Add private _PyUnicode_AsUTF8NoNUL() function (GH-111957)

Like PyUnicode_AsUTF8(), but check for embedded null characters.
This commit is contained in:
Serhiy Storchaka 2023-11-10 21:31:36 +02:00 committed by GitHub
parent 3932b0f7b1
commit 771bd3c94a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 18 additions and 13 deletions

View file

@ -434,6 +434,10 @@ struct _Py_unicode_state {
extern void _PyUnicode_InternInPlace(PyInterpreterState *interp, PyObject **p);
extern void _PyUnicode_ClearInterned(PyInterpreterState *interp);
// Like PyUnicode_AsUTF8(), but check for embedded null characters.
// Export for '_sqlite3' shared extension.
PyAPI_FUNC(const char *) _PyUnicode_AsUTF8NoNUL(PyObject *);
#ifdef __cplusplus
}

View file

@ -1020,15 +1020,10 @@ io_check_errors(PyObject *errors)
return 0;
}
Py_ssize_t name_length;
const char *name = PyUnicode_AsUTF8AndSize(errors, &name_length);
const char *name = _PyUnicode_AsUTF8NoNUL(errors);
if (name == NULL) {
return -1;
}
if (strlen(name) != (size_t)name_length) {
PyErr_SetString(PyExc_ValueError, "embedded null character in errors");
return -1;
}
PyObject *handler = PyCodec_LookupError(name);
if (handler != NULL) {
Py_DECREF(handler);

View file

@ -76,16 +76,10 @@ isolation_level_converter(PyObject *str_or_none, const char **result)
*result = NULL;
}
else if (PyUnicode_Check(str_or_none)) {
Py_ssize_t sz;
const char *str = PyUnicode_AsUTF8AndSize(str_or_none, &sz);
const char *str = _PyUnicode_AsUTF8NoNUL(str_or_none);
if (str == NULL) {
return 0;
}
if (strlen(str) != (size_t)sz) {
PyErr_SetString(PyExc_ValueError, "embedded null character");
return 0;
}
const char *level = get_isolation_level(str);
if (level == NULL) {
return 0;

View file

@ -3847,6 +3847,18 @@ PyUnicode_AsUTF8(PyObject *unicode)
return PyUnicode_AsUTF8AndSize(unicode, NULL);
}
const char *
_PyUnicode_AsUTF8NoNUL(PyObject *unicode)
{
Py_ssize_t size;
const char *s = PyUnicode_AsUTF8AndSize(unicode, &size);
if (s && strlen(s) != (size_t)size) {
PyErr_SetString(PyExc_ValueError, "embedded null character");
return NULL;
}
return s;
}
/*
PyUnicode_GetSize() has been deprecated since Python 3.3
because it returned length of Py_UNICODE.