Update ThreatDOMDetails to be able to collect non-resource HTML Elements based on their attributes.

chrome/browser/safe_browsing/threat_details.cc

Index: chrome/browser/safe_browsing/threat_details.cc
diff --git a/chrome/browser/safe_browsing/threat_details.cc b/chrome/browser/safe_browsing/threat_details.cc
index 830b1f5608635bd8c7589b871638058237053d3c..318178672c3ad39bb8a9550f972d77cabd69ab6c 100644
--- a/chrome/browser/safe_browsing/threat_details.cc
+++ b/chrome/browser/safe_browsing/threat_details.cc
@@ -496,8 +496,10 @@ void ThreatDetails::AddDOMDetails(
   for (size_t i = 0; i < params.size() && i < kMaxDomNodes; ++i) {
     SafeBrowsingHostMsg_ThreatDOMDetails_Node node = params[i];
     DVLOG(1) << node.url << ", " << node.tag_name << ", " << node.parent;
-    ClientSafeBrowsingReportRequest::Resource* resource =
-        AddUrl(node.url, node.parent, node.tag_name, &(node.children));
+    ClientSafeBrowsingReportRequest::Resource* resource = nullptr;
+    if (!node.url.is_empty()) {
+      resource = AddUrl(node.url, node.parent, node.tag_name, &(node.children));
+    }
     if (!node.tag_name.empty()) {

[Nathan Parker, 2017/02/27 23:00:36]

When would there a URL but no tag name? Maybe I'm misunderstanding tag name.

[lpz, 2017/02/28 22:53:28]

The last Node in each IPC is a "summary" node whose URL is the render frame's
document URL and its children are a flat list of all resources in the frame. It
never has a tag name.
Added a comment.

       AddDomElement(frame_tree_node_id, frame_url.spec(), node.node_id,
                     node.tag_name, node.parent_node_id, resource);