def test_set_params(self, mock_open, mock_load_model, mock_base_processor):
self._setup_mock_load_model(mock_load_model)
self._setup_mock_load_processor(mock_base_processor)
# load default
data_labeler = UnstructuredDataLabeler()
# check empty sent
with self.assertRaisesRegex(
ValueError,
re.escape("The params dict must have the following "
"format:\nparams=dict(preprocessor=dict(..."
"), model=dict(...), postprocessor=dict(..."
")), where each sub-dict contains "
"parameters of the specified data_labeler "
"pipeline components."),
):
data_labeler.set_params(None)
with self.assertRaisesRegex(
ValueError,
re.escape("The params dict must have the following "
"format:\nparams=dict(preprocessor=dict(..."
"), model=dict(...), postprocessor=dict(..."
")), where each sub-dict contains "
"parameters of the specified data_labeler "
"pipeline components."),
):
data_labeler.set_params({})
# test if invalid key sent
with self.assertRaisesRegex(
ValueError,
re.escape("The params dict must have the following "
"format:\nparams=dict(preprocessor=dict(..."
"), model=dict(...), postprocessor=dict(..."
")), where each sub-dict contains "
"parameters of the specified data_labeler "
"pipeline components."),
):
data_labeler.set_params({"bad key": None})
# validate no errors occur when correct params are sent
data_labeler._preprocessor.get_parameters.return_value = dict()
data_labeler._model.get_parameters.return_value = dict()
data_labeler._postprocessor.get_parameters.return_value = dict()
data_labeler.set_params({
"preprocessor": {
"test": 1
},
"model": {
"test": 1
},
"postprocessor": {
"test2": 3
},
})
# validate warning on overlaps.
# here we presume parameters are set as dict(test=1), dict(test=2)
data_labeler._preprocessor.get_parameters.return_value = dict(test=1)
data_labeler._model.get_parameters.return_value = dict(test=2)
with self.assertWarnsRegex(
RuntimeWarning,
"Model and preprocessor value for `test` do "
"not match. 2 != 1",
):
data_labeler.set_params({
"preprocessor": {
"test": 1
},
"model": {
"test": 2
}
})
# check if param sent for missing pipeline component
data_labeler._preprocessor = None
with self.assertRaisesRegex(
ValueError,
"Parameters for the preprocessor, model, or"
" postprocessor were specified when one or "
"more of these were not set in the "
"DataLabeler.",
):
data_labeler.set_params({"preprocessor": {"test": 1}})
data_labeler._model = None
with self.assertRaisesRegex(
ValueError,
"Parameters for the preprocessor, model, or"
" postprocessor were specified when one or "
"more of these were not set in the "
"DataLabeler.",
):
data_labeler.set_params({"model": {"test": 1}})
data_labeler._postprocessor = None
with self.assertRaisesRegex(
ValueError,
"Parameters for the preprocessor, model, or"
" postprocessor were specified when one or "
"more of these were not set in the "
"DataLabeler.",
):
data_labeler.set_params({"postprocessor": {"test": 1}})
def test_set_params(self, mock_open, mock_load_model, mock_base_processor):
self._setup_mock_load_model(mock_load_model)
self._setup_mock_load_processor(mock_base_processor)
# load default
data_labeler = UnstructuredDataLabeler()
# check empty sent
with self.assertRaisesRegex(
ValueError,
re.escape('The params dict must have the following '
'format:\nparams=dict(preprocessor=dict(...'
'), model=dict(...), postprocessor=dict(...'
')), where each sub-dict contains '
'parameters of the specified data_labeler '
'pipeline components.')):
data_labeler.set_params(None)
with self.assertRaisesRegex(
ValueError,
re.escape('The params dict must have the following '
'format:\nparams=dict(preprocessor=dict(...'
'), model=dict(...), postprocessor=dict(...'
')), where each sub-dict contains '
'parameters of the specified data_labeler '
'pipeline components.')):
data_labeler.set_params({})
# test if invalid key sent
with self.assertRaisesRegex(
ValueError,
re.escape('The params dict must have the following '
'format:\nparams=dict(preprocessor=dict(...'
'), model=dict(...), postprocessor=dict(...'
')), where each sub-dict contains '
'parameters of the specified data_labeler '
'pipeline components.')):
data_labeler.set_params({'bad key': None})
# validate no errors occur when correct params are sent
data_labeler._preprocessor.get_parameters.return_value = dict()
data_labeler._model.get_parameters.return_value = dict()
data_labeler._postprocessor.get_parameters.return_value = dict()
data_labeler.set_params({
'preprocessor': {
'test': 1
},
'model': {
'test': 1
},
'postprocessor': {
'test2': 3
}
})
# validate warning on overlaps.
# here we presume parameters are set as dict(test=1), dict(test=2)
data_labeler._preprocessor.get_parameters.return_value = dict(test=1)
data_labeler._model.get_parameters.return_value = dict(test=2)
with self.assertWarnsRegex(
RuntimeWarning, 'Model and preprocessor value for `test` do '
'not match. 2 != 1'):
data_labeler.set_params({
'preprocessor': {
'test': 1
},
'model': {
'test': 2
}
})
# check if param sent for missing pipeline component
data_labeler._preprocessor = None
with self.assertRaisesRegex(
ValueError, 'Parameters for the preprocessor, model, or'
' postprocessor were specified when one or '
'more of these were not set in the '
'DataLabeler.'):
data_labeler.set_params({'preprocessor': {'test': 1}})
data_labeler._model = None
with self.assertRaisesRegex(
ValueError, 'Parameters for the preprocessor, model, or'
' postprocessor were specified when one or '
'more of these were not set in the '
'DataLabeler.'):
data_labeler.set_params({'model': {'test': 1}})
data_labeler._postprocessor = None
with self.assertRaisesRegex(
ValueError, 'Parameters for the preprocessor, model, or'
' postprocessor were specified when one or '
'more of these were not set in the '
'DataLabeler.'):
data_labeler.set_params({'postprocessor': {'test': 1}})