def test_find_lcs_length_calls_required_function(self, mock):
"""
Tests that find_lcs_length function
calls fill_lcs_matrix function
"""
patches_sentence = ('the', 'dog', 'is', 'running')
plagiarism_threshold = 0.3
find_lcs_length(patches_sentence, patches_sentence, plagiarism_threshold)
self.assertTrue(mock.called)
def test_find_lcs_length_reversed_behaviour(self):
"""
Tests that find_lcs_length function
can reverse input sentences params
"""
expected = 5
sentence_first = ('the', 'dog', 'is', 'running', 'inside', 'the', 'house')
sentence_second = ('the', 'cat', 'is', 'sleeping', 'inside', 'the', 'house')
plagiarism_threshold = 0.3
actual = find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
actual_reversed = find_lcs_length(sentence_second, sentence_first, plagiarism_threshold)
self.assertEqual(expected, actual)
self.assertEqual(expected, actual_reversed)
def test_find_lcs_length_empty_input(self):
"""
Tests that find_lcs_length function
can handle empty input params
"""
expected = 0
empty_sentence = ()
patches_sentence = ('a', 'boy', 'plays', 'with', 'ball')
plagiarism_threshold = 0.3
actual = find_lcs_length(empty_sentence, patches_sentence, plagiarism_threshold)
actual_reversed = find_lcs_length(patches_sentence, empty_sentence, plagiarism_threshold)
self.assertEqual(expected, actual)
self.assertEqual(expected, actual_reversed)
def test_find_lcs_length_different_sized_inputs(self):
"""
Tests that find_lcs_length function
can handle different sized token inputs
"""
expected = 3
sentence_first = ('the', 'dog', 'is', 'running', 'inside')
sentence_second = ('the', 'cat', 'is', 'sleeping', 'inside', 'the', 'house')
plagiarism_threshold = 0.3
actual = find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
actual_reversed = find_lcs_length(sentence_second, sentence_first, plagiarism_threshold)
self.assertEqual(expected, actual)
self.assertEqual(expected, actual_reversed)
def test_find_lcs_length_incorrect_inputs(self):
"""
Tests that find_lcs_length function
can handle incorrect inputs
"""
expected = -1
bad_inputs = [[], {}, '', 9.22, -1, 0, -6, None, True, (None, None)]
patches_sentence = ('the', 'dog', 'is', 'running')
plagiarism_threshold = 0.3
for bad_input in bad_inputs:
actual = find_lcs_length(bad_input, patches_sentence, plagiarism_threshold)
actual_reversed = find_lcs_length(patches_sentence, bad_input, plagiarism_threshold)
self.assertEqual(expected, actual)
self.assertEqual(expected, actual_reversed)
def test_find_lcs_length_output_check(self):
"""
Tests that find_lcs_length function
can generate correct output according to given params
"""
sentence_first = ('the', 'dog', 'is', 'running', 'here')
sentence_second = ('a', 'boy', 'plays', 'with', 'ball')
plagiarism_threshold = 0.3
actual = find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
self.assertTrue(isinstance(actual, int))
def test_find_lcs_length_threshold_behaviour(self):
"""
Tests that find_lcs_length function
can preprocess threshold inputs
"""
sentence_first = ('the', 'dog', 'is')
sentence_second = ('the', 'cat', 'is')
plagiarism_threshold = 0.3
not_expected = 2/3 # 2/3 < 0.3 = 0
actual = find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
self.assertNotEqual(not_expected, actual)
def test_find_lcs_length_incorrect_threshold(self):
"""
Tests that find_lcs_length function
can handle incorrect threshold input
"""
expected = -1
bad_inputs = [[], {}, '', -1, -6.34, -6, 1.2, None, True, (None, None)]
patches_sentence = ('the', 'dog', 'is', 'running')
for bad_input in bad_inputs:
actual = find_lcs_length(patches_sentence, patches_sentence, bad_input)
self.assertEqual(expected, actual)
def test_find_lcs_length_optimized_works_faster(self):
"""
Tests find_lcs_optimized function
can work faster than find_lcs_length function
"""
sentence_first = ('the', 'dog', 'is', 'running', 'here')
sentence_second = ('a', 'boy', 'plays', 'with', 'ball')
plagiarism_threshold = 0.3
start_time = timeit.default_timer()
find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
end_time = timeit.default_timer()
not_optimized = end_time - start_time
start_time_second = timeit.default_timer()
find_lcs_length_optimized(sentence_first, sentence_second,
plagiarism_threshold)
end_time_second = timeit.default_timer()
optimized = end_time_second - start_time_second
self.assertGreater(not_optimized, optimized)
def test_find_lcs_length_no_diff(self):
"""
Tests that find_lcs_length function
can handle fully different sentences
"""
expected = 0
sentence_first = ('the', 'dog', 'is', 'running', 'here')
sentence_second = ('a', 'boy', 'plays', 'with', 'ball')
plagiarism_threshold = 0.3
actual = find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
self.assertEqual(expected, actual)
def test_find_lcs_length_complex(self):
"""
Tests that find_lcs_length function
can handle complex input case
"""
expected = 5
sentence_first = ('the', 'dog', 'is', 'running', 'inside', 'the', 'house')
sentence_second = ('the', 'cat', 'is', 'sleeping', 'inside', 'the', 'house')
plagiarism_threshold = 0.3
actual = find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
self.assertEqual(expected, actual)
def test_find_lcs_length_ideal(self):
"""
Tests that find_lcs_matrix function
can handle simple input case
"""
expected = 2
sentence_first = ('the', 'dog', 'is', 'running')
sentence_second = ('the', 'cat', 'is', 'sleeping')
plagiarism_threshold = 0.3
actual = find_lcs_length(sentence_first, sentence_second, plagiarism_threshold)
self.assertEqual(expected, actual)
"""
Longest common subsequence implementation starter
"""
from lab_2 import main
if __name__ == "__main__":
original_text = '''The horse is running.
It is fast.'''
second_text = '''The cow is eating.
It is slow.'''
original_tokens = main.tokenize_by_lines(original_text)
print('Original tokens: ', original_tokens)
second_tokens = main.tokenize_by_lines(second_text)
print('Second tokens: ', second_tokens)
plagiarism_threshold = 0.3
zero_matrix = main.create_zero_matrix(len(original_tokens),
len(second_tokens))
lcs_matrix = main.fill_lcs_matrix(original_tokens[0], second_tokens[0])
print('LCS matrix: ', lcs_matrix)
lcs_length = main.find_lcs_length(original_tokens[0], second_tokens[0],
plagiarism_threshold)
print('LCS is', lcs_length)
RESULT = lcs_length
assert RESULT == lcs_length, "Not working"