def visit_method(self, dirname, names):
d = uri.URI.from_path(os.path.join(dirname, self.sys_path(os.curdir)))
c = sys.getfilesystemencoding()
for name in names:
# on windows, c will be mbcs even if strings are unicode
bname = name.encode(c) if is_unicode(name) else name
if bname.startswith(b'??'):
logging.warn("8-bit path tests limited to ASCII file names "
"by %s encoding", c)
continue
join_match = os.path.join(dirname, name)
if is_unicode(name):
seg_name = uri.escape_data(
name.encode(c), uri.is_path_segment_reserved)
else:
seg_name = uri.escape_data(name, uri.is_path_segment_reserved)
u = uri.URI(seg_name)
u = u.resolve(d)
self.assertTrue(isinstance(u, uri.FileURL))
joined = u.get_pathname()
if isinstance(join_match, bytes) and is_unicode(joined):
# if we're walking in 8-bit mode we need to downgrade to
# compare
joined = joined.encode(c)
self.assertTrue(joined == join_match,
"Joined pathnames mismatch:\n%s\n%s" %
(joined, join_match))
def visit_method(self, dirname, names):
d = uri.URI.from_path(os.path.join(dirname, os.curdir))
c = sys.getfilesystemencoding()
for name in names:
if name.startswith('??'):
logging.warn("8-bit path tests limited to ASCII file names "
"by %s encoding", c)
continue
join_match = os.path.join(dirname, name)
if is_unicode(name):
seg_name = uri.escape_data(
name.encode(c), uri.is_path_segment_reserved)
else:
seg_name = uri.escape_data(name, uri.is_path_segment_reserved)
u = uri.URI(seg_name)
u = u.resolve(d)
self.assertTrue(isinstance(u, uri.FileURL))
joined = u.get_pathname()
if isinstance(join_match, bytes) and is_unicode(joined):
# if we're walking in 8-bit mode we need to downgrade to
# compare
joined = joined.encode(c)
self.assertTrue(joined == join_match,
"Joined pathnames mismatch:\n%s\n%s" %
(joined, join_match))
def test_constructor(self):
u = uri.URI(SIMPLE_EXAMPLE)
self.assertTrue(isinstance(u, uri.URI))
self.assertTrue(str(u) == SIMPLE_EXAMPLE)
self.assertTrue(is_unicode(u.octets),
"octets must be a character string")
if py2:
self.assertTrue(to_text(u) == SIMPLE_EXAMPLE)
try:
u = uri.URI(LIST_EXAMPLE)
# we don't support this type of thing any more
# self.assertTrue(str(u)==SIMPLE_EXAMPLE,"Simple from list")
except uri.URIException:
pass
u = uri.URI.from_octets(u8(b'\xe8\x8b\xb1\xe5\x9b\xbd.xml'))
self.assertTrue(
str(u) == '%E8%8B%B1%E5%9B%BD.xml', "Unicode example: %s" % str(u))
self.assertTrue(is_unicode(u.octets),
"octets must be a character string")
try:
u = uri.URI.from_octets(u8(b'\xe8\x8b\xb1\xe5\x9b\xbd.xml'),
strict=True)
self.fail("strict mode requires %-encoding")
except uri.URIException:
pass
# binary string must be US-ASCII clean
try:
u = uri.URI.from_octets(b'Caf\xe9')
self.fail("binary string must be US-ASCII")
except UnicodeDecodeError:
pass
# but URI-encoded is OK even if it is binary
u = uri.URI.from_octets(b'Caf%E9')
self.assertTrue(is_unicode(u.octets),
"octets must be a character string")
def test_constructor(self):
u = uri.URI(SIMPLE_EXAMPLE)
self.assertTrue(isinstance(u, uri.URI))
self.assertTrue(str(u) == SIMPLE_EXAMPLE)
self.assertTrue(is_unicode(u.octets),
"octets must be a character string")
if py2:
self.assertTrue(to_text(u) == SIMPLE_EXAMPLE)
try:
u = uri.URI(LIST_EXAMPLE)
# we don't support this type of thing any more
# self.assertTrue(str(u)==SIMPLE_EXAMPLE,"Simple from list")
except uri.URIException:
pass
u = uri.URI.from_octets(u8(b'\xe8\x8b\xb1\xe5\x9b\xbd.xml'))
self.assertTrue(
str(u) == '%E8%8B%B1%E5%9B%BD.xml', "Unicode example: %s" % str(u))
self.assertTrue(is_unicode(u.octets),
"octets must be a character string")
try:
u = uri.URI.from_octets(u8(b'\xe8\x8b\xb1\xe5\x9b\xbd.xml'),
strict=True)
self.fail("strict mode requires %-encoding")
except uri.URIException:
pass
# binary string must be US-ASCII clean
try:
u = uri.URI.from_octets(b'Caf\xe9')
self.fail("binary string must be US-ASCII")
except UnicodeDecodeError:
pass
# but URI-encoded is OK even if it is binary
u = uri.URI.from_octets(b'Caf%E9')
self.assertTrue(is_unicode(u.octets),
"octets must be a character string")
def test_text(self):
data = "hello"
udata = u"hello"
bdata = b"hello"
xdata = ['hello']
self.assertTrue(py2.is_text(data))
self.assertTrue(py2.is_text(udata))
if sys.version_info[0] < 3:
self.assertTrue(py2.is_text(bdata))
else:
self.assertFalse(py2.is_text(bdata))
self.assertFalse(py2.is_text(xdata))
if sys.version_info[0] < 3:
self.assertFalse(py2.is_unicode(data))
else:
self.assertTrue(py2.is_unicode(data))
self.assertTrue(py2.is_unicode(udata))
self.assertFalse(py2.is_unicode(bdata))
self.assertFalse(py2.is_unicode(xdata))
# force text forces strings to be text
self.assertTrue(data == py2.force_text(data))
self.assertTrue(isinstance(py2.force_text(data), type(u"")))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_text(data), type("")))
else:
self.assertTrue(isinstance(py2.force_text(data), type("")))
self.assertTrue(isinstance(py2.force_text(data), type(u"")))
self.assertTrue(data == py2.force_text(udata))
self.assertTrue(isinstance(py2.force_text(udata), type(u"")))
if sys.version_info[0] < 3:
# force_text will not throw an error in python 2
pass
else:
try:
py2.force_text(bdata)
self.fail("force_text(bytes)")
except TypeError:
pass
# this must work in both python 2 and 3 to prevent accidental
# conversion to string.
try:
py2.force_text(xdata)
self.fail("force_text(object)")
except TypeError:
pass
# conversion to text
self.assertTrue(data == py2.to_text(data))
self.assertTrue(isinstance(py2.to_text(data), type(u"")))
self.assertTrue(data == py2.to_text(udata))
self.assertTrue(isinstance(py2.to_text(udata), type(u"")))
self.assertTrue(data == py2.to_text(bdata))
self.assertTrue(isinstance(py2.to_text(bdata), type(u"")))
if sys.version_info[0] < 3:
self.assertTrue(u"['hello']" == py2.to_text(xdata))
else:
self.assertTrue("['hello']" == py2.to_text(xdata))
def test_constructor(self):
e = structures.XMLEntity(b"<hello>")
self.assertTrue(e.line_num == 1)
self.assertTrue(e.line_pos == 1)
self.assertTrue(is_unicode(e.the_char) and e.the_char == '<')
e = structures.XMLEntity(ul("<hello>"))
self.assertTrue(is_unicode(e.the_char) and e.the_char == '<')
e = structures.XMLEntity(StringIO(ul("<hello>")))
self.assertTrue(e.line_num == 1)
self.assertTrue(e.line_pos == 1)
self.assertTrue(is_unicode(e.the_char) and e.the_char == '<')
def test_constructor(self):
e = structures.XMLEntity(b"<hello>")
self.assertTrue(e.line_num == 1)
self.assertTrue(e.line_pos == 1)
self.assertTrue(is_unicode(e.the_char) and e.the_char == '<')
e = structures.XMLEntity(ul("<hello>"))
self.assertTrue(is_unicode(e.the_char) and e.the_char == '<')
e = structures.XMLEntity(StringIO(ul("<hello>")))
self.assertTrue(e.line_num == 1)
self.assertTrue(e.line_pos == 1)
self.assertTrue(is_unicode(e.the_char) and e.the_char == '<')
def test_constructor(self):
try:
u = urn.URN()
self.fail("empty constructor")
except ValueError:
pass
try:
u = urn.URN(nid="foo")
self.fail("namespace-specific string missing")
except ValueError:
pass
u = urn.URN(nid="foo", nss="a123,456")
self.assertTrue(isinstance(u, urn.URN))
self.assertTrue(str(u) == "urn:foo:a123,456")
self.assertTrue(is_unicode(u.nid))
self.assertTrue(is_unicode(u.nss))
u = uri.URI.from_octets('urn:foo:a123,456')
self.assertTrue(isinstance(u, urn.URN))
self.assertTrue(str(u) == 'urn:foo:a123,456')
def test_unicode_pathnames(self):
if isinstance(self.data_path, bytes):
c = sys.getfilesystemencoding()
data_path = self.data_path.decode(c)
else:
data_path = self.data_path
base = uri.URI.from_path(data_path)
if os.path.supports_unicode_filenames:
data_path2 = base.get_pathname()
self.assertTrue(is_unicode(data_path2),
"Expected get_pathname to return unicode")
self.assertTrue(data_path2 == data_path,
ul("Expected %s found %s") %
(data_path, data_path2))
# os.path.walk(data_path,self.visit_method,None)
for dirpath, dirnames, filenames in os.walk(data_path):
self.visit_method(dirpath, filenames)
else:
data_path2 = base.get_pathname()
self.assertTrue(isinstance(data_path2, bytes),
"Expected get_pathname to return string")
logging.warn("os.path.supports_unicode_filenames is False "
"(skipped unicode path tests)")
def test_unicode_pathnames(self):
if isinstance(self.data_path, bytes):
c = sys.getfilesystemencoding()
data_path = self.data_path.encode(c)
else:
data_path = self.data_path
base = uri.URI.from_path(data_path)
if os.path.supports_unicode_filenames:
data_path2 = base.get_pathname()
self.assertTrue(is_unicode(data_path2),
"Expected get_pathname to return unicode")
self.assertTrue(data_path2 == data_path,
ul("Expected %s found %s") %
(data_path, data_path2))
# os.path.walk(data_path,self.visit_method,None)
for dirpath, dirnames, filenames in os.walk(data_path):
self.visit_method(dirpath, filenames)
else:
data_path2 = base.get_pathname()
self.assertTrue(isinstance(data_path2, bytes),
"Expected get_pathname to return string")
logging.warn("os.path.supports_unicode_filenames is False "
"(skipped unicode path tests)")
def sys_path(self, path):
if is_unicode(path) and not os.path.supports_unicode_filenames:
return path.encode(sys.getfilesystemencoding())
else:
return path
def test_text(self):
data = "hello"
udata = u"hello"
bdata = b"hello"
xdata = ['hello']
self.assertTrue(py2.is_string(data))
self.assertTrue(py2.is_string(udata))
self.assertTrue(py2.is_string(bdata))
self.assertFalse(py2.is_string(xdata))
self.assertTrue(py2.is_text(data))
self.assertTrue(py2.is_text(udata))
if sys.version_info[0] < 3:
self.assertTrue(py2.is_text(bdata))
else:
self.assertFalse(py2.is_text(bdata))
self.assertFalse(py2.is_text(xdata))
if sys.version_info[0] < 3:
self.assertFalse(py2.is_unicode(data))
else:
self.assertTrue(py2.is_unicode(data))
self.assertTrue(py2.is_unicode(udata))
self.assertFalse(py2.is_unicode(bdata))
self.assertFalse(py2.is_unicode(xdata))
# force text forces strings to be text
self.assertTrue(data == py2.force_text(data))
self.assertTrue(isinstance(py2.force_text(data), type(u"")))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_text(data), type("")))
else:
self.assertTrue(isinstance(py2.force_text(data), type("")))
self.assertTrue(data == py2.force_text(udata))
self.assertTrue(isinstance(py2.force_text(udata), type(u"")))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_text(udata), type("")))
else:
self.assertTrue(isinstance(py2.force_text(udata), type("")))
if sys.version_info[0] < 3:
# force_text will not throw an error in python 2
pass
else:
try:
py2.force_text(bdata)
self.fail("force_text(bytes)")
except TypeError:
pass
# this must work in both python 2 and 3 to prevent accidental
# conversion to string.
try:
py2.force_text(xdata)
self.fail("force_text(object)")
except TypeError:
pass
# force ascii forces strings to be ascii text
self.assertTrue(data == py2.force_ascii(data))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_ascii(data), type(u"")))
else:
self.assertTrue(isinstance(py2.force_ascii(data), type(u"")))
self.assertTrue(isinstance(py2.force_ascii(data), type("")))
self.assertTrue(data == py2.force_ascii(udata))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_ascii(udata), type(u"")))
else:
self.assertTrue(isinstance(py2.force_ascii(udata), type(u"")))
self.assertTrue(isinstance(py2.force_ascii(udata), type("")))
if sys.version_info[0] < 3:
self.assertTrue(bdata == py2.force_ascii(bdata))
self.assertFalse(isinstance(py2.force_ascii(bdata), type(u"")))
else:
# can't compare different types in Python 3
self.assertFalse(bdata == py2.force_ascii(bdata))
self.assertTrue(isinstance(py2.force_ascii(bdata), type(u"")))
self.assertTrue(isinstance(py2.force_ascii(bdata), type("")))
# this must work in both python 2 and 3 to prevent accidental
# conversion to string.
try:
py2.force_ascii(xdata)
self.fail("force_ascii(object)")
except TypeError:
pass
# conversion to text
self.assertTrue(data == py2.to_text(data))
self.assertTrue(isinstance(py2.to_text(data), type(u"")))
self.assertTrue(data == py2.to_text(udata))
self.assertTrue(isinstance(py2.to_text(udata), type(u"")))
self.assertTrue(data == py2.to_text(bdata))
self.assertTrue(isinstance(py2.to_text(bdata), type(u"")))
if sys.version_info[0] < 3:
self.assertTrue(u"['hello']" == py2.to_text(xdata))
else:
self.assertTrue("['hello']" == py2.to_text(xdata))
# check the empty text constant:
self.assertTrue(isinstance(py2.uempty, type(u"")))
self.assertFalse(py2.uempty)
self.assertTrue(len(py2.uempty) == 0)
def test_text(self):
data = "hello"
udata = u"hello"
bdata = b"hello"
xdata = ['hello']
self.assertTrue(py2.is_string(data))
self.assertTrue(py2.is_string(udata))
self.assertTrue(py2.is_string(bdata))
self.assertFalse(py2.is_string(xdata))
self.assertTrue(py2.is_text(data))
self.assertTrue(py2.is_text(udata))
if sys.version_info[0] < 3:
self.assertTrue(py2.is_text(bdata))
else:
self.assertFalse(py2.is_text(bdata))
self.assertFalse(py2.is_text(xdata))
if sys.version_info[0] < 3:
self.assertFalse(py2.is_unicode(data))
else:
self.assertTrue(py2.is_unicode(data))
self.assertTrue(py2.is_unicode(udata))
self.assertFalse(py2.is_unicode(bdata))
self.assertFalse(py2.is_unicode(xdata))
# force text forces strings to be text
self.assertTrue(data == py2.force_text(data))
self.assertTrue(isinstance(py2.force_text(data), type(u"")))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_text(data), type("")))
else:
self.assertTrue(isinstance(py2.force_text(data), type("")))
self.assertTrue(data == py2.force_text(udata))
self.assertTrue(isinstance(py2.force_text(udata), type(u"")))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_text(udata), type("")))
else:
self.assertTrue(isinstance(py2.force_text(udata), type("")))
if sys.version_info[0] < 3:
# force_text will not throw an error in python 2
pass
else:
try:
py2.force_text(bdata)
self.fail("force_text(bytes)")
except TypeError:
pass
# this must work in both python 2 and 3 to prevent accidental
# conversion to string.
try:
py2.force_text(xdata)
self.fail("force_text(object)")
except TypeError:
pass
# force ascii forces strings to be ascii text
self.assertTrue(data == py2.force_ascii(data))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_ascii(data), type(u"")))
else:
self.assertTrue(isinstance(py2.force_ascii(data), type(u"")))
self.assertTrue(isinstance(py2.force_ascii(data), type("")))
self.assertTrue(data == py2.force_ascii(udata))
if sys.version_info[0] < 3:
self.assertFalse(isinstance(py2.force_ascii(udata), type(u"")))
else:
self.assertTrue(isinstance(py2.force_ascii(udata), type(u"")))
self.assertTrue(isinstance(py2.force_ascii(udata), type("")))
if sys.version_info[0] < 3:
self.assertTrue(bdata == py2.force_ascii(bdata))
self.assertFalse(isinstance(py2.force_ascii(bdata), type(u"")))
else:
# can't compare different types in Python 3
self.assertFalse(bdata == py2.force_ascii(bdata))
self.assertTrue(isinstance(py2.force_ascii(bdata), type(u"")))
self.assertTrue(isinstance(py2.force_ascii(bdata), type("")))
# this must work in both python 2 and 3 to prevent accidental
# conversion to string.
try:
py2.force_ascii(xdata)
self.fail("force_ascii(object)")
except TypeError:
pass
# conversion to text
self.assertTrue(data == py2.to_text(data))
self.assertTrue(isinstance(py2.to_text(data), type(u"")))
self.assertTrue(data == py2.to_text(udata))
self.assertTrue(isinstance(py2.to_text(udata), type(u"")))
self.assertTrue(data == py2.to_text(bdata))
self.assertTrue(isinstance(py2.to_text(bdata), type(u"")))
if sys.version_info[0] < 3:
self.assertTrue(u"['hello']" == py2.to_text(xdata))
else:
self.assertTrue("['hello']" == py2.to_text(xdata))
# check the empty text constant:
self.assertTrue(isinstance(py2.uempty, type(u"")))
self.assertFalse(py2.uempty)
self.assertTrue(len(py2.uempty) == 0)
def sys_path(self, path):
if is_unicode(path) and not os.path.supports_unicode_filenames:
return path.encode(sys.getfilesystemencoding())
else:
return path