def test_youtube_empty_text(self, mock_get):
mock_get.return_value = Mock(status_code=200, text='')
youtube_id = 'bad_youtube_id'
with self.assertRaises(
transcripts_utils.GetTranscriptsFromYouTubeException):
transcripts_utils.get_transcripts_from_youtube(
youtube_id, settings, translation)
def test_youtube_good_result(self):
response = textwrap.dedent("""<?xml version="1.0" encoding="utf-8" ?>
<transcript>
<text start="0" dur="0.27"></text>
<text start="0.27" dur="2.45">Test text 1.</text>
<text start="2.72">Test text 2.</text>
<text start="5.43" dur="1.73">Test text 3.</text>
</transcript>
""")
expected_transcripts = {
'start': [270, 2720, 5430],
'end': [2720, 2720, 7160],
'text': ['Test text 1.', 'Test text 2.', 'Test text 3.']
}
youtube_id = 'good_youtube_id'
with patch('xmodule.video_module.transcripts_utils.requests.get'
) as mock_get:
mock_get.return_value = Mock(status_code=200,
text=response,
content=response)
transcripts = transcripts_utils.get_transcripts_from_youtube(
youtube_id, settings, translation)
self.assertEqual(transcripts, expected_transcripts)
mock_get.assert_called_with('http://video.google.com/timedtext',
params={
'lang': 'en',
'v': 'good_youtube_id'
})
def __init__(self, course_key, youtube_id, lang):
self.course_key = course_key
self.youtube_id = youtube_id
# Download transcripts from YouTube
# Note: cribbed from common/lib/xmodule/xmodule/video_module/transcripts_utils.py download_youtube_subs()
self.settings = copy.deepcopy(settings)
self.settings.YOUTUBE['TEXT_API']['params']['lang'] = lang
try:
self.subs = transcripts_utils.get_transcripts_from_youtube(self.youtube_id, self.settings, ModuleI18nService())
self.lang = lang
except GetTranscriptsFromYouTubeException:
raise Exception("Can't receive transcripts from Youtube for %s." % self.youtube_id)
def test_youtube_good_result(self):
response = textwrap.dedent("""<?xml version="1.0" encoding="utf-8" ?>
<transcript>
<text start="0" dur="0.27"></text>
<text start="0.27" dur="2.45">Test text 1.</text>
<text start="2.72">Test text 2.</text>
<text start="5.43" dur="1.73">Test text 3.</text>
</transcript>
""")
expected_transcripts = {
'start': [270, 2720, 5430],
'end': [2720, 2720, 7160],
'text': ['Test text 1.', 'Test text 2.', 'Test text 3.']
}
youtube_id = 'good_youtube_id'
with patch('xmodule.video_module.transcripts_utils.requests.get') as mock_get:
mock_get.return_value = Mock(status_code=200, text=response, content=response)
transcripts = transcripts_utils.get_transcripts_from_youtube(youtube_id, settings, translation)
self.assertEqual(transcripts, expected_transcripts)
mock_get.assert_called_with('http://video.google.com/timedtext', params={'lang': 'en', 'v': 'good_youtube_id'})
def test_youtube_good_result(self):
response = textwrap.dedent(
"""<?xml version="1.0" encoding="utf-8" ?>
<transcript>
<text start="0" dur="0.27"></text>
<text start="0.27" dur="2.45">Test text 1.</text>
<text start="2.72">Test text 2.</text>
<text start="5.43" dur="1.73">Test text 3.</text>
</transcript>
"""
)
expected_transcripts = {
"start": [270, 2720, 5430],
"end": [2720, 2720, 7160],
"text": ["Test text 1.", "Test text 2.", "Test text 3."],
}
youtube_id = "good_youtube_id"
with patch("xmodule.video_module.transcripts_utils.requests.get") as mock_get:
mock_get.return_value = Mock(status_code=200, text=response, content=response)
transcripts = transcripts_utils.get_transcripts_from_youtube(youtube_id, settings, translation)
self.assertEqual(transcripts, expected_transcripts)
mock_get.assert_called_with("http://video.google.com/timedtext", params={"lang": "en", "v": "good_youtube_id"})
def check_transcripts(request):
"""
Check state of transcripts availability.
request.GET['data'] has key `videos`, which can contain any of the following::
[
{u'type': u'youtube', u'video': u'OEoXaMPEzfM', u'mode': u'youtube'},
{u'type': u'html5', u'video': u'video1', u'mode': u'mp4'}
{u'type': u'html5', u'video': u'video2', u'mode': u'webm'}
]
`type` is youtube or html5
`video` is html5 or youtube video_id
`mode` is youtube, ,p4 or webm
Returns transcripts_presence dict::
html5_local: list of html5 ids, if subtitles exist locally for them;
is_youtube_mode: bool, if we have youtube_id, and as youtube mode is of higher priority, reflect this with flag;
youtube_local: bool, if youtube transcripts exist locally;
youtube_server: bool, if youtube transcripts exist on server;
youtube_diff: bool, if youtube transcripts exist on youtube server, and are different from local youtube ones;
current_item_subs: string, value of item.sub field;
status: string, 'Error' or 'Success';
subs: string, new value of item.sub field, that should be set in module;
command: string, action to front-end what to do and what to show to user.
"""
transcripts_presence = {
'html5_local': [],
'html5_equal': False,
'is_youtube_mode': False,
'youtube_local': False,
'youtube_server': False,
'youtube_diff': True,
'current_item_subs': None,
'status': 'Error',
}
try:
__, videos, item = _validate_transcripts_data(request)
except TranscriptsRequestValidationException as e:
return error_response(transcripts_presence, e.message)
transcripts_presence['status'] = 'Success'
filename = 'subs_{0}.srt.sjson'.format(item.sub)
content_location = StaticContent.compute_location(
item.location.org, item.location.course, filename
)
try:
local_transcripts = contentstore().find(content_location).data
transcripts_presence['current_item_subs'] = item.sub
except NotFoundError:
pass
# Check for youtube transcripts presence
youtube_id = videos.get('youtube', None)
if youtube_id:
transcripts_presence['is_youtube_mode'] = True
# youtube local
filename = 'subs_{0}.srt.sjson'.format(youtube_id)
content_location = StaticContent.compute_location(
item.location.org, item.location.course, filename
)
try:
local_transcripts = contentstore().find(content_location).data
transcripts_presence['youtube_local'] = True
except NotFoundError:
log.debug("Can't find transcripts in storage for youtube id: %s", youtube_id)
# youtube server
youtube_api = copy.deepcopy(settings.YOUTUBE_API)
youtube_api['params']['v'] = youtube_id
youtube_response = requests.get(youtube_api['url'], params=youtube_api['params'])
if youtube_response.status_code == 200 and youtube_response.text:
transcripts_presence['youtube_server'] = True
#check youtube local and server transcripts for equality
if transcripts_presence['youtube_server'] and transcripts_presence['youtube_local']:
try:
youtube_server_subs = get_transcripts_from_youtube(
youtube_id,
settings,
item.runtime.service(item, "i18n")
)
if json.loads(local_transcripts) == youtube_server_subs: # check transcripts for equality
transcripts_presence['youtube_diff'] = False
except GetTranscriptsFromYouTubeException:
pass
# Check for html5 local transcripts presence
html5_subs = []
for html5_id in videos['html5']:
filename = 'subs_{0}.srt.sjson'.format(html5_id)
content_location = StaticContent.compute_location(
item.location.org, item.location.course, filename
)
try:
html5_subs.append(contentstore().find(content_location).data)
transcripts_presence['html5_local'].append(html5_id)
except NotFoundError:
log.debug("Can't find transcripts in storage for non-youtube video_id: %s", html5_id)
if len(html5_subs) == 2: # check html5 transcripts for equality
transcripts_presence['html5_equal'] = json.loads(html5_subs[0]) == json.loads(html5_subs[1])
command, subs_to_use = _transcripts_logic(transcripts_presence, videos)
transcripts_presence.update({
'command': command,
'subs': subs_to_use,
})
return JsonResponse(transcripts_presence)
def get_transcripts_presence(videos, item):
""" fills in the transcripts_presence dictionary after for a given component
with its list of videos.
Returns transcripts_presence dict:
html5_local: list of html5 ids, if subtitles exist locally for them;
is_youtube_mode: bool, if we have youtube_id, and as youtube mode is of higher priority, reflect this with flag;
youtube_local: bool, if youtube transcripts exist locally;
youtube_server: bool, if youtube transcripts exist on server;
youtube_diff: bool, if youtube transcripts exist on youtube server, and are different from local youtube ones;
current_item_subs: string, value of item.sub field;
status: string, 'Error' or 'Success';
subs: string, new value of item.sub field, that should be set in module;
command: string, action to front-end what to do and what to show to user.
"""
transcripts_presence = {
'html5_local': [],
'html5_equal': False,
'is_youtube_mode': False,
'youtube_local': False,
'youtube_server': False,
'youtube_diff': True,
'current_item_subs': None,
'status': 'Success',
}
filename = 'subs_{0}.srt.sjson'.format(item.sub)
content_location = StaticContent.compute_location(item.location.course_key,
filename)
try:
local_transcripts = contentstore().find(content_location).data
transcripts_presence['current_item_subs'] = item.sub
except NotFoundError:
pass
# Check for youtube transcripts presence
youtube_id = videos.get('youtube', None)
if youtube_id:
transcripts_presence['is_youtube_mode'] = True
# youtube local
filename = 'subs_{0}.srt.sjson'.format(youtube_id)
content_location = StaticContent.compute_location(
item.location.course_key, filename)
try:
local_transcripts = contentstore().find(content_location).data
transcripts_presence['youtube_local'] = True
except NotFoundError:
log.debug("Can't find transcripts in storage for youtube id: %s",
youtube_id)
# youtube server
youtube_text_api = copy.deepcopy(settings.YOUTUBE['TEXT_API'])
youtube_text_api['params']['v'] = youtube_id
youtube_transcript_name = youtube_video_transcript_name(
youtube_text_api)
if youtube_transcript_name:
youtube_text_api['params']['name'] = youtube_transcript_name
youtube_response = requests.get('http://' + youtube_text_api['url'],
params=youtube_text_api['params'])
if youtube_response.status_code == 200 and youtube_response.text:
transcripts_presence['youtube_server'] = True
#check youtube local and server transcripts for equality
if transcripts_presence['youtube_server'] and transcripts_presence[
'youtube_local']:
try:
youtube_server_subs = get_transcripts_from_youtube(
youtube_id, settings, item.runtime.service(item, "i18n"))
if json.loads(
local_transcripts
) == youtube_server_subs: # check transcripts for equality
transcripts_presence['youtube_diff'] = False
except GetTranscriptsFromYouTubeException:
pass
# Check for html5 local transcripts presence
html5_subs = []
for html5_id in videos['html5']:
filename = 'subs_{0}.srt.sjson'.format(html5_id)
content_location = StaticContent.compute_location(
item.location.course_key, filename)
try:
html5_subs.append(contentstore().find(content_location).data)
transcripts_presence['html5_local'].append(html5_id)
except NotFoundError:
log.debug(
"Can't find transcripts in storage for non-youtube video_id: %s",
html5_id)
if len(html5_subs) == 2: # check html5 transcripts for equality
transcripts_presence['html5_equal'] = json.loads(
html5_subs[0]) == json.loads(html5_subs[1])
command, subs_to_use = _transcripts_logic(transcripts_presence, videos)
transcripts_presence.update({
'command': command,
'subs': subs_to_use,
})
return transcripts_presence
def test_youtube_empty_text(self, mock_get):
mock_get.return_value = Mock(status_code=200, text='')
youtube_id = 'bad_youtube_id'
with self.assertRaises(transcripts_utils.GetTranscriptsFromYouTubeException):
transcripts_utils.get_transcripts_from_youtube(youtube_id, settings, translation)
def get_transcripts_presence(videos, item):
""" fills in the transcripts_presence dictionary after for a given component
with its list of videos.
Returns transcripts_presence dict:
html5_local: list of html5 ids, if subtitles exist locally for them;
is_youtube_mode: bool, if we have youtube_id, and as youtube mode is of higher priority, reflect this with flag;
youtube_local: bool, if youtube transcripts exist locally;
youtube_server: bool, if youtube transcripts exist on server;
youtube_diff: bool, if youtube transcripts exist on youtube server, and are different from local youtube ones;
current_item_subs: string, value of item.sub field;
status: string, 'Error' or 'Success';
subs: string, new value of item.sub field, that should be set in module;
command: string, action to front-end what to do and what to show to user.
"""
transcripts_presence = {
'html5_local': [],
'html5_equal': False,
'is_youtube_mode': False,
'youtube_local': False,
'youtube_server': False,
'youtube_diff': True,
'current_item_subs': None,
'status': 'Success',
}
filename = 'subs_{0}.srt.sjson'.format(item.sub)
content_location = StaticContent.compute_location(item.location.course_key, filename)
try:
local_transcripts = contentstore().find(content_location).data
transcripts_presence['current_item_subs'] = item.sub
except NotFoundError:
pass
# Check for youtube transcripts presence
youtube_id = videos.get('youtube', None)
if youtube_id:
transcripts_presence['is_youtube_mode'] = True
# youtube local
filename = 'subs_{0}.srt.sjson'.format(youtube_id)
content_location = StaticContent.compute_location(item.location.course_key, filename)
try:
local_transcripts = contentstore().find(content_location).data
transcripts_presence['youtube_local'] = True
except NotFoundError:
log.debug("Can't find transcripts in storage for youtube id: %s", youtube_id)
# youtube server
youtube_text_api = copy.deepcopy(settings.YOUTUBE['TEXT_API'])
youtube_text_api['params']['v'] = youtube_id
youtube_response = requests.get('http://' + youtube_text_api['url'], params=youtube_text_api['params'])
if youtube_response.status_code == 200 and youtube_response.text:
transcripts_presence['youtube_server'] = True
#check youtube local and server transcripts for equality
if transcripts_presence['youtube_server'] and transcripts_presence['youtube_local']:
try:
youtube_server_subs = get_transcripts_from_youtube(
youtube_id,
settings,
item.runtime.service(item, "i18n")
)
if json.loads(local_transcripts) == youtube_server_subs: # check transcripts for equality
transcripts_presence['youtube_diff'] = False
except GetTranscriptsFromYouTubeException:
pass
# Check for html5 local transcripts presence
html5_subs = []
for html5_id in videos['html5']:
filename = 'subs_{0}.srt.sjson'.format(html5_id)
content_location = StaticContent.compute_location(item.location.course_key, filename)
try:
html5_subs.append(contentstore().find(content_location).data)
transcripts_presence['html5_local'].append(html5_id)
except NotFoundError:
log.debug("Can't find transcripts in storage for non-youtube video_id: %s", html5_id)
if len(html5_subs) == 2: # check html5 transcripts for equality
transcripts_presence['html5_equal'] = json.loads(html5_subs[0]) == json.loads(html5_subs[1])
command, subs_to_use = _transcripts_logic(transcripts_presence, videos)
transcripts_presence.update({
'command': command,
'subs': subs_to_use,
})
return transcripts_presence
def test_youtube_bad_status_code(self, mock_get):
mock_get.return_value = Mock(status_code=404, text="test")
youtube_id = "bad_youtube_id"
with self.assertRaises(transcripts_utils.GetTranscriptsFromYouTubeException):
transcripts_utils.get_transcripts_from_youtube(youtube_id, settings, translation)