1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-22 18:22:21 +00:00
youtube-dl/youtube_dl/extractor/rai.py

500 lines
18 KiB
Python

from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..compat import (
compat_urlparse,
compat_str,
)
from ..utils import (
ExtractorError,
determine_ext,
find_xpath_attr,
fix_xml_ampersands,
GeoRestrictedError,
int_or_none,
parse_duration,
strip_or_none,
try_get,
unescapeHTML,
unified_strdate,
unified_timestamp,
update_url_query,
urljoin,
xpath_text,
)
class RaiBaseIE(InfoExtractor):
_UUID_RE = r'[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}'
_GEO_COUNTRIES = ['IT']
_GEO_BYPASS = False
def _extract_relinker_info(self, relinker_url, video_id):
if not re.match(r'https?://', relinker_url):
return {'formats': [{'url': relinker_url}]}
formats = []
geoprotection = None
is_live = None
duration = None
for platform in ('mon', 'flash', 'native'):
relinker = self._download_xml(
relinker_url, video_id,
note='Downloading XML metadata for platform %s' % platform,
transform_source=fix_xml_ampersands,
query={'output': 45, 'pl': platform},
headers=self.geo_verification_headers())
if not geoprotection:
geoprotection = xpath_text(
relinker, './geoprotection', default=None) == 'Y'
if not is_live:
is_live = xpath_text(
relinker, './is_live', default=None) == 'Y'
if not duration:
duration = parse_duration(xpath_text(
relinker, './duration', default=None))
url_elem = find_xpath_attr(relinker, './url', 'type', 'content')
if url_elem is None:
continue
media_url = url_elem.text
# This does not imply geo restriction (e.g.
# http://www.raisport.rai.it/dl/raiSport/media/rassegna-stampa-04a9f4bd-b563-40cf-82a6-aad3529cb4a9.html)
if media_url == 'http://download.rai.it/video_no_available.mp4':
continue
ext = determine_ext(media_url)
if (ext == 'm3u8' and platform != 'mon') or (ext == 'f4m' and platform != 'flash'):
continue
if ext == 'm3u8':
formats.extend(self._extract_m3u8_formats(
media_url, video_id, 'mp4', 'm3u8_native',
m3u8_id='hls', fatal=False))
elif ext == 'f4m':
manifest_url = update_url_query(
media_url.replace('manifest#live_hds.f4m', 'manifest.f4m'),
{'hdcore': '3.7.0', 'plugin': 'aasp-3.7.0.39.44'})
formats.extend(self._extract_f4m_formats(
manifest_url, video_id, f4m_id='hds', fatal=False))
else:
bitrate = int_or_none(xpath_text(relinker, 'bitrate'))
formats.append({
'url': media_url,
'tbr': bitrate if bitrate > 0 else None,
'format_id': 'http-%d' % bitrate if bitrate > 0 else 'http',
})
if not formats and geoprotection is True:
self.raise_geo_restricted(countries=self._GEO_COUNTRIES)
return dict((k, v) for k, v in {
'is_live': is_live,
'duration': duration,
'formats': formats,
}.items() if v is not None)
@staticmethod
def _extract_subtitles(url, subtitle_url):
subtitles = {}
if subtitle_url and isinstance(subtitle_url, compat_str):
subtitle_url = urljoin(url, subtitle_url)
STL_EXT = '.stl'
SRT_EXT = '.srt'
subtitles['it'] = [{
'ext': 'stl',
'url': subtitle_url,
}]
if subtitle_url.endswith(STL_EXT):
srt_url = subtitle_url[:-len(STL_EXT)] + SRT_EXT
subtitles['it'].append({
'ext': 'srt',
'url': srt_url,
})
return subtitles
class RaiPlayIE(RaiBaseIE):
_VALID_URL = r'(?P<url>https?://(?:www\.)?raiplay\.it/.+?-(?P<id>%s)\.html)' % RaiBaseIE._UUID_RE
_TESTS = [{
'url': 'http://www.raiplay.it/video/2016/10/La-Casa-Bianca-e06118bb-59a9-4636-b914-498e4cfd2c66.html?source=twitter',
'md5': '340aa3b7afb54bfd14a8c11786450d76',
'info_dict': {
'id': 'e06118bb-59a9-4636-b914-498e4cfd2c66',
'ext': 'mp4',
'title': 'La Casa Bianca',
'alt_title': 'S2016 - Puntata del 23/10/2016',
'description': 'md5:a09d45890850458077d1f68bb036e0a5',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'Rai 3',
'creator': 'Rai 3',
'duration': 3278,
'timestamp': 1477764300,
'upload_date': '20161029',
'series': 'La Casa Bianca',
'season': '2016',
},
}, {
'url': 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html',
'md5': '8970abf8caf8aef4696e7b1f2adfc696',
'info_dict': {
'id': 'cb27157f-9dd0-4aee-b788-b1f67643a391',
'ext': 'mp4',
'title': 'Report del 07/04/2014',
'alt_title': 'S2013/14 - Puntata del 07/04/2014',
'description': 'md5:f27c544694cacb46a078db84ec35d2d9',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'Rai 5',
'creator': 'Rai 5',
'duration': 6160,
'series': 'Report',
'season_number': 5,
'season': '2013/14',
},
'params': {
'skip_download': True,
},
}, {
'url': 'http://www.raiplay.it/video/2016/11/gazebotraindesi-efebe701-969c-4593-92f3-285f0d1ce750.html?',
'only_matching': True,
}]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
url, video_id = mobj.group('url', 'id')
media = self._download_json(
'%s?json' % url, video_id, 'Downloading video JSON')
title = media['name']
video = media['video']
relinker_info = self._extract_relinker_info(video['contentUrl'], video_id)
self._sort_formats(relinker_info['formats'])
thumbnails = []
if 'images' in media:
for _, value in media.get('images').items():
if value:
thumbnails.append({
'url': value.replace('[RESOLUTION]', '600x400')
})
timestamp = unified_timestamp(try_get(
media, lambda x: x['availabilities'][0]['start'], compat_str))
subtitles = self._extract_subtitles(url, video.get('subtitles'))
info = {
'id': video_id,
'title': self._live_title(title) if relinker_info.get(
'is_live') else title,
'alt_title': media.get('subtitle'),
'description': media.get('description'),
'uploader': strip_or_none(media.get('channel')),
'creator': strip_or_none(media.get('editor')),
'duration': parse_duration(video.get('duration')),
'timestamp': timestamp,
'thumbnails': thumbnails,
'series': try_get(
media, lambda x: x['isPartOf']['name'], compat_str),
'season_number': int_or_none(try_get(
media, lambda x: x['isPartOf']['numeroStagioni'])),
'season': media.get('stagione') or None,
'subtitles': subtitles,
}
info.update(relinker_info)
return info
class RaiPlayLiveIE(RaiBaseIE):
_VALID_URL = r'https?://(?:www\.)?raiplay\.it/dirette/(?P<id>[^/?#&]+)'
_TEST = {
'url': 'http://www.raiplay.it/dirette/rainews24',
'info_dict': {
'id': 'd784ad40-e0ae-4a69-aa76-37519d238a9c',
'display_id': 'rainews24',
'ext': 'mp4',
'title': 're:^Diretta di Rai News 24 [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
'description': 'md5:6eca31500550f9376819f174e5644754',
'uploader': 'Rai News 24',
'creator': 'Rai News 24',
'is_live': True,
},
'params': {
'skip_download': True,
},
}
def _real_extract(self, url):
display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id)
video_id = self._search_regex(
r'data-uniquename=["\']ContentItem-(%s)' % RaiBaseIE._UUID_RE,
webpage, 'content id')
return {
'_type': 'url_transparent',
'ie_key': RaiPlayIE.ie_key(),
'url': 'http://www.raiplay.it/dirette/ContentItem-%s.html' % video_id,
'id': video_id,
'display_id': display_id,
}
class RaiPlayPlaylistIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?raiplay\.it/programmi/(?P<id>[^/?#&]+)'
_TESTS = [{
'url': 'http://www.raiplay.it/programmi/nondirloalmiocapo/',
'info_dict': {
'id': 'nondirloalmiocapo',
'title': 'Non dirlo al mio capo',
'description': 'md5:9f3d603b2947c1c7abb098f3b14fac86',
},
'playlist_mincount': 12,
}]
def _real_extract(self, url):
playlist_id = self._match_id(url)
webpage = self._download_webpage(url, playlist_id)
title = self._html_search_meta(
('programma', 'nomeProgramma'), webpage, 'title')
description = unescapeHTML(self._html_search_meta(
('description', 'og:description'), webpage, 'description'))
print(description)
entries = []
for mobj in re.finditer(
r'<a\b[^>]+\bhref=(["\'])(?P<path>/raiplay/video/.+?)\1',
webpage):
video_url = urljoin(url, mobj.group('path'))
entries.append(self.url_result(
video_url, ie=RaiPlayIE.ie_key(),
video_id=RaiPlayIE._match_id(video_url)))
return self.playlist_result(entries, playlist_id, title, description)
class RaiIE(RaiBaseIE):
_VALID_URL = r'https?://[^/]+\.(?:rai\.(?:it|tv)|rainews\.it)/dl/.+?-(?P<id>%s)(?:-.+?)?\.html' % RaiBaseIE._UUID_RE
_TESTS = [{
# var uniquename = "ContentItem-..."
# data-id="ContentItem-..."
'url': 'http://www.raisport.rai.it/dl/raiSport/media/rassegna-stampa-04a9f4bd-b563-40cf-82a6-aad3529cb4a9.html',
'info_dict': {
'id': '04a9f4bd-b563-40cf-82a6-aad3529cb4a9',
'ext': 'mp4',
'title': 'TG PRIMO TEMPO',
'thumbnail': r're:^https?://.*\.jpg$',
'duration': 1758,
'upload_date': '20140612',
}
}, {
# with ContentItem in many metas
'url': 'http://www.rainews.it/dl/rainews/media/Weekend-al-cinema-da-Hollywood-arriva-il-thriller-di-Tate-Taylor-La-ragazza-del-treno-1632c009-c843-4836-bb65-80c33084a64b.html',
'info_dict': {
'id': '1632c009-c843-4836-bb65-80c33084a64b',
'ext': 'mp4',
'title': 'Weekend al cinema, da Hollywood arriva il thriller di Tate Taylor "La ragazza del treno"',
'description': 'I film in uscita questa settimana.',
'thumbnail': r're:^https?://.*\.png$',
'duration': 833,
'upload_date': '20161103',
}
}, {
# with ContentItem in og:url
'url': 'http://www.rai.it/dl/RaiTV/programmi/media/ContentItem-efb17665-691c-45d5-a60c-5301333cbb0c.html',
'md5': '11959b4e44fa74de47011b5799490adf',
'info_dict': {
'id': 'efb17665-691c-45d5-a60c-5301333cbb0c',
'ext': 'mp4',
'title': 'TG1 ore 20:00 del 03/11/2016',
'description': 'TG1 edizione integrale ore 20:00 del giorno 03/11/2016',
'thumbnail': r're:^https?://.*\.jpg$',
'duration': 2214,
'upload_date': '20161103',
}
}, {
# drawMediaRaiTV(...)
'url': 'http://www.report.rai.it/dl/Report/puntata/ContentItem-0c7a664b-d0f4-4b2c-8835-3f82e46f433e.html',
'md5': '2dd727e61114e1ee9c47f0da6914e178',
'info_dict': {
'id': '59d69d28-6bb6-409d-a4b5-ed44096560af',
'ext': 'mp4',
'title': 'Il pacco',
'description': 'md5:4b1afae1364115ce5d78ed83cd2e5b3a',
'thumbnail': r're:^https?://.*\.jpg$',
'upload_date': '20141221',
},
}, {
# initEdizione('ContentItem-...'
'url': 'http://www.tg1.rai.it/dl/tg1/2010/edizioni/ContentSet-9b6e0cba-4bef-4aef-8cf0-9f7f665b7dfb-tg1.html?item=undefined',
'info_dict': {
'id': 'c2187016-8484-4e3a-8ac8-35e475b07303',
'ext': 'mp4',
'title': r're:TG1 ore \d{2}:\d{2} del \d{2}/\d{2}/\d{4}',
'duration': 2274,
'upload_date': '20170401',
},
'skip': 'Changes daily',
}, {
# HDS live stream with only relinker URL
'url': 'http://www.rai.tv/dl/RaiTV/dirette/PublishingBlock-1912dbbf-3f96-44c3-b4cf-523681fbacbc.html?channel=EuroNews',
'info_dict': {
'id': '1912dbbf-3f96-44c3-b4cf-523681fbacbc',
'ext': 'flv',
'title': 'EuroNews',
},
'params': {
'skip_download': True,
},
}, {
# HLS live stream with ContentItem in og:url
'url': 'http://www.rainews.it/dl/rainews/live/ContentItem-3156f2f2-dc70-4953-8e2f-70d7489d4ce9.html',
'info_dict': {
'id': '3156f2f2-dc70-4953-8e2f-70d7489d4ce9',
'ext': 'mp4',
'title': 'La diretta di Rainews24',
},
'params': {
'skip_download': True,
},
}, {
# Direct MMS URL
'url': 'http://www.rai.it/dl/RaiTV/programmi/media/ContentItem-b63a4089-ac28-48cf-bca5-9f5b5bc46df5.html',
'only_matching': True,
}]
def _extract_from_content_id(self, content_id, url):
media = self._download_json(
'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-%s.html?json' % content_id,
content_id, 'Downloading video JSON')
title = media['name'].strip()
media_type = media['type']
if 'Audio' in media_type:
relinker_info = {
'formats': [{
'format_id': media.get('formatoAudio'),
'url': media['audioUrl'],
'ext': media.get('formatoAudio'),
}]
}
elif 'Video' in media_type:
relinker_info = self._extract_relinker_info(media['mediaUri'], content_id)
else:
raise ExtractorError('not a media file')
self._sort_formats(relinker_info['formats'])
thumbnails = []
for image_type in ('image', 'image_medium', 'image_300'):
thumbnail_url = media.get(image_type)
if thumbnail_url:
thumbnails.append({
'url': compat_urlparse.urljoin(url, thumbnail_url),
})
subtitles = self._extract_subtitles(url, media.get('subtitlesUrl'))
info = {
'id': content_id,
'title': title,
'description': strip_or_none(media.get('desc')),
'thumbnails': thumbnails,
'uploader': media.get('author'),
'upload_date': unified_strdate(media.get('date')),
'duration': parse_duration(media.get('length')),
'subtitles': subtitles,
}
info.update(relinker_info)
return info
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
content_item_id = None
content_item_url = self._html_search_meta(
('og:url', 'og:video', 'og:video:secure_url', 'twitter:url',
'twitter:player', 'jsonlink'), webpage, default=None)
if content_item_url:
content_item_id = self._search_regex(
r'ContentItem-(%s)' % self._UUID_RE, content_item_url,
'content item id', default=None)
if not content_item_id:
content_item_id = self._search_regex(
r'''(?x)
(?:
(?:initEdizione|drawMediaRaiTV)\(|
<(?:[^>]+\bdata-id|var\s+uniquename)=
)
(["\'])
(?:(?!\1).)*\bContentItem-(?P<id>%s)
''' % self._UUID_RE,
webpage, 'content item id', default=None, group='id')
content_item_ids = set()
if content_item_id:
content_item_ids.add(content_item_id)
if video_id not in content_item_ids:
content_item_ids.add(video_id)
for content_item_id in content_item_ids:
try:
return self._extract_from_content_id(content_item_id, url)
except GeoRestrictedError:
raise
except ExtractorError:
pass
relinker_url = self._search_regex(
r'''(?x)
(?:
var\s+videoURL|
mediaInfo\.mediaUri
)\s*=\s*
([\'"])
(?P<url>
(?:https?:)?
//mediapolis(?:vod)?\.rai\.it/relinker/relinkerServlet\.htm\?
(?:(?!\1).)*\bcont=(?:(?!\1).)+)\1
''',
webpage, 'relinker URL', group='url')
relinker_info = self._extract_relinker_info(
urljoin(url, relinker_url), video_id)
self._sort_formats(relinker_info['formats'])
title = self._search_regex(
r'var\s+videoTitolo\s*=\s*([\'"])(?P<title>[^\'"]+)\1',
webpage, 'title', group='title',
default=None) or self._og_search_title(webpage)
info = {
'id': video_id,
'title': title,
}
info.update(relinker_info)
return info