X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/7df4343e553a9942a290e5e1dffe931261ab844e..4231113a7be907539052ce59df44e23c54d21d38:/youtube_dl/extractor/slideshare.py diff --git a/youtube_dl/extractor/slideshare.py b/youtube_dl/extractor/slideshare.py index 0b717a1..e89ebeb 100644 --- a/youtube_dl/extractor/slideshare.py +++ b/youtube_dl/extractor/slideshare.py @@ -9,11 +9,12 @@ from ..compat import ( ) from ..utils import ( ExtractorError, + get_element_by_id, ) class SlideshareIE(InfoExtractor): - _VALID_URL = r'https?://www\.slideshare\.net/[^/]+?/(?P.+?)($|\?)' + _VALID_URL = r'https?://(?:www\.)?slideshare\.net/[^/]+?/(?P<title>.+?)($|\?)' _TEST = { 'url': 'http://www.slideshare.net/Dataversity/keynote-presentation-managing-scale-and-complexity', @@ -30,7 +31,7 @@ class SlideshareIE(InfoExtractor): page_title = mobj.group('title') webpage = self._download_webpage(url, page_title) slideshare_obj = self._search_regex( - r'\$\.extend\(slideshare_object,\s*(\{.*?\})\);', + r'\$\.extend\(.*?slideshare_object,\s*(\{.*?\})\);', webpage, 'slideshare object') info = json.loads(slideshare_obj) if info['slideshow']['type'] != 'video': @@ -40,7 +41,7 @@ class SlideshareIE(InfoExtractor): bucket = info['jsplayer']['video_bucket'] ext = info['jsplayer']['video_extension'] video_url = compat_urlparse.urljoin(bucket, doc + '-SD.' + ext) - description = self._html_search_regex( + description = get_element_by_id('slideshow-description-paragraph', webpage) or self._html_search_regex( r'(?s)<p[^>]+itemprop="description"[^>]*>(.+?)</p>', webpage, 'description', fatal=False) @@ -51,5 +52,5 @@ class SlideshareIE(InfoExtractor): 'ext': ext, 'url': video_url, 'thumbnail': info['slideshow']['pin_image_url'], - 'description': description, + 'description': description.strip() if description else None, }