laudom/haruhi-dl
2
0
Fork 0
Code Issues 13 Pull requests Packages Projects Releases Wiki Activity

version 2020.11.08

Browse source
This commit is contained in:
Dominika 2020-11-08 00:40:17 +01:00
parent ca3abff9ea
commit 7e9e7a3c91
4 changed files with 22 additions and 14 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
ChangeLog
View file

@ -1,11 +1,18 @@
version 2020.11.08
Extractors
* [youtube] age-gate fixes
* [aliexpress] Added extractor
* [polskieradio] Fixed title extraction
* [tiktok] Added extractor
version 2020.11.06 version 2020.11.06
Extractors Extractors
* youtube fixes * [youtube] fixes
* Added onnetwork extractor * [onnetwork] Added extractor
version 2020.11.01.01 version 2020.11.01.01
Extractors Extractors
* fixed youtube? * [youtube] fixes like/dislike extraction
version 2020.11.01 version 2020.11.01
Core Core

7
haruhi_dl/extractor/polskieradio.py
View file

@ -78,9 +78,11 @@ class PolskieRadioIE(InfoExtractor):
media_urls = set() media_urls = set()
title = self._og_search_title(webpage).strip()
for data_media in re.findall(r'<[^>]+data-media=({[^>]+})', content): for data_media in re.findall(r'<[^>]+data-media=({[^>]+})', content):
media = self._parse_json(data_media, playlist_id, fatal=False) media = self._parse_json(data_media, playlist_id, fatal=False)
if not media.get('file') or not media.get('desc'): if not media.get('file'):
continue continue
media_url = self._proto_relative_url(media['file'], 'http:') media_url = self._proto_relative_url(media['file'], 'http:')
if media_url in media_urls: if media_url in media_urls:
@ -89,14 +91,13 @@ class PolskieRadioIE(InfoExtractor):
entries.append({ entries.append({
'id': compat_str(media['id']), 'id': compat_str(media['id']),
'url': media_url, 'url': media_url,
'title': compat_urllib_parse_unquote(media['desc']), 'title': compat_urllib_parse_unquote(media['desc']) or title,
'duration': int_or_none(media.get('length')), 'duration': int_or_none(media.get('length')),
'vcodec': 'none' if media.get('provider') == 'audio' else None, 'vcodec': 'none' if media.get('provider') == 'audio' else None,
'timestamp': timestamp, 'timestamp': timestamp,
'thumbnail': thumbnail_url 'thumbnail': thumbnail_url
}) })
title = self._og_search_title(webpage).strip()
description = strip_or_none(self._og_search_description(webpage)) description = strip_or_none(self._og_search_description(webpage))
return self.playlist_result(entries, playlist_id, title, description) return self.playlist_result(entries, playlist_id, title, description)

14
haruhi_dl/extractor/youtube.py
View file

@ -1680,8 +1680,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
data = compat_urllib_parse_urlencode({ data = compat_urllib_parse_urlencode({
'video_id': video_id, 'video_id': video_id,
'eurl': 'https://youtube.googleapis.com/v/' + video_id, 'eurl': 'https://youtube.googleapis.com/v/' + video_id,
'sts': self._search_regex( # 'sts': self._search_regex(
r'"sts"\s*:\s*(\d+)', embed_webpage, 'sts', default=''), # r'"sts"\s*:\s*(\d+)', embed_webpage, 'sts', default=''),
}) })
video_info_url = proto + '://www.youtube.com/get_video_info?' + data video_info_url = proto + '://www.youtube.com/get_video_info?' + data
try: try:
@ -1931,11 +1931,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
player_url = self._search_regex( player_url = self._search_regex(
ASSETS_RE, embed_webpage, 'JS player URL') ASSETS_RE, embed_webpage, 'JS player URL')
if player_url is None: #if player_url is None:
player_url_json = self._search_regex( # player_url_json = self._search_regex(
r'ytplayer\.config.*?"url"\s*:\s*("[^"]+")', # r'ytplayer\.config.*?"url"\s*:\s*("[^"]+")',
video_webpage, 'age gate player URL') # video_webpage, 'age gate player URL')
player_url = json.loads(player_url_json) # player_url = json.loads(player_url_json)
if 'sig' in url_data: if 'sig' in url_data:
url += '&signature=' + url_data['sig'][0] url += '&signature=' + url_data['sig'][0]

2
haruhi_dl/version.py
View file

@ -1,3 +1,3 @@
from __future__ import unicode_literals from __future__ import unicode_literals
__version__ = '2020.11.06' __version__ = '2020.11.08'