mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-10-01 06:41:27 +02:00
Compare commits
2 Commits
21f16f6db1
...
5ae98da352
Author | SHA1 | Date | |
---|---|---|---|
|
5ae98da352 | ||
|
76c36cc2dd |
|
@ -9,7 +9,7 @@ from ..utils.traversal import traverse_obj
|
||||||
|
|
||||||
|
|
||||||
class CloudyCDNIE(InfoExtractor):
|
class CloudyCDNIE(InfoExtractor):
|
||||||
_VALID_URL = r'(?:https?:)?//embed\.cloudycdn\.services/(?P<site_id>[^/]+)/media/(?P<id>[\w-]+)'
|
_VALID_URL = r'(?:https?:)?//embed\.cloudycdn\.services/(?P<site_id>[^/?#]+)/media/(?P<id>[\w-]+)'
|
||||||
_EMBED_REGEX = [rf'<iframe[^>]+\bsrc=[\'"](?P<url>{_VALID_URL})']
|
_EMBED_REGEX = [rf'<iframe[^>]+\bsrc=[\'"](?P<url>{_VALID_URL})']
|
||||||
_TESTS = [{
|
_TESTS = [{
|
||||||
'url': 'https://embed.cloudycdn.services/ltv/media/46k_d23-6000-105?',
|
'url': 'https://embed.cloudycdn.services/ltv/media/46k_d23-6000-105?',
|
||||||
|
@ -53,17 +53,16 @@ class CloudyCDNIE(InfoExtractor):
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
site_id, video_id = self._match_valid_url(url).group('site_id', 'id')
|
site_id, video_id = self._match_valid_url(url).group('site_id', 'id')
|
||||||
|
|
||||||
json = self._download_json(
|
data = self._download_json(
|
||||||
f'https://player.cloudycdn.services/player/{site_id}/media/{video_id}/',
|
f'https://player.cloudycdn.services/player/{site_id}/media/{video_id}/',
|
||||||
video_id, data=urlencode_postdata({
|
video_id, data=urlencode_postdata({
|
||||||
'version': '6.4.0',
|
'version': '6.4.0',
|
||||||
'referer': url,
|
'referer': url,
|
||||||
}))
|
}))
|
||||||
|
|
||||||
formats = []
|
formats, subtitles = [], {}
|
||||||
subtitles = {}
|
for m3u8_url in traverse_obj(data, ('source', 'sources', ..., 'src', {url_or_none})):
|
||||||
for url in traverse_obj(json, ('source', 'sources', ..., 'src', {url_or_none})):
|
fmts, subs = self._extract_m3u8_formats_and_subtitles(m3u8_url, video_id, fatal=False)
|
||||||
fmts, subs = self._extract_m3u8_formats_and_subtitles(url, video_id)
|
|
||||||
formats.extend(fmts)
|
formats.extend(fmts)
|
||||||
self._merge_subtitles(subs, target=subtitles)
|
self._merge_subtitles(subs, target=subtitles)
|
||||||
|
|
||||||
|
@ -71,8 +70,8 @@ class CloudyCDNIE(InfoExtractor):
|
||||||
'id': video_id,
|
'id': video_id,
|
||||||
'formats': formats,
|
'formats': formats,
|
||||||
'subtitles': subtitles,
|
'subtitles': subtitles,
|
||||||
**traverse_obj(json, {
|
**traverse_obj(data, {
|
||||||
'title': 'name',
|
'title': ('name', {str}),
|
||||||
'duration': ('duration', {int_or_none}),
|
'duration': ('duration', {int_or_none}),
|
||||||
'timestamp': ('upload_date', {parse_iso8601}),
|
'timestamp': ('upload_date', {parse_iso8601}),
|
||||||
'thumbnail': ('source', 'poster', {url_or_none}),
|
'thumbnail': ('source', 'poster', {url_or_none}),
|
||||||
|
|
|
@ -4,10 +4,12 @@ import urllib.parse
|
||||||
from .common import InfoExtractor
|
from .common import InfoExtractor
|
||||||
from ..utils import (
|
from ..utils import (
|
||||||
ExtractorError,
|
ExtractorError,
|
||||||
|
determine_ext,
|
||||||
int_or_none,
|
int_or_none,
|
||||||
js_to_json,
|
js_to_json,
|
||||||
parse_iso8601,
|
parse_iso8601,
|
||||||
parse_qs,
|
parse_qs,
|
||||||
|
str_or_none,
|
||||||
url_or_none,
|
url_or_none,
|
||||||
urljoin,
|
urljoin,
|
||||||
)
|
)
|
||||||
|
@ -15,7 +17,13 @@ from ..utils.traversal import traverse_obj
|
||||||
|
|
||||||
|
|
||||||
class LSMLREmbedIE(InfoExtractor):
|
class LSMLREmbedIE(InfoExtractor):
|
||||||
_VALID_URL = r'https?://(?:(?:latvijasradio|lr1|lr2|klasika|lr4|naba|radioteatris)\.lsm|pieci)\.lv/[^/]+/(?:pleijeris|embed)'
|
_VALID_URL = r'''(?x)
|
||||||
|
https?://(?:
|
||||||
|
(?:latvijasradio|lr1|lr2|klasika|lr4|naba|radioteatris)\.lsm|
|
||||||
|
pieci
|
||||||
|
)\.lv/[^/?#]+/(?:
|
||||||
|
pleijeris|embed
|
||||||
|
)/?\?(?:[^#]+&)?(?:show|id)=(?P<id>\d+)'''
|
||||||
_TESTS = [{
|
_TESTS = [{
|
||||||
'url': 'https://latvijasradio.lsm.lv/lv/embed/?theme=black&size=16x9&showCaptions=0&id=183522',
|
'url': 'https://latvijasradio.lsm.lv/lv/embed/?theme=black&size=16x9&showCaptions=0&id=183522',
|
||||||
'md5': '719b33875cd1429846eeeaeec6df2830',
|
'md5': '719b33875cd1429846eeeaeec6df2830',
|
||||||
|
@ -29,7 +37,7 @@ class LSMLREmbedIE(InfoExtractor):
|
||||||
}, {
|
}, {
|
||||||
'url': 'https://radioteatris.lsm.lv/lv/embed/?id=&show=1270&theme=white&size=16x9',
|
'url': 'https://radioteatris.lsm.lv/lv/embed/?id=&show=1270&theme=white&size=16x9',
|
||||||
'info_dict': {
|
'info_dict': {
|
||||||
'id': 1270,
|
'id': '1270',
|
||||||
},
|
},
|
||||||
'playlist_count': 3,
|
'playlist_count': 3,
|
||||||
'playlist': [{
|
'playlist': [{
|
||||||
|
@ -55,7 +63,7 @@ class LSMLREmbedIE(InfoExtractor):
|
||||||
}, {
|
}, {
|
||||||
'url': 'https://lr1.lsm.lv/lv/embed/?id=166557&show=0&theme=white&size=16x9',
|
'url': 'https://lr1.lsm.lv/lv/embed/?id=166557&show=0&theme=white&size=16x9',
|
||||||
'info_dict': {
|
'info_dict': {
|
||||||
'id': 166557,
|
'id': '166557',
|
||||||
},
|
},
|
||||||
'playlist_count': 2,
|
'playlist_count': 2,
|
||||||
'playlist': [{
|
'playlist': [{
|
||||||
|
@ -105,40 +113,38 @@ class LSMLREmbedIE(InfoExtractor):
|
||||||
|
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
query = parse_qs(url)
|
query = parse_qs(url)
|
||||||
video_id = traverse_obj(query, ('show', 0, {int_or_none}))
|
video_id = traverse_obj(query, (
|
||||||
|
('show', 'id'), 0, {int_or_none}, {lambda x: x or None}, {str_or_none}), get_all=False)
|
||||||
if not video_id:
|
|
||||||
video_id = traverse_obj(query, ('id', 0, {int_or_none}))
|
|
||||||
|
|
||||||
webpage = self._download_webpage(url, video_id)
|
webpage = self._download_webpage(url, video_id)
|
||||||
|
|
||||||
player_data, media_data = self._html_search_regex(
|
player_data, media_data = self._search_regex(
|
||||||
r'LR\.audio\.Player\s*\([^{]*(?P<player>\{.*?\}),(?P<media>\{.*\})\);',
|
r'LR\.audio\.Player\s*\([^{]*(?P<player>\{.*?\}),(?P<media>\{.*\})\);',
|
||||||
webpage, 'player json', group=('player', 'media'))
|
webpage, 'player json', group=('player', 'media'))
|
||||||
|
|
||||||
player_json = self._parse_json(player_data, video_id, js_to_json)
|
player_json = self._parse_json(
|
||||||
media_json = self._parse_json(media_data, video_id, js_to_json)
|
player_data, video_id, transform_source=js_to_json, fatal=False) or {}
|
||||||
|
media_json = self._parse_json(media_data, video_id, transform_source=js_to_json)
|
||||||
|
|
||||||
entries = []
|
entries = []
|
||||||
for item in traverse_obj(media_json, (('audio', 'video'), ...)):
|
for item in traverse_obj(media_json, (('audio', 'video'), lambda _, v: v['id'])):
|
||||||
formats = []
|
formats = []
|
||||||
for source_url in traverse_obj(item, ('sources', ..., 'file', {url_or_none})):
|
for source_url in traverse_obj(item, ('sources', ..., 'file', {url_or_none})):
|
||||||
if source_url.endswith('.m3u8'):
|
if determine_ext(source_url) == 'm3u8':
|
||||||
formats.extend(self._extract_m3u8_formats(source_url, video_id))
|
formats.extend(self._extract_m3u8_formats(source_url, video_id, fatal=False))
|
||||||
else:
|
else:
|
||||||
formats.append({'url': source_url})
|
formats.append({'url': source_url})
|
||||||
|
|
||||||
id = item.get('id')
|
id_ = item['id']
|
||||||
title = item.get('title')
|
title = item.get('title')
|
||||||
if id and id.startswith('v') and not title:
|
if id_.startswith('v') and not title:
|
||||||
title = traverse_obj(
|
title = traverse_obj(
|
||||||
media_json, ('audio', lambda _, v: v['id'][1:] == id[1:], 'title',
|
media_json, ('audio', lambda _, v: v['id'][1:] == id_[1:], 'title',
|
||||||
{lambda x: x and f'{x} - Video Version'}), get_all=False)
|
{lambda x: x and f'{x} - Video Version'}), get_all=False)
|
||||||
|
|
||||||
entries.append({
|
entries.append({
|
||||||
'formats': formats,
|
'formats': formats,
|
||||||
'thumbnail': urljoin(url, player_json.get('poster')),
|
'thumbnail': urljoin(url, player_json.get('poster')),
|
||||||
'id': id,
|
'id': id_,
|
||||||
'title': title,
|
'title': title,
|
||||||
'duration': traverse_obj(item, ('duration', {int_or_none})),
|
'duration': traverse_obj(item, ('duration', {int_or_none})),
|
||||||
})
|
})
|
||||||
|
@ -150,7 +156,7 @@ class LSMLREmbedIE(InfoExtractor):
|
||||||
|
|
||||||
|
|
||||||
class LSMLTVEmbedIE(InfoExtractor):
|
class LSMLTVEmbedIE(InfoExtractor):
|
||||||
_VALID_URL = r'https?://ltv\.lsm\.lv/embed.*[?&]c=(?P<id>[^&]+)'
|
_VALID_URL = r'https?://ltv\.lsm\.lv/embed\?(?:[^#]+&)?c=(?P<id>[^#&]+)'
|
||||||
_TESTS = [{
|
_TESTS = [{
|
||||||
'url': 'https://ltv.lsm.lv/embed?c=eyJpdiI6IjQzbHVUeHAyaDJiamFjcjdSUUFKdnc9PSIsInZhbHVlIjoiMHl3SnJNRmd2TmFIdnZwOGtGUUpzODFzUEZ4SVVsN2xoRjliSW9vckUyMWZIWG8vbWVzaFFkY0lhNmRjbjRpaCIsIm1hYyI6ImMzNjdhMzFhNTFhZmY1ZmE0NWI5YmFjZGI1YmJiNGEyNjgzNDM4MjUzMWEwM2FmMDMyZDMwYWM1MDFjZmM5MGIiLCJ0YWciOiIifQ==',
|
'url': 'https://ltv.lsm.lv/embed?c=eyJpdiI6IjQzbHVUeHAyaDJiamFjcjdSUUFKdnc9PSIsInZhbHVlIjoiMHl3SnJNRmd2TmFIdnZwOGtGUUpzODFzUEZ4SVVsN2xoRjliSW9vckUyMWZIWG8vbWVzaFFkY0lhNmRjbjRpaCIsIm1hYyI6ImMzNjdhMzFhNTFhZmY1ZmE0NWI5YmFjZGI1YmJiNGEyNjgzNDM4MjUzMWEwM2FmMDMyZDMwYWM1MDFjZmM5MGIiLCJ0YWciOiIifQ==',
|
||||||
'md5': '64f72a360ca530d5ed89c77646c9eee5',
|
'md5': '64f72a360ca530d5ed89c77646c9eee5',
|
||||||
|
@ -198,22 +204,21 @@ class LSMLTVEmbedIE(InfoExtractor):
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
video_id = urllib.parse.unquote(self._match_id(url))
|
video_id = urllib.parse.unquote(self._match_id(url))
|
||||||
webpage = self._download_webpage(url, video_id)
|
webpage = self._download_webpage(url, video_id)
|
||||||
|
data = self._search_json(
|
||||||
json = self._search_json(r'window\.ltvEmbedPayload\s*=', webpage, 'embed json', video_id)
|
r'window\.ltvEmbedPayload\s*=', webpage, 'embed json', video_id)
|
||||||
|
embed_type = traverse_obj(data, ('source', 'name', {str}))
|
||||||
embed_type = traverse_obj(json, ('source', 'name'))
|
|
||||||
|
|
||||||
if embed_type == 'telia':
|
if embed_type == 'telia':
|
||||||
ie_key = 'CloudyCDN'
|
ie_key = 'CloudyCDN'
|
||||||
url = traverse_obj(json, ('source', 'embed_url', {url_or_none}))
|
embed_url = traverse_obj(data, ('source', 'embed_url', {url_or_none}))
|
||||||
elif embed_type == 'youtube':
|
elif embed_type == 'youtube':
|
||||||
ie_key = 'Youtube'
|
ie_key = 'Youtube'
|
||||||
url = traverse_obj(json, ('source', 'id'))
|
embed_url = traverse_obj(data, ('source', 'id', {str}))
|
||||||
else:
|
else:
|
||||||
raise ExtractorError('Unsupported embed type')
|
raise ExtractorError(f'Unsupported embed type {embed_type!r}')
|
||||||
|
|
||||||
return self.url_result(
|
return self.url_result(
|
||||||
url, ie_key, video_id, **traverse_obj(json, {
|
embed_url, ie_key, video_id, **traverse_obj(data, {
|
||||||
'title': ('parentInfo', 'title'),
|
'title': ('parentInfo', 'title'),
|
||||||
'duration': ('parentInfo', 'duration', {int_or_none}),
|
'duration': ('parentInfo', 'duration', {int_or_none}),
|
||||||
'thumbnail': ('source', 'poster', {url_or_none}),
|
'thumbnail': ('source', 'poster', {url_or_none}),
|
||||||
|
@ -221,7 +226,7 @@ class LSMLTVEmbedIE(InfoExtractor):
|
||||||
|
|
||||||
|
|
||||||
class LSMReplayIE(InfoExtractor):
|
class LSMReplayIE(InfoExtractor):
|
||||||
_VALID_URL = r'https?://replay\.lsm\.lv/.*/(?:ieraksts|statja)/[^/]+/(?P<id>\d+)'
|
_VALID_URL = r'https?://replay\.lsm\.lv/[^/?#]+/(?:ieraksts|statja)/[^/?#]+/(?P<id>\d+)'
|
||||||
_TESTS = [{
|
_TESTS = [{
|
||||||
'url': 'https://replay.lsm.lv/lv/ieraksts/ltv/311130/4-studija-zolitudes-tragedija-un-incupes-stacija',
|
'url': 'https://replay.lsm.lv/lv/ieraksts/ltv/311130/4-studija-zolitudes-tragedija-un-incupes-stacija',
|
||||||
'md5': '64f72a360ca530d5ed89c77646c9eee5',
|
'md5': '64f72a360ca530d5ed89c77646c9eee5',
|
||||||
|
@ -253,19 +258,20 @@ class LSMReplayIE(InfoExtractor):
|
||||||
'only_matching': True,
|
'only_matching': True,
|
||||||
}]
|
}]
|
||||||
|
|
||||||
def fix_nuxt_data(self, webpage):
|
def _fix_nuxt_data(self, webpage):
|
||||||
return re.sub(r'Object\.create\(null(?:,(\{.+\}))?\)', lambda m: m.group(1) or 'null', webpage)
|
return re.sub(r'Object\.create\(null(?:,(\{.+\}))?\)', lambda m: m.group(1) or 'null', webpage)
|
||||||
|
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
video_id = self._match_id(url)
|
video_id = self._match_id(url)
|
||||||
webpage = self._download_webpage(url, video_id)
|
webpage = self._download_webpage(url, video_id)
|
||||||
|
|
||||||
json = self._search_nuxt_data(self.fix_nuxt_data(webpage), video_id, context_name='__REPLAY__')
|
data = self._search_nuxt_data(
|
||||||
|
self._fix_nuxt_data(webpage), video_id, context_name='__REPLAY__')
|
||||||
|
|
||||||
return {
|
return {
|
||||||
'_type': 'url_transparent',
|
'_type': 'url_transparent',
|
||||||
'id': video_id,
|
'id': video_id,
|
||||||
**traverse_obj(json, {
|
**traverse_obj(data, {
|
||||||
'url': ('playback', 'service', 'url', {url_or_none}),
|
'url': ('playback', 'service', 'url', {url_or_none}),
|
||||||
'title': ('mediaItem', 'title'),
|
'title': ('mediaItem', 'title'),
|
||||||
'description': ('mediaItem', ('lead', 'body')),
|
'description': ('mediaItem', ('lead', 'body')),
|
||||||
|
|
Loading…
Reference in New Issue
Block a user