def test_tokenize(): result = tokenize(tokenizer, "public void main()") estimated_result = ['public', 'void', 'main', '(', ')'] assert check_torknized_str(result, estimated_result) estimated_result = ['String', 'str', '=', r'"\""'] result = tokenize(tokenizer, r'String str="\""') assert check_torknized_str(result, estimated_result)
def test_create_two_shingles(): seq = tokenize(tokenizer, "public void main()")[1:] result = get_shingles_str(create_two_shingles(seq)) estimated_result = {('public', 'void'), ('void', 'main'), ('main', '('), ('(', ')')} assert result == estimated_result seq = tokenize(tokenizer, "public int main()")[1:] result = get_shingles_str(create_two_shingles(seq)) estimated_result = {('public', 'int'), ('int', 'main'), ('main', '('), ('(', ')')} assert result == estimated_result
def test_tokenize(): result = tokenize(tokenizer, "public void main()") result = tokenize(tokenizer, r'String str="\""') assert False == result
def test_create_two_shingles(): seq = tokenize(tokenizer, "public void main()")[1:] result = create_two_shingles(seq) seq = tokenize(tokenizer, "public int main()")[1:] result = create_two_shingles(seq) assert False == result