core: Show [extractor] and video ID in ERROR messages (Fixes #33254)

test/test_YoutubeDL.py

@@ -1187,5 +1187,111 @@ class TestYoutubeDLCookies(unittest.TestCase):
         self.assertFalse(ydl.cookiejar.get_cookie_header(fmt['url']), msg='Cookies set in cookiejar for wrong domain')
 
 
+class TestExtractorErrorFormatting(unittest.TestCase):
+    """Tests for Issue #33254: [extractor] and video ID in ERROR messages."""
+
+    def test_no_prefix_when_bare(self):
+        """Without ie/video_id the message is unchanged."""
+        e = ExtractorError('test error', expected=True)
+        self.assertEqual(e.orig_msg, 'test error')
+        self.assertIn('test error', str(e))
+        # No bracket prefix before the message
+        self.assertFalse(str(e).startswith('['))
+
+    def test_video_id_only_prefix(self):
+        """Providing video_id prefixes with 'id: msg'."""
+        e = ExtractorError('test error', expected=True, video_id='abc123')
+        self.assertTrue(str(e).startswith('abc123: test error'), str(e))
+        self.assertEqual(e.orig_msg, 'test error')
+        self.assertEqual(e.video_id, 'abc123')
+
+    def test_ie_only_prefix(self):
+        """Providing ie prefixes with '[extractor] msg'."""
+        e = ExtractorError('test error', expected=True, ie='youtube')
+        self.assertTrue(str(e).startswith('[youtube] test error'), str(e))
+        self.assertEqual(e.orig_msg, 'test error')
+        self.assertEqual(e.ie, 'youtube')
+
+    def test_full_prefix(self):
+        """Both ie and video_id produce '[extractor] id: msg'."""
+        e = ExtractorError(
+            'Premieres in 7 hours',
+            expected=True,
+            video_id='-Nb_os-NV5o',
+            ie='youtube',
+        )
+        self.assertEqual(str(e), '[youtube] -Nb_os-NV5o: Premieres in 7 hours')
+        self.assertEqual(e.orig_msg, 'Premieres in 7 hours')
+        self.assertEqual(e.video_id, '-Nb_os-NV5o')
+        self.assertEqual(e.ie, 'youtube')
+
+    def test_setattr_recomputes_msg(self):
+        """Assigning ie/video_id post-construction recomputes str(e)."""
+        e = ExtractorError('test error', expected=True)
+        self.assertNotIn('[youtube]', str(e))
+
+        e.ie = 'youtube'
+        self.assertIn('[youtube]', str(e))
+
+        e.video_id = 'abc123'
+        self.assertIn('abc123', str(e))
+        self.assertIn('[youtube]', str(e))
+        # orig_msg is always preserved
+        self.assertEqual(e.orig_msg, 'test error')
+
+    def test_generic_handler_backfills_ie_and_video_id(self):
+        """InfoExtractor.extract() injects ie/video_id on bare ExtractorError raises."""
+        from youtube_dl.extractor.common import InfoExtractor
+
+        class BareErrorIE(InfoExtractor):
+            IE_NAME = 'bare_test'
+            _VALID_URL = r'http://bare\.test/(?P<id>[^/]+)'
+
+            def _real_extract(self, url):
+                raise ExtractorError('bare error', expected=True)
+
+        ydl = FakeYDL()
+        ie = BareErrorIE(ydl)
+        try:
+            ie.extract('http://bare.test/myvideo')
+            self.fail('ExtractorError not raised')
+        except ExtractorError as e:
+            self.assertEqual(e.ie, 'bare_test')
+            # Falls back to sentinel since no video_id was set
+            self.assertEqual(e.video_id, '**???**')
+            self.assertIn('[bare_test]', str(e))
+
+    def test_raise_no_formats_injects_context(self):
+        """YoutubeDL.raise_no_formats passes extractor and video_id to ExtractorError."""
+        info = {
+            'id': 'testid',
+            'title': 'Test',
+            'extractor': 'testex',
+            'formats': [],
+        }
+        ydl = YDL()
+        try:
+            ydl.raise_no_formats(info, msg='requested format not available', expected=True)
+            self.fail('ExtractorError not raised')
+        except ExtractorError as e:
+            self.assertEqual(e.orig_msg, 'requested format not available')
+            self.assertEqual(e.video_id, 'testid')
+            self.assertEqual(e.ie, 'testex')
+            self.assertIn('[testex]', str(e))
+            self.assertIn('testid', str(e))
+
+    def test_raise_no_formats_missing_id_no_keyerror(self):
+        """raise_no_formats does not raise KeyError when 'id' is absent."""
+        info = {'extractor': 'testex'}  # deliberately missing 'id'
+        ydl = YDL()
+        try:
+            ydl.raise_no_formats(info, msg='Missing \"id\" field in extractor result', forced=True)
+            self.fail('ExtractorError not raised')
+        except ExtractorError as e:
+            self.assertEqual(e.orig_msg, 'Missing "id" field in extractor result')
+            # video_id is None, so it is omitted from the formatted string
+            self.assertIsNone(e.video_id)
+
+
 if __name__ == '__main__':
     unittest.main()

youtube_dl/extractor/common.py

@@ -579,7 +579,10 @@ class InfoExtractor(object):
                     if self.__maybe_fake_ip_and_retry(e.countries):
                         continue
                     raise
-        except ExtractorError:
+        except ExtractorError as e:
+            e.video_id = getattr(e, 'video_id', None) or '**???**'
+            e.ie = getattr(e, 'ie', None) or self.IE_NAME
+            e.traceback = e.traceback or sys.exc_info()[2]
             raise
         except compat_http_client.IncompleteRead as e:
             raise ExtractorError('A network error has occurred.', cause=e, expected=True)

youtube_dl/extractor/youtube.py

@@ -2804,7 +2804,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                     self.raise_login_required(remove_end(reason, 'This helps protect our community. Learn more'))
                 elif traverse_obj(playability_status, ('errorScreen', 'playerCaptchaViewModel', T(dict))):
                     reason += '. YouTube is requiring a captcha challenge before playback'
-                raise ExtractorError(reason, expected=True)
+                raise ExtractorError(reason, expected=True, video_id=video_id)
 
         self._sort_formats(formats)
 

youtube_dl/utils.py

@@ -2416,37 +2416,52 @@ def bug_reports_message(before=';'):
 
 class YoutubeDLError(Exception):
     """Base exception for YoutubeDL errors."""
-    pass
+    msg = None
+
+    def __init__(self, msg=None):
+        self.msg = type(self).__name__ if msg is None else msg
+        super(YoutubeDLError, self).__init__(self.msg)
 
 
 class ExtractorError(YoutubeDLError):
     """Error during info extraction."""
 
-    def __init__(self, msg, tb=None, expected=False, cause=None, video_id=None):
+    def __init__(self, msg, tb=None, expected=False, cause=None, video_id=None, ie=None):
         """ tb, if given, is the original traceback (so that it can be printed out).
         If expected is set, this is a normal error message and most likely not a bug in youtube-dl.
         """
         self.orig_msg = msg
         if sys.exc_info()[0] in (compat_urllib_error.URLError, socket.timeout, UnavailableVideoError):
             expected = True
-        if video_id is not None:
-            msg = video_id + ': ' + msg
-        if cause:
-            msg += ' (caused by %r)' % cause
-        if not expected:
-            msg += bug_reports_message()
-        super(ExtractorError, self).__init__(msg)
 
         self.traceback = tb
         self.exc_info = sys.exc_info()  # preserve original exception
         self.cause = cause
+        self.ie = ie
         self.video_id = video_id
+        self.expected = expected
+        super(ExtractorError, self).__init__(self.__msg())
+
+    def __msg(self):
+        msg = '[%s] ' % (self.ie,) if self.ie else ''
+        msg += join_nonempty(self.video_id, self.orig_msg, delim=': ')
+        if self.cause:
+            msg += ' (caused by %r)' % self.cause
+        if not self.expected:
+            msg += bug_reports_message()
+        return msg
 
     def format_traceback(self):
         if self.traceback is None:
             return None
         return ''.join(traceback.format_tb(self.traceback))
 
+    def __setattr__(self, name, value):
+        super(ExtractorError, self).__setattr__(name, value)
+        if getattr(self, 'msg', None) and name not in ('msg', 'args'):
+            self.msg = self.__msg() or type(self).__name__
+            self.args = (self.msg, )  # Cannot be property
+
 
 class UnsupportedError(ExtractorError):
     def __init__(self, url):