[youtube] Fix extraction.

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

diff --git a/youtube_dl/extractor/piksel.py b/youtube_dl/extractor/piksel.py
index d44edcdfba93a2f05c7bbb2818b9f91da56a66ce..88b6859b01a7c51eebe9f129d759f68005c75ce6 100644 (file)
--- a/youtube_dl/extractor/piksel.py
+++ b/youtube_dl/extractor/piksel.py
@@ -15,19 +15,38 @@ from ..utils import (
 
 
 class PikselIE(InfoExtractor):
-    _VALID_URL = r'https?://player\.piksel\.com/v/(?P<id>[a-z0-9]+)'
-    _TEST = {
-        'url': 'http://player.piksel.com/v/nv60p12f',
-        'md5': 'd9c17bbe9c3386344f9cfd32fad8d235',
-        'info_dict': {
-            'id': 'nv60p12f',
-            'ext': 'mp4',
-            'title': 'فن الحياة  - الحلقة 1',
-            'description': 'احدث برامج الداعية الاسلامي " مصطفي حسني " فى رمضان 2016علي النهار نور',
-            'timestamp': 1465231790,
-            'upload_date': '20160606',
+    _VALID_URL = r'https?://player\.piksel\.com/v/(?:refid/[^/]+/prefid/)?(?P<id>[a-z0-9_]+)'
+    _TESTS = [
+        {
+            'url': 'http://player.piksel.com/v/ums2867l',
+            'md5': '34e34c8d89dc2559976a6079db531e85',
+            'info_dict': {
+                'id': 'ums2867l',
+                'ext': 'mp4',
+                'title': 'GX-005 with Caption',
+                'timestamp': 1481335659,
+                'upload_date': '20161210'
+            }
+        },
+        {
+            # Original source: http://www.uscourts.gov/cameras-courts/state-washington-vs-donald-j-trump-et-al
+            'url': 'https://player.piksel.com/v/v80kqp41',
+            'md5': '753ddcd8cc8e4fa2dda4b7be0e77744d',
+            'info_dict': {
+                'id': 'v80kqp41',
+                'ext': 'mp4',
+                'title': 'WAW- State of Washington vs. Donald J. Trump, et al',
+                'description': 'State of Washington vs. Donald J. Trump, et al, Case Number 17-CV-00141-JLR, TRO Hearing, Civil Rights Case, 02/3/2017, 1:00 PM (PST), Seattle Federal Courthouse, Seattle, WA, Judge James L. Robart presiding.',
+                'timestamp': 1486171129,
+                'upload_date': '20170204'
+            }
+        },
+        {
+            # https://www3.nhk.or.jp/nhkworld/en/ondemand/video/2019240/
+            'url': 'http://player.piksel.com/v/refid/nhkworld/prefid/nw_vod_v_en_2019_240_20190823233000_02_1566873477',
+            'only_matching': True,
         }
-    }
+    ]
 
     @staticmethod
     def _extract_url(webpage):
@@ -38,10 +57,15 @@ class PikselIE(InfoExtractor):
             return mobj.group('url')
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
-        webpage = self._download_webpage(url, video_id)
-        app_token = self._search_regex(
-            r'clientAPI\s*:\s*"([^"]+)"', webpage, 'app token')
+        display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)
+        video_id = self._search_regex(
+            r'data-de-program-uuid=[\'"]([a-z0-9]+)',
+            webpage, 'program uuid', default=display_id)
+        app_token = self._search_regex([
+            r'clientAPI\s*:\s*"([^"]+)"',
+            r'data-de-api-key\s*=\s*"([^"]+)"'
+        ], webpage, 'app token')
         response = self._download_json(
             'http://player.piksel.com/ws/ws_program/api/%s/mode/json/apiv/5' % app_token,
             video_id, query={
@@ -96,6 +120,13 @@ class PikselIE(InfoExtractor):
             })
         self._sort_formats(formats)
 
+        subtitles = {}
+        for caption in video_data.get('captions', []):
+            caption_url = caption.get('url')
+            if caption_url:
+                subtitles.setdefault(caption.get('locale', 'en'), []).append({
+                    'url': caption_url})
+
         return {
             'id': video_id,
             'title': title,
@@ -103,4 +134,5 @@ class PikselIE(InfoExtractor):
             'thumbnail': video_data.get('thumbnailUrl'),
             'timestamp': parse_iso8601(video_data.get('dateadd')),
             'formats': formats,
+            'subtitles': subtitles,
         }