mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-21 20:46:36 -05:00
Update to ytdl-commit-7e79ba7
7e79ba7dd6
[vimeo:album] Fix extraction for albums with number of videos multiple to page sizeThis commit is contained in:
pukkandan
parent
c224251aad
commit
421a459573
5 changed files with 208 additions and 45 deletions
|
|
@ -708,7 +708,10 @@
|
|||
MixcloudUserIE,
|
||||
MixcloudPlaylistIE,
|
||||
)
|
||||
from .mlb import MLBIE
|
||||
from .mlb import (
|
||||
MLBIE,
|
||||
MLBVideoIE,
|
||||
)
|
||||
from .mnet import MnetIE
|
||||
from .moevideo import MoeVideoIE
|
||||
from .mofosex import (
|
||||
|
|
|
|||
|
|
@ -1,15 +1,91 @@
|
|||
from __future__ import unicode_literals
|
||||
|
||||
from .nhl import NHLBaseIE
|
||||
import re
|
||||
|
||||
from .common import InfoExtractor
|
||||
from ..utils import (
|
||||
determine_ext,
|
||||
int_or_none,
|
||||
parse_duration,
|
||||
parse_iso8601,
|
||||
try_get,
|
||||
)
|
||||
|
||||
|
||||
class MLBIE(NHLBaseIE):
|
||||
class MLBBaseIE(InfoExtractor):
|
||||
def _real_extract(self, url):
|
||||
display_id = self._match_id(url)
|
||||
video = self._download_video_data(display_id)
|
||||
video_id = video['id']
|
||||
title = video['title']
|
||||
feed = self._get_feed(video)
|
||||
|
||||
formats = []
|
||||
for playback in (feed.get('playbacks') or []):
|
||||
playback_url = playback.get('url')
|
||||
if not playback_url:
|
||||
continue
|
||||
name = playback.get('name')
|
||||
ext = determine_ext(playback_url)
|
||||
if ext == 'm3u8':
|
||||
formats.extend(self._extract_m3u8_formats(
|
||||
playback_url, video_id, 'mp4',
|
||||
'm3u8_native', m3u8_id=name, fatal=False))
|
||||
else:
|
||||
f = {
|
||||
'format_id': name,
|
||||
'url': playback_url,
|
||||
}
|
||||
mobj = re.search(r'_(\d+)K_(\d+)X(\d+)', name)
|
||||
if mobj:
|
||||
f.update({
|
||||
'height': int(mobj.group(3)),
|
||||
'tbr': int(mobj.group(1)),
|
||||
'width': int(mobj.group(2)),
|
||||
})
|
||||
mobj = re.search(r'_(\d+)x(\d+)_(\d+)_(\d+)K\.mp4', playback_url)
|
||||
if mobj:
|
||||
f.update({
|
||||
'fps': int(mobj.group(3)),
|
||||
'height': int(mobj.group(2)),
|
||||
'tbr': int(mobj.group(4)),
|
||||
'width': int(mobj.group(1)),
|
||||
})
|
||||
formats.append(f)
|
||||
self._sort_formats(formats)
|
||||
|
||||
thumbnails = []
|
||||
for cut in (try_get(feed, lambda x: x['image']['cuts'], list) or []):
|
||||
src = cut.get('src')
|
||||
if not src:
|
||||
continue
|
||||
thumbnails.append({
|
||||
'height': int_or_none(cut.get('height')),
|
||||
'url': src,
|
||||
'width': int_or_none(cut.get('width')),
|
||||
})
|
||||
|
||||
language = (video.get('language') or 'EN').lower()
|
||||
|
||||
return {
|
||||
'id': video_id,
|
||||
'title': title,
|
||||
'formats': formats,
|
||||
'description': video.get('description'),
|
||||
'duration': parse_duration(feed.get('duration')),
|
||||
'thumbnails': thumbnails,
|
||||
'timestamp': parse_iso8601(video.get(self._TIMESTAMP_KEY)),
|
||||
'subtitles': self._extract_mlb_subtitles(feed, language),
|
||||
}
|
||||
|
||||
|
||||
class MLBIE(MLBBaseIE):
|
||||
_VALID_URL = r'''(?x)
|
||||
https?://
|
||||
(?:[\da-z_-]+\.)*(?P<site>mlb)\.com/
|
||||
(?:[\da-z_-]+\.)*mlb\.com/
|
||||
(?:
|
||||
(?:
|
||||
(?:[^/]+/)*c-|
|
||||
(?:[^/]+/)*video/[^/]+/c-|
|
||||
(?:
|
||||
shared/video/embed/(?:embed|m-internal-embed)\.html|
|
||||
(?:[^/]+/)+(?:play|index)\.jsp|
|
||||
|
|
@ -18,7 +94,6 @@ class MLBIE(NHLBaseIE):
|
|||
(?P<id>\d+)
|
||||
)
|
||||
'''
|
||||
_CONTENT_DOMAIN = 'content.mlb.com'
|
||||
_TESTS = [
|
||||
{
|
||||
'url': 'https://www.mlb.com/mariners/video/ackleys-spectacular-catch/c-34698933',
|
||||
|
|
@ -76,18 +151,6 @@ class MLBIE(NHLBaseIE):
|
|||
'thumbnail': r're:^https?://.*\.jpg$',
|
||||
},
|
||||
},
|
||||
{
|
||||
'url': 'https://www.mlb.com/news/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer/c-118550098',
|
||||
'md5': 'e09e37b552351fddbf4d9e699c924d68',
|
||||
'info_dict': {
|
||||
'id': '75609783',
|
||||
'ext': 'mp4',
|
||||
'title': 'Must C: Pillar climbs for catch',
|
||||
'description': '4/15/15: Blue Jays outfielder Kevin Pillar continues his defensive dominance by climbing the wall in left to rob Tim Beckham of a home run',
|
||||
'timestamp': 1429139220,
|
||||
'upload_date': '20150415',
|
||||
}
|
||||
},
|
||||
{
|
||||
'url': 'https://www.mlb.com/video/hargrove-homers-off-caldwell/c-1352023483?tid=67793694',
|
||||
'only_matching': True,
|
||||
|
|
@ -113,8 +176,92 @@ class MLBIE(NHLBaseIE):
|
|||
'url': 'http://mlb.mlb.com/shared/video/embed/m-internal-embed.html?content_id=75609783&property=mlb&autoplay=true&hashmode=false&siteSection=mlb/multimedia/article_118550098/article_embed&club=mlb',
|
||||
'only_matching': True,
|
||||
},
|
||||
{
|
||||
'url': 'https://www.mlb.com/cut4/carlos-gomez-borrowed-sunglasses-from-an-as-fan/c-278912842',
|
||||
'only_matching': True,
|
||||
}
|
||||
]
|
||||
_TIMESTAMP_KEY = 'date'
|
||||
|
||||
@staticmethod
|
||||
def _get_feed(video):
|
||||
return video
|
||||
|
||||
@staticmethod
|
||||
def _extract_mlb_subtitles(feed, language):
|
||||
subtitles = {}
|
||||
for keyword in (feed.get('keywordsAll') or []):
|
||||
keyword_type = keyword.get('type')
|
||||
if keyword_type and keyword_type.startswith('closed_captions_location_'):
|
||||
cc_location = keyword.get('value')
|
||||
if cc_location:
|
||||
subtitles.setdefault(language, []).append({
|
||||
'url': cc_location,
|
||||
})
|
||||
return subtitles
|
||||
|
||||
def _download_video_data(self, display_id):
|
||||
return self._download_json(
|
||||
'http://content.mlb.com/mlb/item/id/v1/%s/details/web-v1.json' % display_id,
|
||||
display_id)
|
||||
|
||||
|
||||
class MLBVideoIE(MLBBaseIE):
|
||||
_VALID_URL = r'https?://(?:www\.)?mlb\.com/(?:[^/]+/)*video/(?P<id>[^/?&#]+)'
|
||||
_TEST = {
|
||||
'url': 'https://www.mlb.com/mariners/video/ackley-s-spectacular-catch-c34698933',
|
||||
'md5': '632358dacfceec06bad823b83d21df2d',
|
||||
'info_dict': {
|
||||
'id': 'c04a8863-f569-42e6-9f87-992393657614',
|
||||
'ext': 'mp4',
|
||||
'title': "Ackley's spectacular catch",
|
||||
'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
|
||||
'duration': 66,
|
||||
'timestamp': 1405995000,
|
||||
'upload_date': '20140722',
|
||||
'thumbnail': r're:^https?://.+',
|
||||
},
|
||||
}
|
||||
_TIMESTAMP_KEY = 'timestamp'
|
||||
|
||||
@classmethod
|
||||
def suitable(cls, url):
|
||||
return False if MLBIE.suitable(url) else super(MLBVideoIE, cls).suitable(url)
|
||||
|
||||
@staticmethod
|
||||
def _get_feed(video):
|
||||
return video['feeds'][0]
|
||||
|
||||
@staticmethod
|
||||
def _extract_mlb_subtitles(feed, language):
|
||||
subtitles = {}
|
||||
for cc_location in (feed.get('closedCaptions') or []):
|
||||
subtitles.setdefault(language, []).append({
|
||||
'url': cc_location,
|
||||
})
|
||||
|
||||
def _download_video_data(self, display_id):
|
||||
# https://www.mlb.com/data-service/en/videos/[SLUG]
|
||||
return self._download_json(
|
||||
'https://fastball-gateway.mlb.com/graphql',
|
||||
display_id, query={
|
||||
'query': '''{
|
||||
mediaPlayback(ids: "%s") {
|
||||
description
|
||||
feeds(types: CMS) {
|
||||
closedCaptions
|
||||
duration
|
||||
image {
|
||||
cuts {
|
||||
width
|
||||
height
|
||||
src
|
||||
}
|
||||
}
|
||||
playbacks {
|
||||
name
|
||||
url
|
||||
}
|
||||
}
|
||||
id
|
||||
timestamp
|
||||
title
|
||||
}
|
||||
}''' % display_id,
|
||||
})['data']['mediaPlayback'][0]
|
||||
|
|
|
|||
|
|
@ -146,18 +146,19 @@ class SVTPlayIE(SVTPlayBaseIE):
|
|||
)
|
||||
(?P<svt_id>[^/?#&]+)|
|
||||
https?://(?:www\.)?(?:svtplay|oppetarkiv)\.se/(?:video|klipp|kanaler)/(?P<id>[^/?#&]+)
|
||||
(?:.*?modalId=(?P<modal_id>[\da-zA-Z-]+))?
|
||||
)
|
||||
'''
|
||||
_TESTS = [{
|
||||
'url': 'https://www.svtplay.se/video/26194546/det-har-ar-himlen',
|
||||
'url': 'https://www.svtplay.se/video/30479064',
|
||||
'md5': '2382036fd6f8c994856c323fe51c426e',
|
||||
'info_dict': {
|
||||
'id': 'jNwpV9P',
|
||||
'id': '8zVbDPA',
|
||||
'ext': 'mp4',
|
||||
'title': 'Det här är himlen',
|
||||
'timestamp': 1586044800,
|
||||
'upload_date': '20200405',
|
||||
'duration': 3515,
|
||||
'title': 'Designdrömmar i Stenungsund',
|
||||
'timestamp': 1615770000,
|
||||
'upload_date': '20210315',
|
||||
'duration': 3519,
|
||||
'thumbnail': r're:^https?://(?:.*[\.-]jpg|www.svtstatic.se/image/.*)$',
|
||||
'age_limit': 0,
|
||||
'subtitles': {
|
||||
|
|
@ -173,6 +174,9 @@ class SVTPlayIE(SVTPlayBaseIE):
|
|||
# AssertionError: Expected test_SVTPlay_jNwpV9P.mp4 to be at least 9.77KiB, but it's only 864.00B
|
||||
'skip_download': True,
|
||||
},
|
||||
}, {
|
||||
'url': 'https://www.svtplay.se/video/30479064/husdrommar/husdrommar-sasong-8-designdrommar-i-stenungsund?modalId=8zVbDPA',
|
||||
'only_matching': True,
|
||||
}, {
|
||||
# geo restricted to Sweden
|
||||
'url': 'http://www.oppetarkiv.se/video/5219710/trollflojten',
|
||||
|
|
@ -219,7 +223,8 @@ def _extract_by_video_id(self, video_id, webpage=None):
|
|||
|
||||
def _real_extract(self, url):
|
||||
mobj = re.match(self._VALID_URL, url)
|
||||
video_id, svt_id = mobj.group('id', 'svt_id')
|
||||
video_id = mobj.group('id')
|
||||
svt_id = mobj.group('svt_id') or mobj.group('modal_id')
|
||||
|
||||
if svt_id:
|
||||
return self._extract_by_video_id(svt_id)
|
||||
|
|
@ -254,6 +259,7 @@ def _real_extract(self, url):
|
|||
if not svt_id:
|
||||
svt_id = self._search_regex(
|
||||
(r'<video[^>]+data-video-id=["\']([\da-zA-Z-]+)',
|
||||
r'<[^>]+\bdata-rt=["\']top-area-play-button["\'][^>]+\bhref=["\'][^"\']*video/%s/[^"\']*\bmodalId=([\da-zA-Z-]+)' % re.escape(video_id),
|
||||
r'["\']videoSvtId["\']\s*:\s*["\']([\da-zA-Z-]+)',
|
||||
r'["\']videoSvtId\\?["\']\s*:\s*\\?["\']([\da-zA-Z-]+)',
|
||||
r'"content"\s*:\s*{.*?"id"\s*:\s*"([\da-zA-Z-]+)"',
|
||||
|
|
|
|||
|
|
@ -991,11 +991,15 @@ def _fetch_page(self, album_id, authorization, hashed_pass, page):
|
|||
}
|
||||
if hashed_pass:
|
||||
query['_hashed_pass'] = hashed_pass
|
||||
videos = self._download_json(
|
||||
'https://api.vimeo.com/albums/%s/videos' % album_id,
|
||||
album_id, 'Downloading page %d' % api_page, query=query, headers={
|
||||
'Authorization': 'jwt ' + authorization,
|
||||
})['data']
|
||||
try:
|
||||
videos = self._download_json(
|
||||
'https://api.vimeo.com/albums/%s/videos' % album_id,
|
||||
album_id, 'Downloading page %d' % api_page, query=query, headers={
|
||||
'Authorization': 'jwt ' + authorization,
|
||||
})['data']
|
||||
except ExtractorError as e:
|
||||
if isinstance(e.cause, compat_HTTPError) and e.cause.code == 400:
|
||||
return
|
||||
for video in videos:
|
||||
link = video.get('link')
|
||||
if not link:
|
||||
|
|
|
|||
|
|
@ -182,17 +182,20 @@ def metadata_from_url(r_url):
|
|||
if not embed_code:
|
||||
continue
|
||||
embed_code = ds(embed_code)
|
||||
if video_type in ('video/rcs', 'video/kenc'):
|
||||
if video_type == 'video/kenc':
|
||||
kenc = self._download_json(
|
||||
'https://www.vvvvid.it/kenc', video_id, query={
|
||||
'action': 'kt',
|
||||
'conn_id': self._conn_id,
|
||||
'url': embed_code,
|
||||
}, fatal=False) or {}
|
||||
kenc_message = kenc.get('message')
|
||||
if kenc_message:
|
||||
embed_code += '?' + ds(kenc_message)
|
||||
if video_type == 'video/kenc':
|
||||
embed_code = re.sub(r'https?(://[^/]+)/z/', r'https\1/i/', embed_code).replace('/manifest.f4m', '/master.m3u8')
|
||||
kenc = self._download_json(
|
||||
'https://www.vvvvid.it/kenc', video_id, query={
|
||||
'action': 'kt',
|
||||
'conn_id': self._conn_id,
|
||||
'url': embed_code,
|
||||
}, fatal=False) or {}
|
||||
kenc_message = kenc.get('message')
|
||||
if kenc_message:
|
||||
embed_code += '?' + ds(kenc_message)
|
||||
formats.extend(self._extract_m3u8_formats(
|
||||
embed_code, video_id, 'mp4', m3u8_id='hls', fatal=False))
|
||||
elif video_type == 'video/rcs':
|
||||
formats.extend(self._extract_akamai_formats(embed_code, video_id))
|
||||
elif video_type == 'video/youtube':
|
||||
info.update({
|
||||
|
|
|
|||
Loading…
Reference in a new issue