From 0bc56fa66a4b0f1b6bf827bd3550a119d3e3b231 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Tue, 27 Aug 2013 12:38:30 +0200 Subject: [PATCH] Add an extractor for NBC news (closes #1320) --- youtube_dl/extractor/__init__.py | 1 + youtube_dl/extractor/nbc.py | 33 ++++++++++++++++++++++++++++++++ 2 files changed, 34 insertions(+) create mode 100644 youtube_dl/extractor/nbc.py diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index ea2af0d0e..27bbcc0f7 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -54,6 +54,7 @@ from .muzu import MuzuTVIE from .myspass import MySpassIE from .myvideo import MyVideoIE from .nba import NBAIE +from .nbc import NBCNewsIE from .ooyala import OoyalaIE from .pbs import PBSIE from .photobucket import PhotobucketIE diff --git a/youtube_dl/extractor/nbc.py b/youtube_dl/extractor/nbc.py new file mode 100644 index 000000000..3bc9dae6d --- /dev/null +++ b/youtube_dl/extractor/nbc.py @@ -0,0 +1,33 @@ +import re +import xml.etree.ElementTree + +from .common import InfoExtractor +from ..utils import find_xpath_attr, compat_str + + +class NBCNewsIE(InfoExtractor): + _VALID_URL = r'https?://www\.nbcnews\.com/video/.+?/(?P\d+)' + + _TEST = { + u'url': u'http://www.nbcnews.com/video/nbc-news/52753292', + u'file': u'52753292.flv', + u'md5': u'47abaac93c6eaf9ad37ee6c4463a5179', + u'info_dict': { + u'title': u'Crew emerges after four-month Mars food study', + u'description': u'md5:24e632ffac72b35f8b67a12d1b6ddfc1', + }, + } + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + video_id = mobj.group('id') + info_xml = self._download_webpage('http://www.nbcnews.com/id/%s/displaymode/1219' % video_id, video_id) + info = xml.etree.ElementTree.fromstring(info_xml.encode('utf-8')).find('video') + + return {'id': video_id, + 'title': info.find('headline').text, + 'ext': 'flv', + 'url': find_xpath_attr(info, 'media', 'type', 'flashVideo').text, + 'description': compat_str(info.find('caption').text), + 'thumbnail': find_xpath_attr(info, 'media', 'type', 'thumbnail').text, + }