mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-12-05 02:53:00 +01:00
41 lines
1.5 KiB
Python
41 lines
1.5 KiB
Python
# coding: utf-8
|
|
from __future__ import unicode_literals
|
|
|
|
from .common import InfoExtractor
|
|
|
|
|
|
class MiaoPaiIE(InfoExtractor):
|
|
_VALID_URL = r'https?://(?:www\.)?miaopai\.com/show/(?P<id>[-A-Za-z0-9~_]+)'
|
|
_TEST = {
|
|
'url': 'http://www.miaopai.com/show/n~0hO7sfV1nBEw4Y29-Hqg__.htm',
|
|
'md5': '095ed3f1cd96b821add957bdc29f845b',
|
|
'info_dict': {
|
|
'id': 'n~0hO7sfV1nBEw4Y29-Hqg__',
|
|
'ext': 'mp4',
|
|
'title': '西游记音乐会的秒拍视频',
|
|
'thumbnail': 're:^https?://.*/n~0hO7sfV1nBEw4Y29-Hqg___m.jpg',
|
|
}
|
|
}
|
|
|
|
_USER_AGENT_IPAD = 'Mozilla/5.0 (iPad; CPU OS 9_1 like Mac OS X) AppleWebKit/601.1.46 (KHTML, like Gecko) Version/9.0 Mobile/13B143 Safari/601.1'
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
webpage = self._download_webpage(
|
|
url, video_id, headers={'User-Agent': self._USER_AGENT_IPAD})
|
|
|
|
title = self._html_search_regex(
|
|
r'<title>([^<]+)</title>', webpage, 'title')
|
|
thumbnail = self._html_search_regex(
|
|
r'<div[^>]+class=(?P<q1>[\'"]).*\bvideo_img\b.*(?P=q1)[^>]+data-url=(?P<q2>[\'"])(?P<url>[^\'"]+)(?P=q2)',
|
|
webpage, 'thumbnail', fatal=False, group='url')
|
|
videos = self._parse_html5_media_entries(url, webpage, video_id)
|
|
info = videos[0]
|
|
|
|
info.update({
|
|
'id': video_id,
|
|
'title': title,
|
|
'thumbnail': thumbnail,
|
|
})
|
|
return info
|