From 28b674ca238fbae29f2ee7bc85202fa2f115a876 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sat, 8 Apr 2017 19:42:09 +0700 Subject: [PATCH] [ceskateleveize:porady] Add extractor (closes #7411, closes #12645) --- youtube_dl/extractor/ceskatelevize.py | 45 +++++++++++++++++++++++++++ youtube_dl/extractor/extractors.py | 5 ++- 2 files changed, 49 insertions(+), 1 deletion(-) diff --git a/youtube_dl/extractor/ceskatelevize.py b/youtube_dl/extractor/ceskatelevize.py index 0daee313f..e250de18c 100644 --- a/youtube_dl/extractor/ceskatelevize.py +++ b/youtube_dl/extractor/ceskatelevize.py @@ -12,6 +12,7 @@ ExtractorError, float_or_none, sanitized_Request, + unescapeHTML, urlencode_postdata, USER_AGENTS, ) @@ -232,3 +233,47 @@ def _fix_subtitle(subtitle): yield line return '\r\n'.join(_fix_subtitle(subtitles)) + + +class CeskaTelevizePoradyIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?ceskatelevize\.cz/porady/(?:[^/?#&]+/)*(?P[^/#?]+)' + _TESTS = [{ + # video with 18+ caution trailer + 'url': 'http://www.ceskatelevize.cz/porady/10520528904-queer/215562210900007-bogotart/', + 'info_dict': { + 'id': '215562210900007-bogotart', + 'title': 'Queer: Bogotart', + 'description': 'Alternativní průvodce současným queer světem', + }, + 'playlist': [{ + 'info_dict': { + 'id': '61924494876844842', + 'ext': 'mp4', + 'title': 'Queer: Bogotart (Varování 18+)', + 'duration': 10.2, + }, + }, { + 'info_dict': { + 'id': '61924494877068022', + 'ext': 'mp4', + 'title': 'Queer: Bogotart (Queer)', + 'thumbnail': r're:^https?://.*\.jpg', + 'duration': 1558.3, + }, + }], + 'params': { + # m3u8 download + 'skip_download': True, + }, + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + + webpage = self._download_webpage(url, video_id) + + data_url = unescapeHTML(self._search_regex( + r']*\bdata-url=(["\'])(?P(?:(?!\1).)+)\1', + webpage, 'iframe player url', group='url')) + + return self.url_result(data_url, ie=CeskaTelevizeIE.ie_key()) diff --git a/youtube_dl/extractor/extractors.py b/youtube_dl/extractor/extractors.py index 2904dd4d1..72728d919 100644 --- a/youtube_dl/extractor/extractors.py +++ b/youtube_dl/extractor/extractors.py @@ -165,7 +165,10 @@ from .ccma import CCMAIE from .cctv import CCTVIE from .cda import CDAIE -from .ceskatelevize import CeskaTelevizeIE +from .ceskatelevize import ( + CeskaTelevizeIE, + CeskaTelevizePoradyIE, +) from .channel9 import Channel9IE from .charlierose import CharlieRoseIE from .chaturbate import ChaturbateIE