blob: 0599821afc9acd17832db57aec67bebc0b7a990b [file] [log] [blame]
// Copyright 2013 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#include <vector>
#include "ui/accessibility/ax_node_data.h"
namespace ui {
// An AXTreeUpdate is a serialized representation of an atomic change
// to an AXTree. The sender and receiver must be in sync; the update
// is only meant to bring the tree from a specific previous state into
// its next state. Trying to apply it to the wrong tree should immediately
// die with a fatal assertion. An AXTreeUpdate is just an ordered vector
// of AXNodeData structures to be applied to the tree in order.
// Suppose that the next AXNodeData to be applied is |node|. The following
// invariants must hold:
// 1. Either || is already in the tree, or else |node| is the new
// root of the tree and |node.role| == WebAXRoleRootWebArea.
// 2. Every child id in |node.child_ids| must either be already a child
// of this node, or a new id not previously in the tree. It is not
// allowed to "reparent" a child to this node without first removing
// that child from its previous parent.
// 3. When a new id appears in |node.child_ids|, the tree should create a
// new uninitialized placeholder node for it immediately. That
// placeholder must be updated within the same AXTreeUpdate, otherwise
// it's a fatal error. This guarantees the tree is always complete
// before or after an AXTreeUpdate.
struct AX_EXPORT AXTreeUpdate {
std::vector<AXNodeData> nodes;
// TODO(dmazzoni): location changes
} // namespace ui