[downloader] Improve downloader selection

youtube_dl/YoutubeDL.py

@@ -1179,7 +1179,7 @@ def process_info(self, info_dict):
         if not self.params.get('skip_download', False):
             try:
                 def dl(name, info):
-                    fd = get_suitable_downloader(info)(self, self.params)
+                    fd = get_suitable_downloader(info, self.params)(self, self.params)
                     for ph in self._progress_hooks:
                         fd.add_progress_hook(ph)
                     if self.params.get('verbose'):

youtube_dl/downloader/__init__.py

@@ -9,27 +9,26 @@
 from .f4m import F4mFD
 
 from ..utils import (
-    determine_ext,
+    determine_protocol,
 )
 
+PROTOCOL_MAP = {
+    'rtmp': RtmpFD,
+    'm3u8_native': NativeHlsFD,
+    'm3u8': HlsFD,
+    'mms': MplayerFD,
+    'rtsp': MplayerFD,
+    'f4m': F4mFD,
+}
 
-def get_suitable_downloader(info_dict):
+
+def get_suitable_downloader(info_dict, params={}):
     """Get the downloader class that can handle the info dict."""
-    url = info_dict['url']
-    protocol = info_dict.get('protocol')
+    protocol = determine_protocol(info_dict)
+    info_dict['protocol'] = protocol
+
+    return PROTOCOL_MAP.get(protocol, HttpFD)
 
-    if url.startswith('rtmp'):
-        return RtmpFD
-    if protocol == 'm3u8_native':
-        return NativeHlsFD
-    if (protocol == 'm3u8') or (protocol is None and determine_ext(url) == 'm3u8'):
-        return HlsFD
-    if url.startswith('mms') or url.startswith('rtsp'):
-        return MplayerFD
-    if determine_ext(url) == 'f4m':
-        return F4mFD
-    else:
-        return HttpFD
 
 __all__ = [
     'get_suitable_downloader',

youtube_dl/utils.py

@@ -1642,3 +1642,25 @@ def is_html(first_bytes):
         s = first_bytes.decode('utf-8', 'replace')
 
     return re.match(r'^\s*<', s)
+
+
+def determine_protocol(info_dict):
+    protocol = info_dict.get('protocol')
+    if protocol is not None:
+        return protocol
+
+    url = info_dict['url']
+    if url.startswith('rtmp'):
+        return 'rtmp'
+    elif url.startswith('mms'):
+        return 'mms'
+    elif url.startswith('rtsp'):
+        return 'rtsp'
+
+    ext = determine_ext(url)
+    if ext == 'm3u8':
+        return 'm3u8'
+    elif ext == 'f4m':
+        return 'f4m'
+
+    return compat_urllib_parse_urlparse(url).scheme