[ie/youtube] Fix comments extraction (#9775)

[yt-dlp.git] / test / test_download.py

diff --git a/test/test_download.py b/test/test_download.py
index 787013c3422102420896d1e9f1fa59b128125c77..2530792493ae078627776d36792883b06f6c4ac8 100755 (executable)
--- a/test/test_download.py
+++ b/test/test_download.py
@@ -10,10 +10,7 @@
 
 import collections
 import hashlib
-import http.client
 import json
-import socket
-import urllib.error
 
 from test.helper import (
     assertGreaterEqual,
@@ -29,10 +26,12 @@
 
 import yt_dlp.YoutubeDL  # isort: split
 from yt_dlp.extractor import get_info_extractor
+from yt_dlp.networking.exceptions import HTTPError, TransportError
 from yt_dlp.utils import (
     DownloadError,
     ExtractorError,
     UnavailableVideoError,
+    YoutubeDLError,
     format_bytes,
     join_nonempty,
 )
@@ -102,14 +101,16 @@ def print_skipping(reason):
             print_skipping('IE marked as not _WORKING')
 
         for tc in test_cases:
+            if tc.get('expected_exception'):
+                continue
             info_dict = tc.get('info_dict', {})
             params = tc.get('params', {})
             if not info_dict.get('id'):
-                raise Exception('Test definition incorrect. \'id\' key is not present')
-            elif not info_dict.get('ext'):
+                raise Exception(f'Test {tname} definition incorrect - "id" key is not present')
+            elif not info_dict.get('ext') and info_dict.get('_type', 'video') == 'video':
                 if params.get('skip_download') and params.get('ignore_no_formats_error'):
                     continue
-                raise Exception('Test definition incorrect. The output file cannot be known. \'ext\' key is not present')
+                raise Exception(f'Test {tname} definition incorrect - "ext" key must be present to define the output file')
 
         if 'skip' in test_case:
             print_skipping(test_case['skip'])
@@ -122,7 +123,8 @@ def print_skipping(reason):
         params['outtmpl'] = tname + '_' + params['outtmpl']
         if is_playlist and 'playlist' not in test_case:
             params.setdefault('extract_flat', 'in_playlist')
-            params.setdefault('playlistend', test_case.get('playlist_mincount'))
+            params.setdefault('playlistend', test_case.get(
+                'playlist_mincount', test_case.get('playlist_count', -2) + 1))
             params.setdefault('skip_download', True)
 
         ydl = YoutubeDL(params, auto_init=False)
@@ -140,6 +142,17 @@ def get_tc_filename(tc):
 
         res_dict = None
 
+        def match_exception(err):
+            expected_exception = test_case.get('expected_exception')
+            if not expected_exception:
+                return False
+            if err.__class__.__name__ == expected_exception:
+                return True
+            for exc in err.exc_info:
+                if exc.__class__.__name__ == expected_exception:
+                    return True
+            return False
+
         def try_rm_tcs_files(tcs=None):
             if tcs is None:
                 tcs = test_cases
@@ -161,7 +174,10 @@ def try_rm_tcs_files(tcs=None):
                         force_generic_extractor=params.get('force_generic_extractor', False))
                 except (DownloadError, ExtractorError) as err:
                     # Check if the exception is not a network related one
-                    if not err.exc_info[0] in (urllib.error.URLError, socket.timeout, UnavailableVideoError, http.client.BadStatusLine) or (err.exc_info[0] == urllib.error.HTTPError and err.exc_info[1].code == 503):
+                    if not isinstance(err.exc_info[1], (TransportError, UnavailableVideoError)) or (isinstance(err.exc_info[1], HTTPError) and err.exc_info[1].status == 503):
+                        if match_exception(err):
+                            return
+                        err.msg = f'{getattr(err, "msg", err)} ({tname})'
                         raise
 
                     if try_num == RETRIES:
@@ -171,6 +187,10 @@ def try_rm_tcs_files(tcs=None):
                     print(f'Retrying: {try_num} failed tries\n\n##########\n\n')
 
                     try_num += 1
+                except YoutubeDLError as err:
+                    if match_exception(err):
+                        return
+                    raise
                 else:
                     break
 
@@ -210,6 +230,8 @@ def try_rm_tcs_files(tcs=None):
                 tc_res_dict = res_dict['entries'][tc_num]
                 # First, check test cases' data against extracted data alone
                 expect_info_dict(self, tc_res_dict, tc.get('info_dict', {}))
+                if tc_res_dict.get('_type', 'video') != 'video':
+                    continue
                 # Now, check downloaded file consistency
                 tc_filename = get_tc_filename(tc)
                 if not test_case.get('params', {}).get('skip_download', False):
@@ -244,7 +266,7 @@ def try_rm_tcs_files(tcs=None):
                 # extractor returns full results even with extract_flat
                 res_tcs = [{'info_dict': e} for e in res_dict['entries']]
                 try_rm_tcs_files(res_tcs)
-
+            ydl.close()
     return test_template