From 1ef1563649374568870e9334cce7055f7c83a817 Mon Sep 17 00:00:00 2001 From: remitamine Date: Fri, 9 Oct 2015 20:08:37 +0100 Subject: [srgssr] Add generic extractor for SRGSSR Group sites --- youtube_dl/extractor/srgssr.py | 130 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 130 insertions(+) create mode 100644 youtube_dl/extractor/srgssr.py (limited to 'youtube_dl/extractor/srgssr.py') diff --git a/youtube_dl/extractor/srgssr.py b/youtube_dl/extractor/srgssr.py new file mode 100644 index 000000000..addf4d26e --- /dev/null +++ b/youtube_dl/extractor/srgssr.py @@ -0,0 +1,130 @@ +# coding: utf-8 +from __future__ import unicode_literals + +import re + +from .common import InfoExtractor +from ..utils import ( + ExtractorError, + parse_iso8601, + qualities, +) + + +class SRGSSRIE(InfoExtractor): + _VALID_URL = r'(?:https?://tp\.srgssr\.ch/p(?:/[^/]+)+\?urn=)?urn:(?Psrf|rts|rsi|rtr|swi):(?:[^:]+:)?(?Pvideo|audio):(?P[0-9a-f\-]{36}|\d+)' + + _ERRORS = { + 'AGERATING12': 'To protect children under the age of 12, this video is only available between 8 p.m. and 6 a.m.', + 'AGERATING18': 'To protect children under the age of 18, this video is only available between 11 p.m. and 5 a.m.', +# 'ENDDATE': 'For legal reasons, this video was only available for a specified period of time.', + 'GEOBLOCK': 'For legal reasons, this video is only available in Switzerland.', + 'LEGAL': 'The video cannot be transmitted for legal reasons.', + 'STARTDATE': 'This video is not yet available. Please try again later.', + } + + def _real_extract(self, url): + bu, media_type, media_id = re.match(self._VALID_URL, url).groups() + + media_data = self._download_json( + 'http://il.srgssr.ch/integrationlayer/1.0/ue/%s/%s/play/%s.json' % (bu, media_type, media_id), + media_id)[media_type.capitalize()] + + if media_data.get('block') and media_data['block'] in self._ERRORS: + raise ExtractorError( + '%s said: %s' % ( + self.IE_NAME, + self._ERRORS[media_data['block']]), + expected=True) + + metadata = media_data['AssetMetadatas']['AssetMetadata'][0] + title = metadata['title'] + description = metadata.get('description') + created_date = media_data.get('createdDate') or metadata.get('createdDate') + timestamp = parse_iso8601(created_date) + + thumbnails = [] + for image in media_data['Image']['ImageRepresentations']['ImageRepresentation']: + thumbnails.append({ + 'id': image.get('id'), + 'url': image['url'], + }) + + preference = qualities(['LQ', 'MQ', 'SD', 'HQ', 'HD']) + formats = [] + for source in media_data['Playlists']['Playlist']: + assets = {} + protocol = source.get('@protocol') + if protocol in ('HTTP-HDS', 'HTTP-HLS'): + for quality in source['url']: + assets[quality['@quality']] = quality['text'] + asset_url = assets.get('HD') or assets.get('HQ') or assets.get('SD') or assets.get('MQ') or assets.get('LQ') + if '.f4m' in asset_url: + formats.extend(self._extract_f4m_formats(asset_url + '?hdcore=3.4.0', media_id, f4m_id='hds')) + elif '.m3u8' in asset_url: + formats.extend(self._extract_m3u8_formats(asset_url, media_id, m3u8_id='hls')) + else: + for asset in source['url']: + asset_url = asset['text'] + ext = None + if asset_url.startswith('rtmp'): + ext = self._search_regex(r'([a-z0-9]+):[^/]+', asset_url, 'ext') + formats.append({ + 'url': asset_url, + 'preference': preference(asset['@quality']), + 'ext': ext, + }) + + downloads = media_data.get('Downloads') + if downloads: + for source in downloads['Download']: + for asset in source['url']: + formats.append({ + 'url': asset['text'], + 'preference': preference(asset['@quality']) + }) + self._sort_formats(formats) + + return { + 'id': media_id, + 'title': title, + 'description': description, + 'timestamp': timestamp, + 'thumbnails': thumbnails, + 'formats': formats, + } + + +class SRGSSRPlayIE(InfoExtractor): + _VALID_URL = r'https?://(?:(?:www|play)\.)?(?Psrf|rts|rsi|rtr|swi)\.ch/play/(?:tv|radio)/[^/]+/(?Pvideo|audio)/[^?]+\?id=(?P[0-9a-f\-]{36}|\d+)' + + _TESTS = [{ + 'url': 'http://www.srf.ch/play/tv/10vor10/video/snowden-beantragt-asyl-in-russland?id=28e1a57d-5b76-4399-8ab3-9097f071e6c5', + 'md5': '4cd93523723beff51bb4bee974ee238d', + 'info_dict': { + 'id': '28e1a57d-5b76-4399-8ab3-9097f071e6c5', + 'ext': 'm4v', + 'upload_date': '20130701', + 'title': 'Snowden beantragt Asyl in Russland', + 'timestamp': 1372713995, + } + }, { + # No Speichern (Save) button + 'url': 'http://www.srf.ch/play/tv/top-gear/video/jaguar-xk120-shadow-und-tornado-dampflokomotive?id=677f5829-e473-4823-ac83-a1087fe97faa', + 'md5': '0a274ce38fda48c53c01890651985bc6', + 'info_dict': { + 'id': '677f5829-e473-4823-ac83-a1087fe97faa', + 'ext': 'flv', + 'upload_date': '20130710', + 'title': 'Jaguar XK120, Shadow und Tornado-Dampflokomotive', + 'description': 'md5:88604432b60d5a38787f152dec89cd56', + 'timestamp': 1373493600, + }, + }, { + 'url': 'http://www.srf.ch/player/tv/10vor10/video/snowden-beantragt-asyl-in-russland?id=28e1a57d-5b76-4399-8ab3-9097f071e6c5', + 'only_matching': True, + }] + + def _real_extract(self, url): + bu, media_type, media_id = re.match(self._VALID_URL, url).groups() + return self.url_result('urn:%s:%s:%s' % (bu, media_type, media_id), 'SRGSSR') -- cgit v1.2.3 From 05ad5409b4fd044169ea0f67b9ae92d555564c4e Mon Sep 17 00:00:00 2001 From: remitamine Date: Fri, 9 Oct 2015 20:34:03 +0100 Subject: [srgssr] fix regex for swissinfo.ch --- youtube_dl/extractor/srgssr.py | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) (limited to 'youtube_dl/extractor/srgssr.py') diff --git a/youtube_dl/extractor/srgssr.py b/youtube_dl/extractor/srgssr.py index addf4d26e..3b5dcc503 100644 --- a/youtube_dl/extractor/srgssr.py +++ b/youtube_dl/extractor/srgssr.py @@ -96,7 +96,7 @@ class SRGSSRIE(InfoExtractor): class SRGSSRPlayIE(InfoExtractor): - _VALID_URL = r'https?://(?:(?:www|play)\.)?(?Psrf|rts|rsi|rtr|swi)\.ch/play/(?:tv|radio)/[^/]+/(?Pvideo|audio)/[^?]+\?id=(?P[0-9a-f\-]{36}|\d+)' + _VALID_URL = r'https?://(?:(?:www|play)\.)?(?Psrf|rts|rsi|rtr|swissinfo)\.ch/play/(?:tv|radio)/[^/]+/(?Pvideo|audio)/[^?]+\?id=(?P[0-9a-f\-]{36}|\d+)' _TESTS = [{ 'url': 'http://www.srf.ch/play/tv/10vor10/video/snowden-beantragt-asyl-in-russland?id=28e1a57d-5b76-4399-8ab3-9097f071e6c5', @@ -120,11 +120,8 @@ class SRGSSRPlayIE(InfoExtractor): 'description': 'md5:88604432b60d5a38787f152dec89cd56', 'timestamp': 1373493600, }, - }, { - 'url': 'http://www.srf.ch/player/tv/10vor10/video/snowden-beantragt-asyl-in-russland?id=28e1a57d-5b76-4399-8ab3-9097f071e6c5', - 'only_matching': True, }] def _real_extract(self, url): bu, media_type, media_id = re.match(self._VALID_URL, url).groups() - return self.url_result('urn:%s:%s:%s' % (bu, media_type, media_id), 'SRGSSR') + return self.url_result('urn:%s:%s:%s' % (bu[:3], media_type, media_id), 'SRGSSR') -- cgit v1.2.3 From e09f58b3bc3af6ce1e541fb7d034fe869fba6e82 Mon Sep 17 00:00:00 2001 From: remitamine Date: Wed, 14 Oct 2015 10:40:54 +0100 Subject: [srgssr] change the url chortcut, fix image extraction ,add a test and extract format id --- youtube_dl/extractor/srgssr.py | 35 +++++++++++++++++++++++++---------- 1 file changed, 25 insertions(+), 10 deletions(-) (limited to 'youtube_dl/extractor/srgssr.py') diff --git a/youtube_dl/extractor/srgssr.py b/youtube_dl/extractor/srgssr.py index 3b5dcc503..f759e5600 100644 --- a/youtube_dl/extractor/srgssr.py +++ b/youtube_dl/extractor/srgssr.py @@ -12,7 +12,7 @@ from ..utils import ( class SRGSSRIE(InfoExtractor): - _VALID_URL = r'(?:https?://tp\.srgssr\.ch/p(?:/[^/]+)+\?urn=)?urn:(?Psrf|rts|rsi|rtr|swi):(?:[^:]+:)?(?Pvideo|audio):(?P[0-9a-f\-]{36}|\d+)' + _VALID_URL = r'(?:https?://tp\.srgssr\.ch/p(?:/[^/]+)+\?urn=urn|srgssr):(?Psrf|rts|rsi|rtr|swi):(?:[^:]+:)?(?Pvideo|audio):(?P[0-9a-f\-]{36}|\d+)' _ERRORS = { 'AGERATING12': 'To protect children under the age of 12, this video is only available between 8 p.m. and 6 a.m.', @@ -44,11 +44,12 @@ class SRGSSRIE(InfoExtractor): timestamp = parse_iso8601(created_date) thumbnails = [] - for image in media_data['Image']['ImageRepresentations']['ImageRepresentation']: - thumbnails.append({ - 'id': image.get('id'), - 'url': image['url'], - }) + if 'Image' in media_data: + for image in media_data['Image']['ImageRepresentations']['ImageRepresentation']: + thumbnails.append({ + 'id': image.get('id'), + 'url': image['url'], + }) preference = qualities(['LQ', 'MQ', 'SD', 'HQ', 'HD']) formats = [] @@ -70,16 +71,17 @@ class SRGSSRIE(InfoExtractor): if asset_url.startswith('rtmp'): ext = self._search_regex(r'([a-z0-9]+):[^/]+', asset_url, 'ext') formats.append({ + 'format_id': asset['@quality'], 'url': asset_url, 'preference': preference(asset['@quality']), 'ext': ext, }) - downloads = media_data.get('Downloads') - if downloads: - for source in downloads['Download']: + if 'Downloads' in media_data: + for source in media_data['Downloads']['Download']: for asset in source['url']: formats.append({ + 'format_id': asset['@quality'], 'url': asset['text'], 'preference': preference(asset['@quality']) }) @@ -120,8 +122,21 @@ class SRGSSRPlayIE(InfoExtractor): 'description': 'md5:88604432b60d5a38787f152dec89cd56', 'timestamp': 1373493600, }, + },{ + 'url': 'http://www.rtr.ch/play/radio/actualitad/audio/saira-tujetsch-tuttina-cuntinuar-cun-sedrun-muster-turissem?id=63cb0778-27f8-49af-9284-8c7a8c6d15fc', + 'info_dict': { + 'id': '63cb0778-27f8-49af-9284-8c7a8c6d15fc', + 'ext': 'mp3', + 'upload_date': '20151013', + 'title': 'Saira: Tujetsch - tuttina cuntinuar cun Sedrun Mustér Turissem', + 'timestamp': 1444750398, + }, + 'params': { + # rtmp download + 'skip_download': True, + }, }] def _real_extract(self, url): bu, media_type, media_id = re.match(self._VALID_URL, url).groups() - return self.url_result('urn:%s:%s:%s' % (bu[:3], media_type, media_id), 'SRGSSR') + return self.url_result('srgssr:%s:%s:%s' % (bu[:3], media_type, media_id), 'SRGSSR') -- cgit v1.2.3 From 7db2897ded3d5970131fb34210f8510a02c672b8 Mon Sep 17 00:00:00 2001 From: remitamine Date: Sat, 5 Dec 2015 15:57:10 +0100 Subject: [srgssr] handle all play urls only in SRGSSRIE and keep RTSIE for articles --- youtube_dl/extractor/srgssr.py | 49 +++++++++++++++++++++++++++++++----------- 1 file changed, 37 insertions(+), 12 deletions(-) (limited to 'youtube_dl/extractor/srgssr.py') diff --git a/youtube_dl/extractor/srgssr.py b/youtube_dl/extractor/srgssr.py index f759e5600..563851b3f 100644 --- a/youtube_dl/extractor/srgssr.py +++ b/youtube_dl/extractor/srgssr.py @@ -17,25 +17,29 @@ class SRGSSRIE(InfoExtractor): _ERRORS = { 'AGERATING12': 'To protect children under the age of 12, this video is only available between 8 p.m. and 6 a.m.', 'AGERATING18': 'To protect children under the age of 18, this video is only available between 11 p.m. and 5 a.m.', -# 'ENDDATE': 'For legal reasons, this video was only available for a specified period of time.', + # 'ENDDATE': 'For legal reasons, this video was only available for a specified period of time.', 'GEOBLOCK': 'For legal reasons, this video is only available in Switzerland.', 'LEGAL': 'The video cannot be transmitted for legal reasons.', 'STARTDATE': 'This video is not yet available. Please try again later.', } - def _real_extract(self, url): - bu, media_type, media_id = re.match(self._VALID_URL, url).groups() - + def get_media_data(self, bu, media_type, media_id): media_data = self._download_json( 'http://il.srgssr.ch/integrationlayer/1.0/ue/%s/%s/play/%s.json' % (bu, media_type, media_id), media_id)[media_type.capitalize()] if media_data.get('block') and media_data['block'] in self._ERRORS: - raise ExtractorError( - '%s said: %s' % ( - self.IE_NAME, - self._ERRORS[media_data['block']]), - expected=True) + raise ExtractorError('%s said: %s' % (self.IE_NAME, self._ERRORS[media_data['block']]), expected=True) + + return media_data + + def _real_extract(self, url): + bu, media_type, media_id = re.match(self._VALID_URL, url).groups() + + if bu == 'rts': + return self.url_result('rts:%s' % media_id, 'RTS') + + media_data = self.get_media_data(bu, media_type, media_id) metadata = media_data['AssetMetadatas']['AssetMetadata'][0] title = metadata['title'] @@ -61,9 +65,13 @@ class SRGSSRIE(InfoExtractor): assets[quality['@quality']] = quality['text'] asset_url = assets.get('HD') or assets.get('HQ') or assets.get('SD') or assets.get('MQ') or assets.get('LQ') if '.f4m' in asset_url: - formats.extend(self._extract_f4m_formats(asset_url + '?hdcore=3.4.0', media_id, f4m_id='hds')) + f4m_formats = formats.extend(self._extract_f4m_formats(asset_url + '?hdcore=3.4.0', media_id, f4m_id='hds', fatal=False)) + if f4m_formats: + formats.extend(f4m_formats) elif '.m3u8' in asset_url: - formats.extend(self._extract_m3u8_formats(asset_url, media_id, m3u8_id='hls')) + m3u8_formats = formats.extend(self._extract_m3u8_formats(asset_url, media_id, m3u8_id='hls', fatal=False)) + if m3u8_formats: + formats.extend(m3u8_formats) else: for asset in source['url']: asset_url = asset['text'] @@ -122,7 +130,7 @@ class SRGSSRPlayIE(InfoExtractor): 'description': 'md5:88604432b60d5a38787f152dec89cd56', 'timestamp': 1373493600, }, - },{ + }, { 'url': 'http://www.rtr.ch/play/radio/actualitad/audio/saira-tujetsch-tuttina-cuntinuar-cun-sedrun-muster-turissem?id=63cb0778-27f8-49af-9284-8c7a8c6d15fc', 'info_dict': { 'id': '63cb0778-27f8-49af-9284-8c7a8c6d15fc', @@ -135,8 +143,25 @@ class SRGSSRPlayIE(InfoExtractor): # rtmp download 'skip_download': True, }, + }, { + 'url': 'http://www.rts.ch/play/tv/-/video/le-19h30?id=6348260', + 'md5': '67a2a9ae4e8e62a68d0e9820cc9782df', + 'info_dict': { + 'id': '6348260', + 'display_id': '6348260', + 'ext': 'flv', + 'duration': 1796, + 'title': 'Le 19h30', + 'description': '', + 'uploader': '19h30', + 'upload_date': '20141201', + 'timestamp': 1417458600, + 'thumbnail': 're:^https?://.*\.image', + 'view_count': int, + }, }] def _real_extract(self, url): bu, media_type, media_id = re.match(self._VALID_URL, url).groups() + # other info can be extracted from url + '&layout=json' return self.url_result('srgssr:%s:%s:%s' % (bu[:3], media_type, media_id), 'SRGSSR') -- cgit v1.2.3 From 422f7c112c12d92bf06c95ff33782c8e9a0df43e Mon Sep 17 00:00:00 2001 From: remitamine Date: Tue, 29 Dec 2015 11:36:04 +0100 Subject: [srgssr] update tests --- youtube_dl/extractor/srgssr.py | 38 +++++++++++++------------------------- 1 file changed, 13 insertions(+), 25 deletions(-) (limited to 'youtube_dl/extractor/srgssr.py') diff --git a/youtube_dl/extractor/srgssr.py b/youtube_dl/extractor/srgssr.py index 563851b3f..4962fe885 100644 --- a/youtube_dl/extractor/srgssr.py +++ b/youtube_dl/extractor/srgssr.py @@ -47,31 +47,24 @@ class SRGSSRIE(InfoExtractor): created_date = media_data.get('createdDate') or metadata.get('createdDate') timestamp = parse_iso8601(created_date) - thumbnails = [] - if 'Image' in media_data: - for image in media_data['Image']['ImageRepresentations']['ImageRepresentation']: - thumbnails.append({ - 'id': image.get('id'), - 'url': image['url'], - }) + thumbnails = [{ + 'id': image.get('id'), + 'url': image['url'], + } for image in media_data.get('Image', {}).get('ImageRepresentations', {}).get('ImageRepresentation', [])] preference = qualities(['LQ', 'MQ', 'SD', 'HQ', 'HD']) formats = [] - for source in media_data['Playlists']['Playlist']: - assets = {} + for source in media_data.get('Playlists', {}).get('Playlist', []) + media_data.get('Downloads', {}).get('Download', []): protocol = source.get('@protocol') if protocol in ('HTTP-HDS', 'HTTP-HLS'): + assets = {} for quality in source['url']: assets[quality['@quality']] = quality['text'] asset_url = assets.get('HD') or assets.get('HQ') or assets.get('SD') or assets.get('MQ') or assets.get('LQ') if '.f4m' in asset_url: - f4m_formats = formats.extend(self._extract_f4m_formats(asset_url + '?hdcore=3.4.0', media_id, f4m_id='hds', fatal=False)) - if f4m_formats: - formats.extend(f4m_formats) + formats.extend(self._extract_f4m_formats(asset_url + '?hdcore=3.4.0', media_id, f4m_id='hds', fatal=False)) elif '.m3u8' in asset_url: - m3u8_formats = formats.extend(self._extract_m3u8_formats(asset_url, media_id, m3u8_id='hls', fatal=False)) - if m3u8_formats: - formats.extend(m3u8_formats) + formats.extend(self._extract_m3u8_formats(asset_url, media_id, m3u8_id='hls', fatal=False)) else: for asset in source['url']: asset_url = asset['text'] @@ -84,15 +77,6 @@ class SRGSSRIE(InfoExtractor): 'preference': preference(asset['@quality']), 'ext': ext, }) - - if 'Downloads' in media_data: - for source in media_data['Downloads']['Download']: - for asset in source['url']: - formats.append({ - 'format_id': asset['@quality'], - 'url': asset['text'], - 'preference': preference(asset['@quality']) - }) self._sort_formats(formats) return { @@ -149,7 +133,7 @@ class SRGSSRPlayIE(InfoExtractor): 'info_dict': { 'id': '6348260', 'display_id': '6348260', - 'ext': 'flv', + 'ext': 'mp4', 'duration': 1796, 'title': 'Le 19h30', 'description': '', @@ -159,6 +143,10 @@ class SRGSSRPlayIE(InfoExtractor): 'thumbnail': 're:^https?://.*\.image', 'view_count': int, }, + 'params': { + # m3u8 download + 'skip_download': True, + } }] def _real_extract(self, url): -- cgit v1.2.3