w/youtube-dl
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

[cnn] Improve regexes and fix test

Browse source
This commit is contained in:
Sergey M․ 2014-12-29 20:27:09 +06:00
parent 974739aab5
commit 7fc2cd819e
1 changed files with 5 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
youtube_dl/extractor/cnn.py
View file

@ -11,14 +11,14 @@ from ..utils import (
class CNNIE(InfoExtractor): class CNNIE(InfoExtractor):
_VALID_URL = r'''(?x)https?://((edition|www)\.)?cnn\.com/video/(data/.+?|\?)/ _VALID_URL = r'''(?x)https?://(?:(?:edition|www)\.)?cnn\.com/video/(?:data/.+?|\?)/
(?P<path>.+?/(?P<title>[^/]+?)(?:\.cnn(-ap)?|(?=&)))''' (?P<path>.+?/(?P<title>[^/]+?)(?:\.cnn(?:-ap)?|(?=&)))'''
_TESTS = [{ _TESTS = [{
'url': 'http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn', 'url': 'http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn',
'md5': '3e6121ea48df7e2259fe73a0628605c4', 'md5': '3e6121ea48df7e2259fe73a0628605c4',
'info_dict': { 'info_dict': {
'id': 'sports_2013_06_09_nadal-1-on-1.cnn', 'id': 'sports/2013/06/09/nadal-1-on-1.cnn',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Nadal wins 8th French Open title', 'title': 'Nadal wins 8th French Open title',
'description': 'World Sport\'s Amanda Davies chats with 2013 French Open champion Rafael Nadal.', 'description': 'World Sport\'s Amanda Davies chats with 2013 French Open champion Rafael Nadal.',
@ -130,7 +130,7 @@ class CNNBlogsIE(InfoExtractor):
class CNNArticleIE(InfoExtractor): class CNNArticleIE(InfoExtractor):
_VALID_URL = r'https?://((edition|www)\.)?cnn\.com/(?!video/)' _VALID_URL = r'https?://(?:(?:edition|www)\.)?cnn\.com/(?!video/)'
_TEST = { _TEST = {
'url': 'http://www.cnn.com/2014/12/21/politics/obama-north-koreas-hack-not-war-but-cyber-vandalism/', 'url': 'http://www.cnn.com/2014/12/21/politics/obama-north-koreas-hack-not-war-but-cyber-vandalism/',
'md5': '275b326f85d80dff7592a9820f5dc887', 'md5': '275b326f85d80dff7592a9820f5dc887',
@ -149,6 +149,6 @@ class CNNArticleIE(InfoExtractor):
cnn_url = self._html_search_regex(r"video:\s*'([^']+)'", webpage, 'cnn url') cnn_url = self._html_search_regex(r"video:\s*'([^']+)'", webpage, 'cnn url')
return { return {
'_type': 'url', '_type': 'url',
'url': r'http://cnn.com/video/?/video/' + cnn_url, 'url': 'http://cnn.com/video/?/video/' + cnn_url,
'ie_key': CNNIE.ie_key(), 'ie_key': CNNIE.ie_key(),
} }