X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Finfoq.py;h=f25f43664e262b25473557c5f11dae91e697e3f6;hb=b0ab0fac49057ea0df3db75fb407af92deaaa453;hp=15cf8ad010fcffd48564b935a7cba19c6957ec6a;hpb=c0a7c608150bfdb36a4916988fc47375db621d59;p=youtube-dl diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py index 15cf8ad01..f25f43664 100644 --- 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, ) @@ -13,9 +12,8 @@ class InfoQIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P[^/]+)$' _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', @@ -25,9 +23,7 @@ class InfoQIE(InfoExtractor): } 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) video_title = self._html_search_regex(r'(.*?)', webpage, 'title') @@ -37,20 +33,32 @@ 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, }