predictors: Add origin learning.

chrome/browser/predictors/resource_prefetch_predictor_tables.h

Index: chrome/browser/predictors/resource_prefetch_predictor_tables.h
diff --git a/chrome/browser/predictors/resource_prefetch_predictor_tables.h b/chrome/browser/predictors/resource_prefetch_predictor_tables.h
index ff9c5ea220f9be08f63e4a8f7477c862d644b032..bf151da4bdfc78886c0891dbe9417f995240f927 100644
--- a/chrome/browser/predictors/resource_prefetch_predictor_tables.h
+++ b/chrome/browser/predictors/resource_prefetch_predictor_tables.h
@@ -9,6 +9,7 @@
 #include <map>
 #include <memory>
 #include <string>
+#include <unordered_map>
 #include <vector>
 
 #include "base/gtest_prod_util.h"
@@ -26,6 +27,7 @@ namespace predictors {
 
 // From resource_prefetch_predictor.proto.
 using RedirectStat = RedirectData_RedirectStat;
+using OriginStat = OriginData_OriginStat;
 
 // Interface for database tables used by the ResourcePrefetchPredictor.
 // All methods except the constructor and destructor need to be called on the DB
@@ -38,23 +40,23 @@ using RedirectStat = RedirectData_RedirectStat;
 //  - HostRedirectTable - redirects per host.
 class ResourcePrefetchPredictorTables : public PredictorTableBase {
  public:
-  typedef std::map<std::string, PrefetchData> PrefetchDataMap;
-  typedef std::map<std::string, RedirectData> RedirectDataMap;
-  typedef std::map<std::string, precache::PrecacheManifest> ManifestDataMap;
+  typedef std::unordered_map<std::string, PrefetchData> PrefetchDataMap;

[alexilin, 2017/04/10 14:58:28]

Wow, this is a big change! Maybe for another CL?
At least, it's worth to mention this in a CL description.

[Benoit L, 2017/04/11 09:04:05]

I really dislike map<>, but unordered_map<> is not really better here,
unfortunately. Wanted to compare the two. Removed the change.

+  typedef std::unordered_map<std::string, RedirectData> RedirectDataMap;
+  typedef std::unordered_map<std::string, precache::PrecacheManifest>
+      ManifestDataMap;
+  typedef std::unordered_map<std::string, OriginData> OriginDataMap;
 
   // Returns data for all Urls and Hosts.
   virtual void GetAllData(PrefetchDataMap* url_data_map,
                           PrefetchDataMap* host_data_map,
                           RedirectDataMap* url_redirect_data_map,
                           RedirectDataMap* host_redirect_data_map,
-                          ManifestDataMap* manifest_map);
+                          ManifestDataMap* manifest_map,
+                          OriginDataMap* origin_data_map);
 
-  // Updates data for a Url and a host. If either of the |url_data| or
-  // |host_data| or |url_redirect_data| or |host_redirect_data| has an empty
+  // Updates data for a Url and a host. If any of the arguments has an empty
   // primary key, it will be ignored.
-  // Note that the Urls and primary key in |url_data|, |host_data|,
-  // |url_redirect_data| and |host_redirect_data| should be less than
-  // |kMaxStringLength| in length.
+  // Note that all the keys should be less |kMaxStringLength| in length.
   virtual void UpdateData(const PrefetchData& url_data,
                           const PrefetchData& host_data,
                           const RedirectData& url_redirect_data,
@@ -65,6 +67,10 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
       const std::string& host,
       const precache::PrecacheManifest& manifest_data);
 
+  // Updates the origin data for a given |host|.
+  virtual void UpdateOriginData(const std::string& host,
+                                const OriginData& origin_data);
+
   // Delete data for the input |urls| and |hosts|.
   virtual void DeleteResourceData(const std::vector<std::string>& urls,
                                   const std::vector<std::string>& hosts);
@@ -84,6 +90,9 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
   // Delete data for the input |hosts|.
   virtual void DeleteManifestData(const std::vector<std::string>& hosts);
 
+  // Deletes the origin data for a list of |hosts|.
+  virtual void DeleteOriginData(const std::vector<std::string>& hosts);
+
   // Deletes all data in all the tables.
   virtual void DeleteAllData();
 
@@ -101,6 +110,16 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
   // misses from |data|.
   static void TrimRedirects(RedirectData* data, size_t max_consecutive_misses);
 
+  // Removes the origins with more than |max_consecutive_misses| consecutive
+  // misses from |data|.
+  static void TrimOrigins(OriginData* data, size_t max_consecutive_misses);
+
+  // Sorts the origins by score, decreasing.
+  static void SortOrigins(OriginData* data);
+
+  // Computes score of |data|.
+  static float ComputeOriginScore(const OriginStat& origin);
+
   // The maximum length of the string that can be stored in the DB.
   static constexpr size_t kMaxStringLength = 1024;
 
@@ -112,7 +131,7 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
 
  private:
   // Represents the type of information that is stored in prefetch database.
-  enum class PrefetchDataType { RESOURCE, REDIRECT, MANIFEST };
+  enum class PrefetchDataType { RESOURCE, REDIRECT, MANIFEST, ORIGIN };
 
   enum class TableOperationType { INSERT, REMOVE };
 
@@ -124,7 +143,7 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
 
   // Database version. Always increment it when any change is made to the data
   // schema (including the .proto).
-  static constexpr int kDatabaseVersion = 6;
+  static constexpr int kDatabaseVersion = 7;
 
   // Helper functions below help perform functions on the Url and host table
   // using the same code.
@@ -133,6 +152,7 @@ class ResourcePrefetchPredictorTables : public PredictorTableBase {
   void GetAllRedirectDataHelper(PrefetchKeyType key_type,
                                 RedirectDataMap* redirect_map);
   void GetAllManifestDataHelper(ManifestDataMap* manifest_map);
+  void GetAllOriginDataHelper(OriginDataMap* manifest_map);
 
   bool UpdateDataHelper(PrefetchKeyType key_type,
                         PrefetchDataType data_type,