Make content text retrieval in ChromeRenderViewObserver never force layout.

chrome/renderer/chrome_render_view_observer.cc

Index: chrome/renderer/chrome_render_view_observer.cc
diff --git a/chrome/renderer/chrome_render_view_observer.cc b/chrome/renderer/chrome_render_view_observer.cc
index f221836df0072563c80fbeb8e36bec974ccc2820..a7302c3daa2f7ae085cf971e68598242ca965e67 100644
--- a/chrome/renderer/chrome_render_view_observer.cc
+++ b/chrome/renderer/chrome_render_view_observer.cc
@@ -80,16 +80,9 @@ using blink::WebView;
 using blink::WebVector;
 using blink::WebWindowFeatures;
 
-// Delay in milliseconds that we'll wait before capturing the page contents
-// and thumbnail.
+// Delay in milliseconds that we'll wait before capturing the page contents.
 static const int kDelayForCaptureMs = 500;
 
-// Typically, we capture the page data once the page is loaded.
-// Sometimes, the page never finishes to load, preventing the page capture
-// To workaround this problem, we always perform a capture after the following
-// delay.
-static const int kDelayForForcedCaptureMs = 6000;
-
 // define to write the time necessary for thumbnail/DOM text retrieval,
 // respectively, into the system debug log
 // #define TIME_TEXT_RETRIEVAL
@@ -272,10 +265,6 @@ void ChromeRenderViewObserver::DidStopLoading() {
         search_provider::AUTODETECTED_PROVIDER));
   }
 
-  // Don't capture pages including refresh meta tag.
-  if (HasRefreshMetaTag(main_frame))
-    return;
-
   CapturePageInfoLater(
       false,  // preliminary_capture
       base::TimeDelta::FromMilliseconds(
@@ -286,16 +275,12 @@ void ChromeRenderViewObserver::DidStopLoading() {
 void ChromeRenderViewObserver::DidCommitProvisionalLoad(
     WebLocalFrame* frame, bool is_new_navigation) {
   // Don't capture pages being not new, or including refresh meta tag.
-  if (!is_new_navigation || HasRefreshMetaTag(frame))
+  if (!is_new_navigation)
     return;
 
   base::debug::SetCrashKeyValue(
       crash_keys::kViewCount,
       base::SizeTToString(content::RenderView::GetRenderViewCount()));
-
-  CapturePageInfoLater(
-      true,  // preliminary_capture
-      base::TimeDelta::FromMilliseconds(kDelayForForcedCaptureMs));
 }
 
 void ChromeRenderViewObserver::CapturePageInfoLater(bool preliminary_capture,
@@ -308,6 +293,10 @@ void ChromeRenderViewObserver::CapturePageInfoLater(bool preliminary_capture,
                  preliminary_capture));
 }
 
+void ChromeRenderViewObserver::DidFirstLayoutAfterFinishedParsing() {
+  CapturePageInfo(true);
+}
+
 void ChromeRenderViewObserver::CapturePageInfo(bool preliminary_capture) {
   if (!render_view()->GetWebView())
     return;
@@ -323,6 +312,10 @@ void ChromeRenderViewObserver::CapturePageInfo(bool preliminary_capture) {
   if (main_frame->isWebRemoteFrame())
     return;
 
+  WebLocalFrame* local_frame = main_frame->toWebLocalFrame();
+  if (local_frame->isNavigationScheduled())
+    return;
+
   // Don't index/capture pages that are in view source mode.
   if (main_frame->isViewSourceModeEnabled())
     return;
@@ -388,30 +381,3 @@ void ChromeRenderViewObserver::CaptureText(WebFrame* frame,
       contents->resize(last_space_index);
   }
 }
-
-bool ChromeRenderViewObserver::HasRefreshMetaTag(WebFrame* frame) {
-  if (!frame)
-    return false;
-  WebElement head = frame->document().head();
-  if (head.isNull() || !head.hasChildNodes())
-    return false;
-
-  const WebString tag_name(base::ASCIIToUTF16("meta"));
-  const WebString attribute_name(base::ASCIIToUTF16("http-equiv"));
-
-  WebNodeList children = head.childNodes();
-  for (size_t i = 0; i < children.length(); ++i) {
-    WebNode node = children.item(i);
-    if (!node.isElementNode())
-      continue;
-    WebElement element = node.to<WebElement>();
-    if (!element.hasHTMLTagName(tag_name))
-      continue;
-    WebString value = element.getAttribute(attribute_name);
-    if (value.isNull() ||
-        !base::LowerCaseEqualsASCII(base::StringPiece16(value), "refresh"))
-      continue;
-    return true;
-  }
-  return false;
-}