Implemented list batching in cPickle.

Lib/pickle.py

@@ -612,6 +612,7 @@ class Pickler:
 
     dispatch[ListType] = save_list
 
+    # Keep in synch with cPickle's BATCHSIZE.
     _BATCHSIZE = 1000
 
     def _batch_appends(self, items):

Lib/test/pickletester.py

@@ -672,23 +672,6 @@ class AbstractPickleTests(unittest.TestCase):
         self.produce_global_ext(0x7fffffff, pickle.EXT4)  # largest EXT4 code
         self.produce_global_ext(0x12abcdef, pickle.EXT4)  # check endianness
 
-# XXX Temporary hack, so long as the C implementation of pickle protocol
-# XXX 2 isn't ready.  When it is, move the methods in TempAbstractPickleTests
-# XXX into AbstractPickleTests above, and get rid of TempAbstractPickleTests
-# XXX along with the references to it in test_pickle.py.
-class TempAbstractPickleTests(unittest.TestCase):
-
-    def test_newobj_list_slots(self):
-        x = SlotList([1, 2, 3])
-        x.foo = 42
-        x.bar = "hello"
-        s = self.dumps(x, 2)
-        y = self.loads(s)
-        self.assertEqual(list(x), list(y))
-        self.assertEqual(x.__dict__, y.__dict__)
-        self.assertEqual(x.foo, y.foo)
-        self.assertEqual(x.bar, y.bar)
-
     def test_list_chunking(self):
         n = 10  # too small to chunk
         x = range(n)
@@ -711,6 +694,23 @@ class TempAbstractPickleTests(unittest.TestCase):
             else:
                 self.failUnless(num_appends >= 2)
 
+# XXX Temporary hack, so long as the C implementation of pickle protocol
+# XXX 2 isn't ready.  When it is, move the methods in TempAbstractPickleTests
+# XXX into AbstractPickleTests above, and get rid of TempAbstractPickleTests
+# XXX along with the references to it in test_pickle.py.
+class TempAbstractPickleTests(unittest.TestCase):
+
+    def test_newobj_list_slots(self):
+        x = SlotList([1, 2, 3])
+        x.foo = 42
+        x.bar = "hello"
+        s = self.dumps(x, 2)
+        y = self.loads(s)
+        self.assertEqual(list(x), list(y))
+        self.assertEqual(x.__dict__, y.__dict__)
+        self.assertEqual(x.foo, y.foo)
+        self.assertEqual(x.bar, y.bar)
+
     def test_dict_chunking(self):
         n = 10  # too small to chunk
         x = dict.fromkeys(range(n))

Modules/cPickle.c

@@ -87,6 +87,11 @@ PyDoc_STRVAR(cPickle_module_documentation,
 #undef FALSE
 #define FALSE       "I00\n"
 
+/* Keep in synch with pickle.Pickler._BATCHSIZE.  This is how many elements
+ * batch_{list, dict} pump out before doing APPENDS/SETITEMS.
+ */
+#define BATCHSIZE 1000
+
 static char MARKv = MARK;
 
 static PyObject *PickleError;
@@ -1614,34 +1619,119 @@ save_tuple(Picklerobject *self, PyObject *args)
 	return res;
 }
 
+/* iter is an iterator giving items, and we batch up chunks of
+ *     MARK item item ... item APPENDS
+ * opcode sequences.  Calling code should have arranged to first create an
+ * empty list, or list-like object, for the APPENDS to operate on.
+ * Returns 0 on success, <0 on error.
+ */
+static int
+batch_list(Picklerobject *self, PyObject *iter)
+{
+	PyObject *obj;
+	PyObject *slice[BATCHSIZE];
+	int i, n;
+
+	static char append = APPEND;
+	static char appends = APPENDS;
+
+	assert(iter != NULL);
+
+	if (self->proto == 0) {
+		/* APPENDS isn't available; do one at a time. */
+		for (;;) {
+			obj = PyIter_Next(iter);
+			if (obj == NULL) {
+				if (PyErr_Occurred())
+					return -1;
+				break;
+			}
+			i = save(self, obj, 0);
+			Py_DECREF(obj);
+			if (i < 0)
+				return -1;
+			if (self->write_func(self, &append, 1) < 0)
+				return -1;
+
+		}
+		return 0;
+	}
+
+	/* proto > 0:  write in batches of BATCHSIZE. */
+	do {
+		/* Get next group of (no more than) BATCHSIZE elements. */
+		for (n = 0; n < BATCHSIZE; ++n) {
+			obj = PyIter_Next(iter);
+			if (obj == NULL) {
+				if (PyErr_Occurred())
+					goto BatchFailed;
+				break;
+			}
+			slice[n] = obj;
+		}
+
+		if (n > 1) {
+			/* Pump out MARK, slice[0:n], APPENDS. */
+			if (self->write_func(self, &MARKv, 1) < 0)
+				goto BatchFailed;
+			for (i = 0; i < n; ++i) {
+				if (save(self, slice[i], 0) < 0)
+					goto BatchFailed;
+			}
+			if (self->write_func(self, &appends, 1) < 0)
+				goto BatchFailed;
+		}
+		else if (n == 1) {
+			if (save(self, slice[0], 0) < 0)
+				goto BatchFailed;
+			if (self->write_func(self, &append, 1) < 0)
+				goto BatchFailed;
+		}
+
+		for (i = 0; i < n; ++i) {
+			Py_DECREF(slice[i]);
+		}
+	}while (n == BATCHSIZE);
+	return 0;
+
+BatchFailed:
+	while (--n >= 0) {
+		Py_DECREF(slice[n]);
+	}
+	return -1;
+}
+
 static int
 save_list(Picklerobject *self, PyObject *args)
 {
-	PyObject *element = 0;
+	int res = -1;
-	int s_len, len, i, using_appends, res = -1;
 	char s[3];
+	int len;
+	PyObject *iter;
 
-	static char append = APPEND, appends = APPENDS;
 
 	if (self->fast && !fast_save_enter(self, args))
 		goto finally;
 
+	/* Create an empty list. */
 	if (self->bin) {
 		s[0] = EMPTY_LIST;
-		s_len = 1;
+		len = 1;
 	}
 	else {
 		s[0] = MARK;
 		s[1] = LIST;
-		s_len = 2;
+		len = 2;
 	}
 
+	if (self->write_func(self, s, len) < 0)
+		goto finally;
+
+	/* Get list length, and bow out early if empty. */
 	if ((len = PyList_Size(args)) < 0)
 		goto finally;
 
-	if (self->write_func(self, s, s_len) < 0)
+	/* Memoize. */
-		goto finally;
-
 	if (len == 0) {
 		if (put(self, args) < 0)
 			goto finally;
@@ -1651,29 +1741,12 @@ save_list(Picklerobject *self, PyObject *args)
 			goto finally;
 	}
 
-	if ((using_appends = (self->bin && (len > 1))))
+	/* Materialize the list elements. */
-		if (self->write_func(self, &MARKv, 1) < 0)
+	iter = PyObject_GetIter(args);
-			goto finally;
+	if (iter == NULL)
-
+		goto finally;
-	for (i = 0; i < len; i++) {
+	res = batch_list(self, iter);
-		if (!( element = PyList_GET_ITEM((PyListObject *)args, i)))
+	Py_DECREF(iter);
-			goto finally;
-
-		if (save(self, element, 0) < 0)
-			goto finally;
-
-		if (!using_appends) {
-			if (self->write_func(self, &append, 1) < 0)
-				goto finally;
-		}
-	}
-
-	if (using_appends) {
-		if (self->write_func(self, &appends, 1) < 0)
-			goto finally;
-	}
-
-	res = 0;
 
   finally:
 	if (self->fast && !fast_save_leave(self, args))