Move DOM-inspecting language detection logic to Blink.

components/translate/content/renderer/translate_helper.cc

Index: components/translate/content/renderer/translate_helper.cc
diff --git a/components/translate/content/renderer/translate_helper.cc b/components/translate/content/renderer/translate_helper.cc
index b42620a4dba10a248bd23ca28a541eb2b5a7e12e..7b80f7e02d76cf5a376bd581ad9465ff01c1968e 100644
--- a/components/translate/content/renderer/translate_helper.cc
+++ b/components/translate/content/renderer/translate_helper.cc
@@ -78,46 +78,6 @@ scoped_ptr<translate::RendererCldDataProvider> CreateDataProvider(
           ->CreateRendererCldDataProvider(render_frame_observer));
 }
 
-// Returns whether the page associated with |document| is a candidate for
-// translation.  Some pages can explictly specify (via a meta-tag) that they
-// should not be translated.
-// TODO(dglazkov): This logic should be moved into Blink.
-bool HasNoTranslateMeta(WebDocument* document) {
-  WebElement head = document->head();
-  if (head.isNull() || !head.hasChildNodes())
-    return false;
-
-  const WebString meta(ASCIIToUTF16("meta"));
-  const WebString name(ASCIIToUTF16("name"));
-  const WebString google(ASCIIToUTF16("google"));
-  const WebString value(ASCIIToUTF16("value"));
-  const WebString content(ASCIIToUTF16("content"));
-
-  for (WebNode child = head.firstChild(); !child.isNull();
-      child = child.nextSibling()) {
-    if (!child.isElementNode())
-      continue;
-    WebElement element = child.to<WebElement>();
-    // Check if a tag is <meta>.
-    if (!element.hasHTMLTagName(meta))
-      continue;
-    // Check if the tag contains name="google".
-    WebString attribute = element.getAttribute(name);
-    if (attribute.isNull() || attribute != google)
-      continue;
-    // Check if the tag contains value="notranslate", or content="notranslate".
-    attribute = element.getAttribute(value);
-    if (attribute.isNull())
-      attribute = element.getAttribute(content);
-    if (attribute.isNull())
-      continue;
-    if (base::LowerCaseEqualsASCII(base::StringPiece16(attribute),
-                                   "notranslate"))
-      return true;
-  }
-  return false;
-}
-
 }  // namespace
 
 namespace translate {
@@ -153,7 +113,8 @@ void TranslateHelper::PrepareForUrl(const GURL& url) {
                                                               page_seq_no_));
   deferred_page_capture_ = false;
   deferred_page_seq_no_ = -1;
-  deferred_contents_.clear();
+  // TODO(dglazkov): This should be unnecessary.
+  detection_details_ = LanguageDetectionDetails();
   if (cld_data_polling_started_)
     return;
 
@@ -177,23 +138,23 @@ void TranslateHelper::PrepareForUrl(const GURL& url) {
   TranslateHelper::SendCldDataRequest(0, 1000);
 }
 
-void TranslateHelper::PageCaptured(const base::string16& contents) {
-  PageCapturedImpl(page_seq_no_, contents);
+void TranslateHelper::LanguageDetectionDetailsCaptured(
+    const LanguageDetectionDetails& details) {
+  detection_details_ = details;
+  DetectLanguage(page_seq_no_);
 }
 
-void TranslateHelper::PageCapturedImpl(int page_seq_no,
-                                       const base::string16& contents) {
-  // Get the document language as set by WebKit from the http-equiv
-  // meta tag for "content-language".  This may or may not also
-  // have a value derived from the actual Content-Language HTTP
+void TranslateHelper::DetectLanguage(int page_seq_no) {
+  // Document language, set by WebKit in http-equiv
+  // meta tag for "content-language", may or may not also
+  // be the same as the value derived from the actual Content-Language HTTP
   // header.  The two actually have different meanings (despite the
   // original intent of http-equiv to be an equivalent) with the former
   // being the language of the document and the latter being the
   // language of the intended audience (a distinction really only
   // relevant for things like langauge textbooks).  This distinction
   // shouldn't affect translation.
-  WebLocalFrame* main_frame = render_frame()->GetWebFrame();
-  if (!main_frame || page_seq_no_ != page_seq_no)
+  if (page_seq_no_ != page_seq_no)
     return;
 
   if (!cld_data_provider_->IsCldDataAvailable()) {
@@ -201,7 +162,6 @@ void TranslateHelper::PageCapturedImpl(int page_seq_no,
     // is loaded, if ever.
     deferred_page_capture_ = true;
     deferred_page_seq_no_ = page_seq_no;
-    deferred_contents_ = contents;
     RecordLanguageDetectionTiming(DEFERRED);
     return;
   }
@@ -215,41 +175,31 @@ void TranslateHelper::PageCapturedImpl(int page_seq_no,
     RecordLanguageDetectionTiming(RESUMED);
   }
 
-  WebDocument document = main_frame->document();
-  std::string content_language = document.contentLanguage().utf8();
-  WebElement html_element = document.documentElement();
-  std::string html_lang;
-  // |html_element| can be null element, e.g. in
-  // BrowserTest.WindowOpenClose.
-  if (!html_element.isNull())
-    html_lang = html_element.getAttribute("lang").utf8();
   std::string cld_language;
   bool is_cld_reliable;
   std::string language = DeterminePageLanguage(
-      content_language, html_lang, contents, &cld_language, &is_cld_reliable);
+      detection_details_.content_language,
+      detection_details_.html_root_language, detection_details_.contents,
+      &cld_language, &is_cld_reliable);
 
   if (language.empty())
     return;
 
   language_determined_time_ = base::TimeTicks::Now();
+  detection_details_.time = base::Time::Now();
+  detection_details_.cld_language = cld_language;
+  detection_details_.is_cld_reliable = is_cld_reliable;
+  detection_details_.adopted_language = language;
 
-  GURL url(document.url());
-  LanguageDetectionDetails details;
-  details.time = base::Time::Now();
-  details.url = url;
-  details.content_language = content_language;
-  details.cld_language = cld_language;
-  details.is_cld_reliable = is_cld_reliable;
-  details.has_notranslate = HasNoTranslateMeta(&document);
-  details.html_root_language = html_lang;
-  details.adopted_language = language;
-
-  // TODO(hajimehoshi): If this affects performance, it should be set only if
-  // translate-internals tab exists.
-  details.contents = contents;
+  // TODO(hajimehoshi): If this affects performance, details.contents should be
+  // sent
+  //  only if translate-internals tab exists.
 
   Send(new ChromeFrameHostMsg_TranslateLanguageDetermined(
-      routing_id(), details, !details.has_notranslate && !language.empty()));
+      routing_id(), detection_details_,
+      !detection_details_.has_notranslate && !language.empty()));
+
+  detection_details_ = LanguageDetectionDetails();  // Clean up for sanity
 }
 
 void TranslateHelper::CancelPendingTranslation() {
@@ -606,9 +556,8 @@ void TranslateHelper::SendCldDataRequest(const int delay_millis,
 void TranslateHelper::OnCldDataAvailable() {
   if (deferred_page_capture_) {
     deferred_page_capture_ = false; // Don't do this a second time.
-    PageCapturedImpl(deferred_page_seq_no_, deferred_contents_);
+    DetectLanguage(deferred_page_seq_no_);
     deferred_page_seq_no_ = -1; // Clean up for sanity
-    deferred_contents_.clear(); // Clean up for sanity
   }
 }