def WriteAdjective(outfile, degree, lemma, adjective, standard=None):
synset = lemma.synset.name
count = base.GetCompoundCount(lemma)
standard = '+' if standard else '-'
terminals = base.LemmaToTerminals(adjective)
outfile.write('Adj[DEG=%s,SNS="%s",FRQ=%d,%sstd] -> %s\n' %
(degree, synset, count, standard, terminals))
def WriteProperNoun(outfile, lemma, gender):
if ShouldBeIgnored(lemma): return
noun = base.LemmaToTerminals(lemma.name)
count = base.GetCompoundCount(lemma) - 1
if gender:
outfile.write('PrpN[NUM=sg,SNS="%s",SEX=%s,FRQ=%d] -> %s\n' %
(lemma.synset.name, gender, count, noun))
else:
outfile.write('PrpN[NUM=sg,SNS="%s",FRQ=%d] -> %s\n' %
(lemma.synset.name, count, noun))
def WriteNoun(outfile, lemma, gender, number, name_override=None):
if ShouldBeIgnored(name_override or lemma): return
noun = base.LemmaToTerminals(name_override or lemma.name)
count = base.GetCompoundCount(lemma)
if gender:
outfile.write('Noun[NUM=%s,SNS="%s",SEX=%s,FRQ=%s] -> %s\n' %
(number, lemma.synset.name, gender, count, noun))
else:
outfile.write('Noun[NUM=%s,SNS="%s",FRQ=%s] -> %s\n' %
(number, lemma.synset.name, count, noun))
def GetVerbRules(patterns):
conjugation = pickle.load(open(VERBS_LIST))
rules = []
with_children = collections.defaultdict(set)
for cls in nltk.corpus.verbnet.classids():
if cls.count('-') > 1:
name, number, suffix = cls.split('-', 2)
base_class = name + '-' + number
for number in suffix.split('-'):
with_children[base_class].add(cls)
base_class += '-' + number
with_children[base_class].add(cls)
else:
with_children[cls].add(cls)
for index, pattern, classes in patterns:
for cls in classes:
for frame_cls in with_children[cls]:
frame_cls = nltk.corpus.verbnet.vnclass(frame_cls)
for member in frame_cls.findall('MEMBERS/MEMBER'):
verb = member.attrib['name']
lemmas = member.attrib['wn'].replace('?', '').split()
if lemmas:
lemmas = [
nltk.corpus.wordnet.lemma_from_key(i + '::')
for i in lemmas
]
else:
lemmas = nltk.corpus.wordnet.lemmas(verb, 'v')
if len(lemmas) > 1:
lemmas = []
for lemma in lemmas:
synset = lemma.synset.name
count = base.GetCompoundCount(lemma)
if verb not in conjugation: continue
for form, conjugated_verb in zip(
VERB_FORMS, conjugation[verb]):
conjugated_verb = base.LemmaToTerminals(
conjugated_verb)
args = (form, index, cls, synset, count,
conjugated_verb)
rules.append(VERB_TEMPLATE % args)
return rules
def ShouldBeIgnored(lemma):
return ((IGNORE_PROPER_MULTIWORD and '_' in lemma.name)
or (IGNORE_UNCOMMON and not base.GetCompoundCount(lemma))
or lemma.name in EXCEPTIONS or len(lemma.name) == 1)