def test_create_tokens_with_profanity():
"""Testing that the create_tokens function will not create
a token from profanity"""
list_responses = ['test this code', ' for bad words', 'such as shit']
check = edurate_gensim.create_tokens(list_responses)
assert check == [['test', 'code'], ['bad', 'words']]
assert ("shit" in check) is False
def test_create_tokens_with_no_repeating_words():
"""Testing that the create_tokens function will create tokens
from a list of strings. A given word must have a length
of at least 3 characters to be considered a token. In this
test, we have zero repeating words"""
list_responses = ['I am testing', 'this is a test', 'make my tokens']
check = edurate_gensim.create_tokens(list_responses)
assert check == [['testing'], ['test'], ['make', 'tokens']]
assert ("am" in check) is False
assert ("my" in check) is False
def test_create_tokens_with_repeating_words():
"""Testing the create_tokens function works properly when
given a list with repeating words."""
list_responses = [
'I am testing', 'testing testing testing', 'make my tokens'
]
check = edurate_gensim.create_tokens(list_responses)
assert check == [['testing'], ['testing', 'testing', 'testing'],
['make', 'tokens']]
assert ("I" in check) is False
assert ("am" in check) is False
assert ("my" in check) is False