Merge branch 'main' into issue-152631

Author: sobolevn

Doc/library/email.charset.rst

@@ -183,8 +183,8 @@ new entries to the global character set, alias, and codec registries:
 
    Optional *output_charset* is the character set that the output should be in.
    Conversions will proceed from input charset, to Unicode, to the output charset
-   when the method :meth:`Charset.convert` is called.  The default is to output in
-   the same character set as the input.
+   when the method :meth:`Charset.body_encode` is called.  The default is to
+   output in the same character set as the input.
 
    Both *input_charset* and *output_charset* must have Unicode codec entries in the
    module's character set-to-codec mapping; use :func:`add_codec` to add codecs the

Doc/library/email.parser.rst

@@ -48,7 +48,7 @@ methods.
 FeedParser API
 ^^^^^^^^^^^^^^
 
-The :class:`BytesFeedParser`, imported from the :mod:`email.feedparser` module,
+The :class:`BytesFeedParser`, imported from the :mod:`!email.feedparser` module,
 provides an API that is conducive to incremental parsing of email messages,
 such as would be necessary when reading the text of an email message from a
 source that can block (such as a socket).  The :class:`BytesFeedParser` can of
@@ -86,7 +86,7 @@ Here is the API for the :class:`BytesFeedParser`:
    more information on what else *policy* controls, see the
    :mod:`~email.policy` documentation.
 
-   Note: **The policy keyword should always be specified**; The default will
+   Note: **The policy keyword should always be specified**. The default will
    change to :data:`email.policy.default` in a future version of Python.
 
    .. versionadded:: 3.2
@@ -142,7 +142,7 @@ message body, instead setting the payload to the raw body.
    arguments have the same meaning and semantics as the *_factory*
    and *policy* arguments of :class:`BytesFeedParser`.
 
-   Note: **The policy keyword should always be specified**; The default will
+   Note: **The policy keyword should always be specified**. The default will
    change to :data:`email.policy.default` in a future version of Python.
 
    .. versionchanged:: 3.3

Doc/tools/.nitignore

@@ -4,8 +4,6 @@
 
 Doc/library/ast.rst
 Doc/library/asyncio-extending.rst
-Doc/library/email.charset.rst
-Doc/library/email.parser.rst
 Doc/library/logging.config.rst
 Doc/library/logging.handlers.rst
 Doc/library/multiprocessing.rst

Include/internal/pycore_pyerrors.h

@@ -186,7 +186,6 @@ extern int _PyErr_CheckSignalsTstate(PyThreadState *tstate);
 extern void _Py_DumpExtensionModules(int fd, PyInterpreterState *interp);
 // Exported for external JIT support
 PyAPI_FUNC(PyObject *) _Py_CalculateSuggestions(PyObject *dir, PyObject *name);
-extern PyObject* _Py_Offer_Suggestions(PyObject* exception);
 
 // Export for '_testinternalcapi' shared extension
 PyAPI_FUNC(Py_ssize_t) _Py_UTF8_Edit_Cost(PyObject *str_a, PyObject *str_b,

Lib/email/charset.py

@@ -151,7 +151,7 @@ def add_charset(charset, header_enc=None, body_enc=None, output_charset=None):
 
     Optional output_charset is the character set that the output should be
     in.  Conversions will proceed from input charset, to Unicode, to the
-    output charset when the method Charset.convert() is called.  The default
+    output charset when the method Charset.body_encode() is called.  The default
     is to output in the same character set as the input.
 
     Both input_charset and output_charset must have Unicode codec entries in

Lib/gettext.py

@@ -362,6 +362,8 @@ def _parse(self, fp):
         self.plural = lambda n: int(n != 1) # germanic plural by default
         buf = fp.read()
         buflen = len(buf)
+        if buflen < 20:
+            raise OSError(0, 'File is corrupt', filename)
         # Are we big endian or little endian?
         magic = unpack('<I', buf[:4])[0]
         if magic == self.LE_MAGIC:
@@ -378,6 +380,11 @@ def _parse(self, fp):
         if major_version not in self.VERSIONS:
             raise OSError(0, 'Bad version number ' + str(major_version), filename)
 
+        # The seek tables must lie within the file.
+        if (masteridx + 8 * msgcount > buflen
+                or transidx + 8 * msgcount > buflen):
+            raise OSError(0, 'File is corrupt', filename)
+
         # Now put all messages from the .mo file buffer into the catalog
         # dictionary.
         for i in range(0, msgcount):

Lib/test/test_ast/test_ast.py

@@ -103,6 +103,13 @@ def test_AST_objects(self):
             # "ast.AST constructor takes 0 positional arguments"
             ast.AST(2)
 
+    @unittest.skipIf(_testinternalcapi is None, "requires _testinternalcapi")
+    def test_nodes_have_inline_values(self):
+        tree = ast.parse("x = f(y, z=1)")
+        for node in ast.walk(tree):
+            with self.subTest(node=type(node).__name__):
+                self.assertTrue(_testinternalcapi.has_inline_values(node))
+
     def test_AST_fields_NULL_check(self):
         # See: https://github.com/python/cpython/issues/126105
         old_value = ast.AST._fields

Lib/test/test_gettext.py

@@ -1,6 +1,7 @@
 import os
 import base64
 import gettext
+import io
 import unittest
 import unittest.mock
 from functools import partial
@@ -39,7 +40,7 @@
 '''
 
 # .mo file with an invalid magic number
-GNU_MO_DATA_BAD_MAGIC_NUMBER = base64.b64encode(b'ABCD')
+GNU_MO_DATA_BAD_MAGIC_NUMBER = base64.b64encode(b'ABCD' + bytes(16))
 
 # This data contains an invalid major version number (5)
 # An unexpected major version number should be treated as an error when
@@ -314,6 +315,39 @@ def test_corrupt_file(self):
             self.assertEqual(exception.strerror, "File is corrupt")
             self.assertEqual(exception.filename, MOFILE_CORRUPT)
 
+    def test_truncated_header(self):
+        magic = b'\xde\x12\x04\x95'
+        for buf in (b'', magic, magic + bytes(15)):
+            with self.subTest(buflen=len(buf)):
+                with self.assertRaises(OSError) as cm:
+                    gettext.GNUTranslations(io.BytesIO(buf))
+                self.assertEqual(cm.exception.errno, 0)
+                self.assertEqual(cm.exception.strerror, "File is corrupt")
+
+    def test_offsets_out_of_bounds(self):
+        buf = bytes([
+            0xde, 0x12, 0x04, 0x95,  # Magic
+            0x00, 0x00, 0x00, 0x00,  # Version
+            0x01, 0x00, 0x00, 0x00,  # Message count
+            0xe8, 0x03, 0x00, 0x00,  # Message offset (past EOF)
+            0xd0, 0x07, 0x00, 0x00,  # Translation offset (past EOF)
+        ])
+        with self.assertRaises(OSError) as cm:
+            gettext.GNUTranslations(io.BytesIO(buf))
+        self.assertEqual(cm.exception.errno, 0)
+        self.assertEqual(cm.exception.strerror, "File is corrupt")
+
+    def test_empty_catalog(self):
+        buf = bytes([
+            0xde, 0x12, 0x04, 0x95,  # Magic
+            0x00, 0x00, 0x00, 0x00,  # Version
+            0x00, 0x00, 0x00, 0x00,  # Message count
+            0x14, 0x00, 0x00, 0x00,  # Message offset
+            0x14, 0x00, 0x00, 0x00,  # Translation offset
+        ])
+        t = gettext.GNUTranslations(io.BytesIO(buf))
+        self.assertEqual(t.gettext('foo'), 'foo')
+
     def test_big_endian_file(self):
         with open(MOFILE_BIG_ENDIAN, 'rb') as fp:
             t = gettext.GNUTranslations(fp)

Lib/test/test_sys.py

@@ -2042,7 +2042,7 @@ def test_pythontypes(self):
         check = self.check_sizeof
         # _ast.AST
         import _ast
-        check(_ast.Module(), size('3P'))
+        check(_ast.Module(), size('2P'))
         try:
             raise TypeError
         except TypeError as e:

Misc/NEWS.d/next/Core_and_Builtins/2026-10-23-23-45-00.gh-issue-153568.astinl.rst

@@ -0,0 +1,2 @@
+Speed up :func:`ast.parse` and reduce the memory used by AST nodes by storing
+node attributes as inline values instead of a separate per-node dictionary.