Page distiller core

This change adds code to load a page in a WebContents, execute JavaScript in that context, parse the result of executing the JS, fetch images referenced in the result, and store the images and other details of the result in a proto.

BUG=288015, 283846

Review URL: https://codereview.chromium.org/26358008

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@234921 0039d316-1c4b-4281-b951-d872f2087c98

18 files changed, 920 insertions, 24 deletions

diff --git a/components/components_tests.gyp b/components/components_tests.gyp
index eaa08b7..21319a2 100644
--- a/components/components_tests.gyp
+++ b/components/components_tests.gyp
@@ -23,6 +23,7 @@
             'autofill/core/common/password_form_fill_data_unittest.cc',
             'browser_context_keyed_service/browser_context_dependency_manager_unittest.cc',
             'browser_context_keyed_service/dependency_graph_unittest.cc',
+            'dom_distiller/core/distiller_unittest.cc',
             'dom_distiller/core/distiller_url_fetcher_unittest.cc',
             'dom_distiller/core/dom_distiller_database_unittest.cc',
             'dom_distiller/core/dom_distiller_model_unittest.cc',
@@ -222,5 +223,74 @@
         }],
       ],
     }],
+    ['OS!="ios"', {
+      'targets': [
+        {
+          'target_name': 'components_browsertests',
+          'type': '<(gtest_target_type)',
+          'defines!': ['CONTENT_IMPLEMENTATION'],
+          'dependencies': [
+            '../content/content_shell_and_tests.gyp:content_browser_test_support',
+            '../content/content_shell_and_tests.gyp:test_support_content',
+            '../skia/skia.gyp:skia',
+            '../testing/gtest.gyp:gtest',
+            'components.gyp:dom_distiller_content',
+            'components.gyp:dom_distiller_core',
+          ],
+          'include_dirs': [
+            '..',
+          ],
+          'defines': [
+            'HAS_OUT_OF_PROC_TEST_RUNNER',
+          ],
+          'sources': [
+            '../content/test/content_test_launcher.cc',
+            'dom_distiller/content/distiller_page_web_contents_browsertest.cc',
+          ],
+          'conditions': [
+            ['OS=="win"', {
+              'resource_include_dirs': [
+                '<(SHARED_INTERMEDIATE_DIR)/webkit',
+              ],
+              'sources': [
+                '../content/shell/app/resource.h',
+                '../content/shell/app/shell.rc',
+                # TODO:  It would be nice to have these pulled in
+                # automatically from direct_dependent_settings in
+                # their various targets (net.gyp:net_resources, etc.),
+                # but that causes errors in other targets when
+                # resulting .res files get referenced multiple times.
+                '<(SHARED_INTERMEDIATE_DIR)/net/net_resources.rc',
+                '<(SHARED_INTERMEDIATE_DIR)/webkit/blink_resources.rc',
+                '<(SHARED_INTERMEDIATE_DIR)/webkit/webkit_strings_en-US.rc',
+              ],
+              'dependencies': [
+                '<(DEPTH)/net/net.gyp:net_resources',
+                '<(DEPTH)/third_party/iaccessible2/iaccessible2.gyp:iaccessible2',
+                '<(DEPTH)/third_party/isimpledom/isimpledom.gyp:isimpledom',
+                '<(DEPTH)/webkit/webkit_resources.gyp:webkit_strings',
+                '<(DEPTH)/webkit/webkit_resources.gyp:webkit_resources',
+              ],
+              'configurations': {
+                'Debug_Base': {
+                  'msvs_settings': {
+                    'VCLinkerTool': {
+                      'LinkIncremental': '<(msvs_large_module_debug_link_mode)',
+                    },
+                  },
+                },
+              },
+              # TODO(jschuh): crbug.com/167187 fix size_t to int truncations.
+              'msvs_disabled_warnings': [ 4267, ],
+            }],
+            ['OS=="win" and win_use_allocator_shim==1', {
+              'dependencies': [
+                '../base/allocator/allocator.gyp:allocator',
+              ],
+            }],
+          ],
+        },
+      ],
+    }],
   ],
 }
diff --git a/components/dom_distiller.gypi b/components/dom_distiller.gypi
index abece3f..74a9b1d 100644
--- a/components/dom_distiller.gypi
+++ b/components/dom_distiller.gypi
@@ -32,12 +32,15 @@
           'type': 'static_library',
           'dependencies': [
             'dom_distiller_core',
+            '../skia/skia.gyp:skia',
             '../sync/sync.gyp:sync',
           ],
           'include_dirs': [
             '..',
           ],
           'sources': [
+            'dom_distiller/content/distiller_page_web_contents.h',
+            'dom_distiller/content/distiller_page_web_contents.cc',
             'dom_distiller/content/dom_distiller_service_factory.h',
             'dom_distiller/content/dom_distiller_service_factory.cc',
           ],
@@ -64,8 +67,9 @@
           'type': 'static_library',
           'dependencies': [
             'distilled_page_proto',
+            'dom_distiller_resources',
             '../base/base.gyp:base',
-            #'../skia/skia.gyp:skia',
+            '../skia/skia.gyp:skia',
             '../sync/sync.gyp:sync',
             '../third_party/protobuf/protobuf.gyp:protobuf_lite',
             '../third_party/leveldatabase/leveldatabase.gyp:leveldatabase',
@@ -76,6 +80,10 @@
           'sources': [
             'dom_distiller/core/article_entry.cc',
             'dom_distiller/core/article_entry.h',
+            'dom_distiller/core/distiller.cc',
+            'dom_distiller/core/distiller.h',
+            'dom_distiller/core/distiller_page.cc',
+            'dom_distiller/core/distiller_page.h',
             'dom_distiller/core/distiller_url_fetcher.cc',
             'dom_distiller/core/distiller_url_fetcher.h',
             'dom_distiller/core/dom_distiller_constants.cc',
diff --git a/components/dom_distiller/DEPS b/components/dom_distiller/DEPS
index 99f902a..1c12a78 100644
--- a/components/dom_distiller/DEPS
+++ b/components/dom_distiller/DEPS
@@ -5,6 +5,9 @@ include_rules = [
   "+third_party/leveldatabase/src/include",
   "+net/http",
   "+net/url_request",
+  "+ui/base/android",
+  "+ui/base/resource",
+  "+ui/gfx/android",
 
   # The dom distiller is a layered component; subdirectories must explicitly
   # introduce the ability to use the content layer as appropriate.
diff --git a/components/dom_distiller/content/DEPS b/components/dom_distiller/content/DEPS
index f3ade60..6be787a 100644
--- a/components/dom_distiller/content/DEPS
+++ b/components/dom_distiller/content/DEPS
@@ -1,4 +1,7 @@
 include_rules = [
   "+components/browser_context_keyed_service",
-  "+content/public/browser",
+  "+content/public",
+  "+content/shell",
+  "+content/test",
+  "+net/test",
 ]
diff --git a/components/dom_distiller/content/distiller_page_web_contents.cc b/components/dom_distiller/content/distiller_page_web_contents.cc
new file mode 100644
index 0000000..0138d37
--- /dev/null
+++ b/components/dom_distiller/content/distiller_page_web_contents.cc
@@ -0,0 +1,78 @@
+// Copyright 2013 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/distiller_page_web_contents.h"
+
+#include "base/callback.h"
+#include "base/memory/scoped_ptr.h"
+#include "base/strings/utf_string_conversions.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_view_host.h"
+#include "content/public/browser/web_contents.h"
+#include "content/public/browser/web_contents_observer.h"
+#include "url/gurl.h"
+
+namespace dom_distiller {
+
+scoped_ptr<DistillerPage> DistillerPageWebContentsFactory::CreateDistillerPage(
+    DistillerPage::Delegate* delegate) const {
+  DCHECK(browser_context_);
+  return scoped_ptr<DistillerPage>(
+      new DistillerPageWebContents(delegate, browser_context_));
+}
+
+DistillerPageWebContents::DistillerPageWebContents(
+    DistillerPage::Delegate* delegate,
+    content::BrowserContext* browser_context)
+  : DistillerPage(delegate),
+    browser_context_(browser_context) {}
+
+DistillerPageWebContents::~DistillerPageWebContents() {
+}
+
+void DistillerPageWebContents::InitImpl() {
+  DCHECK(browser_context_);
+  web_contents_.reset(
+      content::WebContents::Create(
+          content::WebContents::CreateParams(browser_context_)));
+}
+
+void DistillerPageWebContents::LoadURLImpl(const GURL& gurl) {
+  DCHECK(web_contents_.get());
+  content::WebContentsObserver::Observe(web_contents_.get());
+  content::NavigationController::LoadURLParams params(gurl);
+  web_contents_->GetController().LoadURLWithParams(params);
+}
+
+void DistillerPageWebContents::ExecuteJavaScriptImpl(
+    const std::string& script) {
+  content::RenderViewHost* host = web_contents_->GetRenderViewHost();
+  DCHECK(host);
+  host->ExecuteJavascriptInWebFrameCallbackResult(
+      string16(),  // frame_xpath
+      UTF8ToUTF16(script),
+      base::Bind(&DistillerPage::OnExecuteJavaScriptDone,
+                 base::Unretained(this)));
+}
+
+void DistillerPageWebContents::DidFinishLoad(int64 frame_id,
+                                             const GURL& validated_url,
+                                             bool is_main_frame,
+                                             RenderViewHost* render_view_host) {
+  content::WebContentsObserver::Observe(NULL);
+  OnLoadURLDone();
+}
+void DistillerPageWebContents::DidFailLoad(int64 frame_id,
+                                           const GURL& validated_url,
+                                           bool is_main_frame,
+                                           int error_code,
+                                           const string16& error_description,
+                                           RenderViewHost* render_view_host) {
+  content::WebContentsObserver::Observe(NULL);
+  OnLoadURLFailed();
+}
+
+}  // namespace dom_distiller
diff --git a/components/dom_distiller/content/distiller_page_web_contents.h b/components/dom_distiller/content/distiller_page_web_contents.h
new file mode 100644
index 0000000..65694ff
--- /dev/null
+++ b/components/dom_distiller/content/distiller_page_web_contents.h
@@ -0,0 +1,71 @@
+// Copyright 2013 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_DISTILLER_PAGE_WEB_CONTENTS_H_
+#define COMPONENTS_DOM_DISTILLER_CONTENT_DISTILLER_PAGE_WEB_CONTENTS_H_
+
+#include "base/memory/scoped_ptr.h"
+#include "components/dom_distiller/core/distiller_page.h"
+#include "content/public/browser/web_contents.h"
+#include "content/public/browser/web_contents_observer.h"
+#include "url/gurl.h"
+
+namespace content {
+class RenderViewHost;
+}
+
+using content::RenderViewHost;
+
+namespace dom_distiller {
+
+class DistillerContext;
+
+class DistillerPageWebContentsFactory : public DistillerPageFactory {
+ public:
+  DistillerPageWebContentsFactory(content::BrowserContext* browser_context)
+    :  DistillerPageFactory(), browser_context_(browser_context) {}
+  virtual ~DistillerPageWebContentsFactory() {}
+
+  virtual scoped_ptr<DistillerPage> CreateDistillerPage(
+      DistillerPage::Delegate* delegate) const OVERRIDE;
+
+ private:
+  content::BrowserContext* browser_context_;
+};
+
+
+class DistillerPageWebContents : public DistillerPage,
+                                 public content::WebContentsObserver {
+ public:
+  DistillerPageWebContents(DistillerPage::Delegate* delegate,
+                           content::BrowserContext* browser_context);
+  virtual ~DistillerPageWebContents();
+
+  // content::WebContentsObserver implementation.
+  virtual void DidFinishLoad(int64 frame_id,
+                             const GURL& validated_url,
+                             bool is_main_frame,
+                             RenderViewHost* render_view_host) OVERRIDE;
+
+  virtual void DidFailLoad(int64 frame_id,
+                           const GURL& validated_url,
+                           bool is_main_frame,
+                           int error_code,
+                           const string16& error_description,
+                           RenderViewHost* render_view_host) OVERRIDE;
+
+ protected:
+  virtual void InitImpl() OVERRIDE;
+  virtual void LoadURLImpl(const GURL& gurl) OVERRIDE;
+  virtual void ExecuteJavaScriptImpl(const std::string& script) OVERRIDE;
+
+ private:
+  scoped_ptr<content::WebContents> web_contents_;
+  content::BrowserContext* browser_context_;
+  DISALLOW_COPY_AND_ASSIGN(DistillerPageWebContents);
+};
+
+}  // namespace dom_distiller
+
+#endif  // COMPONENTS_DOM_DISTILLER_CONTENT_DISTILLER_PAGE_WEB_CONTENTS_H_
diff --git a/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc b/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc
new file mode 100644
index 0000000..13f6a49
--- /dev/null
+++ b/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc
@@ -0,0 +1,86 @@
+// Copyright 2013 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 "base/run_loop.h"
+#include "base/values.h"
+#include "components/dom_distiller/content/distiller_page_web_contents.h"
+#include "components/dom_distiller/core/distiller_page.h"
+#include "content/public/browser/browser_context.h"
+#include "content/shell/browser/shell.h"
+#include "content/test/content_browser_test.h"
+#include "net/test/embedded_test_server/embedded_test_server.h"
+
+using content::ContentBrowserTest;
+
+namespace {
+  // TODO(bengr): Once JavaScript has landed to extract article content from
+  //              a loaded page, test the interaction of that script with
+  //              DistillerPageWebContents.
+  static const char kTitle[] = "Test Page Title";
+  static const char kHtml[] =
+      "<body>T<img src='http://t.com/t.jpg' id='0'></body>";
+  static const char kImageUrl[] = "http://t.com/t.jpg";
+
+  static const char kScript[] =
+      " (function () {"
+      "   var result = new Array(3);"
+      "   result[0] = \"Test Page Title\";"
+      "   result[1] = \"<body>T<img src='http://t.com/t.jpg' id='0'></body>\";"
+      "   result[2] = \"http://t.com/t.jpg\";"
+      "   return result;"
+      " }())";
+}
+
+namespace dom_distiller {
+
+class DistillerPageWebContentsTest
+  : public ContentBrowserTest,
+    public DistillerPage::Delegate {
+ public:
+  void DistillPage(const base::Closure& quit_closure, const std::string& url) {
+    quit_closure_ = quit_closure;
+    distiller_page_->LoadURL(
+        embedded_test_server()->GetURL(url));
+  }
+
+  virtual void OnLoadURLDone() OVERRIDE {
+    distiller_page_->ExecuteJavaScript(kScript);
+  }
+
+  virtual void OnExecuteJavaScriptDone(const base::Value* value) OVERRIDE {
+    value_ = value->DeepCopy();
+    quit_closure_.Run();
+  }
+
+ protected:
+  DistillerPageWebContents* distiller_page_;
+  base::Closure quit_closure_;
+  const base::Value* value_;
+};
+
+IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, LoadPage) {
+  ASSERT_TRUE(embedded_test_server()->InitializeAndWaitUntilReady());
+  DistillerPageWebContents distiller_page(
+      this, shell()->web_contents()->GetBrowserContext());
+  distiller_page_ = &distiller_page;
+  distiller_page.Init();
+  base::RunLoop run_loop;
+  DistillPage(run_loop.QuitClosure(), "/simple_page.html");
+  run_loop.Run();
+
+  const base::ListValue* result_list = NULL;
+  ASSERT_TRUE(value_->GetAsList(&result_list));
+  ASSERT_EQ(3u, result_list->GetSize());
+  std::string title;
+  result_list->GetString(0, &title);
+  ASSERT_EQ(kTitle, title);
+  std::string html;
+  result_list->GetString(1, &html);
+  ASSERT_EQ(kHtml, html);
+  std::string image_url;
+  result_list->GetString(2, &image_url);
+  ASSERT_EQ(kImageUrl, image_url);
+}
+
+}  // namespace dom_distiller
diff --git a/components/dom_distiller/content/dom_distiller_service_factory.cc b/components/dom_distiller/content/dom_distiller_service_factory.cc
index 956451f..1d030e6 100644
--- a/components/dom_distiller/content/dom_distiller_service_factory.cc
+++ b/components/dom_distiller/content/dom_distiller_service_factory.cc
@@ -5,8 +5,10 @@
 #include "components/dom_distiller/content/dom_distiller_service_factory.h"
 
 #include "components/browser_context_keyed_service/browser_context_dependency_manager.h"
+#include "components/dom_distiller/content/distiller_page_web_contents.h"
 #include "components/dom_distiller/core/distiller.h"
 #include "components/dom_distiller/core/dom_distiller_store.h"
+#include "content/public/browser/browser_context.h"
 
 namespace dom_distiller {
 
@@ -38,7 +40,13 @@ DomDistillerServiceFactory::~DomDistillerServiceFactory() {}
 BrowserContextKeyedService* DomDistillerServiceFactory::BuildServiceInstanceFor(
     content::BrowserContext* profile) const {
   scoped_ptr<DomDistillerStoreInterface> dom_distiller_store;
-  scoped_ptr<DistillerFactory> distiller_factory;
+  scoped_ptr<DistillerPageFactory> distiller_page_factory(
+      new DistillerPageWebContentsFactory(profile));
+  scoped_ptr<DistillerURLFetcherFactory> distiller_url_fetcher_factory(
+      new DistillerURLFetcherFactory(profile->GetRequestContext()));
+  scoped_ptr<DistillerFactory> distiller_factory(
+      new DistillerFactory(distiller_page_factory.Pass(),
+                           distiller_url_fetcher_factory.Pass()));
   return new DomDistillerContextKeyedService(dom_distiller_store.Pass(),
                                              distiller_factory.Pass());
 }
diff --git a/components/dom_distiller/core/distiller.cc b/components/dom_distiller/core/distiller.cc
new file mode 100644
index 0000000..11189e5
--- /dev/null
+++ b/components/dom_distiller/core/distiller.cc
@@ -0,0 +1,131 @@
+// Copyright 2013 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/core/distiller.h"
+
+#include <map>
+
+#include "base/bind.h"
+#include "base/callback.h"
+#include "base/strings/stringprintf.h"
+#include "base/strings/utf_string_conversions.h"
+#include "base/values.h"
+#include "components/dom_distiller/core/distiller_page.h"
+#include "components/dom_distiller/core/distiller_url_fetcher.h"
+#include "components/dom_distiller/core/proto/distilled_page.pb.h"
+#include "grit/dom_distiller_resources.h"
+#include "net/url_request/url_request_context_getter.h"
+#include "ui/base/resource/resource_bundle.h"
+#include "url/gurl.h"
+
+namespace dom_distiller {
+
+DistillerFactory::DistillerFactory(
+    scoped_ptr<DistillerPageFactory> distiller_page_factory,
+    scoped_ptr<DistillerURLFetcherFactory> distiller_url_fetcher_factory)
+  : distiller_page_factory_(distiller_page_factory.Pass()),
+    distiller_url_fetcher_factory_(distiller_url_fetcher_factory.Pass()) {}
+
+DistillerFactory::~DistillerFactory() {}
+
+Distiller* DistillerFactory::CreateDistiller() {
+  return new DistillerImpl(*distiller_page_factory_,
+                           *distiller_url_fetcher_factory_);
+}
+
+DistillerImpl::DistillerImpl(
+    const DistillerPageFactory& distiller_page_factory,
+    const DistillerURLFetcherFactory& distiller_url_fetcher_factory)
+  : distiller_page_factory_(distiller_page_factory),
+    distiller_url_fetcher_factory_(distiller_url_fetcher_factory) {
+  distiller_page_ = distiller_page_factory_.CreateDistillerPage(this).Pass();
+}
+
+DistillerImpl::~DistillerImpl() {
+}
+
+void DistillerImpl::Init() {
+  distiller_page_->Init();
+}
+
+void DistillerImpl::DistillPage(const GURL& url,
+                            const DistillerCallback& distillation_cb) {
+  distillation_cb_ = distillation_cb;
+  proto_.reset(new DistilledPageProto());
+  proto_->set_url(url.spec());
+  LoadURL(url);
+}
+
+void DistillerImpl::LoadURL(const GURL& url) {
+  distiller_page_->LoadURL(url);
+}
+
+void DistillerImpl::OnLoadURLDone() {
+  GetDistilledContent();
+}
+
+void DistillerImpl::GetDistilledContent() {
+  std::string script =
+      ResourceBundle::GetSharedInstance().GetRawDataResource(
+          IDR_DISTILLER_JS).as_string();
+  distiller_page_->ExecuteJavaScript(script);
+}
+
+void DistillerImpl::OnExecuteJavaScriptDone(const base::Value* value) {
+  std::string result;
+  const base::ListValue* result_list = NULL;
+  if (!value->GetAsList(&result_list)) {
+    DCHECK(proto_);
+    distillation_cb_.Run(proto_.Pass());
+    return;
+  }
+  int i = 0;
+  for (base::ListValue::const_iterator iter = result_list->begin();
+      iter != result_list->end(); ++iter, ++i) {
+    std::string item;
+    (*iter)->GetAsString(&item);
+    // The JavaScript returns an array where the first element is the title,
+    // the second element is the article content HTML, and the remaining
+    // elements are image URLs referenced in the HTML.
+    switch (i) {
+      case 0:
+        proto_->set_title(item);
+        break;
+      case 1:
+        proto_->set_html(item);
+        break;
+      default:
+        int image_number = i - 2;
+        std::string image_id = base::StringPrintf("%d", image_number);
+        FetchImage(image_id, item);
+    }
+  }
+}
+
+void DistillerImpl::FetchImage(const std::string& image_id,
+                           const std::string& item) {
+  DistillerURLFetcher* fetcher =
+      distiller_url_fetcher_factory_.CreateDistillerURLFetcher();
+  image_fetchers_[image_id] = fetcher;
+  fetcher->FetchURL(item,
+                    base::Bind(&DistillerImpl::OnFetchImageDone,
+                               base::Unretained(this), image_id));
+}
+
+void DistillerImpl::OnFetchImageDone(const std::string& id,
+                                 const std::string& response) {
+  DCHECK(proto_);
+  DistilledPageProto_Image* image = proto_->add_image();
+  image->set_name(id);
+  image->set_data(response);
+  DCHECK(image_fetchers_.end() != image_fetchers_.find(id));
+  delete image_fetchers_[id];
+  int result = image_fetchers_.erase(id);
+  DCHECK_EQ(1, result);
+  if (image_fetchers_.empty()) {
+    distillation_cb_.Run(proto_.Pass());
+  }
+}
+
+}  // namespace dom_distiller
diff --git a/components/dom_distiller/core/distiller.h b/components/dom_distiller/core/distiller.h
index b4c4ea4..6fe8228 100644
--- a/components/dom_distiller/core/distiller.h
+++ b/components/dom_distiller/core/distiller.h
@@ -8,28 +8,87 @@
 #include <map>
 
 #include "base/callback.h"
+#include "base/gtest_prod_util.h"
+#include "base/memory/ref_counted.h"
 #include "base/values.h"
+#include "components/dom_distiller/core/distiller_page.h"
+#include "components/dom_distiller/core/distiller_url_fetcher.h"
+#include "components/dom_distiller/core/proto/distilled_page.pb.h"
+#include "net/url_request/url_request_context_getter.h"
 #include "url/gurl.h"
 
 namespace dom_distiller {
 
-class DistilledPageProto;
+class DistillerImpl;
 
-class DistillerInterface {
+class Distiller {
  public:
-  typedef base::Callback<void(DistilledPageProto*)> DistillerCallback;
-  virtual ~DistillerInterface() {}
+  typedef base::Callback<void(
+      scoped_ptr<DistilledPageProto>)> DistillerCallback;
+  virtual ~Distiller() {}
 
-  // Distills a page, and asynchronously returns the article HTML to the
+  // Distills a page, and asynchrounously returns the article HTML to the
   // supplied callback.
   virtual void DistillPage(const GURL& url,
                            const DistillerCallback& callback) = 0;
 };
 
+
+// Factory for creating a Distiller.
 class DistillerFactory {
  public:
-  virtual ~DistillerFactory() {};
-  virtual scoped_ptr<DistillerInterface> CreateDistiller() = 0;
+  DistillerFactory(
+      scoped_ptr<DistillerPageFactory> distiller_page_factory,
+      scoped_ptr<DistillerURLFetcherFactory> distiller_url_fetcher_factory);
+  virtual ~DistillerFactory();
+  virtual Distiller* CreateDistiller();
+
+ private:
+  scoped_ptr<DistillerPageFactory> distiller_page_factory_;
+  scoped_ptr<DistillerURLFetcherFactory> distiller_url_fetcher_factory_;
+};
+
+
+// Distills a article from a page and associated pages.
+class DistillerImpl : public Distiller,
+                      public DistillerPage::Delegate {
+ public:
+  DistillerImpl(
+      const DistillerPageFactory& distiller_page_factory,
+      const DistillerURLFetcherFactory& distiller_url_fetcher_factory);
+  virtual ~DistillerImpl();
+
+  // Creates an execution context. This must be called once before any calls are
+  // made to distill the page.
+  virtual void Init();
+
+  virtual void DistillPage(const GURL& url,
+                           const DistillerCallback& callback) OVERRIDE;
+
+  // PageDistillerContext::Delegate
+  virtual void OnLoadURLDone() OVERRIDE;
+  virtual void OnExecuteJavaScriptDone(const base::Value* value) OVERRIDE;
+
+  void OnFetchImageDone(const std::string& id, const std::string& response);
+
+ private:
+  virtual void LoadURL(const GURL& url);
+  virtual void FetchImage(const std::string& image_id, const std::string& item);
+
+  // Injects JavaScript to distill a loaded page down to its important content,
+  // e.g., extracting a news article from its surrounding boilerplate.
+  void GetDistilledContent();
+
+  const DistillerPageFactory& distiller_page_factory_;
+  const DistillerURLFetcherFactory& distiller_url_fetcher_factory_;
+  scoped_ptr<DistillerPage> distiller_page_;
+  DistillerCallback distillation_cb_;
+
+  std::map<std::string, DistillerURLFetcher* > image_fetchers_;
+
+  scoped_ptr<DistilledPageProto> proto_;
+
+  DISALLOW_COPY_AND_ASSIGN(DistillerImpl);
 };
 
 }  // namespace dom_distiller
diff --git a/components/dom_distiller/core/distiller_page.cc b/components/dom_distiller/core/distiller_page.cc
new file mode 100644
index 0000000..cc486c7
--- /dev/null
+++ b/components/dom_distiller/core/distiller_page.cc
@@ -0,0 +1,67 @@
+// Copyright 2013 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/core/distiller_page.h"
+
+#include "base/logging.h"
+#include "url/gurl.h"
+
+namespace dom_distiller {
+
+DistillerPageFactory::~DistillerPageFactory() {}
+
+DistillerPage::DistillerPage(
+    DistillerPage::Delegate* delegate)
+    : state_(NO_CONTEXT), delegate_(delegate) {
+}
+
+DistillerPage::~DistillerPage() {
+}
+
+void DistillerPage::Init() {
+  DCHECK_EQ(NO_CONTEXT, state_);
+  InitImpl();
+  state_ = IDLE;
+}
+
+void DistillerPage::LoadURL(const GURL& gurl) {
+  DCHECK(state_ == IDLE ||
+         state_ == PAGE_AVAILABLE ||
+         state_ == PAGELOAD_FAILED);
+  state_ = LOADING_PAGE;
+  LoadURLImpl(gurl);
+}
+
+void DistillerPage::ExecuteJavaScript(const std::string& script) {
+  DCHECK_EQ(PAGE_AVAILABLE, state_);
+  state_ = EXECUTING_JAVASCRIPT;
+  ExecuteJavaScriptImpl(script);
+}
+
+void DistillerPage::OnLoadURLDone() {
+  DCHECK_EQ(LOADING_PAGE, state_);
+  state_ = PAGE_AVAILABLE;
+  if (!delegate_)
+    return;
+  delegate_->OnLoadURLDone();
+}
+
+void DistillerPage::OnLoadURLFailed() {
+  state_ = PAGELOAD_FAILED;
+  scoped_ptr<base::Value> empty(base::Value::CreateNullValue());
+  if (!delegate_)
+    return;
+  delegate_->OnExecuteJavaScriptDone(empty.get());
+}
+
+void DistillerPage::OnExecuteJavaScriptDone(
+    const base::Value* value) {
+  DCHECK_EQ(EXECUTING_JAVASCRIPT, state_);
+  state_ = PAGE_AVAILABLE;
+  if (!delegate_)
+    return;
+  delegate_->OnExecuteJavaScriptDone(value);
+}
+
+}  // namespace dom_distiller
diff --git a/components/dom_distiller/core/distiller_page.h b/components/dom_distiller/core/distiller_page.h
new file mode 100644
index 0000000..8769a0e
--- /dev/null
+++ b/components/dom_distiller/core/distiller_page.h
@@ -0,0 +1,100 @@
+// Copyright 2013 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_CORE_DISTILLER_PAGE_H_
+#define COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_
+
+#include "base/memory/scoped_ptr.h"
+#include "base/values.h"
+#include "url/gurl.h"
+
+namespace dom_distiller {
+
+// 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.
+class DistillerPage {
+ public:
+  class Delegate {
+   public:
+    virtual ~Delegate() {}
+    virtual void OnLoadURLDone() {}
+    virtual void OnExecuteJavaScriptDone(const base::Value* value) {}
+  };
+
+  DistillerPage(Delegate* delegate);
+
+  virtual ~DistillerPage();
+
+
+  // Initializes a |DistillerPage|. It must be called before any
+  // other functions, and must only be called once.
+  void Init();
+
+  // Loads a URL. |OnLoadURLDone| is called when the load completes or fails.
+  // May be called when the distiller is idle or a page is available.
+  void LoadURL(const GURL& url);
+  virtual void OnLoadURLDone();
+  virtual void OnLoadURLFailed();
+
+  // Injects and executes JavaScript in the context of a loaded page. |LoadURL|
+  // must complete before this function is called. May be called only when
+  // a page is available.
+  void ExecuteJavaScript(const std::string& script);
+
+  // Called when the JavaScript execution completes. |value| contains data
+  // returned by the script.
+  virtual void OnExecuteJavaScriptDone(const base::Value* value);
+
+ protected:
+  enum State {
+    // No context has yet been set in which to load or distill a page.
+    NO_CONTEXT,
+    // The page distiller has been initialized and is idle.
+    IDLE,
+    // A page is currently loading.
+    LOADING_PAGE,
+    // A page has loaded within the specified context.
+    PAGE_AVAILABLE,
+    // There was an error processing the page.
+    PAGELOAD_FAILED,
+    // JavaScript is executing within the context of the page. When the
+    // JavaScript completes, the state will be returned to |PAGE_AVAILABLE|.
+    EXECUTING_JAVASCRIPT
+  };
+
+  // Called by |Init| to do plaform-specific initialization work set up an
+  // environment in which a page can be loaded.
+  virtual void InitImpl() = 0;
+
+  // Called by |LoadURL| to carry out platform-specific instructions to load a
+  // page.
+  virtual void LoadURLImpl(const GURL& gurl) = 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(const std::string& script) = 0;
+
+
+
+  // The current state of the |DistillerPage|, initially |NO_CONTEXT|.
+  State state_;
+
+ private:
+  Delegate* delegate_;
+  DISALLOW_COPY_AND_ASSIGN(DistillerPage);
+};
+
+// Factory for generating a |DistillerPage|.
+class DistillerPageFactory {
+ public:
+  virtual ~DistillerPageFactory();
+
+  virtual scoped_ptr<DistillerPage> CreateDistillerPage(
+      DistillerPage::Delegate* delegate) const = 0;
+};
+
+}  // namespace dom_distiller
+
+#endif  // COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_
diff --git a/components/dom_distiller/core/distiller_unittest.cc b/components/dom_distiller/core/distiller_unittest.cc
new file mode 100644
index 0000000..144b41e
--- /dev/null
+++ b/components/dom_distiller/core/distiller_unittest.cc
@@ -0,0 +1,169 @@
+// Copyright 2013 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 <map>
+
+#include "base/bind.h"
+#include "base/bind_helpers.h"
+#include "base/location.h"
+#include "base/memory/scoped_ptr.h"
+#include "base/message_loop/message_loop.h"
+#include "base/values.h"
+#include "components/dom_distiller/core/distiller.h"
+#include "components/dom_distiller/core/distiller_page.h"
+#include "components/dom_distiller/core/proto/distilled_page.pb.h"
+#include "net/url_request/url_request_context_getter.h"
+#include "testing/gmock/include/gmock/gmock.h"
+#include "testing/gtest/include/gtest/gtest.h"
+#include "ui/base/resource/resource_bundle.h"
+
+#if defined(OS_ANDROID)
+#include "base/android/jni_android.h"
+#include "ui/base/android/ui_jni_registrar.h"
+#include "ui/gfx/android/gfx_jni_registrar.h"
+#endif
+
+using::testing::Invoke;
+using::testing::Return;
+using::testing::_;
+
+namespace {
+  const char kTitle[] = "Title";
+  const char kContent[] = "Content";
+  const char kURL[] = "http://a.com/";
+  const char kId0[] = "0";
+  const char kId1[] = "1";
+  const char kImageURL0[] = "http://a.com/img1.jpg";
+  const char kImageURL1[] = "http://a.com/img2.jpg";
+  const char kImageData0[] = { 'a', 'b', 'c', 'd', 'e', 0 };
+  const char kImageData1[] = { '1', '2', '3', '4', '5', 0 };
+}
+
+namespace dom_distiller {
+
+class TestDistillerURLFetcher : public DistillerURLFetcher {
+ public:
+
+  TestDistillerURLFetcher() : DistillerURLFetcher(NULL) {
+    responses_[kImageURL0] = std::string(kImageData0);
+    responses_[kImageURL1] = std::string(kImageData1);
+  }
+
+  void CallCallback(std::string url, const URLFetcherCallback& callback) {
+    callback.Run(responses_[url]);
+  }
+
+  virtual void FetchURL(const std::string& url,
+                        const URLFetcherCallback& callback) OVERRIDE {
+    ASSERT_TRUE(base::MessageLoop::current());
+    base::MessageLoop::current()->PostTask(
+        FROM_HERE,
+        base::Bind(&TestDistillerURLFetcher::CallCallback,
+                   base::Unretained(this), url, callback));
+  }
+
+  std::map<std::string, std::string> responses_;
+};
+
+
+class TestDistillerURLFetcherFactory : public DistillerURLFetcherFactory {
+ public:
+  TestDistillerURLFetcherFactory() : DistillerURLFetcherFactory(NULL) {}
+  virtual ~TestDistillerURLFetcherFactory() {}
+  virtual DistillerURLFetcher* CreateDistillerURLFetcher() const OVERRIDE {
+    return new TestDistillerURLFetcher();
+  }
+};
+
+
+class MockDistillerPage : public DistillerPage {
+ public:
+  MOCK_METHOD0(InitImpl, void());
+  MOCK_METHOD1(LoadURLImpl, void(const GURL& gurl));
+  MOCK_METHOD1(ExecuteJavaScriptImpl, void(const std::string& script));
+
+  MockDistillerPage(DistillerPage::Delegate* delegate)
+    : DistillerPage(delegate) {
+  }
+};
+
+
+class MockDistillerPageFactory : public DistillerPageFactory {
+ public:
+  MOCK_CONST_METHOD1(
+      CreateDistillerPageMock,
+      DistillerPage*(DistillerPage::Delegate* delegate));
+
+  virtual scoped_ptr<DistillerPage> CreateDistillerPage(
+      DistillerPage::Delegate* delegate) const OVERRIDE {
+    return scoped_ptr<DistillerPage>(CreateDistillerPageMock(delegate));
+  }
+};
+
+
+class DistillerTest : public testing::Test {
+ public:
+  virtual ~DistillerTest() {}
+  void OnDistillPageDone(scoped_ptr<DistilledPageProto> proto) {
+    proto_ = proto.Pass();
+  }
+
+ protected:
+  scoped_ptr<DistillerImpl> distiller_;
+  scoped_ptr<DistilledPageProto> proto_;
+  MockDistillerPageFactory page_factory_;
+  TestDistillerURLFetcherFactory url_fetcher_factory_;
+};
+
+ACTION_P2(DistillerPageOnExecuteJavaScriptDone, distiller_page, list) {
+  distiller_page->OnExecuteJavaScriptDone(list);
+}
+
+ACTION_P2(CreateMockDistillerPage, list, kurl) {
+  DistillerPage::Delegate* delegate = arg0;
+  MockDistillerPage* distiller_page = new MockDistillerPage(delegate);
+  EXPECT_CALL(*distiller_page, InitImpl());
+  EXPECT_CALL(*distiller_page, LoadURLImpl(kurl))
+      .WillOnce(testing::InvokeWithoutArgs(distiller_page,
+                                           &DistillerPage::OnLoadURLDone));
+  EXPECT_CALL(*distiller_page, ExecuteJavaScriptImpl(_))
+      .WillOnce(DistillerPageOnExecuteJavaScriptDone(distiller_page, list));
+  return distiller_page;
+}
+
+TEST_F(DistillerTest, DistillPage) {
+#if defined(OS_ANDROID)
+  // Register JNI bindings for android.
+  JNIEnv* env = base::android::AttachCurrentThread();
+  gfx::android::RegisterJni(env);
+  ui::android::RegisterJni(env);
+#endif
+
+  ui::ResourceBundle::InitSharedInstanceWithLocale("en-US", NULL);
+  base::MessageLoop loop(base::MessageLoop::TYPE_UI);
+  scoped_ptr<base::ListValue> list(new base::ListValue());
+  list->AppendString(kTitle);
+  list->AppendString(kContent);
+  list->AppendString(kImageURL0);
+  list->AppendString(kImageURL1);
+  EXPECT_CALL(page_factory_,
+              CreateDistillerPageMock(_)).WillOnce(
+                  CreateMockDistillerPage(list.get(), GURL(kURL)));
+  distiller_.reset(new DistillerImpl(page_factory_, url_fetcher_factory_));
+  distiller_->Init();
+  distiller_->DistillPage(
+      GURL(kURL),
+      base::Bind(&DistillerTest::OnDistillPageDone, base::Unretained(this)));
+  base::MessageLoop::current()->RunUntilIdle();
+  EXPECT_EQ(kTitle, proto_->title());
+  EXPECT_EQ(kContent, proto_->html());
+  EXPECT_EQ(kURL, proto_->url());
+  EXPECT_EQ(2, proto_->image_size());
+  EXPECT_EQ(kImageData0, proto_->image(0).data());
+  EXPECT_EQ(kId0, proto_->image(0).name());
+  EXPECT_EQ(kImageData1, proto_->image(1).data());
+  EXPECT_EQ(kId1, proto_->image(1).name());
+}
+
+}  // namespace dom_distiller
diff --git a/components/dom_distiller/core/distiller_url_fetcher.cc b/components/dom_distiller/core/distiller_url_fetcher.cc
index 4472768..854a34b 100644
--- a/components/dom_distiller/core/distiller_url_fetcher.cc
+++ b/components/dom_distiller/core/distiller_url_fetcher.cc
@@ -15,20 +15,31 @@ using net::URLFetcher;
 
 namespace dom_distiller {
 
-DistillerURLFetcher::DistillerURLFetcher() {
+DistillerURLFetcherFactory::DistillerURLFetcherFactory(
+    net::URLRequestContextGetter* context_getter)
+  : context_getter_(context_getter) {
+}
+
+DistillerURLFetcher*
+DistillerURLFetcherFactory::CreateDistillerURLFetcher() const {
+  return new DistillerURLFetcher(context_getter_);
+}
+
+
+DistillerURLFetcher::DistillerURLFetcher(
+    net::URLRequestContextGetter* context_getter)
+  : context_getter_(context_getter) {
 }
 
 DistillerURLFetcher::~DistillerURLFetcher() {
 }
 
-void DistillerURLFetcher::FetchURL(
-    net::URLRequestContextGetter* context_getter,
-    const std::string& url,
-    const URLFetcherCallback& callback) {
+void DistillerURLFetcher::FetchURL(const std::string& url,
+                                   const URLFetcherCallback& callback) {
   // Don't allow a fetch if one is pending.
   DCHECK(!url_fetcher_ || !url_fetcher_->GetStatus().is_io_pending());
   callback_ = callback;
-  url_fetcher_.reset(CreateURLFetcher(context_getter, url));
+  url_fetcher_.reset(CreateURLFetcher(context_getter_, url));
   url_fetcher_->Start();
 }
 
diff --git a/components/dom_distiller/core/distiller_url_fetcher.h b/components/dom_distiller/core/distiller_url_fetcher.h
index c6c654b..cb1dad7 100644
--- a/components/dom_distiller/core/distiller_url_fetcher.h
+++ b/components/dom_distiller/core/distiller_url_fetcher.h
@@ -12,20 +12,32 @@
 
 namespace dom_distiller {
 
+class DistillerURLFetcher;
+
+// Class for creating a DistillerURLFetcher.
+class DistillerURLFetcherFactory {
+ public:
+  DistillerURLFetcherFactory(net::URLRequestContextGetter* context_getter);
+  virtual ~DistillerURLFetcherFactory() {}
+  virtual DistillerURLFetcher* CreateDistillerURLFetcher() const;
+
+ private:
+  net::URLRequestContextGetter* context_getter_;
+};
+
 // This class fetches a URL, and notifies the caller when the operation
 // completes or fails. If the request fails, an empty string will be returned.
 class DistillerURLFetcher : public net::URLFetcherDelegate {
  public:
-  DistillerURLFetcher();
+  DistillerURLFetcher(net::URLRequestContextGetter* context_getter);
   virtual ~DistillerURLFetcher();
 
   // Indicates when a fetch is done.
   typedef base::Callback<void(const std::string& data)> URLFetcherCallback;
 
   // Fetches a |url|. Notifies when the fetch is done via |callback|.
-  void FetchURL(net::URLRequestContextGetter* context_getter,
-                const std::string& url,
-                const URLFetcherCallback& callback);
+  virtual void FetchURL(const std::string& url,
+                        const URLFetcherCallback& callback);
 
  protected:
   virtual net::URLFetcher* CreateURLFetcher(
@@ -38,6 +50,7 @@ class DistillerURLFetcher : public net::URLFetcherDelegate {
 
   scoped_ptr<net::URLFetcher> url_fetcher_;
   URLFetcherCallback callback_;
+  net::URLRequestContextGetter* context_getter_;
   DISALLOW_COPY_AND_ASSIGN(DistillerURLFetcher);
 };
 
diff --git a/components/dom_distiller/core/distiller_url_fetcher_unittest.cc b/components/dom_distiller/core/distiller_url_fetcher_unittest.cc
index 9f4a806..b8e6367 100644
--- a/components/dom_distiller/core/distiller_url_fetcher_unittest.cc
+++ b/components/dom_distiller/core/distiller_url_fetcher_unittest.cc
@@ -29,23 +29,25 @@ public:
  protected:
   // testing::Test implementation:
   virtual void SetUp() OVERRIDE {
-    url_fetcher_.reset(new dom_distiller::DistillerURLFetcher());
+    url_fetcher_.reset(new dom_distiller::DistillerURLFetcher(NULL));
     factory_.reset(new net::FakeURLFetcherFactory(NULL));
     factory_->SetFakeResponse(
         GURL(kTestPageA),
         std::string(kTestPageAResponse, sizeof(kTestPageAResponse)),
-        net::HTTP_OK, net::URLRequestStatus::SUCCESS);
+        net::HTTP_OK,
+        net::URLRequestStatus::SUCCESS);
     factory_->SetFakeResponse(
         GURL(kTestPageB),
         std::string(kTestPageBResponse, sizeof(kTestPageBResponse)),
-        net::HTTP_INTERNAL_SERVER_ERROR, net::URLRequestStatus::FAILED);
+        net::HTTP_INTERNAL_SERVER_ERROR,
+        net::URLRequestStatus::SUCCESS);
   }
 
   void Fetch(const std::string& url,
              const std::string& expected_response) {
     base::MessageLoop loop(base::MessageLoop::TYPE_UI);
     url_fetcher_->FetchURL(
-        NULL, url,
+        url,
         base::Bind(&DistillerURLFetcherTest::FetcherCallback,
                    base::Unretained(this)));
     loop.RunUntilIdle();
diff --git a/components/dom_distiller/core/resources/distiller.js b/components/dom_distiller/core/resources/distiller.js
new file mode 100644
index 0000000..4088cc7
--- /dev/null
+++ b/components/dom_distiller/core/resources/distiller.js
@@ -0,0 +1,16 @@
+// Copyright 2013 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.
+
+(function () {
+  // TODO(bengr): Replace this JavaScript, which is only a placeholder, with
+  //              code that will extract article content from a loaded page.
+  var result = new Array(4);
+  result[0] = "Rain in Seattle";
+  result[1] = "<i>Seattle</i> It is raining. " +
+              "<img src='http://www.a.com/img0.jpg' id='0'>" +
+              "<img src='http://www.a.com/img1.jpg' id='1'>";
+  result[2] = "http://www.a.com/img0.jpg";
+  result[3] = "http://www.a.com/img1.jpg";
+  return result;
+}())
diff --git a/components/dom_distiller_resources.grd b/components/dom_distiller_resources.grd
index dd2a5f5..8713589 100644
--- a/components/dom_distiller_resources.grd
+++ b/components/dom_distiller_resources.grd
@@ -12,6 +12,7 @@
       <include name="IDR_ABOUT_DOM_DISTILLER_HTML" file="dom_distiller/webui/resources/about_dom_distiller.html" flattenhtml="true" allowexternalscript="true" type="BINDATA" />
       <include name="IDR_ABOUT_DOM_DISTILLER_CSS" file="dom_distiller/webui/resources/about_dom_distiller.css" type="BINDATA" />
       <include name="IDR_ABOUT_DOM_DISTILLER_JS" file="dom_distiller/webui/resources/about_dom_distiller.js" type="BINDATA" />
+      <include name="IDR_DISTILLER_JS" file="dom_distiller/core/resources/distiller.js" type="BINDATA" />
     </includes>
   </release>
 </grit>