summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--chrome/renderer/render_view.cc38
-rw-r--r--chrome/renderer/render_view.h6
-rw-r--r--chrome/renderer/render_view_browsertest.cc69
-rw-r--r--chrome/renderer/translate_helper.cc56
-rw-r--r--chrome/renderer/translate_helper.h13
-rw-r--r--webkit/glue/dom_operations.cc11
-rw-r--r--webkit/glue/dom_operations.h11
7 files changed, 167 insertions, 37 deletions
diff --git a/chrome/renderer/render_view.cc b/chrome/renderer/render_view.cc
index 861722a..287b44c 100644
--- a/chrome/renderer/render_view.cc
+++ b/chrome/renderer/render_view.cc
@@ -889,15 +889,22 @@ void RenderView::CapturePageInfo(int load_id, bool preliminary_capture) {
string16 contents;
CaptureText(main_frame, &contents);
if (contents.size()) {
- base::TimeTicks begin_time = base::TimeTicks::Now();
- std::string language = DetermineTextLanguage(contents);
- UMA_HISTOGRAM_MEDIUM_TIMES("Renderer4.LanguageDetection",
- base::TimeTicks::Now() - begin_time);
WebKit::WebDocument document = main_frame->document();
+ // If the page explicitly specifies a language, use it, otherwise we'll
+ // determine it based on the text content using the CLD.
+ std::string language =
+ TranslateHelper::GetPageLanguageFromMetaTag(&document);
+ if (language.empty()) {
+ base::TimeTicks begin_time = base::TimeTicks::Now();
+ language = DetermineTextLanguage(contents);
+ UMA_HISTOGRAM_MEDIUM_TIMES("Renderer4.LanguageDetection",
+ base::TimeTicks::Now() - begin_time);
+ }
// Send the text to the browser for indexing (the browser might decide not
// to index, if the URL is HTTPS for instance) and language discovery.
- Send(new ViewHostMsg_PageContents(routing_id_, url, load_id, contents,
- language, IsPageTranslatable(&document)));
+ Send(new ViewHostMsg_PageContents(
+ routing_id_, url, load_id, contents, language,
+ TranslateHelper::IsPageTranslatable(&document)));
}
OnCaptureThumbnail();
@@ -1799,7 +1806,7 @@ void RenderView::didStopLoading() {
FROM_HERE,
page_info_method_factory_.NewRunnableMethod(
&RenderView::CapturePageInfo, page_id_, false),
- kDelayForCaptureMs);
+ send_content_state_immediately_ ? 0 : kDelayForCaptureMs);
}
bool RenderView::isSmartInsertDeleteEnabled() {
@@ -2089,23 +2096,6 @@ void RenderView::UpdateTargetURL(const GURL& url, const GURL& fallback_url) {
}
}
-bool RenderView::IsPageTranslatable(WebKit::WebDocument* document) {
- std::vector<WebKit::WebElement> meta_elements;
- webkit_glue::GetMetaElementsWithName(document, ASCIIToUTF16("google"),
- &meta_elements);
- std::vector<WebKit::WebElement>::const_iterator iter;
- for (iter = meta_elements.begin(); iter != meta_elements.end(); ++iter) {
- WebString attribute = iter->getAttribute("value");
- if (attribute.isNull()) // We support both 'value' and 'content'.
- attribute = iter->getAttribute("content");
- if (attribute.isNull())
- continue;
- if (LowerCaseEqualsASCII(attribute, "notranslate"))
- return false;
- }
- return true;
-}
-
void RenderView::StartNavStateSyncTimerIfNecessary() {
int delay;
if (send_content_state_immediately_)
diff --git a/chrome/renderer/render_view.h b/chrome/renderer/render_view.h
index 8e4d807..b318548 100644
--- a/chrome/renderer/render_view.h
+++ b/chrome/renderer/render_view.h
@@ -1005,9 +1005,9 @@ class RenderView : public RenderWidget,
HostContentSettings host_content_settings_;
HostZoomLevels host_zoom_levels_;
- // Whether content state (such as form state and scroll position) should be
- // sent to the browser immediately. This is normally false, but set to true
- // by some tests.
+ // Whether content state (such as form state, scroll position and page
+ // contents) should be sent to the browser immediately. This is normally
+ // false, but set to true by some tests.
bool send_content_state_immediately_;
// Stores if loading of images, scripts, and plugins is allowed.
diff --git a/chrome/renderer/render_view_browsertest.cc b/chrome/renderer/render_view_browsertest.cc
index f5539b4..35cdb6d 100644
--- a/chrome/renderer/render_view_browsertest.cc
+++ b/chrome/renderer/render_view_browsertest.cc
@@ -1089,3 +1089,72 @@ TEST_F(RenderViewTest, FillFormElement) {
ASSERT_EQ(static_cast<IPC::Message*>(NULL), message2);
EXPECT_EQ(firstname.value(), WebKit::WebString::fromUTF8("David"));
}
+
+// Tests that we send the right translatable for a page and that we respect the
+// "no translate" meta-tag.
+TEST_F(RenderViewTest, TranslatablePage) {
+ // Suppress the normal delay that occurs when the page is loaded before which
+ // the renderer sends the page contents to the browser.
+ view_->set_send_content_state_immediately(true);
+
+ LoadHTML("<html><body>A random page with random content.</body></html>");
+ ProcessPendingMessages();
+ const IPC::Message* message = render_thread_.sink().GetUniqueMessageMatching(
+ ViewHostMsg_PageContents::ID);
+ ASSERT_NE(static_cast<IPC::Message*>(NULL), message);
+ ViewHostMsg_PageContents::Param params;
+ ViewHostMsg_PageContents::Read(message, &params);
+ EXPECT_TRUE(params.e); // Translatable should be true.
+ render_thread_.sink().ClearMessages();
+
+ // Now the page specifies the META tag to prevent translation.
+ LoadHTML("<html><head><meta name=\"google\" value=\"notranslate\"></head>"
+ "<body>A random page with random content.</body></html>");
+ ProcessPendingMessages();
+ message = render_thread_.sink().GetUniqueMessageMatching(
+ ViewHostMsg_PageContents::ID);
+ ASSERT_NE(static_cast<IPC::Message*>(NULL), message);
+ ViewHostMsg_PageContents::Read(message, &params);
+ EXPECT_FALSE(params.e); // Translatable should be false.
+ render_thread_.sink().ClearMessages();
+
+ // Try the alternate version of the META tag (content instead of value).
+ LoadHTML("<html><head><meta name=\"google\" content=\"notranslate\"></head>"
+ "<body>A random page with random content.</body></html>");
+ ProcessPendingMessages();
+ message = render_thread_.sink().GetUniqueMessageMatching(
+ ViewHostMsg_PageContents::ID);
+ ASSERT_NE(static_cast<IPC::Message*>(NULL), message);
+ ViewHostMsg_PageContents::Read(message, &params);
+ EXPECT_FALSE(params.e); // Translatable should be false.
+}
+
+// Tests that the language meta tag takes precedence over the CLD when reporting
+// the page's language.
+TEST_F(RenderViewTest, LanguageMetaTag) {
+ // Suppress the normal delay that occurs when the page is loaded before which
+ // the renderer sends the page contents to the browser.
+ view_->set_send_content_state_immediately(true);
+
+ LoadHTML("<html><head><meta http-equiv=\"content-language\" content=\"es\">"
+ "</head><body>A random page with random content.</body></html>");
+ ProcessPendingMessages();
+ const IPC::Message* message = render_thread_.sink().GetUniqueMessageMatching(
+ ViewHostMsg_PageContents::ID);
+ ASSERT_NE(static_cast<IPC::Message*>(NULL), message);
+ ViewHostMsg_PageContents::Param params;
+ ViewHostMsg_PageContents::Read(message, &params);
+ EXPECT_EQ("es", params.d);
+ render_thread_.sink().ClearMessages();
+
+ // Makes sure we support multiple languages specified.
+ LoadHTML("<html><head><meta http-equiv=\"content-language\" "
+ "content=\" fr , es,en \">"
+ "</head><body>A random page with random content.</body></html>");
+ ProcessPendingMessages();
+ message = render_thread_.sink().GetUniqueMessageMatching(
+ ViewHostMsg_PageContents::ID);
+ ASSERT_NE(static_cast<IPC::Message*>(NULL), message);
+ ViewHostMsg_PageContents::Read(message, &params);
+ EXPECT_EQ("fr", params.d);
+}
diff --git a/chrome/renderer/translate_helper.cc b/chrome/renderer/translate_helper.cc
index d27f107..fafd52d 100644
--- a/chrome/renderer/translate_helper.cc
+++ b/chrome/renderer/translate_helper.cc
@@ -12,6 +12,7 @@
#include "third_party/WebKit/WebKit/chromium/public/WebScriptSource.h"
#include "third_party/WebKit/WebKit/chromium/public/WebView.h"
#include "v8/include/v8.h"
+#include "webkit/glue/dom_operations.h"
using WebKit::WebFrame;
using WebKit::WebScriptSource;
@@ -103,6 +104,61 @@ void TranslateHelper::CancelPendingTranslation() {
target_lang_.clear();
}
+// static
+bool TranslateHelper::IsPageTranslatable(WebKit::WebDocument* document) {
+ std::vector<WebKit::WebElement> meta_elements;
+ webkit_glue::GetMetaElementsWithAttribute(document,
+ ASCIIToUTF16("name"),
+ ASCIIToUTF16("google"),
+ &meta_elements);
+ std::vector<WebKit::WebElement>::const_iterator iter;
+ for (iter = meta_elements.begin(); iter != meta_elements.end(); ++iter) {
+ WebKit::WebString attribute = iter->getAttribute("value");
+ if (attribute.isNull()) // We support both 'value' and 'content'.
+ attribute = iter->getAttribute("content");
+ if (attribute.isNull())
+ continue;
+ if (LowerCaseEqualsASCII(attribute, "notranslate"))
+ return false;
+ }
+ return true;
+}
+
+// static
+std::string TranslateHelper::GetPageLanguageFromMetaTag(
+ WebKit::WebDocument* document) {
+ // The META language tag looks like:
+ // <meta http-equiv="content-language" content="en">
+ // It can contain more than one language:
+ // <meta http-equiv="content-language" content="en, fr">
+ std::vector<WebKit::WebElement> meta_elements;
+ webkit_glue::GetMetaElementsWithAttribute(document,
+ ASCIIToUTF16("http-equiv"),
+ ASCIIToUTF16("content-language"),
+ &meta_elements);
+ if (meta_elements.empty())
+ return std::string();
+
+ // We don't expect more than one such tag. If there are several, just use the
+ // first one.
+ WebKit::WebString attribute = meta_elements[0].getAttribute("content");
+ if (attribute.isEmpty())
+ return std::string();
+
+ // The value is supposed to be ASCII.
+ if (!IsStringASCII(attribute))
+ return std::string();
+
+ std::string language = StringToLowerASCII(UTF16ToASCII(attribute));
+ size_t coma_index = language.find(',');
+ if (coma_index != std::string::npos) {
+ // There are more than 1 language specified, just keep the first one.
+ language = language.substr(0, coma_index);
+ }
+ TrimWhitespaceASCII(language, TRIM_ALL, &language);
+ return language;
+}
+
////////////////////////////////////////////////////////////////////////////////
// TranslateHelper, protected:
//
diff --git a/chrome/renderer/translate_helper.h b/chrome/renderer/translate_helper.h
index bec9e02..0e3a989 100644
--- a/chrome/renderer/translate_helper.h
+++ b/chrome/renderer/translate_helper.h
@@ -13,6 +13,7 @@
class RenderView;
namespace WebKit {
+class WebDocument;
class WebFrame;
}
@@ -39,6 +40,18 @@ class TranslateHelper {
// revert existing translations.
void CancelPendingTranslation();
+ // 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.
+ static bool IsPageTranslatable(WebKit::WebDocument* document);
+
+ // Returns the language specified in the language meta tag of |document|, or
+ // an empty string if no such tag was found.
+ // The tag may specify several languages, the first one is returned.
+ // Example of such meta-tag:
+ // <meta http-equiv="content-language" content="en, fr">
+ static std::string GetPageLanguageFromMetaTag(WebKit::WebDocument* document);
+
protected:
// The following methods are protected so they can be overridden in
// unit-tests.
diff --git a/webkit/glue/dom_operations.cc b/webkit/glue/dom_operations.cc
index b39943f..724e200 100644
--- a/webkit/glue/dom_operations.cc
+++ b/webkit/glue/dom_operations.cc
@@ -592,9 +592,10 @@ int NumberOfActiveAnimations(WebView* view) {
return controller->numberOfActiveAnimations();
}
-void GetMetaElementsWithName(WebDocument* document,
- const string16& name,
- std::vector<WebElement>* meta_elements) {
+void GetMetaElementsWithAttribute(WebDocument* document,
+ const string16& attribute_name,
+ const string16& attribute_value,
+ std::vector<WebElement>* meta_elements) {
DCHECK(document);
DCHECK(meta_elements);
meta_elements->clear();
@@ -610,8 +611,8 @@ void GetMetaElementsWithName(WebDocument* document,
WebElement element = node.to<WebElement>();
if (!element.hasTagName("meta"))
continue;
- WebString meta_name = element.getAttribute("name");
- if (meta_name.isNull() || meta_name != name)
+ WebString value = element.getAttribute(attribute_name);
+ if (value.isNull() || value != attribute_value)
continue;
meta_elements->push_back(element);
}
diff --git a/webkit/glue/dom_operations.h b/webkit/glue/dom_operations.h
index a2dac8c..20084f7 100644
--- a/webkit/glue/dom_operations.h
+++ b/webkit/glue/dom_operations.h
@@ -132,11 +132,12 @@ int NumberOfActiveAnimations(WebKit::WebView* view);
WebKit::WebString GetSubResourceLinkFromElement(
const WebKit::WebElement& element);
-// Puts the meta-elements of |document| that have the specified |name| in
-// |meta_elements|.
-void GetMetaElementsWithName(WebKit::WebDocument* document,
- const string16& name,
- std::vector<WebKit::WebElement>* meta_elements);
+// Puts the meta-elements of |document| that have the attribute |attribute_name|
+// with a value of |attribute_value| in |meta_elements|.
+void GetMetaElementsWithAttribute(WebKit::WebDocument* document,
+ const string16& attribute_name,
+ const string16& atribute_value,
+ std::vector<WebKit::WebElement>* meta_elements);
} // namespace webkit_glue
generated by cgit v1.1 at 2025-01-23 06:13:22 +0000