Add custom action support

ui/accessibility/ax_enums.idl

 // Copyright 2014 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.
 
 // TODO(nektar): Migrate entire file to Mojoq.
 // These should be kept in sync with third_party/WebKit/public/web/WebAXEnums.h
 // until the Chromium and Blink trees are merged.
 [camel_case_enum_to_string=true] namespace ui {
 
   // For new entries to the following four enums, also add to
@@ skipping 514 matching lines @@
 
     // Used for caching. Do not read directly. Use
     // |AXNode::GetOrComputeLineStartOffsets|
     // For all text fields and content editable roots: A list of the start
     // offsets of each line inside this object.
     cached_line_starts,
 
     // For inline text. These int lists must be the same size; they represent
     // the start and end character offset of each word within this text.
     word_starts,
-    word_ends
+    word_ends,
+
+    custom_action_ids

[David Tseng, 2017/05/15 15:20:00]

nit: .. // arc

+  };
+
+  [cpp_enum_prefix_override="ax_attr"] enum AXStringListAttribute {
+    custom_action_descriptions

[David Tseng, 2017/05/15 15:20:00]

You can avoid doing this if you delimit via commas and add this as a string
attribute. You could then avoid adding string attribute support throughout

 Optional though, as I think this might be useful for other attributes in the
future.

[yawano, 2017/05/15 23:18:16]

I think adding string list attribute is more straightforward. An application can
use arbitrary characters for its description including commas. If we use comma
separated list, we need to workaround for those descriptions.

[David Tseng, 2017/05/16 22:18:55]

Fair enough. You'll have to add string list attribute support through.

   };
 
   enum AXMarkerType {
     // Assignments are ignored by the parser, but are kept here for clarity.
     spelling = 1,
     grammar = 2,
     spelling_grammar = 3,
     text_match = 4,
     spelling_text_match = 5,
     grammar_text_match = 6,
@@ skipping 117 matching lines @@
     // First node is before the second one.
     before,
 
     // Nodes are the same.
     equal,
 
     // First node is after the second one.
     after
   };
 };