Moving the CLD code to the renderer so it runs sanboxed.

BUG=
TEST=Run the unit-tests.

Review URL: http://codereview.chromium.org/541013

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@36045 0039d316-1c4b-4281-b951-d872f2087c98

8 files changed, 118 insertions, 129 deletions

diff --git a/chrome/browser/extensions/extension_tabs_module.cc b/chrome/browser/extensions/extension_tabs_module.cc
index 1e36294..edfd5f5 100644
--- a/chrome/browser/extensions/extension_tabs_module.cc
+++ b/chrome/browser/extensions/extension_tabs_module.cc
@@ -905,7 +905,7 @@ bool DetectTabLanguageFunction::RunImpl() {
   // object that the language has been received.
   contents->GetPageLanguage();
   registrar_.Add(this, NotificationType::TAB_LANGUAGE_DETERMINED,
-      NotificationService::AllSources());
+                 Source<RenderViewHost>(contents->render_view_host()));
   AddRef();  // balanced in Observe()
   return true;
 }
diff --git a/chrome/browser/renderer_host/render_view_host.cc b/chrome/browser/renderer_host/render_view_host.cc
index fa2c4b2..5d46ccc 100644
--- a/chrome/browser/renderer_host/render_view_host.cc
+++ b/chrome/browser/renderer_host/render_view_host.cc
@@ -47,8 +47,6 @@
 #if defined(OS_WIN)
 // TODO(port): accessibility not yet implemented. See http://crbug.com/8288.
 #include "chrome/browser/browser_accessibility_manager.h"
-// TODO(port): The compact language detection library works only for Windows.
-#include "third_party/cld/bar/toolbar/cld/i18n/encodings/compact_lang_det/win/cld_unicodetext.h"
 #endif
 
 using base::TimeDelta;
@@ -425,7 +423,7 @@ void RenderViewHost::StopFinding(bool clear_selection) {
 }
 
 void RenderViewHost::GetPageLanguage() {
-  Send(new ViewMsg_DeterminePageText(routing_id()));
+  Send(new ViewMsg_DeterminePageLanguage(routing_id()));
 }
 
 void RenderViewHost::Zoom(PageZoom::Function function) {
@@ -765,8 +763,8 @@ void RenderViewHost::OnMessageReceived(const IPC::Message& msg) {
     IPC_MESSAGE_HANDLER(ViewHostMsg_DidFailProvisionalLoadWithError,
                         OnMsgDidFailProvisionalLoadWithError)
     IPC_MESSAGE_HANDLER(ViewHostMsg_Find_Reply, OnMsgFindReply)
-    IPC_MESSAGE_HANDLER(ViewMsg_DeterminePageText_Reply,
-                        OnDeterminePageTextReply)
+    IPC_MESSAGE_HANDLER(ViewHostMsg_PageLanguageDetermined,
+                        OnPageLanguageDetermined)
     IPC_MESSAGE_HANDLER(ViewMsg_ExecuteCodeFinished,
                         OnExecuteCodeFinished)
     IPC_MESSAGE_HANDLER(ViewHostMsg_UpdateFavIconURL, OnMsgUpdateFavIconURL)
@@ -1163,20 +1161,12 @@ void RenderViewHost::OnMsgFindReply(int request_id,
   Send(new ViewMsg_FindReplyACK(routing_id()));
 }
 
-void RenderViewHost::OnDeterminePageTextReply(
-    const std::wstring& page_text) {
-#if defined(OS_WIN)  // Only for windows.
-    int num_languages = 0;
-    bool is_reliable = false;
-    const char* language_iso_code = LanguageCodeISO639_1(
-        DetectLanguageOfUnicodeText(NULL, page_text.c_str(), true, &is_reliable,
-                                    &num_languages, NULL));
-    std::string language(language_iso_code);
-    NotificationService::current()->Notify(
-        NotificationType::TAB_LANGUAGE_DETERMINED,
-        Source<RenderViewHost>(this),
-        Details<std::string>(&language));
-#endif
+void RenderViewHost::OnPageLanguageDetermined(const std::string& language) {
+  std::string lang(language);
+  NotificationService::current()->Notify(
+      NotificationType::TAB_LANGUAGE_DETERMINED,
+      Source<RenderViewHost>(this),
+      Details<std::string>(&lang));
 }
 
 void RenderViewHost::OnExecuteCodeFinished(int request_id, bool success) {
diff --git a/chrome/browser/renderer_host/render_view_host.h b/chrome/browser/renderer_host/render_view_host.h
index 7c3dbb3..cfd1a77 100644
--- a/chrome/browser/renderer_host/render_view_host.h
+++ b/chrome/browser/renderer_host/render_view_host.h
@@ -212,9 +212,9 @@ class RenderViewHost : public RenderWidgetHost,
   // clear the selection on the focused frame.
   void StopFinding(bool clear_selection);
 
-  // Get the most probable language of the text content in the tab. This sends
-  // a message to the render view to get the content of the page as text. The
-  // caller gets the language via the NotificationService by registering to the
+  // Gets the most probable language of the text content in the tab. (This sends
+  // a message to the render view.)  The caller gets the language via the
+  // NotificationService by registering to the
   // NotificationType TAB_LANGUAGE_DETERMINED.
   void GetPageLanguage();
 
@@ -499,7 +499,7 @@ class RenderViewHost : public RenderWidgetHost,
                       const gfx::Rect& selection_rect,
                       int active_match_ordinal,
                       bool final_update);
-  void OnDeterminePageTextReply(const std::wstring& tab_text);
+  void OnPageLanguageDetermined(const std::string& language);
   void OnExecuteCodeFinished(int request_id, bool success);
   void OnMsgUpdateFavIconURL(int32 page_id, const GURL& icon_url);
   void OnMsgDidDownloadFavIcon(int id,
diff --git a/chrome/chrome.gyp b/chrome/chrome.gyp
index 61910b5..49003b4 100755
--- a/chrome/chrome.gyp
+++ b/chrome/chrome.gyp
@@ -1199,7 +1199,6 @@
             '../third_party/bsdiff/bsdiff.gyp:*',
             '../third_party/bspatch/bspatch.gyp:*',
             '../third_party/bzip2/bzip2.gyp:*',
-            '../third_party/cld/cld.gyp:cld',
             '../third_party/codesighs/codesighs.gyp:*',
             '../third_party/icu/icu.gyp:*',
             '../third_party/libjpeg/libjpeg.gyp:*',
diff --git a/chrome/chrome_renderer.gypi b/chrome/chrome_renderer.gypi
index 40a066a..b6a44b1 100755
--- a/chrome/chrome_renderer.gypi
+++ b/chrome/chrome_renderer.gypi
@@ -163,8 +163,12 @@
         # Windows-specific rules.
         ['OS=="win"', {
           'include_dirs': [
+            '../third_party/cld',
             'third_party/wtl/include',
           ],
+          'dependencies': [
+            '../third_party/cld/cld.gyp:cld',
+          ],
           'conditions': [
             ['win_use_allocator_shim==1', {
               'dependencies': [
diff --git a/chrome/common/render_messages_internal.h b/chrome/common/render_messages_internal.h
index b4e1e92..3ab66e2 100644
--- a/chrome/common/render_messages_internal.h
+++ b/chrome/common/render_messages_internal.h
@@ -257,13 +257,8 @@ IPC_BEGIN_MESSAGES(View)
                       string16 /* search_text */,
                       WebKit::WebFindOptions)
 
-  // Send from the browser to the rendered to get the text content of the page.
-  IPC_MESSAGE_ROUTED0(ViewMsg_DeterminePageText)
-
-  // Send from the renderer to the browser to return the text content of the
-  // page.
-  IPC_MESSAGE_ROUTED1(ViewMsg_DeterminePageText_Reply,
-                      std::wstring /* the language */)
+  // Asks the renderer for the language of the current page.
+  IPC_MESSAGE_ROUTED0(ViewMsg_DeterminePageLanguage)
 
   // Send from the renderer to the browser to return the script running result.
   IPC_MESSAGE_ROUTED2(ViewMsg_ExecuteCodeFinished,
@@ -1939,6 +1934,11 @@ IPC_BEGIN_MESSAGES(ViewHost)
                        string16 /* origin identifier */,
                        string16 /* database name */)
 
+  // Notifies the browser of the language (ISO 639_1 code language, such as fr,
+  // en, zh...) of the current page.
+  IPC_MESSAGE_ROUTED1(ViewHostMsg_PageLanguageDetermined,
+                      std::string /* the language */)
+
   //---------------------------------------------------------------------------
   // Socket Stream messages:
   // These are messages from the SocketStreamHandle to the browser.
diff --git a/chrome/renderer/render_view.cc b/chrome/renderer/render_view.cc
index 2dc6b75..6371372 100644
--- a/chrome/renderer/render_view.cc
+++ b/chrome/renderer/render_view.cc
@@ -17,12 +17,10 @@
 #include "base/command_line.h"
 #include "base/compiler_specific.h"
 #include "base/field_trial.h"
-#include "base/histogram.h"
 #include "base/process_util.h"
 #include "base/singleton.h"
 #include "base/string_piece.h"
 #include "base/string_util.h"
-#include "base/time.h"
 #include "build/build_config.h"
 #include "chrome/common/bindings_policy.h"
 #include "chrome/common/child_process_logging.h"
@@ -66,6 +64,10 @@
 #include "net/base/net_errors.h"
 #include "skia/ext/bitmap_platform_device.h"
 #include "skia/ext/image_operations.h"
+#if defined(OS_WIN)
+// TODO(port): The compact language detection library works only for Windows.
+#include "third_party/cld/bar/toolbar/cld/i18n/encodings/compact_lang_det/win/cld_unicodetext.h"
+#endif
 #include "third_party/WebKit/WebKit/chromium/public/WebAccessibilityCache.h"
 #include "third_party/WebKit/WebKit/chromium/public/WebAccessibilityObject.h"
 #include "third_party/WebKit/WebKit/chromium/public/WebCString.h"
@@ -174,6 +176,7 @@ using WebKit::WebWorkerClient;
 
 // define to write the time necessary for thumbnail/DOM text retrieval,
 // respectively, into the system debug log
+// #define TIME_BITMAP_RETRIEVAL
 // #define TIME_TEXT_RETRIEVAL
 
 // maximum number of characters in the document to index, any text beyond this
@@ -213,6 +216,9 @@ static const char* const kUnreachableWebDataURL =
 
 static const char* const kBackForwardNavigationScheme = "history";
 
+// The string returned in DetectLanguage if we failed to detect the language.
+static const char* const kUnknownLanguageCode = "unknown";
+
 static void GetRedirectChain(WebDataSource* ds, std::vector<GURL>* result) {
   WebVector<WebURL> urls;
   ds->redirectChain(urls);
@@ -231,35 +237,6 @@ static bool UrlMatchesPermissions(
   return false;
 }
 
-static bool PaintViewIntoCanvas(WebView* view,
-                                skia::PlatformCanvas& canvas) {
-  view->layout();
-  const WebSize& size = view->size();
-
-  if (!canvas.initialize(size.width, size.height, true))
-    return false;
-
-  view->paint(webkit_glue::ToWebCanvas(&canvas),
-              WebRect(0, 0, size.width, size.height));
-  // TODO: Add a way to snapshot the whole page, not just the currently
-  // visible part.
-
-  return true;
-}
-
-// Calculates how "boring" a thumbnail is. The boring score is the
-// 0,1 ranged percentage of pixels that are the most common
-// luma. Higher boring scores indicate that a higher percentage of a
-// bitmap are all the same brightness.
-static double CalculateBoringScore(SkBitmap* bitmap) {
-  int histogram[256] = {0};
-  color_utils::BuildLumaHistogram(bitmap, histogram);
-
-  int color_count = *std::max_element(histogram, histogram + 256);
-  int pixel_count = bitmap->width() * bitmap->height();
-  return static_cast<double>(color_count) / pixel_count;
-}
-
 ///////////////////////////////////////////////////////////////////////////////
 
 int32 RenderView::next_page_id_ = 1;
@@ -452,7 +429,6 @@ void RenderView::OnMessageReceived(const IPC::Message& message) {
 
   IPC_BEGIN_MESSAGE_MAP(RenderView, message)
     IPC_MESSAGE_HANDLER(ViewMsg_CaptureThumbnail, SendThumbnail)
-    IPC_MESSAGE_HANDLER(ViewMsg_CaptureSnapshot, SendSnapshot)
     IPC_MESSAGE_HANDLER(ViewMsg_PrintPages, OnPrintPages)
     IPC_MESSAGE_HANDLER(ViewMsg_PrintingDone, OnPrintingDone)
     IPC_MESSAGE_HANDLER(ViewMsg_Navigate, OnNavigate)
@@ -477,7 +453,7 @@ void RenderView::OnMessageReceived(const IPC::Message& message) {
     IPC_MESSAGE_HANDLER(ViewMsg_CopyImageAt, OnCopyImageAt)
     IPC_MESSAGE_HANDLER(ViewMsg_ExecuteEditCommand, OnExecuteEditCommand)
     IPC_MESSAGE_HANDLER(ViewMsg_Find, OnFind)
-    IPC_MESSAGE_HANDLER(ViewMsg_DeterminePageText, OnDeterminePageText)
+    IPC_MESSAGE_HANDLER(ViewMsg_DeterminePageLanguage, OnDeterminePageLanguage)
     IPC_MESSAGE_HANDLER(ViewMsg_Zoom, OnZoom)
     IPC_MESSAGE_HANDLER(ViewMsg_SetZoomLevelForLoadingHost,
                         OnSetZoomLevelForLoadingHost)
@@ -582,24 +558,6 @@ void RenderView::SendThumbnail() {
   Send(new ViewHostMsg_Thumbnail(routing_id_, url, score, thumbnail));
 }
 
-void RenderView::SendSnapshot() {
-  SkBitmap snapshot;
-  bool error = false;
-
-  WebFrame* main_frame = webview()->mainFrame();
-  if (!main_frame)
-    error = true;
-
-  if (!error && !CaptureSnapshot(webview(), &snapshot))
-    error = true;
-
-  DCHECK(error == snapshot.empty()) <<
-      "Snapshot should be empty on error, non-empty otherwise.";
-
-  // Send the snapshot to the browser process.
-  Send(new ViewHostMsg_Snapshot(routing_id_, snapshot));
-}
-
 void RenderView::OnPrintPages() {
   DCHECK(webview());
   if (webview()) {
@@ -660,10 +618,11 @@ void RenderView::CapturePageInfo(int load_id, bool preliminary_capture) {
     Send(new ViewHostMsg_PageContents(url, load_id, contents));
   }
 
-  // Send over text content of this page to the browser.
+  // Now that we have the contents, we can determine the language if necessary.
   if (determine_page_text_after_loading_stops_) {
     determine_page_text_after_loading_stops_ = false;
-    Send(new ViewMsg_DeterminePageText_Reply(routing_id_, contents));
+    Send(new ViewHostMsg_PageLanguageDetermined(
+        routing_id_, DetermineTextLanguage(contents)));
   }
 
   // thumbnail
@@ -715,13 +674,18 @@ bool RenderView::CaptureThumbnail(WebView* view,
                                   int h,
                                   SkBitmap* thumbnail,
                                   ThumbnailScore* score) {
-  base::TimeTicks beginning_time = base::TimeTicks::Now();
+#ifdef TIME_BITMAP_RETRIEVAL
+  double begin = time_util::GetHighResolutionTimeNow();
+#endif
 
-  skia::PlatformCanvas canvas;
+  view->layout();
+  const WebSize& size = view->size();
 
-  // Paint |view| into |canvas|.
-  if (!PaintViewIntoCanvas(view, canvas))
+  skia::PlatformCanvas canvas;
+  if (!canvas.initialize(size.width, size.height, true))
     return false;
+  view->paint(webkit_glue::ToWebCanvas(&canvas),
+              WebRect(0, 0, size.width, size.height));
 
   skia::BitmapPlatformDevice& device =
       static_cast<skia::BitmapPlatformDevice&>(canvas.getTopPlatformDevice());
@@ -769,28 +733,22 @@ bool RenderView::CaptureThumbnail(WebView* view,
 
   score->boring_score = CalculateBoringScore(thumbnail);
 
-  HISTOGRAM_TIMES("Renderer4.Thumbnail",
-                  base::TimeTicks::Now() - beginning_time);
+#ifdef TIME_BITMAP_RETRIEVAL
+  double end = time_util::GetHighResolutionTimeNow();
+  char buf[128];
+  sprintf_s(buf, "thumbnail in %gms\n", (end - begin) * 1000);
+  OutputDebugStringA(buf);
+#endif
   return true;
 }
 
-bool RenderView::CaptureSnapshot(WebView* view, SkBitmap* snapshot) {
-  base::TimeTicks beginning_time = base::TimeTicks::Now();
-
-  skia::PlatformCanvas canvas;
-  if (!PaintViewIntoCanvas(view, canvas))
-    return false;
-
-  skia::BitmapPlatformDevice& device =
-      static_cast<skia::BitmapPlatformDevice&>(canvas.getTopPlatformDevice());
-
-  const SkBitmap& bitmap = device.accessBitmap(false);
-  if (!bitmap.copyTo(snapshot, SkBitmap::kARGB_8888_Config))
-    return false;
+double RenderView::CalculateBoringScore(SkBitmap* bitmap) {
+  int histogram[256] = {0};
+  color_utils::BuildLumaHistogram(bitmap, histogram);
 
-  HISTOGRAM_TIMES("Renderer4.Snapshot",
-                  base::TimeTicks::Now() - beginning_time);
-  return true;
+  int color_count = *std::max_element(histogram, histogram + 256);
+  int pixel_count = bitmap->width() * bitmap->height();
+  return static_cast<double>(color_count) / pixel_count;
 }
 
 void RenderView::OnNavigate(const ViewMsg_Navigate_Params& params) {
@@ -2724,21 +2682,21 @@ webkit_glue::WebPluginDelegate* RenderView::CreatePluginDelegate(
     }
   }
   if (in_process_plugin) {
+#if defined(OS_WIN)  // In-proc plugins aren't supported on Linux or Mac.
     if (use_pepper_host) {
       return WebPluginDelegatePepper::Create(
           path,
           *mime_type_to_use,
           AsWeakPtr(),
-          0);
+          gfx::NativeViewFromId(host_window_));
     } else {
-#if defined(OS_WIN)  // In-proc plugins aren't supported on Linux or Mac.
       return WebPluginDelegateImpl::Create(
           path, *mime_type_to_use, gfx::NativeViewFromId(host_window_));
+    }
 #else
-      NOTIMPLEMENTED();
-      return NULL;
+    NOTIMPLEMENTED();
+    return NULL;
 #endif
-    }
   }
 
   return new WebPluginDelegateProxy(*mime_type_to_use, AsWeakPtr());
@@ -3032,21 +2990,47 @@ void RenderView::OnFind(int request_id, const string16& search_text,
   }
 }
 
-void RenderView::OnDeterminePageText() {
-  if (!is_loading_) {
-    if (!webview())
-      return;
-    WebFrame* main_frame = webview()->mainFrame();
-    std::wstring contents;
-    CaptureText(main_frame, &contents);
-    Send(new ViewMsg_DeterminePageText_Reply(routing_id_, contents));
-    determine_page_text_after_loading_stops_ = false;
+void RenderView::OnDeterminePageLanguage() {
+  if (is_loading_) {
+    // Wait for the page to finish loading before trying to determine the
+    // language.
+    determine_page_text_after_loading_stops_ = true;
     return;
   }
 
-  // We set |determine_page_text_after_loading_stops_| true here so that,
-  // after page has been loaded completely, the text in the page is captured.
-  determine_page_text_after_loading_stops_ = true;
+  Send(new ViewHostMsg_PageLanguageDetermined(routing_id_, DetectLanguage()));
+}
+
+std::string RenderView::DetectLanguage() {
+  if (!webview() || is_loading_)
+    return kUnknownLanguageCode;
+
+  std::string language = kUnknownLanguageCode;
+#if defined(OS_WIN)  // CLD is only available on Windows at this time.
+  WebFrame* main_frame = webview()->mainFrame();
+  std::wstring contents;
+  CaptureText(main_frame, &contents);
+  language = DetermineTextLanguage(contents);
+#endif
+
+  return language;
+}
+
+// static
+std::string RenderView::DetermineTextLanguage(const std::wstring& text) {
+  std::string language = kUnknownLanguageCode;
+#if defined(OS_WIN)  // CLD is only available on Windows at this time.
+  int num_languages = 0;
+  bool is_reliable = false;
+  Language cld_language =
+      DetectLanguageOfUnicodeText(NULL, text.c_str(), true, &is_reliable,
+                                  &num_languages, NULL);
+  if (cld_language != NUM_LANGUAGES && cld_language != UNKNOWN_LANGUAGE &&
+      cld_language != TG_UNKNOWN_LANGUAGE) {
+    language = LanguageCodeISO639_1(cld_language);
+  }
+#endif
+  return language;
 }
 
 void RenderView::DnsPrefetch(const std::vector<std::string>& host_names) {
diff --git a/chrome/renderer/render_view.h b/chrome/renderer/render_view.h
index 012c7cc..1174b3a 100644
--- a/chrome/renderer/render_view.h
+++ b/chrome/renderer/render_view.h
@@ -430,6 +430,11 @@ class RenderView : public RenderWidget,
   // UserScript::DOCUMENT_IDLE.
   void OnUserScriptIdleTriggered(WebKit::WebFrame* frame);
 
+  // Returns the ISO 639_1 language code of the current page
+  // (ex: en, fr, zh...).  Returns 'unknown' if the language could not be
+  // determined.
+  std::string DetectLanguage();
+
  protected:
   // RenderWidget overrides:
   virtual void Close();
@@ -508,9 +513,11 @@ class RenderView : public RenderWidget,
                         SkBitmap* thumbnail,
                         ThumbnailScore* score);
 
-  // Capture a snapshot of a view.  This is used to allow an extension
-  // to get a snapshot of a tab using chrome.tabs.captureVisibleTab().
-  bool CaptureSnapshot(WebKit::WebView* view, SkBitmap* snapshot);
+  // Calculates how "boring" a thumbnail is. The boring score is the
+  // 0,1 ranged percentage of pixels that are the most common
+  // luma. Higher boring scores indicate that a higher percentage of a
+  // bitmap are all the same brightness.
+  double CalculateBoringScore(SkBitmap* bitmap);
 
   bool RunJavaScriptMessage(int type,
                             const std::wstring& message,
@@ -524,7 +531,6 @@ class RenderView : public RenderWidget,
 
   // RenderView IPC message handlers
   void SendThumbnail();
-  void SendSnapshot();
   void OnPrintPages();
   void OnPrintingDone(int document_cookie, bool success);
   void OnNavigate(const ViewMsg_Navigate_Params& params);
@@ -556,7 +562,7 @@ class RenderView : public RenderWidget,
   void OnSetupDevToolsClient();
   void OnCancelDownload(int32 download_id);
   void OnFind(int request_id, const string16&, const WebKit::WebFindOptions&);
-  void OnDeterminePageText();
+  void OnDeterminePageLanguage();
   void OnZoom(PageZoom::Function function);
   void OnSetZoomLevelForLoadingHost(std::string host, int zoom_level);
   void OnSetPageEncoding(const std::string& encoding_name);
@@ -749,6 +755,12 @@ class RenderView : public RenderWidget,
   // Starts nav_state_sync_timer_ if it isn't already running.
   void StartNavStateSyncTimerIfNecessary();
 
+  // Returns the ISO 639_1 language code of the specified |text|, or 'unknown'
+  // if it failed.
+  // Note this only works on Windows at this time.  It always returns 'unknown'
+  // on other platforms.
+  static std::string DetermineTextLanguage(const std::wstring& text);
+
   // Bitwise-ORed set of extra bindings that have been enabled.  See
   // BindingsPolicy for details.
   int enabled_bindings_;