[inc] Add extractor
authorDéstin Reed <TRox1972@noreply.github.com>
Sun, 8 Jan 2017 18:28:16 +0000 (19:28 +0100)
committerSergey M․ <dstftw@gmail.com>
Mon, 9 Jan 2017 15:57:14 +0000 (22:57 +0700)
youtube_dl/extractor/extractors.py
youtube_dl/extractor/inc.py [new file with mode: 0644]

index 2e80ef1524f4d120a82d8b4f28d5f92bd0f22683..5ba8efb0eaaeaf7915ff9e36ab03f24740625a4d 100644 (file)
@@ -395,6 +395,7 @@ from .imgur import (
     ImgurAlbumIE,
 )
 from .ina import InaIE
+from .inc import IncIE
 from .indavideo import (
     IndavideoIE,
     IndavideoEmbedIE,
diff --git a/youtube_dl/extractor/inc.py b/youtube_dl/extractor/inc.py
new file mode 100644 (file)
index 0000000..279e53c
--- /dev/null
@@ -0,0 +1,40 @@
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+
+
+class IncIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?inc\.com(?:/[\w-]+)+/(?P<id>[\w-]+)(?:\.html)?'
+    _TESTS = [{
+        'url': 'http://www.inc.com/tip-sheet/bill-gates-says-these-5-books-will-make-you-smarter.html',
+        'md5': '7416739c9c16438c09fa35619d6ba5cb',
+        'info_dict': {
+            'id': '1_wqig47aq',
+            'ext': 'mov',
+            'title': 'Bill Gates Says These 5 Books Will Make You Smarter',
+            'description': 'md5:bea7ff6cce100886fc1995acb743237e',
+            'timestamp': 1474414430,
+            'upload_date': '20160920',
+            'uploader_id': 'video@inc.com',
+        },
+    }, {
+        'url': 'http://www.inc.com/video/david-whitford/founders-forum-tripadvisor-steve-kaufer-most-enjoyable-moment-for-entrepreneur.html',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)
+
+        partner_id = self._search_regex(
+            r'var\s+_bizo_data_partner_id\s*=\s*"(\d+)";',
+            webpage,
+            'partner id')
+
+        kaltura_id = self._parse_json(self._search_regex(
+            r'pageInfo\.videos\s*=\s*\[(.+)\];',
+            webpage,
+            'kaltura id'),
+            display_id)['vid_kaltura_id']
+
+        return self.url_result('kaltura:%s:%s' % (partner_id, kaltura_id), 'Kaltura')