mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-10-04 08:11:25 +02:00
Compare commits
3 Commits
13f116fce6
...
5b148c0f79
Author | SHA1 | Date | |
---|---|---|---|
|
5b148c0f79 | ||
|
85b739eba7 | ||
|
a634e7c5d1 |
|
@ -8,12 +8,16 @@ from .youtube import YoutubeIE
|
||||||
from ..utils import (
|
from ..utils import (
|
||||||
OnDemandPagedList,
|
OnDemandPagedList,
|
||||||
clean_html,
|
clean_html,
|
||||||
|
dict_get,
|
||||||
extract_attributes,
|
extract_attributes,
|
||||||
ExtractorError,
|
ExtractorError,
|
||||||
get_element_by_class,
|
get_element_by_class,
|
||||||
get_elements_html_by_class,
|
get_elements_html_by_class,
|
||||||
int_or_none,
|
int_or_none,
|
||||||
|
js_to_json,
|
||||||
|
merge_dicts,
|
||||||
parse_duration,
|
parse_duration,
|
||||||
|
str_or_none,
|
||||||
traverse_obj,
|
traverse_obj,
|
||||||
urlencode_postdata,
|
urlencode_postdata,
|
||||||
url_or_none,
|
url_or_none,
|
||||||
|
@ -74,8 +78,7 @@ class PromoDJBaseIE(InfoExtractor):
|
||||||
def _fetch_page(self, url, allowed_media_cats, playlist_id, page):
|
def _fetch_page(self, url, allowed_media_cats, playlist_id, page):
|
||||||
page_url = self._set_url_page(url, page + 1)
|
page_url = self._set_url_page(url, page + 1)
|
||||||
html = self._download_webpage(page_url, f'{playlist_id}-page-{page + 1}')
|
html = self._download_webpage(page_url, f'{playlist_id}-page-{page + 1}')
|
||||||
current_page = int(clean_html(get_element_by_class('NavigatorCurrentPage', html)) or '1')
|
if self._get_current_page(html) != page + 1:
|
||||||
if current_page != page + 1:
|
|
||||||
return
|
return
|
||||||
|
|
||||||
for a in get_elements_html_by_class('player_standard_tool__play', html):
|
for a in get_elements_html_by_class('player_standard_tool__play', html):
|
||||||
|
@ -115,16 +118,59 @@ class PromoDJBaseIE(InfoExtractor):
|
||||||
is_default_playlist = '/groups/' not in url
|
is_default_playlist = '/groups/' not in url
|
||||||
return 30 if is_default_playlist else 20
|
return 30 if is_default_playlist else 20
|
||||||
|
|
||||||
|
def _get_current_page(self, html):
|
||||||
|
return int(clean_html(get_element_by_class('NavigatorCurrentPage', html)) or '1')
|
||||||
|
|
||||||
def _fetch_media_data(self, ids, video_id):
|
def _fetch_media_data(self, ids, video_id):
|
||||||
data = {}
|
data = {}
|
||||||
for i, id in enumerate(ids):
|
for i, id in enumerate(ids):
|
||||||
data[f'multi[{i}][method]'] = 'players/config'
|
data[f'multi[{i}][method]'] = 'players/config'
|
||||||
data[f'multi[{i}][params][kind]'] = 'standalone.big'
|
data[f'multi[{i}][params][kind]'] = 'cover.big'
|
||||||
data[f'multi[{i}][params][fileID]'] = id
|
data[f'multi[{i}][params][fileID]'] = id
|
||||||
return self._download_json(
|
return self._download_json(
|
||||||
'https://promodj.com/api/multi.json', video_id, data=urlencode_postdata(data),
|
'https://promodj.com/api/multi.json', video_id, data=urlencode_postdata(data),
|
||||||
headers={'Content-Type': 'application/x-www-form-urlencoded'})
|
headers={'Content-Type': 'application/x-www-form-urlencoded'})
|
||||||
|
|
||||||
|
def _parse_media_data(self, media_data, id):
|
||||||
|
if player_error := media_data.get('player_error'):
|
||||||
|
raise ExtractorError(player_error, expected=True)
|
||||||
|
|
||||||
|
if media_data.get('video'):
|
||||||
|
video = traverse_obj(
|
||||||
|
self._parse_json(media_data['config'], id), ('playlist', 'item', 0))
|
||||||
|
formats = [{
|
||||||
|
'url': traverse_obj(video, ('play', '@url')).replace('?returnurl=1', ''),
|
||||||
|
**traverse_obj(media_data, {
|
||||||
|
'width': ('width', {int_or_none}),
|
||||||
|
'height': ('height', {int_or_none}),
|
||||||
|
})
|
||||||
|
}]
|
||||||
|
return {
|
||||||
|
'id': id,
|
||||||
|
'formats': formats,
|
||||||
|
**traverse_obj(video, {
|
||||||
|
'title': ('title', 'line', 1, 0, '$', {str_or_none}),
|
||||||
|
'webpage_url': ('title', '@ico_url', {url_or_none}),
|
||||||
|
'duration': ('play', '@duration', {int_or_none}),
|
||||||
|
'thumbnail': ('background', '@url', {url_or_none}),
|
||||||
|
'channel': ('title', 'line', 0, 0, '$', {str_or_none}),
|
||||||
|
'channel_url': ('title', 'line', 0, 0, '@url', {url_or_none}),
|
||||||
|
})
|
||||||
|
}
|
||||||
|
|
||||||
|
formats = [traverse_obj(source, {
|
||||||
|
'url': ('URL', {url_or_none}),
|
||||||
|
'size': ('size', {int_or_none}),
|
||||||
|
}) for source in traverse_obj(media_data, ('sources'))]
|
||||||
|
thumbnails = [{'url': url} for url in traverse_obj(media_data, ('coverURL', ('600', '1200', '2000'))) if url_or_none]
|
||||||
|
return {
|
||||||
|
'id': id,
|
||||||
|
'title': clean_html(dict_get(media_data, ('title_html', 'title'))),
|
||||||
|
'formats': formats,
|
||||||
|
'thumbnails': thumbnails,
|
||||||
|
'webpage_url': traverse_obj(media_data, ('titleURL', {url_or_none}))
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
class PromoDJPageIE(PromoDJBaseIE):
|
class PromoDJPageIE(PromoDJBaseIE):
|
||||||
_PAGES_RE = '|'.join(PromoDJBaseIE._PAGES)
|
_PAGES_RE = '|'.join(PromoDJBaseIE._PAGES)
|
||||||
|
@ -232,11 +278,10 @@ class PromoDJUserPagesIE(PromoDJBaseIE):
|
||||||
for page_url, page_title in re.findall(r'<a href=\"([^\"]+)\">([^<]+)</a>', content_html):
|
for page_url, page_title in re.findall(r'<a href=\"([^\"]+)\">([^<]+)</a>', content_html):
|
||||||
yield self.url_result(page_url, PromoDJUserPageIE, video_title=page_title)
|
yield self.url_result(page_url, PromoDJUserPageIE, video_title=page_title)
|
||||||
|
|
||||||
def _fetch_blog_page(self, url, playlist_id, page):
|
def _fetch_blogs_page(self, url, playlist_id, page):
|
||||||
page_url = self._set_url_page(url, page + 1)
|
page_url = self._set_url_page(url, page + 1)
|
||||||
html = self._download_webpage(page_url, f'{playlist_id}-page-{page + 1}')
|
html = self._download_webpage(page_url, f'{playlist_id}-page-{page + 1}')
|
||||||
current_page = int(clean_html(get_element_by_class('NavigatorCurrentPage', html)) or '1')
|
if self._get_current_page(html) != page + 1:
|
||||||
if current_page != page + 1:
|
|
||||||
return
|
return
|
||||||
|
|
||||||
for a in get_elements_html_by_class('post_title_moderated', html):
|
for a in get_elements_html_by_class('post_title_moderated', html):
|
||||||
|
@ -250,7 +295,7 @@ class PromoDJUserPagesIE(PromoDJBaseIE):
|
||||||
entries = self._parse_pages(url, playlist_id)
|
entries = self._parse_pages(url, playlist_id)
|
||||||
elif type == 'blog':
|
elif type == 'blog':
|
||||||
entries = OnDemandPagedList(
|
entries = OnDemandPagedList(
|
||||||
functools.partial(self._fetch_blog_page, url, playlist_id),
|
functools.partial(self._fetch_blogs_page, url, playlist_id),
|
||||||
self._PAGE_SIZE)
|
self._PAGE_SIZE)
|
||||||
return self.playlist_result(entries, playlist_id)
|
return self.playlist_result(entries, playlist_id)
|
||||||
|
|
||||||
|
@ -372,57 +417,222 @@ class PromoDJIE(PromoDJBaseIE):
|
||||||
_VALID_URL = rf'{PromoDJBaseIE._BASE_URL_RE}/{PromoDJBaseIE._LOGIN_RE}/(?P<type>{PromoDJBaseIE._MEDIA_TYPES_RE})/(?P<id>\d+)(?:/\w+)?',
|
_VALID_URL = rf'{PromoDJBaseIE._BASE_URL_RE}/{PromoDJBaseIE._LOGIN_RE}/(?P<type>{PromoDJBaseIE._MEDIA_TYPES_RE})/(?P<id>\d+)(?:/\w+)?',
|
||||||
_TESTS = [{
|
_TESTS = [{
|
||||||
'url': 'https://promodj.com/antonpavlovsky/remixes/6259208/David_Usher_Black_Black_Heart_Anton_Pavlovsky_Cover',
|
'url': 'https://promodj.com/antonpavlovsky/remixes/6259208/David_Usher_Black_Black_Heart_Anton_Pavlovsky_Cover',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '6259208',
|
||||||
|
'ext': 'mp3',
|
||||||
|
'title': 'David Usher - Black Black Heart (Anton Pavlovsky Cover)',
|
||||||
|
'tags': ['Lounge', 'Deep House'],
|
||||||
|
'upload_date': '20170323',
|
||||||
|
'timestamp': 1490258400.0,
|
||||||
|
'duration': 173.0,
|
||||||
|
'size': 7654604,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
'url': 'https://promodj.com/j-factory/samples/7560171/Amedici_BW1_Intro',
|
'url': 'https://promodj.com/j-factory/samples/7560171/Amedici_BW1_Intro',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '7560171',
|
||||||
|
'ext': 'mp3',
|
||||||
|
'title': 'Amedici - BW1 - Intro',
|
||||||
|
'tags': ['Multitrack master', 'Fx'],
|
||||||
|
'upload_date': '20240212',
|
||||||
|
'timestamp': 1707748800.0,
|
||||||
|
'duration': 21.0,
|
||||||
|
'size': 838041,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# music: no download links in html
|
# music: no download links in html
|
||||||
'url': 'https://promodj.com/gluk/tracks/4713922/DJ_Glyuk_Folk_ing_DJ_Steven_Smile_Remix_2005',
|
'url': 'https://promodj.com/gluk/tracks/4713922/DJ_Glyuk_Folk_ing_DJ_Steven_Smile_Remix_2005',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '4713922',
|
||||||
|
'ext': 'mp3',
|
||||||
|
'title': 'DJ Глюк - Folk\'ing [DJ Steven Smile Remix] (2005)',
|
||||||
|
'tags': ['Pumping House', 'Hard House'],
|
||||||
|
'upload_date': '20140404',
|
||||||
|
'timestamp': 1396605480.0,
|
||||||
|
'duration': 299.0,
|
||||||
|
'size': 12058624,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# video: no download link in html
|
# video: no download link in html
|
||||||
'url': 'https://promodj.com/psywanderer/videos/7559147/Chu_de_sa',
|
'url': 'https://promodj.com/psywanderer/videos/7559147/Chu_de_sa',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '7559147',
|
||||||
|
'ext': 'mp4',
|
||||||
|
'title': 'Чу де са',
|
||||||
|
'tags': ['Jazz-Rap', 'Jazzstep'],
|
||||||
|
'thumbnail': r're:^https?://',
|
||||||
|
'upload_date': '20240210',
|
||||||
|
'timestamp': 1707533820.0,
|
||||||
|
'duration': 388720,
|
||||||
|
'view_count': int,
|
||||||
|
'channel': 'PsyWanderer',
|
||||||
|
'channel_url': 'https://promodj.com/psywanderer',
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# no player
|
# no player (external link)
|
||||||
'url': 'https://promodj.com/gluk/tracks/420310/IMpulse_Zakat',
|
'url': 'https://promodj.com/gluk/tracks/420310/IMpulse_Zakat',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '420310',
|
||||||
|
'ext': 'mp3',
|
||||||
|
'title': 'IMpulse - Закат',
|
||||||
|
'tags': ['House', 'Electro House'],
|
||||||
|
'thumbnail': r're:^https?://',
|
||||||
|
'upload_date': '20081024',
|
||||||
|
'timestamp': 1224846120.0,
|
||||||
|
'duration': 133.0,
|
||||||
|
'size': 1048576,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
|
'params': {
|
||||||
|
'skip_download': 'Link is broken',
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# without slug
|
# without slug
|
||||||
'url': 'https://promodj.com/djlykov/tracks/7551590',
|
'url': 'https://promodj.com/djlykov/tracks/7551590',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '7551590',
|
||||||
|
'ext': 'mp3',
|
||||||
|
'title': 'Lykov - Benjamin (Radio Edit) [MOUSE-P]',
|
||||||
|
'tags': ['Dance Pop', 'Eurodance'],
|
||||||
|
'upload_date': '20240122',
|
||||||
|
'timestamp': 1705919280.0,
|
||||||
|
'duration': 233.0,
|
||||||
|
'size': 9332326,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# lossless
|
# lossless wav
|
||||||
'url': 'https://promodj.com/modi-glu/tracks/6081339/Modi_Glyu_Anabel',
|
'url': 'https://promodj.com/modi-glu/tracks/6081339/Modi_Glyu_Anabel',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '6081339',
|
||||||
|
'ext': 'wav',
|
||||||
|
'title': 'Моди Глю " Анабель"',
|
||||||
|
'tags': ['Chillout', 'Downtempo'],
|
||||||
|
'upload_date': '20161029',
|
||||||
|
'timestamp': 1477767780.0,
|
||||||
|
'duration': 236.0,
|
||||||
|
'size': 42257612,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# paid audio
|
# lossless flac
|
||||||
|
'url': 'https://promodj.com/sashaorbeat/mixes/7422493/Sasha_Orbeat_Pure_Love_3',
|
||||||
|
'info_dict': {
|
||||||
|
'id': '7422493',
|
||||||
|
'ext': 'flac',
|
||||||
|
'title': 'Sasha Orbeat — Pure Love 3',
|
||||||
|
'tags': ['Lo-Fi', 'Downtempo'],
|
||||||
|
'upload_date': '20230213',
|
||||||
|
'timestamp': 1676306160.0,
|
||||||
|
'duration': 3631.0,
|
||||||
|
'size': 685139558,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
|
}, {
|
||||||
|
# paid lossless
|
||||||
'url': 'https://promodj.com/boyko/tracks/1435682/Dj_Boyko_Katy_Queen_Nad_Oblakami',
|
'url': 'https://promodj.com/boyko/tracks/1435682/Dj_Boyko_Katy_Queen_Nad_Oblakami',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '1435682',
|
||||||
|
'ext': 'mp3',
|
||||||
|
'title': 'Dj Boyko & Katy Queen - Над Облаками',
|
||||||
|
'tags': ['House', 'Trance'],
|
||||||
|
'upload_date': '20100404',
|
||||||
|
'timestamp': 1270376700.0,
|
||||||
|
'duration': 321.0,
|
||||||
|
'size': 56623104,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
'url': 'https://promodj.com/sergeyfedotov306/videos/7457627/V_Matrice_Sboy',
|
# paid lossy
|
||||||
'only_matching': True,
|
'url': 'https://promodj.com/tesla/tracks/342938/Library_Of_Bugs',
|
||||||
|
'info_dict': {
|
||||||
|
'id': '342938',
|
||||||
|
'ext': 'mp3',
|
||||||
|
'title': 'Library Of Bugs',
|
||||||
|
'tags': ['Minimal Techno', 'Tech House'],
|
||||||
|
'upload_date': '20080827',
|
||||||
|
'timestamp': 1219841220.0,
|
||||||
|
'duration': 64.0,
|
||||||
|
'size': 2097152,
|
||||||
|
'view_count': int,
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
|
# mp4
|
||||||
'url': 'https://promodj.com/djperetse/videos/5868236/Fatalist_Project_feat_DJ_Peretse_Den_pobedi_Videoklip',
|
'url': 'https://promodj.com/djperetse/videos/5868236/Fatalist_Project_feat_DJ_Peretse_Den_pobedi_Videoklip',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '5868236',
|
||||||
|
'ext': 'mp4',
|
||||||
|
'title': 'Fatalist Project feat. DJ Peretse - День победы (Видеоклип)',
|
||||||
|
'tags': ['House', 'Progressive House'],
|
||||||
|
'thumbnail': r're:^https?://',
|
||||||
|
'upload_date': '20160505',
|
||||||
|
'timestamp': 1462419720.0,
|
||||||
|
'duration': 265045,
|
||||||
|
'size': 165465292,
|
||||||
|
'view_count': int,
|
||||||
|
'channel': 'DJ Peretse',
|
||||||
|
'channel_url': 'https://promodj.com/djperetse',
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# avi
|
# avi
|
||||||
'url': 'https://promodj.com/djmikis/videos/5311597/Mikis_Live_SDJ_Show',
|
'url': 'https://promodj.com/djmikis/videos/5311597/Mikis_Live_SDJ_Show',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '5311597',
|
||||||
|
'ext': 'avi',
|
||||||
|
'title': 'Mikis Live @ SDJ Show',
|
||||||
|
'tags': ['Club House'],
|
||||||
|
'thumbnail': r're:^https?://',
|
||||||
|
'upload_date': '20150409',
|
||||||
|
'timestamp': 1428579840.0,
|
||||||
|
'duration': 1716240,
|
||||||
|
'size': 371195904,
|
||||||
|
'view_count': int,
|
||||||
|
'channel': 'MIKIS',
|
||||||
|
'channel_url': 'https://promodj.com/djmikis',
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# asf
|
# asf
|
||||||
'url': 'https://promodj.com/gigsiphonic/videos/7559341/Gigsiphonic_PODCAST_309_Extended_video_version',
|
'url': 'https://promodj.com/gigsiphonic/videos/7559341/Gigsiphonic_PODCAST_309_Extended_video_version',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '7559341',
|
||||||
|
'ext': 'asf',
|
||||||
|
'title': 'Gigsiphonic - PODCAST 309 (Extended video version)',
|
||||||
|
'tags': ['Synthwave', 'Synth-Pop'],
|
||||||
|
'thumbnail': r're:^https?://',
|
||||||
|
'upload_date': '20240210',
|
||||||
|
'timestamp': 1707580080.0,
|
||||||
|
'duration': 4309200,
|
||||||
|
'size': 3715146711,
|
||||||
|
'view_count': int,
|
||||||
|
'channel': 'Gigsiphonic',
|
||||||
|
'channel_url': 'https://promodj.com/gigsiphonic',
|
||||||
|
},
|
||||||
}, {
|
}, {
|
||||||
# not valid html
|
# not valid html
|
||||||
'url': 'https://promodj.com/martin.sehnal/videos/7555841/Martin_Sehnal_CII_33_Plus_CII_32_Clothes_on_the_peg_2_020_2_024_02_01th',
|
'url': 'https://promodj.com/martin.sehnal/videos/7555841/Martin_Sehnal_CII_33_Plus_CII_32_Clothes_on_the_peg_2_020_2_024_02_01th',
|
||||||
'only_matching': True,
|
'info_dict': {
|
||||||
|
'id': '7555841',
|
||||||
|
'ext': 'avi',
|
||||||
|
'title': 'Martin Sehnal - CII 33 ( Plus CII 32 ) Clothes on the peg 2 020 ( 2 024 02. 01th ) )',
|
||||||
|
'tags': ['Easy Listening', 'Drum & Bass'],
|
||||||
|
'thumbnail': r're:^https?://',
|
||||||
|
'upload_date': '20240201',
|
||||||
|
'timestamp': 1706827560.0,
|
||||||
|
'duration': 30000,
|
||||||
|
'size': 2340757176,
|
||||||
|
'view_count': int,
|
||||||
|
'channel_url': 'https://promodj.com/martin.sehnal',
|
||||||
|
'channel': 'Martin Sehnal',
|
||||||
|
},
|
||||||
}]
|
}]
|
||||||
|
|
||||||
_IS_PAID_RE = r'<b>Цена:</b>'
|
_IS_PAID_RE = r'<b>Цена:</b>'
|
||||||
# examples: MP3, 320 Кбит | MP4, 20157 Кбит | WAV, 1412 Кбит | AVI, 1731 Кбит | ASF, 6905 Кбит
|
# examples: MP3, 320 Кбит | MP4, 20157 Кбит | WAV, 1412 Кбит | AVI, 1731 Кбит | ASF, 6905 Кбит | FLAC, 1509 Кбит
|
||||||
_FORMATS_RE = r'<a\s+href=\"(?P<url>[^\"]+)\">\s*(?P<format>\w+), (?P<bitrate>\d+) Кбит\s*</a>'
|
# https://regex101.com/r/2AuaxB/1
|
||||||
|
_FORMATS_RE = r'(?:<a\s+href=\"(?P<url>[^\"]+)\">)?\s*(?P<format>\w+), (?P<bitrate>\d+) Кбит'
|
||||||
_VIEW_COUNT_RE = r'<b>(?:Прослушиваний|Просмотров):</b>\s*(\d+)'
|
_VIEW_COUNT_RE = r'<b>(?:Прослушиваний|Просмотров):</b>\s*(\d+)'
|
||||||
# examples: 0:21 | 1:07 | 74:38
|
# examples: 0:21 | 1:07 | 74:38
|
||||||
_DURATION_RE = r'<b>Продолжительность:</b>\s*(\d+:\d{2})'
|
_DURATION_RE = r'<b>Продолжительность:</b>\s*(\d+:\d{2})'
|
||||||
|
@ -461,55 +671,64 @@ class PromoDJIE(PromoDJBaseIE):
|
||||||
size, size_unit = raw_size
|
size, size_unit = raw_size
|
||||||
return int(float(size) * pow(1024, RU_SIZE_UNITS.index(size_unit)))
|
return int(float(size) * pow(1024, RU_SIZE_UNITS.index(size_unit)))
|
||||||
|
|
||||||
def _parse_media(self, html, id, type):
|
# music: always have lossy format (mp3), sometimes have lossless (wav or flac) format
|
||||||
|
# video: sometimes have source format (mp4, avi, asf), always have converted for web format (mp4)
|
||||||
|
def _real_extract(self, url):
|
||||||
|
type, id = self._match_valid_url(url).groups()
|
||||||
|
html = self._download_webpage(url, id)
|
||||||
|
|
||||||
|
# always returns only one format: lossy mp3 for music or converted mp4 for video
|
||||||
|
media_data = self._search_json(
|
||||||
|
'', html, 'media data', id,
|
||||||
|
contains_pattern=self._VIDEO_DATA_REGEX if type == 'videos' else self._MUSIC_DATA_REGEX,
|
||||||
|
transform_source=js_to_json, fatal=False, default=None)
|
||||||
|
if not media_data:
|
||||||
|
media_data = self._fetch_media_data([id], id)[0]
|
||||||
|
metadata = self._parse_media_data(media_data, id)
|
||||||
|
|
||||||
# html can be invalid
|
# html can be invalid
|
||||||
try:
|
try:
|
||||||
meta_html = get_elements_html_by_class('dj_universal', html)[1]
|
meta_html = get_elements_html_by_class('dj_universal', html)[1]
|
||||||
except Exception:
|
except Exception:
|
||||||
meta_html = html
|
meta_html = html
|
||||||
|
|
||||||
|
# music: lossy format or lossless and lossy formats
|
||||||
|
# video: source format
|
||||||
|
# download links can be missing
|
||||||
|
# best quality format always comes first
|
||||||
formats_from_html = re.findall(self._FORMATS_RE, meta_html)
|
formats_from_html = re.findall(self._FORMATS_RE, meta_html)
|
||||||
has_formats = len(formats_from_html) != 0
|
|
||||||
is_paid = re.search(self._IS_PAID_RE, meta_html)
|
is_paid = re.search(self._IS_PAID_RE, meta_html)
|
||||||
|
# size field describes best quality
|
||||||
if not has_formats and is_paid:
|
size = self._parse_ru_size(re.search(self._SIZE_RE, meta_html).groups())
|
||||||
media_data_raw = self._search_regex(self._MUSIC_DATA_REGEX, html, 'media data')
|
if type == 'videos':
|
||||||
media_data = self._parse_json(media_data_raw, id)
|
for url, _, bitrate in formats_from_html:
|
||||||
formats = [{
|
if url_or_none(url):
|
||||||
'url': source.get('URL'),
|
metadata['formats'].append({
|
||||||
'size': int_or_none(source.get('size')),
|
|
||||||
} for source in traverse_obj(media_data, ('sources')) if url_or_none(source.get('URL'))]
|
|
||||||
elif not has_formats and type == 'videos':
|
|
||||||
media_data_raw = self._search_regex(self._VIDEO_DATA_REGEX, html, 'media data')
|
|
||||||
media_data = self._parse_json(media_data_raw, id)
|
|
||||||
video_config = self._parse_json(media_data['config'], id)
|
|
||||||
video = traverse_obj(video_config, ('playlist', 'item', 0))
|
|
||||||
formats = [{
|
|
||||||
'url': traverse_obj(video, ('play', '@url', {url_or_none})),
|
|
||||||
}]
|
|
||||||
else:
|
|
||||||
formats = [{
|
|
||||||
'url': url,
|
'url': url,
|
||||||
'format': format.lower(),
|
|
||||||
'tbr': int(bitrate),
|
'tbr': int(bitrate),
|
||||||
} for url, format, bitrate in formats_from_html if url_or_none(url)]
|
'size': size,
|
||||||
# size field describes best quality. best quality always comes first
|
'quality': 1,
|
||||||
formats[0]['size'] = self._parse_ru_size(re.findall(self._SIZE_RE, meta_html)[0])
|
})
|
||||||
|
else:
|
||||||
|
for i, match in enumerate(formats_from_html):
|
||||||
|
url, _, bitrate = match
|
||||||
|
is_last = i == len(formats_from_html) - 1
|
||||||
|
if is_last:
|
||||||
|
metadata['formats'][0]['abr'] = int(bitrate)
|
||||||
|
elif url_or_none(url) and not is_paid:
|
||||||
|
metadata['formats'].append({
|
||||||
|
'url': url,
|
||||||
|
'abr': int(bitrate),
|
||||||
|
})
|
||||||
|
metadata['formats'][-1]['size'] = size
|
||||||
|
|
||||||
return {
|
return merge_dicts(metadata, {
|
||||||
'id': id,
|
|
||||||
'title': clean_html(get_element_by_class('file_title', html)),
|
'title': clean_html(get_element_by_class('file_title', html)),
|
||||||
'formats': formats,
|
|
||||||
'view_count': int_or_none(self._search_regex(self._VIEW_COUNT_RE, meta_html, 'view_count', default=None)),
|
'view_count': int_or_none(self._search_regex(self._VIEW_COUNT_RE, meta_html, 'view_count', default=None)),
|
||||||
'duration': parse_duration(self._search_regex(self._DURATION_RE, meta_html, 'duration')),
|
'duration': parse_duration(self._search_regex(self._DURATION_RE, meta_html, 'duration')),
|
||||||
'timestamp': self._parse_ru_date(re.findall(self._TIMESTAMP_RE, meta_html)[0]),
|
'timestamp': self._parse_ru_date(re.findall(self._TIMESTAMP_RE, meta_html)[0]),
|
||||||
'tags': self._html_search_regex(self._TAGS_RE, meta_html, 'tags').split(', '),
|
'tags': self._html_search_regex(self._TAGS_RE, meta_html, 'tags').split(', '),
|
||||||
}
|
})
|
||||||
|
|
||||||
def _real_extract(self, url):
|
|
||||||
type, id = self._match_valid_url(url).groups()
|
|
||||||
html = self._download_webpage(url, id)
|
|
||||||
return self._parse_media(html, id, type)
|
|
||||||
|
|
||||||
|
|
||||||
class PromoDJEmbedIE(PromoDJBaseIE):
|
class PromoDJEmbedIE(PromoDJBaseIE):
|
||||||
|
@ -534,21 +753,11 @@ class PromoDJEmbedIE(PromoDJBaseIE):
|
||||||
'only_matching': True,
|
'only_matching': True,
|
||||||
}]
|
}]
|
||||||
|
|
||||||
def _get_full_url(self, media_data, id):
|
|
||||||
if player_error := media_data.get('player_error'):
|
|
||||||
raise ExtractorError(player_error, expected=True)
|
|
||||||
|
|
||||||
if media_data.get('video'):
|
|
||||||
video_config = self._parse_json(media_data['config'], id)
|
|
||||||
video = traverse_obj(video_config, ('playlist', 'item', 0))
|
|
||||||
return traverse_obj(video, ('title', '@ico_url'))
|
|
||||||
else:
|
|
||||||
return media_data.get('titleURL')
|
|
||||||
|
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
id = self._match_id(url)
|
id = self._match_id(url)
|
||||||
url = self._get_full_url(self._fetch_media_data([id], id)[0], id)
|
metadata = self._parse_media_data(
|
||||||
return self.url_result(url, PromoDJIE, id)
|
self._fetch_media_data([id], id)[0], id)
|
||||||
|
return self.url_result(metadata['webpage_url'], PromoDJIE, id)
|
||||||
|
|
||||||
|
|
||||||
class PromoDJShortIE(PromoDJBaseIE):
|
class PromoDJShortIE(PromoDJBaseIE):
|
||||||
|
@ -558,12 +767,13 @@ class PromoDJShortIE(PromoDJBaseIE):
|
||||||
'only_matching': True,
|
'only_matching': True,
|
||||||
}]
|
}]
|
||||||
|
|
||||||
_PAGE_URL_REGEX = r'<meta property="og:url"\s*content="(?P<url>[^"]+)"'
|
|
||||||
|
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
id = self._match_id(url)
|
id = self._match_id(url)
|
||||||
html = self._download_webpage(url, id)
|
html = self._download_webpage(url, id)
|
||||||
url = re.findall(self._PAGE_URL_REGEX, html)[0]
|
try:
|
||||||
|
url = self._og_search_url(html)
|
||||||
|
except Exception:
|
||||||
|
raise ExtractorError('Unable to extract full URL')
|
||||||
return self.url_result(url, PromoDJIE, id)
|
return self.url_result(url, PromoDJIE, id)
|
||||||
|
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue
Block a user