def transform(self, relations):
for i, relation in enumerate(relations):
if relation.rel == 'parataxis':
if relation.word == 'mean' \
and len(relation.deps) == 1\
and relations[relation.deps[0]].word == "I":
delete_indices(relations, [i, relation.deps[0]])
def transform(self, relations):
for index, relation in enumerate(relations):
if relation.rel in ('null', 'ROOT', 'xcomp', 'rcmod')\
and relation.tag in ('VBZ', 'VBD', 'VBP', 'VB')\
and relation.word in self.verb_forms:
xcomp_indices = Relation.get_children_with_dep(
'xcomp', relations, index)
if xcomp_indices == []:
continue
else:
xcomp_index = xcomp_indices[0]
if relations[xcomp_index].tag == 'VB':
aux_indices = Relation.get_children_with_dep(
'aux', relations, xcomp_index)
to_index = [
index for index in aux_indices
if relations[index].tag == 'TO'
][0]
# Append 'to' and the xcomp head to main verb.
relations[index].word += ' to ' + \
relations[xcomp_index].word
# Remove 'aux' and 'xcomp' relations.
delete_indices(relations, [to_index, xcomp_index])
# Change head of relations pointing to xcomp to point to
# the main verb.
for i, rel in enumerate(relations):
if rel.head == xcomp_index - 2:
rel.head = index
relations[index].deps.append(i)
relations[index].deps.sort()
def transform(self, relations):
for index, relation in enumerate(relations):
if relation.rel in ('null', 'ROOT', 'xcomp', 'rcmod')\
and relation.tag in ('VBZ', 'VBD', 'VBP', 'VB')\
and relation.word in self.verb_forms:
xcomp_indices = Relation.get_children_with_dep('xcomp',
relations,
index)
if xcomp_indices == []:
continue
else:
xcomp_index = xcomp_indices[0]
if relations[xcomp_index].tag == 'VB':
aux_indices = Relation.get_children_with_dep('aux',
relations,
xcomp_index)
to_index = [index for index in aux_indices
if relations[index].tag == 'TO'][0]
# Append 'to' and the xcomp head to main verb.
relations[index].word += ' to ' + \
relations[xcomp_index].word
# Remove 'aux' and 'xcomp' relations.
delete_indices(relations, [to_index, xcomp_index])
# Change head of relations pointing to xcomp to point to
# the main verb.
for i, rel in enumerate(relations):
if rel.head == xcomp_index - 2:
rel.head = index
relations[index].deps.append(i)
relations[index].deps.sort()
def transform(self, relations):
for i, relation in enumerate(relations):
if relation.rel == 'parataxis':
if relation.word == 'mean' \
and len(relation.deps) == 1\
and relations[relation.deps[0]].word == "I":
delete_indices(relations, [i, relation.deps[0]])
def transform(self, relations):
for i, relation in enumerate(relations):
if relation.word\
and relation.word.lower() == 'first'\
and i + 2 < len(relations)\
and relations[i + 1].word == 'of'\
and relations[i + 2].word == 'all':
relation.word += ' of all'
relation.rel = 'advmod'
delete_indices(relations, [i + 1, i + 2])
def transform(self, relations):
for i, relation in enumerate(relations):
if relation.word\
and relation.word.lower() == 'first'\
and i + 2 < len(relations)\
and relations[i + 1].word == 'of'\
and relations[i + 2].word == 'all':
relation.word += ' of all'
relation.rel = 'advmod'
delete_indices(relations, [i + 1, i + 2])
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].tag == "EX":
relations[i + 1].word = relations[i].word +\
' ' + relations[i + 1].word
indices_to_delete.append(i)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].tag == "EX":
relations[i + 1].word = relations[i].word +\
' ' + relations[i + 1].word
indices_to_delete.append(i)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].word == 'at'\
and i + 1 < len(relations)\
and relations[i + 1].word == 'all'\
and not relations[i + 1].deps:
relations[i].word = 'at all'
indices_to_delete.append(i + 1)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].word == 'at'\
and i + 1 < len(relations)\
and relations[i + 1].word == 'all'\
and not relations[i + 1].deps:
relations[i].word = 'at all'
indices_to_delete.append(i + 1)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].word == 'up'\
and i + 1 < len(relations)\
and relations[i + 1].word == 'to':
relations[i].word = 'up to'
relations[i].tag = 'RB'
indices_to_delete.append(i + 1)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].word == 'up'\
and i + 1 < len(relations)\
and relations[i + 1].word == 'to':
relations[i].word = 'up to'
relations[i].tag = 'RB'
indices_to_delete.append(i + 1)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(1, len(relations)):
if relations[i].word.lower() in ('up', 'down', 'right')\
and i + 1 < len(relations)\
and relations[i + 1].word == 'there':
relations[i + 1].word = relations[i].word +\
' ' + relations[i + 1].word
indices_to_delete.append(i)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(1, len(relations)):
if relations[i].word.lower() in ('up', 'down', 'right')\
and i + 1 < len(relations)\
and relations[i + 1].word == 'there':
relations[i + 1].word = relations[i].word +\
' ' + relations[i + 1].word
indices_to_delete.append(i)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(1, len(relations)):
if relations[i].word.lower() == 'not'\
and relations[i].tag == 'RB'\
and relations[i].head == i + 1\
and relations[i + 1].word == 'even'\
and relations[i + 1].tag == 'RB':
relations[i + 1].word = relations[i].word +\
' ' + relations[i + 1].word
indices_to_delete.append(i)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(1, len(relations)):
if relations[i].word.lower() == 'not'\
and relations[i].tag == 'RB'\
and relations[i].head == i + 1\
and relations[i + 1].word == 'even'\
and relations[i + 1].tag == 'RB':
relations[i + 1].word = relations[i].word +\
' ' + relations[i + 1].word
indices_to_delete.append(i)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
mwes = {}
for relation in relations:
if relation.rel == 'mwe':
if relation.head not in mwes:
mwes[relation.head] = [relation.address]
else:
mwes[relation.head].append(relation.address)
for head, deps in mwes.items():
new_word = ' '.join([relations[i].word
for i in sorted(deps + [head])])
relations[head].word = new_word
delete_indices(relations, deps)
def transform(self, relations):
mwes = {}
for relation in relations:
if relation.rel == 'mwe':
if relation.head not in mwes:
mwes[relation.head] = [relation.address]
else:
mwes[relation.head].append(relation.address)
for head, deps in mwes.items():
new_word = ' '.join(
[relations[i].word for i in sorted(deps + [head])])
relations[head].word = new_word
delete_indices(relations, deps)
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].tag == "IN":
if i + 1 < len(relations) and relations[i + 1].tag == "IN":
relations[i].word += ' ' + relations[i + 1].word
for j in range(i + 1, len(relations)):
if relations[j].head == i + 1:
relations[j].head = i
relations[i].deps.append(j)
relations[i].deps.sort()
indices_to_delete.append(i + 1)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
indices_to_delete = []
for i in range(len(relations)):
if relations[i].tag == "IN":
if i + 1 < len(relations) and relations[i + 1].tag == "IN":
relations[i].word += ' ' + relations[i + 1].word
for j in range(i + 1, len(relations)):
if relations[j].head == i + 1:
relations[j].head = i
relations[i].deps.append(j)
relations[i].deps.sort()
indices_to_delete.append(i + 1)
delete_indices(relations, indices_to_delete)
def transform(self, relations):
for i, relation in enumerate(relations):
if relation.word\
and relation.word.lower() == 'a'\
and i + 2 < len(relations)\
and relations[i + 1].word == 'number'\
and relations[i + 2].word == 'of':
k = Relation.get_children_with_dep('adpobj', relations,
i + 2)[0]
relations[i].head = k
relations[i].word = 'a number of'
relations[k].head = relations[i + 1].head
relations[k].rel = relations[i + 1].rel
for j in range(i + 4, len(relations)):
if relations[j].head == i + 1:
relations[j].head = k
delete_indices(relations, [i + 1, i + 2])
def transform(self, relations):
for i, relation in enumerate(relations):
if relation.word\
and relation.word.lower() == 'a'\
and i + 2 < len(relations)\
and relations[i + 1].word == 'couple'\
and relations[i + 2].word == 'of':
k = Relation.get_children_with_dep('adpobj', relations,
i + 2)[0]
relations[i].head = k
relations[i].word = 'a couple of'
relations[k].head = relations[i + 1].head
relations[k].rel = relations[i + 1].rel
for j in range(i + 4, len(relations)):
if relations[j].head == i + 1:
relations[j].head = k
delete_indices(relations, [i + 1, i + 2])
def transform(self, relations):
for i, relation in enumerate(relations):
if relation.word == 'day' and relation.rel == 'nmod'\
and i - 1 >= 0 and relations[i - 1].word.lower() == 'one':
relation.word = relations[i - 1].word + ' ' + relation.word
delete_indices(relations, [i - 1])
def transform(self, relations):
if relations[1].word.lower() in ('and', 'then'):
delete_indices(relations, [1])
def transform(self, relations):
for i, relation in enumerate(relations):
if relation.word == 'day' and relation.rel == 'nmod'\
and i - 1 >= 0 and relations[i - 1].word.lower() == 'one':
relation.word = relations[i - 1].word + ' ' + relation.word
delete_indices(relations, [i - 1])
def transform(self, relations):
if relations[1].word.lower() in ('and', 'then'):
delete_indices(relations, [1])
def transform(self, relations):
if relations[1].word.lower() in ('e', 'aí', 'ai'):
delete_indices(relations, [1])
def transform(self, relations):
if relations[1].word.lower() in ('e', 'aí', 'ai'):
delete_indices(relations, [1])