Add MetaTagObserver for retrieving information about meta tags

Add an observer that grabs information about meta tags when the
page has finished loading.

BUG=332281

Review URL: https://codereview.chromium.org/139413006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@248327 0039d316-1c4b-4281-b951-d872f2087c98

8 files changed, 222 insertions, 37 deletions

diff --git a/chrome/browser/android/meta_tag_observer.cc b/chrome/browser/android/meta_tag_observer.cc
new file mode 100644
index 0000000..77425b8
--- /dev/null
+++ b/chrome/browser/android/meta_tag_observer.cc
@@ -0,0 +1,56 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "chrome/browser/android/meta_tag_observer.h"
+
+#include "chrome/common/chrome_constants.h"
+#include "chrome/common/render_messages.h"
+#include "content/public/browser/render_view_host.h"
+#include "content/public/browser/web_contents.h"
+#include "content/public/browser/web_contents_observer.h"
+
+MetaTagObserver::MetaTagObserver(const std::string& meta_tag)
+    : meta_tag_(meta_tag) {
+  if (meta_tag_.size() > chrome::kMaxMetaTagAttributeLength) {
+    VLOG(1) << "Length of the <meta> name attribute is too large.";
+    NOTREACHED();
+  }
+}
+
+MetaTagObserver::~MetaTagObserver() {
+}
+
+void MetaTagObserver::DidFinishLoad(
+    int64 frame_id,
+    const GURL& validated_url,
+    bool is_main_frame,
+    content::RenderViewHost* render_view_host) {
+  if (!is_main_frame) return;
+  validated_url_ = validated_url;
+  Send(new ChromeViewMsg_RetrieveMetaTagContent(routing_id(),
+                                                validated_url,
+                                                meta_tag_));
+}
+
+bool MetaTagObserver::OnMessageReceived(const IPC::Message& message) {
+  bool handled = true;
+  IPC_BEGIN_MESSAGE_MAP(MetaTagObserver, message)
+    IPC_MESSAGE_HANDLER(ChromeViewHostMsg_DidRetrieveMetaTagContent,
+                        OnDidRetrieveMetaTagContent)
+    IPC_MESSAGE_UNHANDLED(handled = false)
+  IPC_END_MESSAGE_MAP()
+  return handled;
+}
+
+void MetaTagObserver::OnDidRetrieveMetaTagContent(
+    bool success,
+    const std::string& tag_name,
+    const std::string& tag_content,
+    const GURL& expected_url) {
+  if (!success || tag_name != meta_tag_ || validated_url_ != expected_url ||
+      tag_content.size() >= chrome::kMaxMetaTagAttributeLength) {
+    return;
+  }
+  HandleMetaTagContent(tag_content, expected_url);
+}
diff --git a/chrome/browser/android/meta_tag_observer.h b/chrome/browser/android/meta_tag_observer.h
new file mode 100644
index 0000000..419fa57
--- /dev/null
+++ b/chrome/browser/android/meta_tag_observer.h
@@ -0,0 +1,47 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef CHROME_BROWSER_ANDROID_META_TAG_OBSERVER_H_
+#define CHROME_BROWSER_ANDROID_META_TAG_OBSERVER_H_
+
+#include "content/public/browser/web_contents_observer.h"
+
+class GURL;
+
+// Retrieves the content of a particular meta tag when a page has fully loaded.
+// Subclasses must implement |HandleMetaTagContent|, which gets called after
+// the DOM has been scanned for the tag.
+class MetaTagObserver : public content::WebContentsObserver {
+ public:
+  explicit MetaTagObserver(const std::string& meta_tag);
+  virtual ~MetaTagObserver();
+
+  // content::WebContentsObserver overrides:
+  virtual void DidFinishLoad(
+      int64 frame_id,
+      const GURL& validated_url,
+      bool is_main_frame,
+      content::RenderViewHost* render_view_host) OVERRIDE;
+
+  virtual bool OnMessageReceived(const IPC::Message& message) OVERRIDE;
+
+ protected:
+  // Called when the meta tag's content has been retrieved successfully.
+  virtual void HandleMetaTagContent(const std::string& tag_content,
+                                    const GURL& expected_url) = 0;
+
+ private:
+  // Called when the renderer has returned information about the meta tag.
+  void OnDidRetrieveMetaTagContent(bool success,
+                                   const std::string& tag_name,
+                                   const std::string& tag_content,
+                                   const GURL& expected_url);
+
+  const std::string meta_tag_;
+  GURL validated_url_;
+
+  DISALLOW_COPY_AND_ASSIGN(MetaTagObserver);
+};
+
+#endif  // CHROME_BROWSER_ANDROID_META_TAG_OBSERVER_H_
diff --git a/chrome/chrome_browser.gypi b/chrome/chrome_browser.gypi
index 3aca70a..643d574 100644
--- a/chrome/chrome_browser.gypi
+++ b/chrome/chrome_browser.gypi
@@ -121,6 +121,8 @@
         'browser/android/field_trial_helper.h',
         'browser/android/foreign_session_helper.cc',
         'browser/android/foreign_session_helper.h',
+        'browser/android/meta_tag_observer.cc',
+        'browser/android/meta_tag_observer.h',
         'browser/android/new_tab_page_prefs.cc',
         'browser/android/new_tab_page_prefs.h',
         'browser/android/google_location_settings_helper.h',
diff --git a/chrome/common/chrome_constants.cc b/chrome/common/chrome_constants.cc
index bb629a4..31e7f11 100644
--- a/chrome/common/chrome_constants.cc
+++ b/chrome/common/chrome_constants.cc
@@ -267,6 +267,10 @@ const char kTestUserProfileDir[] = "test-user";
 const char kApplicationClientIDStringForAVScanning[] =
     "2676A9A2-D919-4FEE-9187-152100393AB2";
 
+#if defined(OS_ANDROID)
+const size_t kMaxMetaTagAttributeLength = 2000;
+#endif
+
 }  // namespace chrome
 
 #undef FPL
diff --git a/chrome/common/chrome_constants.h b/chrome/common/chrome_constants.h
index 5e7b481..f2a6770 100644
--- a/chrome/common/chrome_constants.h
+++ b/chrome/common/chrome_constants.h
@@ -155,6 +155,11 @@ extern const char kTestUserProfileDir[];
 // Used to identify the application to the system AV function in Windows.
 extern const char kApplicationClientIDStringForAVScanning[];
 
+#if defined(OS_ANDROID)
+// The largest reasonable length we'd assume for a meta tag attribute.
+extern const size_t kMaxMetaTagAttributeLength;
+#endif
+
 }  // namespace chrome
 
 #endif  // CHROME_COMMON_CHROME_CONSTANTS_H_
diff --git a/chrome/common/render_messages.h b/chrome/common/render_messages.h
index 5381fab..26035c9 100644
--- a/chrome/common/render_messages.h
+++ b/chrome/common/render_messages.h
@@ -403,6 +403,11 @@ IPC_MESSAGE_ROUTED2(ChromeViewHostMsg_RequestThumbnailForContextNode_ACK,
 // be treated as a webapp.
 IPC_MESSAGE_ROUTED1(ChromeViewMsg_RetrieveWebappInformation,
                     GURL /* expected_url */)
+
+// Asks the renderer to return information about the given meta tag.
+IPC_MESSAGE_ROUTED2(ChromeViewMsg_RetrieveMetaTagContent,
+                    GURL /* expected_url */,
+                    std::string /* tag_name */ )
 #endif  // defined(OS_ANDROID)
 
 // chrome.principals messages ------------------------------------------------
@@ -654,6 +659,12 @@ IPC_MESSAGE_ROUTED4(ChromeViewHostMsg_DidRetrieveWebappInformation,
                     bool /* is_mobile_webapp_capable */,
                     bool /* is_apple_mobile_webapp_capable */,
                     GURL /* expected_url */)
+
+IPC_MESSAGE_ROUTED4(ChromeViewHostMsg_DidRetrieveMetaTagContent,
+                    bool /* success */,
+                    std::string /* tag_name */,
+                    std::string /* tag_content */,
+                    GURL /* expected_url */)
 #endif  // defined(OS_ANDROID)
 
 // The currently displayed PDF has an unsupported feature.
diff --git a/chrome/renderer/chrome_render_view_observer.cc b/chrome/renderer/chrome_render_view_observer.cc
index 47509ba..5527529 100644
--- a/chrome/renderer/chrome_render_view_observer.cc
+++ b/chrome/renderer/chrome_render_view_observer.cc
@@ -196,6 +196,56 @@ extensions::StackTrace GetStackTraceFromMessage(
   return result;
 }
 
+#if defined(OS_ANDROID)
+// Parses the DOM for a <meta> tag with a particular name.
+// |meta_tag_content| is set to the contents of the 'content' attribute.
+// |found_tag| is set to true if the tag was successfully found.
+// Returns true if the document was parsed without errors.
+bool RetrieveMetaTagContent(const WebFrame* main_frame,
+                            const GURL& expected_url,
+                            const std::string& meta_tag_name,
+                            bool* found_tag,
+                            std::string* meta_tag_content) {
+  WebDocument document =
+      main_frame ? main_frame->document() : WebDocument();
+  WebElement head = document.isNull() ? WebElement() : document.head();
+  GURL document_url = document.isNull() ? GURL() : GURL(document.url());
+
+  // Search the DOM for the <meta> tag with the given name.
+  *found_tag = false;
+  *meta_tag_content = "";
+  if (!head.isNull()) {
+    WebNodeList children = head.childNodes();
+    for (unsigned i = 0; i < children.length(); ++i) {
+      WebNode child = children.item(i);
+      if (!child.isElementNode())
+        continue;
+      WebElement elem = child.to<WebElement>();
+      if (elem.hasTagName("meta")) {
+        if (elem.hasAttribute("name") && elem.hasAttribute("content")) {
+          std::string name = elem.getAttribute("name").utf8();
+          if (name == meta_tag_name) {
+            *meta_tag_content = elem.getAttribute("content").utf8();
+            *found_tag = true;
+            break;
+          }
+        }
+      }
+    }
+  }
+
+  // Make sure we're checking the right page and that the length of the content
+  // string is reasonable.
+  bool success = document_url == expected_url;
+  if (meta_tag_content->size() > chrome::kMaxMetaTagAttributeLength) {
+    *meta_tag_content = "";
+    success = false;
+  }
+
+  return success;
+}
+#endif
+
 }  // namespace
 
 ChromeRenderViewObserver::ChromeRenderViewObserver(
@@ -233,6 +283,8 @@ bool ChromeRenderViewObserver::OnMessageReceived(const IPC::Message& message) {
                         OnUpdateTopControlsState)
     IPC_MESSAGE_HANDLER(ChromeViewMsg_RetrieveWebappInformation,
                         OnRetrieveWebappInformation)
+    IPC_MESSAGE_HANDLER(ChromeViewMsg_RetrieveMetaTagContent,
+                        OnRetrieveMetaTagContent)
 #endif
     IPC_MESSAGE_HANDLER(ChromeViewMsg_SetWindowFeatures, OnSetWindowFeatures)
     IPC_MESSAGE_UNHANDLED(handled = false)
@@ -273,42 +325,28 @@ void ChromeRenderViewObserver::OnUpdateTopControlsState(
 void ChromeRenderViewObserver::OnRetrieveWebappInformation(
     const GURL& expected_url) {
   WebFrame* main_frame = render_view()->GetWebView()->mainFrame();
-  WebDocument document =
-      main_frame ? main_frame->document() : WebDocument();
-
-  WebElement head = document.isNull() ? WebElement() : document.head();
-  GURL document_url = document.isNull() ? GURL() : GURL(document.url());
-
-  // Make sure we're checking the right page.
-  bool success = document_url == expected_url;
-
-  bool is_mobile_webapp_capable = false;
-  bool is_apple_mobile_webapp_capable = false;
-
-  // Search the DOM for the webapp <meta> tags.
-  if (!head.isNull()) {
-    WebNodeList children = head.childNodes();
-    for (unsigned i = 0; i < children.length(); ++i) {
-      WebNode child = children.item(i);
-      if (!child.isElementNode())
-        continue;
-      WebElement elem = child.to<WebElement>();
-
-      if (elem.hasTagName("meta") && elem.hasAttribute("name")) {
-        std::string name = elem.getAttribute("name").utf8();
-        WebString content = elem.getAttribute("content");
-        if (LowerCaseEqualsASCII(content, "yes")) {
-          if (name == "mobile-web-app-capable") {
-            is_mobile_webapp_capable = true;
-          } else if (name == "apple-mobile-web-app-capable") {
-            is_apple_mobile_webapp_capable = true;
-          }
-        }
-      }
-    }
-  } else {
-    success = false;
-  }
+  bool found_tag;
+  std::string content_str;
+
+  // Search for the "mobile-web-app-capable" tag.
+  bool mobile_parse_success = RetrieveMetaTagContent(
+      main_frame,
+      expected_url,
+      "mobile-web-app-capable",
+      &found_tag,
+      &content_str);
+  bool is_mobile_webapp_capable = mobile_parse_success && found_tag &&
+      LowerCaseEqualsASCII(content_str, "yes");
+
+  // Search for the "apple-mobile-web-app-capable" tag.
+  bool apple_parse_success = RetrieveMetaTagContent(
+      main_frame,
+      expected_url,
+      "apple-mobile-web-app-capable",
+      &found_tag,
+      &content_str);
+  bool is_apple_mobile_webapp_capable = apple_parse_success && found_tag &&
+      LowerCaseEqualsASCII(content_str, "yes");
 
   bool is_only_apple_mobile_webapp_capable =
       is_apple_mobile_webapp_capable && !is_mobile_webapp_capable;
@@ -324,11 +362,31 @@ void ChromeRenderViewObserver::OnRetrieveWebappInformation(
 
   Send(new ChromeViewHostMsg_DidRetrieveWebappInformation(
       routing_id(),
-      success,
+      mobile_parse_success && apple_parse_success,
       is_mobile_webapp_capable,
       is_apple_mobile_webapp_capable,
       expected_url));
 }
+
+void ChromeRenderViewObserver::OnRetrieveMetaTagContent(
+    const GURL& expected_url,
+    const std::string tag_name) {
+  bool found_tag;
+  std::string content_str;
+  bool parsed_successfully = RetrieveMetaTagContent(
+      render_view()->GetWebView()->mainFrame(),
+      expected_url,
+      tag_name,
+      &found_tag,
+      &content_str);
+
+  Send(new ChromeViewHostMsg_DidRetrieveMetaTagContent(
+      routing_id(),
+      parsed_successfully && found_tag,
+      tag_name,
+      content_str,
+      expected_url));
+}
 #endif
 
 void ChromeRenderViewObserver::OnSetWindowFeatures(
diff --git a/chrome/renderer/chrome_render_view_observer.h b/chrome/renderer/chrome_render_view_observer.h
index 786079e..a2e7f68 100644
--- a/chrome/renderer/chrome_render_view_observer.h
+++ b/chrome/renderer/chrome_render_view_observer.h
@@ -82,6 +82,8 @@ class ChromeRenderViewObserver : public content::RenderViewObserver {
                                 content::TopControlsState current,
                                 bool animate);
   void OnRetrieveWebappInformation(const GURL& expected_url);
+  void OnRetrieveMetaTagContent(const GURL& expected_url,
+                                const std::string tag_name);
 #endif
   void OnSetWindowFeatures(const blink::WebWindowFeatures& window_features);