youtube-dl

commit 38ad119f97cba871d34b057050547ba56b3e54c6
parent 4e415288d73f3ec15a0b2854de79c7359d1ae6fe
Author: Philipp Hagemeister <phihag@phihag.de>
Date:   Fri, 11 Jul 2014 13:34:19 +0200

[screencast] Add new extractor (Fixes #3236)

Diffstat:
M
youtube_dl/extractor/__init__.py
 | 
1
+
A
youtube_dl/extractor/screencast.py
 | 
48
++++++++++++++++++++++++++++++++++++++++++++++++

2 files changed, 49 insertions(+), 0 deletions(-)
diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
@@ -248,6 +248,7 @@ from .rutube import (
 from .rutv import RUTVIE
 from .savefrom import SaveFromIE
 from .scivee import SciVeeIE
+from .screencast import ScreencastIE
 from .servingsys import ServingSysIE
 from .sina import SinaIE
 from .slideshare import SlideshareIE
diff --git a/youtube_dl/extractor/screencast.py b/youtube_dl/extractor/screencast.py
@@ -0,0 +1,48 @@
+# -*- coding: utf-8 -*-
+from __future__ import unicode_literals
+
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    compat_parse_qs,
+    compat_urllib_request,
+)
+
+
+class ScreencastIE(InfoExtractor):
+    _VALID_URL = r'https?://www\.screencast\.com/t/(?P<id>[a-zA-Z0-9]+)'
+    _TEST = {
+        'url': 'http://www.screencast.com/t/3ZEjQXlT',
+        'md5': '917df1c13798a3e96211dd1561fded83',
+        'info_dict': {
+            'id': '3ZEjQXlT',
+            'ext': 'm4v',
+            'title': 'Color Measurement with Ocean Optics Spectrometers',
+            'description': 'md5:240369cde69d8bed61349a199c5fb153',
+            'thumbnail': 're:^https?://.*\.jpg$'
+        }
+    }
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id')
+        webpage = self._download_webpage(url, video_id)
+
+        flash_vars_s = self._html_search_regex(
+            r'<param name="flashVars" value="([^"]+)"', webpage, 'flash vars')
+        flash_vars = compat_parse_qs(flash_vars_s)
+
+        thumbnail = flash_vars.get('thumb', [None])[0]
+        video_url_raw = compat_urllib_request.quote(flash_vars['content'][0])
+        video_url = video_url_raw.replace('http%3A', 'http:')
+        title = self._og_search_title(webpage)
+        description = self._og_search_description(webpage)
+
+        return {
+            'id': video_id,
+            'url': video_url,
+            'title': title,
+            'description': description,
+            'thumbnail': thumbnail,
+        }