Update changelog

[youtubedl] / youtube_dl / extractor / liveleak.py

diff --git a/youtube_dl/extractor/liveleak.py b/youtube_dl/extractor/liveleak.py
index 35822067f908f0567e8dcb8c9c8265df4d3421c2..b84e4dd6c20415a1f8bf7a562a61f80bbdbd5786 100644 (file)
--- a/youtube_dl/extractor/liveleak.py
+++ b/youtube_dl/extractor/liveleak.py
@@ -17,7 +17,8 @@ class LiveLeakIE(InfoExtractor):
             'ext': 'flv',
             'description': 'extremely bad day for this guy..!',
             'uploader': 'ljfriel2',
-            'title': 'Most unlucky car accident'
+            'title': 'Most unlucky car accident',
+            'thumbnail': 're:^https?://.*\.jpg$'
         }
     }, {
         'url': 'http://www.liveleak.com/view?i=f93_1390833151',
@@ -28,6 +29,7 @@ class LiveLeakIE(InfoExtractor):
             'description': 'German Television Channel NDR does an exclusive interview with Edward Snowden.\r\nUploaded on LiveLeak cause German Television thinks the rest of the world isn\'t intereseted in Edward Snowden.',
             'uploader': 'ARD_Stinkt',
             'title': 'German Television does first Edward Snowden Interview (ENGLISH)',
+            'thumbnail': 're:^https?://.*\.jpg$'
         }
     }, {
         'url': 'http://www.liveleak.com/view?i=4f7_1392687779',
@@ -40,8 +42,44 @@ class LiveLeakIE(InfoExtractor):
             'title': 'Man is Fatally Struck by Reckless Car While Packing up a Moving Truck',
             'age_limit': 18,
         }
+    }, {
+        # Covers https://github.com/rg3/youtube-dl/pull/5983
+        'url': 'http://www.liveleak.com/view?i=801_1409392012',
+        'md5': '0b3bec2d888c20728ca2ad3642f0ef15',
+        'info_dict': {
+            'id': '801_1409392012',
+            'ext': 'mp4',
+            'description': 'Happened on 27.7.2014. \r\nAt 0:53 you can see people still swimming at near beach.',
+            'uploader': 'bony333',
+            'title': 'Crazy Hungarian tourist films close call waterspout in Croatia',
+            'thumbnail': 're:^https?://.*\.jpg$'
+        }
+    }, {
+        # Covers https://github.com/rg3/youtube-dl/pull/10664#issuecomment-247439521
+        'url': 'http://m.liveleak.com/view?i=763_1473349649',
+        'add_ie': ['Youtube'],
+        'info_dict': {
+            'id': '763_1473349649',
+            'ext': 'mp4',
+            'title': 'Reporters and public officials ignore epidemic of black on asian violence in Sacramento | Colin Flaherty',
+            'description': 'Colin being the warrior he is and showing the injustice Asians in Sacramento are being subjected to.',
+            'uploader': 'Ziz',
+            'upload_date': '20160908',
+            'uploader_id': 'UCEbta5E_jqlZmEJsriTEtnw'
+        },
+        'params': {
+            'skip_download': True,
+        },
     }]
 
+    @staticmethod
+    def _extract_url(webpage):
+        mobj = re.search(
+            r'<iframe[^>]+src="https?://(?:\w+\.)?liveleak\.com/ll_embed\?(?:.*?)i=(?P<id>[\w_]+)(?:.*)',
+            webpage)
+        if mobj:
+            return 'http://www.liveleak.com/view?i=%s' % mobj.group('id')
+
     def _real_extract(self, url):
         video_id = self._match_id(url)
         webpage = self._download_webpage(url, video_id)
@@ -53,6 +91,7 @@ class LiveLeakIE(InfoExtractor):
         age_limit = int_or_none(self._search_regex(
             r'you confirm that you are ([0-9]+) years and over.',
             webpage, 'age limit', default=None))
+        video_thumbnail = self._og_search_thumbnail(webpage)
 
         sources_raw = self._search_regex(
             r'(?s)sources:\s*(\[.*?\]),', webpage, 'video URLs', default=None)
@@ -64,7 +103,7 @@ class LiveLeakIE(InfoExtractor):
             else:
                 # Maybe an embed?
                 embed_url = self._search_regex(
-                    r'<iframe[^>]+src="(http://www.prochan.com/embed\?[^"]+)"',
+                    r'<iframe[^>]+src="(https?://(?:www\.)?(?:prochan|youtube)\.com/embed[^"]+)"',
                     webpage, 'embed URL')
                 return {
                     '_type': 'url_transparent',
@@ -84,8 +123,12 @@ class LiveLeakIE(InfoExtractor):
             'format_note': s.get('label'),
             'url': s['file'],
         } for i, s in enumerate(sources)]
+
         for i, s in enumerate(sources):
-            orig_url = s['file'].replace('.h264_base.mp4', '')
+            # Removing '.h264_*.mp4' gives the raw video, which is essentially
+            # the same video without the LiveLeak logo at the top (see
+            # https://github.com/rg3/youtube-dl/pull/4768)
+            orig_url = re.sub(r'\.h264_.+?\.mp4', '', s['file'])
             if s['file'] != orig_url:
                 formats.append({
                     'format_id': 'original-%s' % i,
@@ -102,4 +145,5 @@ class LiveLeakIE(InfoExtractor):
             'uploader': video_uploader,
             'formats': formats,
             'age_limit': age_limit,
+            'thumbnail': video_thumbnail,
         }