def mk_punct(s):
return {'.': terminator('.'),
'?': terminator('?'),
'!': terminator('!'),
',': separator(','),
';': separator(';'),
'-': sequence(abut, literal('-'), abut),
'--': literal('--')}[s]
def mk_punct(s):
return {
'.': terminator('.'),
'?': terminator('?'),
'!': terminator('!'),
',': separator(','),
';': separator(';'),
'-': sequence(abut, literal('-'), abut),
'--': literal('--')
}[s]
def parse(grammar):
rules = {
'-a-': a_an,
'-an-': a_an,
'-a-an-': a_an,
'-adjoining-': abut,
'-capitalize-': capitalize
}
parser = Parser(
r"""
grammar = _ rules
rules = rule rules |
rule = name [=] _ exp hug
exp = alts mk_choice
alts = alt [/] _ alts
| alt
alt = \[ number \] _ seq mk_weight
| seq mk_unit
seq = factor seq mk_seq
| mk_empty
factor = name ![=] mk_ref
| punct mk_punct
| \( _ exp \) _
| { _ alts } _ mk_shuffle
| word { _ alts } _ mk_fixed
| word mk_literal
punct = ([.,;?!]) _
| (--?)\s _
word = ([A-Za-z0-9']+) _
name = (-[A-Za-z0-9'-]+-) _
number = (\d+) _ int
_ = (?:\s|#.*)*
""",
hug=hug,
int=int,
mk_choice=lambda *pairs: pairs[0][0]
if 1 == len(pairs) else weighted_choice(dict(pairs)),
mk_empty=lambda: empty,
mk_fixed=lambda tag, *pairs: fixed(tag)(dict(pairs)),
mk_literal=literal,
mk_punct=mk_punct,
mk_ref=lambda name: delay(lambda: rules[name]),
mk_seq=lambda p, q: sequence(p, q) if q is not empty else p,
mk_shuffle=lambda *pairs: shuffled(dict(pairs)),
mk_unit=lambda p: (p, 1),
mk_weight=lambda w, p: (p, w),
)
rules.update(parser(grammar))
return rules
def parse(grammar):
rules = {'-a-': a_an,
'-an-': a_an,
'-a-an-': a_an,
'-adjoining-': abut,
'-capitalize-': capitalize}
parser = Parser(r"""
grammar = _ rules
rules = rule rules |
rule = name [=] _ exp hug
exp = alts mk_choice
alts = alt [/] _ alts
| alt
alt = \[ number \] _ seq mk_weight
| seq mk_unit
seq = factor seq mk_seq
| mk_empty
factor = name ![=] mk_ref
| punct mk_punct
| \( _ exp \) _
| { _ alts } _ mk_shuffle
| word { _ alts } _ mk_fixed
| word mk_literal
punct = ([.,;?!]) _
| (--?)\s _
word = ([A-Za-z0-9']+) _
name = (-[A-Za-z0-9'-]+-) _
number = (\d+) _ int
_ = (?:\s|#.*)*
""",
hug=hug,
int=int,
mk_choice = lambda *pairs: pairs[0][0] if 1 == len(pairs) else weighted_choice(dict(pairs)),
mk_empty = lambda: empty,
mk_fixed = lambda tag, *pairs: fixed(tag)(dict(pairs)),
mk_literal = literal,
mk_punct = mk_punct,
mk_ref = lambda name: delay(lambda: rules[name]),
mk_seq = lambda p, q: sequence(p, q) if q is not empty else p,
mk_shuffle = lambda *pairs: shuffled(dict(pairs)),
mk_unit = lambda p: (p, 1),
mk_weight = lambda w, p: (p, w),
)
rules.update(parser(grammar))
return rules
