[3.12] gh-105390: Correctly raise TokenError instead of SyntaxError f…

…or tokenize errors (GH-105399) (#105439)

Doc/library/tokenize.rst

@@ -139,11 +139,6 @@ function it uses to do this is available:
        2,
        3
 
-Note that unclosed single-quoted strings do not cause an error to be
-raised. They are tokenized as :data:`~token.ERRORTOKEN`, followed by the
-tokenization of their contents.
-
-
 .. _tokenize-cli:
 
 Command-Line Usage

Doc/whatsnew/3.12.rst

@@ -1489,14 +1489,15 @@ Changes in the Python API
   Additionally, there may be some minor behavioral changes as a consecuence of the
   changes required to support :pep:`701`. Some of these changes include:
 
-  * Some final ``DEDENT`` tokens are now emitted within the bounds of the
-    input. This means that for a file containing 3 lines, the old version of the
-    tokenizer returned a ``DEDENT`` token in line 4 whilst the new version returns
-    the token in line 3.
-
   * The ``type`` attribute of the tokens emitted when tokenizing some invalid Python
     characters such as ``!`` has changed from ``ERRORTOKEN`` to ``OP``.
 
+  * Incomplete single-line strings now also raise :exc:`tokenize.TokenError` as incomplete
+    multiline strings do.
+
+  * Some incomplete or invalid Python code now raises :exc:`tokenize.TokenError` instead of
+    returning arbitrary ``ERRORTOKEN`` tokens when tokenizing it.
+
 Build Changes
 =============
 

Lib/test/test_tokenize.py

@@ -3,7 +3,8 @@
 from tokenize import (tokenize, untokenize, NUMBER, NAME, OP,
                      STRING, ENDMARKER, ENCODING, tok_name, detect_encoding,
                      open as tokenize_open, Untokenizer, generate_tokens,
-                     NEWLINE, _generate_tokens_from_c_tokenizer, DEDENT, TokenInfo)
+                     NEWLINE, _generate_tokens_from_c_tokenizer, DEDENT, TokenInfo,
+                     TokenError)
 from io import BytesIO, StringIO
 import unittest
 from textwrap import dedent
@@ -286,7 +287,7 @@ def number_token(s):
         for lit in INVALID_UNDERSCORE_LITERALS:
             try:
                 number_token(lit)
-            except SyntaxError:
+            except TokenError:
                 continue
             self.assertNotEqual(number_token(lit), lit)
 
@@ -1379,7 +1380,7 @@ def test_latin1_normalization(self):
                 self.assertEqual(found, "iso-8859-1")
 
     def test_syntaxerror_latin1(self):
-        # Issue 14629: need to raise SyntaxError if the first
+        # Issue 14629: need to raise TokenError if the first
         # line(s) have non-UTF-8 characters
         lines = (
             b'print("\xdf")', # Latin-1: LATIN SMALL LETTER SHARP S
@@ -2754,7 +2755,7 @@ def get_tokens(string):
             "]",
         ]:
             with self.subTest(case=case):
-                self.assertRaises(SyntaxError, get_tokens, case)
+                self.assertRaises(TokenError, get_tokens, case)
 
     def test_max_indent(self):
         MAXINDENT = 100
@@ -2773,7 +2774,7 @@ def generate_source(indents):
 
         invalid = generate_source(MAXINDENT)
         the_input = StringIO(invalid)
-        self.assertRaises(SyntaxError, lambda: list(_generate_tokens_from_c_tokenizer(the_input.readline)))
+        self.assertRaises(IndentationError, lambda: list(_generate_tokens_from_c_tokenizer(the_input.readline)))
         self.assertRaises(
             IndentationError, compile, invalid, "<string>", "exec"
         )

Lib/tokenize.py

@@ -517,14 +517,30 @@ def error(message, filename=None, location=None):
         perror("unexpected error: %s" % err)
         raise
 
+def _transform_msg(msg):
+    """Transform error messages from the C tokenizer into the Python tokenize
+
+    The C tokenizer is more picky than the Python one, so we need to massage
+    the error messages a bit for backwards compatibility.
+    """
+    if "unterminated triple-quoted string literal" in msg:
+        return "EOF in multi-line string"
+    return msg
+
 def _generate_tokens_from_c_tokenizer(source, encoding=None, extra_tokens=False):
     """Tokenize a source reading Python code as unicode strings using the internal C tokenizer"""
     if encoding is None:
         it = _tokenize.TokenizerIter(source, extra_tokens=extra_tokens)
     else:
         it = _tokenize.TokenizerIter(source, encoding=encoding, extra_tokens=extra_tokens)
-    for info in it:
-        yield TokenInfo._make(info)
+    try:
+        for info in it:
+            yield TokenInfo._make(info)
+    except SyntaxError as e:
+        if type(e) != SyntaxError:
+            raise e from None
+        msg = _transform_msg(e.msg)
+        raise TokenError(msg, (e.lineno, e.offset)) from None
 
 
 if __name__ == "__main__":

Misc/NEWS.d/next/Core and Builtins/2023-06-06-17-10-42.gh-issue-105390.DvqI-e.rst

@@ -0,0 +1,3 @@
+Correctly raise :exc:`tokenize.TokenError` exceptions instead of
+:exc:`SyntaxError` for tokenize errors such as incomplete input. Patch by
+Pablo Galindo

Python/Python-tokenize.c

@@ -84,13 +84,8 @@ _tokenizer_error(struct tok_state *tok)
             msg = "invalid token";
             break;
         case E_EOF:
-            if (tok->level > 0) {
-                    PyErr_Format(PyExc_SyntaxError,
-                                 "parenthesis '%c' was never closed",
-                                tok->parenstack[tok->level-1]);
-            } else {
-                PyErr_SetString(PyExc_SyntaxError, "unexpected EOF while parsing");
-            }
+            PyErr_SetString(PyExc_SyntaxError, "unexpected EOF in multi-line statement");
+            PyErr_SyntaxLocationObject(tok->filename, tok->lineno, tok->inp - tok->buf < 0 ? 0 : tok->inp - tok->buf);
             return -1;
         case E_DEDENT:
             msg = "unindent does not match any outer indentation level";