mirror of
https://github.com/ytdl-org/youtube-dl.git
synced 2024-11-14 06:36:44 +01:00
52 lines
1.8 KiB
Python
52 lines
1.8 KiB
Python
# coding: utf-8
|
|
from __future__ import unicode_literals
|
|
|
|
import re
|
|
|
|
from .common import InfoExtractor
|
|
from ..utils import int_or_none
|
|
|
|
|
|
class PlaysTVIE(InfoExtractor):
|
|
_VALID_URL = r'https?://(?:www\.)?plays\.tv/video/(?P<id>[0-9a-f]{18})'
|
|
_TEST = {
|
|
'url': 'http://plays.tv/video/56af17f56c95335490/when-you-outplay-the-azir-wall',
|
|
'md5': 'dfeac1198506652b5257a62762cec7bc',
|
|
'info_dict': {
|
|
'id': '56af17f56c95335490',
|
|
'ext': 'mp4',
|
|
'title': 'When you outplay the Azir wall',
|
|
'description': 'Posted by Bjergsen',
|
|
}
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
webpage = self._download_webpage(url, video_id)
|
|
|
|
title = self._og_search_title(webpage)
|
|
content = self._parse_json(
|
|
self._search_regex(
|
|
r'R\.bindContent\(({.+?})\);', webpage,
|
|
'content'), video_id)['content']
|
|
mpd_url, sources = re.search(
|
|
r'(?s)<video[^>]+data-mpd="([^"]+)"[^>]*>(.+?)</video>',
|
|
content).groups()
|
|
formats = self._extract_mpd_formats(
|
|
self._proto_relative_url(mpd_url), video_id, mpd_id='DASH')
|
|
for format_id, height, format_url in re.findall(r'<source\s+res="((\d+)h?)"\s+src="([^"]+)"', sources):
|
|
formats.append({
|
|
'url': self._proto_relative_url(format_url),
|
|
'format_id': 'http-' + format_id,
|
|
'height': int_or_none(height),
|
|
})
|
|
self._sort_formats(formats)
|
|
|
|
return {
|
|
'id': video_id,
|
|
'title': title,
|
|
'description': self._og_search_description(webpage),
|
|
'thumbnail': self._og_search_thumbnail(webpage),
|
|
'formats': formats,
|
|
}
|