Delay initialization of InfoExtractors until they are needed

youtube_dl/YoutubeDL.py

@@ -82,7 +82,7 @@
     YoutubeDLHandler,
 )
 from .cache import Cache
-from .extractor import get_info_extractor, gen_extractors
+from .extractor import get_info_extractor, gen_extractor_classes
 from .downloader import get_suitable_downloader
 from .downloader.rtmp import rtmpdump_version
 from .postprocessor import (
@@ -378,8 +378,9 @@ def warn_if_short_id(self, argv):
     def add_info_extractor(self, ie):
         """Add an InfoExtractor object to the end of the list."""
         self._ies.append(ie)
-        self._ies_instances[ie.ie_key()] = ie
-        ie.set_downloader(self)
+        if not isinstance(ie, type):
+            self._ies_instances[ie.ie_key()] = ie
+            ie.set_downloader(self)
 
     def get_info_extractor(self, ie_key):
         """
@@ -397,7 +398,7 @@ def add_default_info_extractors(self):
         """
         Add the InfoExtractors returned by gen_extractors to the end of the list
         """
-        for ie in gen_extractors():
+        for ie in gen_extractor_classes():
             self.add_info_extractor(ie)
 
     def add_post_processor(self, pp):
@@ -661,6 +662,7 @@ def extract_info(self, url, download=True, ie_key=None, extra_info={},
             if not ie.suitable(url):
                 continue
 
+            ie = self.get_info_extractor(ie.ie_key())
             if not ie.working():
                 self.report_warning('The program functionality for this site has been marked as broken, '
                                     'and will probably not work.')

youtube_dl/extractor/__init__.py

@@ -997,11 +997,18 @@
 _ALL_CLASSES.append(GenericIE)
 
 
+def gen_extractor_classes():
+    """ Return a list of supported extractors.
+    The order does matter; the first extractor matched is the one handling the URL.
+    """
+    return _ALL_CLASSES
+
+
 def gen_extractors():
     """ Return a list of an instance of every supported extractor.
     The order does matter; the first extractor matched is the one handling the URL.
     """
-    return [klass() for klass in _ALL_CLASSES]
+    return [klass() for klass in gen_extractor_classes()]
 
 
 def list_extractors(age_limit):