From 91effe22a091035bc5abace2fcf562a0db89090f Mon Sep 17 00:00:00 2001 From: Remita Amine Date: Fri, 8 Feb 2019 07:21:31 +0100 Subject: [PATCH] [linkedin:learning] extract chapter_number and chapter_id(closes #19162) --- youtube_dl/extractor/linkedin.py | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) diff --git a/youtube_dl/extractor/linkedin.py b/youtube_dl/extractor/linkedin.py index 259fc4c5e..5a86b0064 100644 --- a/youtube_dl/extractor/linkedin.py +++ b/youtube_dl/extractor/linkedin.py @@ -34,12 +34,15 @@ class LinkedInLearningBaseIE(InfoExtractor): 'Csrf-Token': self._get_cookies(api_url)['JSESSIONID'].value, }, query=query)['elements'][0] - def _get_video_id(self, urn, course_slug, video_slug): + def _get_urn_id(self, video_data): + urn = video_data.get('urn') if urn: mobj = re.search(r'urn:li:lyndaCourse:\d+,(\d+)', urn) if mobj: return mobj.group(1) - return '%s/%s' % (course_slug, video_slug) + + def _get_video_id(self, video_data, course_slug, video_slug): + return self._get_urn_id(video_data) or '%s/%s' % (course_slug, video_slug) def _real_initialize(self): email, password = self._get_login_info() @@ -123,7 +126,7 @@ class LinkedInLearningIE(LinkedInLearningBaseIE): self._sort_formats(formats, ('width', 'height', 'source_preference', 'tbr', 'abr')) return { - 'id': self._get_video_id(video_data.get('urn'), course_slug, video_slug), + 'id': self._get_video_id(video_data, course_slug, video_slug), 'title': title, 'formats': formats, 'thumbnail': video_data.get('defaultThumbnail'), @@ -154,18 +157,21 @@ class LinkedInLearningCourseIE(LinkedInLearningBaseIE): course_data = self._call_api(course_slug, 'chapters,description,title') entries = [] - for chapter in course_data.get('chapters', []): + for chapter_number, chapter in enumerate(course_data.get('chapters', []), 1): chapter_title = chapter.get('title') + chapter_id = self._get_urn_id(chapter) for video in chapter.get('videos', []): video_slug = video.get('slug') if not video_slug: continue entries.append({ '_type': 'url_transparent', - 'id': self._get_video_id(video.get('urn'), course_slug, video_slug), + 'id': self._get_video_id(video, course_slug, video_slug), 'title': video.get('title'), 'url': 'https://www.linkedin.com/learning/%s/%s' % (course_slug, video_slug), 'chapter': chapter_title, + 'chapter_number': chapter_number, + 'chapter_id': chapter_id, 'ie_key': LinkedInLearningIE.ie_key(), })