Merge remote-tracking branch 'Dineshs91/f4m-2.0'

[youtube-dl] / youtube_dl / extractor / infoq.py

diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py
index 55f7608b5b257ecc7be179dd9386f731cf65c79a..f25f43664e262b25473557c5f11dae91e697e3f6 100644 (file)
--- a/youtube_dl/extractor/infoq.py
+++ b/youtube_dl/extractor/infoq.py
@@ -1,10 +1,9 @@
 from __future__ import unicode_literals
 
 import base64
-import re
 
 from .common import InfoExtractor
-from ..utils import (
+from ..compat import (
     compat_urllib_parse,
 )
 
@@ -13,42 +12,53 @@ class InfoQIE(InfoExtractor):
     _VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P<id>[^/]+)$'
 
     _TEST = {
-        u'name': u'InfoQ',
-        u'url': u'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things',
-        u'md5': u'fcaa3d995e04080dcb9465d86b5eef62',
-        u'info_dict': {
-            u'id': u'12-jan-pythonthings',
-            u'ext': u'mp4',
-            u'description': u'Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.',
-            u'title': u'A Few of My Favorite [Python] Things',
+        'url': 'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things',
+        'md5': 'b5ca0e0a8c1fed93b0e65e48e462f9a2',
+        'info_dict': {
+            'id': '12-jan-pythonthings',
+            'ext': 'mp4',
+            'description': 'Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.',
+            'title': 'A Few of My Favorite [Python] Things',
         },
     }
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
-
+        video_id = self._match_id(url)
         webpage = self._download_webpage(url, video_id)
 
-        self.report_extraction(video_id)
-
         video_title = self._html_search_regex(r'<title>(.*?)</title>', webpage, 'title')
         video_description = self._html_search_meta('description', webpage, 'description')
 
+        # The server URL is hardcoded
         video_url = 'rtmpe://video.infoq.com/cfx/st/'
-        base64playpath = self._search_regex(r"jsclassref = '([^']*)'", webpage, 'jsclassref')
-        playpath = 'mp4:' + base64.b64decode(base64playpath).decode('utf-8')
+
+        # Extract video URL
+        encoded_id = self._search_regex(
+            r"jsclassref\s*=\s*'([^']*)'", webpage, 'encoded id')
+        real_id = compat_urllib_parse.unquote(base64.b64decode(encoded_id.encode('ascii')).decode('utf-8'))
+        playpath = 'mp4:' + real_id
 
         video_filename = playpath.split('/')[-1]
         video_id, extension = video_filename.split('.')
 
-        return [{
+        http_base = self._search_regex(
+            r'EXPRESSINSTALL_SWF\s*=\s*"(https?://[^/"]+/)', webpage,
+            'HTTP base URL')
+
+        formats = [{
+            'format_id': 'rtmp',
+            'url': video_url,
+            'ext': extension,
+            'play_path': playpath,
+        }, {
+            'format_id': 'http',
+            'url': http_base + real_id,
+        }]
+        self._sort_formats(formats)
+
+        return {
             'id': video_id,
             'title': video_title,
             'description': video_description,
-            'formats': [{
-                'url': video_url,
-                'ext': extension,
-                'play_path': playpath,
-            }],
-        }]
+            'formats': formats,
+        }