X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/b8d8e13c1f9e4d3cdd7d41c5c9d711a36dd5f9c3..68a5583f34cad0cbaf50fa60ad5c6f8c934d0f1b:/youtube_dl/extractor/pluralsight.py diff --git a/youtube_dl/extractor/pluralsight.py b/youtube_dl/extractor/pluralsight.py index 5c798e8..aacc5d4 100644 --- a/youtube_dl/extractor/pluralsight.py +++ b/youtube_dl/extractor/pluralsight.py @@ -18,6 +18,8 @@ from ..utils import ( parse_duration, qualities, srt_subtitles_timecode, + try_get, + update_url_query, urlencode_postdata, ) @@ -25,6 +27,39 @@ from ..utils import ( class PluralsightBaseIE(InfoExtractor): _API_BASE = 'https://app.pluralsight.com' + def _download_course(self, course_id, url, display_id): + try: + return self._download_course_rpc(course_id, url, display_id) + except ExtractorError: + # Old API fallback + return self._download_json( + 'https://app.pluralsight.com/player/user/api/v1/player/payload', + display_id, data=urlencode_postdata({'courseId': course_id}), + headers={'Referer': url}) + + def _download_course_rpc(self, course_id, url, display_id): + response = self._download_json( + '%s/player/functions/rpc' % self._API_BASE, display_id, + 'Downloading course JSON', + data=json.dumps({ + 'fn': 'bootstrapPlayer', + 'payload': { + 'courseId': course_id, + }, + }).encode('utf-8'), + headers={ + 'Content-Type': 'application/json;charset=utf-8', + 'Referer': url, + }) + + course = try_get(response, lambda x: x['payload']['course'], dict) + if course: + return course + + raise ExtractorError( + '%s said: %s' % (self.IE_NAME, response['error']['message']), + expected=True) + class PluralsightIE(PluralsightBaseIE): IE_NAME = 'pluralsight' @@ -39,7 +74,7 @@ class PluralsightIE(PluralsightBaseIE): 'info_dict': { 'id': 'hosting-sql-server-windows-azure-iaas-m7-mgmt-04', 'ext': 'mp4', - 'title': 'Management of SQL Server - Demo Monitoring', + 'title': 'Demo Monitoring', 'duration': 338, }, 'skip': 'Requires pluralsight account credentials', @@ -81,7 +116,7 @@ class PluralsightIE(PluralsightBaseIE): post_url = compat_urlparse.urljoin(self._LOGIN_URL, post_url) response = self._download_webpage( - post_url, None, 'Logging in as %s' % username, + post_url, None, 'Logging in', data=urlencode_postdata(login_form), headers={'Content-Type': 'application/x-www-form-urlencoded'}) @@ -92,6 +127,17 @@ class PluralsightIE(PluralsightBaseIE): raise ExtractorError('Unable to login: %s' % error, expected=True) if all(p not in response for p in ('__INITIAL_STATE__', '"currentUser"')): + BLOCKED = 'Your account has been blocked due to suspicious activity' + if BLOCKED in response: + raise ExtractorError( + 'Unable to login: %s' % BLOCKED, expected=True) + MUST_AGREE = 'To continue using Pluralsight, you must agree to' + if any(p in response for p in (MUST_AGREE, '>Disagree<', '>Agree<')): + raise ExtractorError( + 'Unable to login: %s some documents. Go to pluralsight.com, ' + 'log in and agree with what Pluralsight requires.' + % MUST_AGREE, expected=True) + raise ExtractorError('Unable to log in') def _get_subtitles(self, author, clip_id, lang, name, duration, video_id): @@ -125,12 +171,12 @@ class PluralsightIE(PluralsightBaseIE): for num, current in enumerate(subs): current = subs[num] start, text = ( - float_or_none(dict_get(current, TIME_OFFSET_KEYS)), + float_or_none(dict_get(current, TIME_OFFSET_KEYS, skip_false_values=False)), dict_get(current, TEXT_KEYS)) if start is None or text is None: continue end = duration if num == len(subs) - 1 else float_or_none( - dict_get(subs[num + 1], TIME_OFFSET_KEYS)) + dict_get(subs[num + 1], TIME_OFFSET_KEYS, skip_false_values=False)) if end is None: continue srt += os.linesep.join( @@ -157,18 +203,14 @@ class PluralsightIE(PluralsightBaseIE): display_id = '%s-%s' % (name, clip_id) - course = self._download_json( - 'https://app.pluralsight.com/player/user/api/v1/player/payload', - display_id, data=urlencode_postdata({'courseId': course_name}), - headers={'Referer': url}) + course = self._download_course(course_name, url, display_id) collection = course['modules'] - module, clip = None, None + clip = None for module_ in collection: if name in (module_.get('moduleName'), module_.get('name')): - module = module_ for clip_ in module_.get('clips', []): clip_index = clip_.get('clipIndex') if clip_index is None: @@ -182,7 +224,7 @@ class PluralsightIE(PluralsightBaseIE): if not clip: raise ExtractorError('Unable to resolve clip') - title = '%s - %s' % (module['title'], clip['title']) + title = clip['title'] QUALITIES = { 'low': {'width': 640, 'height': 480}, @@ -220,6 +262,7 @@ class PluralsightIE(PluralsightBaseIE): req_format_split = req_format.split('-', 1) if len(req_format_split) > 1: req_ext, req_quality = req_format_split + req_quality = '-'.join(req_quality.split('-')[:2]) for allowed_quality in ALLOWED_QUALITIES: if req_ext == allowed_quality.ext and req_quality in allowed_quality.qualities: return (AllowedQuality(req_ext, (req_quality, )), ) @@ -326,26 +369,34 @@ class PluralsightCourseIE(PluralsightBaseIE): # TODO: PSM cookie - course = self._download_json( - '%s/data/course/%s' % (self._API_BASE, course_id), - course_id, 'Downloading course JSON') + course = self._download_course(course_id, url, course_id) title = course['title'] + course_name = course['name'] + course_data = course['modules'] description = course.get('description') or course.get('shortDescription') - course_data = self._download_json( - '%s/data/course/content/%s' % (self._API_BASE, course_id), - course_id, 'Downloading course data JSON') - entries = [] for num, module in enumerate(course_data, 1): + author = module.get('author') + module_name = module.get('name') + if not author or not module_name: + continue for clip in module.get('clips', []): - player_parameters = clip.get('playerParameters') - if not player_parameters: + clip_index = int_or_none(clip.get('index')) + if clip_index is None: continue + clip_url = update_url_query( + '%s/player' % self._API_BASE, query={ + 'mode': 'live', + 'course': course_name, + 'author': author, + 'name': module_name, + 'clip': clip_index, + }) entries.append({ '_type': 'url_transparent', - 'url': '%s/training/player?%s' % (self._API_BASE, player_parameters), + 'url': clip_url, 'ie_key': PluralsightIE.ie_key(), 'chapter': module.get('title'), 'chapter_number': num,