[core] Fix format selection parse error for CPython 3.12 (#8797)

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

diff --git a/yt_dlp/YoutubeDL.py b/yt_dlp/YoutubeDL.py
index 29dd761862faa362af0e9b0b7c3816d2894f80fa..5e28fd0e21c00c6d1b1d563ea5cf38e5a0b05545 100644 (file)
--- a/yt_dlp/YoutubeDL.py
+++ b/yt_dlp/YoutubeDL.py
@@ -1201,6 +1201,15 @@ def prepare_outtmpl(self, outtmpl, info_dict, sanitize=False):
                 (?:\|(?P<default>.*?))?
             )$''')
 
+        def _from_user_input(field):
+            if field == ':':
+                return ...
+            elif ':' in field:
+                return slice(*map(int_or_none, field.split(':')))
+            elif int_or_none(field) is not None:
+                return int(field)
+            return field
+
         def _traverse_infodict(fields):
             fields = [f for x in re.split(r'\.({.+?})\.?', fields)
                       for f in ([x] if x.startswith('{') else x.split('.'))]
@@ -1210,11 +1219,12 @@ def _traverse_infodict(fields):
 
             for i, f in enumerate(fields):
                 if not f.startswith('{'):
+                    fields[i] = _from_user_input(f)
                     continue
                 assert f.endswith('}'), f'No closing brace for {f} in {fields}'
-                fields[i] = {k: k.split('.') for k in f[1:-1].split(',')}
+                fields[i] = {k: list(map(_from_user_input, k.split('.'))) for k in f[1:-1].split(',')}
 
-            return traverse_obj(info_dict, fields, is_user_input=True, traverse_string=True)
+            return traverse_obj(info_dict, fields, traverse_string=True)
 
         def get_value(mdict):
             # Object traversal
@@ -2455,9 +2465,16 @@ def final_selector(ctx):
                 return selector_function(ctx_copy)
             return final_selector
 
-        stream = io.BytesIO(format_spec.encode())
+        # HACK: Python 3.12 changed the underlying parser, rendering '7_a' invalid
+        #       Prefix numbers with random letters to avoid it being classified as a number
+        #       See: https://github.com/yt-dlp/yt-dlp/pulls/8797
+        # TODO: Implement parser not reliant on tokenize.tokenize
+        prefix = ''.join(random.choices(string.ascii_letters, k=32))
+        stream = io.BytesIO(re.sub(r'\d[_\d]*', rf'{prefix}\g<0>', format_spec).encode())
         try:
-            tokens = list(_remove_unused_ops(tokenize.tokenize(stream.readline)))
+            tokens = list(_remove_unused_ops(
+                token._replace(string=token.string.replace(prefix, ''))
+                for token in tokenize.tokenize(stream.readline)))
         except tokenize.TokenError:
             raise syntax_error('Missing closing/opening brackets or parenthesis', (0, len(format_spec)))