Source Code for Module lxml.tests.test_unicode

 1  # -*- coding: utf-8 -*- 
 2  import unittest, doctest 
 3   
 4  from common_imports import StringIO, etree, SillyFileLike, HelperTestCase 
 5   
 6  ascii_uni = u'a' 
 7   
 8  klingon = u"\uF8D2" # not valid for XML names 
 9   
10  invalid_tag = "test" + klingon 
11   
12  uni = u'Ã\u0680\u3120' # some non-ASCII characters 
13   
14  uxml = u"<test><title>test á\u3120</title><h1>page á\u3120 title</h1></test>" 
15   
16 -class UnicodeTestCase(HelperTestCase): 
17 -    def test_unicode_xml(self): 
18          tree = etree.XML(u'<p>%s</p>' % uni) 
19          self.assertEquals(uni, tree.text) 
20   
21 -    def test_unicode_xml_broken(self): 
22          uxml = u'<?xml version="1.0" encoding="UTF-8"?>' + \ 
23                 u'<p>%s</p>' % uni 
24          self.assertRaises(ValueError, etree.XML, uxml) 
25   
26 -    def test_unicode_tag(self): 
27          el = etree.Element(uni) 
28          self.assertEquals(uni, el.tag) 
29   
30 -    def test_unicode_tag_invalid(self): 
31          # sadly, Klingon is not well-formed 
32          self.assertRaises(ValueError, etree.Element, invalid_tag) 
33   
34 -    def test_unicode_nstag(self): 
35          tag = u"{%s}%s" % (uni, uni) 
36          el = etree.Element(tag) 
37          self.assertEquals(tag, el.tag) 
38   
39 -    def test_unicode_nstag_invalid(self): 
40          # sadly, Klingon is not well-formed 
41          tag = u"{%s}%s" % (uni, invalid_tag) 
42          self.assertRaises(ValueError, etree.Element, tag) 
43   
44 -    def test_unicode_qname(self): 
45          qname = etree.QName(uni, uni) 
46          tag = u"{%s}%s" % (uni, uni) 
47          self.assertEquals(qname.text, tag) 
48          self.assertEquals(unicode(qname), tag) 
49   
50 -    def test_unicode_qname_invalid(self): 
51          self.assertRaises(ValueError, etree.QName, invalid_tag) 
52   
53 -    def test_unicode_attr(self): 
54          el = etree.Element('foo', {'bar': uni}) 
55          self.assertEquals(uni, el.attrib['bar']) 
56   
57 -    def test_unicode_comment(self): 
58          el = etree.Comment(uni) 
59          self.assertEquals(uni, el.text) 
60   
61 -    def test_unicode_parse_stringio(self): 
62          el = etree.parse(StringIO(u'<p>%s</p>' % uni)).getroot() 
63          self.assertEquals(uni, el.text) 
64   
65  ##     def test_parse_fileobject_unicode(self): 
66  ##         # parse unicode from unamed file object (not support by ElementTree) 
67  ##         f = SillyFileLike(uxml) 
68  ##         root = etree.parse(f).getroot() 
69  ##         self.assertEquals(unicode(etree.tostring(root, 'UTF-8'), 'UTF-8'), 
70  ##                           uxml) 
71   
72 -def test_suite(): 
73      suite = unittest.TestSuite() 
74      suite.addTests([unittest.makeSuite(UnicodeTestCase)]) 
75      return suite 
76