_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',
},
}
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,
+ }