Issue #16350: Fix zlib decompressor handling of unused_data with multiple calls to decompress() after EOF.
Patch by Serhiy Storchaka.
diff --git a/Lib/test/test_zlib.py b/Lib/test/test_zlib.py
index 7f63143..eb3dd6e 100644
--- a/Lib/test/test_zlib.py
+++ b/Lib/test/test_zlib.py
@@ -426,6 +426,19 @@
c.flush()
self.assertRaises(ValueError, c.copy)
+ def test_decompress_unused_data(self):
+ # Repeated calls to decompress() after EOF should accumulate data in
+ # dco.unused_data, instead of just storing the arg to the last call.
+ x = zlib.compress(HAMLET_SCENE) + HAMLET_SCENE
+ for step in 1, 2, 100:
+ dco = zlib.decompressobj()
+ data = b''.join(dco.decompress(x[i : i + step])
+ for i in range(0, len(x), step))
+ data += dco.flush()
+
+ self.assertEqual(data, HAMLET_SCENE)
+ self.assertEqual(dco.unused_data, HAMLET_SCENE)
+
if hasattr(zlib.decompressobj(), "copy"):
def test_decompresscopy(self):
# Test copying a decompression object
diff --git a/Misc/NEWS b/Misc/NEWS
index 883bd93..b2ad14b 100644
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -137,6 +137,10 @@
Library
-------
+- Issue #16350: zlib.Decompress.decompress() now accumulates data from
+ successive calls after EOF in unused_data, instead of only saving the argument
+ to the last call. Patch by Serhiy Storchaka.
+
- Issue #12759: sre_parse now raises a proper error when the name of the group
is missing. Initial patch by Serhiy Storchaka.
diff --git a/Modules/zlibmodule.c b/Modules/zlibmodule.c
index 035aa8e..95e85b5 100644
--- a/Modules/zlibmodule.c
+++ b/Modules/zlibmodule.c
@@ -566,12 +566,29 @@
preserved.
*/
if (err == Z_STREAM_END) {
- Py_XDECREF(self->unused_data); /* Free original empty string */
- self->unused_data = PyString_FromStringAndSize(
- (char *)self->zst.next_in, self->zst.avail_in);
- if (self->unused_data == NULL) {
- Py_DECREF(RetVal);
- goto error;
+ if (self->zst.avail_in > 0) {
+ /* Append the leftover data to the existing value of unused_data. */
+ Py_ssize_t old_size = PyString_GET_SIZE(self->unused_data);
+ Py_ssize_t new_size = old_size + self->zst.avail_in;
+ PyObject *new_data;
+ if (new_size <= old_size) { /* Check for overflow. */
+ PyErr_NoMemory();
+ Py_DECREF(RetVal);
+ RetVal = NULL;
+ goto error;
+ }
+ new_data = PyString_FromStringAndSize(NULL, new_size);
+ if (new_data == NULL) {
+ Py_DECREF(RetVal);
+ RetVal = NULL;
+ goto error;
+ }
+ Py_MEMCPY(PyString_AS_STRING(new_data),
+ PyString_AS_STRING(self->unused_data), old_size);
+ Py_MEMCPY(PyString_AS_STRING(new_data) + old_size,
+ self->zst.next_in, self->zst.avail_in);
+ Py_DECREF(self->unused_data);
+ self->unused_data = new_data;
}
/* We will only get Z_BUF_ERROR if the output buffer was full
but there wasn't more output when we tried again, so it is