Sync pre-release CPython 3.15 branch from GitHub (2026-08-31)

Summary:
Imported python/cpython `3.15.0rc1+dev` from upstream rev [`b93576a`](https://www.github.com/python/cpython/commit/b93576a2b66d98ec8fe2bf64de6fd27a7851e1ed) (committed 2026-08-31 19:05:08+00:00).

# Commit Info

- Base: (`3.15.0rc1+dev`) - [`5136a68`](https://www.github.com/python/cpython/commit/5136a686beda8566da8592b90cea4b8e734724b6) (commit date: 2026-08-30 18:20:30+00:00)
- Imported: (`3.15.0rc1+dev`) - [`b93576a`](https://www.github.com/python/cpython/commit/b93576a2b66d98ec8fe2bf64de6fd27a7851e1ed) (commit date: 2026-08-31 19:05:08+00:00)

# Noteworthy file changes

- Low-signal files (6 added, 1 removed) (NEWS.d, docs, .github)

Complete list of added/removed files: https://www.internalfb.com/intern/everpaste/?color=0&handle=GL-nlisr4yDjBMgDALm7qp6M0e4Obr0LAAAz

Reviewed By: itamaro

Differential Revision: D118238648

fbshipit-source-id: 13acc96c08b89681b084230f61a5d8c4191292cd

Author: generatedunixname1734921407115435

Doc/library/mimetypes.rst

@@ -201,8 +201,8 @@ than one MIME-type database; it provides an interface similar to the one of the
 .. class:: MimeTypes(filenames=(), strict=True)
 
    This class represents a MIME-types database.  By default, it provides access to
-   the same database as the rest of this module. The initial database is a copy of
-   that provided by the module, and may be extended by loading additional
+   the same database as the rest of this module. The initial database is created from
+   Python's built-in MIME type tables. It may be extended by loading additional
    :file:`mime.types`\ -style files into the database using the :meth:`read` or
    :meth:`readfp` methods.  The mapping dictionaries may also be cleared before
    loading additional data if the default data is not desired.
@@ -216,30 +216,32 @@ than one MIME-type database; it provides an interface similar to the one of the
       Dictionary mapping suffixes to suffixes.  This is used to allow recognition of
       encoded files for which the encoding and the type are indicated by the same
       extension.  For example, the :file:`.tgz` extension is mapped to :file:`.tar.gz`
-      to allow the encoding and type to be recognized separately.  This is initially a
-      copy of the global :data:`suffix_map` defined in the module.
+      to allow the encoding and type to be recognized separately.
+      This is initialized with some predefined values.
 
 
    .. attribute:: MimeTypes.encodings_map
 
-      Dictionary mapping filename extensions to encoding types.  This is initially a
-      copy of the global :data:`encodings_map` defined in the module.
+      Dictionary mapping filename extensions to encoding types.
+      This is initialized with some predefined values.
 
 
    .. attribute:: MimeTypes.types_map
 
       Tuple containing two dictionaries, mapping filename extensions to MIME types:
       the first dictionary is for the non-standards types and the second one is for
-      the standard types. They are initialized by :data:`common_types` and
-      :data:`types_map`.
+      the standard types.
+      They are initialized with some predefined values and MIME type
+      information loaded from files specified by the *filenames* argument.
 
 
    .. attribute:: MimeTypes.types_map_inv
 
       Tuple containing two dictionaries, mapping MIME types to a list of filename
       extensions: the first dictionary is for the non-standards types and the
-      second one is for the standard types. They are initialized by
-      :data:`common_types` and :data:`types_map`.
+      second one is for the standard types.
+      They are initialized with some predefined values and MIME type
+      information loaded from files specified by the *filenames* argument.
 
 
    .. method:: MimeTypes.guess_extension(type, strict=True)

Doc/library/tarfile.rst

@@ -1107,6 +1107,10 @@ reused in custom filters:
     paths (in case the name is absolute
     even after stripping slashes, e.g. ``C:/foo`` on Windows).
     This raises :class:`~tarfile.AbsolutePathError`.
+  - Normalize filenames (:attr:`TarInfo.name`) that contain ``..`` components
+    using :func:`os.path.normpath`.
+    Note that this removes internal ``..`` components, which may change the
+    meaning of the name if it traverses symbolic links.
   - :ref:`Refuse <tarfile-extraction-refuse>` to extract files whose absolute
     path (after following symlinks) would end up outside the destination.
     This raises :class:`~tarfile.OutsideDestinationError`.
@@ -1115,6 +1119,10 @@ reused in custom filters:
 
   Return the modified ``TarInfo`` member.
 
+  .. versionchanged:: next
+
+     Filenames containing ``..`` components are now normalized.
+
 .. function:: data_filter(member, path)
 
   Implements the ``'data'`` filter.

Doc/library/xml.dom.minidom.rst

@@ -264,6 +264,11 @@ rules apply:
 * :attr:`~xml.dom.Document.strictErrorChecking` and
   :attr:`~xml.dom.Attr.specified` are always ``False``.
 
+* :meth:`~xml.dom.Element.removeAttribute` and
+  :meth:`~xml.dom.Element.removeAttributeNS` raise
+  :exc:`~xml.dom.NotFoundErr` if there is no matching attribute,
+  while the DOM specifies that this has no effect.
+
 The following interfaces have no implementation in :mod:`!xml.dom.minidom`:
 
 * :class:`DOMTimeStamp`

Doc/library/xml.dom.rst

@@ -153,6 +153,11 @@ Objects in the DOM
 
 The definitive documentation for the DOM is the DOM specification from the W3C.
 
+The names documented in this section are DOM interfaces.
+With the exception of :class:`Node` and the exception classes,
+they are not provided by the :mod:`!xml.dom` module itself,
+but by concrete DOM implementations, such as :mod:`xml.dom.minidom`.
+
 Note that DOM attributes may also be manipulated as nodes instead of as simple
 strings.  It is fairly rare that you must do this, however, so this usage is not
 yet documented.
@@ -461,23 +466,23 @@ for each node type are:
 .. method:: Node.insertBefore(newChild, refChild)
 
    Insert a new child node before an existing child.  It must be the case that
-   *refChild* is a child of this node; if not, :exc:`ValueError` is raised.
+   *refChild* is a child of this node; if not, :exc:`NotFoundErr` is raised.
    *newChild* is returned. If *refChild* is ``None``, it inserts *newChild* at the
    end of the children's list.
 
 
 .. method:: Node.removeChild(oldChild)
 
    Remove a child node.  *oldChild* must be a child of this node; if not,
-   :exc:`ValueError` is raised.  *oldChild* is returned on success.  If *oldChild*
+   :exc:`NotFoundErr` is raised.  *oldChild* is returned on success.  If *oldChild*
    will not be used further, its :meth:`~xml.dom.minidom.Node.unlink` method
    should be called.
 
 
 .. method:: Node.replaceChild(newChild, oldChild)
 
    Replace an existing node with a new node. It must be the case that  *oldChild*
-   is a child of this node; if not, :exc:`ValueError` is raised.
+   is a child of this node; if not, :exc:`NotFoundErr` is raised.
 
 
 .. method:: Node.normalize()
@@ -826,8 +831,7 @@ of that class.
 
 .. method:: Element.removeAttribute(name)
 
-   Remove an attribute by name.  If there is no matching attribute, a
-   :exc:`NotFoundErr` is raised.
+   Remove an attribute by name.
 
 
 .. method:: Element.removeAttributeNode(oldAttr)
@@ -838,8 +842,7 @@ of that class.
 
 .. method:: Element.removeAttributeNS(namespaceURI, localName)
 
-   Remove an attribute by name.  Note that it uses a localName, not a qname.  No
-   exception is raised if there is no matching attribute.
+   Remove an attribute by name.  Note that it uses a localName, not a qname.
 
 
 .. method:: Element.setAttribute(name, value)

Include/internal/pycore_pystate.h

@@ -360,6 +360,10 @@ struct PyInterpreterView {
 PyAPI_FUNC(Py_ssize_t) _PyInterpreterState_GuardCountdown(PyInterpreterState *interp);
 PyAPI_FUNC(PyInterpreterState *) _PyInterpreterGuard_GetInterpreter(PyInterpreterGuard *guard);
 
+extern int _PyInterpreterGuard_TryAcquire(PyInterpreterState *interp,
+                                          PyInterpreterGuard *guard);
+extern void _PyInterpreterGuard_Release(PyInterpreterGuard *guard);
+
 #ifdef __cplusplus
 }
 #endif

Include/internal/pycore_tracemalloc.h

@@ -44,9 +44,10 @@ struct
 __attribute__((packed))
 #endif
 tracemalloc_frame {
-    /* filename cannot be NULL: "<unknown>" is used if the Python frame
-       filename is NULL */
-    PyObject *filename;
+    /* Interned NUL terminated UTF-8 (surrogatepass) string.
+       Cannot be NULL: "<unknown>" is used if the Python frame filename
+       cannot be captured. */
+    const char *filename;
     unsigned int lineno;
 };
 
@@ -85,7 +86,7 @@ struct _tracemalloc_runtime_state {
        Protected by TABLES_LOCK(). */
     size_t peak_traced_memory;
     /* Hash table used as a set to intern filenames:
-       PyObject* => PyObject*.
+       char* (NUL terminated UTF-8 string) => NULL.
        Protected by the TABLES_LOCK(). */
     _Py_hashtable_t *filenames;
     /* Buffer to store a new traceback in traceback_new().

Lib/asyncio/graph.py

@@ -58,7 +58,8 @@ def _build_graph_for_future(
     while coro is not None:
         if hasattr(coro, 'cr_await'):
             # A native coroutine or duck-type compatible iterator
-            st.append(FrameCallGraphEntry(coro.cr_frame))
+            if coro.cr_frame is not None:
+                st.append(FrameCallGraphEntry(coro.cr_frame))
             coro = coro.cr_await
         elif hasattr(coro, 'ag_await'):
             # A native async generator or duck-type compatible iterator

Lib/netrc.py

@@ -48,11 +48,12 @@ def _read_char(self):
     def get_token(self):
         if self.pushback:
             return self.pushback.pop(0)
-        token = ""
+        token = None
         fiter = iter(self._read_char, "")
         for ch in fiter:
             if ch in self.whitespace:
                 continue
+            token = ""
             if ch == '"':
                 for ch in fiter:
                     if ch == '"':
@@ -96,9 +97,9 @@ def _parse(self, file, fp, default_netrc):
             # Look for a machine, default, or macdef top-level keyword
             saved_lineno = lexer.lineno
             tt = lexer.get_token()
-            if not tt:
+            if tt is None:
                 break
-            elif tt[0] == '#':
+            elif tt.startswith('#'):
                 if lexer.lineno == saved_lineno and len(tt) == 1:
                     lexer.instream.readline()
                 continue
@@ -135,20 +136,20 @@ def _parse(self, file, fp, default_netrc):
             while 1:
                 prev_lineno = lexer.lineno
                 tt = lexer.get_token()
-                if tt.startswith('#'):
+                if tt is not None and tt.startswith('#'):
                     if lexer.lineno == prev_lineno:
                         lexer.instream.readline()
                     continue
-                if tt in {'', 'machine', 'default', 'macdef'}:
+                if tt in {None, 'machine', 'default', 'macdef'}:
                     self.hosts[entryname] = (login, account, password)
                     lexer.push_token(tt)
                     break
                 elif tt == 'login' or tt == 'user':
-                    login = lexer.get_token()
+                    login = lexer.get_token() or ''
                 elif tt == 'account':
-                    account = lexer.get_token()
+                    account = lexer.get_token() or ''
                 elif tt == 'password':
-                    password = lexer.get_token()
+                    password = lexer.get_token() or ''
                 else:
                     raise NetrcParseError("bad follower token %r" % tt,
                                           file, lexer.lineno)

Lib/tarfile.py

@@ -833,6 +833,13 @@ def _get_filtered_attrs(member, dest_path, for_data=True):
         # For example, 'C:/foo' on Windows.
         raise AbsolutePathError(member)
     # Ensure we stay in the destination
+    if '..' in name.replace(os.sep, '/').split('/'):
+        # Directories are created from the name as given, so a name that
+        # leaves the destination part-way through would create them
+        # outside it even if the resolved path stays inside.
+        normalized = os.path.normpath(name)
+        if normalized != name:
+            name = new_attrs['name'] = normalized
     target_path = os.path.realpath(os.path.join(dest_path, name),
                                    strict=os.path.ALLOW_MISSING)
     if os.path.commonpath([target_path, dest_path]) != dest_path:

Lib/test/support/os_helper.py

@@ -434,7 +434,10 @@ def _rmtree_inner(path):
                           file=sys.__stderr__)
                     mode = 0
                 if stat.S_ISDIR(mode):
-                    _waitfor(_rmtree_inner, fullname, waitall=True)
+                    # Do not follow junctions, which os.lstat() reports
+                    # as directories.
+                    if not os.path.isjunction(fullname):
+                        _waitfor(_rmtree_inner, fullname, waitall=True)
                     _force_run(fullname, os.rmdir, fullname)
                 else:
                     _force_run(fullname, os.unlink, fullname)