def get_transcript_language_parameters(lang_code):
"""
Get parameters of a transcript's language, having checked on consistency with settings.
Arguments:
lang_code (str): Raw language code of a transcript, fetched from the video platform.
Returns:
lang_code (str): Pre-configured language code, e.g. 'br'
lang_label (str): Pre-configured language label, e.g. 'Breton'
"""
# Delete region subtags
# Reference: https://github.com/edx/edx-platform/blob/release-2017-02-16-12.24/lms/envs/common.py#L861
lang_code = lang_code[0:2]
# Check on consistency with the pre-configured ALL_LANGUAGES
if lang_code not in [
language[0] for language in settings.ALL_LANGUAGES
]:
raise VideoXBlockException(
_('Not all the languages of transcripts fetched from video platform are consistent '
'with the pre-configured ALL_LANGUAGES'))
lang_label = [
language[1] for language in settings.ALL_LANGUAGES
if language[0] == lang_code
][0]
return lang_code, lang_label
def create_credentials(self):
"""
Mock `get_client_credentials` returned value.
"""
if self.event == 'auth_failed':
self.side_effect = VideoXBlockException(
self.ordered_results[self.event]['error_message'])
self.return_value = (self.ordered_results[self.event]['client_secret'],
self.ordered_results[self.event]['client_id'],
self.ordered_results[self.event]['error_message'])
return self
def download_default_transcript(self, url=None, language_code=None): # pylint: disable=unused-argument
"""
Download default transcript from a video platform API in WebVVT format.
Arguments:
url (str): Transcript download url.
Returns:
sub (unicode): Transcripts formatted per WebVTT format https://w3c.github.io/webvtt/
"""
log.debug("BC: downloading default transcript from url:{}".format(url))
if url is None:
raise VideoXBlockException(_('`url` parameter is required.'))
data = requests.get(url)
return remove_escaping(data.content)
def download_default_transcript(self, url=None, language_code=None): # pylint: disable=unused-argument
"""
Download default transcript from Youtube API and format it to WebVTT-like unicode.
Reference to `get_transcripts_from_youtube()`:
https://github.com/edx/edx-platform/blob/ecc3473d36b3c7a360e260f8962e21cb01eb1c39/common/lib/xmodule/xmodule/video_module/transcripts_utils.py#L122
"""
if url is None:
raise VideoXBlockException(_('`url` parameter is required.'))
utf8_parser = etree.XMLParser(encoding='utf-8')
data = requests.get(url)
xmltree = etree.fromstring(data.content, parser=utf8_parser)
sub = [
self.format_transcript_element(element, i)
for i, element in enumerate(xmltree, 1)
]
sub = "".join(sub)
sub = "WEBVTT\n\n" + str(sub) if "WEBVTT" not in sub else str(sub)
return sub
