components/browsing_topics/browsing_topics_calculator.h

   1 // Copyright 2022 The Chromium Authors
   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 COMPONENTS_BROWSING_TOPICS_BROWSING_TOPICS_CALCULATOR_H_
   6 #define COMPONENTS_BROWSING_TOPICS_BROWSING_TOPICS_CALCULATOR_H_
   7
   8 #include <map>
   9 #include <set>
  10
  11 #include "base/containers/queue.h"
  12 #include "base/functional/callback.h"
  13 #include "base/memory/raw_ptr.h"
  14 #include "base/task/cancelable_task_tracker.h"
  15 #include "base/time/time.h"
  16 #include "components/browsing_topics/annotator.h"
  17 #include "components/browsing_topics/common/common_types.h"
  18 #include "components/browsing_topics/epoch_topics.h"
  19 #include "components/history/core/browser/history_types.h"
  20
  21 namespace privacy_sandbox {
  22 class PrivacySandboxSettings;
  23 }  // namespace privacy_sandbox
  24
  25 namespace history {
  26 class HistoryService;
  27 }  // namespace history
  28
  29 namespace content {
  30 class BrowsingTopicsSiteDataManager;
  31 }  // namespace content
  32
  33 namespace browsing_topics {
  34
  35 // Responsible for doing a one-off browsing topics calculation. It will:
  36 // 1) Check the user settings for calculation permissions.
  37 // 2) Query the `BrowsingTopicsSiteDataManager` for the contexts where the
  38 // Topics API was called on.
  39 // 3) Query the `HistoryService` for the hosts of the pages the API was called
  40 // on.
  41 // 4) Query the `Annotator` with a set of hosts, to get the corresponding
  42 // topics.
  43 // 5) Derive `EpochTopics` (i.e. the top topics and the their observed-by
  44 // contexts), and return it as the final result.
  45 class BrowsingTopicsCalculator {
  46  public:
  47   // These values are persisted to logs. Entries should not be renumbered and
  48   // numeric values should never be reused.
  49   enum class CalculatorResultStatus {
  50     kSuccess = 0,
  51     kFailurePermissionDenied = 1,
  52     kFailureApiUsageContextQueryError = 2,
  53     kFailureAnnotationExecutionError = 3,
  54     kFailureTaxonomyVersionNotSupportedInBinary = 4,
  55
  56     kMaxValue = kFailureTaxonomyVersionNotSupportedInBinary,
  57   };
  58
  59   using CalculateCompletedCallback = base::OnceCallback<void(EpochTopics)>;
  60
  61   BrowsingTopicsCalculator(
  62       privacy_sandbox::PrivacySandboxSettings* privacy_sandbox_settings,
  63       history::HistoryService* history_service,
  64       content::BrowsingTopicsSiteDataManager* site_data_manager,
  65       Annotator* annotator,
  66       const base::circular_deque<EpochTopics>& epochs,
  67       bool is_manually_triggered,
  68       CalculateCompletedCallback callback);
  69
  70   BrowsingTopicsCalculator(const BrowsingTopicsCalculator&) = delete;
  71   BrowsingTopicsCalculator& operator=(const BrowsingTopicsCalculator&) = delete;
  72   BrowsingTopicsCalculator(BrowsingTopicsCalculator&&) = delete;
  73   BrowsingTopicsCalculator& operator=(BrowsingTopicsCalculator&&) = delete;
  74
  75   virtual ~BrowsingTopicsCalculator();
  76
  77   bool is_manually_triggered() const { return is_manually_triggered_; }
  78
  79  protected:
  80   // This method exists for the purposes of overriding in tests.
  81   virtual uint64_t GenerateRandUint64();
  82   virtual void CheckCanCalculate();
  83
  84  private:
  85   // Get the top `kBrowsingTopicsNumberOfTopTopicsPerEpoch` topics. If there
  86   // aren't enough topics, pad with random ones. Return the result topics, and
  87   // the starting index of the padded topics (or
  88   // `kBrowsingTopicsNumberOfTopTopicsPerEpoch` if there's no padded topics),
  89   // and the number of topics associated with `history_hosts_count`.
  90   void DeriveTopTopics(
  91       const std::map<HashedHost, size_t>& history_hosts_count,
  92       const std::map<HashedHost, std::set<Topic>>& host_topics_map,
  93       std::vector<Topic>& top_topics,
  94       size_t& padded_top_topics_start_index,
  95       size_t& history_topics_count);
  96
  97   void OnGetRecentBrowsingTopicsApiUsagesCompleted(
  98       browsing_topics::ApiUsageContextQueryResult result);
  99
 100   void OnGetRecentlyVisitedURLsCompleted(history::QueryResults results);
 101
 102   void OnRequestModelCompleted(std::vector<std::string> raw_hosts);
 103
 104   void OnGetTopicsForHostsCompleted(const std::vector<Annotation>& results);
 105
 106   void OnCalculateCompleted(CalculatorResultStatus status,
 107                             EpochTopics epoch_topics);
 108
 109   // Those pointers are safe to hold and use throughout the lifetime of
 110   // `BrowsingTopicsService`, which owns this object.
 111   raw_ptr<privacy_sandbox::PrivacySandboxSettings> privacy_sandbox_settings_;
 112   raw_ptr<history::HistoryService> history_service_;
 113   raw_ptr<content::BrowsingTopicsSiteDataManager> site_data_manager_;
 114   raw_ptr<Annotator> annotator_;
 115
 116   CalculateCompletedCallback calculate_completed_callback_;
 117
 118   // The calculation start time.
 119   base::Time calculation_time_;
 120
 121   base::Time history_data_start_time_;
 122   base::Time api_usage_context_data_start_time_;
 123
 124   // The history hosts over
 125   // `kBrowsingTopicsNumberOfEpochsOfObservationDataToUseForFiltering` epochs,
 126   // and the calling context domains that used the Topics API in each main frame
 127   // host.
 128   std::map<HashedHost, std::vector<HashedDomain>> host_context_domains_map_;
 129
 130   // The hashed history hosts and their count over the last epoch.
 131   std::map<HashedHost, size_t> history_hosts_count_;
 132
 133   // Used for the async tasks querying the HistoryService.
 134   base::CancelableTaskTracker history_task_tracker_;
 135
 136   // Whether this calculator was generated via the topics-internals page rather
 137   // than via a scheduled task.
 138   bool is_manually_triggered_;
 139
 140   base::WeakPtrFactory<BrowsingTopicsCalculator> weak_ptr_factory_{this};
 141 };
 142
 143 }  // namespace browsing_topics
 144
 145 #endif  // COMPONENTS_BROWSING_TOPICS_BROWSING_TOPICS_CALCULATOR_H_