conversion_utils/tests/test_jos_msd_to_properties.py

import os.path
import lxml.etree as lxml
import unittest

from conversion_utils.jos_msds_and_properties import SpecificationsBuilder, MsdToProperties, Msd

class JosMsdToPropertiesTestCase(unittest.TestCase):

    def setUp(self):
        specifications_file_name = os.path.join(os.path.dirname(__file__), '../resources/msd-sl.spc.xml')
        builder = SpecificationsBuilder()
        self.specifications = builder.build(specifications_file_name)
        self.converter = MsdToProperties()

    def test_en_en(self):
        properties = self.converter.convert(self.specifications, Msd('Ncfpd', 'en'), 'miza', 'en')
        self.assertEqual(properties.language, 'en')
        self.assertEqual(properties.category, 'noun')
        self.assertEqual(properties.lexeme_feature_map, {'type':'common', 'gender':'feminine'})
        self.assertEqual(properties.form_feature_map, {'number':'plural', 'case':'dative'})

    def test_en_sl(self):
        properties = self.converter.convert(self.specifications, Msd('Ncfpd', 'en'), 'miza', 'sl')
        self.assertEqual(properties.language, 'sl')
        self.assertEqual(properties.category, 'samostalnik')
        self.assertEqual(properties.lexeme_feature_map, {'vrsta':'občno_ime', 'spol':'ženski'})
        self.assertEqual(properties.form_feature_map, {'število':'množina', 'sklon':'dajalnik'})

    def test_sl_en(self):
        properties = self.converter.convert(self.specifications, Msd('Sozmd', 'sl'), 'miza', 'en')
        self.assertEqual(properties.language, 'en')
        self.assertEqual(properties.category, 'noun')
        self.assertEqual(properties.lexeme_feature_map, {'type':'common', 'gender':'feminine'})
        self.assertEqual(properties.form_feature_map, {'number':'plural', 'case':'dative'})

    def test_sl_sl(self):
        properties = self.converter.convert(self.specifications, Msd('Sozmd', 'sl'), 'miza', 'sl')
        self.assertEqual(properties.language, 'sl')
        self.assertEqual(properties.category, 'samostalnik')
        self.assertEqual(properties.lexeme_feature_map, {'vrsta':'občno_ime', 'spol':'ženski'})
        self.assertEqual(properties.form_feature_map, {'število':'množina', 'sklon':'dajalnik'})

    def test_exception_feature_level(self): # testing level exception: ('pronoun', 8, 'ti')
        properties = self.converter.convert(self.specifications, Msd('Pp2-sd--y', 'en'), 'ti', 'sl')
        self.assertEqual(properties.language, 'sl')
        self.assertEqual(properties.category, 'zaimek')
        self.assertEqual(properties.lexeme_feature_map, {'vrsta':'osebni', 'oseba':'druga'})
        self.assertEqual(properties.form_feature_map, {'število':'ednina', 'sklon':'dajalnik', 'naslonskost':'klitična'})

    def test_normal_feature_level(self): # invalid msd, but useful for testing contrast with previous test
        properties = self.converter.convert(self.specifications, Msd('Pp2-pd--y', 'en'), 'vi', 'sl')
        self.assertEqual(properties.language, 'sl')
        self.assertEqual(properties.category, 'zaimek')
        self.assertEqual(properties.lexeme_feature_map, {'vrsta':'osebni', 'oseba':'druga', 'naslonskost':'klitična'})
        self.assertEqual(properties.form_feature_map, {'število':'množina', 'sklon':'dajalnik'})

    def test_featureless(self):
        properties = self.converter.convert(self.specifications, Msd('V', 'sl'), 'in', 'en')
        self.assertEqual(properties.language, 'en')
        self.assertEqual(properties.category, 'conjunction')
        self.assertEqual(properties.lexeme_feature_map, {})
        self.assertEqual(properties.form_feature_map, {})