diff options
Diffstat (limited to 'youtube_dl/extractor/gaskrank.py')
| -rw-r--r-- | youtube_dl/extractor/gaskrank.py | 106 |
1 files changed, 42 insertions, 64 deletions
diff --git a/youtube_dl/extractor/gaskrank.py b/youtube_dl/extractor/gaskrank.py index 36ba7d8cf..1726a6704 100644 --- a/youtube_dl/extractor/gaskrank.py +++ b/youtube_dl/extractor/gaskrank.py @@ -6,62 +6,52 @@ from .common import InfoExtractor from ..utils import ( float_or_none, int_or_none, - js_to_json, unified_strdate, ) class GaskrankIE(InfoExtractor): - """InfoExtractor for gaskrank.tv""" - _VALID_URL = r'https?://(?:www\.)?gaskrank\.tv/tv/(?P<categories>[^/]+)/(?P<id>[^/]+)\.html?' - _TESTS = [ - { - 'url': 'http://www.gaskrank.tv/tv/motorrad-fun/strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden.htm', - 'md5': '1ae88dbac97887d85ebd1157a95fc4f9', - 'info_dict': { - 'id': '201601/26955', - 'ext': 'mp4', - 'title': 'Strike! Einparken können nur Männer - Flurschaden hält sich in Grenzen *lol*', - 'thumbnail': r're:^https?://.*\.jpg$', - 'categories': ['motorrad-fun'], - 'display_id': 'strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden', - 'uploader_id': 'Bikefun', - 'upload_date': '20170110', - 'uploader_url': None, - } - }, - { - 'url': 'http://www.gaskrank.tv/tv/racing/isle-of-man-tt-2011-michael-du-15920.htm', - 'md5': 'c33ee32c711bc6c8224bfcbe62b23095', - 'info_dict': { - 'id': '201106/15920', - 'ext': 'mp4', - 'title': 'Isle of Man - Michael Dunlop vs Guy Martin - schwindelig kucken', - 'thumbnail': r're:^https?://.*\.jpg$', - 'categories': ['racing'], - 'display_id': 'isle-of-man-tt-2011-michael-du-15920', - 'uploader_id': 'IOM', - 'upload_date': '20160506', - 'uploader_url': 'www.iomtt.com', - } + _VALID_URL = r'https?://(?:www\.)?gaskrank\.tv/tv/(?P<categories>[^/]+)/(?P<id>[^/]+)\.htm' + _TESTS = [{ + 'url': 'http://www.gaskrank.tv/tv/motorrad-fun/strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden.htm', + 'md5': '1ae88dbac97887d85ebd1157a95fc4f9', + 'info_dict': { + 'id': '201601/26955', + 'ext': 'mp4', + 'title': 'Strike! Einparken können nur Männer - Flurschaden hält sich in Grenzen *lol*', + 'thumbnail': r're:^https?://.*\.jpg$', + 'categories': ['motorrad-fun'], + 'display_id': 'strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden', + 'uploader_id': 'Bikefun', + 'upload_date': '20170110', + 'uploader_url': None, } - ] + }, { + 'url': 'http://www.gaskrank.tv/tv/racing/isle-of-man-tt-2011-michael-du-15920.htm', + 'md5': 'c33ee32c711bc6c8224bfcbe62b23095', + 'info_dict': { + 'id': '201106/15920', + 'ext': 'mp4', + 'title': 'Isle of Man - Michael Dunlop vs Guy Martin - schwindelig kucken', + 'thumbnail': r're:^https?://.*\.jpg$', + 'categories': ['racing'], + 'display_id': 'isle-of-man-tt-2011-michael-du-15920', + 'uploader_id': 'IOM', + 'upload_date': '20170523', + 'uploader_url': 'www.iomtt.com', + } + }] def _real_extract(self, url): - """extract information from gaskrank.tv""" - def fix_json(code): - """Removes trailing comma in json: {{},} --> {{}}""" - return re.sub(r',\s*}', r'}', js_to_json(code)) - display_id = self._match_id(url) + webpage = self._download_webpage(url, display_id) + + title = self._og_search_title( + webpage, default=None) or self._html_search_meta( + 'title', webpage, fatal=True) + categories = [re.match(self._VALID_URL, url).group('categories')] - title = self._search_regex( - r'movieName\s*:\s*\'([^\']*)\'', - webpage, 'title') - thumbnail = self._search_regex( - r'poster\s*:\s*\'([^\']*)\'', - webpage, 'thumbnail', default=None) mobj = re.search( r'Video von:\s*(?P<uploader_id>[^|]*?)\s*\|\s*vom:\s*(?P<upload_date>[0-9][0-9]\.[0-9][0-9]\.[0-9][0-9][0-9][0-9])', @@ -89,29 +79,14 @@ class GaskrankIE(InfoExtractor): if average_rating: average_rating = float_or_none(average_rating.replace(',', '.')) - playlist = self._parse_json( - self._search_regex( - r'playlist\s*:\s*\[([^\]]*)\]', - webpage, 'playlist', default='{}'), - display_id, transform_source=fix_json, fatal=False) - video_id = self._search_regex( r'https?://movies\.gaskrank\.tv/([^-]*?)(-[^\.]*)?\.mp4', - playlist.get('0').get('src'), 'video id') - - formats = [] - for key in playlist: - formats.append({ - 'url': playlist[key]['src'], - 'format_id': key, - 'quality': playlist[key].get('quality')}) - self._sort_formats(formats, field_preference=['format_id']) + webpage, 'video id', default=display_id) - return { + entry = self._parse_html5_media_entries(url, webpage, video_id)[0] + entry.update({ 'id': video_id, 'title': title, - 'formats': formats, - 'thumbnail': thumbnail, 'categories': categories, 'display_id': display_id, 'uploader_id': uploader_id, @@ -120,4 +95,7 @@ class GaskrankIE(InfoExtractor): 'tags': tags, 'view_count': view_count, 'average_rating': average_rating, - } + }) + self._sort_formats(entry['formats']) + + return entry |