# Only used as the top of the appendix hierarchy
a1 = Word(string.digits).setResultsName("a1")
aI = Word("IVXLCDM").setResultsName("aI")
# Catches the A in 12A but not in 12Awesome
markerless_upper = Word(string.ascii_uppercase).setResultsName(
'markerless_upper') + ~FollowedBy(Word(string.ascii_lowercase))
paren_upper = parenthesize(string.ascii_uppercase, "paren_upper")
paren_lower = parenthesize(string.ascii_lowercase, "paren_lower")
paren_digit = parenthesize(string.digits, "paren_digit")
period_upper = decimalize(string.ascii_uppercase, "period_upper")
period_lower = decimalize(string.ascii_lowercase, "period_lower")
period_digit = decimalize(string.digits, "period_digit")
section = (atomic.section_marker.copy().leaveWhitespace() +
unified.part_section + SkipTo(LineEnd()))
par = (atomic.section.copy().leaveWhitespace() + unified.depth1_p +
SkipTo(LineEnd()))
marker_par = (atomic.paragraph_marker.copy().leaveWhitespace() +
atomic.section + unified.depth1_p)
appendix = (atomic.appendix_marker.copy().leaveWhitespace() + atomic.appendix +
SkipTo(LineEnd()))
headers = utils.QuickSearchable(LineStart() +
(section | marker_par | par | appendix))
def _compare_search(self, grammar, text):
quick_grammar = utils.QuickSearchable(grammar)
self.assertEqual([str(m) for m in grammar.scanString(text)],
[str(m) for m in quick_grammar.scanString(text)])
class Delayed:
pass
effective_date = (utils.Marker("effective") +
utils.Marker("date")).setParseAction(lambda: EffectiveDate())
notice_citation = (
Word(string.digits) + utils.Marker('FR') +
Word(string.digits)).setParseAction(lambda m: Notice(int(m[0]), int(m[1])))
delayed = utils.Marker("delayed").setParseAction(lambda: Delayed())
def int2Month(m):
month = date(2000, m, 1)
month = month.strftime('%B')
token = utils.Marker(month)
return token.setParseAction(lambda: m)
months = reduce(lambda l, r: l | r, (int2Month(i) for i in range(2, 13)))
date_parser = (months + Word(string.digits) + Suppress(Optional(",")) + Word(
string.digits)).setParseAction(lambda m: date(int(m[2]), m[0], int(m[1])))
tokenizer = utils.QuickSearchable(effective_date | notice_citation | delayed
| date_parser)