def test_retokenize(self, conjugation):
doc = Doc(TEXT, conjugation)
doc.retokenize()
assert doc.get_text() == TEXT
text = "今日はいい天気ですね。"
doc.retokenize(text)
assert doc.get_text() == text
def test():
doc = Doc("本を書きました。")
# print surface forms of the tokens.
surfaces = [word.surface for word in doc.words]
print("/".join(surfaces)) # 本/を/書き/まし/た/。
# print plain text
print(doc.get_text()) # 本を書きました。
# delete a word
doc.delete(3) # Word conjugation will be done as needed.
print(doc.get_text()) # 本を書いた。
# update a word
word = doc.conjugation.tokenize("読む")
# In addition to conjugation, transform the peripheral words as needed.
doc.update(2, word)
print(doc.get_text()) # 本を読んだ。
def test_delete(self, conjugation, text, interval, expect):
doc = Doc(text, conjugation)
doc.delete(interval)
assert doc.get_text() == expect
def test_insert(self, conjugation, text, surfaces, i, expect):
doc = Doc(text, conjugation)
words = conjugation.tokenize(surfaces)
doc.insert(i, words)
assert doc.get_text() == expect
def test_insert_a_single_word(self, conjugation):
doc = Doc(TEXT, conjugation)
text = "公園"
word = conjugation.tokenize(text)[0]
doc.insert(1, word)
assert doc.get_text() == SURFACES[0] + text + "".join(SURFACES[1:])
def test_conjugate_word_in_doc(self, conjugation, text, i, c_form, expect):
doc = Doc(text, conjugation)
doc.conjugate(i, c_form)
assert doc.get_text() == expect
def test_should_be_able_to_extract_text(self, conjugation, interval, expect):
doc = Doc(TEXT, conjugation)
assert doc.get_text(interval) == expect
def test_should_be_able_to_initialize_with_only_a_string(self):
doc = Doc(TEXT)
assert doc is not None
def test_should_be_able_to_initialize_with_string(self, conjugation):
doc = Doc(TEXT, conjugation)
surfaces = [word.surface for word in doc.words]
assert surfaces == SURFACES
def test_to_word_list(self):
doc = Doc(TEXT)
for dic in doc.to_word_list():
assert type(dic) == dict
def test_simple_view(self):
doc = Doc(TEXT)
assert len(doc.simple_view()) > 0
def test_update_surfaces(self, conjugation, text, interval, surfaces, expect):
doc = Doc(text, conjugation)
doc.update_surfaces(interval, surfaces)
assert doc.get_text() == expect
def test_update(self, conjugation, text, interval, surfaces, expect):
doc = Doc(text, conjugation)
words = conjugation.tokenize(surfaces)
doc.update(interval, words)
assert doc.get_text() == expect
def test_update_a_single_word(self, conjugation):
doc = Doc(TEXT, conjugation)
text = "毎週"
word = conjugation.tokenize(text)[0]
doc.update(0, word)
assert doc.get_text() == "毎週" + "".join(SURFACES[1:])