def test_provide_oov006(self):
cinfo = MeCabOovPlugin.CategoryInfo()
cinfo.type_ = CategoryType.KANJI
cinfo.is_invoke = False
cinfo.is_group = False
cinfo.length = 6
self.plugin.categories[CategoryType.KANJI] = cinfo
mock_inputtext.set_category_type(0, 3, CategoryType.KANJI)
nodes = self.plugin.provide_oov(self.mocked_input_text, 0, False)
self.assertEqual(3, len(nodes))
n = nodes[0]
self.assertEqual('あ', n.get_word_info().surface)
self.assertEqual(1, n.get_word_info().length())
self.assertEqual(1, n.get_word_info().pos_id)
n = nodes[1]
self.assertEqual('あい', n.get_word_info().surface)
self.assertEqual(2, n.get_word_info().length())
self.assertEqual(1, n.get_word_info().pos_id)
n = nodes[2]
self.assertEqual('あいう', n.get_word_info().surface)
self.assertEqual(3, n.get_word_info().length())
self.assertEqual(1, n.get_word_info().pos_id)
nodes = self.plugin.provide_oov(self.mocked_input_text, 0, True)
self.assertEqual(0, len(nodes))
def test_read_oov_with_category_not_in_character_property(self):
input_ = os.path.join(self.test_dir, 'test.txt')
with open(input_, 'w') as wf:
wf.write("ALPHA,1,2,3,補助記号,一般,*,*,*,*\n")
plugin = MeCabOovPlugin()
plugin.categories[CategoryType.DEFAULT] = MeCabOovPlugin.CategoryInfo()
with self.assertRaises(ValueError) as cm:
plugin.read_oov(input_, mock_grammar.mocked_grammar)
self.assertEqual('`ALPHA` is undefined at line 1', cm.exception.args[0])
def test_read_oov_with_undefined_type(self):
input_ = os.path.join(self.test_dir, 'test.txt')
with open(input_, 'w') as wf:
wf.write("FOO,1,2,3,補助記号,一般,*,*,*,*\n")
plugin = MeCabOovPlugin()
plugin.categories[CategoryType.DEFAULT] = MeCabOovPlugin.CategoryInfo()
with self.assertRaises(ValueError) as cm:
plugin.read_oov(input_, mock_grammar.mocked_grammar)
self.assertEqual('`FOO` is invalid type at line 1', cm.exception.args[0])
def test_read_oov_with_too_few_columns(self):
input_ = os.path.join(self.test_dir, 'test.txt')
with open(input_, 'w') as wf:
wf.write("DEFAULT,1,2,3\n")
plugin = MeCabOovPlugin()
plugin.categories[CategoryType.DEFAULT] = MeCabOovPlugin.CategoryInfo()
with self.assertRaises(ValueError) as cm:
plugin.read_oov(input_, mock_grammar.mocked_grammar)
self.assertEqual('invalid format at line 1', cm.exception.args[0])
def test_provide_oov_without_oov_list(self):
cinfo = MeCabOovPlugin.CategoryInfo()
cinfo.type_ = CategoryType.HIRAGANA
cinfo.is_invoke = False
cinfo.is_group = True
cinfo.length = 0
self.plugin.categories[CategoryType.HIRAGANA] = cinfo
mock_inputtext.set_category_type(0, 3, CategoryType.HIRAGANA)
nodes = self.plugin.provide_oov(self.mocked_input_text, 0, False)
self.assertEqual(0, len(nodes))
def test_read_oov(self):
oov = os.path.join(self.test_dir, 'test.txt')
with open(oov, 'w') as wf:
wf.write("DEFAULT,1,2,3,補助記号,一般,*,*,*,*\n")
wf.write("DEFAULT,3,4,5,補助記号,一般,*,*,*,*\n")
plugin = MeCabOovPlugin()
plugin.categories[CategoryType.DEFAULT] = MeCabOovPlugin.CategoryInfo()
plugin.read_oov(oov, mock_grammar.mocked_grammar)
self.assertEqual(1, len(plugin.oov_list))
self.assertEqual(2, len(plugin.oov_list[CategoryType.DEFAULT]))
self.assertEqual(1, plugin.oov_list[CategoryType.DEFAULT][0].left_id)
self.assertEqual(2, plugin.oov_list[CategoryType.DEFAULT][0].right_id)
self.assertEqual(3, plugin.oov_list[CategoryType.DEFAULT][0].cost)
self.assertEqual(0, plugin.oov_list[CategoryType.DEFAULT][0].pos_id)
def test_provide_oov100(self):
cinfo = MeCabOovPlugin.CategoryInfo()
cinfo.type_ = CategoryType.KANJI
cinfo.is_invoke = True
cinfo.is_group = False
cinfo.length = 0
self.plugin.categories[CategoryType.KANJI] = cinfo
mock_inputtext.set_category_type(0, 3, CategoryType.KANJI)
nodes = self.plugin.provide_oov(self.mocked_input_text, 0, False)
self.assertEqual(0, len(nodes))
nodes = self.plugin.provide_oov(self.mocked_input_text, 0, True)
self.assertEqual(0, len(nodes))