Add a cache for recently viewed webpages.

chrome/android/java/src/org/chromium/chrome/browser/AppIndexingUtil.java

Index: chrome/android/java/src/org/chromium/chrome/browser/AppIndexingUtil.java
diff --git a/chrome/android/java/src/org/chromium/chrome/browser/AppIndexingUtil.java b/chrome/android/java/src/org/chromium/chrome/browser/AppIndexingUtil.java
index 4e4aca058b446417c865412c358189ba45edb456..ef7ff2010ac20f4249b38311cf05e5997cd30e6b 100644
--- a/chrome/android/java/src/org/chromium/chrome/browser/AppIndexingUtil.java
+++ b/chrome/android/java/src/org/chromium/chrome/browser/AppIndexingUtil.java
@@ -4,9 +4,12 @@
 
 package org.chromium.chrome.browser;
 
+import android.os.SystemClock;
+import android.util.LruCache;
 import android.webkit.URLUtil;
 
 import org.chromium.base.SysUtils;
+import org.chromium.base.VisibleForTesting;
 import org.chromium.blink.mojom.document_metadata.CopylessPaste;
 import org.chromium.blink.mojom.document_metadata.WebPage;
 import org.chromium.chrome.browser.historyreport.AppIndexingReporter;
@@ -19,31 +22,118 @@ import org.chromium.services.service_manager.InterfaceProvider;
  * This is the top-level CopylessPaste metadata extraction for AppIndexing.
  */
 public class AppIndexingUtil {
-    public static void extractCopylessPasteMetadata(final Tab tab) {
-        String url = tab.getUrl();
+    private static final int CACHE_SIZE = 100;
+    private static final int CACHE_VISIT_CUTOFF_MS = 60 * 60 * 1000; // 1 hour
+    // Cache of recently seen urls. If a url is among the CACHE_SIZE most recent pages visited, and
+    // the visit was in the last CACHE_VISIT_CUTOFF_MS milliseconds, then we don't parse the page,

[wychen, 2017/04/13 07:05:00]

and the *parsing* was in the last ...

Our expiration is based on parsing, not visiting, right?

[dproctor, 2017/04/13 17:23:09]

Thanks, yeah, the comparison is made based on the last time the page was parsed,
but cache eviction is based on the last visit time.

+    // and instead just report the view (not the content) to App Indexing.
+    private final LruCache<String, CacheEntry> mPageCache;
+
+    public AppIndexingUtil() {
+        this(CACHE_SIZE);
+    }
+
+    private AppIndexingUtil(int cacheSize) {
+        mPageCache = new LruCache<String, CacheEntry>(cacheSize);
+    }
+
+    public void extractCopylessPasteMetadata(final Tab tab) {
+        final String url = tab.getUrl();
         boolean isHttpOrHttps = URLUtil.isHttpsUrl(url) || URLUtil.isHttpUrl(url);
-        if (SysUtils.isLowEndDevice() || tab.isIncognito()
-                || !ChromeFeatureList.isEnabled(ChromeFeatureList.COPYLESS_PASTE)
-                || !isHttpOrHttps) {
+        if (!isEnabledForDevice() || tab.isIncognito() || !isHttpOrHttps) {
             return;
         }
 
+        // There are three conditions that can occur with respect to the cache.

[wychen, 2017/04/13 07:05:00]

We might want to add UMA to track the distribution of these conditions.

[dproctor, 2017/04/13 17:34:58]

How about we add this in a follow-up cl?

[wychen, 2017/04/13 18:04:45]

No problem.

+        // 1. Cache hit, and an entity was found previously. Report only the page view to App
+        //    Indexing.
+        // 2. Cache hit, but no entity was found. Ignore.
+        // 3. Cache miss, we need to parse the page.
+        if (wasPageVisitedRecently(url)) {
+            if (lastPageVisitContainedEntity(url)) {
+                // Condition 1
+                getAppIndexingReporter().reportWebPageView(url, tab.getTitle());
+            }
+            // Condition 2
+        } else {
+            // Condition 3
+            CopylessPaste copylessPaste = getCopylessPasteInterface(tab);
+            if (copylessPaste == null) {
+                return;
+            }
+            copylessPaste.getEntities(new CopylessPaste.GetEntitiesResponse() {
+                @Override
+                public void call(WebPage webpage) {
+                    putCacheEntry(url, webpage != null);
+                    if (webpage == null) return;
+                    getAppIndexingReporter().reportWebPage(webpage);
+                }
+            });
+        }
+    }
+
+    private boolean wasPageVisitedRecently(String url) {
+        if (url == null) {
+            return false;
+        }
+        CacheEntry entry = mPageCache.get(url);
+        if (entry == null || getElapsedTime() - entry.lastSeenTimeMs > CACHE_VISIT_CUTOFF_MS) {
+            return false;
+        }
+        return true;
+    }
+
+    // Returns true if the page is in the cache and it contained an entity the last time it was
+    // visited.

[wychen, 2017/04/13 07:05:00]

it was *parsed*

[dproctor, 2017/04/13 17:23:09]

Done.

+    private boolean lastPageVisitContainedEntity(String url) {
+        if (url == null) {
+            return false;
+        }
+        CacheEntry entry = mPageCache.get(url);
+        if (entry == null || !entry.containedEntity) {
+            return false;
+        }
+        return true;
+    }
+
+    private void putCacheEntry(String url, boolean containedEntity) {
+        CacheEntry entry = new CacheEntry();
+        entry.lastSeenTimeMs = getElapsedTime();
+        entry.containedEntity = containedEntity;
+        mPageCache.put(url, entry);
+    }
+
+    @VisibleForTesting
+    AppIndexingReporter getAppIndexingReporter() {
+        return AppIndexingReporter.getInstance();
+    }
+
+    @VisibleForTesting
+    CopylessPaste getCopylessPasteInterface(Tab tab) {
         WebContents webContents = tab.getWebContents();
-        if (webContents == null) return;
+        if (webContents == null) return null;
 
         RenderFrameHost mainFrame = webContents.getMainFrame();
-        if (mainFrame == null) return;
+        if (mainFrame == null) return null;
 
         InterfaceProvider interfaces = mainFrame.getRemoteInterfaces();
-        if (interfaces == null) return;
-
-        CopylessPaste copylesspaste = interfaces.getInterface(CopylessPaste.MANAGER);
-        copylesspaste.getEntities(new CopylessPaste.GetEntitiesResponse() {
-            @Override
-            public void call(WebPage webpage) {
-                if (webpage == null) return;
-                AppIndexingReporter.getInstance().reportWebPage(webpage);
-            }
-        });
+        if (interfaces == null) return null;
+
+        return interfaces.getInterface(CopylessPaste.MANAGER);
+    }
+
+    @VisibleForTesting
+    long getElapsedTime() {
+        return SystemClock.elapsedRealtime();
+    }
+
+    boolean isEnabledForDevice() {
+        return !SysUtils.isLowEndDevice()
+                && ChromeFeatureList.isEnabled(ChromeFeatureList.COPYLESS_PASTE);
+    }
+
+    private static class CacheEntry {
+        public long lastSeenTimeMs;
+        public boolean containedEntity;
     }
 }