[ML] Refactor initialization of the bucket gatherer

.clang-tidy

@@ -23,6 +23,7 @@ Checks: >
   -modernize-use-trailing-return-type,
   -modernize-concat-nested-namespaces,
   -modernize-use-nodiscard,
+  -modernize-use-ranges,
 
   performance-*,
 

include/model/CBucketGatherer.h

@@ -132,6 +132,26 @@ class MODEL_EXPORT CBucketGatherer {
     using TTimeVec = std::vector<core_t::TTime>;
     using TTimeVecCItr = TTimeVec::const_iterator;
 
+    struct SBucketGathererInitData {
+        // The name of the field holding the summary count.
+        const std::string& s_SummaryCountFieldName;
+        // The name of the field which identifies people.
+        const std::string& s_PersonFieldName;
+        // The name of the field which defines the person attributes.
+        const std::string& s_AttributeFieldName;
+        // The name of the field which contains the metric values.
+        const std::string& s_ValueFieldName;
+        // The field names for which we will compute influences.
+        const TStrVec& s_InfluenceFieldNames;
+        // The start of the time interval for which to gather data.
+        core_t::TTime s_StartTime;
+        // Override for the number of measurements
+        // in a statistic. (Note that this is intended for testing only.)
+        // A zero value means that the data gatherer class will determine
+        // an appropriate value for the bucket length and data rate.
+        unsigned int s_SampleOverrideCount;
+    };
+
 public:
     static const std::string EVENTRATE_BUCKET_GATHERER_TAG;
     static const std::string METRIC_BUCKET_GATHERER_TAG;
@@ -142,11 +162,10 @@ class MODEL_EXPORT CBucketGatherer {
     //! Create a new data series gatherer.
     //!
     //! \param[in] dataGatherer The owning data gatherer.
-    //! \param[in] startTime The start of the time interval for which
-    //! to gather data.
-    //! \param[in] numberInfluencers The number of result influencers
-    //! for which to gather data.
-    CBucketGatherer(CDataGatherer& dataGatherer, core_t::TTime startTime, std::size_t numberInfluencers);
+    //! \param[in] bucketGathererInitData The parameter initialization object
+    //! for the bucket gatherer.
+    CBucketGatherer(CDataGatherer& dataGatherer,
+                    const SBucketGathererInitData& bucketGathererInitData);
 
     //! Create a copy that will result in the same persisted state as the
     //! original.  This is effectively a copy constructor that creates a

include/model/CDataGatherer.h

@@ -146,51 +146,26 @@ class MODEL_EXPORT CDataGatherer {
     //! \param[in] summaryMode Indicates whether the data being gathered
     //! are already summarized by an external aggregation process.
     //! \param[in] modelParams The global configuration parameters.
-    //! \param[in] summaryCountFieldName If \p summaryMode is E_Manual
-    //! then this is the name of the field holding the summary count.
     //! \param[in] partitionFieldValue The value of the field which splits
     //! the data.
-    //! \param[in] personFieldName The name of the field which identifies
-    //! people.
-    //! \param[in] attributeFieldName The name of the field which defines
-    //! the person attributes.
-    //! \param[in] valueFieldName The name of the field which contains
-    //! the metric values.
-    //! \param[in] influenceFieldNames The field names for which we will
-    //! compute influences.
     //! \param[in] key The key of the search for which to gatherer data.
     //! \param[in] features The features of the data to model.
-    //! \param[in] startTime The start of the time interval for which
-    //! to gather data.
-    //! \param[in] sampleCountOverride for the number of measurements
-    //! in a statistic. (Note that this is intended for testing only.)
-    //! A zero value means that the data gatherer class will determine
-    //! an appropriate value for the bucket length and data rate.
+    //! \param[in] bucketGathererInitData The parameter initialization object for the bucket gatherer.
     CDataGatherer(model_t::EAnalysisCategory gathererType,
                   model_t::ESummaryMode summaryMode,
                   const SModelParams& modelParams,
-                  const std::string& summaryCountFieldName,
                   const std::string& partitionFieldValue,
-                  const std::string& personFieldName,
-                  const std::string& attributeFieldName,
-                  const std::string& valueFieldName,
-                  const TStrVec& influenceFieldNames,
                   const CSearchKey& key,
                   const TFeatureVec& features,
-                  core_t::TTime startTime,
-                  int sampleCountOverride);
+                  const CBucketGatherer::SBucketGathererInitData& bucketGathererInitData);
 
     //! Construct from a state document.
     CDataGatherer(model_t::EAnalysisCategory gathererType,
                   model_t::ESummaryMode summaryMode,
                   const SModelParams& modelParams,
-                  const std::string& summaryCountFieldName,
                   const std::string& partitionFieldValue,
-                  const std::string& personFieldName,
-                  const std::string& attributeFieldName,
-                  const std::string& valueFieldName,
-                  const TStrVec& influenceFieldNames,
                   const CSearchKey& key,
+                  const CBucketGatherer::SBucketGathererInitData& bucketGathererInitData,
                   core::CStateRestoreTraverser& traverser);
 
     //! Create a copy that will result in the same persisted state as the
@@ -330,6 +305,7 @@ class MODEL_EXPORT CDataGatherer {
     //! containing \p time.
     //!
     //! \param[in] time The time of interest.
+    //! \param[in] bucketLength The length of the bucketing interval.
     //! \param[out] result Filled in with the feature data at \p time.
     //! \tparam T The type of the feature data.
     template<typename T>
@@ -675,19 +651,11 @@ class MODEL_EXPORT CDataGatherer {
 
 private:
     //! Restore state from supplied traverser.
-    bool acceptRestoreTraverser(const std::string& summaryCountFieldName,
-                                const std::string& personFieldName,
-                                const std::string& attributeFieldName,
-                                const std::string& valueFieldName,
-                                const TStrVec& influenceFieldNames,
+    bool acceptRestoreTraverser(const CBucketGatherer::SBucketGathererInitData& bucketGathererInitData,
                                 core::CStateRestoreTraverser& traverser);
 
     //! Restore a bucket gatherer from the supplied traverser.
-    bool restoreBucketGatherer(const std::string& summaryCountFieldName,
-                               const std::string& personFieldName,
-                               const std::string& attributeFieldName,
-                               const std::string& valueFieldName,
-                               const TStrVec& influenceFieldNames,
+    bool restoreBucketGatherer(const CBucketGatherer::SBucketGathererInitData& bucketGathererInitData,
                                core::CStateRestoreTraverser& traverser);
 
     //! Persist a bucket gatherer by passing information to the supplied
@@ -696,13 +664,7 @@ class MODEL_EXPORT CDataGatherer {
 
     //! Create the bucket specific data gatherer.
     void createBucketGatherer(model_t::EAnalysisCategory gathererType,
-                              const std::string& summaryCountFieldName,
-                              const std::string& personFieldName,
-                              const std::string& attributeFieldName,
-                              const std::string& valueFieldName,
-                              const TStrVec& influenceFieldNames,
-                              core_t::TTime startTime,
-                              unsigned int sampleCountOverride);
+                              const CBucketGatherer::SBucketGathererInitData& initData);
 
 private:
     //! The type of the bucket gatherer(s) used.

include/model/CEventRateBucketGatherer.h

@@ -111,33 +111,13 @@ class MODEL_EXPORT CEventRateBucketGatherer final : public CBucketGatherer {
     //! Create an event rate bucket gatherer.
     //!
     //! \param[in] dataGatherer The owning data gatherer.
-    //! \param[in] summaryCountFieldName If summaryMode is E_Manual
-    //! then this is the name of the field holding the summary count.
-    //! \param[in] personFieldName The name of the field which identifies
-    //! people.
-    //! \param[in] attributeFieldName The name of the field which defines
-    //! the person attributes.
-    //! \param[in] valueFieldName The name of the field which contains
-    //! the metric values.
-    //! \param[in] influenceFieldNames The field names for which we will
-    //! compute influences.
-    //! \param[in] startTime The start of the time interval for which
-    //! to gather data.
+    //! \param[in] bucketGathererInitData The parameter initialization object.
     CEventRateBucketGatherer(CDataGatherer& dataGatherer,
-                             const std::string& summaryCountFieldName,
-                             const std::string& personFieldName,
-                             const std::string& attributeFieldName,
-                             const std::string& valueFieldName,
-                             const TStrVec& influenceFieldNames,
-                             core_t::TTime startTime);
+                             const SBucketGathererInitData& bucketGathererInitData);
 
     //! Construct from a state document.
     CEventRateBucketGatherer(CDataGatherer& dataGatherer,
-                             const std::string& summaryCountFieldName,
-                             const std::string& personFieldName,
-                             const std::string& attributeFieldName,
-                             const std::string& valueFieldName,
-                             const TStrVec& influenceFieldNames,
+                             const SBucketGathererInitData& bucketGathererInitData,
                              core::CStateRestoreTraverser& traverser);
 
     //! Create a copy that will result in the same persisted state as the
@@ -444,11 +424,7 @@ class MODEL_EXPORT CEventRateBucketGatherer final : public CBucketGatherer {
     void startNewBucket(core_t::TTime time, bool skipUpdates) override;
 
     //! Initialize the field names collection.
-    void initializeFieldNames(const std::string& personFieldName,
-                              const std::string& attributeFieldName,
-                              const std::string& valueFieldName,
-                              const std::string& summaryCountFieldName,
-                              const TStrVec& influenceFieldNames);
+    void initializeFieldNames(const CBucketGatherer::SBucketGathererInitData& initData);
 
     //! Initialize the feature data gatherers.
     void initializeFeatureData();

include/model/CMetricBucketGatherer.h

@@ -53,33 +53,13 @@ class MODEL_EXPORT CMetricBucketGatherer final : public CBucketGatherer {
     //! Create a new population metric data gatherer.
     //!
     //! \param[in] dataGatherer The owning data gatherer.
-    //! \param[in] summaryCountFieldName If \p summaryMode is E_Manual
-    //! then this is the name of the field holding the summary count.
-    //! \param[in] personFieldName The name of the field which identifies
-    //! people.
-    //! \param[in] attributeFieldName The name of the field which defines
-    //! the person attributes.
-    //! \param[in] valueFieldName The name of the field which contains
-    //! the metric values.
-    //! \param[in] influenceFieldNames The field names for which we will
-    //! compute influences.
-    //! \param[in] startTime The start of the time interval for which
-    //! to gather data.
-    CMetricBucketGatherer(CDataGatherer& dataGatherer,
-                          const std::string& summaryCountFieldName,
-                          const std::string& personFieldName,
-                          const std::string& attributeFieldName,
-                          const std::string& valueFieldName,
-                          const TStrVec& influenceFieldNames,
-                          core_t::TTime startTime);
+    //! \param[in] initData The parameter initialization object for the bucket
+    //! gatherer.
+    CMetricBucketGatherer(CDataGatherer& dataGatherer, const SBucketGathererInitData& initData);
 
     //! Construct from a state document.
     CMetricBucketGatherer(CDataGatherer& dataGatherer,
-                          const std::string& summaryCountFieldName,
-                          const std::string& personFieldName,
-                          const std::string& attributeFieldName,
-                          const std::string& valueFieldName,
-                          const TStrVec& influenceFieldNames,
+                          const SBucketGathererInitData& initData,
                           core::CStateRestoreTraverser& traverser);
 
     //! Create a copy that will result in the same persisted state as the
@@ -266,9 +246,7 @@ class MODEL_EXPORT CMetricBucketGatherer final : public CBucketGatherer {
     //! 1) initializeFieldNamesPart1()
     //! 2) restore state
     //! 3) initializeFieldNamesPart2()
-    void initializeFieldNamesPart1(const std::string& personFieldName,
-                                   const std::string& attributeFieldName,
-                                   const TStrVec& influenceFieldNames);
+    void initializeFieldNamesPart1(const SBucketGathererInitData& initData);
 
     //! Initialize the field names collection.
     //! initializeFieldNamesPart1() must be called before this.
@@ -277,8 +255,7 @@ class MODEL_EXPORT CMetricBucketGatherer final : public CBucketGatherer {
     //! 1) initializeFieldNamesPart1()
     //! 2) restore state
     //! 3) initializeFieldNamesPart2()
-    void initializeFieldNamesPart2(const std::string& valueFieldName,
-                                   const std::string& summaryCountFieldName);
+    void initializeFieldNamesPart2(const SBucketGathererInitData& initData);
 
     //! Initialize the feature data gatherers.
     void initializeFeatureData();

include/model/CModelFactory.h

@@ -17,14 +17,14 @@
 #include <maths/common/COrderings.h>
 #include <maths/common/MathsTypes.h>
 
+#include <model/CResourceMonitor.h>
 #include <model/ImportExport.h>
 #include <model/ModelTypes.h>
 #include <model/SModelParams.h>
 
 #include <map>
 #include <memory>
 #include <optional>
-#include <utility>
 #include <vector>
 
 namespace ml {
@@ -104,6 +104,8 @@ class MODEL_EXPORT CModelFactory {
     using TStrDetectionRulePr = std::pair<std::string, model::CDetectionRule>;
     using TStrDetectionRulePrVec = std::vector<TStrDetectionRulePr>;
     using TStrDetectionRulePrVecCRef = std::reference_wrapper<const TStrDetectionRulePrVec>;
+    using TResourceMonitorCRef = std::reference_wrapper<const CResourceMonitor>;
+    using TOptionalResourceMonitorCRef = std::optional<TResourceMonitorCRef>;
 
 public:
     //! Wrapper around the model initialization data.

lib/model/CBucketGatherer.cc

@@ -202,21 +202,22 @@ const std::string CBucketGatherer::EVENTRATE_BUCKET_GATHERER_TAG("a");
 const std::string CBucketGatherer::METRIC_BUCKET_GATHERER_TAG("b");
 
 CBucketGatherer::CBucketGatherer(CDataGatherer& dataGatherer,
-                                 core_t::TTime startTime,
-                                 std::size_t numberInfluencers)
-    : m_DataGatherer(dataGatherer), m_EarliestTime(startTime), m_BucketStart(startTime),
+                                 const SBucketGathererInitData& initData)
+    : m_DataGatherer(dataGatherer), m_EarliestTime(initData.s_StartTime),
+      m_BucketStart(initData.s_StartTime),
       m_PersonAttributeCounts(dataGatherer.params().s_LatencyBuckets,
                               dataGatherer.params().s_BucketLength,
-                              startTime,
+                              initData.s_StartTime,
                               TSizeSizePrUInt64UMap(1)),
       m_PersonAttributeExplicitNulls(dataGatherer.params().s_LatencyBuckets,
                                      dataGatherer.params().s_BucketLength,
-                                     startTime,
+                                     initData.s_StartTime,
                                      TSizeSizePrUSet(1)),
       m_InfluencerCounts(dataGatherer.params().s_LatencyBuckets + 3,
                          dataGatherer.params().s_BucketLength,
-                         startTime,
-                         TSizeSizePrOptionalStrPrUInt64UMapVec(numberInfluencers)) {
+                         initData.s_StartTime,
+                         TSizeSizePrOptionalStrPrUInt64UMapVec(
+                             initData.s_InfluenceFieldNames.size())) {
 }
 
 CBucketGatherer::CBucketGatherer(bool isForPersistence, const CBucketGatherer& other)

lib/model/CCountingModelFactory.cc

@@ -61,19 +61,27 @@ CCountingModelFactory::makeModel(const SModelInitializationData& initData,
 
 CDataGatherer*
 CCountingModelFactory::makeDataGatherer(const SGathererInitializationData& initData) const {
-    return new CDataGatherer(model_t::E_EventRate, m_SummaryMode, this->modelParams(),
-                             m_SummaryCountFieldName, initData.s_PartitionFieldValue,
-                             m_PersonFieldName, EMPTY_STRING, EMPTY_STRING, {},
-                             this->searchKey(), m_Features, initData.s_StartTime, 0);
+    const CBucketGatherer::SBucketGathererInitData bucketGathererInitData{
+        m_SummaryCountFieldName,
+        m_PersonFieldName,
+        EMPTY_STRING,
+        EMPTY_STRING,
+        {},
+        initData.s_StartTime,
+        0};
+    return new CDataGatherer(model_t::E_EventRate, m_SummaryMode,
+                             this->modelParams(), initData.s_PartitionFieldValue,
+                             this->searchKey(), m_Features, bucketGathererInitData);
 }
 
 CDataGatherer*
 CCountingModelFactory::makeDataGatherer(const std::string& partitionFieldValue,
                                         core::CStateRestoreTraverser& traverser) const {
+    CBucketGatherer::SBucketGathererInitData bucketGathererInitData{
+        m_SummaryCountFieldName, m_PersonFieldName, EMPTY_STRING, EMPTY_STRING, {}, 0, 0};
     return new CDataGatherer(model_t::E_EventRate, m_SummaryMode,
-                             this->modelParams(), m_SummaryCountFieldName,
-                             partitionFieldValue, m_PersonFieldName, EMPTY_STRING,
-                             EMPTY_STRING, {}, this->searchKey(), traverser);
+                             this->modelParams(), partitionFieldValue,
+                             this->searchKey(), bucketGathererInitData, traverser);
 }
 
 CCountingModelFactory::TPriorPtr