[egghead:lesson] Add extractor (#6635)

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

diff --git a/youtube_dl/extractor/egghead.py b/youtube_dl/extractor/egghead.py
index c86f52319a46464dc9c2f329ceee6e19f8124c7e..e4a3046af573fec2961ae84a2137a9395bf68f0b 100644 (file)
--- a/youtube_dl/extractor/egghead.py
+++ b/youtube_dl/extractor/egghead.py
@@ -2,6 +2,11 @@
 from __future__ import unicode_literals
 
 from .common import InfoExtractor
+from ..utils import (
+    int_or_none,
+    try_get,
+    unified_timestamp,
+)
 
 
 class EggheadCourseIE(InfoExtractor):
@@ -33,3 +38,47 @@ class EggheadCourseIE(InfoExtractor):
         return self.playlist_result(
             entries, playlist_id, course.get('title'),
             course.get('description'))
+
+
+class EggheadLessonIE(InfoExtractor):
+    IE_DESC = 'egghead.io lesson'
+    IE_NAME = 'egghead:lesson'
+    _VALID_URL = r'https://egghead\.io/lessons/(?P<id>[^/?#&]+)'
+    _TEST = {
+        'url': 'https://egghead.io/lessons/javascript-linear-data-flow-with-container-style-types-box',
+        'info_dict': {
+            'id': 'fv5yotjxcg',
+            'ext': 'mp4',
+            'title': 'Create linear data flow with container style types (Box)',
+            'description': 'md5:9aa2cdb6f9878ed4c39ec09e85a8150e',
+            'thumbnail': r're:^https?:.*\.jpg$',
+            'timestamp': 1481296768,
+            'upload_date': '20161209',
+            'duration': 304,
+            'view_count': 0,
+            'tags': ['javascript', 'free'],
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }
+
+    def _real_extract(self, url):
+        lesson_id = self._match_id(url)
+
+        lesson = self._download_json(
+            'https://egghead.io/api/v1/lessons/%s' % lesson_id, lesson_id)
+
+        return {
+            '_type': 'url_transparent',
+            'ie_key': 'Wistia',
+            'url': 'wistia:%s' % lesson['wistia_id'],
+            'id': lesson['wistia_id'],
+            'title': lesson.get('title'),
+            'description': lesson.get('summary'),
+            'thumbnail': lesson.get('thumb_nail'),
+            'timestamp': unified_timestamp(lesson.get('published_at')),
+            'duration': int_or_none(lesson.get('duration')),
+            'view_count': int_or_none(lesson.get('plays_count')),
+            'tags': try_get(lesson, lambda x: x['tag_list'], list),
+        }