diff options
author | Philipp Hagemeister <[email protected]> | 2014-04-21 03:21:34 +0200 |
---|---|---|
committer | Philipp Hagemeister <[email protected]> | 2014-04-21 03:21:34 +0200 |
commit | edec83a02579007da0f1043f7340ff2fe252a84b (patch) | |
tree | c877bc05c0b49e133493ad73badb5c89bbf216a1 /youtube_dl/extractor/infoq.py | |
parent | c0a7c608150bfdb36a4916988fc47375db621d59 (diff) | |
download | youtube-dl-edec83a02579007da0f1043f7340ff2fe252a84b.tar.gz youtube-dl-edec83a02579007da0f1043f7340ff2fe252a84b.zip |
[infoq] Add support for HTTP downloads (Fixes #722)
Diffstat (limited to 'youtube_dl/extractor/infoq.py')
-rw-r--r-- | youtube_dl/extractor/infoq.py | 27 |
1 files changed, 19 insertions, 8 deletions
diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py index 15cf8ad01..e76dd222d 100644 --- a/youtube_dl/extractor/infoq.py +++ b/youtube_dl/extractor/infoq.py @@ -13,9 +13,8 @@ class InfoQIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P<id>[^/]+)$' _TEST = { - 'name': 'InfoQ', 'url': 'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things', - 'md5': 'fcaa3d995e04080dcb9465d86b5eef62', + 'md5': 'b5ca0e0a8c1fed93b0e65e48e462f9a2', 'info_dict': { 'id': '12-jan-pythonthings', 'ext': 'mp4', @@ -37,20 +36,32 @@ class InfoQIE(InfoExtractor): video_url = 'rtmpe://video.infoq.com/cfx/st/' # Extract video URL - encoded_id = self._search_regex(r"jsclassref ?= ?'([^']*)'", webpage, 'encoded id') + 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('.') + 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, } |