def get_dataframe(self, columns: Columns = None) -> DataFrame:
if pd and get_use_objects_for_output():
dataframe = DataFrame(self.get_items())
if arg.is_defined(columns):
columns = arg.get_names(columns)
dataframe = dataframe[columns]
return dataframe
def show(self,
count: Optional[int] = None,
as_dataframe: Union[bool, Auto] = AUTO) -> Optional[DataFrame]:
as_dataframe = Auto.acquire(as_dataframe, get_use_objects_for_output())
if as_dataframe:
return self.get_dataframe()
else:
return self.describe(as_dataframe=False)
def get_dataframe(self, columns: Optional[Iterable] = None) -> DataFrame:
if pd and get_use_objects_for_output():
if columns:
dataframe = DataFrame(self.get_items(), columns=columns)
columns = arg.get_names(columns)
dataframe = dataframe[columns]
else:
dataframe = DataFrame(self.get_items())
return dataframe
def describe(self,
*filters,
take_struct_from_source: bool = False,
count: Count = 10,
columns: Columns = None,
show_header: bool = True,
struct_as_dataframe: bool = False,
separate_by_tabs: bool = False,
allow_collect: bool = True,
**filter_kwargs):
if show_header:
for line in self.get_str_headers():
self.log(line)
example = self.example(*filters, **filter_kwargs, count=count)
assert isinstance(example, ColumnarMixin)
if hasattr(self, 'get_struct'):
expected_struct = self.get_struct()
source_str = 'native'
elif take_struct_from_source:
expected_struct = self.get_source_struct()
source_str = 'from source {}'.format(self.get_source().__repr__())
else:
expected_struct = self.get_detected_struct()
source_str = 'detected from example items'
expected_struct = fc.FlatStruct.convert_to_native(expected_struct)
detected_struct = example.get_detected_struct(count)
detected_struct.validate_about(expected_struct)
message = '{} {}'.format(source_str,
expected_struct.get_validation_message())
struct_as_dataframe = struct_as_dataframe and get_use_objects_for_output(
)
struct_dataframe = expected_struct.describe(
as_dataframe=struct_as_dataframe,
show_header=False,
logger=self.get_logger(),
separate_by_tabs=separate_by_tabs,
example=example.get_one_item(),
comment=message,
)
if struct_as_dataframe:
return struct_dataframe
else:
return example.get_demo_example(
as_dataframe=get_use_objects_for_output())
def describe(
self, *filters,
take_struct_from_source: bool = False,
count: Count = DEFAULT_SHOW_COUNT,
columns: Columns = None,
allow_collect: bool = True,
show_header: bool = True,
struct_as_dataframe: bool = False,
delimiter: str = ' ',
output=AUTO,
**filter_kwargs
):
output = Auto.delayed_acquire(output, self.get_logger)
if show_header:
for line in self.get_str_headers():
self.output_line(line, output=output)
example = self.example(*filters, **filter_kwargs, count=count)
if hasattr(self, 'get_struct'):
expected_struct = self.get_struct()
source_str = 'native'
elif take_struct_from_source:
expected_struct = self.get_source_struct()
source_str = 'from source {}'.format(self.get_source().__repr__())
else:
expected_struct = self.get_detected_struct()
source_str = 'detected from example items'
expected_struct = fc.FlatStruct.convert_to_native(expected_struct)
detected_struct = example.get_detected_struct(count)
assert isinstance(expected_struct, fc.FlatStruct) or hasattr(expected_struct, 'describe'), expected_struct
assert isinstance(detected_struct, fc.FlatStruct) or hasattr(expected_struct, 'describe'), expected_struct
detected_struct.validate_about(expected_struct)
validation_message = '{} {}'.format(source_str, expected_struct.get_validation_message())
struct_as_dataframe = struct_as_dataframe and get_use_objects_for_output()
struct_dataframe = expected_struct.describe(
as_dataframe=struct_as_dataframe, show_header=False, output=output,
delimiter=delimiter, example=example.get_one_item(), comment=validation_message,
)
if struct_as_dataframe:
return struct_dataframe
else:
return example.get_demo_example(as_dataframe=get_use_objects_for_output())
def get_demo_example(
self, count: int = DEFAULT_SHOW_COUNT,
as_dataframe: AutoBool = AUTO,
filters: Optional[Array] = None, columns: Optional[Array] = None,
) -> Union[DataFrame, Iterable]:
as_dataframe = arg.acquire(as_dataframe, get_use_objects_for_output())
sm_sample = self.filter(*filters or []) if filters else self
sm_sample = sm_sample.take(count)
if hasattr(sm_sample, 'get_dataframe') and as_dataframe:
return sm_sample.get_dataframe(columns)
elif hasattr(sm_sample, 'select') and columns:
return sm_sample.select(*columns).get_items()
elif hasattr(sm_sample, 'get_items'):
return sm_sample.get_items()
def get_demo_example(
self,
count: Count = DEFAULT_EXAMPLE_COUNT,
filters: Columns = None,
columns: Columns = None,
as_dataframe: AutoBool = AUTO,
) -> Union[DataFrame, list, None]:
as_dataframe = arg.acquire(as_dataframe, get_use_objects_for_output())
sm_sample = self.filter(*filters) if filters else self
sm_sample = sm_sample.take(count)
if as_dataframe:
return sm_sample.get_dataframe(columns)
elif hasattr(sm_sample, 'get_list'):
return sm_sample.get_list()
