Add extractor for Engadget and 5min (closes #2465)
[youtube-dl] / youtube_dl / extractor / fivemin.py
1 from __future__ import unicode_literals
2
3 import re
4
5 from .common import InfoExtractor
6 from ..utils import (
7     compat_str,
8 )
9
10
11 class FiveMinIE(InfoExtractor):
12     IE_NAME = '5min'
13     _VALID_URL = r'''(?x)
14         (?:https?://[^/]*?5min\.com/Scripts/PlayerSeed\.js\?(.*?&)?playList=|
15             5min:)
16         (?P<id>\d+)
17         '''
18
19     _TEST = {
20         # From http://www.engadget.com/2013/11/15/ipad-mini-retina-display-review/
21         'url': 'http://pshared.5min.com/Scripts/PlayerSeed.js?sid=281&width=560&height=345&playList=518013791',
22         'md5': '4f7b0b79bf1a470e5004f7112385941d',
23         'info_dict': {
24             'id': '518013791',
25             'ext': 'mp4',
26             'title': 'iPad Mini with Retina Display Review',
27         },
28     }
29
30     @classmethod
31     def _build_result(cls, video_id):
32         return cls.url_result('5min:%s' % video_id, cls.ie_key())
33
34     def _real_extract(self, url):
35         mobj = re.match(self._VALID_URL, url)
36         video_id = mobj.group('id')
37         info = self._download_json(
38             'https://syn.5min.com/handlers/SenseHandler.ashx?func=GetResults&'
39             'playlist=%s&url=https' % video_id,
40             video_id)['binding'][0]
41
42         second_id = compat_str(int(video_id[:-2]) + 1)
43         formats = []
44         for quality, height in [(1, 320), (2, 480), (4, 720), (8, 1080)]:
45             if any(r['ID'] == quality for r in info['Renditions']):
46                 formats.append({
47                     'format_id': compat_str(quality),
48                     'url': 'http://avideos.5min.com/%s/%s/%s_%s.mp4' % (second_id[-3:], second_id, video_id, quality),
49                     'height': height,
50                 })
51
52         return {
53             'id': video_id,
54             'title': info['Title'],
55             'formats': formats,
56         }