youtube-dl

commit de691a498dd334ee5f8d237a4fc7ced314d86f44
parent 2e6e742c3cda5ed4846bff4ef894aac21434e3d4
Author: Sergey M․ <dstftw@gmail.com>
Date:   Mon, 25 Jan 2016 22:18:34 +0600

[facebook:post] Add extractor (Closes #8321)

Diffstat:
M
youtube_dl/extractor/__init__.py
 | 
5
++++-
M
youtube_dl/extractor/facebook.py
 | 
30
++++++++++++++++++++++++++++++

2 files changed, 34 insertions(+), 1 deletion(-)
diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
@@ -196,7 +196,10 @@ from .everyonesmixtape import EveryonesMixtapeIE
 from .exfm import ExfmIE
 from .expotv import ExpoTVIE
 from .extremetube import ExtremeTubeIE
-from .facebook import FacebookIE
+from .facebook import (
+    FacebookIE,
+    FacebookPostIE,
+)
 from .faz import FazIE
 from .fc2 import FC2IE
 from .fczenit import FczenitIE
diff --git a/youtube_dl/extractor/facebook.py b/youtube_dl/extractor/facebook.py
@@ -199,3 +199,33 @@ class FacebookIE(InfoExtractor):
             'formats': formats,
             'uploader': uploader,
         }
+
+
+class FacebookPostIE(InfoExtractor):
+    IE_NAME = 'facebook:post'
+    _VALID_URL = r'https?://(?:\w+\.)?facebook\.com/[^/]+/posts/(?P<id>\d+)'
+    _TEST = {
+        'url': 'https://www.facebook.com/maxlayn/posts/10153807558977570',
+        'md5': '037b1fa7f3c2d02b7a0d7bc16031ecc6',
+        'info_dict': {
+            'id': '544765982287235',
+            'ext': 'mp4',
+            'title': '"What are you doing running in the snow?"',
+            'uploader': 'FailArmy',
+        }
+    }
+
+    def _real_extract(self, url):
+        post_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, post_id)
+
+        entries = [
+            self.url_result('facebook:%s' % video_id, FacebookIE.ie_key())
+            for video_id in self._parse_json(
+                self._search_regex(
+                    r'(["\'])video_ids\1\s*:\s*(?P<ids>\[.+?\])',
+                    webpage, 'video ids', group='ids'),
+                post_id)]
+
+        return self.playlist_result(entries, post_id)