[3.12] gh-88943: Improve syntax error for non-ASCII character that follows a numerical literal (GH-109081) (#109090)

gh-88943: Improve syntax error for non-ASCII character that follows a numerical literal (GH-109081)

It now points on the invalid non-ASCII character, not on the valid numerical literal.
(cherry picked from commit b2729e9)

Co-authored-by: Serhiy Storchaka <[email protected]>

Author: miss-islington

Lib/test/test_grammar.py

@@ -236,6 +236,10 @@ def check(test, error=False):
             check(f"[{num}for x in ()]")
             check(f"{num}spam", error=True)
 
+            # gh-88943: Invalid non-ASCII character following a numerical literal.
+            with self.assertRaisesRegex(SyntaxError, r"invalid character '⁄' \(U\+2044\)"):
+                compile(f"{num}⁄7", "<testcase>", "eval")
+
             with self.assertWarnsRegex(SyntaxWarning, r'invalid \w+ literal'):
                 compile(f"{num}is x", "<testcase>", "eval")
             with warnings.catch_warnings():

Misc/NEWS.d/next/Core and Builtins/2023-09-07-16-05-36.gh-issue-88943.rH_X3W.rst

@@ -0,0 +1,3 @@
+Improve syntax error for non-ASCII character that follows a numerical
+literal. It now points on the invalid non-ASCII character, not on the valid
+numerical literal.

Parser/tokenizer.c

@@ -1648,7 +1648,7 @@ verify_end_of_number(struct tok_state *tok, int c, const char *kind) {
         tok_nextc(tok);
     }
     else /* In future releases, only error will remain. */
-    if (is_potential_identifier_char(c)) {
+    if (c < 128 && is_potential_identifier_char(c)) {
         tok_backup(tok, c);
         syntaxerror(tok, "invalid %s literal", kind);
         return 0;