def test_invalid_format(self):
expected_language = languages.getlang_by_name('English')
converter = build_subtitle_converter_from_file(os.path.join(test_files_dir, 'not.txt'))
with self.assertRaises(InvalidSubtitleFormatError):
converter.convert(expected_language.code)
def test_invalid_format__empty(self):
expected_language = languages.getlang_by_name('English')
converter = build_subtitle_converter_from_file(os.path.join(test_files_dir, 'empty.ttml'))
with self.assertRaises(InvalidSubtitleFormatError, msg='Caption file is empty'):
converter.convert(expected_language.code)
def test_invalid_language(self):
expected_language = languages.getlang_by_name('Spanish')
converter = build_subtitle_converter_from_file(
os.path.join(test_files_dir, 'encapsulated.sami'))
with self.assertRaises(InvalidSubtitleLanguageError):
converter.convert(expected_language.code)
def test_not_expected_type(self):
expected_format = file_formats.SCC
expected_language = languages.getlang_by_name('Arabic')
converter = build_subtitle_converter_from_file(
os.path.join(test_files_dir, 'basic.srt'), in_format=expected_format)
with self.assertRaises(InvalidSubtitleFormatError):
converter.convert(expected_language.code)
def test_srt_conversion(self):
expected_file = os.path.join(test_files_dir, 'basic.vtt')
expected_language = languages.getlang_by_name('Arabic')
converter = build_subtitle_converter_from_file(os.path.join(test_files_dir, 'basic.srt'))
converter.replace_unknown_language(expected_language.code)
with tempfile.NamedTemporaryFile() as actual_file:
converter.write(actual_file.name, expected_language.code)
self.assertFileHashesEqual(expected_file, actual_file.name)
def test_replace_unknown_language(self):
expected_language = languages.getlang_by_name('Arabic')
converter = build_subtitle_converter_from_file(os.path.join(test_files_dir, 'basic.srt'))
self.assertTrue(converter.has_language(LANGUAGE_CODE_UNKNOWN))
converter.replace_unknown_language(expected_language.code)
self.assertTrue(converter.has_language(expected_language.code))
self.assertFalse(converter.has_language(LANGUAGE_CODE_UNKNOWN))
def test_valid_language(self):
expected_file = os.path.join(test_files_dir, 'encapsulated.vtt')
expected_language = languages.getlang_by_name('English')
converter = build_subtitle_converter_from_file(
os.path.join(test_files_dir, 'encapsulated.sami'))
self.assertTrue(converter.has_language(expected_language.code))
with tempfile.NamedTemporaryFile() as actual_file:
converter.write(actual_file.name, expected_language.code)
self.assertFileHashesEqual(expected_file, actual_file.name)
def download_and_transform_file(self, path):
"""
Download subtitles file at `path` and transform it to `.vtt` if necessary.
Args: path (URL or local path)
Returns: filename of final .vtt file
"""
key = "DOWNLOAD:{}".format(path)
cache_file = get_cache_filename(key)
if not config.UPDATE and not cache_is_outdated(path, cache_file):
return cache_file
config.LOGGER.info("\tDownloading {}".format(path))
with tempfile.NamedTemporaryFile() as temp_in_file,\
tempfile.NamedTemporaryFile() as temp_out_file:
write_and_get_hash(path, temp_in_file)
temp_in_file.seek(0)
converter = build_subtitle_converter_from_file(
temp_in_file.name, self.subtitlesformat)
# We'll assume the provided file is in the passed language in this case
if len(converter.get_language_codes()) == 1 \
and converter.has_language(LANGUAGE_CODE_UNKNOWN):
converter.replace_unknown_language(self.language)
convert_lang_code = self.language
# Language is not present, let's try different codes
if not converter.has_language(self.language):
for lang_code in converter.get_language_codes():
language = languages.getlang_by_alpha2(lang_code)
if language and language.code == self.language:
convert_lang_code = lang_code
break
else:
raise InvalidSubtitleLanguageError(
"Missing language '{}' in subtitle file".format(
self.language))
converter.write(temp_out_file.name, convert_lang_code)
temp_out_file.seek(0)
file_hash = get_hash(temp_out_file.name)
filename = '{0}.{ext}'.format(file_hash, ext=file_formats.VTT)
temp_out_file.seek(0)
copy_file_to_storage(filename, temp_out_file)
FILECACHE.set(key, bytes(filename, "utf-8"))
return filename