mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-16 21:33:19 +00:00
86e5f3ed2e
Using https://github.com/asottile/pyupgrade 1. `__future__` imports and `coding: utf-8` were removed 2. Files were rewritten with `pyupgrade --py36-plus --keep-percent-format` 3. f-strings were cherry-picked from `pyupgrade --py36-plus` Extractors are left untouched (except removing header) to avoid unnecessary merge conflicts
133 lines
4.6 KiB
Python
133 lines
4.6 KiB
Python
from .common import InfoExtractor
|
|
from ..compat import compat_str
|
|
from ..utils import (
|
|
dict_get,
|
|
int_or_none,
|
|
try_get,
|
|
)
|
|
|
|
|
|
class ABCOTVSIE(InfoExtractor):
|
|
IE_NAME = 'abcotvs'
|
|
IE_DESC = 'ABC Owned Television Stations'
|
|
_VALID_URL = r'https?://(?P<site>abc(?:7(?:news|ny|chicago)?|11|13|30)|6abc)\.com(?:(?:/[^/]+)*/(?P<display_id>[^/]+))?/(?P<id>\d+)'
|
|
_TESTS = [
|
|
{
|
|
'url': 'http://abc7news.com/entertainment/east-bay-museum-celebrates-vintage-synthesizers/472581/',
|
|
'info_dict': {
|
|
'id': '472548',
|
|
'display_id': 'east-bay-museum-celebrates-vintage-synthesizers',
|
|
'ext': 'mp4',
|
|
'title': 'East Bay museum celebrates synthesized music',
|
|
'description': 'md5:24ed2bd527096ec2a5c67b9d5a9005f3',
|
|
'thumbnail': r're:^https?://.*\.jpg$',
|
|
'timestamp': 1421118520,
|
|
'upload_date': '20150113',
|
|
},
|
|
'params': {
|
|
# m3u8 download
|
|
'skip_download': True,
|
|
},
|
|
},
|
|
{
|
|
'url': 'http://abc7news.com/472581',
|
|
'only_matching': True,
|
|
},
|
|
{
|
|
'url': 'https://6abc.com/man-75-killed-after-being-struck-by-vehicle-in-chester/5725182/',
|
|
'only_matching': True,
|
|
},
|
|
]
|
|
_SITE_MAP = {
|
|
'6abc': 'wpvi',
|
|
'abc11': 'wtvd',
|
|
'abc13': 'ktrk',
|
|
'abc30': 'kfsn',
|
|
'abc7': 'kabc',
|
|
'abc7chicago': 'wls',
|
|
'abc7news': 'kgo',
|
|
'abc7ny': 'wabc',
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
site, display_id, video_id = self._match_valid_url(url).groups()
|
|
display_id = display_id or video_id
|
|
station = self._SITE_MAP[site]
|
|
|
|
data = self._download_json(
|
|
'https://api.abcotvs.com/v2/content', display_id, query={
|
|
'id': video_id,
|
|
'key': 'otv.web.%s.story' % station,
|
|
'station': station,
|
|
})['data']
|
|
video = try_get(data, lambda x: x['featuredMedia']['video'], dict) or data
|
|
video_id = compat_str(dict_get(video, ('id', 'publishedKey'), video_id))
|
|
title = video.get('title') or video['linkText']
|
|
|
|
formats = []
|
|
m3u8_url = video.get('m3u8')
|
|
if m3u8_url:
|
|
formats = self._extract_m3u8_formats(
|
|
video['m3u8'].split('?')[0], display_id, 'mp4', m3u8_id='hls', fatal=False)
|
|
mp4_url = video.get('mp4')
|
|
if mp4_url:
|
|
formats.append({
|
|
'abr': 128,
|
|
'format_id': 'https',
|
|
'height': 360,
|
|
'url': mp4_url,
|
|
'width': 640,
|
|
})
|
|
self._sort_formats(formats)
|
|
|
|
image = video.get('image') or {}
|
|
|
|
return {
|
|
'id': video_id,
|
|
'display_id': display_id,
|
|
'title': title,
|
|
'description': dict_get(video, ('description', 'caption'), try_get(video, lambda x: x['meta']['description'])),
|
|
'thumbnail': dict_get(image, ('source', 'dynamicSource')),
|
|
'timestamp': int_or_none(video.get('date')),
|
|
'duration': int_or_none(video.get('length')),
|
|
'formats': formats,
|
|
}
|
|
|
|
|
|
class ABCOTVSClipsIE(InfoExtractor):
|
|
IE_NAME = 'abcotvs:clips'
|
|
_VALID_URL = r'https?://clips\.abcotvs\.com/(?:[^/]+/)*video/(?P<id>\d+)'
|
|
_TEST = {
|
|
'url': 'https://clips.abcotvs.com/kabc/video/214814',
|
|
'info_dict': {
|
|
'id': '214814',
|
|
'ext': 'mp4',
|
|
'title': 'SpaceX launch pad explosion destroys rocket, satellite',
|
|
'description': 'md5:9f186e5ad8f490f65409965ee9c7be1b',
|
|
'upload_date': '20160901',
|
|
'timestamp': 1472756695,
|
|
},
|
|
'params': {
|
|
# m3u8 download
|
|
'skip_download': True,
|
|
},
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
video_data = self._download_json('https://clips.abcotvs.com/vogo/video/getByIds?ids=' + video_id, video_id)['results'][0]
|
|
title = video_data['title']
|
|
formats = self._extract_m3u8_formats(
|
|
video_data['videoURL'].split('?')[0], video_id, 'mp4')
|
|
self._sort_formats(formats)
|
|
|
|
return {
|
|
'id': video_id,
|
|
'title': title,
|
|
'description': video_data.get('description'),
|
|
'thumbnail': video_data.get('thumbnailURL'),
|
|
'duration': int_or_none(video_data.get('duration')),
|
|
'timestamp': int_or_none(video_data.get('pubDate')),
|
|
'formats': formats,
|
|
}
|