X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fted.py;h=4bca62ba003e325ebedd0fcc74c953bd64120cd5;hb=c0f9969b9e297a2c4ceb87017081792e13f4cdc6;hp=239e2a4482ea6bbf03089181e1301f9a12cbde5b;hpb=a9a3876d55be943a7eaf505cbeb8fb862514db6c;p=youtube-dl
diff --git a/youtube_dl/extractor/ted.py b/youtube_dl/extractor/ted.py
index 239e2a448..4bca62ba0 100644
--- a/youtube_dl/extractor/ted.py
+++ b/youtube_dl/extractor/ted.py
@@ -3,6 +3,10 @@ import re
from .subtitles import SubtitlesInfoExtractor
+from ..utils import (
+ RegexNotFoundError,
+)
+
class TEDIE(SubtitlesInfoExtractor):
_VALID_URL=r'''http://www\.ted\.com/
(
@@ -31,33 +35,32 @@ class TEDIE(SubtitlesInfoExtractor):
def _real_extract(self, url):
m=re.match(self._VALID_URL, url, re.VERBOSE)
if m.group('type_talk'):
- return [self._talk_info(url)]
+ return self._talk_info(url)
else :
playlist_id=m.group('playlist_id')
name=m.group('name')
self.to_screen(u'Getting info of playlist %s: "%s"' % (playlist_id,name))
return [self._playlist_videos_info(url,name,playlist_id)]
- def _playlist_videos_info(self,url,name,playlist_id=0):
+
+ def _playlist_videos_info(self, url, name, playlist_id):
'''Returns the videos of the playlist'''
- video_RE=r'''
-
\d+)"
- ([.\s]*?)data-playlist_item_id="(\d+)"
- ([.\s]*?)data-mediaslug="(?P.+?)"
- '''
- video_name_RE=r'(?P.+?)
'
- webpage=self._download_webpage(url, playlist_id, 'Downloading playlist webpage')
- m_videos=re.finditer(video_RE,webpage,re.VERBOSE)
- m_names=re.finditer(video_name_RE,webpage)
+
+ webpage = self._download_webpage(
+ url, playlist_id, u'Downloading playlist webpage')
+ matches = re.finditer(
+ r'/talks/[^"]+\.html)">[^<]*
',
+ webpage)
playlist_title = self._html_search_regex(r'div class="headline">\s*?\s*?(.*?)',
webpage, 'playlist title')
- playlist_entries = []
- for m_video, m_name in zip(m_videos,m_names):
- talk_url='http://www.ted.com%s' % m_name.group('talk_url')
- playlist_entries.append(self.url_result(talk_url, 'TED'))
- return self.playlist_result(playlist_entries, playlist_id = playlist_id, playlist_title = playlist_title)
+ playlist_entries = [
+ self.url_result(u'http://www.ted.com' + m.group('talk_url'), 'TED')
+ for m in matches
+ ]
+ return self.playlist_result(
+ playlist_entries, playlist_id=playlist_id, playlist_title=playlist_title)
def _talk_info(self, url, video_id=0):
"""Return the video for the talk in the url"""
@@ -80,7 +83,7 @@ class TEDIE(SubtitlesInfoExtractor):
'ext': 'mp4',
'url': stream['file'],
'format': stream['id']
- } for stream in info['htmlStreams']]
+ } for stream in info['htmlStreams']]
video_id = info['id']
@@ -90,7 +93,7 @@ class TEDIE(SubtitlesInfoExtractor):
self._list_available_subtitles(video_id, webpage)
return
- info = {
+ return {
'id': video_id,
'title': title,
'thumbnail': thumbnail,
@@ -99,18 +102,16 @@ class TEDIE(SubtitlesInfoExtractor):
'formats': formats,
}
- # TODO: Remove when #980 has been merged
- info.update(info['formats'][-1])
-
- return info
-
def _get_available_subtitles(self, video_id, webpage):
- options = self._search_regex(r'(?:)', webpage, 'subtitles_language_select', flags=re.DOTALL)
- languages = re.findall(r'(?: