summaryrefslogtreecommitdiff
path: root/src/lxml/tests/test_unicode.py
diff options
context:
space:
mode:
authorStefan Behnel <stefan_ml@behnel.de>2014-01-02 12:34:51 +0100
committerStefan Behnel <stefan_ml@behnel.de>2014-01-02 12:34:51 +0100
commit0c339d4277b6857a358c06f071bad6864cb84fef (patch)
tree1e920b21be2a3cd32b419598712c42304ac442ad /src/lxml/tests/test_unicode.py
parent3043a9e728c1147ffa525a67ce427688475d01d1 (diff)
downloadpython-lxml-0c339d4277b6857a358c06f071bad6864cb84fef.tar.gz
fix and extend Unicode parsing tests
Diffstat (limited to 'src/lxml/tests/test_unicode.py')
-rw-r--r--src/lxml/tests/test_unicode.py20
1 files changed, 10 insertions, 10 deletions
diff --git a/src/lxml/tests/test_unicode.py b/src/lxml/tests/test_unicode.py
index 55777d79..0ae482b2 100644
--- a/src/lxml/tests/test_unicode.py
+++ b/src/lxml/tests/test_unicode.py
@@ -13,11 +13,11 @@ try:
except NameError:
unicode = str
-ascii_uni = _str('a')
+ascii_uni = _bytes('a').decode('utf8')
klingon = _bytes("\\uF8D2").decode("unicode_escape") # not valid for XML names
-invalid_tag = _str("test") + klingon
+invalid_tag = _bytes("test").decode('utf8') + klingon
uni = _bytes('\\xc3\\u0680\\u3120').decode("unicode_escape") # some non-ASCII characters
@@ -26,12 +26,12 @@ uxml = _bytes("<test><title>test \\xc3\\xa1\\u3120</title><h1>page \\xc3\\xa1\\u
class UnicodeTestCase(HelperTestCase):
def test_unicode_xml(self):
- tree = etree.XML(_str('<p>%s</p>') % uni)
+ tree = etree.XML('<p>%s</p>' % uni)
self.assertEqual(uni, tree.text)
def test_unicode_xml_broken(self):
- uxml = _str('<?xml version="1.0" encoding="UTF-8"?>') + \
- _str('<p>%s</p>') % uni
+ uxml = ('<?xml version="1.0" encoding="UTF-8"?>' +
+ '<p>%s</p>' % uni)
self.assertRaises(ValueError, etree.XML, uxml)
def test_unicode_tag(self):
@@ -43,23 +43,23 @@ class UnicodeTestCase(HelperTestCase):
self.assertRaises(ValueError, etree.Element, invalid_tag)
def test_unicode_nstag(self):
- tag = _str("{http://abc/}%s") % uni
+ tag = "{http://abc/}%s" % uni
el = etree.Element(tag)
self.assertEqual(tag, el.tag)
def test_unicode_ns_invalid(self):
# namespace URIs must conform to RFC 3986
- tag = _str("{http://%s/}abc") % uni
+ tag = "{http://%s/}abc" % uni
self.assertRaises(ValueError, etree.Element, tag)
def test_unicode_nstag_invalid(self):
# sadly, Klingon is not well-formed
- tag = _str("{http://abc/}%s") % invalid_tag
+ tag = "{http://abc/}%s" % invalid_tag
self.assertRaises(ValueError, etree.Element, tag)
def test_unicode_qname(self):
qname = etree.QName(uni, uni)
- tag = _str("{%s}%s") % (uni, uni)
+ tag = "{%s}%s" % (uni, uni)
self.assertEqual(qname.text, tag)
self.assertEqual(unicode(qname), tag)
@@ -75,7 +75,7 @@ class UnicodeTestCase(HelperTestCase):
self.assertEqual(uni, el.text)
def test_unicode_parse_stringio(self):
- el = etree.parse(StringIO(_str('<p>%s</p>') % uni)).getroot()
+ el = etree.parse(StringIO('<p>%s</p>' % uni)).getroot()
self.assertEqual(uni, el.text)
## def test_parse_fileobject_unicode(self):