_ Git - youtube-dl/blob - youtube_dl/extractor/musicplayon.py

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_urlparse
   6 from ..utils import (
   7     int_or_none,
   8     js_to_json,
   9     mimetype2ext,
  10 )
  11
  12
  13 class MusicPlayOnIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:.+?\.)?musicplayon\.com/play(?:-touch)?\?(?:v|pl=100&play)=(?P<id>\d+)'
  15
  16     _TEST = {
  17         'url': 'http://en.musicplayon.com/play?v=433377',
  18         'md5': '00cdcdea1726abdf500d1e7fd6dd59bb',
  19         'info_dict': {
  20             'id': '433377',
  21             'ext': 'mp4',
  22             'title': 'Rick Ross - Interview On Chelsea Lately (2014)',
  23             'description': 'Rick Ross Interview On Chelsea Lately',
  24             'duration': 342,
  25             'uploader': 'ultrafish',
  26         },
  27     }
  28
  29     def _real_extract(self, url):
  30         video_id = self._match_id(url)
  31
  32         page = self._download_webpage(url, video_id)
  33
  34         title = self._og_search_title(page)
  35         description = self._og_search_description(page)
  36         thumbnail = self._og_search_thumbnail(page)
  37         duration = self._html_search_meta('video:duration', page, 'duration', fatal=False)
  38         view_count = self._og_search_property('count', page, fatal=False)
  39         uploader = self._html_search_regex(
  40             r'<div>by&nbsp;<a href="[^"]+" class="purple">([^<]+)</a></div>', page, 'uploader', fatal=False)
  41
  42         sources = self._parse_json(
  43             self._search_regex(r'setup\[\'_sources\'\]\s*=\s*([^;]+);', page, 'video sources'),
  44             video_id, transform_source=js_to_json)
  45         formats = [{
  46             'url': compat_urlparse.urljoin(url, source['src']),
  47             'ext': mimetype2ext(source.get('type')),
  48             'format_note': source.get('data-res'),
  49         } for source in sources]
  50
  51         return {
  52             'id': video_id,
  53             'title': title,
  54             'description': description,
  55             'thumbnail': thumbnail,
  56             'uploader': uploader,
  57             'duration': int_or_none(duration),
  58             'view_count': int_or_none(view_count),
  59             'formats': formats,
  60         }