objects: Allow tree entries to be sorted by name.

Change-Id: I31e00df32b2a331337b577a55311c5764c04c37c

dulwich/_objects.c

@@ -146,18 +146,32 @@ int cmp_tree_item(const void *_a, const void *_b)
 	return strcmp(remain_a, remain_b);
 }
 
-static PyObject *py_sorted_tree_items(PyObject *self, PyObject *entries)
+int cmp_tree_item_name_order(const void *_a, const void *_b) {
+	const struct tree_item *a = _a, *b = _b;
+	return strcmp(a->name, b->name);
+}
+
+static PyObject *py_sorted_tree_items(PyObject *self, PyObject *args)
 {
 	struct tree_item *qsort_entries = NULL;
-	int num_entries, n = 0, i;
-	PyObject *ret, *key, *value, *py_mode, *py_sha;
+	int name_order, num_entries, n = 0, i;
+	PyObject *entries, *py_name_order, *ret, *key, *value, *py_mode, *py_sha;
 	Py_ssize_t pos = 0;
+	int (*cmp)(const void *, const void *);
+
+	if (!PyArg_ParseTuple(args, "OO", &entries, &py_name_order))
+		goto error;
 
 	if (!PyDict_Check(entries)) {
 		PyErr_SetString(PyExc_TypeError, "Argument not a dictionary");
 		goto error;
 	}
 
+	name_order = PyObject_IsTrue(py_name_order);
+	if (name_order == -1)
+		goto error;
+	cmp = name_order ? cmp_tree_item_name_order : cmp_tree_item;
+
 	num_entries = PyDict_Size(entries);
 	if (PyErr_Occurred())
 		goto error;
@@ -199,7 +213,7 @@ static PyObject *py_sorted_tree_items(PyObject *self, PyObject *entries)
 		n++;
 	}
 
-	qsort(qsort_entries, num_entries, sizeof(struct tree_item), cmp_tree_item);
+	qsort(qsort_entries, num_entries, sizeof(struct tree_item), cmp);
 
 	ret = PyList_New(num_entries);
 	if (ret == NULL) {
@@ -223,7 +237,7 @@ error:
 
 static PyMethodDef py_objects_methods[] = {
 	{ "parse_tree", (PyCFunction)py_parse_tree, METH_VARARGS, NULL },
-	{ "sorted_tree_items", (PyCFunction)py_sorted_tree_items, METH_O, NULL },
+	{ "sorted_tree_items", py_sorted_tree_items, METH_VARARGS, NULL },
 	{ NULL, NULL, 0, NULL }
 };
 

dulwich/objects.py

@@ -730,14 +730,17 @@ def serialize_tree(items):
         yield "%04o %s\0%s" % (mode, name, hex_to_sha(hexsha))
 
 
-def sorted_tree_items(entries):
-    """Iterate over a tree entries dictionary in the order in which
-    the items would be serialized.
+def sorted_tree_items(entries, name_order):
+    """Iterate over a tree entries dictionary.
 
+    :param name_order: If True, iterate entries in order of their name. If
+        False, iterate entries in tree order, that is, treat subtree entries as
+        having '/' appended.
     :param entries: Dictionary mapping names to (mode, sha) tuples
     :return: Iterator over (name, mode, hexsha)
     """
-    for name, entry in sorted(entries.iteritems(), cmp=cmp_entry):
+    cmp_func = name_order and cmp_entry_name_order or cmp_entry
+    for name, entry in sorted(entries.iteritems(), cmp=cmp_func):
         mode, hexsha = entry
         # Stricter type checks than normal to mirror checks in the C version.
         mode = int(mode)
@@ -747,7 +750,7 @@ def sorted_tree_items(entries):
 
 
 def cmp_entry((name1, value1), (name2, value2)):
-    """Compare two tree entries."""
+    """Compare two tree entries in tree order."""
     if stat.S_ISDIR(value1[0]):
         name1 += "/"
     if stat.S_ISDIR(value2[0]):
@@ -755,6 +758,11 @@ def cmp_entry((name1, value1), (name2, value2)):
     return cmp(name1, name2)
 
 
+def cmp_entry_name_order(entry1, entry2):
+    """Compare two tree entries in name order."""
+    return cmp(entry1[0], entry2[0])
+
+
 class Tree(ShaFile):
     """A Git tree object"""
 
@@ -833,13 +841,14 @@ class Tree(ShaFile):
         return [
             (mode, name, hexsha) for (name, mode, hexsha) in self.iteritems()]
 
-    def iteritems(self):
-        """Iterate over entries in the order in which they would be serialized.
+    def iteritems(self, name_order=False):
+        """Iterate over entries.
 
+        :param name_order: If True, iterate in name order instead of tree order.
         :return: Iterator over (name, mode, sha) tuples
         """
         self._ensure_parsed()
-        return sorted_tree_items(self._entries)
+        return sorted_tree_items(self._entries, name_order)
 
     def items(self):
         """Return the sorted entries in this tree.

dulwich/tests/test_objects.py

@@ -476,7 +476,7 @@ class TreeTests(ShaFileCheckTests):
 
     def _do_test_sorted_tree_items(self, sorted_tree_items):
         def do_sort(entries):
-            return list(sorted_tree_items(entries))
+            return list(sorted_tree_items(entries, False))
 
         actual = do_sort(_TREE_ITEMS)
         self.assertEqual(_SORTED_TREE_ITEMS, actual)
@@ -502,6 +502,23 @@ class TreeTests(ShaFileCheckTests):
             raise TestSkipped('sorted_tree_items extension not found')
         self._do_test_sorted_tree_items(sorted_tree_items)
 
+    def _do_test_sorted_tree_items_name_order(self, sorted_tree_items):
+        self.assertEqual([
+          TreeEntry('a', stat.S_IFDIR,
+                    'd80c186a03f423a81b39df39dc87fd269736ca86'),
+          TreeEntry('a.c', 0100755, 'd80c186a03f423a81b39df39dc87fd269736ca86'),
+          TreeEntry('a/c', stat.S_IFDIR,
+                    'd80c186a03f423a81b39df39dc87fd269736ca86'),
+          ], list(sorted_tree_items(_TREE_ITEMS, True)))
+
+    def test_sorted_tree_items_name_order(self):
+        self._do_test_sorted_tree_items_name_order(_sorted_tree_items_py)
+
+    def test_sorted_tree_items_name_order_extension(self):
+        if sorted_tree_items is _sorted_tree_items_py:
+            raise TestSkipped('sorted_tree_items extension not found')
+        self._do_test_sorted_tree_items_name_order(sorted_tree_items)
+
     def test_check(self):
         t = Tree
         sha = hex_to_sha(a_sha)