[Python-checkins] cpython (3.2): keep the buffer object around while we're using it (closes #14212)
benjamin.peterson
python-checkins at python.org
Wed Mar 7 21:59:27 CET 2012
http://hg.python.org/cpython/rev/10a79a33d09b
changeset: 75476:10a79a33d09b
branch: 3.2
parent: 75467:7b65ea9c9972
user: Benjamin Peterson <benjamin at python.org>
date: Wed Mar 07 14:50:25 2012 -0600
summary:
keep the buffer object around while we're using it (closes #14212)
files:
Lib/test/test_re.py | 14 +++++-
Misc/NEWS | 3 +
Modules/_sre.c | 82 +++++++++++++++++++-------------
Modules/sre.h | 2 +
4 files changed, 66 insertions(+), 35 deletions(-)
diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
--- a/Lib/test/test_re.py
+++ b/Lib/test/test_re.py
@@ -1,4 +1,5 @@
-from test.support import verbose, run_unittest
+from test.support import verbose, run_unittest, gc_collect
+import io
import re
from re import Scanner
import sys
@@ -16,6 +17,17 @@
class ReTests(unittest.TestCase):
+ def test_keep_buffer(self):
+ # See bug 14212
+ b = bytearray(b'x')
+ it = re.finditer(b'a', b)
+ with self.assertRaises(BufferError):
+ b.extend(b'x'*400)
+ list(it)
+ del it
+ gc_collect()
+ b.extend(b'x'*400)
+
def test_weakref(self):
s = 'QabbbcR'
x = re.compile('ab+c')
diff --git a/Misc/NEWS b/Misc/NEWS
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -549,6 +549,9 @@
Extension Modules
-----------------
+- Issue #14212: The re module didn't retain a reference to buffers it was
+ scanning, resulting in segfaults.
+
- Issue #13840: The error message produced by ctypes.create_string_buffer
when given a Unicode string has been fixed.
diff --git a/Modules/_sre.c b/Modules/_sre.c
--- a/Modules/_sre.c
+++ b/Modules/_sre.c
@@ -1664,7 +1664,7 @@
}
static void*
-getstring(PyObject* string, Py_ssize_t* p_length, int* p_charsize)
+getstring(PyObject* string, Py_ssize_t* p_length, int* p_charsize, Py_buffer *view)
{
/* given a python object, return a data pointer, a length (in
characters), and a character size. return NULL if the object
@@ -1674,7 +1674,6 @@
Py_ssize_t size, bytes;
int charsize;
void* ptr;
- Py_buffer view;
/* Unicode objects do not support the buffer API. So, get the data
directly instead. */
@@ -1686,26 +1685,21 @@
}
/* get pointer to string buffer */
- view.len = -1;
+ view->len = -1;
buffer = Py_TYPE(string)->tp_as_buffer;
if (!buffer || !buffer->bf_getbuffer ||
- (*buffer->bf_getbuffer)(string, &view, PyBUF_SIMPLE) < 0) {
+ (*buffer->bf_getbuffer)(string, view, PyBUF_SIMPLE) < 0) {
PyErr_SetString(PyExc_TypeError, "expected string or buffer");
return NULL;
}
/* determine buffer size */
- bytes = view.len;
- ptr = view.buf;
-
- /* Release the buffer immediately --- possibly dangerous
- but doing something else would require some re-factoring
- */
- PyBuffer_Release(&view);
+ bytes = view->len;
+ ptr = view->buf;
if (bytes < 0) {
PyErr_SetString(PyExc_TypeError, "buffer has negative size");
- return NULL;
+ goto err;
}
/* determine character size */
@@ -1719,7 +1713,7 @@
#endif
else {
PyErr_SetString(PyExc_TypeError, "buffer size mismatch");
- return NULL;
+ goto err;
}
*p_length = size;
@@ -1728,8 +1722,13 @@
if (ptr == NULL) {
PyErr_SetString(PyExc_ValueError,
"Buffer is NULL");
+ goto err;
}
return ptr;
+ err:
+ PyBuffer_Release(view);
+ view->buf = NULL;
+ return NULL;
}
LOCAL(PyObject*)
@@ -1747,20 +1746,21 @@
state->lastmark = -1;
state->lastindex = -1;
- ptr = getstring(string, &length, &charsize);
+ state->buffer.buf = NULL;
+ ptr = getstring(string, &length, &charsize, &state->buffer);
if (!ptr)
- return NULL;
-
- if (charsize == 1 && pattern->charsize > 1) {
- PyErr_SetString(PyExc_TypeError,
+ goto err;
+
+ if (charsize == 1 && pattern->charsize > 1) {
+ PyErr_SetString(PyExc_TypeError,
"can't use a string pattern on a bytes-like object");
- return NULL;
- }
- if (charsize > 1 && pattern->charsize == 1) {
- PyErr_SetString(PyExc_TypeError,
+ goto err;
+ }
+ if (charsize > 1 && pattern->charsize == 1) {
+ PyErr_SetString(PyExc_TypeError,
"can't use a bytes pattern on a string-like object");
- return NULL;
- }
+ goto err;
+ }
/* adjust boundaries */
if (start < 0)
@@ -1797,11 +1797,17 @@
state->lower = sre_lower;
return string;
+ err:
+ if (state->buffer.buf)
+ PyBuffer_Release(&state->buffer);
+ return NULL;
}
LOCAL(void)
state_fini(SRE_STATE* state)
{
+ if (state->buffer.buf)
+ PyBuffer_Release(&state->buffer);
Py_XDECREF(state->string);
data_stack_dealloc(state);
}
@@ -1863,6 +1869,8 @@
{
if (self->weakreflist != NULL)
PyObject_ClearWeakRefs((PyObject *) self);
+ if (self->view.buf)
+ PyBuffer_Release(&self->view);
Py_XDECREF(self->pattern);
Py_XDECREF(self->groupindex);
Py_XDECREF(self->indexgroup);
@@ -2297,6 +2305,7 @@
Py_ssize_t i, b, e;
int bint;
int filter_is_callable;
+ Py_buffer view;
if (PyCallable_Check(ptemplate)) {
/* sub/subn takes either a function or a template */
@@ -2306,7 +2315,8 @@
} else {
/* if not callable, check if it's a literal string */
int literal;
- ptr = getstring(ptemplate, &n, &bint);
+ view.buf = NULL;
+ ptr = getstring(ptemplate, &n, &bint, &view);
b = bint;
if (ptr) {
if (b == 1) {
@@ -2320,6 +2330,8 @@
PyErr_Clear();
literal = 0;
}
+ if (view.buf)
+ PyBuffer_Release(&view);
if (literal) {
filter = ptemplate;
Py_INCREF(filter);
@@ -2661,6 +2673,7 @@
Py_ssize_t groups = 0;
PyObject* groupindex = NULL;
PyObject* indexgroup = NULL;
+
if (!PyArg_ParseTuple(args, "OiO!|nOO", &pattern, &flags,
&PyList_Type, &code, &groups,
&groupindex, &indexgroup))
@@ -2675,6 +2688,7 @@
self->pattern = NULL;
self->groupindex = NULL;
self->indexgroup = NULL;
+ self->view.buf = NULL;
self->codesize = n;
@@ -2694,15 +2708,15 @@
return NULL;
}
- if (pattern == Py_None)
- self->charsize = -1;
- else {
- Py_ssize_t p_length;
- if (!getstring(pattern, &p_length, &self->charsize)) {
- Py_DECREF(self);
- return NULL;
- }
- }
+ if (pattern == Py_None)
+ self->charsize = -1;
+ else {
+ Py_ssize_t p_length;
+ if (!getstring(pattern, &p_length, &self->charsize, &self->view)) {
+ Py_DECREF(self);
+ return NULL;
+ }
+ }
Py_INCREF(pattern);
self->pattern = pattern;
diff --git a/Modules/sre.h b/Modules/sre.h
--- a/Modules/sre.h
+++ b/Modules/sre.h
@@ -31,6 +31,7 @@
int flags; /* flags used when compiling pattern source */
PyObject *weakreflist; /* List of weak references */
int charsize; /* pattern charsize (or -1) */
+ Py_buffer view;
/* pattern code */
Py_ssize_t codesize;
SRE_CODE code[1];
@@ -80,6 +81,7 @@
char* data_stack;
size_t data_stack_size;
size_t data_stack_base;
+ Py_buffer buffer;
/* current repeat context */
SRE_REPEAT *repeat;
/* hooks */
--
Repository URL: http://hg.python.org/cpython
More information about the Python-checkins
mailing list