from __future__ import unicode_literals import re import json import os from .common import InfoExtractor from ..compat import ( compat_urlparse, compat_urllib_parse_urlencode, compat_urllib_parse_urlparse, compat_str, ) from ..utils import ( unified_strdate, determine_ext, int_or_none, parse_iso8601, parse_duration, ) class NHLBaseInfoExtractor(InfoExtractor): @staticmethod def _fix_json(json_string): return json_string.replace('\\\'', '\'') def _real_extract_video(self, video_id): vid_parts = video_id.split(',') if len(vid_parts) == 3: video_id = '%s0%s%s-X-h' % (vid_parts[0][:4], vid_parts[1], vid_parts[2].rjust(4, '0')) json_url = 'http://video.nhl.com/videocenter/servlets/playlist?ids=%s&format=json' % video_id data = self._download_json( json_url, video_id, transform_source=self._fix_json) return self._extract_video(data[0]) def _extract_video(self, info): video_id = info['id'] self.report_extraction(video_id) initial_video_url = info['publishPoint'] if info['formats'] == '1': parsed_url = compat_urllib_parse_urlparse(initial_video_url) filename, ext = os.path.splitext(parsed_url.path) path = '%s_sd%s' % (filename, ext) data = compat_urllib_parse_urlencode({ 'type': 'fvod', 'path': compat_urlparse.urlunparse(parsed_url[:2] + (path,) + parsed_url[3:]) }) path_url = 'http://video.nhl.com/videocenter/servlets/encryptvideopath?' + data path_doc = self._download_xml( path_url, video_id, 'Downloading final video url') video_url = path_doc.find('path').text else: video_url = initial_video_url join = compat_urlparse.urljoin ret = { 'id': video_id, 'title': info['name'], 'url': video_url, 'description': info['description'], 'duration': int(info['duration']), 'thumbnail': join(join(video_url, '/u/'), info['bigImage']), 'upload_date': unified_strdate(info['releaseDate'].split('.')[0]), } if video_url.startswith('rtmp:'): mobj = re.match(r'(?P<tc_url>rtmp://[^/]+/(?P<app>[a-z0-9/]+))/(?P<play_path>mp4:.*)', video_url) ret.update({ 'tc_url': mobj.group('tc_url'), 'play_path': mobj.group('play_path'), 'app': mobj.group('app'), 'no_resume': True, }) return ret class NHLVideocenterIE(NHLBaseInfoExtractor): IE_NAME = 'nhl.com:videocenter' _VALID_URL = r'https?://video(?P<team>\.[^.]*)?\.nhl\.com/videocenter/(?:console|embed)?(?:\?(?:.*?[?&])?)(?:id|hlg|playlist)=(?P<id>[-0-9a-zA-Z,]+)' _TESTS = [{ 'url': 'http://video.canucks.nhl.com/videocenter/console?catid=6?id=453614', 'md5': 'db704a4ea09e8d3988c85e36cc892d09', 'info_dict': { 'id': '453614', 'ext': 'mp4', 'title': 'Quick clip: Weise 4-3 goal vs Flames', 'description': 'Dale Weise scores his first of the season to put the Canucks up 4-3.', 'duration': 18, 'upload_date': '20131006', }, }, { 'url': 'http://video.nhl.com/videocenter/console?id=2014020024-628-h', 'md5': 'd22e82bc592f52d37d24b03531ee9696', 'info_dict': { 'id': '2014020024-628-h', 'ext': 'mp4', 'title': 'Alex Galchenyuk Goal on Ray Emery (14:40/3rd)', 'description': 'Home broadcast - Montreal Canadiens at Philadelphia Flyers - October 11, 2014', 'duration': 0, 'upload_date': '20141011', }, }, { 'url': 'http://video.mapleleafs.nhl.com/videocenter/console?id=58665&catid=802', 'md5': 'c78fc64ea01777e426cfc202b746c825', 'info_dict': { 'id': '58665', 'ext': 'flv', 'title': 'Classic Game In Six - April 22, 1979', 'description': 'It was the last playoff game for the Leafs in the decade, and the last time the Leafs and Habs played in the playoffs. Great game, not a great ending.', 'duration': 400, 'upload_date': '20100129' }, }, { 'url': 'http://video.flames.nhl.com/videocenter/console?id=630616', 'only_matching': True, }, { 'url': 'http://video.nhl.com/videocenter/?id=736722', 'only_matching': True, }, { 'url': 'http://video.nhl.com/videocenter/console?hlg=20142015,2,299&lang=en', 'md5': '076fcb88c255154aacbf0a7accc3f340', 'info_dict': { 'id': '2014020299-X-h', 'ext': 'mp4', 'title': 'Penguins at Islanders / Game Highlights', 'description': 'Home broadcast - Pittsburgh Penguins at New York Islanders - November 22, 2014', 'duration': 268, 'upload_date': '20141122', } }, { 'url': 'http://video.oilers.nhl.com/videocenter/console?id=691469&catid=4', 'info_dict': { 'id': '691469', 'ext': 'mp4', 'title': 'RAW | Craig MacTavish Full Press Conference', 'description': 'Oilers GM Craig MacTavish addresses the media at Rexall Place on Friday.', 'upload_date': '20141205', }, 'params': { 'skip_download': True, # Requires rtmpdump } }, { 'url': 'http://video.nhl.com/videocenter/embed?playlist=836127', 'only_matching': True, }] def _real_extract(self, url): video_id = self._match_id(url) return self._real_extract_video(video_id) class NHLNewsIE(NHLBaseInfoExtractor): IE_NAME = 'nhl.com:news' IE_DESC = 'NHL news' _VALID_URL = r'https?://(?:.+?\.)?nhl\.com/(?:ice|club)/news\.html?(?:\?(?:.*?[?&])?)id=(?P<id>[-0-9a-zA-Z]+)' _TESTS = [{ 'url': 'http://www.nhl.com/ice/news.htm?id=750727', 'md5': '4b3d1262e177687a3009937bd9ec0be8', 'info_dict': { 'id': '736722', 'ext': 'mp4', 'title': 'Cal Clutterbuck has been fined $2,000', 'description': 'md5:45fe547d30edab88b23e0dd0ab1ed9e6', 'duration': 37, 'upload_date': '20150128', }, }, { # iframe embed 'url': 'http://sabres.nhl.com/club/news.htm?id=780189', 'md5': '9f663d1c006c90ac9fb82777d4294e12', 'info_dict': { 'id': '836127', 'ext': 'mp4', 'title': 'Morning Skate: OTT vs. BUF (9/23/15)', 'description': "Brian Duff chats with Tyler Ennis prior to Buffalo's first preseason home game.", 'duration': 93, 'upload_date': '20150923', }, }] def _real_extract(self, url): news_id = self._match_id(url) webpage = self._download_webpage(url, news_id) video_id = self._search_regex( [r'pVid(\d+)', r"nlid\s*:\s*'(\d+)'", r'<iframe[^>]+src=["\']https?://video.*?\.nhl\.com/videocenter/embed\?.*\bplaylist=(\d+)'], webpage, 'video id') return self._real_extract_video(video_id) class NHLVideocenterCategoryIE(NHLBaseInfoExtractor): IE_NAME = 'nhl.com:videocenter:category' IE_DESC = 'NHL videocenter category' _VALID_URL = r'https?://video\.(?P<team>[^.]*)\.nhl\.com/videocenter/(console\?[^(id=)]*catid=(?P<catid>[0-9]+)(?![&?]id=).*?)?$' _TEST = { 'url': 'http://video.canucks.nhl.com/videocenter/console?catid=999', 'info_dict': { 'id': '999', 'title': 'Highlights', }, 'playlist_count': 12, } def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) team = mobj.group('team') webpage = self._download_webpage(url, team) cat_id = self._search_regex( [r'var defaultCatId = "(.+?)";', r'{statusIndex:0,index:0,.*?id:(.*?),'], webpage, 'category id') playlist_title = self._html_search_regex( r'tab0"[^>]*?>(.*?)</td>', webpage, 'playlist title', flags=re.DOTALL).lower().capitalize() data = compat_urllib_parse_urlencode({ 'cid': cat_id, # This is the default value 'count': 12, 'ptrs': 3, 'format': 'json', }) path = '/videocenter/servlets/browse?' + data request_url = compat_urlparse.urljoin(url, path) response = self._download_webpage(request_url, playlist_title) response = self._fix_json(response) if not response.strip(): self._downloader.report_warning('Got an empty response, trying ' 'adding the "newvideos" parameter') response = self._download_webpage(request_url + '&newvideos=true', playlist_title) response = self._fix_json(response) videos = json.loads(response) return { '_type': 'playlist', 'title': playlist_title, 'id': cat_id, 'entries': [self._extract_video(v) for v in videos], } class NHLIE(InfoExtractor): IE_NAME = 'nhl.com' _VALID_URL = r'https?://(?:www\.)?(?P<site>nhl|wch2016)\.com/(?:[^/]+/)*c-(?P<id>\d+)' _SITES_MAP = { 'nhl': 'nhl', 'wch2016': 'wch', } _TESTS = [{ # type=video 'url': 'https://www.nhl.com/video/anisimov-cleans-up-mess/t-277752844/c-43663503', 'md5': '0f7b9a8f986fb4b4eeeece9a56416eaf', 'info_dict': { 'id': '43663503', 'ext': 'mp4', 'title': 'Anisimov cleans up mess', 'description': 'md5:a02354acdfe900e940ce40706939ca63', 'timestamp': 1461288600, 'upload_date': '20160422', }, }, { # type=article 'url': 'https://www.nhl.com/news/dennis-wideman-suspended/c-278258934', 'md5': '1f39f4ea74c1394dea110699a25b366c', 'info_dict': { 'id': '40784403', 'ext': 'mp4', 'title': 'Wideman suspended by NHL', 'description': 'Flames defenseman Dennis Wideman was banned 20 games for violation of Rule 40 (Physical Abuse of Officials)', 'upload_date': '20160204', 'timestamp': 1454544904, }, }, { # Some m3u8 URLs are invalid (https://github.com/rg3/youtube-dl/issues/10713) 'url': 'https://www.nhl.com/predators/video/poile-laviolette-on-subban-trade/t-277437416/c-44315003', 'md5': '50b2bb47f405121484dda3ccbea25459', 'info_dict': { 'id': '44315003', 'ext': 'mp4', 'title': 'Poile, Laviolette on Subban trade', 'description': 'General manager David Poile and head coach Peter Laviolette share their thoughts on acquiring P.K. Subban from Montreal (06/29/16)', 'timestamp': 1467242866, 'upload_date': '20160629', }, }, { 'url': 'https://www.wch2016.com/video/caneur-best-of-game-2-micd-up/t-281230378/c-44983703', 'only_matching': True, }, { 'url': 'https://www.wch2016.com/news/3-stars-team-europe-vs-team-canada/c-282195068', 'only_matching': True, }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) tmp_id, site = mobj.group('id'), mobj.group('site') video_data = self._download_json( 'https://nhl.bamcontent.com/%s/id/v1/%s/details/web-v1.json' % (self._SITES_MAP[site], tmp_id), tmp_id) if video_data.get('type') == 'article': video_data = video_data['media'] video_id = compat_str(video_data['id']) title = video_data['title'] formats = [] for playback in video_data.get('playbacks', []): playback_url = playback.get('url') if not playback_url: continue ext = determine_ext(playback_url) if ext == 'm3u8': m3u8_formats = self._extract_m3u8_formats( playback_url, video_id, 'mp4', 'm3u8_native', m3u8_id=playback.get('name', 'hls'), fatal=False) self._check_formats(m3u8_formats, video_id) formats.extend(m3u8_formats) else: height = int_or_none(playback.get('height')) formats.append({ 'format_id': playback.get('name', 'http' + ('-%dp' % height if height else '')), 'url': playback_url, 'width': int_or_none(playback.get('width')), 'height': height, }) self._sort_formats(formats, ('preference', 'width', 'height', 'tbr', 'format_id')) thumbnails = [] for thumbnail_id, thumbnail_data in video_data.get('image', {}).get('cuts', {}).items(): thumbnail_url = thumbnail_data.get('src') if not thumbnail_url: continue thumbnails.append({ 'id': thumbnail_id, 'url': thumbnail_url, 'width': int_or_none(thumbnail_data.get('width')), 'height': int_or_none(thumbnail_data.get('height')), }) return { 'id': video_id, 'title': title, 'description': video_data.get('description'), 'timestamp': parse_iso8601(video_data.get('date')), 'duration': parse_duration(video_data.get('duration')), 'thumbnails': thumbnails, 'formats': formats, }