Updated version 2017.10.15.1 from 'upstream/2017.10.15.1'

[youtubedl] / youtube_dl / extractor / karaoketv.py

diff --git a/youtube_dl/extractor/karaoketv.py b/youtube_dl/extractor/karaoketv.py
index e3b43ff8dbfec5e065aa069a0fb140dcfb5822c9..bfccf89b0fda0be1100764290681a53e022947e0 100644 (file)
--- a/youtube_dl/extractor/karaoketv.py
+++ b/youtube_dl/extractor/karaoketv.py
@@ -2,39 +2,63 @@
 from __future__ import unicode_literals
 
 from .common import InfoExtractor
 from __future__ import unicode_literals
 
 from .common import InfoExtractor

-from ..compat import compat_urllib_parse
-from ..utils import (
-    js_to_json,
-)

 
 
 class KaraoketvIE(InfoExtractor):
 
 
 class KaraoketvIE(InfoExtractor):

-    _VALID_URL = r'http://karaoketv\.co\.il/\?container=songs&id=(?P<id>[0-9]+)'
+    _VALID_URL = r'https?://(?:www\.)?karaoketv\.co\.il/[^/]+/(?P<id>\d+)'

     _TEST = {
     _TEST = {

-        'url': 'http://karaoketv.co.il/?container=songs&id=171568',
+        'url': 'http://www.karaoketv.co.il/%D7%A9%D7%99%D7%A8%D7%99_%D7%A7%D7%A8%D7%99%D7%95%D7%A7%D7%99/58356/%D7%90%D7%99%D7%96%D7%95%D7%9F',

         'info_dict': {
         'info_dict': {

-            'id': '171568',
-            'ext': 'mp4',
-            'title': 'אל העולם שלך - רותם כהן - שרים קריוקי',
+            'id': '58356',
+            'ext': 'flv',
+            'title': 'קריוקי של איזון',
+        },
+        'params': {
+            # rtmp download
+            'skip_download': True,

         }
     }
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
         }
     }
 
     def _real_extract(self, url):
         video_id = self._match_id(url)

+

         webpage = self._download_webpage(url, video_id)
         webpage = self._download_webpage(url, video_id)

+        api_page_url = self._search_regex(
+            r'<iframe[^>]+src=(["\'])(?P<url>https?://www\.karaoke\.co\.il/api_play\.php\?.+?)\1',
+            webpage, 'API play URL', group='url')
+
+        api_page = self._download_webpage(api_page_url, video_id)
+        video_cdn_url = self._search_regex(
+            r'<iframe[^>]+src=(["\'])(?P<url>https?://www\.video-cdn\.com/embed/iframe/.+?)\1',
+            api_page, 'video cdn URL', group='url')
+
+        video_cdn = self._download_webpage(video_cdn_url, video_id)
+        play_path = self._parse_json(
+            self._search_regex(
+                r'var\s+options\s*=\s*({.+?});', video_cdn, 'options'),
+            video_id)['clip']['url']

 
 

-        page_video_url = self._og_search_video_url(webpage, video_id)
-        config_json = compat_urllib_parse.unquote_plus(self._search_regex(
-            r'config=(.*)', page_video_url, 'configuration'))
+        settings = self._parse_json(
+            self._search_regex(
+                r'var\s+settings\s*=\s*({.+?});', video_cdn, 'servers', default='{}'),
+            video_id, fatal=False) or {}

 
 

-        urls_info_json = self._download_json(
-            config_json, video_id, 'Downloading configuration',
-            transform_source=js_to_json)
+        servers = settings.get('servers')
+        if not servers or not isinstance(servers, list):
+            servers = ('wowzail.video-cdn.com:80/vodcdn', )

 
 

-        url = urls_info_json['playlist'][0]['url']
+        formats = [{
+            'url': 'rtmp://%s' % server if not server.startswith('rtmp') else server,
+            'play_path': play_path,
+            'app': 'vodcdn',
+            'page_url': video_cdn_url,
+            'player_url': 'http://www.video-cdn.com/assets/flowplayer/flowplayer.commercial-3.2.18.swf',
+            'rtmp_real_time': True,
+            'ext': 'flv',
+        } for server in servers]

 
         return {
             'id': video_id,
             'title': self._og_search_title(webpage),
 
         return {
             'id': video_id,
             'title': self._og_search_title(webpage),

-            'url': url,
+            'formats': formats,

         }
         }