gh-151464: exclude '<>' token from tokenize output

Doc/whatsnew/3.16.rst

@@ -461,6 +461,13 @@ symtable
   (Contributed by Serhiy Storchaka in :gh:`153844`.)
 
 
+tokenize
+--------
+
+* Invalid token ``<>`` excluded from :mod:`tokenize` output.
+  (Contributed by Sergey B Kirpichev in :gh:`151464`.)
+
+
 tkinter
 -------
 

Grammar/python.gram

@@ -807,6 +807,7 @@ noteq_bitwise_or[CmpopExprPair*]:
     | (tok='!=' { _PyPegen_check_barry_as_flufl(p, tok) ? NULL : tok}) a=bitwise_or {_PyPegen_cmpop_expr_pair(p, NotEq, a) }
 lte_bitwise_or[CmpopExprPair*]: '<=' a=bitwise_or { _PyPegen_cmpop_expr_pair(p, LtE, a) }
 lt_bitwise_or[CmpopExprPair*]: '<' a=bitwise_or { _PyPegen_cmpop_expr_pair(p, Lt, a) }
+    | invalid_noteq
 gte_bitwise_or[CmpopExprPair*]: '>=' a=bitwise_or { _PyPegen_cmpop_expr_pair(p, GtE, a) }
 gt_bitwise_or[CmpopExprPair*]: '>' a=bitwise_or { _PyPegen_cmpop_expr_pair(p, Gt, a) }
 notin_bitwise_or[CmpopExprPair*]: 'not' 'in' a=bitwise_or { _PyPegen_cmpop_expr_pair(p, NotIn, a) }
@@ -1659,3 +1660,10 @@ invalid_bitwise_or:
         ? RAISE_SYNTAX_ERROR_KNOWN_RANGE(b, c, "invalid syntax.  Maybe you meant 'or' or '|' instead of '||'?")
         : NULL
     }
+
+invalid_noteq:
+    | a='<' b='>' {
+      _PyPegen_tokens_are_adjacent(a, b)
+      ? RAISE_SYNTAX_ERROR_KNOWN_RANGE(a, b, "invalid syntax.  Maybe you meant '!=' instead of '<>'?")
+      : NULL
+    }

Include/internal/pycore_token.h

@@ -101,7 +101,7 @@ extern "C" {
 // Export these 4 symbols for 'test_peg_generator'
 PyAPI_DATA(const char * const) _PyParser_TokenNames[]; /* Token names */
 PyAPI_FUNC(int) _PyToken_OneChar(int);
-PyAPI_FUNC(int) _PyToken_TwoChars(int, int);
+PyAPI_FUNC(int) _PyToken_TwoChars(int, int, int);
 PyAPI_FUNC(int) _PyToken_ThreeChars(int, int, int);
 
 #ifdef __cplusplus

Lib/test/test_flufl.py

@@ -59,7 +59,5 @@ def test_barry_as_bdfl_relative_import(self):
         self.assertEqual(cm.exception.offset, len(code) - 4)
 
 
-
-
 if __name__ == '__main__':
     unittest.main()

Lib/test/test_syntax.py

@@ -3608,6 +3608,31 @@ def test_ifexp_body_stmt_else_stmt(self):
         ]:
             self._check_error(f"x = {lhs_stmt} if 1 else {rhs_stmt}", msg)
 
+    def test_diamond_operator(self):
+        self._check_error(
+            "1<>2",
+            r"Maybe you meant '!=' instead of '<>'\?",
+            lineno=1,
+            end_lineno=1,
+            offset=2,
+            end_offset=4,
+        )
+
+    def test_diamond_operator_barry_as_flufl(self):
+        # Under barry_as_FLUFL, '<>' is the valid "not equal" operator
+        compile(
+            "from __future__ import barry_as_FLUFL\n1<>2",
+            "<test>", "exec",
+        )
+        self._check_error(
+            "from __future__ import barry_as_FLUFL\na != b",
+            "with Barry as BDFL, use '<>' instead of '!='",
+            lineno=2,
+            end_lineno=2,
+            offset=3,
+            end_offset=5,
+        )
+
     def test_double_ampersand(self):
         self._check_error(
             "a && b",

Lib/test/test_tokenize.py

@@ -1233,6 +1233,20 @@ def test_multiline_non_ascii_fstring_with_expr(self):
     FSTRING_END \'"\'           (2, 2) (2, 3)
     """)
 
+    def test_ineq_tokens(self):
+        self.check_tokenize("1 != 2", """\
+    NUMBER     '1'           (1, 0) (1, 1)
+    OP         '!='          (1, 2) (1, 4)
+    NUMBER     '2'           (1, 5) (1, 6)
+""")
+        self.check_tokenize("1 <> 2", """\
+    NUMBER     '1'           (1, 0) (1, 1)
+    OP         '<'           (1, 2) (1, 3)
+    OP         '>'           (1, 3) (1, 4)
+    NUMBER     '2'           (1, 5) (1, 6)
+""")
+
+
 class GenerateTokensTest(TokenizeTest):
     def check_tokenize(self, s, expected):
         # Format the tokens in s in a table format.

Misc/NEWS.d/next/Core_and_Builtins/2026-07-29-08-43-14.gh-issue-151464.o_mtmz.rst

@@ -0,0 +1,2 @@
+Exclude invalid token ``<>`` from :mod:`tokenize` output.  :exc:`SyntaxError`
+for ``<>`` now suggests ``!=``.

Parser/action_helpers.c

@@ -2076,6 +2076,7 @@ _PyPegen_checked_from_import(Parser *p, asdl_seq *dots, expr_ty module_name,
             alias_ty alias = asdl_seq_GET(names, i);
             if (PyUnicode_CompareWithASCIIString(alias->name, "barry_as_FLUFL") == 0) {
                 p->flags |= PyPARSE_BARRY_AS_BDFL;
+                p->tok->barry_as_bdfl = 1;
             }
         }
     }

Parser/lexer/lexer.c

@@ -562,7 +562,7 @@ _PyLexer_get_normal_mode(struct tok_state *tok, tokenizer_mode* current_tok, str
     /* Check for two-character token */
     {
         int c2 = tok_nextc(tok);
-        int current_token = _PyToken_TwoChars(c, c2);
+        int current_token = _PyToken_TwoChars(c, c2, tok->barry_as_bdfl);
         if (current_token != OP) {
             int c3 = tok_nextc(tok);
             int current_token3 = _PyToken_ThreeChars(c, c2, c3);

Parser/lexer/state.c

@@ -63,6 +63,7 @@ _PyTokenizer_tok_new(void)
 #ifdef Py_DEBUG
     tok->debug = _Py_GetConfig()->parser_debug;
 #endif
+    tok->barry_as_bdfl = 0;
     return tok;
 }
 

Parser/lexer/state.h

@@ -137,6 +137,7 @@ struct tok_state {
 #ifdef Py_DEBUG
     int debug;
 #endif
+    int barry_as_bdfl;
 };
 
 int _PyLexer_type_comment_token_setup(struct tok_state *tok, struct token *token, int type, int col_offset,