parse_duration,
xpath_attr,
update_url_query,
- compat_urlparse,
+ ExtractorError,
)
class TurnerBaseIE(InfoExtractor):
+ def _extract_timestamp(self, video_data):
+ return int_or_none(xpath_attr(video_data, 'dateCreated', 'uts'))
+
def _extract_cvp_info(self, data_src, video_id, path_data={}):
video_data = self._download_xml(data_src, video_id)
- video_id = video_data.attrib['id'].split('/')[-1].split('.')[0]
+ video_id = video_data.attrib['id']
title = xpath_text(video_data, 'headline', fatal=True)
+ content_id = xpath_text(video_data, 'contentId') or video_id
# rtmp_src = xpath_text(video_data, 'akamai/src')
# if rtmp_src:
# splited_rtmp_src = rtmp_src.split(',')
# auth = self._download_webpage(
# protected_path_data['tokenizer_src'], query={
# 'path': protected_path,
- # 'videoId': video_id,
+ # 'videoId': content_id,
# 'aifp': aifp,
# })
# token = xpath_text(auth, 'token')
auth = self._download_xml(
secure_path_data['tokenizer_src'], video_id, query={
'path': secure_path,
- 'videoId': video_id,
+ 'videoId': content_id,
})
+ error_msg = xpath_text(auth, 'error/msg')
+ if error_msg:
+ raise ExtractorError(error_msg, expected=True)
token = xpath_text(auth, 'token')
if not token:
continue
formats.extend(self._extract_smil_formats(
video_url, video_id, fatal=False))
elif ext == 'm3u8':
- m3u8_formats = self._extract_m3u8_formats(
- video_url, video_id, 'mp4', m3u8_id=format_id or 'hls',
- fatal=False)
- if m3u8_formats:
- # Sometimes final URLs inside m3u8 are unsigned, let's fix this
- # ourselves
- qs = compat_urlparse.urlparse(video_url).query
- if qs:
- query = compat_urlparse.parse_qs(qs)
- for m3u8_format in m3u8_formats:
- m3u8_format['url'] = update_url_query(m3u8_format['url'], query)
- m3u8_format['extra_param_to_segment_url'] = qs
- formats.extend(m3u8_formats)
+ formats.extend(self._extract_m3u8_formats(
+ video_url, video_id, 'mp4',
+ m3u8_id=format_id or 'hls', fatal=False))
elif ext == 'f4m':
formats.extend(self._extract_f4m_formats(
update_url_query(video_url, {'hdcore': '3.7.0'}),
for source in video_data.findall('closedCaptions/source'):
for track in source.findall('track'):
track_url = track.get('url')
- if not track_url:
+ if not isinstance(track_url, compat_str) or track_url.endswith('/big'):
continue
lang = track.get('lang') or track.get('label') or 'en'
subtitles.setdefault(lang, []).append({
'height': int_or_none(image.get('height')),
} for image in video_data.findall('images/image')]
- timestamp = None
- if 'cnn.com' not in data_src:
- timestamp = int_or_none(xpath_attr(video_data, 'dateCreated', 'uts'))
-
return {
'id': video_id,
'title': title,
'thumbnails': thumbnails,
'description': xpath_text(video_data, 'description'),
'duration': parse_duration(xpath_text(video_data, 'length') or xpath_text(video_data, 'trt')),
- 'timestamp': timestamp,
+ 'timestamp': self._extract_timestamp(video_data),
'upload_date': xpath_attr(video_data, 'metas', 'version'),
'series': xpath_text(video_data, 'showTitle'),
'season_number': int_or_none(xpath_text(video_data, 'seasonNumber')),