def test_predict(self, mock_stdout):
parameters = {
'regex_patterns': {
'PAD': [r'\W'],
'BACKGROUND': [r'\w']
},
'ignore_case': True,
'default_label': 'BACKGROUND',
}
model = RegexModel(label_mapping=self.label_mapping,
parameters=parameters)
# test only pad and background separate
expected_output = {
'pred': [
np.array([[1, 0, 0], [1, 0, 0], [1, 0, 0]]),
np.array([[0, 1, 0], [0, 1, 0], [0, 1, 0], [0, 1, 0],
[0, 1, 0]])
]
}
model_output = model.predict([' ', 'hello'])
self.assertIn('pred', model_output)
for expected, output in zip(expected_output['pred'],
model_output['pred']):
self.assertTrue(np.array_equal(expected, output))
# check verbose printing
self.assertIn('Data Samples', mock_stdout.getvalue())
# test pad with background
expected_output = {
'pred': [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
]
}
model_output = model.predict([' h w.'])
self.assertIn('pred', model_output)
for expected, output in zip(expected_output['pred'],
model_output['pred']):
self.assertTrue(np.array_equal(expected, output))
# test show confidences
expected_output = {
'pred': [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
],
'conf': [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
]
}
model_output = model.predict([' h w.'], show_confidences=True)
self.assertIn('pred', model_output)
self.assertIn('conf', model_output)
for expected, output in zip(expected_output['pred'],
model_output['pred']):
self.assertTrue(np.array_equal(expected, output))
for expected, output in zip(expected_output['conf'],
model_output['conf']):
self.assertTrue(np.array_equal(expected, output))
# test verbose = False
# clear stdout
mock_stdout.seek(0)
mock_stdout.truncate(0)
model_output = model.predict(['hello world.'], verbose=False)
self.assertNotIn('Data Samples', mock_stdout.getvalue())
def test_predict(self, mock_stdout):
parameters = {
"regex_patterns": {
"PAD": [r"\W"],
"UNKNOWN": [r"\w"]
},
"ignore_case": True,
"default_label": "UNKNOWN",
}
model = RegexModel(label_mapping=self.label_mapping,
parameters=parameters)
# test only pad and background separate
expected_output = {
"pred": [
np.array([[1, 0, 0], [1, 0, 0], [1, 0, 0]]),
np.array([[0, 1, 0], [0, 1, 0], [0, 1, 0], [0, 1, 0],
[0, 1, 0]]),
]
}
with self.assertLogs("DataProfiler.labelers.regex_model",
level="INFO") as logs:
model_output = model.predict([" ", "hello"])
self.assertIn("pred", model_output)
for expected, output in zip(expected_output["pred"],
model_output["pred"]):
self.assertTrue(np.array_equal(expected, output))
# check verbose printing
self.assertIn("Data Samples", mock_stdout.getvalue())
# check verbose logging
self.assertTrue(len(logs.output))
# test pad with background
expected_output = {
"pred": [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
]
}
model_output = model.predict([" h w."])
self.assertIn("pred", model_output)
for expected, output in zip(expected_output["pred"],
model_output["pred"]):
self.assertTrue(np.array_equal(expected, output))
# test show confidences
expected_output = {
"pred": [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
],
"conf": [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
],
}
model_output = model.predict([" h w."], show_confidences=True)
self.assertIn("pred", model_output)
self.assertIn("conf", model_output)
for expected, output in zip(expected_output["pred"],
model_output["pred"]):
self.assertTrue(np.array_equal(expected, output))
for expected, output in zip(expected_output["conf"],
model_output["conf"]):
self.assertTrue(np.array_equal(expected, output))
# clear stdout
mock_stdout.seek(0)
mock_stdout.truncate(0)
# test verbose = False
# Want to ensure no INFO logged
with self.assertRaisesRegex(
AssertionError,
"no logs of level INFO or higher triggered "
"on DataProfiler.labelers.regex_model",
):
with self.assertLogs("DataProfiler.labelers.regex_model",
level="INFO"):
model.predict(["hello world."], verbose=False)
# Not in stdout
self.assertNotIn("Data Samples", mock_stdout.getvalue())
def test_predict(self, mock_stdout):
parameters = {
'regex_patterns': {
'PAD': [r'\W'],
'UNKNOWN': [r'\w']
},
'ignore_case': True,
'default_label': 'UNKNOWN',
}
model = RegexModel(label_mapping=self.label_mapping,
parameters=parameters)
# test only pad and background separate
expected_output = {
'pred': [
np.array([[1, 0, 0], [1, 0, 0], [1, 0, 0]]),
np.array([[0, 1, 0], [0, 1, 0], [0, 1, 0], [0, 1, 0],
[0, 1, 0]])
]
}
with self.assertLogs('DataProfiler.labelers.regex_model',
level='INFO') as logs:
model_output = model.predict([' ', 'hello'])
self.assertIn('pred', model_output)
for expected, output in zip(expected_output['pred'],
model_output['pred']):
self.assertTrue(np.array_equal(expected, output))
# check verbose printing
self.assertIn('Data Samples', mock_stdout.getvalue())
# check verbose logging
self.assertTrue(len(logs.output))
# test pad with background
expected_output = {
'pred': [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
]
}
model_output = model.predict([' h w.'])
self.assertIn('pred', model_output)
for expected, output in zip(expected_output['pred'],
model_output['pred']):
self.assertTrue(np.array_equal(expected, output))
# test show confidences
expected_output = {
'pred': [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
],
'conf': [
np.array([[1, 0, 0], [0, 1, 0], [1, 0, 0], [0, 1, 0],
[1, 0, 0]])
]
}
model_output = model.predict([' h w.'], show_confidences=True)
self.assertIn('pred', model_output)
self.assertIn('conf', model_output)
for expected, output in zip(expected_output['pred'],
model_output['pred']):
self.assertTrue(np.array_equal(expected, output))
for expected, output in zip(expected_output['conf'],
model_output['conf']):
self.assertTrue(np.array_equal(expected, output))
# clear stdout
mock_stdout.seek(0)
mock_stdout.truncate(0)
# test verbose = False
# Want to ensure no INFO logged
with self.assertRaisesRegex(
AssertionError, 'no logs of level INFO or higher triggered '
'on DataProfiler.labelers.regex_model'):
with self.assertLogs('DataProfiler.labelers.regex_model',
level='INFO'):
model.predict(['hello world.'], verbose=False)
# Not in stdout
self.assertNotIn('Data Samples', mock_stdout.getvalue())