X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Ftvnoe.py;h=6d5c748264197ba99a2be8fefccf375761818610;hb=4a76b73c6c797ec7c63432680b38198ba5ef5b9f;hp=d50261dddb7f83f1e2240c80e07bd19c252fe28f;hpb=78e762d23c48f85c61a8afcae29307912000a7dd;p=youtube-dl diff --git a/youtube_dl/extractor/tvnoe.py b/youtube_dl/extractor/tvnoe.py index d50261ddd..6d5c74826 100644 --- a/youtube_dl/extractor/tvnoe.py +++ b/youtube_dl/extractor/tvnoe.py @@ -2,11 +2,15 @@ from __future__ import unicode_literals from .jwplatform import JWPlatformBaseIE -from ..utils import clean_html, get_element_by_class, js_to_json +from ..utils import ( + clean_html, + get_element_by_class, + js_to_json, +) class TVNoeIE(JWPlatformBaseIE): - _VALID_URL = r'https?://(www\.)?tvnoe\.cz/video/(?P[0-9]+)' + _VALID_URL = r'https?://(?:www\.)?tvnoe\.cz/video/(?P[0-9]+)' _TEST = { 'url': 'http://www.tvnoe.cz/video/10362', 'md5': 'aee983f279aab96ec45ab6e2abb3c2ca', @@ -14,8 +18,7 @@ class TVNoeIE(JWPlatformBaseIE): 'id': '10362', 'ext': 'mp4', 'series': 'Noční univerzita', - 'title': 'prof. Tomáš Halík, Th.D. - ' + - 'Návrat náboženství a střet civilizací', + 'title': 'prof. Tomáš Halík, Th.D. - Návrat náboženství a střet civilizací', 'description': 'md5:f337bae384e1a531a52c55ebc50fff41', } } @@ -24,21 +27,23 @@ class TVNoeIE(JWPlatformBaseIE): video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) - iframe_url = self._search_regex(r']+src="([^"]+)"', - webpage, 'iframe src attribute') + iframe_url = self._search_regex( + r']+src="([^"]+)"', webpage, 'iframe URL') ifs_page = self._download_webpage(iframe_url, video_id) - jwplayer_data = self._parse_json(self._find_jwplayer_data(ifs_page), - video_id, transform_source=js_to_json) + jwplayer_data = self._parse_json( + self._find_jwplayer_data(ifs_page), + video_id, transform_source=js_to_json) info_dict = self._parse_jwplayer_data( jwplayer_data, video_id, require_title=False, base_url=iframe_url) info_dict.update({ 'id': video_id, - 'title': clean_html( - get_element_by_class('field-name-field-podnazev', webpage)), - 'description': clean_html(get_element_by_class('field-name-body', - webpage)), + 'title': clean_html(get_element_by_class( + 'field-name-field-podnazev', webpage)), + 'description': clean_html(get_element_by_class( + 'field-name-body', webpage)), 'series': clean_html(get_element_by_class('title', webpage)) }) + return info_dict