def _process_pred_annotations(
self,
pack: DataPack,
labels: List[str],
word_begin: int,
word_end: int,
current_pred_arg: List[Stack],
verbal_pred_args: List[List[Tuple[PredicateArgument, str]]],
) -> None:
"""
Various cases will be handled regarding nested spans including:
case 1:
(xxx(xxx*)*)
case 2:
(xxx(xxx*)
*
*)
case 3:
(xxx(xxx*
*
*)
*
*)
case 4:
(xxx*
*
(xxx
*
*)
*
*)
Where the `xxx` represents an argument type.
A stack will be maintained to parse nested spans.
"""
# TODO: currently all nested spans will be parsed but only outer most
# span will be stored into datapack.
for label_index, label in enumerate(labels):
label = label.strip()
arg_type: str = ""
i: int = 0
while i < len(label):
c: str = label[i]
if c == '*':
i += 1
elif c == '(':
# New argument span.
j: int = i + 1
while j < len(label):
arg_type += label[j]
if j == len(label) - 1 or label[j + 1] in ['(', ')']:
break
j += 1
i = j + 1
arg_type = arg_type.strip("* ")
stack: Stack = current_pred_arg[label_index]
stack.append((word_begin, arg_type))
arg_type = ""
elif c == ')':
# End of current argument span.
stack = current_pred_arg[label_index]
assert len(stack) > 0, \
"invalid parsing state: mismatch argument span"
arg_begin, arg_type = stack.pop()
if not stack:
# The outer most span will be stored into data pack
if arg_type != "V":
pred_arg = PredicateArgument(
pack, arg_begin, word_end)
verbal_pred_args[label_index].append(
(pred_arg, arg_type))
i += 1