def _parse_index(
index: Optional[Union[int, list, tuple, slice, str]] = None
) -> Optional[Union[int, slice]]:
if index is None:
return None
elif isinstance(index, (int, slice)):
return index
elif isinstance(index, (list, tuple)):
if len(index) > 3:
raise ge_exceptions.BatchFilterError(
f"""The number of index slice components must be between 1 and 3 (the given number is
{len(index)}).
""")
if len(index) == 1:
return index[0]
if len(index) == 2:
return slice(index[0], index[1], None)
if len(index) == 3:
return slice(index[0], index[1], index[2])
elif isinstance(index, str):
if is_int(value=index):
return _parse_index(index=int(index))
return _parse_index(
index=[int(idx_str) for idx_str in index.split(":")])
else:
raise ge_exceptions.BatchFilterError(
f"""The type of index must be an integer (Python "int"), or a list (Python "list") or a tuple
(Python "tuple"), or a Python "slice" object, or a string that has the format of a single integer or a slice argument.
The type given is "{str(type(index))}", which is illegal.
""")
def get_batch_key(self, batch_definition: BatchDefinition) -> Any:
batch_identifiers: dict = batch_definition.batch_identifiers
batch_value: Any = batch_identifiers[self.name]
if not is_numeric(value=batch_value):
raise ge_exceptions.SorterError(
# what is the identifying characteristic of batch_definition?
f"""BatchDefinition with IDDict "{self.name}" with value "{batch_value}" has value
"{batch_value}" which cannot be part of numeric sort.
""")
if is_int(value=batch_value):
return int(batch_value)
# The case of strings having floating point number format used as references to partitions should be rare.
return round(float(batch_value))