def test_reload(self):
storage_factory = StorageFactory()
file_store_config = FileStorageConfiguration()
file_store_config._gender_storage = FileStoreConfiguration(
file=os.path.dirname(__file__) + os.sep + "test_files" + os.sep +
"gender.txt",
format="text",
extension="txt",
encoding="utf-8",
delete_on_start=False)
storage_engine = FileStorageEngine(file_store_config)
storage_factory._storage_engines[
StorageFactory.GENDER] = storage_engine
storage_factory._store_to_engine_map[
StorageFactory.GENDER] = storage_engine
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.load(storage_factory)
self.assertEqual(collection.gender("WITH HIM"), 'with her')
self.assertEqual(collection.genderise_string(None, "This is with him"),
"This is with her")
collection.reload(storage_factory)
self.assertEqual(collection.gender("WITH HIM"), 'with her')
self.assertEqual(collection.genderise_string(None, "This is with him"),
"This is with her")
def test_reload_jp(self):
storage_factory = StorageFactory()
tokenizer = TokenizerJP()
file_store_config = FileStorageConfiguration()
file_store_config._gender_storage = FileStoreConfiguration(
file=os.path.dirname(__file__) + os.sep + "test_files" + os.sep +
"gender_jp.txt",
format="text",
extension="txt",
encoding="utf-8",
delete_on_start=False)
storage_engine = FileStorageEngine(file_store_config)
storage_factory._storage_engines[
StorageFactory.GENDER] = storage_engine
storage_factory._store_to_engine_map[
StorageFactory.GENDER] = storage_engine
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.load(storage_factory)
self.assertEqual(collection.gender("彼"), '彼女')
self.assertEqual(collection.genderise_string(tokenizer, "彼が来た"),
"彼女が来た")
collection.reload(storage_factory)
self.assertEqual(collection.gender("彼"), '彼女')
self.assertEqual(collection.genderise_string(tokenizer, "彼が来た"),
"彼女が来た")
def test_collection(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
count = collection.load_from_text("""
" with him "," with her "
" with her "," with him "
" to him "," to her "
" to her "," to him "
" on him "," on her "
" on her "," on him "
" in him "," in her "
" in her "," in him "
" for him "," for her "
" for her "," for him "
" he "," she "
" his "," her "
" him "," her "
" her "," his "
" she "," he "
""")
self.assertEqual(count, 15)
self.assertEqual(collection.genderise_string("This is with him "),
"This is with her")
self.assertEqual("(^with him | with him | with him$)",
collection.gender(" with him "))
self.assertIsNone(collection.gender("unknown"))
def assert_upload_from_text(self, store):
store.empty()
store.upload_from_text(
None, """
" with him "," with her "
" with her "," with him "
" to him "," to her "
" to her "," to him "
" on him "," on her "
" on her "," on him "
" in him "," in her "
" in her "," in him "
" for him "," for her "
" for her "," for him "
" he "," she "
" his "," her "
" him "," her "
" her "," his "
" she "," he "
""")
collection = GenderCollection()
store.load(collection)
self.assertEqual(collection.gender(" WITH HIM "), [
re.compile('(^WITH HIM | WITH HIM | WITH HIM$)', re.IGNORECASE),
' WITH HER '
])
self.assertEqual(collection.genderise_string("This is with him "),
"This is with her")
def test_collection(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
count = collection.load_from_text("""
" with him "," with her "
" with her "," with him "
" to him "," to her "
" to her "," to him "
" on him "," on her "
" on her "," on him "
" in him "," in her "
" in her "," in him "
" for him "," for her "
" for her "," for him "
" he "," she "
" his "," her "
" him "," her "
" her "," his "
" she "," he "
""")
self.assertEqual(count, 15)
self.assertEqual(collection.genderise_string("This is with him "), "This is with her")
#self.assertEqual("(^with him | with him | with him$)", collection.gender(" with him "))
self.assertIsNone(collection.gender("unknown"))
def test_collection_duplicate(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("with him", 'with her')
collection.add_to_lookup("with him", 'with you')
self.assertEqual(collection.gender("WITH HIM"), 'with her')
def test_collection_duplicate_jp(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("彼", '彼女')
collection.add_to_lookup("彼", '彼氏')
self.assertEqual(collection.gender("彼"), '彼女')
def assert_upload_csv_file(self, store, filename):
store.empty()
store.upload_from_file(filename, format=Store.CSV_FORMAT)
collection = GenderCollection()
store.load(collection)
self.assertEqual(collection.gender(" WITH HIM "), [re.compile('(^WITH HIM | WITH HIM | WITH HIM$)', re.IGNORECASE), ' WITH HER '])
self.assertEqual(collection.genderise_string("This is with him "), "This is with her")
def assert_upload_from_text_file(self, store):
store.empty()
store.upload_from_file(os.path.dirname(__file__) + os.sep + "data" + os.sep + "lookups" + os.sep + "text" + os.sep + "gender.txt")
collection = GenderCollection()
store.load(collection)
self.assertEqual(collection.gender(" WITH HIM "), [re.compile('(^WITH HIM | WITH HIM | WITH HIM$)', re.IGNORECASE), ' WITH HER '])
self.assertEqual(collection.genderise_string("This is with him "), "This is with her")
def test_collection_operations(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup(" WITH HIM ", [re.compile('(^WITH HIM | WITH HIM | WITH HIM$)', re.IGNORECASE), ' WITH HER '])
self.assertTrue(collection.has_key(" WITH HIM "))
self.assertEqual([re.compile('(^WITH HIM | WITH HIM | WITH HIM$)', re.IGNORECASE), ' WITH HER '], collection.value(" WITH HIM "))
self.assertEqual(collection.gender(" WITH HIM "), [re.compile('(^WITH HIM | WITH HIM | WITH HIM$)', re.IGNORECASE), ' WITH HER '])
self.assertEqual(collection.genderise_string("This is with him "), "This is with her")
def test_collection_invalid(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("with him", 'with her')
self.assertFalse(collection.has_keyVal("WITH YOU"))
self.assertIsNone(collection.value("WITH YOU"))
self.assertIsNone(collection.gender("WITH YOU"))
self.assertEqual(collection.genderise_string(None, "This is with you"),
"This is with you")
def test_collection_operations(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("with him", 'with her')
self.assertTrue(collection.has_keyVal("WITH HIM"))
self.assertEqual('with her', collection.value("WITH HIM"))
self.assertEqual(collection.gender("WITH HIM"), 'with her')
self.assertEqual(collection.genderise_string(None, "This is with him"),
"This is with her")
def test_load_from_file(self):
config = FileStorageConfiguration()
config._gender_storage = FileStoreConfiguration(file=os.path.dirname(__file__) + os.sep + "data" + os.sep + "lookups" + os.sep + "text" + os.sep + "gender.txt", fileformat="text", encoding="utf-8", delete_on_start=False)
engine = FileStorageEngine(config)
engine.initialise()
store = FileGenderStore(engine)
gender_collection = GenderCollection()
store.load(gender_collection)
self.assertEqual(gender_collection.gender(" WITH HIM "), [re.compile('(^WITH HIM | WITH HIM | WITH HIM$)', re.IGNORECASE), ' WITH HER '])
self.assertEqual(gender_collection.genderise_string("This is with him "), "This is with her")
def test_collection_invalid_JP(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("彼", '彼女')
self.assertFalse(collection.has_keyVal("彼氏"))
self.assertIsNone(collection.value("彼氏"))
tokenizer = TokenizerJP()
self.assertIsNone(collection.gender("彼氏"))
self.assertEqual(collection.genderise_string(tokenizer, "彼氏が来た"),
"彼氏が来た")
def test_collection_operations_JP(self):
collection = GenderCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("彼", '彼女')
tokenizer = TokenizerJP()
self.assertTrue(collection.has_keyVal("彼"))
self.assertEqual('彼女', collection.value("彼"))
self.assertEqual(collection.gender("彼"), '彼女')
self.assertEqual(collection.genderise_string(tokenizer, "彼が来た"),
"彼女が来た")
