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   virtual ~CountingPolicy();
  27
  28   virtual void ProcessAction(scoped_refptr<Action> action) OVERRIDE;
  29
  30   virtual 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
  38           <void(scoped_ptr<Action::ActionVector>)>& callback) OVERRIDE;
  39
  40   virtual 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   // Clean the URL data stored for this policy.
  49   virtual void RemoveURLs(const std::vector<GURL>&) OVERRIDE;
  50
  51   // Clean the data related to this extension for this policy.
  52   virtual void RemoveExtensionData(const std::string& extension_id) OVERRIDE;
  53
  54   // Delete everything in the database.
  55   virtual void DeleteDatabase() OVERRIDE;
  56
  57   // The main database table, and the name for a read-only view that
  58   // decompresses string values for easier parsing.
  59   static const char* kTableName;
  60   static const char* kReadViewName;
  61
  62  protected:
  63   // The ActivityDatabase::Delegate interface.  These are always called from
  64   // the database thread.
  65   virtual bool InitDatabase(sql::Connection* db) OVERRIDE;
  66   virtual bool FlushDatabase(sql::Connection* db) OVERRIDE;
  67   virtual void OnDatabaseFailure() OVERRIDE;
  68   virtual void OnDatabaseClose() OVERRIDE;
  69
  70  private:
  71   // A type used to track pending writes to the database.  The key is an action
  72   // to write; the value is the amount by which the count field should be
  73   // incremented in the database.
  74   typedef std::map<scoped_refptr<Action>, int, ActionComparatorExcludingTime>
  75       ActionQueue;
  76
  77   // Adds an Action to those to be written out; this is an internal method used
  78   // by ProcessAction and is called on the database thread.
  79   void QueueAction(scoped_refptr<Action> action);
  80
  81   // Internal method to read data from the database; called on the database
  82   // thread.
  83   scoped_ptr<Action::ActionVector> DoReadFilteredData(
  84       const std::string& extension_id,
  85       const Action::ActionType type,
  86       const std::string& api_name,
  87       const std::string& page_url,
  88       const std::string& arg_url,
  89       const int days_ago);
  90
  91   // The implementation of RemoveURLs; this must only run on the database
  92   // thread.
  93   void DoRemoveURLs(const std::vector<GURL>& restrict_urls);
  94
  95   // The implementation of RemoveExtensionData; this must only run on the
  96   // database thread.
  97   void DoRemoveExtensionData(const std::string& extension_id);
  98
  99   // The implementation of DeleteDatabase; called on the database thread.
 100   void DoDeleteDatabase();
 101
 102   // Cleans old records from the activity log database.
 103   bool CleanOlderThan(sql::Connection* db, const base::Time& cutoff);
 104
 105   // Cleans unused interned strings from the database.  This should be run
 106   // after deleting rows from the main log table to clean out stale values.
 107   bool CleanStringTables(sql::Connection* db);
 108
 109   // API calls for which complete arguments should be logged.
 110   Util::ApiSet api_arg_whitelist_;
 111
 112   // Tables for mapping strings to integers for shrinking database storage
 113   // requirements.  URLs are kept in a separate table from other strings to
 114   // make history clearing simpler.
 115   DatabaseStringTable string_table_;
 116   DatabaseStringTable url_table_;
 117
 118   // Tracks any pending updates to be written to the database, if write
 119   // batching is turned on.  Should only be accessed from the database thread.
 120   ActionQueue queued_actions_;
 121
 122   // All queued actions must fall on the same day, so that we do not
 123   // accidentally aggregate actions that should be kept separate.
 124   // queued_actions_date_ is the date (timestamp at local midnight) of all the
 125   // actions in queued_actions_.
 126   base::Time queued_actions_date_;
 127
 128   // The amount of time old activity log records should be kept in the
 129   // database.  This time is subtracted from the current time, rounded down to
 130   // midnight, and rows older than this are deleted from the database when
 131   // cleaning runs.
 132   base::TimeDelta retention_time_;
 133
 134   // The time at which old activity log records were last cleaned out of the
 135   // database (only tracked for this browser session).  Old records are deleted
 136   // on the first database flush, and then every 12 hours subsequently.
 137   base::Time last_database_cleaning_time_;
 138
 139   friend class CountingPolicyTest;
 140   FRIEND_TEST_ALL_PREFIXES(CountingPolicyTest, EarlyFlush);
 141   FRIEND_TEST_ALL_PREFIXES(CountingPolicyTest, MergingAndExpiring);
 142   FRIEND_TEST_ALL_PREFIXES(CountingPolicyTest, StringTableCleaning);
 143 };
 144
 145 }  // namespace extensions
 146
 147 #endif  // CHROME_BROWSER_EXTENSIONS_ACTIVITY_LOG_COUNTING_POLICY_H_