def _parse_compare(self, op_tokens, arg_tokens):
if len(op_tokens) >= 3:
if normalize(op_tokens[1]) == 'more':
return (normalize(op_tokens[2]), False)
elif normalize(op_tokens[1]) == 'less':
return (normalize(op_tokens[2]), True)
token = normalize(op_tokens[1])
if token.endswith('er'):
token = token[:-2]
if token.endswith('i'):
token = token[:-1] + 'y'
return (token, False)
def parse_question(self, question, discard_global):
# Filter out 'global' questions
if discard_global and question['semantic'][0][
'operation'] == 'select' and question['semantic'][0][
'argument'] == 'scene':
return None
ops, args, deps = self.parse_program(question['semantic'])
if None in ops or None in args:
return None
# Combine all 'verify_attr-and' to 'verify_attrs'
trace, _ = self._compute_op_trace(ops, deps)
ops, args, deps, trace = self._combine_verify(ops, args, deps, trace)
if self._is_batch_format:
question['program'] = self._de_branch_program(ops, args, deps)
question['program'] = self._fix_logical_branches(
question['program'])
else:
question['operators'] = ops
question['arguments'] = args
question['dependencies'] = deps
question['answer'] = normalize(question['answer'])
# print(question['semantic'], ' ---> ', ops, args, deps)
return question
def _parse_two_different(self, op_tokens, arg_tokens):
op_tokens = [normalize(t) for t in op_tokens[1:]]
return (' '.join(op_tokens), )
def _parse_all_different(self, op_tokens, arg_tokens):
return (normalize(arg_tokens[0]), )
def _parse_all_same(self, op_tokens, arg_tokens):
return (normalize(arg_tokens[0]), )
def _parse_verify_rel(self, op_tokens, arg_tokens):
return (normalize(arg_tokens[1]), arg_tokens[2] == 's',
normalize(arg_tokens[0]))
def _parse_choose_rel(self, op_tokens, arg_tokens):
rels = re.split('\|', arg_tokens[1])
rels = [normalize(r) for r in rels]
return (rels, arg_tokens[2] == 's', normalize(arg_tokens[0]))
def _parse_verify_attrs(self, op_tokens, arg_tokens):
return ([normalize(t) for t in arg_tokens], )
def _parse_verify_attr(self, op_tokens, arg_tokens):
return (normalize(arg_tokens[0]), )
def _parse_choose_attr(self, op_tokens, arg_tokens):
arg_tokens = re.split('\|', arg_tokens[0])
return ([normalize(t) for t in arg_tokens], )
def _parse_filter(self, op_tokens, arg_tokens):
return (normalize(arg_tokens[0]), )
def _parse_select(self, op_tokens, arg_tokens):
return (normalize(arg_tokens[0]), )
