bpo-46054: Fix parsing error when parsing non-utf8 characters in source files (GH-30068)

Lib/test/test_exceptions.py

@@ -2387,6 +2387,18 @@ class SyntaxErrorTests(unittest.TestCase):
         finally:
             unlink(TESTFN)
 
+    def test_non_utf8(self):
+        # Check non utf-8 characters
+        try:
+            with open(TESTFN, 'bw') as testfile:
+                testfile.write(b'\x7fELF\x02\x01\x01\x00\x00\x00')
+            rc, out, err = script_helper.assert_python_failure('-Wd', '-X', 'utf8', TESTFN)
+            err = err.decode('utf-8').splitlines()
+
+            self.assertEqual(err[-1], "SyntaxError: invalid non-printable character U+007F")
+        finally:
+            unlink(TESTFN)
+
     def test_attributes_new_constructor(self):
         args = ("bad.py", 1, 2, "abcdefg", 1, 100)
         the_exception = SyntaxError("bad bad", args)

Misc/NEWS.d/next/Core and Builtins/2021-12-12-05-30-21.bpo-46054.2P-foG.rst

@@ -0,0 +1,2 @@
+Fix parser error when parsing non-utf8 characters in source files. Patch by
+Pablo Galindo.

Parser/tokenizer.c

@@ -819,10 +819,10 @@ tok_readline_raw(struct tok_state *tok)
             tok_concatenate_interactive_new_line(tok, line) == -1) {
             return 0;
         }
-        if (*tok->inp == '\0') {
+        tok->inp = strchr(tok->inp, '\0');
+        if (tok->inp == tok->buf) {
             return 0;
         }
-        tok->inp = strchr(tok->inp, '\0');
     } while (tok->inp[-1] != '\n');
     return 1;
 }
@@ -984,13 +984,10 @@ tok_underflow_file(struct tok_state *tok) {
     }
     /* The default encoding is UTF-8, so make sure we don't have any
        non-UTF-8 sequences in it. */
-    if (!tok->encoding
+    if (!tok->encoding && !ensure_utf8(tok->cur, tok)) {
-        && (tok->decoding_state != STATE_NORMAL || tok->lineno >= 2)) {
-        if (!ensure_utf8(tok->cur, tok)) {
         error_ret(tok);
         return 0;
     }
-    }
     assert(tok->done == E_OK);
     return tok->done == E_OK;
 }