7
0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-12-15 09:46:46 +00:00
youtube-dl/youtube_dl/extractor/sexu.py

64 lines
2.0 KiB
Python
Raw Normal View History

2014-11-12 19:41:13 +00:00
from __future__ import unicode_literals
from .common import InfoExtractor
class SexuIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?sexu\.com/(?P<id>\d+)'
_TEST = {
'url': 'http://sexu.com/961791/',
'md5': 'ff615aca9691053c94f8f10d96cd7884',
'info_dict': {
'id': '961791',
'ext': 'mp4',
2014-11-13 13:02:53 +00:00
'title': 'md5:4d05a19a5fc049a63dbbaf05fb71d91b',
2016-03-06 10:57:41 +00:00
'description': 'md5:2b75327061310a3afb3fbd7d09e2e403',
2014-11-12 19:41:13 +00:00
'categories': list, # NSFW
'thumbnail': r're:https?://.*\.jpg$',
2014-11-12 19:41:13 +00:00
'age_limit': 18,
}
}
def _real_extract(self, url):
2014-11-13 14:20:49 +00:00
video_id = self._match_id(url)
2014-11-12 19:41:13 +00:00
webpage = self._download_webpage(url, video_id)
jwvideo = self._parse_json(
self._search_regex(r'\.setup\(\s*({.+?})\s*\);', webpage, 'jwvideo'),
video_id)
sources = jwvideo['sources']
2014-11-12 19:41:13 +00:00
formats = [{
'url': source['file'].replace('\\', ''),
'format_id': source.get('label'),
2017-06-08 17:30:23 +00:00
'height': int(self._search_regex(
r'^(\d+)[pP]', source.get('label', ''), 'height',
default=None)),
} for source in sources if source.get('file')]
2014-11-12 19:41:13 +00:00
self._sort_formats(formats)
title = self._html_search_regex(
2014-11-13 14:20:49 +00:00
r'<title>([^<]+)\s*-\s*Sexu\.Com</title>', webpage, 'title')
2014-11-12 19:41:13 +00:00
2014-11-13 14:20:49 +00:00
description = self._html_search_meta(
'description', webpage, 'description')
2014-11-12 19:41:13 +00:00
thumbnail = jwvideo.get('image')
2014-11-12 19:41:13 +00:00
2014-11-13 14:20:49 +00:00
categories_str = self._html_search_meta(
'keywords', webpage, 'categories')
2014-11-12 19:41:13 +00:00
categories = (
None if categories_str is None
else categories_str.split(','))
return {
'id': video_id,
'title': title,
'description': description,
'thumbnail': thumbnail,
'categories': categories,
'formats': formats,
'age_limit': 18,
}