OOPIFs: Deduplicating MHTML parts across frames.

content/browser/download/mhtml_generation_manager.cc

Index: content/browser/download/mhtml_generation_manager.cc
diff --git a/content/browser/download/mhtml_generation_manager.cc b/content/browser/download/mhtml_generation_manager.cc
index e011f42dae39635230366e9daa7947ffb863d6d8..9d112ffb716b1a29caf5bb17a657b4f0d8d94a2c 100644
--- a/content/browser/download/mhtml_generation_manager.cc
+++ b/content/browser/download/mhtml_generation_manager.cc
@@ -20,6 +20,8 @@
 #include "content/public/browser/render_process_host.h"
 #include "content/public/browser/render_process_host_observer.h"
 #include "content/public/browser/web_contents.h"
+#include "crypto/sha2.h"
+#include "url/gurl.h"
 
 namespace content {
 
@@ -34,15 +36,18 @@ class MHTMLGenerationManager::Job : public RenderProcessHostObserver {
 
   GenerateMHTMLCallback callback() const { return callback_; }
 
-  // Sends IPC to the renderer, asking for MHTML generation of the next frame.
-  //
-  // Returns true if the message was sent successfully; false otherwise.
-  //
-  // See FrameMsg_SerializeAsMHTML IPC message for description of
-  // |mhtml_boundary_marker| parameter.
-  bool SendToNextRenderFrame(const std::string& mhtml_boundary_marker);
+  // Sends IPC to the renderer of the first frame, asking it to start
+  // MHTML serialization.
+  // Returns |true| upon success; |false| otherwise.
+  bool SendToFirstFrame();
 
-  // Indicates if more calls to SendToNextRenderFrame are needed.
+  // Handler for ViewHostMsg_SavedPageAsMHTML (a notification from the renderer
+  // that the MHTML generation for previous frame has finished).
+  // Returns |true| upon success; |false| otherwise.
+  bool OnSavedPageAsMHTML(std::string mhtml_boundary_marker,
+                          const std::set<GURL>& uris_of_generated_mhtml_parts);
+
+  // Indicates if the job has completed.
   bool GotMoreFramesToProcess() {
     return !pending_frame_tree_node_ids_.empty();
   }
@@ -66,6 +71,14 @@ class MHTMLGenerationManager::Job : public RenderProcessHostObserver {
   std::map<int, std::string> CreateFrameRoutingIdToContentId(
       SiteInstance* site_instance);
 
+  // Sends IPC to the renderer, asking for MHTML generation of the next frame.
+  //
+  // Returns true if the message was sent successfully; false otherwise.
+  //
+  // See FrameMsg_SerializeAsMHTML IPC message for description of
+  // |mhtml_boundary_marker| parameter.
+  bool SendToNextRenderFrame(const std::string& mhtml_boundary_marker);
+
   // Id used to map renderer responses to jobs.
   // See also MHTMLGenerationManager::id_to_job_ map.
   int job_id_;
@@ -80,6 +93,9 @@ class MHTMLGenerationManager::Job : public RenderProcessHostObserver {
   // for more details about what "content ids" are and how they are used).
   std::map<int, std::string> frame_tree_node_to_content_id_;
 
+  // URIs of already generated MHTML parts.
+  std::set<GURL> already_serialized_uris_;
+
   // The callback to call once generation is complete.
   GenerateMHTMLCallback callback_;
 
@@ -138,9 +154,13 @@ bool MHTMLGenerationManager::Job::SendToNextRenderFrame(
   DCHECK(browser_file_.IsValid());
   DCHECK_LT(0u, pending_frame_tree_node_ids_.size());
 
+  FrameMsg_SerializeAsMHTML_Params ipc_params;
+  ipc_params.job_id = job_id_;
+  ipc_params.mhtml_boundary_marker = mhtml_boundary_marker;
+
   int frame_tree_node_id = pending_frame_tree_node_ids_.front();
   pending_frame_tree_node_ids_.pop();
-  bool is_last_frame = pending_frame_tree_node_ids_.empty();
+  ipc_params.is_last_frame = pending_frame_tree_node_ids_.empty();
 
   FrameTreeNode* ftn = FrameTreeNode::GloballyFindByID(frame_tree_node_id);
   if (!ftn)  // The contents went away.
@@ -151,12 +171,20 @@ bool MHTMLGenerationManager::Job::SendToNextRenderFrame(
   observed_renderer_process_host_.RemoveAll();
   observed_renderer_process_host_.Add(rfh->GetProcess());
 
-  IPC::PlatformFileForTransit renderer_file = IPC::GetFileHandleForProcess(
+  // Tell the renderer to skip (= deduplicate) already covered MHTML parts.
+  std::set<std::string> digests_of_uris_to_skip;

[ncarter (slow), 2015/12/04 21:52:39]

This is unused.

[Łukasz Anforowicz, 2015/12/04 22:55:43]

Thanks for catching this.  I must have missed this when converting things to use
FrameMsg_SerializeAsMHTML_Params.  Removed.

+  ipc_params.salt = base::GenerateGUID();
+  for (const GURL& uri : already_serialized_uris_) {
+    std::string digest = crypto::SHA256HashString(ipc_params.salt + uri.spec());
+    ipc_params.digests_of_uris_to_skip.insert(digest);
+  }
+
+  ipc_params.destination_file = IPC::GetFileHandleForProcess(
       browser_file_.GetPlatformFile(), rfh->GetProcess()->GetHandle(),
       false);  // |close_source_handle|.
-  rfh->Send(new FrameMsg_SerializeAsMHTML(
-      rfh->GetRoutingID(), job_id_, renderer_file, mhtml_boundary_marker,
-      CreateFrameRoutingIdToContentId(rfh->GetSiteInstance()), is_last_frame));
+  ipc_params.frame_routing_id_to_content_id =
+      CreateFrameRoutingIdToContentId(rfh->GetSiteInstance());
+  rfh->Send(new FrameMsg_SerializeAsMHTML(rfh->GetRoutingID(), ipc_params));
   return true;
 }
 
@@ -202,6 +230,27 @@ void MHTMLGenerationManager::Job::CloseFile(
       callback);
 }
 
+bool MHTMLGenerationManager::Job::SendToFirstFrame() {
+  // We don't yet have an mhtml boundary marker when serializing the 1st frame.
+  std::string initial_mhtml_boundary_marker = "";
+
+  return SendToNextRenderFrame(initial_mhtml_boundary_marker);
+}
+
+bool MHTMLGenerationManager::Job::OnSavedPageAsMHTML(
+    std::string mhtml_boundary_marker,
+    const std::set<GURL>& uris_of_generated_mhtml_parts) {
+  bool success = true;
+
+  already_serialized_uris_.insert(uris_of_generated_mhtml_parts.begin(),
+                                  uris_of_generated_mhtml_parts.end());
+
+  if (GotMoreFramesToProcess())
+    success = SendToNextRenderFrame(mhtml_boundary_marker);
+
+  return success;
+}
+
 // static
 int64 MHTMLGenerationManager::Job::CloseFileOnFileThread(base::File file) {
   DCHECK_CURRENTLY_ON(BrowserThread::FILE);
@@ -239,7 +288,8 @@ void MHTMLGenerationManager::SaveMHTML(WebContents* web_contents,
 void MHTMLGenerationManager::OnSavedPageAsMHTML(
     int job_id,
     bool mhtml_generation_in_renderer_succeeded,
-    const std::string& mhtml_boundary_marker) {
+    const std::string& mhtml_boundary_marker,
+    const std::set<GURL>& uris_of_generated_mhtml_parts) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
 
   if (!mhtml_generation_in_renderer_succeeded) {
@@ -251,14 +301,15 @@ void MHTMLGenerationManager::OnSavedPageAsMHTML(
   if (!job)
     return;
 
-  if (job->GotMoreFramesToProcess()) {
-    if (!job->SendToNextRenderFrame(mhtml_boundary_marker)) {
-      JobFinished(job_id, JobStatus::FAILURE);
-    }
+  if (!job->GotMoreFramesToProcess()) {
+    JobFinished(job_id, JobStatus::SUCCESS);
     return;
   }
 
-  JobFinished(job_id, JobStatus::SUCCESS);
+  if (!job->OnSavedPageAsMHTML(mhtml_boundary_marker,
+                               uris_of_generated_mhtml_parts)) {
+    JobFinished(job_id, JobStatus::FAILURE);
+  }
 }
 
 // static
@@ -297,9 +348,7 @@ void MHTMLGenerationManager::OnFileAvailable(int job_id,
 
   job->set_browser_file(browser_file.Pass());
 
-  // We don't yet have an mhtml boundary marker when serializing the 1st frame.
-  std::string initial_mhtml_boundary_marker = "";
-  if (!job->SendToNextRenderFrame(initial_mhtml_boundary_marker)) {
+  if (!job->SendToFirstFrame()) {
     JobFinished(job_id, JobStatus::FAILURE);
   }
 }