[bbc] add support for BBC Reel videos(closes #21870, closes #23660, closes #28268)

This commit is contained in:
dirkf 2021-02-24 11:52:30 +00:00 committed by Remita Amine
parent 7c06216abf
commit e465b25c1f

View file

@ -5,10 +5,15 @@ import itertools
import re import re
from .common import InfoExtractor from .common import InfoExtractor
from ..compat import (
compat_etree_Element,
compat_HTTPError,
compat_urlparse,
)
from ..utils import ( from ..utils import (
ExtractorError,
clean_html, clean_html,
dict_get, dict_get,
ExtractorError,
float_or_none, float_or_none,
get_element_by_class, get_element_by_class,
int_or_none, int_or_none,
@ -21,11 +26,6 @@ from ..utils import (
urlencode_postdata, urlencode_postdata,
urljoin, urljoin,
) )
from ..compat import (
compat_etree_Element,
compat_HTTPError,
compat_urlparse,
)
class BBCCoUkIE(InfoExtractor): class BBCCoUkIE(InfoExtractor):
@ -793,6 +793,20 @@ class BBCIE(BBCCoUkIE):
'description': 'Learn English words and phrases from this story', 'description': 'Learn English words and phrases from this story',
}, },
'add_ie': [BBCCoUkIE.ie_key()], 'add_ie': [BBCCoUkIE.ie_key()],
}, {
# BBC Reel
'url': 'https://www.bbc.com/reel/video/p07c6sb6/how-positive-thinking-is-harming-your-happiness',
'info_dict': {
'id': 'p07c6sb9',
'ext': 'mp4',
'title': 'How positive thinking is harming your happiness',
'alt_title': 'The downsides of positive thinking',
'description': 'md5:fad74b31da60d83b8265954ee42d85b4',
'duration': 235,
'thumbnail': r're:https?://.+/p07c9dsr.jpg',
'upload_date': '20190604',
'categories': ['Psychology'],
},
}] }]
@classmethod @classmethod
@ -980,6 +994,37 @@ class BBCIE(BBCCoUkIE):
'subtitles': subtitles, 'subtitles': subtitles,
} }
# bbc reel (e.g. https://www.bbc.com/reel/video/p07c6sb6/how-positive-thinking-is-harming-your-happiness)
initial_data = self._parse_json(self._html_search_regex(
r'<script[^>]+id=(["\'])initial-data\1[^>]+data-json=(["\'])(?P<json>(?:(?!\2).)+)',
webpage, 'initial data', default='{}', group='json'), playlist_id, fatal=False)
if initial_data:
init_data = try_get(
initial_data, lambda x: x['initData']['items'][0], dict) or {}
smp_data = init_data.get('smpData') or {}
clip_data = try_get(smp_data, lambda x: x['items'][0], dict) or {}
version_id = clip_data.get('versionID')
if version_id:
title = smp_data['title']
formats, subtitles = self._download_media_selector(version_id)
self._sort_formats(formats)
image_url = smp_data.get('holdingImageURL')
display_date = init_data.get('displayDate')
topic_title = init_data.get('topicTitle')
return {
'id': version_id,
'title': title,
'formats': formats,
'alt_title': init_data.get('shortTitle'),
'thumbnail': image_url.replace('$recipe', 'raw') if image_url else None,
'description': smp_data.get('summary') or init_data.get('shortSummary'),
'upload_date': display_date.replace('-', '') if display_date else None,
'subtitles': subtitles,
'duration': int_or_none(clip_data.get('duration')),
'categories': [topic_title] if topic_title else None,
}
# Morph based embed (e.g. http://www.bbc.co.uk/sport/live/olympics/36895975) # Morph based embed (e.g. http://www.bbc.co.uk/sport/live/olympics/36895975)
# There are several setPayload calls may be present but the video # There are several setPayload calls may be present but the video
# seems to be always related to the first one # seems to be always related to the first one
@ -1041,7 +1086,7 @@ class BBCIE(BBCCoUkIE):
thumbnail = None thumbnail = None
image_url = current_programme.get('image_url') image_url = current_programme.get('image_url')
if image_url: if image_url:
thumbnail = image_url.replace('{recipe}', '1920x1920') thumbnail = image_url.replace('{recipe}', 'raw')
return { return {
'id': programme_id, 'id': programme_id,
'title': title, 'title': title,