From 365343131d752bece96d2279a3e0bcd7e9f0000f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Mon, 1 Oct 2018 21:45:24 +0700 Subject: [PATCH] [pluralsight] Fix subtitles extraction (closes #17726, closes #17728) --- youtube_dl/extractor/pluralsight.py | 34 ++++++++++++++++++----------- 1 file changed, 21 insertions(+), 13 deletions(-) diff --git a/youtube_dl/extractor/pluralsight.py b/youtube_dl/extractor/pluralsight.py index ec67381bbd..daf1725702 100644 --- a/youtube_dl/extractor/pluralsight.py +++ b/youtube_dl/extractor/pluralsight.py @@ -210,18 +210,26 @@ def _login(self): raise ExtractorError('Unable to log in') - def _get_subtitles(self, author, clip_idx, lang, name, duration, video_id): - captions_post = { - 'a': author, - 'cn': int(clip_idx), - 'lc': lang, - 'm': name, - } - captions = self._download_json( - '%s/player/retrieve-captions' % self._API_BASE, video_id, - 'Downloading captions JSON', 'Unable to download captions JSON', - fatal=False, data=json.dumps(captions_post).encode('utf-8'), - headers={'Content-Type': 'application/json;charset=utf-8'}) + def _get_subtitles(self, author, clip_idx, clip_id, lang, name, duration, video_id): + captions = None + if clip_id: + captions = self._download_json( + '%s/transcript/api/v1/caption/json/%s/%s' + % (self._API_BASE, clip_id, lang), video_id, + 'Downloading captions JSON', 'Unable to download captions JSON', + fatal=False) + if not captions: + captions_post = { + 'a': author, + 'cn': int(clip_idx), + 'lc': lang, + 'm': name, + } + captions = self._download_json( + '%s/player/retrieve-captions' % self._API_BASE, video_id, + 'Downloading captions JSON', 'Unable to download captions JSON', + fatal=False, data=json.dumps(captions_post).encode('utf-8'), + headers={'Content-Type': 'application/json;charset=utf-8'}) if captions: return { lang: [{ @@ -413,7 +421,7 @@ def guess_allowed_qualities(): # TODO: other languages? subtitles = self.extract_subtitles( - author, clip_idx, 'en', name, duration, display_id) + author, clip_idx, clip.get('clipId'), 'en', name, duration, display_id) return { 'id': clip_id,