[pluralsight] Clarify and randomize ViewClip sleep interval

[youtube-dl] / youtube_dl / extractor / pluralsight.py

diff --git a/youtube_dl/extractor/pluralsight.py b/youtube_dl/extractor/pluralsight.py
index 93244f41dba6228c86a8af129c065518a756cc61..8481a10b0c23900bd8197fa271dc9711caf0b5cf 100644 (file)
--- a/youtube_dl/extractor/pluralsight.py
+++ b/youtube_dl/extractor/pluralsight.py
@@ -1,6 +1,7 @@
 from __future__ import unicode_literals
 
 import json
+import random
 
 from .common import InfoExtractor
 from ..compat import (
@@ -16,10 +17,15 @@ from ..utils import (
 )
 
 
-class PluralsightIE(InfoExtractor):
+class PluralsightBaseIE(InfoExtractor):
+    _API_BASE = 'http://app.pluralsight.com'
+
+
+class PluralsightIE(PluralsightBaseIE):
     IE_NAME = 'pluralsight'
     _VALID_URL = r'https?://(?:(?:www|app)\.)?pluralsight\.com/training/player\?'
-    _LOGIN_URL = 'https://www.pluralsight.com/id/'
+    _LOGIN_URL = 'https://app.pluralsight.com/id/'
+
     _NETRC_MACHINE = 'pluralsight'
 
     _TESTS = [{
@@ -79,6 +85,9 @@ class PluralsightIE(InfoExtractor):
         if error:
             raise ExtractorError('Unable to login: %s' % error, expected=True)
 
+        if all(p not in response for p in ('__INITIAL_STATE__', '"currentUser"')):
+            raise ExtractorError('Unable to log in')
+
     def _real_extract(self, url):
         qs = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
 
@@ -142,12 +151,23 @@ class PluralsightIE(InfoExtractor):
                     'q': '%dx%d' % (f['width'], f['height']),
                 }
                 request = compat_urllib_request.Request(
-                    'http://app.pluralsight.com/training/Player/ViewClip',
+                    '%s/training/Player/ViewClip' % self._API_BASE,
                     json.dumps(clip_post).encode('utf-8'))
                 request.add_header('Content-Type', 'application/json;charset=utf-8')
                 format_id = '%s-%s' % (ext, quality)
                 clip_url = self._download_webpage(
                     request, display_id, 'Downloading %s URL' % format_id, fatal=False)
+
+                # Pluralsight tracks multiple sequential calls to ViewClip API and start
+                # to return 429 HTTP errors after some time (see
+                # https://github.com/rg3/youtube-dl/pull/6989). Moreover it may even lead
+                # to account ban (see https://github.com/rg3/youtube-dl/issues/6842).
+                # To somewhat reduce the probability of these consequences
+                # we will sleep random amount of time before each call to ViewClip.
+                self._sleep(
+                    random.randint(2, 5), display_id,
+                    '%(video_id)s: Waiting for %(timeout)s seconds to avoid throttling')
+
                 if not clip_url:
                     continue
                 f.update({
@@ -173,7 +193,7 @@ class PluralsightIE(InfoExtractor):
         }
 
 
-class PluralsightCourseIE(InfoExtractor):
+class PluralsightCourseIE(PluralsightBaseIE):
     IE_NAME = 'pluralsight:course'
     _VALID_URL = r'https?://(?:(?:www|app)\.)?pluralsight\.com/(?:library/)?courses/(?P<id>[^/]+)'
     _TESTS = [{
@@ -201,14 +221,14 @@ class PluralsightCourseIE(InfoExtractor):
         # TODO: PSM cookie
 
         course = self._download_json(
-            'http://www.pluralsight.com/data/course/%s' % course_id,
+            '%s/data/course/%s' % (self._API_BASE, course_id),
             course_id, 'Downloading course JSON')
 
         title = course['title']
         description = course.get('description') or course.get('shortDescription')
 
         course_data = self._download_json(
-            'http://www.pluralsight.com/data/course/content/%s' % course_id,
+            '%s/data/course/content/%s' % (self._API_BASE, course_id),
             course_id, 'Downloading course data JSON')
 
         entries = []
@@ -218,7 +238,7 @@ class PluralsightCourseIE(InfoExtractor):
                 if not player_parameters:
                     continue
                 entries.append(self.url_result(
-                    'http://www.pluralsight.com/training/player?%s' % player_parameters,
+                    '%s/training/player?%s' % (self._API_BASE, player_parameters),
                     'Pluralsight'))
 
         return self.playlist_result(entries, course_id, title, description)