2025-05-27 19:09:08 +00:00
committed by Git OBS Bridge
parent 5ffcff295f
commit 3d0b1fd2f3

View File

@@ -20,13 +20,13 @@ Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
---
Include/cpython/bytesobject.h | 4
Include/cpython/unicodeobject.h | 13 ++
Lib/test/test_codeccallbacks.py | 36 +++++
Lib/test/test_codeccallbacks.py | 39 ++++++
Lib/test/test_codecs.py | 52 ++++++--
Misc/NEWS.d/next/Security/2025-05-09-20-22-54.gh-issue-133767.kN2i3Q.rst | 2
Objects/bytesobject.c | 54 +++++---
Objects/unicodeobject.c | 61 +++++++---
Parser/string_parser.c | 26 ++--
8 files changed, 191 insertions(+), 57 deletions(-)
8 files changed, 194 insertions(+), 57 deletions(-)
create mode 100644 Misc/NEWS.d/next/Security/2025-05-09-20-22-54.gh-issue-133767.kN2i3Q.rst
Index: Python-3.12.10/Include/cpython/bytesobject.h
@@ -71,7 +71,7 @@ Index: Python-3.12.10/Include/cpython/unicodeobject.h
Index: Python-3.12.10/Lib/test/test_codeccallbacks.py
===================================================================
--- Python-3.12.10.orig/Lib/test/test_codeccallbacks.py 2025-05-27 17:27:04.334768069 +0200
+++ Python-3.12.10/Lib/test/test_codeccallbacks.py 2025-05-27 20:45:06.747118127 +0200
+++ Python-3.12.10/Lib/test/test_codeccallbacks.py 2025-05-27 21:05:08.061522938 +0200
@@ -1,6 +1,7 @@
import codecs
import html.entities
@@ -89,7 +89,7 @@ Index: Python-3.12.10/Lib/test/test_codeccallbacks.py
baddata = [
("ascii", b"\xff"),
("utf-7", b"++"),
@@ -1159,6 +1160,39 @@
@@ -1159,6 +1160,42 @@
for (encoding, data) in baddata:
self.assertEqual(data.decode(encoding, "test.mutating"), "\u4242")
@@ -114,17 +114,20 @@ Index: Python-3.12.10/Lib/test/test_codeccallbacks.py
+ self.assertEqual(decode(input, 'test.mutating2'), (expected, len(input)))
+ self.assertIn(msg, str(cm.warning))
+
+ # The warning message for invalid escape sequences dynamically escapes the backslash.
+ # Using re.escape to handle the backslash and then formatting the character directly.
+ check(br'\x0n\z', '\u0404\n\\z', r"invalid escape sequence '%s'" % r'\z')
+ check(br'\x0n\501', '\u0404\n\u0141', r"invalid octal escape sequence '\\501'")
+ check(br'\x0n\z', '\u0404\n\\z', r"invalid escape sequence '\z'")
+ check(br'\x0n\501', '\u0404\n\u0141', r"invalid octal escape sequence '\501'")
+ check(br'\x0z', '\u0404\\z', r"invalid escape sequence '\z'")
+
+ check(br'\x0z', '\u0404\\z', r"invalid escape sequence '%s'" % r'\z')
+ check(br'\x3n\zr', '\u0404\n\\zr', r"invalid escape sequence '%s'" % r'\z')
+ check(br'\x3zr', '\u0404\\zr', r"invalid escape sequence '%s'" % r'\z')
+ check(br'\x3z5', '\u0404\\z5', r"invalid escape sequence '%s'" % r'\z')
+ check(memoryview(br'\x3z5x')[:-1], '\u0404\\z5', r"invalid escape sequence '%s'" % r'\z')
+ check(memoryview(br'\x3z5xy')[:-2], '\u0404\\z5', r"invalid escape sequence '%s'" % r'\z')
+ check(br'\x3n\zr', '\u0404\n\\zr', r"invalid escape sequence '\z'")
+ check(br'\x3zr', '\u0404\\zr', r"invalid escape sequence '\z'")
+ check(br'\x3z5', '\u0404\\z5', r"invalid escape sequence '\z'")
+ check(memoryview(br'\x3z5x')[:-1], '\u0404\\z5', r"invalid escape sequence '\z'")
+ check(memoryview(br'\x3z5xy')[:-2], '\u0404\\z5', r"invalid escape sequence '\z'")
+
+ check(br'\x5n\z', '\u0404\n\\z', r"invalid escape sequence '\z'")
+ check(br'\x5n\501', '\u0404\n\u0141', r"invalid octal escape sequence '\501'")
+ check(br'\x5z', '\u0404\\z', r"invalid escape sequence '\z'")
+ check(memoryview(br'\x5zy')[:-1], '\u0404\\z', r"invalid escape sequence '\z'")
+
# issue32583
def test_crashing_decode_handler(self):
@@ -234,7 +237,7 @@ Index: Python-3.12.10/Misc/NEWS.d/next/Security/2025-05-09-20-22-54.gh-issue-133
Index: Python-3.12.10/Objects/bytesobject.c
===================================================================
--- Python-3.12.10.orig/Objects/bytesobject.c 2025-04-08 13:35:47.000000000 +0200
+++ Python-3.12.10/Objects/bytesobject.c 2025-05-27 17:27:08.581189399 +0200
+++ Python-3.12.10/Objects/bytesobject.c 2025-05-27 21:08:20.056554751 +0200
@@ -1048,10 +1048,11 @@
}
@@ -328,7 +331,7 @@ Index: Python-3.12.10/Objects/bytesobject.c
if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1,
- "invalid octal escape sequence '\\%.3s'",
- first_invalid_escape) < 0)
+ "invalid octal escape sequence '\\%o'",
+ "invalid octal escape sequence '\\\\%o'",
+ first_invalid_escape_char) < 0)
{
Py_DECREF(result);
@@ -345,7 +348,7 @@ Index: Python-3.12.10/Objects/bytesobject.c
Index: Python-3.12.10/Objects/unicodeobject.c
===================================================================
--- Python-3.12.10.orig/Objects/unicodeobject.c 2025-04-08 13:35:47.000000000 +0200
+++ Python-3.12.10/Objects/unicodeobject.c 2025-05-27 17:27:08.583127021 +0200
+++ Python-3.12.10/Objects/unicodeobject.c 2025-05-27 21:08:02.745636052 +0200
@@ -6046,13 +6046,15 @@
/* --- Unicode Escape Codec ----------------------------------------------- */
@@ -450,7 +453,7 @@ Index: Python-3.12.10/Objects/unicodeobject.c
if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1,
- "invalid octal escape sequence '\\%.3s'",
- first_invalid_escape) < 0)
+ "invalid octal escape sequence '\\%o'",
+ "invalid octal escape sequence '\\\\%o'",
+ first_invalid_escape_char) < 0)
{
Py_DECREF(result);