Add saveImageFromDataURL() and use it in saveImageAt().

Source/web/WebViewImpl.cpp

Index: Source/web/WebViewImpl.cpp
diff --git a/Source/web/WebViewImpl.cpp b/Source/web/WebViewImpl.cpp
index 572e42bd4c3d09d6e3479a593b63e3b8d3a00b5e..e7ffa0fc2bd3df2950bc6a1519b5ffe63e9b0350 100644
--- a/Source/web/WebViewImpl.cpp
+++ b/Source/web/WebViewImpl.cpp
@@ -3296,18 +3296,18 @@ void WebViewImpl::copyImageAt(const WebPoint& point)
 
 void WebViewImpl::saveImageAt(const WebPoint& point)
 {
-    if (!m_page)
+    if (!m_client)
         return;
 
-    KURL url = hitTestResultForWindowPos(point).absoluteImageURLIncludingCanvasDataURL();
+    Node* node = hitTestResultForWindowPos(point).innerNonSharedNode();
+    if (!node || !(isHTMLCanvasElement(*node) || isHTMLImageElement(*node)))
+        return;
 
-    if (url.isEmpty())
+    String url = toElement(*node).imageSourceURL();
+    if (!KURL(KURL(), url).protocolIsData())
         return;

[Justin Novosad, 2014/09/05 17:22:46]

If the user attempts to save an image with an http URL, we just return? Isn't
that a regression?  I think what kbr was suggestion is that in the case we
should do things the old way.

[zino, 2014/09/05 18:00:26]

On 2014/09/05 17:22:46, junov wrote:

If the user attempts to save an image with an http URL, we don't use this code
path. In that case, browser process can immediately start to download the image
without this API. because the process already knows the URL.

<canvas> : use saveImageAt()
<img> with a large data url : use saveImageAt()
<img> with http, https, and so on : use WebContents::saveFrame()

[Ken Russell (switch to Gerrit), 2014/09/11 00:43:06]

Thanks for clarifying this.

 
-    ResourceRequest request(url);
-    request.setRequestContext(WebURLRequest::RequestContextDownload);
-    m_page->deprecatedLocalMainFrame()->loader().client()->loadURLExternally(
-        request, NavigationPolicyDownloadTo, WebString());
+    m_client->saveImageFromDataURL(url);
 }
 
 void WebViewImpl::dragSourceEndedAt(