Add support for distilling current WebContents

This CL adds the utilities needed for using the current WebContents when
distilling web pages.

BUG=361939

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=272611

Review URL: https://codereview.chromium.org/266073003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@273101 0039d316-1c4b-4281-b951-d872f2087c98

11 files changed, 435 insertions, 21 deletions

diff --git a/components/dom_distiller.gypi b/components/dom_distiller.gypi
index 092a26a..cc7bae5 100644
--- a/components/dom_distiller.gypi
+++ b/components/dom_distiller.gypi
@@ -156,6 +156,8 @@
                 'dom_distiller/content/distiller_page_web_contents.h',
                 'dom_distiller/content/dom_distiller_viewer_source.cc',
                 'dom_distiller/content/dom_distiller_viewer_source.h',
+                'dom_distiller/content/web_contents_main_frame_observer.cc',
+                'dom_distiller/content/web_contents_main_frame_observer.h',
               ],
             },
           ],
diff --git a/components/dom_distiller/content/distiller_page_web_contents.cc b/components/dom_distiller/content/distiller_page_web_contents.cc
index baa9977..527c0eb 100644
--- a/components/dom_distiller/content/distiller_page_web_contents.cc
+++ b/components/dom_distiller/content/distiller_page_web_contents.cc
@@ -7,7 +7,9 @@
 #include "base/callback.h"
 #include "base/memory/scoped_ptr.h"
 #include "base/strings/utf_string_conversions.h"
+#include "components/dom_distiller/content/web_contents_main_frame_observer.h"
 #include "components/dom_distiller/core/distiller_page.h"
+#include "components/dom_distiller/core/dom_distiller_service.h"
 #include "content/public/browser/browser_context.h"
 #include "content/public/browser/navigation_controller.h"
 #include "content/public/browser/render_frame_host.h"
@@ -18,18 +20,48 @@
 
 namespace dom_distiller {
 
+SourcePageHandleWebContents::SourcePageHandleWebContents(
+    scoped_ptr<content::WebContents> web_contents)
+    : web_contents_(web_contents.Pass()) {
+  DCHECK(web_contents_);
+}
+
+SourcePageHandleWebContents::~SourcePageHandleWebContents() {
+}
+
+scoped_ptr<content::WebContents> SourcePageHandleWebContents::GetWebContents() {
+  return web_contents_.Pass();
+}
+
 scoped_ptr<DistillerPage> DistillerPageWebContentsFactory::CreateDistillerPage()
     const {
   DCHECK(browser_context_);
-  return scoped_ptr<DistillerPage>(
-      new DistillerPageWebContents(browser_context_));
+  return scoped_ptr<DistillerPage>(new DistillerPageWebContents(
+      browser_context_, scoped_ptr<SourcePageHandleWebContents>()));
+}
+
+scoped_ptr<DistillerPage>
+DistillerPageWebContentsFactory::CreateDistillerPageWithHandle(
+    scoped_ptr<SourcePageHandle> handle) const {
+  DCHECK(browser_context_);
+  scoped_ptr<SourcePageHandleWebContents> web_contents_handle =
+      scoped_ptr<SourcePageHandleWebContents>(
+          static_cast<SourcePageHandleWebContents*>(handle.release()));
+  return scoped_ptr<DistillerPage>(new DistillerPageWebContents(
+      browser_context_, web_contents_handle.Pass()));
 }
 
 DistillerPageWebContents::DistillerPageWebContents(
-    content::BrowserContext* browser_context)
-    : state_(IDLE), browser_context_(browser_context) {}
+    content::BrowserContext* browser_context,
+    scoped_ptr<SourcePageHandleWebContents> optional_web_contents_handle)
+    : state_(IDLE), browser_context_(browser_context) {
+  if (optional_web_contents_handle) {
+    web_contents_ = optional_web_contents_handle->GetWebContents().Pass();
+  }
+}
 
-DistillerPageWebContents::~DistillerPageWebContents() {}
+DistillerPageWebContents::~DistillerPageWebContents() {
+}
 
 void DistillerPageWebContents::DistillPageImpl(const GURL& url,
                                                const std::string& script) {
@@ -38,6 +70,31 @@ void DistillerPageWebContents::DistillPageImpl(const GURL& url,
   state_ = LOADING_PAGE;
   script_ = script;
 
+  if (web_contents_ && web_contents_->GetLastCommittedURL() == url) {
+    WebContentsMainFrameObserver* main_frame_observer =
+        WebContentsMainFrameObserver::FromWebContents(web_contents_.get());
+    if (main_frame_observer && main_frame_observer->is_initialized()) {
+      if (main_frame_observer->is_document_loaded_in_main_frame()) {
+        // Main frame has already loaded for the current WebContents, so execute
+        // JavaScript immediately.
+        ExecuteJavaScript();
+      } else {
+        // Main frame document has not loaded yet, so wait until it has before
+        // executing JavaScript. It will trigger after DocumentLoadedInFrame is
+        // called for the main frame.
+        content::WebContentsObserver::Observe(web_contents_.get());
+      }
+    } else {
+      // The WebContentsMainFrameObserver has not been correctly initialized,
+      // so fall back to creating a new WebContents.
+      CreateNewWebContents(url);
+    }
+  } else {
+    CreateNewWebContents(url);
+  }
+}
+
+void DistillerPageWebContents::CreateNewWebContents(const GURL& url) {
   // Create new WebContents to use for distilling the content.
   content::WebContents::CreateParams create_params(browser_context_);
   create_params.initially_hidden = true;
@@ -54,8 +111,6 @@ void DistillerPageWebContents::DocumentLoadedInFrame(
     int64 frame_id,
     RenderViewHost* render_view_host) {
   if (frame_id == web_contents_->GetMainFrame()->GetRoutingID()) {
-    content::WebContentsObserver::Observe(NULL);
-    web_contents_->Stop();
     ExecuteJavaScript();
   }
 }
@@ -81,6 +136,8 @@ void DistillerPageWebContents::ExecuteJavaScript() {
   DCHECK(frame);
   DCHECK_EQ(LOADING_PAGE, state_);
   state_ = EXECUTING_JAVASCRIPT;
+  content::WebContentsObserver::Observe(NULL);
+  web_contents_->Stop();
   DVLOG(1) << "Beginning distillation";
   frame->ExecuteJavaScript(
       base::UTF8ToUTF16(script_),
diff --git a/components/dom_distiller/content/distiller_page_web_contents.h b/components/dom_distiller/content/distiller_page_web_contents.h
index 52509f5..aed9789 100644
--- a/components/dom_distiller/content/distiller_page_web_contents.h
+++ b/components/dom_distiller/content/distiller_page_web_contents.h
@@ -21,7 +21,18 @@ using content::RenderViewHost;
 
 namespace dom_distiller {
 
-class DistillerContext;
+class SourcePageHandleWebContents : public SourcePageHandle {
+ public:
+  explicit SourcePageHandleWebContents(
+      scoped_ptr<content::WebContents> web_contents);
+  virtual ~SourcePageHandleWebContents();
+
+  scoped_ptr<content::WebContents> GetWebContents();
+
+ private:
+  // The WebContents this class owns.
+  scoped_ptr<content::WebContents> web_contents_;
+};
 
 class DistillerPageWebContentsFactory : public DistillerPageFactory {
  public:
@@ -31,6 +42,8 @@ class DistillerPageWebContentsFactory : public DistillerPageFactory {
   virtual ~DistillerPageWebContentsFactory() {}
 
   virtual scoped_ptr<DistillerPage> CreateDistillerPage() const OVERRIDE;
+  virtual scoped_ptr<DistillerPage> CreateDistillerPageWithHandle(
+      scoped_ptr<SourcePageHandle> handle) const OVERRIDE;
 
  private:
   content::BrowserContext* browser_context_;
@@ -39,7 +52,9 @@ class DistillerPageWebContentsFactory : public DistillerPageFactory {
 class DistillerPageWebContents : public DistillerPage,
                                  public content::WebContentsObserver {
  public:
-  DistillerPageWebContents(content::BrowserContext* browser_context);
+  DistillerPageWebContents(
+      content::BrowserContext* browser_context,
+      scoped_ptr<SourcePageHandleWebContents> optional_web_contents_handle);
   virtual ~DistillerPageWebContents();
 
   // content::WebContentsObserver implementation.
@@ -58,6 +73,8 @@ class DistillerPageWebContents : public DistillerPage,
                                const std::string& script) OVERRIDE;
 
  private:
+  friend class TestDistillerPageWebContents;
+
   enum State {
     // The page distiller is idle.
     IDLE,
@@ -70,6 +87,10 @@ class DistillerPageWebContents : public DistillerPage,
     EXECUTING_JAVASCRIPT
   };
 
+  // Creates a new WebContents, adds |this| as an observer, and loads the
+  // |url|.
+  virtual void CreateNewWebContents(const GURL& url);
+
   // Injects and executes JavaScript in the context of a loaded page. This
   // must only be called after the page has successfully loaded.
   void ExecuteJavaScript();
diff --git a/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc b/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc
index 5d1cd2a..e319097 100644
--- a/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc
+++ b/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc
@@ -7,8 +7,12 @@
 #include "base/run_loop.h"
 #include "base/values.h"
 #include "components/dom_distiller/content/distiller_page_web_contents.h"
+#include "components/dom_distiller/content/web_contents_main_frame_observer.h"
 #include "components/dom_distiller/core/distiller_page.h"
 #include "content/public/browser/browser_context.h"
+#include "content/public/browser/navigation_controller.h"
+#include "content/public/browser/render_frame_host.h"
+#include "content/public/browser/web_contents_observer.h"
 #include "content/public/test/content_browser_test.h"
 #include "content/shell/browser/shell.h"
 #include "grit/component_resources.h"
@@ -23,6 +27,8 @@ using testing::Not;
 
 namespace dom_distiller {
 
+const char* kSimpleArticlePath = "/simple_article.html";
+
 class DistillerPageWebContentsTest : public ContentBrowserTest {
  public:
   // ContentBrowserTest:
@@ -66,18 +72,108 @@ class DistillerPageWebContentsTest : public ContentBrowserTest {
   }
 
  protected:
+  void RunUseCurrentWebContentsTest(const std::string& url,
+                                    bool expect_new_web_contents,
+                                    bool setup_main_frame_observer,
+                                    bool wait_for_document_loaded);
+
   DistillerPageWebContents* distiller_page_;
   base::Closure quit_closure_;
   scoped_ptr<DistilledPageInfo> page_info_;
 };
 
+// Use this class to be able to leak the WebContents, which is needed for when
+// the current WebContents is used for distillation.
+class TestDistillerPageWebContents : public DistillerPageWebContents {
+ public:
+  TestDistillerPageWebContents(
+      content::BrowserContext* browser_context,
+      scoped_ptr<SourcePageHandleWebContents> optional_web_contents_handle,
+      bool expect_new_web_contents)
+      : DistillerPageWebContents(browser_context,
+                                 optional_web_contents_handle.Pass()),
+        expect_new_web_contents_(expect_new_web_contents),
+        new_web_contents_created_(false) {}
+
+  virtual void CreateNewWebContents(const GURL& url) OVERRIDE {
+    ASSERT_EQ(true, expect_new_web_contents_);
+    new_web_contents_created_ = true;
+    // DistillerPageWebContents::CreateNewWebContents resets the scoped_ptr to
+    // the WebContents, so intentionally leak WebContents here, since it is
+    // owned by the shell.
+    content::WebContents* web_contents = web_contents_.release();
+    web_contents->GetLastCommittedURL();
+    DistillerPageWebContents::CreateNewWebContents(url);
+  }
+
+  virtual ~TestDistillerPageWebContents() {
+    if (!expect_new_web_contents_) {
+      // Intentionally leaking WebContents, since it is owned by the shell.
+      content::WebContents* web_contents = web_contents_.release();
+      web_contents->GetLastCommittedURL();
+    }
+  }
+
+  bool new_web_contents_created() { return new_web_contents_created_; }
+
+ private:
+  bool expect_new_web_contents_;
+  bool new_web_contents_created_;
+};
+
+// Helper class to know how far in the loading process the current WebContents
+// has come. It will call the callback either after
+// DidCommitProvisionalLoadForFrame or DocumentLoadedInFrame is called for the
+// main frame, based on the value of |wait_for_document_loaded|.
+class WebContentsMainFrameHelper : public content::WebContentsObserver {
+ public:
+  WebContentsMainFrameHelper(content::WebContents* web_contents,
+                             const base::Closure& callback,
+                             bool wait_for_document_loaded)
+      : web_contents_(web_contents),
+        callback_(callback),
+        wait_for_document_loaded_(wait_for_document_loaded) {
+    content::WebContentsObserver::Observe(web_contents);
+  }
+
+  virtual void DidCommitProvisionalLoadForFrame(
+      int64 frame_id,
+      const base::string16& frame_unique_name,
+      bool is_main_frame,
+      const GURL& url,
+      content::PageTransition transition_type,
+      content::RenderViewHost* render_view_host) OVERRIDE {
+    if (wait_for_document_loaded_)
+      return;
+    if (is_main_frame)
+      callback_.Run();
+  }
+
+  virtual void DocumentLoadedInFrame(
+      int64 frame_id,
+      content::RenderViewHost* render_view_host) OVERRIDE {
+    if (wait_for_document_loaded_) {
+      if (web_contents_ &&
+          frame_id == web_contents_->GetMainFrame()->GetRoutingID()) {
+        callback_.Run();
+      }
+    }
+  }
+
+ private:
+  content::WebContents* web_contents_;
+  base::Closure callback_;
+  bool wait_for_document_loaded_;
+};
+
 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, BasicDistillationWorks) {
   DistillerPageWebContents distiller_page(
-      shell()->web_contents()->GetBrowserContext());
+      shell()->web_contents()->GetBrowserContext(),
+      scoped_ptr<SourcePageHandleWebContents>());
   distiller_page_ = &distiller_page;
 
   base::RunLoop run_loop;
-  DistillPage(run_loop.QuitClosure(), "/simple_article.html");
+  DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
   run_loop.Run();
 
   EXPECT_EQ("Test Page Title", page_info_.get()->title);
@@ -89,11 +185,12 @@ IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, BasicDistillationWorks) {
 
 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, HandlesRelativeLinks) {
   DistillerPageWebContents distiller_page(
-      shell()->web_contents()->GetBrowserContext());
+      shell()->web_contents()->GetBrowserContext(),
+      scoped_ptr<SourcePageHandleWebContents>());
   distiller_page_ = &distiller_page;
 
   base::RunLoop run_loop;
-  DistillPage(run_loop.QuitClosure(), "/simple_article.html");
+  DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
   run_loop.Run();
 
   // A relative link should've been updated.
@@ -105,11 +202,12 @@ IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, HandlesRelativeLinks) {
 
 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, HandlesRelativeImages) {
   DistillerPageWebContents distiller_page(
-      shell()->web_contents()->GetBrowserContext());
+      shell()->web_contents()->GetBrowserContext(),
+      scoped_ptr<SourcePageHandleWebContents>());
   distiller_page_ = &distiller_page;
 
   base::RunLoop run_loop;
-  DistillPage(run_loop.QuitClosure(), "/simple_article.html");
+  DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
   run_loop.Run();
 
   // A relative link should've been updated.
@@ -121,7 +219,8 @@ IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, HandlesRelativeImages) {
 
 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, VisibilityDetection) {
   DistillerPageWebContents distiller_page(
-      shell()->web_contents()->GetBrowserContext());
+      shell()->web_contents()->GetBrowserContext(),
+      scoped_ptr<SourcePageHandleWebContents>());
   distiller_page_ = &distiller_page;
 
   // visble_style.html and invisible_style.html only differ by the visibility
@@ -142,4 +241,92 @@ IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, VisibilityDetection) {
   }
 }
 
+IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
+                       UsingCurrentWebContentsWrongUrl) {
+  std::string url("/bogus");
+  bool expect_new_web_contents = true;
+  bool setup_main_frame_observer = true;
+  bool wait_for_document_loaded = true;
+  RunUseCurrentWebContentsTest(url,
+                               expect_new_web_contents,
+                               setup_main_frame_observer,
+                               wait_for_document_loaded);
+}
+
+IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
+                       UsingCurrentWebContentsNoMainFrameObserver) {
+  std::string url(kSimpleArticlePath);
+  bool expect_new_web_contents = true;
+  bool setup_main_frame_observer = false;
+  bool wait_for_document_loaded = true;
+  RunUseCurrentWebContentsTest(url,
+                               expect_new_web_contents,
+                               setup_main_frame_observer,
+                               wait_for_document_loaded);
+}
+
+IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
+                       UsingCurrentWebContentsNotFinishedLoadingYet) {
+  std::string url(kSimpleArticlePath);
+  bool expect_new_web_contents = false;
+  bool setup_main_frame_observer = true;
+  bool wait_for_document_loaded = false;
+  RunUseCurrentWebContentsTest(url,
+                               expect_new_web_contents,
+                               setup_main_frame_observer,
+                               wait_for_document_loaded);
+}
+
+IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
+                       UsingCurrentWebContentsReadyForDistillation) {
+  std::string url(kSimpleArticlePath);
+  bool expect_new_web_contents = false;
+  bool setup_main_frame_observer = true;
+  bool wait_for_document_loaded = true;
+  RunUseCurrentWebContentsTest(url,
+                               expect_new_web_contents,
+                               setup_main_frame_observer,
+                               wait_for_document_loaded);
+}
+
+void DistillerPageWebContentsTest::RunUseCurrentWebContentsTest(
+    const std::string& url,
+    bool expect_new_web_contents,
+    bool setup_main_frame_observer,
+    bool wait_for_document_loaded) {
+  content::WebContents* current_web_contents = shell()->web_contents();
+  if (setup_main_frame_observer) {
+    dom_distiller::WebContentsMainFrameObserver::CreateForWebContents(
+        current_web_contents);
+  }
+  base::RunLoop url_loaded_runner;
+  WebContentsMainFrameHelper main_frame_loaded(current_web_contents,
+                                               url_loaded_runner.QuitClosure(),
+                                               wait_for_document_loaded);
+  current_web_contents->GetController().LoadURL(
+      embedded_test_server()->GetURL(url),
+      content::Referrer(),
+      content::PAGE_TRANSITION_TYPED,
+      std::string());
+  url_loaded_runner.Run();
+
+  scoped_ptr<content::WebContents> old_web_contents_sptr(current_web_contents);
+  scoped_ptr<SourcePageHandleWebContents> source_page_handle(
+      new SourcePageHandleWebContents(old_web_contents_sptr.Pass()));
+
+  TestDistillerPageWebContents distiller_page(
+      shell()->web_contents()->GetBrowserContext(),
+      source_page_handle.Pass(),
+      expect_new_web_contents);
+  distiller_page_ = &distiller_page;
+
+  base::RunLoop run_loop;
+  DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
+  run_loop.Run();
+
+  // Sanity check of distillation process.
+  EXPECT_EQ(expect_new_web_contents, distiller_page.new_web_contents_created());
+  EXPECT_EQ("Test Page Title", page_info_.get()->title);
+}
+
 }  // namespace dom_distiller
diff --git a/components/dom_distiller/content/web_contents_main_frame_observer.cc b/components/dom_distiller/content/web_contents_main_frame_observer.cc
new file mode 100644
index 0000000..ed1bbab
--- /dev/null
+++ b/components/dom_distiller/content/web_contents_main_frame_observer.cc
@@ -0,0 +1,61 @@
+// Copyright 2014 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.
+
+#include "components/dom_distiller/content/web_contents_main_frame_observer.h"
+
+#include "content/public/browser/navigation_details.h"
+#include "content/public/browser/render_frame_host.h"
+#include "content/public/browser/web_contents.h"
+#include "content/public/browser/web_contents_observer.h"
+#include "content/public/browser/web_contents_user_data.h"
+
+DEFINE_WEB_CONTENTS_USER_DATA_KEY(dom_distiller::WebContentsMainFrameObserver);
+
+namespace dom_distiller {
+
+WebContentsMainFrameObserver::WebContentsMainFrameObserver(
+    content::WebContents* web_contents)
+    : is_document_loaded_in_main_frame_(false),
+      is_initialized_(false),
+      web_contents_(web_contents) {
+  content::WebContentsObserver::Observe(web_contents);
+}
+
+WebContentsMainFrameObserver::~WebContentsMainFrameObserver() {
+  CleanUp();
+}
+
+void WebContentsMainFrameObserver::DocumentLoadedInFrame(
+    int64 frame_id,
+    content::RenderViewHost* render_view_host) {
+  if (web_contents_ &&
+      frame_id == web_contents_->GetMainFrame()->GetRoutingID()) {
+    is_document_loaded_in_main_frame_ = true;
+  }
+}
+
+void WebContentsMainFrameObserver::DidNavigateMainFrame(
+    const content::LoadCommittedDetails& details,
+    const content::FrameNavigateParams& params) {
+  if (details.is_main_frame) {
+    is_document_loaded_in_main_frame_ = false;
+    is_initialized_ = true;
+  }
+}
+
+void WebContentsMainFrameObserver::RenderProcessGone(
+    base::TerminationStatus status) {
+  CleanUp();
+}
+
+void WebContentsMainFrameObserver::WebContentsDestroyed() {
+  CleanUp();
+}
+
+void WebContentsMainFrameObserver::CleanUp() {
+  content::WebContentsObserver::Observe(NULL);
+  web_contents_ = NULL;
+}
+
+}  // namespace dom_distiller
diff --git a/components/dom_distiller/content/web_contents_main_frame_observer.h b/components/dom_distiller/content/web_contents_main_frame_observer.h
new file mode 100644
index 0000000..7343dba
--- /dev/null
+++ b/components/dom_distiller/content/web_contents_main_frame_observer.h
@@ -0,0 +1,63 @@
+// Copyright 2014 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 COMPONENTS_DOM_DISTILLER_CONTENT_WEB_CONTENTS_MAIN_FRAME_OBSERVER_H_
+#define COMPONENTS_DOM_DISTILLER_CONTENT_WEB_CONTENTS_MAIN_FRAME_OBSERVER_H_
+
+#include "content/public/browser/web_contents.h"
+#include "content/public/browser/web_contents_observer.h"
+#include "content/public/browser/web_contents_user_data.h"
+
+namespace dom_distiller {
+
+// Tracks whether DocumentLoadedInFrame has been called for the main frame for
+// the current main frame for the given WebContents. It removes itself as an
+// observer if the WebContents is destroyed or the render process is gone.
+class WebContentsMainFrameObserver
+    : public content::WebContentsObserver,
+      public content::WebContentsUserData<WebContentsMainFrameObserver> {
+ public:
+  virtual ~WebContentsMainFrameObserver();
+
+  bool is_document_loaded_in_main_frame() {
+    return is_document_loaded_in_main_frame_;
+  }
+
+  bool is_initialized() { return is_initialized_; }
+
+  // content::WebContentsObserver implementation.
+  virtual void DocumentLoadedInFrame(
+      int64 frame_id,
+      content::RenderViewHost* render_view_host) OVERRIDE;
+  virtual void DidNavigateMainFrame(
+      const content::LoadCommittedDetails& details,
+      const content::FrameNavigateParams& params) OVERRIDE;
+  virtual void RenderProcessGone(base::TerminationStatus status) OVERRIDE;
+  virtual void WebContentsDestroyed() OVERRIDE;
+
+ private:
+  explicit WebContentsMainFrameObserver(content::WebContents* web_contents);
+  friend class content::WebContentsUserData<WebContentsMainFrameObserver>;
+
+  // Removes the observer and clears the WebContents member.
+  void CleanUp();
+
+  // Whether DocumentLoadedInFrame has been called for the tracked WebContents
+  // for the current main frame. This is cleared when the main frame navigates,
+  // and set again when DocumentLoadedInFrame is called for the main frame.
+  bool is_document_loaded_in_main_frame_;
+
+  // Whether this object has been correctly initialized. This is set as soon as
+  // at least one call to DidNavigateMainFrame has happened.
+  bool is_initialized_;
+
+  // The WebContents this class is tracking.
+  content::WebContents* web_contents_;
+
+  DISALLOW_COPY_AND_ASSIGN(WebContentsMainFrameObserver);
+};
+
+}  // namespace dom_distiller
+
+#endif  // COMPONENTS_DOM_DISTILLER_CONTENT_WEB_CONTENTS_MAIN_FRAME_OBSERVER_H_
diff --git a/components/dom_distiller/core/distiller_page.h b/components/dom_distiller/core/distiller_page.h
index 429a676..b950df7 100644
--- a/components/dom_distiller/core/distiller_page.h
+++ b/components/dom_distiller/core/distiller_page.h
@@ -29,6 +29,11 @@ struct DistilledPageInfo {
   DISALLOW_COPY_AND_ASSIGN(DistilledPageInfo);
 };
 
+class SourcePageHandle {
+ public:
+  virtual ~SourcePageHandle() {}
+};
+
 // Injects JavaScript into a page, and uses it to extract and return long-form
 // content. The class can be reused to load and distill multiple pages,
 // following the state transitions described along with the class's states.
@@ -62,10 +67,6 @@ class DistillerPage {
   // should be the same regardless of the DistillerPage implementation.
   virtual void DistillPageImpl(const GURL& url, const std::string& script) = 0;
 
-  // Called by |ExecuteJavaScript| to carry out platform-specific instructions
-  // to inject and execute JavaScript within the context of the loaded page.
-  //virtual void ExecuteJavaScriptImpl() = 0;
-
  private:
   bool ready_;
   DistillerPageCallback distiller_page_callback_;
@@ -81,6 +82,8 @@ class DistillerPageFactory {
   // should be very cheap, since the pages can be thrown away without being
   // used.
   virtual scoped_ptr<DistillerPage> CreateDistillerPage() const = 0;
+  virtual scoped_ptr<DistillerPage> CreateDistillerPageWithHandle(
+      scoped_ptr<SourcePageHandle> handle) const = 0;
 };
 
 }  // namespace dom_distiller
diff --git a/components/dom_distiller/core/dom_distiller_service.cc b/components/dom_distiller/core/dom_distiller_service.cc
index 848adaa..a3cbbda 100644
--- a/components/dom_distiller/core/dom_distiller_service.cc
+++ b/components/dom_distiller/core/dom_distiller_service.cc
@@ -46,7 +46,8 @@ DomDistillerService::DomDistillerService(
       distiller_page_factory_(distiller_page_factory.Pass()) {
 }
 
-DomDistillerService::~DomDistillerService() {}
+DomDistillerService::~DomDistillerService() {
+}
 
 syncer::SyncableService* DomDistillerService::GetSyncableService() const {
   return store_->GetSyncableService();
@@ -56,6 +57,13 @@ scoped_ptr<DistillerPage> DomDistillerService::CreateDefaultDistillerPage() {
   return distiller_page_factory_->CreateDistillerPage().Pass();
 }
 
+scoped_ptr<DistillerPage>
+DomDistillerService::CreateDefaultDistillerPageWithHandle(
+    scoped_ptr<SourcePageHandle> handle) {
+  return distiller_page_factory_->CreateDistillerPageWithHandle(handle.Pass())
+      .Pass();
+}
+
 const std::string DomDistillerService::AddToList(
     const GURL& url,
     scoped_ptr<DistillerPage> distiller_page,
diff --git a/components/dom_distiller/core/dom_distiller_service.h b/components/dom_distiller/core/dom_distiller_service.h
index b2698b9e..cd59950 100644
--- a/components/dom_distiller/core/dom_distiller_service.h
+++ b/components/dom_distiller/core/dom_distiller_service.h
@@ -84,6 +84,8 @@ class DomDistillerServiceInterface {
 
   // Creates a default DistillerPage.
   virtual scoped_ptr<DistillerPage> CreateDefaultDistillerPage() = 0;
+  virtual scoped_ptr<DistillerPage> CreateDefaultDistillerPageWithHandle(
+      scoped_ptr<SourcePageHandle> handle) = 0;
 
   virtual void AddObserver(DomDistillerObserver* observer) = 0;
   virtual void RemoveObserver(DomDistillerObserver* observer) = 0;
@@ -121,6 +123,8 @@ class DomDistillerService : public DomDistillerServiceInterface {
       scoped_ptr<DistillerPage> distiller_page,
       const GURL& url) OVERRIDE;
   virtual scoped_ptr<DistillerPage> CreateDefaultDistillerPage() OVERRIDE;
+  virtual scoped_ptr<DistillerPage> CreateDefaultDistillerPageWithHandle(
+      scoped_ptr<SourcePageHandle> handle) OVERRIDE;
   virtual void AddObserver(DomDistillerObserver* observer) OVERRIDE;
   virtual void RemoveObserver(DomDistillerObserver* observer) OVERRIDE;
 
diff --git a/components/dom_distiller/core/fake_distiller_page.h b/components/dom_distiller/core/fake_distiller_page.h
index 245d23e..2e88389 100644
--- a/components/dom_distiller/core/fake_distiller_page.h
+++ b/components/dom_distiller/core/fake_distiller_page.h
@@ -19,6 +19,10 @@ class MockDistillerPageFactory : public DistillerPageFactory {
   virtual scoped_ptr<DistillerPage> CreateDistillerPage() const OVERRIDE {
     return scoped_ptr<DistillerPage>(CreateDistillerPageImpl());
   }
+  virtual scoped_ptr<DistillerPage> CreateDistillerPageWithHandle(
+      scoped_ptr<SourcePageHandle> handle) const OVERRIDE {
+    return scoped_ptr<DistillerPage>(CreateDistillerPageImpl());
+  }
 };
 
 class MockDistillerPage : public DistillerPage {
diff --git a/components/dom_distiller/core/viewer_unittest.cc b/components/dom_distiller/core/viewer_unittest.cc
index b0cea2a..7c1b075 100644
--- a/components/dom_distiller/core/viewer_unittest.cc
+++ b/components/dom_distiller/core/viewer_unittest.cc
@@ -62,6 +62,10 @@ class TestDomDistillerService : public DomDistillerServiceInterface {
   virtual scoped_ptr<DistillerPage> CreateDefaultDistillerPage() {
     return scoped_ptr<DistillerPage>();
   }
+  virtual scoped_ptr<DistillerPage> CreateDefaultDistillerPageWithHandle(
+      scoped_ptr<SourcePageHandle> handle) {
+    return scoped_ptr<DistillerPage>();
+  }
 };
 
 class DomDistillerViewerTest : public testing::Test {