How to use the sudachipy.plugin.oov.MeCabOovPlugin function in SudachiPy

To help you get started, we’ve selected a few SudachiPy examples, based on popular ways it is used in public projects.

Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.

github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
def test_provide_oov012(self):
        cinfo = MeCabOovPlugin.CategoryInfo()
        cinfo.type_ = CategoryType.KANJI
        cinfo.is_invoke = False
        cinfo.is_group = True
        cinfo.length = 2
        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('あいう', n.get_word_info().surface)
        self.assertEqual(3, n.get_word_info().length())
        self.assertEqual(1, n.get_word_info().pos_id)
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
def test_read_character_property_duplicate_definitions(self):
        input_ = os.path.join(self.test_dir, 'test.txt')
        with open(input_, 'w') as wf:
            wf.write("DEFAULT 0 1 2\nDEFAULT 1 1 2")
        plugin = MeCabOovPlugin()
        with self.assertRaises(ValueError) as cm:
            plugin.read_character_property(input_)
        self.assertEqual('`DEFAULT` is already defined at line 2', cm.exception.args[0])
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
def test_read_character_property_with_too_few_columns(self):
        input_ = os.path.join(self.test_dir, 'test.txt')
        with open(input_, 'w') as wf:
            wf.write("DEFAULT 0 1\n")
        plugin = MeCabOovPlugin()
        with self.assertRaises(ValueError) as cm:
            plugin.read_character_property(input_)
        self.assertEqual('invalid format at line 1', cm.exception.args[0])
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
def setUp(self):
        self.plugin = MeCabOovPlugin()
        oov1 = MeCabOovPlugin.OOV()
        oov1.pos_id = 1
        oov2 = MeCabOovPlugin.OOV()
        oov2.pos_id = 2
        self.plugin.oov_list[CategoryType.KANJI] = [oov1]
        self.plugin.oov_list[CategoryType.KANJINUMERIC] = [oov1, oov2]
        self.mocked_input_text = mock_inputtext.mocked_input_text
        mock_inputtext.set_text('あいうえお')

        self.test_dir = tempfile.mkdtemp()
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
def test_read_character_property(self):
        input_ = os.path.join(self.test_dir, 'test.txt')
        with open(input_, 'w') as wf:
            wf.write("#\n  \nDEFAULT 0 1 2\nALPHA 1 0 0\n0x0000...0x0002 ALPHA")
        plugin = MeCabOovPlugin()
        plugin.read_character_property(input_)
        self.assertFalse(plugin.categories[CategoryType.DEFAULT].is_invoke)
        self.assertTrue(plugin.categories[CategoryType.DEFAULT].is_group)
        self.assertEqual(2, plugin.categories[CategoryType.DEFAULT].length)
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
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)
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
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])
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
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])
github WorksApplications / SudachiPy / tests / plugin / test_mecab_oov_plugin.py View on Github external
def test_read_character_property_with_undefined_type(self):
        input_ = os.path.join(self.test_dir, 'test.txt')
        with open(input_, 'w') as wf:
            wf.write("FOO 0 1 2\n")
        plugin = MeCabOovPlugin()
        with self.assertRaises(ValueError) as cm:
            plugin.read_character_property(input_)
        self.assertEqual('`FOO` is invalid type at line 1', cm.exception.args[0])
github WorksApplications / SudachiPy / sudachipy / plugin / oov / utils.py View on Github external
def get_oov_plugin(json_obj) -> OovProviderPlugin:
    # In the future, users can define plugin by themselves
    try:
        if json_obj['class'] == 'sudachipy.plugin.oov.MeCabOovProviderPlugin':
            return MeCabOovPlugin(json_obj)
        if json_obj['class'] == 'sudachipy.plugin.oov.SimpleOovProviderPlugin':
            return SimpleOovPlugin(json_obj)
        raise ValueError('{} is invalid OovProviderPlugin class'.format(json_obj['class']))
    except KeyError:
        raise ValueError('config file is invalid format')