X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/532a08904ffbacc5e5ccf99edb660c5f37ddb213..81657e6063f6dd0b3ee6243d446b8df1b13adfbc:/youtube_dl/extractor/turner.py?ds=inline diff --git a/youtube_dl/extractor/turner.py b/youtube_dl/extractor/turner.py index efeb677..e73b64a 100644 --- a/youtube_dl/extractor/turner.py +++ b/youtube_dl/extractor/turner.py @@ -18,9 +18,32 @@ from ..utils import ( class TurnerBaseIE(AdobePassIE): + _AKAMAI_SPE_TOKEN_CACHE = {} + def _extract_timestamp(self, video_data): return int_or_none(xpath_attr(video_data, 'dateCreated', 'uts')) + def _add_akamai_spe_token(self, tokenizer_src, video_url, content_id, ap_data): + secure_path = self._search_regex(r'https?://[^/]+(.+/)', video_url, 'secure path') + '*' + token = self._AKAMAI_SPE_TOKEN_CACHE.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'], content_id, ap_data['site_name'], ap_data['site_name']) + auth = self._download_xml( + tokenizer_src, content_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: + return video_url + self._AKAMAI_SPE_TOKEN_CACHE[secure_path] = token + return video_url + '?hdnea=' + token + 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'] @@ -33,7 +56,6 @@ class TurnerBaseIE(AdobePassIE): # rtmp_src = splited_rtmp_src[1] # aifp = xpath_text(video_data, 'akamai/aifp', default='') - tokens = {} urls = [] formats = [] rex = re.compile( @@ -67,26 +89,10 @@ class TurnerBaseIE(AdobePassIE): secure_path_data = path_data.get('secure') if not secure_path_data: continue - video_url = secure_path_data['media_src'] + video_url - 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=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 - tokens[secure_path] = token - video_url = video_url + '?hdnea=' + token + video_url = self._add_akamai_spe_token( + secure_path_data['tokenizer_src'], + secure_path_data['media_src'] + video_url, + content_id, ap_data) elif not re.match('https?://', video_url): base_path_data = path_data.get(ext, path_data.get('default', {})) media_src = base_path_data.get('media_src')