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..8162a61e9707737957ab72e931d98a0f93592c0c
--- /dev/null
+++ b/headless/public/util/dom_tree_extractor.h
@@ -0,0 +1,66 @@
+// 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).

[Sami, 2016/09/30 10:56:03]

Could you briefly document what the received structure looks like?

[alex clarke (OOO till 29th), 2016/09/30 13:16:33]

Is that still needed? It should be obvious now.

+class DomTreeExtractor : public dom::Observer {
+ public:
+  explicit DomTreeExtractor(HeadlessDevToolsClient* devtools_client);
+  ~DomTreeExtractor() override;
+
+  using DomResultCB =
+      base::Callback<void(std::vector<std::unique_ptr<base::DictionaryValue>>)>;
+
+  // Extracts all nodes from the DOM.  This is an asynchronous operation and its
+  // an error to call ExtractDom while a previous operation is in flight.
+  void ExtractDom(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 ExtractDomTree();
+
+  DomResultCB callback_;
+
+  using NodeId = int;
+  using Index = size_t;
+
+  std::unique_ptr<dom::GetDocumentResult> document_result_;
+  std::unique_ptr<dom::GetLayoutTreeNodesResult> layout_tree_result_;
+  std::vector<const dom::Node*> nodes_;
+  std::unordered_map<NodeId, size_t> node_id_to_index_;
+  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_