def __init__(self):
characters = string.letters + string.digits + "_-.'"
word = parser_.CharacterSeries(characters)
whitespace = parser_.Whitespace()
elements = (
word,
parser_.Repetition(parser_.Sequence((whitespace, word)), min=0),
)
parser_.Sequence.__init__(self, elements)
def initialize(self):
self._element_identifier = _ElementRef()
self._action_identifier = _ActionRef()
alternatives = parser_.Alternative((
_Literal(),
_Optional(),
_Group(),
self._element_identifier,
))
elements = [
parser_.Whitespace(optional=True),
alternatives,
parser_.Optional(
parser_.Sequence((
parser_.Whitespace(optional=True),
self._action_identifier,
))),
parser_.Whitespace(optional=True),
]
parser_.Sequence.__init__(self, elements)
def __init__(self):
# Use a pattern to allow ascii and Unicode alphanumeric characters plus a
# few special characters.
pattern = re.compile(r"[\w_\-.',]", re.UNICODE)
word = parser_.CharacterSeries(None, pattern=pattern)
whitespace = parser_.Whitespace()
elements = (
word,
parser_.Repetition(parser_.Sequence((whitespace, word)), min=0),
)
parser_.Sequence.__init__(self, elements)
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)
