[soundcloud] fix download url extraction(closes #24394)

youtube_dl/extractor/soundcloud.py

@@ -27,6 +27,7 @@
     unified_timestamp,
     update_url_query,
     url_or_none,
+    urlhandle_detect_ext,
 )
 
 
@@ -96,7 +97,7 @@ class SoundcloudIE(InfoExtractor):
                 'repost_count': int,
             }
         },
-        # not streamable song, preview
+        # geo-restricted
         {
             'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep',
             'info_dict': {
@@ -108,17 +109,13 @@ class SoundcloudIE(InfoExtractor):
                 'uploader_id': '9615865',
                 'timestamp': 1337635207,
                 'upload_date': '20120521',
-                'duration': 30,
+                'duration': 227.155,
                 'license': 'all-rights-reserved',
                 'view_count': int,
                 'like_count': int,
                 'comment_count': int,
                 'repost_count': int,
             },
-            'params': {
-                # rtmp
-                'skip_download': True,
-            },
         },
         # private link
         {
@@ -229,7 +226,6 @@ class SoundcloudIE(InfoExtractor):
                 'skip_download': True,
             },
         },
-        # not available via api.soundcloud.com/i1/tracks/id/streams
         {
             'url': 'https://soundcloud.com/giovannisarani/mezzo-valzer',
             'md5': 'e22aecd2bc88e0e4e432d7dcc0a1abf7',
@@ -250,11 +246,9 @@ class SoundcloudIE(InfoExtractor):
                 'comment_count': int,
                 'repost_count': int,
             },
-            'expected_warnings': ['Unable to download JSON metadata'],
         }
     ]
 
-    _API_BASE = 'https://api.soundcloud.com/'
     _API_V2_BASE = 'https://api-v2.soundcloud.com/'
     _BASE_URL = 'https://soundcloud.com/'
     _IMAGE_REPL_RE = r'-([0-9a-z]+)\.jpg'
@@ -316,10 +310,9 @@ def _real_initialize(self):
     def _resolv_url(cls, url):
         return SoundcloudIE._API_V2_BASE + 'resolve?url=' + url
 
-    def _extract_info_dict(self, info, full_title=None, secret_token=None, version=2):
+    def _extract_info_dict(self, info, full_title=None, secret_token=None):
         track_id = compat_str(info['id'])
         title = info['title']
-        track_base_url = self._API_BASE + 'tracks/%s' % track_id
 
         format_urls = set()
         formats = []
@@ -328,21 +321,22 @@ def _extract_info_dict(self, info, full_title=None, secret_token=None, version=2
             query['secret_token'] = secret_token
 
         if info.get('downloadable') and info.get('has_downloads_left'):
-            format_url = update_url_query(
+            download_url = update_url_query(
-                info.get('download_url') or track_base_url + '/download', query)
+                self._API_V2_BASE + 'tracks/' + track_id + '/download', query)
-            format_urls.add(format_url)
+            redirect_url = (self._download_json(download_url, track_id, fatal=False) or {}).get('redirectUri')
-            if version == 2:
+            if redirect_url:
-                v1_info = self._download_json(
+                urlh = self._request_webpage(
-                    track_base_url, track_id, query=query, fatal=False) or {}
+                    HEADRequest(redirect_url), track_id, fatal=False)
-            else:
+                if urlh:
-                v1_info = info
+                    format_url = urlh.geturl()
-            formats.append({
+                    format_urls.add(format_url)
-                'format_id': 'download',
+                    formats.append({
-                'ext': v1_info.get('original_format') or 'mp3',
+                        'format_id': 'download',
-                'filesize': int_or_none(v1_info.get('original_content_size')),
+                        'ext': urlhandle_detect_ext(urlh) or 'mp3',
-                'url': format_url,
+                        'filesize': int_or_none(urlh.headers.get('Content-Length')),
-                'preference': 10,
+                        'url': format_url,
-            })
+                        'preference': 10,
+                    })
 
         def invalid_url(url):
             return not url or url in format_urls
@@ -406,42 +400,11 @@ def add_format(f, protocol, is_preview=False):
             }, 'http' if protocol == 'progressive' else protocol,
                 t.get('snipped') or '/preview/' in format_url)
 
-        if not formats:
-            # Old API, does not work for some tracks (e.g.
-            # https://soundcloud.com/giovannisarani/mezzo-valzer)
-            # and might serve preview URLs (e.g.
-            # http://www.soundcloud.com/snbrn/ele)
-            format_dict = self._download_json(
-                track_base_url + '/streams', track_id,
-                'Downloading track url', query=query, fatal=False) or {}
-
-            for key, stream_url in format_dict.items():
-                if invalid_url(stream_url):
-                    continue
-                format_urls.add(stream_url)
-                mobj = re.search(r'(http|hls)_([^_]+)_(\d+)_url', key)
-                if mobj:
-                    protocol, ext, abr = mobj.groups()
-                    add_format({
-                        'abr': abr,
-                        'ext': ext,
-                        'url': stream_url,
-                    }, protocol)
-
-        if not formats:
-            # We fallback to the stream_url in the original info, this
-            # cannot be always used, sometimes it can give an HTTP 404 error
-            urlh = self._request_webpage(
-                HEADRequest(info.get('stream_url') or track_base_url + '/stream'),
-                track_id, query=query, fatal=False)
-            if urlh:
-                stream_url = urlh.geturl()
-                if not invalid_url(stream_url):
-                    add_format({'url': stream_url}, 'http')
-
         for f in formats:
             f['vcodec'] = 'none'
 
+        if not formats and info.get('policy') == 'BLOCK':
+            self.raise_geo_restricted()
         self._sort_formats(formats)
 
         user = info.get('user') or {}
@@ -511,16 +474,10 @@ def _real_extract(self, url):
                 resolve_title += '/%s' % token
             info_json_url = self._resolv_url(self._BASE_URL + resolve_title)
 
-        version = 2
         info = self._download_json(
-            info_json_url, full_title, 'Downloading info JSON', query=query, fatal=False)
+            info_json_url, full_title, 'Downloading info JSON', query=query)
-        if not info:
-            info = self._download_json(
-                info_json_url.replace(self._API_V2_BASE, self._API_BASE),
-                full_title, 'Downloading info JSON', query=query)
-            version = 1
 
-        return self._extract_info_dict(info, full_title, token, version)
+        return self._extract_info_dict(info, full_title, token)
 
 
 class SoundcloudPlaylistBaseIE(SoundcloudIE):