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

   1 # -*- coding: utf-8 -*-
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     ExtractorError,
   9     compat_parse_qs,
  10     compat_urllib_request,
  11 )
  12
  13
  14 class ScreencastIE(InfoExtractor):
  15     _VALID_URL = r'https?://www\.screencast\.com/t/(?P<id>[a-zA-Z0-9]+)'
  16     _TESTS = [{
  17         'url': 'http://www.screencast.com/t/3ZEjQXlT',
  18         'md5': '917df1c13798a3e96211dd1561fded83',
  19         'info_dict': {
  20             'id': '3ZEjQXlT',
  21             'ext': 'm4v',
  22             'title': 'Color Measurement with Ocean Optics Spectrometers',
  23             'description': 'md5:240369cde69d8bed61349a199c5fb153',
  24             'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
  25         }
  26     }, {
  27         'url': 'http://www.screencast.com/t/V2uXehPJa1ZI',
  28         'md5': 'e8e4b375a7660a9e7e35c33973410d34',
  29         'info_dict': {
  30             'id': 'V2uXehPJa1ZI',
  31             'ext': 'mov',
  32             'title': 'The Amadeus Spectrometer',
  33             'description': 're:^In this video, our friends at.*To learn more about Amadeus, visit',
  34             'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
  35         }
  36     }, {
  37         'url': 'http://www.screencast.com/t/aAB3iowa',
  38         'md5': 'dedb2734ed00c9755761ccaee88527cd',
  39         'info_dict': {
  40             'id': 'aAB3iowa',
  41             'ext': 'mp4',
  42             'title': 'Google Earth Export',
  43             'description': 'Provides a demo of a CommunityViz export to Google Earth, one of the 3D viewing options.',
  44             'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
  45         }
  46     },
  47     ]
  48
  49     def _real_extract(self, url):
  50         mobj = re.match(self._VALID_URL, url)
  51         video_id = mobj.group('id')
  52         webpage = self._download_webpage(url, video_id)
  53
  54         video_url = self._html_search_regex(
  55             r'<embed name="Video".*?src="([^"]+)"', webpage,
  56             'QuickTime embed', default=None)
  57
  58         if video_url is None:
  59             flash_vars_s = self._html_search_regex(
  60                 r'<param name="flashVars" value="([^"]+)"', webpage, 'flash vars',
  61                 default=None)
  62             if flash_vars_s:
  63                 flash_vars = compat_parse_qs(flash_vars_s)
  64                 video_url_raw = compat_urllib_request.quote(
  65                     flash_vars['content'][0])
  66                 video_url = video_url_raw.replace('http%3A', 'http:')
  67
  68         if video_url is None:
  69             video_meta = self._html_search_meta(
  70                 'og:video', webpage, default=None)
  71             if video_meta:
  72                 video_url = self._search_regex(
  73                     r'src=(.*?)(?:$|&)', video_meta,
  74                     'meta tag video URL', default=None)
  75
  76         if video_url is None:
  77             raise ExtractorError('Cannot find video')
  78
  79         title = self._og_search_title(webpage, default=None)
  80         if title is None:
  81             title = self._html_search_regex(
  82                 r'class="tabSeperator">></span><span class="tabText">(.*?)<',
  83                 webpage, 'title')
  84         thumbnail = self._og_search_thumbnail(webpage)
  85         description = self._og_search_description(webpage, default=None)
  86         if description is None:
  87             description = self._html_search_meta('description', webpage)
  88
  89         return {
  90             'id': video_id,
  91             'url': video_url,
  92             'title': title,
  93             'description': description,
  94             'thumbnail': thumbnail,
  95         }