[facebook] Add support for DASH manifests

[youtube-dl] / youtube_dl / extractor / daum.py

diff --git a/youtube_dl/extractor/daum.py b/youtube_dl/extractor/daum.py
index 567abdc3904c1f4317c557ef4f07415f1dee89b4..9bc345f60766617aa5877b32f2e092b6eedb4c5f 100644 (file)
--- a/youtube_dl/extractor/daum.py
+++ b/youtube_dl/extractor/daum.py
@@ -5,7 +5,10 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..compat import compat_urllib_parse
+from ..compat import (
+    compat_urllib_parse,
+    compat_urllib_parse_unquote,
+)
 from ..utils import (
     int_or_none,
     str_to_int,
@@ -14,7 +17,7 @@ from ..utils import (
 
 
 class DaumIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/v/(?P<id>[^?#&]+)'
+    _VALID_URL = r'https?://(?:(?:m\.)?tvpot\.daum\.net/v/|videofarm\.daum\.net/controller/player/VodPlayer\.swf\?vid=)(?P<id>[^?#&]+)'
     IE_NAME = 'daum.net'
 
     _TESTS = [{
@@ -25,6 +28,7 @@ class DaumIE(InfoExtractor):
             'title': '마크 헌트 vs 안토니오 실바',
             'description': 'Mark Hunt vs Antonio Silva',
             'upload_date': '20131217',
+            'thumbnail': 're:^https?://.*\.(?:jpg|png)',
             'duration': 2117,
             'view_count': int,
             'comment_count': int,
@@ -37,6 +41,7 @@ class DaumIE(InfoExtractor):
             'title': 'md5:a100d65d09cec246d8aa9bde7de45aed',
             'description': 'md5:79794514261164ff27e36a21ad229fc5',
             'upload_date': '20150604',
+            'thumbnail': 're:^https?://.*\.(?:jpg|png)',
             'duration': 154,
             'view_count': int,
             'comment_count': int,
@@ -44,14 +49,24 @@ class DaumIE(InfoExtractor):
     }, {
         'url': 'http://tvpot.daum.net/v/07dXWRka62Y%24',
         'only_matching': True,
+    }, {
+        'url': 'http://videofarm.daum.net/controller/player/VodPlayer.swf?vid=vwIpVpCQsT8%24&ref=',
+        'info_dict': {
+            'id': 'vwIpVpCQsT8$',
+            'ext': 'flv',
+            'title': '01-Korean War ( Trouble on the horizon )',
+            'description': '\nKorean War 01\nTrouble on the horizon\n전쟁의 먹구름',
+            'upload_date': '20080223',
+            'thumbnail': 're:^https?://.*\.(?:jpg|png)',
+            'duration': 249,
+            'view_count': int,
+            'comment_count': int,
+        },
     }]
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
+        video_id = compat_urllib_parse_unquote(self._match_id(url))
         query = compat_urllib_parse.urlencode({'vid': video_id})
-        info = self._download_xml(
-            'http://tvpot.daum.net/clip/ClipInfoXml.do?' + query, video_id,
-            'Downloading video info')
         movie_data = self._download_json(
             'http://videofarm.daum.net/controller/api/closed/v1_2/IntegratedMovieData.json?' + query,
             video_id, 'Downloading video formats info')
@@ -60,6 +75,10 @@ class DaumIE(InfoExtractor):
         if not movie_data.get('output_list', {}).get('output_list') and re.match(r'^\d+$', video_id):
             return self.url_result('http://tvpot.daum.net/clip/ClipView.do?clipid=%s' % video_id)
 
+        info = self._download_xml(
+            'http://tvpot.daum.net/clip/ClipInfoXml.do?' + query, video_id,
+            'Downloading video info')
+
         formats = []
         for format_el in movie_data['output_list']['output_list']:
             profile = format_el['profile']
@@ -105,6 +124,7 @@ class DaumClipIE(InfoExtractor):
             'title': 'DOTA 2GETHER 시즌2 6회 - 2부',
             'description': 'DOTA 2GETHER 시즌2 6회 - 2부',
             'upload_date': '20130831',
+            'thumbnail': 're:^https?://.*\.(?:jpg|png)',
             'duration': 3868,
             'view_count': int,
         },