gh-105549: Tokenize separately NUMBER and NAME tokens and allow 0-prefixed literals

Lib/test/test_tokenize.py

@@ -284,7 +284,12 @@ def number_token(s):
                 # this won't work with compound complex inputs
                 continue
             self.assertEqual(number_token(lit), lit)
+        # Valid cases with extra underscores in the tokenize module
+        # See gh-105549 for context
+        extra_valid_cases = {"0_7", "09_99"}
         for lit in INVALID_UNDERSCORE_LITERALS:
+            if lit in extra_valid_cases:
+                continue
             try:
                 number_token(lit)
             except TokenError:
@@ -1873,6 +1878,34 @@ def test_indentation_semantics_retained(self):
         self.check_roundtrip(code)
 
 
+class InvalidPythonTests(TestCase):
+    def test_number_followed_by_name(self):
+        # See issue #gh-105549
+        source = "2sin(x)"
+        expected_tokens = [
+            TokenInfo(type=token.NUMBER, string='2', start=(1, 0), end=(1, 1), line='2sin(x)'),
+            TokenInfo(type=token.NAME, string='sin', start=(1, 1), end=(1, 4), line='2sin(x)'),
+            TokenInfo(type=token.OP, string='(', start=(1, 4), end=(1, 5), line='2sin(x)'),
+            TokenInfo(type=token.NAME, string='x', start=(1, 5), end=(1, 6), line='2sin(x)'),
+            TokenInfo(type=token.OP, string=')', start=(1, 6), end=(1, 7), line='2sin(x)'),
+            TokenInfo(type=token.NEWLINE, string='', start=(1, 7), end=(1, 8), line='2sin(x)'),
+            TokenInfo(type=token.ENDMARKER, string='', start=(2, 0), end=(2, 0), line='')
+        ]
+
+        tokens = list(generate_tokens(StringIO(source).readline))
+        self.assertEqual(tokens, expected_tokens)
+
+    def test_number_starting_with_zero(self):
+        source = "01234"
+        expected_tokens = [
+            TokenInfo(type=token.NUMBER, string='01234', start=(1, 0), end=(1, 5), line='01234'),
+            TokenInfo(type=token.NEWLINE, string='', start=(1, 5), end=(1, 6), line='01234'),
+            TokenInfo(type=token.ENDMARKER, string='', start=(2, 0), end=(2, 0), line='')
+        ]
+
+        tokens = list(generate_tokens(StringIO(source).readline))
+        self.assertEqual(tokens, expected_tokens)
+
 class CTokenizeTest(TestCase):
     def check_tokenize(self, s, expected):
         # Format the tokens in s in a table format.

Misc/NEWS.d/next/Core and Builtins/2023-06-09-12-59-18.gh-issue-105549.PYfTNp.rst

@@ -0,0 +1,2 @@
+Tokenize separately `NUMBER` and `NAME` tokens that are not ambiguous. Patch
+by Pablo Galindo

Parser/tokenizer.c

@@ -1600,8 +1600,12 @@ lookahead(struct tok_state *tok, const char *test)
 }
 
 static int
-verify_end_of_number(struct tok_state *tok, int c, const char *kind)
-{
+verify_end_of_number(struct tok_state *tok, int c, const char *kind) {
+    if (tok->tok_extra_tokens) {
+        // When we are parsing extra tokens, we don't want to emit warnings
+        // about invalid literals, because we want to be a bit more liberal.
+        return 1;
+    }
     /* Emit a deprecation warning only if the numeric literal is immediately
      * followed by one of keywords which can occur after a numeric literal
      * in valid code: "and", "else", "for", "if", "in", "is" and "or".
@@ -1659,6 +1663,9 @@ verify_end_of_number(struct tok_state *tok, int c, const char *kind)
 static int
 verify_identifier(struct tok_state *tok)
 {
+    if (tok->tok_extra_tokens) {
+        return 1;
+    }
     PyObject *s;
     if (tok->decoding_erred)
         return 0;
@@ -2318,7 +2325,7 @@ tok_get_normal_mode(struct tok_state *tok, tokenizer_mode* current_tok, struct t
                 else if (c == 'j' || c == 'J') {
                     goto imaginary;
                 }
-                else if (nonzero) {
+                else if (nonzero && !tok->tok_extra_tokens) {
                     /* Old-style octal: now disallowed. */
                     tok_backup(tok, c);
                     return MAKE_TOKEN(syntaxerror_known_range(