def test_prepare_drop_hyphen_and_apostrophe_at_start_or_end_word(self):
self.stream = StringIO(
'date some- \'text ’date text'
)
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual({'some': 1, 'text': 2, 'date': 2}, words)
def test_get_content(self):
# stream = open("myfile.txt", "r", encoding="utf-8")
# stream = io.StringIO("some initial text data")
text = 'Some text date.'
self.stream = StringIO(text)
dic = DictionaryForText(self.stream)
self.assertEqual(text, dic.text)
def test_prepare_simple(self):
self.stream = StringIO('text date \nand more text')
dic = DictionaryForText(self.stream)
self.assertEqual(
{'text': 2, 'date': 1, 'and': 1, 'more': 1},
dic.prepare()
)
def test_get_content_after_close_stream(self):
text = 'Some text date.'
self.stream = StringIO(text)
dic = DictionaryForText(self.stream)
self.stream.close()
self.assertEqual(text, dic.text)
def test_prepare_text_drop_short_words(self):
self.stream = StringIO(
'cat is a word'
)
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual({'cat': 1, 'word': 1}, words)
self.assertEqual({'is': 1, 'a': 1}, dic.get_drop_short())
def test_prepare_text_with_punctuation_mark_at_end(self):
self.stream = StringIO(
'text, date. and? more! text: text; text…; text — text.'
)
dic = DictionaryForText(self.stream)
self.assertEqual(
{'text': 6, 'date': 1, 'and': 1, 'more': 1},
dic.prepare()
)
def test_prepare_drop_end_apostrophe_s(self):
self.stream = StringIO('that cat and this cat are cat\'s cats')
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual(
{'that': 1, 'cat': 4, 'and': 1, 'this': 1, 'are': 1},
words
)
self.assertEqual({'cat\'s': 'cat'}, dic.get_drop_end_apostrophe_s())
def test_prepare_drop_end_s(self):
self.stream = StringIO('two cats are cat and cat')
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual(
{'two': 1, 'cat': 3, 'are': 1, 'and': 1},
words
)
self.assertEqual({'cats': 'cat'}, dic.get_drop_end_s())
def test_prepare_drop_proper_name(self):
self.stream = StringIO('Two cats are two tails, Murzik and Venik.')
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual(
{'two': 2, 'cats': 1, 'are': 1, 'tails': 1, 'and': 1},
words
)
self.assertEqual({'Murzik': 1, 'Venik': 1}, dic.get_drop_proper_name())
def test_prepare_text_with_punctuation_marks_and_other_symbols(self):
self.stream = StringIO('''
text, «date». and? {more}! [text]: "text"; text…; (text)
“text” — text 2date
''')
dic = DictionaryForText(self.stream)
self.assertEqual(
{'text': 7, 'date': 2, 'and': 1, 'more': 1},
dic.prepare()
)
def test_prepare_drop_end_es(self):
self.stream = StringIO(
'classes are object of class class'
)
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual(
{'are': 1, 'object': 1, 'class': 3},
words
)
self.assertEqual({'classes': 'class'}, dic.get_drop_end_es())
def test_prepare_drop_end_ed(self):
self.stream = StringIO(
'the call called. create it. it will be created.'
)
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual(2, words['call'])
self.assertEqual(2, words['create'])
self.assertFalse('called' in words)
self.assertFalse('created' in words)
self.assertEqual(
{'called': 'call', 'created': 'create'}, dic.get_drop_end_ed()
)
def test_prepare_drop_end_ing(self):
self.stream = StringIO(
'th the thing are running during I am run too. Music is dur'
)
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual(1, words['thing'])
self.assertEqual(1, words['during'])
self.assertEqual(2, words['run'])
self.assertFalse('running' in words)
self.assertEqual(
{'running': 'run'}, dic.get_drop_end_ing()
)
def test_prepare_drop_end_ies(self):
self.stream = StringIO(
'those goodies are one goody for you and one one for you'
)
dic = DictionaryForText(self.stream)
words = dic.prepare()
self.assertEqual(
{
'those': 1, 'are': 1, 'one': 3, 'goody': 2, 'for': 2,
'you': 2, 'and': 1
},
words
)
self.assertEqual({'goodies': 'goody'}, dic.get_drop_end_ies())