Add a utility class for extracting details of the DOM

headless/public/util/dom_tree_extractor.h

Index: headless/public/util/dom_tree_extractor.h
diff --git a/headless/public/util/dom_tree_extractor.h b/headless/public/util/dom_tree_extractor.h
new file mode 100644
index 0000000000000000000000000000000000000000..8645b40c98f53a3b7264a8d4bd86648784659f22
--- /dev/null
+++ b/headless/public/util/dom_tree_extractor.h
@@ -0,0 +1,85 @@
+// Copyright 2016 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.
+
+#ifndef HEADLESS_PUBLIC_UTIL_DOM_TREE_EXTRACTOR_H_
+#define HEADLESS_PUBLIC_UTIL_DOM_TREE_EXTRACTOR_H_
+
+#include <unordered_map>
+#include <vector>
+
+#include "base/macros.h"
+#include "headless/public/domains/dom.h"
+
+namespace headless {
+class HeadlessDevToolsClient;
+
+// A utility class for extracting information from the DOM via DevTools. In
+// addition it also extracts details of bounding boxes and layout text (NB the
+// exact layout should not be regarded as stable, it's subject to change without
+// notice).
+class DomTreeExtractor : public dom::Observer {
+ public:
+  explicit DomTreeExtractor(HeadlessDevToolsClient* devtools_client);
+  ~DomTreeExtractor() override;
+
+  using NodeId = int;
+  using Index = size_t;
+
+  class DomTree {
+   public:
+    DomTree();
+    DomTree(DomTree&& other);
+    ~DomTree();
+
+    // Flattened dom tree.
+    std::vector<const dom::Node*> dom_nodes_;

[Sami, 2016/10/03 10:42:18]

The root (document) is always guaranteed to be first, right?

[alex clarke (OOO till 29th), 2016/10/19 16:23:26]

Yes. I've added a comment.

+
+    // Map of node IDs to indexes into |dom_nodes_|.
+    std::unordered_map<NodeId, Index> node_id_to_index_;
+
+    std::vector<const dom::LayoutTreeNode*> layout_tree_nodes_;
+
+   private:
+    friend class DomTreeExtractor;
+
+    // Owns the raw pointers in |dom_nodes_|.
+    std::unique_ptr<dom::GetDocumentResult> document_result_;
+
+    // Owns the raw pointers in |layout_tree_nodes_|.
+    std::unique_ptr<dom::GetLayoutTreeNodesResult> layout_tree_result_;
+  };

[Sami, 2016/10/03 10:42:18]

DISALLOW_COPY_AND_ASSIGN?

[alex clarke (OOO till 29th), 2016/10/19 16:23:26]

Done.

+
+  using DomResultCB = base::Callback<void(DomTree)>;
+
+  // Extracts all nodes from the DOM.  This is an asynchronous operation and its

[Sami, 2016/10/03 10:42:18]

s/its/it's/

[alex clarke (OOO till 29th), 2016/10/19 16:23:26]

Done.

+  // an error to call ExtractDom while a previous operation is in flight.
+  void ExtractDomTree(DomResultCB callback);
+
+  // headless::dom::Observer implementation:
+  void OnSetChildNodes(const dom::SetChildNodesParams& params) override;
+
+ private:
+  void OnRootDocumentFetched(std::unique_ptr<dom::GetDocumentResult> result);
+
+  void OnLayoutTreeNodesFetched(
+      std::unique_ptr<dom::GetLayoutTreeNodesResult> result);
+
+  void MaybeExtractDomTree();
+  void EnumerateNodes(const dom::Node* node);
+  void ExtractLayoutTreeNodes();
+
+  DomResultCB callback_;
+  DomTree dom_tree_;
+  bool child_nodes_fetched_;
+  bool dom_observer_registered_;
+  bool work_in_progress_;
+  HeadlessDevToolsClient* devtools_client_;  // NOT OWNED
+  base::WeakPtrFactory<DomTreeExtractor> weak_factory_;
+
+  DISALLOW_COPY_AND_ASSIGN(DomTreeExtractor);
+};
+
+}  // namespace headless
+
+#endif  // HEADLESS_PUBLIC_UTIL_DOM_TREE_EXTRACTOR_H_