def test_get_actual_utterance_multiple_occurences_of_each(self):
"""Test with 2 shortenings, 2 fragments and 2 replacements."""
utterance = ('(A)mu:(ğ)ça yarasam [: yorosom] '
'&ab yarasam [: yorosom] &ac')
actual_output = Extractor.to_actual_utterance(utterance)
desired_output = 'mu:ça yarasam ab yarasam ac'
self.assertEqual(actual_output, desired_output)
def get_target_utterance(self):
"""Get the target form of the utterance.
Returns: str
"""
utterance = self.get_utterance()
return ActualTargetUtteranceExtractor.to_target_utterance(utterance)
def test_get_replacement_target_multiple_replacements(self):
"""Test get_replacement_target with 3 replacements."""
utterance = 'yarasam [: yorosom] yarasam [: yorosom] ' \
'yarasam [: yorosom] .'
actual_output = Extractor.get_replacement_target(utterance)
desired_output = 'yorosom yorosom yorosom .'
self.assertEqual(actual_output, desired_output)
def test_to_target_utterance_multiple_occurrences_of_each(self):
"""Test with 2 shortenings, 2 fragments and 2 replacements."""
utterance = ('yarasam [: yorosom] '
'&ab (a)mu:(ğ)ça &ac yarasam [: yorosom]')
actual_output = Extractor.to_target_utterance(utterance)
desired_output = 'yorosom xxx amu:ğça xxx yorosom'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_actual_retracing_several_words_no_whitespace(self):
"""Test retracing with several words and missing whitespace."""
utterance = '<this is>[/] this is a test'
actual_output = Extractor.get_retracing_actual(utterance)
desired_output = 'this is this is a test'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_target_correction_no_whitespace(self):
"""Test with correction and missing whitespace."""
utterance = 'this us[//] is a test'
actual_output = Extractor.get_retracing_target(utterance)
desired_output = 'this is is a test'
self.assertEqual(actual_output, desired_output)
def test_get_shortening_target_empty_string(self):
"""Test get_shortening_target with an empty string."""
utterance = ''
actual_output = Extractor.get_shortening_target(utterance)
desired_output = ''
self.assertEqual(actual_output, desired_output)
def test_get_shortening_target_special_characters(self):
"""Test get_shortening_target with special chars in parentheses."""
utterance = 'Mu:(ğ)ça .'
actual_output = Extractor.get_shortening_target(utterance)
desired_output = 'Mu:ğça .'
self.assertEqual(actual_output, desired_output)
def test_get_shortening_target_multiple_shortenings(self):
"""Test get_shortening_target with 3 shortening occurence."""
utterance = '(o)na:(ra)da dükäm lan(da) [?] [>] ?'
actual_output = Extractor.get_shortening_target(utterance)
desired_output = 'ona:rada dükäm landa [?] [>] ?'
self.assertEqual(actual_output, desired_output)
def test_get_actual_utterance_nesting(self):
"""Test with nesting."""
utterance = '<it shoulda[: should have]> [<2] ?'
actual_output = Extractor.to_actual_utterance(utterance)
desired_output = '<it shoulda> [<2] ?'
self.assertEqual(actual_output, desired_output)
def test_to_target_utterance_no_occurrences(self):
"""Test to_target_utterance using an utterance without occurrences."""
utterance = 'mu:ça yarasam ab yarasam ac'
actual_output = Extractor.to_target_utterance(utterance)
desired_output = 'mu:ça yarasam ab yarasam ac'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_actual_reformulation(self):
"""Test get_retracing_actual."""
utterance = 'what [///] why do you eat'
actual_output = Extractor.get_retracing_actual(utterance)
desired_output = 'what why do you eat'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_actual_no_whitespace(self):
"""Test with missing whitespace."""
utterance = 'this is[/] is a test'
actual_output = Extractor.get_retracing_actual(utterance)
desired_output = 'this is is a test'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_actual_correction(self):
"""Test get_retracing_actual."""
utterance = 'this us [//] is a test'
actual_output = Extractor.get_retracing_actual(utterance)
desired_output = 'this us is a test'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_actual_retracing_several_words(self):
"""Test get_retracing_actual."""
utterance = '<this is> [/] this is a test'
actual_output = Extractor.get_retracing_actual(utterance)
desired_output = 'this is this is a test'
self.assertEqual(actual_output, desired_output)
def test_get_fragment_target_ampersand_inside_word(self):
"""Test with ampersand occurring inside the word."""
utterance = 'dont&handle'
actual_output = Extractor.get_fragment_target(utterance)
desired_output = 'dont&handle'
self.assertEqual(actual_output, desired_output)
def test_get_fragment_target_one_char_fragment(self):
"""Test with a fragment consisting of one character."""
utterance = 'This is &a test'
actual_output = Extractor.get_fragment_target(utterance)
desired_output = 'This is xxx test'
self.assertEqual(actual_output, desired_output)
def test_get_fragment_target_ampersand_outside(self):
"""Test get_fragment_target with ampersand outside fragment."""
utterance = '&=laugh &wow &-um'
actual_output = Extractor.get_fragment_target(utterance)
desired_output = '&=laugh xxx &-um'
self.assertEqual(actual_output, desired_output)
def test_to_target_utterance_empty_string(self):
utterance = ''
actual_output = Extractor.to_target_utterance(utterance)
desired_output = ''
self.assertEqual(actual_output, desired_output)
def test_get_retracing_target_retracing(self):
"""Test get_retracing_target."""
utterance = 'this is [/] is a test'
actual_output = Extractor.get_retracing_target(utterance)
desired_output = 'this is is a test'
self.assertEqual(actual_output, desired_output)
def test_to_target_utterance_one_occurrence_of_each(self):
"""Test with 1 shortening, 1 fragment and 1 replacement."""
utterance = 'Mu:(ğ)ça &ab yarasam [: yorosom]'
actual_output = Extractor.to_target_utterance(utterance)
desired_output = 'Mu:ğça xxx yorosom'
self.assertEqual(actual_output, desired_output)
def test_get_shortening_actual_no_shortening(self):
"""Test get_shortening_actual using utt without shortening."""
utterance = 'Tu:çe .'
actual_output = Extractor.get_shortening_actual(utterance)
desired_output = 'Tu:çe .'
self.assertEqual(actual_output, desired_output)
def test_get_shortening_target_standard_case(self):
"""Test get_shortening_target with 1 shortening occurence."""
utterance = 'na:(ra)da <dükäm lan> [?] [>] ?'
actual_output = Extractor.get_shortening_target(utterance)
desired_output = 'na:rada <dükäm lan> [?] [>] ?'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_target_correction_several_words(self):
"""Test get_retracing_target."""
utterance = '<this us> [//] this is a test'
actual_output = Extractor.get_retracing_target(utterance)
desired_output = 'this us this is a test'
self.assertEqual(actual_output, desired_output)
def test_get_shortening_target_non_shortening_parentheses(self):
"""Test get_shortening_target with non shortening parentheses."""
utterance = 'mo:(ra)da (.) mu ?'
actual_output = Extractor.get_shortening_target(utterance)
desired_output = 'mo:rada (.) mu ?'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_target_reformulation_several_words(self):
"""Test get_retracing_target."""
utterance = '<for what> [///] why do you eat'
actual_output = Extractor.get_retracing_target(utterance)
desired_output = 'for what why do you eat'
self.assertEqual(actual_output, desired_output)
def test_get_shortening_target_no_shortening(self):
"""Test get_shortening_target using utt without a shortening."""
utterance = 'Mu:ça .'
actual_output = Extractor.get_shortening_target(utterance)
desired_output = 'Mu:ça .'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_target_false_start(self):
"""Test get_retracing_target."""
utterance = 'I want [/-] would do that'
actual_output = Extractor.get_retracing_target(utterance)
desired_output = 'I want would do that'
self.assertEqual(actual_output, desired_output)
def test_get_replacement_actual_one_replacement(self):
"""Test get_replacement_actual with 1 replacement."""
utterance = 'yarasam [: yorosom] .'
actual_output = Extractor.get_replacement_actual(utterance)
desired_output = 'yarasam .'
self.assertEqual(actual_output, desired_output)
def test_get_retracing_target_false_start_several_words(self):
"""Test get_retracing_target."""
utterance = '<I want> [///] what do you eat'
actual_output = Extractor.get_retracing_target(utterance)
desired_output = 'I want what do you eat'
self.assertEqual(actual_output, desired_output)