def apply_annotators(self, chart, tokens, i, j):
"""Add parses to chart cell (i, j) by applying annotators."""
if self.annotators:
words = [t['word'] for t in tokens]
for annotator in self.annotators:
for category, semantics in annotator.annotate(tokens[i:j]):
rule = Rule(category, tuple(words[i:j]), semantics)
chart[(i, j)].append(Parse(rule, words[i:j]))
def apply_aliases(self, chart, words, i, j):
"""Add parses to chart cell (i, j) by applying user lists."""
if self.aliases:
key = ' '.join(words[i:j])
if key in self.aliases:
lhs = '$UserList'
rhs = tuple(key.split())
semantics = ('.alias', ('.string', key))
rule = Rule(lhs, rhs, semantics)
chart[(i, j)].append(Parse(rule, words[i:j]))
def apply_absorb_rules(self, chart, i, j):
"""Add parses to chart cell (i, j) that require absorbing."""
if j - i > 2: # Otherwise, there's no chance for absorption
for m in range(i + 1, j - 1):
for n in range(m + 1, j):
for parse_1, parse_2 in product(chart[(i, m)], chart[(n, j)]):
# Don't absorb unmatched quote marks
if sum(parse.rule.lhs=='$Quote' for p in range(m, n) for parse in chart[(p, p+1)]) % 2 != 0:
break
for rule in self.binary_rules[(parse_1.rule.lhs, parse_2.rule.lhs)]:
# Don't allow $StringStub to absorb (to control growth)
if rule.lhs=='$StringStub':
continue
absorbed = n - m
chart[(i, j)].append(Parse(rule, [parse_1, parse_2], absorbed))
def apply_unary_rules(self, chart, i, j):
"""Add parses to chart cell (i, j) by applying unary rules."""
for parse in chart[(i, j)]:
for rule in self.unary_rules[(parse.rule.lhs,)]:
chart[(i, j)].append(Parse(rule, [parse]))
def apply_binary_rules(self, chart, i, j):
"""Add parses to chart cell (i, j) by applying binary rules."""
for k in range(i + 1, j):
for parse_1, parse_2 in product(chart[(i, k)], chart[(k, j)]):
for rule in self.binary_rules[(parse_1.rule.lhs, parse_2.rule.lhs)]:
chart[(i, j)].append(Parse(rule, [parse_1, parse_2]))
def apply_lexical_rules(self, chart, words, i, j):
"""Add parses to chart cell (i, j) by applying lexical rules."""
for rule in self.lexical_rules[tuple(words[i:j])]:
chart[(i, j)].append(Parse(rule, words[i:j]))