def test_character_series(self):
""" Test CharacterSeries parser classes. """
# Test with ascii characters
self._test_multiple(
parser.CharacterSeries(string.ascii_letters),
[
("abc", ["abc"]),
],
must_finish=False)
# Test the Letters and Alphanumerics classes with a few inputs and
# outputs
input_outputs = [
# Unicode strings
(u"abc", [u"abc"]),
# Mix of non-ascii characters
(u"éèàâêùôöëäïüû", [u"éèàâêùôöëäïüû"]),
(u"touché", [u"touché"]),
]
self._test_multiple(
parser.Letters(),
input_outputs + [(string.digits, [])],
must_finish=False
)
self._test_multiple(
parser.Alphanumerics(),
input_outputs + [(string.digits, [string.digits])],
must_finish=False
)
def test_other_alphabets(self):
""" Test that other alphabets also work. """
# While the DNS and WSR engines are mostly limited to western
# European languages (i.e. windows-1252), that doesn't necessarily
# mean the parser needs to be.
input_outputs = [
(u"йцукенгшщзхъфывапролджэячсмитьбю",
[u"йцукенгшщзхъфывапролджэячсмитьбю"]),
]
self._test_multiple(parser.Letters(), input_outputs, must_finish=False)
self._test_multiple(parser.Alphanumerics(),
input_outputs,
must_finish=False)
def test_character_series(self):
""" Test CharacterSeries parser class. """
# Test with ascii characters
self._test_multiple(
parser.CharacterSeries(string.letters),
[
("abc", ["abc"]),
],
must_finish=False)
# Test with Unicode characters
self._test_multiple(
parser.Letters(),
[
(u"abc", [u"abc"]),
],
must_finish=False
)
def test_repetition(self):
""" Test repetition parser class. """
word = parser.Letters()
whitespace = parser.Whitespace()
p = parser.Repetition(parser.Alternative((word, whitespace)))
# Test with ascii letters
input_output = (
("abc", ["abc"]),
("abc abc", ["abc", " ", "abc"]),
("abc abc\t\t\n cba", ["abc", " ", "abc", "\t\t\n ", "cba"]),
)
self._test_single(p, input_output)
# Test with non-ascii letters
input_output = (
(u"êùö", [u"êùö"]),
(u"êùö êùö", [u"êùö", u" ", u"êùö"]),
(u"êùö êùö\t\t\n öùê",
[u"êùö", " ", u"êùö", u"\t\t\n ", u"öùê"]),
)
self._test_single(p, input_output)
