SolarFM/plugins/youtube_download/yt_dlp/extractor/daftsex.py

151 lines
5.8 KiB
Python
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

from .common import InfoExtractor
from ..compat import compat_b64decode
from ..utils import (
ExtractorError,
int_or_none,
js_to_json,
parse_count,
parse_duration,
traverse_obj,
try_get,
unified_timestamp,
)
class DaftsexIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?daft\.sex/watch/(?P<id>-?\d+_\d+)'
_TESTS = [{
'url': 'https://daft.sex/watch/-35370899_456246186',
'md5': '64c04ef7b4c7b04b308f3b0c78efe7cd',
'info_dict': {
'id': '-35370899_456246186',
'ext': 'mp4',
'title': 'just relaxing',
'description': 'just relaxing Watch video Watch video in high quality',
'upload_date': '20201113',
'timestamp': 1605261911,
'thumbnail': r're:^https?://.*\.jpg$',
'age_limit': 18,
'duration': 15.0,
'view_count': int
},
}, {
'url': 'https://daft.sex/watch/-156601359_456242791',
'info_dict': {
'id': '-156601359_456242791',
'ext': 'mp4',
'title': 'Skye Blue - Dinner And A Show',
'description': 'Skye Blue - Dinner And A Show - Watch video Watch video in high quality',
'upload_date': '20200916',
'timestamp': 1600250735,
'thumbnail': 'https://psv153-1.crazycloud.ru/videos/-156601359/456242791/thumb.jpg?extra=i3D32KaBbBFf9TqDRMAVmQ',
},
'skip': 'deleted / private'
}]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
title = self._html_search_meta('name', webpage, 'title')
timestamp = unified_timestamp(self._html_search_meta('uploadDate', webpage, 'Upload Date', default=None))
description = self._html_search_meta('description', webpage, 'Description', default=None)
duration = parse_duration(self._search_regex(
r'Duration: ((?:[0-9]{2}:){0,2}[0-9]{2})',
webpage, 'duration', fatal=False))
views = parse_count(self._search_regex(
r'Views: ([0-9 ]+)',
webpage, 'views', fatal=False))
player_hash = self._search_regex(
r'DaxabPlayer\.Init\({[\s\S]*hash:\s*"([0-9a-zA-Z_\-]+)"[\s\S]*}',
webpage, 'player hash')
player_color = self._search_regex(
r'DaxabPlayer\.Init\({[\s\S]*color:\s*"([0-9a-z]+)"[\s\S]*}',
webpage, 'player color', fatal=False) or ''
embed_page = self._download_webpage(
'https://dxb.to/player/%s?color=%s' % (player_hash, player_color),
video_id, headers={'Referer': url})
video_params = self._parse_json(
self._search_regex(
r'window\.globParams\s*=\s*({[\S\s]+})\s*;\s*<\/script>',
embed_page, 'video parameters'),
video_id, transform_source=js_to_json)
server_domain = 'https://%s' % compat_b64decode(video_params['server'][::-1]).decode('utf-8')
cdn_files = traverse_obj(video_params, ('video', 'cdn_files')) or {}
if cdn_files:
formats = []
for format_id, format_data in cdn_files.items():
ext, height = format_id.split('_')
formats.append({
'format_id': format_id,
'url': f'{server_domain}/videos/{video_id.replace("_", "/")}/{height}.mp4?extra={format_data.split(".")[-1]}',
'height': int_or_none(height),
'ext': ext,
})
return {
'id': video_id,
'title': title,
'formats': formats,
'description': description,
'duration': duration,
'thumbnail': try_get(video_params, lambda vi: 'https:' + compat_b64decode(vi['video']['thumb']).decode('utf-8')),
'timestamp': timestamp,
'view_count': views,
'age_limit': 18,
}
items = self._download_json(
f'{server_domain}/method/video.get/{video_id}', video_id,
headers={'Referer': url}, query={
'token': video_params['video']['access_token'],
'videos': video_id,
'ckey': video_params['c_key'],
'credentials': video_params['video']['credentials'],
})['response']['items']
if not items:
raise ExtractorError('Video is not available', video_id=video_id, expected=True)
item = items[0]
formats = []
for f_id, f_url in item.get('files', {}).items():
if f_id == 'external':
return self.url_result(f_url)
ext, height = f_id.split('_')
height_extra_key = traverse_obj(video_params, ('video', 'partial', 'quality', height))
if height_extra_key:
formats.append({
'format_id': f'{height}p',
'url': f'{server_domain}/{f_url[8:]}&videos={video_id}&extra_key={height_extra_key}',
'height': int_or_none(height),
'ext': ext,
})
thumbnails = []
for k, v in item.items():
if k.startswith('photo_') and v:
width = k.replace('photo_', '')
thumbnails.append({
'id': width,
'url': v,
'width': int_or_none(width),
})
return {
'id': video_id,
'title': title,
'formats': formats,
'comment_count': int_or_none(item.get('comments')),
'description': description,
'duration': duration,
'thumbnails': thumbnails,
'timestamp': timestamp,
'view_count': views,
'age_limit': 18,
}