def create_main_feature(qid, word, token, text, entities, tags):
tmp_bow = {}
tmp_bow['qid'] = qid
tmp_bow['TIENE_RAIZ'] = has_root(word)
tmp_bow['FULL_MAYUSCULAS'] = full_upper(word)
tmp_bow['FULL_MINUSCULAS'] = full_lower(word)
tmp_bow['INICIO_MAYUSCULAS_RESTO_MINUSCULAS'] = is_capitalized(word)
# print word.encode('utf-8')
cat = sin_cat(word, tags)
tmp_bow['CAT_SINTACTICA_' + cat] = 1
tmp_bow['PALABRA_'+word] = 1
tmp_bow['PALABRA_LARGO'] = word_len(word)
tmp_bow['ES_TOKEN'] = token
return tmp_bow
def test_is_not_capitalized(self):
word_for_test = "WAs"
result = is_capitalized(word_for_test)
self.assertEqual(0, result)
def test_is_capitalized_one_letter(self):
word_for_test = "I"
result = is_capitalized(word_for_test)
self.assertEqual(1, result)