gh-96670: Raise SyntaxError when parsing NULL bytes (#97594)

This commit is contained in:
Pablo Galindo Salgado 2022-09-27 23:23:42 +01:00 committed by GitHub
parent dd53b79de0
commit aab01e3524
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
10 changed files with 65 additions and 21 deletions

View file

@ -86,6 +86,12 @@ Other Language Changes
* :class:`memoryview` now supports the half-float type (the "e" format code). * :class:`memoryview` now supports the half-float type (the "e" format code).
(Contributed by Dong-hee Na and Antoine Pitrou in :gh:`90751`.) (Contributed by Dong-hee Na and Antoine Pitrou in :gh:`90751`.)
* The parser now raises :exc:`SyntaxError` when parsing source code containing
null bytes. (Contributed by Pablo Galindo in :gh:`96670`.)
* :func:`ast.parse` now raises :exc:`SyntaxError` instead of :exc:`ValueError`
when parsing source code containing null bytes. (Contributed by Pablo Galindo
in :gh:`96670`.)
New Modules New Modules
=========== ===========

View file

@ -3,6 +3,7 @@
#endif #endif
PyAPI_FUNC(char *) Py_UniversalNewlineFgets(char *, int, FILE*, PyObject *); PyAPI_FUNC(char *) Py_UniversalNewlineFgets(char *, int, FILE*, PyObject *);
PyAPI_FUNC(char *) _Py_UniversalNewlineFgetsWithSize(char *, int, FILE*, PyObject *, size_t*);
/* The std printer acts as a preliminary sys.stderr until the new io /* The std printer acts as a preliminary sys.stderr until the new io
infrastructure is in place. */ infrastructure is in place. */

View file

@ -844,6 +844,10 @@ class AST_Tests(unittest.TestCase):
check_limit("a", "[0]") check_limit("a", "[0]")
check_limit("a", "*a") check_limit("a", "*a")
def test_null_bytes(self):
with self.assertRaises(SyntaxError,
msg="source code string cannot contain null bytes"):
ast.parse("a\0b")
class ASTHelpers_Test(unittest.TestCase): class ASTHelpers_Test(unittest.TestCase):
maxDiff = None maxDiff = None

View file

@ -334,11 +334,10 @@ class BuiltinTest(unittest.TestCase):
self.assertRaises(TypeError, compile) self.assertRaises(TypeError, compile)
self.assertRaises(ValueError, compile, 'print(42)\n', '<string>', 'badmode') self.assertRaises(ValueError, compile, 'print(42)\n', '<string>', 'badmode')
self.assertRaises(ValueError, compile, 'print(42)\n', '<string>', 'single', 0xff) self.assertRaises(ValueError, compile, 'print(42)\n', '<string>', 'single', 0xff)
self.assertRaises(ValueError, compile, chr(0), 'f', 'exec')
self.assertRaises(TypeError, compile, 'pass', '?', 'exec', self.assertRaises(TypeError, compile, 'pass', '?', 'exec',
mode='eval', source='0', filename='tmp') mode='eval', source='0', filename='tmp')
compile('print("\xe5")\n', '', 'exec') compile('print("\xe5")\n', '', 'exec')
self.assertRaises(ValueError, compile, chr(0), 'f', 'exec') self.assertRaises(SyntaxError, compile, chr(0), 'f', 'exec')
self.assertRaises(ValueError, compile, str('a = 1'), 'f', 'bad') self.assertRaises(ValueError, compile, str('a = 1'), 'f', 'bad')
# test the optimize argument # test the optimize argument

View file

@ -657,6 +657,18 @@ class CmdLineTest(unittest.TestCase):
], ],
) )
def test_syntaxerror_null_bytes(self):
script = "x = '\0' nothing to see here\n';import os;os.system('echo pwnd')\n"
with os_helper.temp_dir() as script_dir:
script_name = _make_test_script(script_dir, 'script', script)
exitcode, stdout, stderr = assert_python_failure(script_name)
self.assertEqual(
stderr.splitlines()[-2:],
[ b" x = '",
b'SyntaxError: source code cannot contain null bytes'
],
)
def test_consistent_sys_path_for_direct_execution(self): def test_consistent_sys_path_for_direct_execution(self):
# This test case ensures that the following all give the same # This test case ensures that the following all give the same
# sys.path configuration: # sys.path configuration:

View file

@ -544,7 +544,7 @@ if 1:
with open(fn, "wb") as fp: with open(fn, "wb") as fp:
fp.write(src) fp.write(src)
res = script_helper.run_python_until_end(fn)[0] res = script_helper.run_python_until_end(fn)[0]
self.assertIn(b"Non-UTF-8", res.err) self.assertIn(b"source code cannot contain null bytes", res.err)
def test_yet_more_evil_still_undecodable(self): def test_yet_more_evil_still_undecodable(self):
# Issue #25388 # Issue #25388
@ -554,7 +554,7 @@ if 1:
with open(fn, "wb") as fp: with open(fn, "wb") as fp:
fp.write(src) fp.write(src)
res = script_helper.run_python_until_end(fn)[0] res = script_helper.run_python_until_end(fn)[0]
self.assertIn(b"Non-UTF-8", res.err) self.assertIn(b"source code cannot contain null bytes", res.err)
@support.cpython_only @support.cpython_only
@unittest.skipIf(support.is_wasi, "exhausts limited stack on WASI") @unittest.skipIf(support.is_wasi, "exhausts limited stack on WASI")
@ -591,9 +591,9 @@ if 1:
def test_null_terminated(self): def test_null_terminated(self):
# The source code is null-terminated internally, but bytes-like # The source code is null-terminated internally, but bytes-like
# objects are accepted, which could be not terminated. # objects are accepted, which could be not terminated.
with self.assertRaisesRegex(ValueError, "cannot contain null"): with self.assertRaisesRegex(SyntaxError, "cannot contain null"):
compile("123\x00", "<dummy>", "eval") compile("123\x00", "<dummy>", "eval")
with self.assertRaisesRegex(ValueError, "cannot contain null"): with self.assertRaisesRegex(SyntaxError, "cannot contain null"):
compile(memoryview(b"123\x00"), "<dummy>", "eval") compile(memoryview(b"123\x00"), "<dummy>", "eval")
code = compile(memoryview(b"123\x00")[1:-1], "<dummy>", "eval") code = compile(memoryview(b"123\x00")[1:-1], "<dummy>", "eval")
self.assertEqual(eval(code), 23) self.assertEqual(eval(code), 23)

View file

@ -0,0 +1,2 @@
The parser now raises :exc:`SyntaxError` when parsing source code containing
null bytes. Patch by Pablo Galindo

View file

@ -230,16 +230,8 @@ _PyLong_FileDescriptor_Converter(PyObject *o, void *ptr)
return 1; return 1;
} }
/*
** Py_UniversalNewlineFgets is an fgets variation that understands
** all of \r, \n and \r\n conventions.
** The stream should be opened in binary mode.
** The fobj parameter exists solely for legacy reasons and must be NULL.
** Note that we need no error handling: fgets() treats error and eof
** identically.
*/
char * char *
Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj) _Py_UniversalNewlineFgetsWithSize(char *buf, int n, FILE *stream, PyObject *fobj, size_t* size)
{ {
char *p = buf; char *p = buf;
int c; int c;
@ -265,11 +257,28 @@ Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj)
} }
FUNLOCKFILE(stream); FUNLOCKFILE(stream);
*p = '\0'; *p = '\0';
if (p == buf) if (p == buf) {
return NULL; return NULL;
}
*size = p - buf;
return buf; return buf;
} }
/*
** Py_UniversalNewlineFgets is an fgets variation that understands
** all of \r, \n and \r\n conventions.
** The stream should be opened in binary mode.
** The fobj parameter exists solely for legacy reasons and must be NULL.
** Note that we need no error handling: fgets() treats error and eof
** identically.
*/
char *
Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj) {
size_t size;
return _Py_UniversalNewlineFgetsWithSize(buf, n, stream, fobj, &size);
}
/* **************************** std printer **************************** /* **************************** std printer ****************************
* The stdprinter is used during the boot strapping phase as a preliminary * The stdprinter is used during the boot strapping phase as a preliminary
* file like object for sys.stderr. * file like object for sys.stderr.

View file

@ -378,6 +378,11 @@ tok_reserve_buf(struct tok_state *tok, Py_ssize_t size)
return 1; return 1;
} }
static inline int
contains_null_bytes(const char* str, size_t size) {
return memchr(str, 0, size) != NULL;
}
static int static int
tok_readline_recode(struct tok_state *tok) { tok_readline_recode(struct tok_state *tok) {
PyObject *line; PyObject *line;
@ -829,9 +834,9 @@ tok_readline_raw(struct tok_state *tok)
if (!tok_reserve_buf(tok, BUFSIZ)) { if (!tok_reserve_buf(tok, BUFSIZ)) {
return 0; return 0;
} }
char *line = Py_UniversalNewlineFgets(tok->inp, int n_chars = (int)(tok->end - tok->inp);
(int)(tok->end - tok->inp), size_t line_size = 0;
tok->fp, NULL); char *line = _Py_UniversalNewlineFgetsWithSize(tok->inp, n_chars, tok->fp, NULL, &line_size);
if (line == NULL) { if (line == NULL) {
return 1; return 1;
} }
@ -839,7 +844,7 @@ tok_readline_raw(struct tok_state *tok)
tok_concatenate_interactive_new_line(tok, line) == -1) { tok_concatenate_interactive_new_line(tok, line) == -1) {
return 0; return 0;
} }
tok->inp = strchr(tok->inp, '\0'); tok->inp += line_size;
if (tok->inp == tok->buf) { if (tok->inp == tok->buf) {
return 0; return 0;
} }
@ -1075,6 +1080,12 @@ tok_nextc(struct tok_state *tok)
return EOF; return EOF;
} }
tok->line_start = tok->cur; tok->line_start = tok->cur;
if (contains_null_bytes(tok->line_start, tok->inp - tok->line_start)) {
syntaxerror(tok, "source code cannot contain null bytes");
tok->cur = tok->inp;
return EOF;
}
} }
Py_UNREACHABLE(); Py_UNREACHABLE();
} }

View file

@ -1858,7 +1858,7 @@ _Py_SourceAsString(PyObject *cmd, const char *funcname, const char *what, PyComp
} }
if (strlen(str) != (size_t)size) { if (strlen(str) != (size_t)size) {
PyErr_SetString(PyExc_ValueError, PyErr_SetString(PyExc_SyntaxError,
"source code string cannot contain null bytes"); "source code string cannot contain null bytes");
Py_CLEAR(*cmd_copy); Py_CLEAR(*cmd_copy);
return NULL; return NULL;