]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/mit.py
Update changelog.
[youtubedl] / youtube_dl / extractor / mit.py
index 3c61a850f296c32861cdfd35095746c2cf1ef4ad..e1506a745a69e6c300f1e36cdc6b3e4ed8206bf0 100644 (file)
@@ -5,9 +5,6 @@ import json
 
 from .common import InfoExtractor
 from .youtube import YoutubeIE
-from ..compat import (
-    compat_urlparse,
-)
 from ..utils import (
     clean_html,
     ExtractorError,
@@ -21,12 +18,12 @@ class TechTVMITIE(InfoExtractor):
 
     _TEST = {
         'url': 'http://techtv.mit.edu/videos/25418-mit-dna-learning-center-set',
-        'md5': '1f8cb3e170d41fd74add04d3c9330e5f',
+        'md5': '00a3a27ee20d44bcaa0933ccec4a2cf7',
         'info_dict': {
             'id': '25418',
             'ext': 'mp4',
-            'title': 'MIT DNA Learning Center Set',
-            'description': 'md5:82313335e8a8a3f243351ba55bc1b474',
+            'title': 'MIT DNA and Protein Sets',
+            'description': 'md5:46f5c69ce434f0a97e7c628cc142802d',
         },
     }
 
@@ -36,8 +33,8 @@ class TechTVMITIE(InfoExtractor):
             'http://techtv.mit.edu/videos/%s' % video_id, video_id)
         clean_page = re.compile(r'<!--.*?-->', re.S).sub('', raw_page)
 
-        base_url = self._search_regex(
-            r'ipadUrl: \'(.+?cloudfront.net/)', raw_page, 'base url')
+        base_url = self._proto_relative_url(self._search_regex(
+            r'ipadUrl: \'(.+?cloudfront.net/)', raw_page, 'base url'), 'http:')
         formats_json = self._search_regex(
             r'bitrates: (\[.+?\])', raw_page, 'video formats')
         formats_mit = json.loads(formats_json)
@@ -68,33 +65,9 @@ class TechTVMITIE(InfoExtractor):
         }
 
 
-class MITIE(TechTVMITIE):
-    IE_NAME = 'video.mit.edu'
-    _VALID_URL = r'https?://video\.mit\.edu/watch/(?P<title>[^/]+)'
-
-    _TEST = {
-        'url': 'http://video.mit.edu/watch/the-government-is-profiling-you-13222/',
-        'md5': '7db01d5ccc1895fc5010e9c9e13648da',
-        'info_dict': {
-            'id': '21783',
-            'ext': 'mp4',
-            'title': 'The Government is Profiling You',
-            'description': 'md5:ad5795fe1e1623b73620dbfd47df9afd',
-        },
-    }
-
-    def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        page_title = mobj.group('title')
-        webpage = self._download_webpage(url, page_title)
-        embed_url = self._search_regex(
-            r'<iframe .*?src="(.+?)"', webpage, 'embed url')
-        return self.url_result(embed_url, ie='TechTVMIT')
-
-
 class OCWMITIE(InfoExtractor):
     IE_NAME = 'ocw.mit.edu'
-    _VALID_URL = r'^http://ocw\.mit\.edu/courses/(?P<topic>[a-z0-9\-]+)'
+    _VALID_URL = r'^https?://ocw\.mit\.edu/courses/(?P<topic>[a-z0-9\-]+)'
     _BASE_URL = 'http://ocw.mit.edu/'
 
     _TESTS = [
@@ -102,13 +75,12 @@ class OCWMITIE(InfoExtractor):
             'url': 'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/',
             'info_dict': {
                 'id': 'EObHWIEKGjA',
-                'ext': 'mp4',
+                'ext': 'webm',
                 'title': 'Lecture 7: Multiple Discrete Random Variables: Expectations, Conditioning, Independence',
                 'description': 'In this lecture, the professor discussed multiple random variables, expectations, and binomial distribution.',
                 'upload_date': '20121109',
                 'uploader_id': 'MIT',
                 'uploader': 'MIT OpenCourseWare',
-                # 'subtitles': 'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/MIT6_041F11_lec07_300k.mp4.srt'
             }
         },
         {
@@ -121,7 +93,6 @@ class OCWMITIE(InfoExtractor):
                 'uploader_id': 'MIT',
                 'uploader': 'MIT OpenCourseWare',
                 'description': 'This section contains lecture video excerpts, lecture notes, an interactive mathlet with supporting documents, and problem solving videos.',
-                # 'subtitles': 'http://ocw.mit.edu//courses/mathematics/18-01sc-single-variable-calculus-fall-2010/ocw-18.01-f07-lec01_300k.SRT'
             }
         }
     ]
@@ -140,7 +111,6 @@ class OCWMITIE(InfoExtractor):
             metadata = re.sub(r'[\'"]', '', embed_chapter_media.group(1))
             metadata = re.split(r', ?', metadata)
             yt = metadata[1]
-            subs = compat_urlparse.urljoin(self._BASE_URL, metadata[7])
         else:
             # search for call to ocw_embed_chapter_media(container_id, media_url, provider, page_url, image_url, captions_file)
             embed_media = re.search(r'ocw_embed_media\((.+?)\)', webpage)
@@ -148,7 +118,6 @@ class OCWMITIE(InfoExtractor):
                 metadata = re.sub(r'[\'"]', '', embed_media.group(1))
                 metadata = re.split(r', ?', metadata)
                 yt = metadata[1]
-                subs = compat_urlparse.urljoin(self._BASE_URL, metadata[5])
             else:
                 raise ExtractorError('Unable to find embedded YouTube video.')
         video_id = YoutubeIE.extract_id(yt)
@@ -159,7 +128,5 @@ class OCWMITIE(InfoExtractor):
             'title': title,
             'description': description,
             'url': yt,
-            'url_transparent'
-            'subtitles': subs,
             'ie_key': 'Youtube',
         }