Only track http and https in duplicate resource handler metrics

content/browser/renderer_host/duplicate_content_resource_handler.cc

Index: content/browser/renderer_host/duplicate_content_resource_handler.cc
diff --git a/content/browser/renderer_host/duplicate_content_resource_handler.cc b/content/browser/renderer_host/duplicate_content_resource_handler.cc
index 0c7a8ba880ea28635801f8a6df42a4cf4be14e17..d63641a7adf082b90b383297e99eb9ae59c19451 100644
--- a/content/browser/renderer_host/duplicate_content_resource_handler.cc
+++ b/content/browser/renderer_host/duplicate_content_resource_handler.cc
@@ -89,17 +89,23 @@ bool DuplicateContentResourceHandler::OnResponseCompleted(
 }
 
 void DuplicateContentResourceHandler::RecordContentMetrics() {
+  // Ignore everything that's not http/https. Specifically, exclude data and
+  // blob URLs which can be generated by content and cause the maintained sets
+  // to grow without bounds.
+  if (!request_->url().SchemeIs("http") && !request_->url().SchemeIs("https")) {

[gavinp, 2012/08/09 20:20:45]

I'm taking my LGTM back. This code calculates the hash on every resource, when
it could save some time by just setting a bool early on.

[scottmg, 2012/08/09 20:24:56]

Sorry, you lost me? There's a return here, so if the request is not http and not
https we early out.

[gavinp, 2012/08/12 16:17:14]

See my comment on line 74, above.

On 2012/08/09 20:24:56, scottmg wrote:

+    bytes_read_ = 0;
+    read_buffer_ = NULL;
+    return;
+  }
   MH_UINT32 contents_hash = PMurHash32_Result(pmurhash_ph1_,
                                               pmurhash_pcarry_, bytes_read_);
-
   // Combine the contents_hash with the url, so we can test if future content
   // identical resources have the same original url or not.
-  MH_UINT32 hashed_with_url;
   const std::string& url_spec = request_->url().spec();
   PMurHash32_Process(&pmurhash_ph1_, &pmurhash_pcarry_,
                      url_spec.data(), url_spec.length());
-  hashed_with_url = PMurHash32_Result(pmurhash_ph1_, pmurhash_pcarry_,
-                                      url_spec.length() + bytes_read_);
+  MH_UINT32 hashed_with_url = PMurHash32_Result(

[gavinp, 2012/08/09 18:32:00]

nice cleanup.

+      pmurhash_ph1_, pmurhash_pcarry_, url_spec.length() + bytes_read_);
 
   DVLOG(4) << "url: " << url_spec;
   DVLOG(4) << "contents hash: " << contents_hash;