X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/b238854ce845f3796daac74edab2e8a373e8ba1a..46113edab215c2211a604c06245c16d5d4e57dcf:/youtube_dl/extractor/gamespot.py?ds=sidebyside diff --git a/youtube_dl/extractor/gamespot.py b/youtube_dl/extractor/gamespot.py index 26b7d2a..3d67b9d 100644 --- a/youtube_dl/extractor/gamespot.py +++ b/youtube_dl/extractor/gamespot.py @@ -1,3 +1,5 @@ +from __future__ import unicode_literals + import re import json @@ -13,12 +15,13 @@ from ..utils import ( class GameSpotIE(InfoExtractor): _VALID_URL = r'(?:http://)?(?:www\.)?gamespot\.com/.*-(?P\d+)/?' _TEST = { - u"url": u"http://www.gamespot.com/arma-iii/videos/arma-iii-community-guide-sitrep-i-6410818/", - u"file": u"gs-2300-6410818.mp4", - u"md5": u"b2a30deaa8654fcccd43713a6b6a4825", - u"info_dict": { - u"title": u"Arma 3 - Community Guide: SITREP I", - u'description': u'Check out this video where some of the basics of Arma 3 is explained.', + 'url': 'http://www.gamespot.com/videos/arma-3-community-guide-sitrep-i/2300-6410818/', + 'md5': 'b2a30deaa8654fcccd43713a6b6a4825', + 'info_dict': { + 'id': 'gs-2300-6410818', + 'ext': 'mp4', + 'title': 'Arma 3 - Community Guide: SITREP I', + 'description': 'Check out this video where some of the basics of Arma 3 is explained.', } } @@ -26,7 +29,7 @@ class GameSpotIE(InfoExtractor): mobj = re.match(self._VALID_URL, url) page_id = mobj.group('page_id') webpage = self._download_webpage(url, page_id) - data_video_json = self._search_regex(r'data-video=\'(.*?)\'', webpage, u'data video') + data_video_json = self._search_regex(r'data-video=["\'](.*?)["\']', webpage, 'data video') data_video = json.loads(unescapeHTML(data_video_json)) # Transform the manifest url to a link to the mp4 files @@ -34,7 +37,7 @@ class GameSpotIE(InfoExtractor): f4m_url = data_video['videoStreams']['f4m_stream'] f4m_path = compat_urlparse.urlparse(f4m_url).path QUALITIES_RE = r'((,\d+)+,?)' - qualities = self._search_regex(QUALITIES_RE, f4m_path, u'qualities').strip(',').split(',') + qualities = self._search_regex(QUALITIES_RE, f4m_path, 'qualities').strip(',').split(',') http_path = f4m_path[1:].split('/', 1)[1] http_template = re.sub(QUALITIES_RE, r'%s', http_path) http_template = http_template.replace('.csmil/manifest.f4m', '')