def identifier(first=partial(one_of, ascii_letters),
consecutive=partial(one_of, ascii_letters + digit_chars + '_'),
must_contain=set(digit_chars)):
"""Expects a letter followed by one or more alphanumerical
characters. If ``must_contain`` is given, the following letters
must include one from this set.
The default option is to expect a letter followed by a number of
letters and digits, but with a requirement of at least one digit
(this allows an easy distinction between names and identifiers).
>>> parse(identifier, 'abc123')
'abc123'
>>> parse(identifier, 'abc') # doctest: +ELLIPSIS
Traceback (most recent call last):
...
NoMatch: ...
>>> parse(partial(identifier, must_contain=None), 'abc')
'abc'
"""
result = []
if first is not None:
result.append(first())
if must_contain is None:
chars = many(consecutive)
else:
chars = many1(partial(choice, consecutive, partial(one_of, must_contain)))
if not set(chars) & must_contain:
fail()
result.extend(chars)
return result
def name():
"""Parses one or more names separated by whitespace, and
concatenates with a single space.
>>> parse(name, 'John')
'John'
>>> parse(name, 'John Smith')
'John Smith'
>>> parse(name, 'John, Smith')
'John'
>>> parse(name, 'John Smith ')
'John Smith'
"""
names = map(partial("".join), sep1(
partial(many, partial(satisfies, lambda l: l and l.isalpha())), whitespace1))
name = " ".join(names).strip()
if not name:
fail()
return name
def _parse_date_format(date_format):
tokens = tuple(date_format.replace('%', ''))
parsers = map(tri, map(_date_format_mapping.get, tokens))
date_string = "".join(itertools.chain(*[parser() for parser in parsers]))
try:
return datetime.datetime.strptime(date_string, date_format)
except:
fail()
def end_element(name):
whitespace()
string("</")
commit()
if name != xml_name():
fail()
whitespace()
close_angle()
def end_element(name):
whitespace()
string("</")
commit()
if name != xml_name():
fail()
whitespace()
close_angle()
def parse(cls):
if optional(any_token, None):
fail()
def named_entity():
name = build_string(many1(partial(not_one_of, ";#")))
if name not in named_entities:
fail()
return named_entities[name]
def named_entity():
name = build_string(many1(partial(not_one_of,';#')))
if name not in named_entities: fail()
return named_entities[name]
