Let `--match-filter` reject entries early

[yt-dlp.git] / yt_dlp / utils.py

diff --git a/yt_dlp/utils.py b/yt_dlp/utils.py
index 0e8392fdffe7d6679e146648284679981dfae487..6276ac726be379f628e056b8e76b5d7b8523d0b8 100644 (file)
--- a/yt_dlp/utils.py
+++ b/yt_dlp/utils.py
@@ -4657,7 +4657,7 @@ def filter_using_list(row, filterArray):
     return '\n'.join(format_str % tuple(row) for row in table)
 
 
-def _match_one(filter_part, dct):
+def _match_one(filter_part, dct, incomplete):
     # TODO: Generalize code with YoutubeDL._build_format_filter
     STRING_OPERATORS = {
         '*=': operator.contains,
@@ -4718,7 +4718,7 @@ def _match_one(filter_part, dct):
                         'Invalid integer value %r in filter part %r' % (
                             m.group('intval'), filter_part))
         if actual_value is None:
-            return m.group('none_inclusive')
+            return incomplete or m.group('none_inclusive')
         return op(actual_value, comparison_value)
 
     UNARY_OPERATORS = {
@@ -4733,22 +4733,25 @@ def _match_one(filter_part, dct):
     if m:
         op = UNARY_OPERATORS[m.group('op')]
         actual_value = dct.get(m.group('key'))
+        if incomplete and actual_value is None:
+            return True
         return op(actual_value)
 
     raise ValueError('Invalid filter part %r' % filter_part)
 
 
-def match_str(filter_str, dct):
-    """ Filter a dictionary with a simple string syntax. Returns True (=passes filter) or false """
-
+def match_str(filter_str, dct, incomplete=False):
+    """ Filter a dictionary with a simple string syntax. Returns True (=passes filter) or false
+        When incomplete, all conditions passes on missing fields
+    """
     return all(
-        _match_one(filter_part.replace(r'\&', '&'), dct)
+        _match_one(filter_part.replace(r'\&', '&'), dct, incomplete)
         for filter_part in re.split(r'(?<!\\)&', filter_str))
 
 
 def match_filter_func(filter_str):
-    def _match_func(info_dict):
-        if match_str(filter_str, info_dict):
+    def _match_func(info_dict, *args, **kwargs):
+        if match_str(filter_str, info_dict, *args, **kwargs):
             return None
         else:
             video_title = info_dict.get('title', info_dict.get('id', 'video'))