youtube-dl

commit 53a664edf4bf713df0159e604bbc131dde5ed1e6
parent 264e77c406a3b14f15aafcd036524cb6fe86aa20
Author: Sergey M․ <dstftw@gmail.com>
Date:   Sat, 24 Dec 2016 22:46:27 +0700

[brightcove:legacy] Improve embeds detection (closes #11523)

Diffstat:
M
youtube_dl/extractor/brightcove.py
 | 
13
++++++++-----
M
youtube_dl/extractor/generic.py
 | 
24
+++++++++++++++++++++---

2 files changed, 29 insertions(+), 8 deletions(-)
diff --git a/youtube_dl/extractor/brightcove.py b/youtube_dl/extractor/brightcove.py
@@ -232,13 +232,16 @@ class BrightcoveLegacyIE(InfoExtractor):
         """Return a list of all Brightcove URLs from the webpage """
 
         url_m = re.search(
-            r'<meta\s+property=[\'"]og:video[\'"]\s+content=[\'"](https?://(?:secure|c)\.brightcove.com/[^\'"]+)[\'"]',
-            webpage)
+            r'''(?x)
+                <meta\s+
+                    (?:property|itemprop)=([\'"])(?:og:video|embedURL)\1[^>]+
+                    content=([\'"])(?P<url>https?://(?:secure|c)\.brightcove.com/(?:(?!\2).)+)\2
+            ''', webpage)
         if url_m:
-            url = unescapeHTML(url_m.group(1))
+            url = unescapeHTML(url_m.group('url'))
             # Some sites don't add it, we can't download with this url, for example:
             # http://www.ktvu.com/videos/news/raw-video-caltrain-releases-video-of-man-almost/vCTZdY/
-            if 'playerKey' in url or 'videoId' in url:
+            if 'playerKey' in url or 'videoId' in url or 'idVideo' in url:
                 return [url]
 
         matches = re.findall(
@@ -259,7 +262,7 @@ class BrightcoveLegacyIE(InfoExtractor):
         url, smuggled_data = unsmuggle_url(url, {})
 
         # Change the 'videoId' and others field to '@videoPlayer'
-        url = re.sub(r'(?<=[?&])(videoI(d|D)|bctid)', '%40videoPlayer', url)
+        url = re.sub(r'(?<=[?&])(videoI(d|D)|idVideo|bctid)', '%40videoPlayer', url)
         # Change bckey (used by bcove.me urls) to playerKey
         url = re.sub(r'(?<=[?&])bckey', 'playerKey', url)
         mobj = re.match(self._VALID_URL, url)
diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
@@ -344,10 +344,10 @@ class GenericIE(InfoExtractor):
             },
             'skip': 'There is a limit of 200 free downloads / month for the test song',
         },
-        # embedded brightcove video
-        # it also tests brightcove videos that need to set the 'Referer' in the
-        # http requests
         {
+            # embedded brightcove video
+            # it also tests brightcove videos that need to set the 'Referer'
+            # in the http requests
             'add_ie': ['BrightcoveLegacy'],
             'url': 'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
             'info_dict': {
@@ -362,6 +362,24 @@ class GenericIE(InfoExtractor):
             },
         },
         {
+            # embedded with itemprop embedURL and video id spelled as `idVideo`
+            'add_id': ['BrightcoveLegacy'],
+            'url': 'http://bfmbusiness.bfmtv.com/mediaplayer/chroniques/olivier-delamarche/',
+            'info_dict': {
+                'id': '5255628253001',
+                'ext': 'mp4',
+                'title': 'md5:37c519b1128915607601e75a87995fc0',
+                'description': 'md5:37f7f888b434bb8f8cc8dbd4f7a4cf26',
+                'uploader': 'BFM BUSINESS',
+                'uploader_id': '876450612001',
+                'timestamp': 1482255315,
+                'upload_date': '20161220',
+            },
+            'params': {
+                'skip_download': True,
+            },
+        },
+        {
             # https://github.com/rg3/youtube-dl/issues/2253
             'url': 'http://bcove.me/i6nfkrc3',
             'md5': '0ba9446db037002366bab3b3eb30c88c',