Adding hosts based prefetching.

Currently the ResourcePrefetchPredictor learns about resource required on a per URL basis. This change enables it to learn the same on a per Host basis.

NOTE: Tests will be fixed once the changes looks ok.

BUG=149743


Review URL: https://chromiumcodereview.appspot.com/11369080

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@171325 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 100 insertions, 33 deletions

diff --git a/chrome/browser/predictors/resource_prefetch_predictor_tables.h b/chrome/browser/predictors/resource_prefetch_predictor_tables.h
index 7d6f8e0..9e7e18a 100644
--- a/chrome/browser/predictors/resource_prefetch_predictor_tables.h
+++ b/chrome/browser/predictors/resource_prefetch_predictor_tables.h
@@ -5,40 +5,54 @@
 #ifndef CHROME_BROWSER_PREDICTORS_RESOURCE_PREFETCH_PREDICTOR_TABLES_H_
 #define CHROME_BROWSER_PREDICTORS_RESOURCE_PREFETCH_PREDICTOR_TABLES_H_
 
-#include "chrome/browser/predictors/predictor_table_base.h"
 
+#include <map>
 #include <string>
 #include <vector>
 
 #include "base/time.h"
+#include "chrome/browser/predictors/predictor_table_base.h"
+#include "chrome/browser/predictors/resource_prefetch_common.h"
 #include "googleurl/src/gurl.h"
 #include "webkit/glue/resource_type.h"
 
+namespace sql {
+class Statement;
+}
+
 namespace predictors {
 
 // Interface for database tables used by the ResourcePrefetchPredictor.
 // All methods except the constructor and destructor need to be called on the DB
 // thread.
 //
-// Currently manages two tables:
-//  - UrlResourceTable - Keeps track of resources per Urls.
-//  - UrlMetadataTable - Keeps misc data for Urls (like last visit time).
+// Currently manages:
+//  - UrlResourceTable - resources per Urls.
+//  - UrlMetadataTable - misc data for Urls (like last visit time).
+//  - HostResourceTable - resources per host.
+//  - HostMetadataTable - misc data for hosts.
 class ResourcePrefetchPredictorTables : public PredictorTableBase {
  public:
-  struct UrlResourceRow {
-    UrlResourceRow();
-    UrlResourceRow(const UrlResourceRow& other);
-    UrlResourceRow(const std::string& main_frame_url,
-                   const std::string& resource_url,
-                   ResourceType::Type resource_type,
-                   int number_of_hits,
-                   int number_of_misses,
-                   int consecutive_misses,
-                   double average_position);
+  // Used in the UrlResourceTable and HostResourceTable to store resources
+  // required for the page or host.
+  struct ResourceRow {
+    ResourceRow();
+    ResourceRow(const ResourceRow& other);
+    ResourceRow(const std::string& main_frame_url,
+                const std::string& resource_url,
+                ResourceType::Type resource_type,
+                int number_of_hits,
+                int number_of_misses,
+                int consecutive_misses,
+                double average_position);
     void UpdateScore();
-    bool operator==(const UrlResourceRow& rhs) const;
+    bool operator==(const ResourceRow& rhs) const;
+
+    // Stores the host for host based data, main frame Url for the Url based
+    // data. This field is cleared for efficiency reasons and the code outside
+    // this class should not assume it is set.
+    std::string primary_key;
 
-    GURL main_frame_url;
     GURL resource_url;
     ResourceType::Type resource_type;
     int number_of_hits;
@@ -49,30 +63,58 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
     // Not stored.
     float score;
   };
-  typedef std::vector<UrlResourceRow> UrlResourceRows;
+  typedef std::vector<ResourceRow> ResourceRows;
 
-  // Sorts the UrlResourceRows by score, descending.
-  struct UrlResourceRowSorter {
-    bool operator()(const UrlResourceRow& x, const UrlResourceRow& y) const;
+  // Sorts the ResourceRows by score, descending.
+  struct ResourceRowSorter {
+    bool operator()(const ResourceRow& x, const ResourceRow& y) const;
   };
 
-  // Data from both the tables for a particular main frame url.
-  struct UrlData {
-    explicit UrlData(const GURL& main_frame_url);
-    UrlData(const UrlData& other);
-    ~UrlData();
-    bool operator==(const UrlData& rhs) const;
+  // Aggregated data for a Url or Host. Although the data differs slightly, we
+  // store them in the same structure, because most of the fields are common and
+  // it allows us to use the same functions.
+  struct PrefetchData {
+    PrefetchData(PrefetchKeyType key_type, const std::string& primary_key);
+    PrefetchData(const PrefetchData& other);
+    ~PrefetchData();
+    bool operator==(const PrefetchData& rhs) const;
+
+    bool is_host() const { return key_type == PREFETCH_KEY_TYPE_HOST; }
+
+    // Is the data a host as opposed to a Url?
+    PrefetchKeyType key_type;  // Not const to be able to assign.
+    std::string primary_key;  // is_host() ? main frame url : host.
 
-    GURL main_frame_url;
     base::Time last_visit;
-    UrlResourceRows resources;
+    ResourceRows resources;
   };
+  // Map from primary key to PrefetchData for the key.
+  typedef std::map<std::string, PrefetchData> PrefetchDataMap;
+
+  // Returns data for all Urls and Hosts.
+  virtual void GetAllData(PrefetchDataMap* url_data_map,
+                          PrefetchDataMap* host_data_map);
 
-  // |url_data_buffer| should be empty for GetAllRows.
-  virtual void GetAllUrlData(std::vector<UrlData>* url_data_buffer);
-  virtual void UpdateDataForUrl(const UrlData& url_data);
-  virtual void DeleteDataForUrls(const std::vector<GURL>& urls);
-  virtual void DeleteAllUrlData();
+  // Updates data for a Url and a host. If either of the |url_data| or
+  // |host_data| has an empty primary key, it will be ignored.
+  // Note that the Urls and primary key in |url_data| and |host_data| should be
+  // less than |kMaxStringLength| in length.
+  virtual void UpdateData(const PrefetchData& url_data,
+                          const PrefetchData& host_data);
+
+  // Delete data for the input |urls| and |hosts|.
+  virtual void DeleteData(const std::vector<std::string>& urls,
+                  const std::vector<std::string>& hosts);
+
+  // Wrapper over DeleteData for convenience.
+  virtual void DeleteSingleDataPoint(const std::string& key,
+                                     PrefetchKeyType key_type);
+
+  // Deletes all data in all the tables.
+  virtual void DeleteAllData();
+
+  // The maximum length of the string that can be stored in the DB.
+  static const size_t kMaxStringLength;
 
  private:
   friend class PredictorDatabaseInternal;
@@ -81,10 +123,35 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
   ResourcePrefetchPredictorTables();
   virtual ~ResourcePrefetchPredictorTables();
 
+  // Helper functions below help perform functions on the Url and host table
+  // using the same code.
+  void GetAllDataHelper(PrefetchKeyType key_type,
+                        PrefetchDataMap* data_map,
+                        std::vector<std::string>* to_delete);
+  bool UpdateDataHelper(const PrefetchData& data);
+  void DeleteDataHelper(PrefetchKeyType key_type,
+                        const std::vector<std::string>& keys);
+
+  // Returns true if the strings in the |data| are less than |kMaxStringLength|
+  // in length.
+  bool StringsAreSmallerThanDBLimit(const PrefetchData& data) const;
+
   // PredictorTableBase methods.
   virtual void CreateTableIfNonExistent() OVERRIDE;
   virtual void LogDatabaseStats() OVERRIDE;
 
+  // Helpers to return Statements for cached Statements. The caller must take
+  // ownership of the return Statements.
+  sql::Statement* GetUrlResourceDeleteStatement();
+  sql::Statement* GetUrlResourceUpdateStatement();
+  sql::Statement* GetUrlMetadataDeleteStatement();
+  sql::Statement* GetUrlMetadataUpdateStatement();
+
+  sql::Statement* GetHostResourceDeleteStatement();
+  sql::Statement* GetHostResourceUpdateStatement();
+  sql::Statement* GetHostMetadataDeleteStatement();
+  sql::Statement* GetHostMetadataUpdateStatement();
+
   DISALLOW_COPY_AND_ASSIGN(ResourcePrefetchPredictorTables);
 };