chrome/browser/extensions/activity_log/counting_policy.h

   1 // Copyright 2013 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #ifndef CHROME_BROWSER_EXTENSIONS_ACTIVITY_LOG_COUNTING_POLICY_H_
   6 #define CHROME_BROWSER_EXTENSIONS_ACTIVITY_LOG_COUNTING_POLICY_H_
   7
   8 #include <string>
   9
  10 #include "base/containers/hash_tables.h"
  11 #include "chrome/browser/extensions/activity_log/activity_database.h"
  12 #include "chrome/browser/extensions/activity_log/activity_log_policy.h"
  13 #include "chrome/browser/extensions/activity_log/database_string_table.h"
  14
  15 namespace extensions {
  16
  17 // A policy for logging the stream of actions, but with most arguments stripped
  18 // out (to improve privacy and reduce database size) and with multiple
  19 // identical rows combined together using a count column to track the total
  20 // number of repetitions.  Identical rows within the same day are merged, but
  21 // actions on separate days are kept distinct.  Data is kept for up to a few
  22 // days then deleted.
  23 class CountingPolicy : public ActivityLogDatabasePolicy {
  24  public:
  25   explicit CountingPolicy(Profile* profile);
  26   ~CountingPolicy() override;
  27
  28   void ProcessAction(scoped_refptr<Action> action) override;
  29
  30   void ReadFilteredData(
  31       const std::string& extension_id,
  32       const Action::ActionType type,
  33       const std::string& api_name,
  34       const std::string& page_url,
  35       const std::string& arg_url,
  36       const int days_ago,
  37       const base::Callback<void(scoped_ptr<Action::ActionVector>)>& callback)
  38       override;
  39
  40   void Close() override;
  41
  42   // Gets or sets the amount of time that old records are kept in the database.
  43   const base::TimeDelta& retention_time() const { return retention_time_; }
  44   void set_retention_time(const base::TimeDelta& delta) {
  45     retention_time_ = delta;
  46   }
  47
  48   // Remove actions (rows) which IDs are specified in the action_ids array.
  49   void RemoveActions(const std::vector<int64>& action_ids) override;
  50
  51   // Clean the URL data stored for this policy.
  52   void RemoveURLs(const std::vector<GURL>&) override;
  53
  54   // Clean the data related to this extension for this policy.
  55   void RemoveExtensionData(const std::string& extension_id) override;
  56
  57   // Delete everything in the database.
  58   void DeleteDatabase() override;
  59
  60   // The main database table, and the name for a read-only view that
  61   // decompresses string values for easier parsing.
  62   static const char kTableName[];
  63   static const char kReadViewName[];
  64
  65  protected:
  66   // The ActivityDatabase::Delegate interface.  These are always called from
  67   // the database thread.
  68   bool InitDatabase(sql::Connection* db) override;
  69   bool FlushDatabase(sql::Connection* db) override;
  70   void OnDatabaseFailure() override;
  71   void OnDatabaseClose() override;
  72
  73  private:
  74   // A type used to track pending writes to the database.  The key is an action
  75   // to write; the value is the amount by which the count field should be
  76   // incremented in the database.
  77   typedef std::map<
  78       scoped_refptr<Action>,
  79       int,
  80       ActionComparatorExcludingTimeAndActionId>
  81       ActionQueue;
  82
  83   // Adds an Action to those to be written out; this is an internal method used
  84   // by ProcessAction and is called on the database thread.
  85   void QueueAction(scoped_refptr<Action> action);
  86
  87   // Internal method to read data from the database; called on the database
  88   // thread.
  89   scoped_ptr<Action::ActionVector> DoReadFilteredData(
  90       const std::string& extension_id,
  91       const Action::ActionType type,
  92       const std::string& api_name,
  93       const std::string& page_url,
  94       const std::string& arg_url,
  95       const int days_ago);
  96
  97   // The implementation of RemoveActions; this must only run on the database
  98   // thread.
  99   void DoRemoveActions(const std::vector<int64>& action_ids);
 100
 101   // The implementation of RemoveURLs; this must only run on the database
 102   // thread.
 103   void DoRemoveURLs(const std::vector<GURL>& restrict_urls);
 104
 105   // The implementation of RemoveExtensionData; this must only run on the
 106   // database thread.
 107   void DoRemoveExtensionData(const std::string& extension_id);
 108
 109   // The implementation of DeleteDatabase; called on the database thread.
 110   void DoDeleteDatabase();
 111
 112   // Cleans old records from the activity log database.
 113   bool CleanOlderThan(sql::Connection* db, const base::Time& cutoff);
 114
 115   // Cleans unused interned strings from the database.  This should be run
 116   // after deleting rows from the main log table to clean out stale values.
 117   bool CleanStringTables(sql::Connection* db);
 118
 119   // API calls for which complete arguments should be logged.
 120   Util::ApiSet api_arg_whitelist_;
 121
 122   // Tables for mapping strings to integers for shrinking database storage
 123   // requirements.  URLs are kept in a separate table from other strings to
 124   // make history clearing simpler.
 125   DatabaseStringTable string_table_;
 126   DatabaseStringTable url_table_;
 127
 128   // Tracks any pending updates to be written to the database, if write
 129   // batching is turned on.  Should only be accessed from the database thread.
 130   ActionQueue queued_actions_;
 131
 132   // All queued actions must fall on the same day, so that we do not
 133   // accidentally aggregate actions that should be kept separate.
 134   // queued_actions_date_ is the date (timestamp at local midnight) of all the
 135   // actions in queued_actions_.
 136   base::Time queued_actions_date_;
 137
 138   // The amount of time old activity log records should be kept in the
 139   // database.  This time is subtracted from the current time, rounded down to
 140   // midnight, and rows older than this are deleted from the database when
 141   // cleaning runs.
 142   base::TimeDelta retention_time_;
 143
 144   // The time at which old activity log records were last cleaned out of the
 145   // database (only tracked for this browser session).  Old records are deleted
 146   // on the first database flush, and then every 12 hours subsequently.
 147   base::Time last_database_cleaning_time_;
 148
 149   friend class CountingPolicyTest;
 150   FRIEND_TEST_ALL_PREFIXES(CountingPolicyTest, EarlyFlush);
 151   FRIEND_TEST_ALL_PREFIXES(CountingPolicyTest, MergingAndExpiring);
 152   FRIEND_TEST_ALL_PREFIXES(CountingPolicyTest, StringTableCleaning);
 153 };
 154
 155 }  // namespace extensions
 156
 157 #endif  // CHROME_BROWSER_EXTENSIONS_ACTIVITY_LOG_COUNTING_POLICY_H_