[Auengine] corrected extractions logic
[youtube-dl] / youtube_dl / extractor / auengine.py
1 from __future__ import unicode_literals
2
3 import re
4
5 from .common import InfoExtractor
6 from ..compat import compat_urllib_parse
7 from ..utils import (
8     determine_ext,
9     ExtractorError,
10 )
11
12
13 class AUEngineIE(InfoExtractor):
14     _VALID_URL = r'http://(?:www\.)?auengine\.com/embed\.php\?.*?file=(?P<id>[^&]+).*?'
15
16     _TEST = {
17         'url': 'http://auengine.com/embed.php?file=lfvlytY6&w=650&h=370',
18         'md5': '48972bdbcf1a3a2f5533e62425b41d4f',
19         'info_dict': {
20             'id': 'lfvlytY6',
21             'ext': 'mp4',
22             'title': '[Commie]The Legend of the Legendary Heroes - 03 - Replication Eye (Alpha Stigma)[F9410F5A]'
23         }
24     }
25
26     def _real_extract(self, url):
27         video_id = self._match_id(url)
28
29         webpage = self._download_webpage(url, video_id)
30         title = self._html_search_regex(r'<title>(?P<title>.+?)</title>', webpage, 'title')
31         title = title.strip()
32         video_url = re.findall(r'http://\w+.auengine.com/vod/.*[^\W]', webpage)
33         video_url = map(compat_urllib_parse.unquote, video_url)[0]
34         thumbnail = re.findall(r'http://\w+.auengine.com/thumb/.*[^\W]', webpage)
35         thumbnail = map(compat_urllib_parse.unquote, thumbnail)[0]
36
37         if video_url == "" and thumbnail =="":
38             raise ExtractorError('Could not find video URL')
39         ext = '.' + determine_ext(video_url)
40         if ext == title[-len(ext):]:
41             title = title[:-len(ext)]
42
43         return {
44             'id': video_id,
45             'url': video_url,
46             'title': title,
47             'thumbnail': thumbnail,
48             'http_referer': 'http://www.auengine.com/flowplayer/flowplayer.commercial-3.2.14.swf',
49         }
50