X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fturner.py;h=1c0be9fc6aa97260622b1148763a57dbf25ce50f;hb=36fce54816eb1f1d792ac7ed4d07e292d44d62f5;hp=d69977b566345974cd4b696bf7f989632d75e093;hpb=547993dcd09dd46fda2fd429ed0ed72db7263503;p=youtube-dl diff --git a/youtube_dl/extractor/turner.py b/youtube_dl/extractor/turner.py index d69977b56..1c0be9fc6 100644 --- a/youtube_dl/extractor/turner.py +++ b/youtube_dl/extractor/turner.py @@ -3,7 +3,8 @@ from __future__ import unicode_literals import re -from .common import InfoExtractor +from .adobepass import AdobePassIE +from ..compat import compat_str from ..utils import ( xpath_text, int_or_none, @@ -11,15 +12,19 @@ from ..utils import ( parse_duration, xpath_attr, update_url_query, - compat_urlparse, + ExtractorError, ) -class TurnerBaseIE(InfoExtractor): - def _extract_cvp_info(self, data_src, video_id, path_data={}): +class TurnerBaseIE(AdobePassIE): + 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={}, ap_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(',') @@ -30,11 +35,11 @@ class TurnerBaseIE(InfoExtractor): tokens = {} urls = [] formats = [] - rex = re.compile(r'''(?x) - (?P[0-9]+)x(?P[0-9]+) - (?:_(?P[0-9]+))? - ''') - for video_file in video_data.findall('files/file'): + rex = re.compile( + r'(?P[0-9]+)x(?P[0-9]+)(?:_(?P[0-9]+))?') + # Possible formats locations: files/file, files/groupFiles/files + # and maybe others + for video_file in video_data.findall('.//file'): video_url = video_file.text.strip() if not video_url: continue @@ -50,7 +55,7 @@ class TurnerBaseIE(InfoExtractor): # 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') @@ -65,11 +70,17 @@ class TurnerBaseIE(InfoExtractor): secure_path = self._search_regex(r'https?://[^/]+(.+/)', video_url, 'secure path') + '*' token = tokens.get(secure_path) if not token: + query = { + 'path': secure_path, + 'videoId': content_id, + } + if ap_data.get('auth_required'): + query['accessToken'] = self._extract_mvpd_auth(ap_data['url'], video_id, ap_data['site_name'], ap_data['site_name']) auth = self._download_xml( - secure_path_data['tokenizer_src'], video_id, query={ - 'path': secure_path, - 'videoId': video_id, - }) + secure_path_data['tokenizer_src'], video_id, query=query) + 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 @@ -84,26 +95,22 @@ class TurnerBaseIE(InfoExtractor): if video_url in urls: continue urls.append(video_url) - format_id = video_file.attrib['bitrate'] + format_id = video_file.get('bitrate') if ext == 'smil': - formats.extend(self._extract_smil_formats(video_url, video_id, fatal=False)) + 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, 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) + video_url, video_id, 'mp4', + m3u8_id=format_id or 'hls', fatal=False) + if '/secure/' in video_url and '?hdnea=' in video_url: + for f in m3u8_formats: + f['_seekable'] = False + formats.extend(m3u8_formats) elif ext == 'f4m': formats.extend(self._extract_f4m_formats( update_url_query(video_url, {'hdcore': '3.7.0'}), - video_id, f4m_id=format_id, fatal=False)) + video_id, f4m_id=format_id or 'hds', fatal=False)) else: f = { 'format_id': format_id, @@ -117,18 +124,19 @@ class TurnerBaseIE(InfoExtractor): 'height': int(mobj.group('height')), 'tbr': int_or_none(mobj.group('bitrate')), }) - elif format_id.isdigit(): - f['tbr'] = int(format_id) - else: - mobj = re.match(r'ios_(audio|[0-9]+)$', format_id) - if mobj: - if mobj.group(1) == 'audio': - f.update({ - 'vcodec': 'none', - 'ext': 'm4a', - }) - else: - f['tbr'] = int(mobj.group(1)) + elif isinstance(format_id, compat_str): + if format_id.isdigit(): + f['tbr'] = int(format_id) + else: + mobj = re.match(r'ios_(audio|[0-9]+)$', format_id) + if mobj: + if mobj.group(1) == 'audio': + f.update({ + 'vcodec': 'none', + 'ext': 'm4a', + }) + else: + f['tbr'] = int(mobj.group(1)) formats.append(f) self._sort_formats(formats) @@ -136,7 +144,7 @@ class TurnerBaseIE(InfoExtractor): 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({ @@ -155,10 +163,6 @@ class TurnerBaseIE(InfoExtractor): '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, @@ -167,7 +171,7 @@ class TurnerBaseIE(InfoExtractor): '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')),