empty_attrs(AttributesImpl({}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000711
Guido van Rossumd8faa362007-04-27 19:54:29 +0000712 def test_attrs_wattr(self):
713 self.verify_attrs_wattr(AttributesImpl({"attr" : "val"}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000714
Guido van Rossumd8faa362007-04-27 19:54:29 +0000715 def test_nsattrs_empty(self):
716 self.verify_empty_nsattrs(AttributesNSImpl({}, {}))
Lars Gustäbelab647872000-09-24 18:40:52 +0000717
Guido van Rossumd8faa362007-04-27 19:54:29 +0000718 def test_nsattrs_wattr(self):
719 attrs = AttributesNSImpl({(ns_uri, "attr") : "val"},
720 {(ns_uri, "attr") : "ns:attr"})
Fred Drake004d5e62000-10-23 17:22:08 +0000721
Guido van Rossumd8faa362007-04-27 19:54:29 +0000722 self.assertEquals(attrs.getLength(), 1)
723 self.assertEquals(attrs.getNames(), [(ns_uri, "attr")])
724 self.assertEquals(attrs.getQNames(), ["ns:attr"])
725 self.assertEquals(len(attrs), 1)
Benjamin Peterson577473f2010-01-19 00:09:57 +0000726 self.assertIn((ns_uri, "attr"), attrs)
Guido van Rossumd8faa362007-04-27 19:54:29 +0000727 self.assertEquals(list(attrs.keys()), [(ns_uri, "attr")])
728 self.assertEquals(attrs.get((ns_uri, "attr")), "val")
729 self.assertEquals(attrs.get((ns_uri, "attr"), 25), "val")
730 self.assertEquals(list(attrs.items()), [((ns_uri, "attr"), "val")])
731 self.assertEquals(list(attrs.values()), ["val"])
732 self.assertEquals(attrs.getValue((ns_uri, "attr")), "val")
733 self.assertEquals(attrs.getValueByQName("ns:attr"), "val")
734 self.assertEquals(attrs.getNameByQName("ns:attr"), (ns_uri, "attr"))
735 self.assertEquals(attrs[(ns_uri, "attr")], "val")
736 self.assertEquals(attrs.getQNameByName((ns_uri, "attr")), "ns:attr")
Fred Drake004d5e62000-10-23 17:22:08 +0000737
Lars Gustäbelab647872000-09-24 18:40:52 +0000738
Guido van Rossumd8faa362007-04-27 19:54:29 +0000739 # During the development of Python 2.5, an attempt to move the "xml"
740 # package implementation to a new package ("xmlcore") proved painful.
741 # The goal of this change was to allow applications to be able to
742 # obtain and rely on behavior in the standard library implementation
743 # of the XML support without needing to be concerned about the
744 # availability of the PyXML implementation.
745 #
746 # While the existing import hackery in Lib/xml/__init__.py can cause
747 # PyXML's _xmlpus package to supplant the "xml" package, that only
748 # works because either implementation uses the "xml" package name for
749 # imports.
750 #
751 # The move resulted in a number of problems related to the fact that
752 # the import machinery's "package context" is based on the name that's
753 # being imported rather than the __name__ of the actual package
754 # containment; it wasn't possible for the "xml" package to be replaced
755 # by a simple module that indirected imports to the "xmlcore" package.
756 #
757 # The following two tests exercised bugs that were introduced in that
758 # attempt. Keeping these tests around will help detect problems with
759 # other attempts to provide reliable access to the standard library's
760 # implementation of the XML support.
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000761
Guido van Rossumd8faa362007-04-27 19:54:29 +0000762 def test_sf_1511497(self):
763 # Bug report: http://www.python.org/sf/1511497
764 import sys
765 old_modules = sys.modules.copy()
766 for modname in list(sys.modules.keys()):
767 if modname.startswith("xml."):
768 del sys.modules[modname]
769 try:
770 import xml.sax.expatreader
771 module = xml.sax.expatreader
772 self.assertEquals(module.__name__, "xml.sax.expatreader")
773 finally:
774 sys.modules.update(old_modules)
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000775
Guido van Rossumd8faa362007-04-27 19:54:29 +0000776 def test_sf_1513611(self):
777 # Bug report: http://www.python.org/sf/1513611
778 sio = StringIO("invalid")
779 parser = make_parser()
780 from xml.sax import SAXParseException
781 self.assertRaises(SAXParseException, parser.parse, sio)
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000782
Lars Gustäbel96753b32000-09-24 12:24:24 +0000783
Christian Heimesbbe741d2008-03-28 10:53:29 +0000784def test_main():
Guido van Rossumd8faa362007-04-27 19:54:29 +0000785 run_unittest(MakeParserTest,
786 SaxutilsTest,
787 XmlgenTest,
788 ExpatReaderTest,
789 ErrorReportingTest,
790 XmlReaderTest)
Lars Gustäbelb7536d52000-09-24 18:53:56 +0000791
Guido van Rossumd8faa362007-04-27 19:54:29 +0000792if __name__ == "__main__":
Christian Heimesbbe741d2008-03-28 10:53:29 +0000793 test_main()