Merge pull request #3 from unit193/master

[youtubedl] / youtube_dl / extractor / tenplay.py

diff --git a/youtube_dl/extractor/tenplay.py b/youtube_dl/extractor/tenplay.py
index f6694149b8e3509b4446458300824a3f3d5fc5de..dff44a4e2f4f5e4105def539c186dc630b411b34 100644 (file)
--- a/youtube_dl/extractor/tenplay.py
+++ b/youtube_dl/extractor/tenplay.py
@@ -3,88 +3,53 @@ from __future__ import unicode_literals
 
 from .common import InfoExtractor
 from ..utils import (
 
 from .common import InfoExtractor
 from ..utils import (

-    int_or_none,
-    float_or_none,
+    parse_age_limit,
+    parse_iso8601,
+    smuggle_url,

 )
 
 
 class TenPlayIE(InfoExtractor):
 )
 
 
 class TenPlayIE(InfoExtractor):

-    _VALID_URL = r'https?://(?:www\.)?ten(play)?\.com\.au/.+'
+    _VALID_URL = r'https?://(?:www\.)?10play\.com\.au/[^/]+/episodes/[^/]+/[^/]+/(?P<id>tpv\d{6}[a-z]{5})'

     _TEST = {
     _TEST = {

-        'url': 'http://tenplay.com.au/ten-insider/extra/season-2013/tenplay-tv-your-way',
+        'url': 'https://10play.com.au/masterchef/episodes/season-1/masterchef-s1-ep-1/tpv190718kwzga',

         'info_dict': {
         'info_dict': {

-            'id': '2695695426001',
-            'ext': 'flv',
-            'title': 'TENplay: TV your way',
-            'description': 'Welcome to a new TV experience. Enjoy a taste of the TENplay benefits.',
-            'timestamp': 1380150606.889,
-            'upload_date': '20130925',
-            'uploader': 'TENplay',
+            'id': '6060533435001',
+            'ext': 'mp4',
+            'title': 'MasterChef - S1 Ep. 1',
+            'description': 'md5:4fe7b78e28af8f2d900cd20d900ef95c',
+            'age_limit': 10,
+            'timestamp': 1240828200,
+            'upload_date': '20090427',
+            'uploader_id': '2199827728001',

         },
         'params': {
         },
         'params': {

-            'skip_download': True,  # Requires rtmpdump
+            'format': 'bestvideo',
+            'skip_download': True,

         }
     }
         }
     }

-
-    _video_fields = [
-        "id", "name", "shortDescription", "longDescription", "creationDate",
-        "publishedDate", "lastModifiedDate", "customFields", "videoStillURL",
-        "thumbnailURL", "referenceId", "length", "playsTotal",
-        "playsTrailingWeek", "renditions", "captioning", "startDate", "endDate"]
+    BRIGHTCOVE_URL_TEMPLATE = 'https://players.brightcove.net/2199827728001/cN6vRtRQt_default/index.html?videoId=%s'

 
     def _real_extract(self, url):
 
     def _real_extract(self, url):

-        webpage = self._download_webpage(url, url)
-        video_id = self._html_search_regex(
-            r'videoID: "(\d+?)"', webpage, 'video_id')
-        api_token = self._html_search_regex(
-            r'apiToken: "([a-zA-Z0-9-_\.]+?)"', webpage, 'api_token')
-        title = self._html_search_regex(
-            r'<meta property="og:title" content="\s*(.*?)\s*"\s*/?\s*>',
-            webpage, 'title')
-
-        json = self._download_json('https://api.brightcove.com/services/library?command=find_video_by_id&video_id=%s&token=%s&video_fields=%s' % (video_id, api_token, ','.join(self._video_fields)), title)
-
-        formats = []
-        for rendition in json['renditions']:
-            url = rendition['remoteUrl'] or rendition['url']
-            protocol = 'rtmp' if url.startswith('rtmp') else 'http'
-            ext = 'flv' if protocol == 'rtmp' else rendition['videoContainer'].lower()
-
-            if protocol == 'rtmp':
-                url = url.replace('&mp4:', '')
-
-                tbr = int_or_none(rendition.get('encodingRate'), 1000)
-
-            formats.append({
-                'format_id': '_'.join(
-                    ['rtmp', rendition['videoContainer'].lower(),
-                     rendition['videoCodec'].lower(), '%sk' % tbr]),
-                'width': int_or_none(rendition['frameWidth']),
-                'height': int_or_none(rendition['frameHeight']),
-                'tbr': tbr,
-                'filesize': int_or_none(rendition['size']),
-                'protocol': protocol,
-                'ext': ext,
-                'vcodec': rendition['videoCodec'].lower(),
-                'container': rendition['videoContainer'].lower(),
-                'url': url,
-            })
-        self._sort_formats(formats)
+        content_id = self._match_id(url)
+        data = self._download_json(
+            'https://10play.com.au/api/video/' + content_id, content_id)
+        video = data.get('video') or {}
+        metadata = data.get('metaData') or {}
+        brightcove_id = video.get('videoId') or metadata['showContentVideoId']
+        brightcove_url = smuggle_url(
+            self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
+            {'geo_countries': ['AU']})

 
         return {
 
         return {

-            'id': video_id,
-            'display_id': json['referenceId'],
-            'title': json['name'],
-            'description': json['shortDescription'] or json['longDescription'],
-            'formats': formats,
-            'thumbnails': [{
-                'url': json['videoStillURL']
-            }, {
-                'url': json['thumbnailURL']
-            }],
-            'thumbnail': json['videoStillURL'],
-            'duration': float_or_none(json.get('length'), 1000),
-            'timestamp': float_or_none(json.get('creationDate'), 1000),
-            'uploader': json.get('customFields', {}).get('production_company_distributor') or 'TENplay',
-            'view_count': int_or_none(json.get('playsTotal')),
+            '_type': 'url_transparent',
+            'url': brightcove_url,
+            'id': content_id,
+            'title': video.get('title') or metadata.get('pageContentName') or metadata.get('showContentName'),
+            'description': video.get('description'),
+            'age_limit': parse_age_limit(video.get('showRatingClassification') or metadata.get('showProgramClassification')),
+            'series': metadata.get('showName'),
+            'season': metadata.get('showContentSeason'),
+            'timestamp': parse_iso8601(metadata.get('contentPublishDate') or metadata.get('pageContentPublishDate')),
+            'ie_key': 'BrightcoveNew',

         }
         }