From a7b8467ac0baecd02a815b1f57731ae9bb10ab87 Mon Sep 17 00:00:00 2001 From: Vitaliy Syrchikov Date: Fri, 15 May 2015 16:52:11 +0400 Subject: [PATCH 1/8] Sportbox extractor fix. --- youtube_dl/extractor/sportbox.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py index becdf658f..830220543 100644 --- a/youtube_dl/extractor/sportbox.py +++ b/youtube_dl/extractor/sportbox.py @@ -11,7 +11,7 @@ from ..utils import ( class SportBoxIE(InfoExtractor): - _VALID_URL = r'https?://news\.sportbox\.ru/Vidy_sporta/(?:[^/]+/)+spbvideo_NI\d+_(?P.+)' + _VALID_URL = r'https?://news\.sportbox\.ru/(?:[^/]+/)+spbvideo_NI\d+_(?P.+)' _TESTS = [ { 'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S', @@ -50,7 +50,7 @@ class SportBoxIE(InfoExtractor): display_id, 'Downloading player webpage') hls = self._search_regex( - r"var\s+original_hls_file\s*=\s*'([^']+)'", player, 'hls file') + r"sportboxPlayer\.jwplayer_common_params\.file\s*=\s*['\"]+([^\"]+)['\"]+", player, 'hls file') formats = self._extract_m3u8_formats(hls, display_id, 'mp4') From ae670a6ed8019f1b69bbe345621f51c8b32789ec Mon Sep 17 00:00:00 2001 From: Vitaliy Syrchikov Date: Fri, 15 May 2015 17:53:05 +0400 Subject: [PATCH 2/8] Sportbox source fix. HD videos support. --- youtube_dl/extractor/sportbox.py | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py index 830220543..695b3ff82 100644 --- a/youtube_dl/extractor/sportbox.py +++ b/youtube_dl/extractor/sportbox.py @@ -14,7 +14,7 @@ class SportBoxIE(InfoExtractor): _VALID_URL = r'https?://news\.sportbox\.ru/(?:[^/]+/)+spbvideo_NI\d+_(?P.+)' _TESTS = [ { - 'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S', + 'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S', 'md5': 'ff56a598c2cf411a9a38a69709e97079', 'info_dict': { 'id': '80822', @@ -42,11 +42,15 @@ class SportBoxIE(InfoExtractor): webpage = self._download_webpage(url, display_id) - video_id = self._search_regex( - r'src="/vdl/player/media/(\d+)"', webpage, 'video id') + sobj = re.search(r'src="/vdl/player/(?P\w+)/(?P\d+)"', webpage) + if (sobj): + video_id = sobj.group('video_id') + media_type = sobj.group('media_type') + else: + raise RegexNotFoundError('Unable to extract video_id') player = self._download_webpage( - 'http://news.sportbox.ru/vdl/player/media/%s' % video_id, + 'http://news.sportbox.ru/vdl/player/%s/%s' % (media_type, video_id), display_id, 'Downloading player webpage') hls = self._search_regex( From e9ca615a9872e85a6986061fdf54257244ce1f77 Mon Sep 17 00:00:00 2001 From: Vitaliy Syrchikov Date: Fri, 15 May 2015 19:57:54 +0400 Subject: [PATCH 3/8] New test --- youtube_dl/extractor/sportbox.py | 23 +++++++++++++++++++++-- 1 file changed, 21 insertions(+), 2 deletions(-) diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py index 695b3ff82..cb1515eff 100644 --- a/youtube_dl/extractor/sportbox.py +++ b/youtube_dl/extractor/sportbox.py @@ -30,10 +30,29 @@ class SportBoxIE(InfoExtractor): # m3u8 download 'skip_download': True, }, - }, { + }, + { + 'url': 'http://news.sportbox.ru/video/no_ads/spbvideo_NI536574_V_Novorossijske_proshel_detskij_turnir_Pole_slavy_bojevoj?ci=211355', + 'md5': 'ff56a598c2cf411a9a38a69709e97079', + 'info_dict': { + 'id': '211355', + 'ext': 'mp4', + 'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»', + 'description': '16 детских коллективов приняли участие в суперфинале турнира «Поле славы боевой».', + 'thumbnail': 're:^https?://.*\.jpg$', + 'timestamp': 1426237001, + 'upload_date': '20150313', + 'duration': 292, + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + }, + { 'url': 'http://news.sportbox.ru/Vidy_sporta/billiard/spbvideo_NI486287_CHempionat-mira-po-dinamichnoy-piramide-4', 'only_matching': True, - } + }, ] def _real_extract(self, url): From 3a7382950b6f498f50173c8813f6cb1db3739277 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Fri, 15 May 2015 22:50:44 +0600 Subject: [PATCH 4/8] [sportbox:embed] Add extractor --- youtube_dl/extractor/__init__.py | 5 +- youtube_dl/extractor/sportbox.py | 138 +++++++++++++++++-------------- 2 files changed, 82 insertions(+), 61 deletions(-) diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index 8ec0c1032..f293bc2a4 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -502,7 +502,10 @@ from .spiegel import SpiegelIE, SpiegelArticleIE from .spiegeltv import SpiegeltvIE from .spike import SpikeIE from .sport5 import Sport5IE -from .sportbox import SportBoxIE +from .sportbox import ( + SportBoxIE, + SportBoxEmbedIE, +) from .sportdeutschland import SportDeutschlandIE from .srf import SrfIE from .srmediathek import SRMediathekIE diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py index cb1515eff..10c45eb74 100644 --- a/youtube_dl/extractor/sportbox.py +++ b/youtube_dl/extractor/sportbox.py @@ -4,6 +4,7 @@ from __future__ import unicode_literals import re from .common import InfoExtractor +from ..compat import compat_urlparse from ..utils import ( parse_duration, parse_iso8601, @@ -12,48 +13,30 @@ from ..utils import ( class SportBoxIE(InfoExtractor): _VALID_URL = r'https?://news\.sportbox\.ru/(?:[^/]+/)+spbvideo_NI\d+_(?P.+)' - _TESTS = [ - { - 'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S', - 'md5': 'ff56a598c2cf411a9a38a69709e97079', - 'info_dict': { - 'id': '80822', - 'ext': 'mp4', - 'title': 'Гонка 2 заезд ««Объединенный 2000»: классы Туринг и Супер-продакшн', - 'description': 'md5:81715fa9c4ea3d9e7915dc8180c778ed', - 'thumbnail': 're:^https?://.*\.jpg$', - 'timestamp': 1411896237, - 'upload_date': '20140928', - 'duration': 4846, - }, - 'params': { - # m3u8 download - 'skip_download': True, - }, - }, - { - 'url': 'http://news.sportbox.ru/video/no_ads/spbvideo_NI536574_V_Novorossijske_proshel_detskij_turnir_Pole_slavy_bojevoj?ci=211355', - 'md5': 'ff56a598c2cf411a9a38a69709e97079', - 'info_dict': { - 'id': '211355', - 'ext': 'mp4', - 'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»', - 'description': '16 детских коллективов приняли участие в суперфинале турнира «Поле славы боевой».', - 'thumbnail': 're:^https?://.*\.jpg$', - 'timestamp': 1426237001, - 'upload_date': '20150313', - 'duration': 292, - }, - 'params': { - # m3u8 download - 'skip_download': True, - }, - }, - { - 'url': 'http://news.sportbox.ru/Vidy_sporta/billiard/spbvideo_NI486287_CHempionat-mira-po-dinamichnoy-piramide-4', - 'only_matching': True, + _TESTS = [{ + 'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S', + 'md5': 'ff56a598c2cf411a9a38a69709e97079', + 'info_dict': { + 'id': '80822', + 'ext': 'mp4', + 'title': 'Гонка 2 заезд ««Объединенный 2000»: классы Туринг и Супер-продакшн', + 'description': 'md5:81715fa9c4ea3d9e7915dc8180c778ed', + 'thumbnail': 're:^https?://.*\.jpg$', + 'timestamp': 1411896237, + 'upload_date': '20140928', + 'duration': 4846, }, - ] + 'params': { + # m3u8 download + 'skip_download': True, + }, + }, { + 'url': 'http://news.sportbox.ru/Vidy_sporta/billiard/spbvideo_NI486287_CHempionat-mira-po-dinamichnoy-piramide-4', + 'only_matching': True, + }, { + 'url': 'http://news.sportbox.ru/video/no_ads/spbvideo_NI536574_V_Novorossijske_proshel_detskij_turnir_Pole_slavy_bojevoj?ci=211355', + 'only_matching': True, + }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) @@ -61,39 +44,74 @@ class SportBoxIE(InfoExtractor): webpage = self._download_webpage(url, display_id) - sobj = re.search(r'src="/vdl/player/(?P\w+)/(?P\d+)"', webpage) - if (sobj): - video_id = sobj.group('video_id') - media_type = sobj.group('media_type') - else: - raise RegexNotFoundError('Unable to extract video_id') - - player = self._download_webpage( - 'http://news.sportbox.ru/vdl/player/%s/%s' % (media_type, video_id), - display_id, 'Downloading player webpage') - - hls = self._search_regex( - r"sportboxPlayer\.jwplayer_common_params\.file\s*=\s*['\"]+([^\"]+)['\"]+", player, 'hls file') - - formats = self._extract_m3u8_formats(hls, display_id, 'mp4') + player = self._search_regex( + r'src="/?(vdl/player/[^"]+)"', webpage, 'player') title = self._html_search_regex( r'

([^<]+)

', webpage, 'title') description = self._html_search_regex( - r'(?s)
(.+?)
', webpage, 'description', fatal=False) + r'(?s)
(.+?)
', + webpage, 'description', fatal=False) thumbnail = self._og_search_thumbnail(webpage) timestamp = parse_iso8601(self._search_regex( - r'([^<]+)', webpage, 'timestamp', fatal=False)) + r'([^<]+)', + webpage, 'timestamp', fatal=False)) duration = parse_duration(self._html_search_regex( - r'', webpage, 'duration', fatal=False)) + r'', + webpage, 'duration', fatal=False)) return { - 'id': video_id, + '_type': 'url_transparent', + 'url': compat_urlparse.urljoin(url, '/%s' % player), 'display_id': display_id, 'title': title, 'description': description, 'thumbnail': thumbnail, 'timestamp': timestamp, 'duration': duration, + } + + +class SportBoxEmbedIE(InfoExtractor): + _VALID_URL = r'https?://news\.sportbox\.ru/vdl/player(?:/[^/]+/|\?.*?\bn?id=)(?P\d+)' + _TESTS = [{ + 'url': 'http://news.sportbox.ru/vdl/player/ci/211355', + 'info_dict': { + 'id': '211355', + 'ext': 'mp4', + 'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»', + 'thumbnail': 're:^https?://.*\.jpg$', + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + }, { + 'url': 'http://news.sportbox.ru/vdl/player?nid=370908&only_player=1&autostart=false&playeri=2&height=340&width=580', + 'only_matching': True, + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + + webpage = self._download_webpage(url, video_id) + + hls = self._search_regex( + r"sportboxPlayer\.jwplayer_common_params\.file\s*=\s*['\"]([^'\"]+)['\"]", + webpage, 'hls file') + + formats = self._extract_m3u8_formats(hls, video_id, 'mp4') + + title = self._search_regex( + r'sportboxPlayer\.node_title\s*=\s*"([^"]+)"', webpage, 'title') + + thumbnail = self._search_regex( + r'sportboxPlayer\.jwplayer_common_params\.image\s*=\s*"([^"]+)"', + webpage, 'thumbnail', fatal=False) + + return { + 'id': video_id, + 'title': title, + 'thumbnail': thumbnail, 'formats': formats, } From 1436a6835e0f3489a4c37cca3da5087567b68158 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Fri, 15 May 2015 23:08:44 +0600 Subject: [PATCH 5/8] [sportbox:embed] Add `_extract_urls` --- youtube_dl/extractor/sportbox.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py index 10c45eb74..a869a1b25 100644 --- a/youtube_dl/extractor/sportbox.py +++ b/youtube_dl/extractor/sportbox.py @@ -91,6 +91,12 @@ class SportBoxEmbedIE(InfoExtractor): 'only_matching': True, }] + @staticmethod + def _extract_urls(webpage): + return re.findall( + r']+src="(https?://news\.sportbox\.ru/vdl/player[^"]+)"', + webpage) + def _real_extract(self, url): video_id = self._match_id(url) From ef28a6cb26630f8f198a72eee34a2b5c8bd2f802 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Fri, 15 May 2015 23:09:10 +0600 Subject: [PATCH 6/8] [sportbox:embed] Relax thumbnail --- youtube_dl/extractor/sportbox.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py index a869a1b25..8686f9d11 100644 --- a/youtube_dl/extractor/sportbox.py +++ b/youtube_dl/extractor/sportbox.py @@ -113,7 +113,7 @@ class SportBoxEmbedIE(InfoExtractor): thumbnail = self._search_regex( r'sportboxPlayer\.jwplayer_common_params\.image\s*=\s*"([^"]+)"', - webpage, 'thumbnail', fatal=False) + webpage, 'thumbnail', default=None) return { 'id': video_id, From d40a3b5b55973d7ed65538179b71990c1828845a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Fri, 15 May 2015 23:09:34 +0600 Subject: [PATCH 7/8] [generic] Add support for sportbox embeds --- youtube_dl/extractor/generic.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index 3d756e848..9230c3bb0 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -32,6 +32,7 @@ from .brightcove import BrightcoveIE from .nbc import NBCSportsVPlayerIE from .ooyala import OoyalaIE from .rutv import RUTVIE +from .sportbox import SportBoxEmbedIE from .smotri import SmotriIE from .condenast import CondeNastIE from .udn import UDNEmbedIE @@ -1229,6 +1230,11 @@ class GenericIE(InfoExtractor): if rutv_url: return self.url_result(rutv_url, 'RUTV') + # Look for embedded SportBox player + sportbox_urls = SportBoxEmbedIE._extract_urls(webpage) + if sportbox_urls: + return _playlist_from_matches(sportbox_urls, ie='SportBoxEmbed') + # Look for embedded TED player mobj = re.search( r']+?src=(["\'])(?Phttps?://embed(?:-ssl)?\.ted\.com/.+?)\1', webpage) From b827a6015c145d67a4d4e9ea38aa54ebe347d3fe Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Fri, 15 May 2015 23:18:21 +0600 Subject: [PATCH 8/8] [generic] Add test for sportbox embeds --- youtube_dl/extractor/generic.py | 31 +++++++++++++++++++++++++++++++ 1 file changed, 31 insertions(+) diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index 9230c3bb0..610e33091 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -225,6 +225,37 @@ class GenericIE(InfoExtractor): 'skip_download': True, }, }, + # SportBox embed + { + 'url': 'http://www.vestifinance.ru/articles/25753', + 'info_dict': { + 'id': '25753', + 'title': 'Вести Экономика ― Прямые трансляции с Форума-выставки "Госзаказ-2013"', + }, + 'playlist': [{ + 'info_dict': { + 'id': '370908', + 'title': 'Госзаказ. День 3', + 'ext': 'mp4', + } + }, { + 'info_dict': { + 'id': '370905', + 'title': 'Госзаказ. День 2', + 'ext': 'mp4', + } + }, { + 'info_dict': { + 'id': '370902', + 'title': 'Госзаказ. День 1', + 'ext': 'mp4', + } + }], + 'params': { + # m3u8 download + 'skip_download': True, + }, + }, # Embedded TED video { 'url': 'http://en.support.wordpress.com/videos/ted-talks/',