[cjsw] Add extractor
authorChristopher Smith <ctsmi@ucalgary.ca>
Thu, 29 Jun 2017 19:10:45 +0000 (13:10 -0600)
committerSergey M․ <dstftw@gmail.com>
Sun, 9 Jul 2017 10:01:00 +0000 (17:01 +0700)
youtube_dl/extractor/cjsw.py [new file with mode: 0644]
youtube_dl/extractor/extractors.py

diff --git a/youtube_dl/extractor/cjsw.py b/youtube_dl/extractor/cjsw.py
new file mode 100644 (file)
index 0000000..087cac9
--- /dev/null
@@ -0,0 +1,41 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+
+
+class CJSWIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?cjsw\.com/program/\S+/(?P<id>[0-9]+)'
+    IE_NAME = 'cjsw'
+    _TEST = {
+        'url': 'http://cjsw.com/program/freshly-squeezed/episode/20170620',
+        'md5': 'cee14d40f1e9433632c56e3d14977120',
+        'info_dict': {
+            'id': '20170620',
+            'ext': 'mp3',
+            'title': 'Freshly Squeezed',
+            'description': 'Sled Island artists featured // Live session with Phi Pho, followed by a live session with Sinzere & The Late Nights! // Stay Fresh Y\'all!!',
+        }
+    }
+
+    def _real_extract(self, url):
+        episode_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, episode_id)
+
+        title = self._search_regex(
+            r'<button[^>]+data-showname=(["\'])(?P<title>(?!\1).+?)\1[^>]*>', webpage, 'title', group='title')
+        description = self._html_search_regex(
+            r'<p>(?P<description>.+?)</p>', webpage, 'description', fatal=False)
+        formats = [{
+            'url': self._search_regex(
+                r'<button[^>]+data-audio-src=(["\'])(?P<audio_url>(?!\1).+?)\1[^>]*>', webpage, 'audio_url', group='audio_url'),
+            'ext': 'mp3',
+            'vcodec': 'none',
+        }]
+        return {
+            'id': episode_id,
+            'title': title,
+            'description': description,
+            'formats': formats,
+        }
index b83c3aba596c0390a302504221125224e8d66006..4524fa68766721e24de0da1a669e4a391bd4855e 100644 (file)
@@ -185,6 +185,7 @@ from .chirbit import (
     ChirbitProfileIE,
 )
 from .cinchcast import CinchcastIE
+from .cjsw import CJSWIE
 from .clipfish import ClipfishIE
 from .cliphunter import CliphunterIE
 from .cliprs import ClipRsIE