New class QgsDateTimeStatisticalSummary

Calculates summary statistics on lists of datetime values, such as count, count distinct, min, max and range
qgis · May 11, 2016 · b053ad8 · b053ad8
1 parent 0d165e5
commit b053ad8
Show file tree

Hide file tree

Showing 7 changed files with 528 additions and 0 deletions.
diff --git a/python/core/core.sip b/python/core/core.sip
@@ -39,6 +39,7 @@
 %Include qgsdataitemproviderregistry.sip
 %Include qgsdataprovider.sip
 %Include qgsdatasourceuri.sip
+%Include qgsdatetimestatisticalsummary.sip
 %Include qgsdatumtransformstore.sip
 %Include qgsdbfilterproxymodel.sip
 %Include qgsdistancearea.sip

diff --git a/python/core/qgsdatetimestatisticalsummary.sip b/python/core/qgsdatetimestatisticalsummary.sip
@@ -0,0 +1,104 @@
+/** \ingroup core
+ * \class QgsDateTimeStatisticalSummary
+ * \brief Calculator for summary statistics and aggregates for a list of datetimes.
+ *
+ * Statistics are calculated by calling @link calculate @endlink and passing a list of datetimes. The
+ * individual statistics can then be retrieved using the associated methods. Note that not all statistics
+ * are calculated by default. Statistics which require slower computations are only calculated by
+ * specifying the statistic in the constructor or via @link setStatistics @endlink.
+ *
+ * \note Added in version 2.16
+ */
+
+class QgsDateTimeStatisticalSummary
+{
+%TypeHeaderCode
+#include <qgsdatetimestatisticalsummary.h>
+%End
+
+  public:
+
+    //! Enumeration of flags that specify statistics to be calculated
+    enum Statistic
+    {
+      Count,  //!< Count
+      CountDistinct,  //!< Number of distinct datetime values
+      CountMissing,  //!< Number of missing (null) values
+      Min, //!< Minimum (earliest) datetime value
+      Max, //!< Maximum (latest) datetime value
+      Range, //!< Interval between earliest and latest datetime value
+      All, //! All statistics
+    };
+    typedef QFlags<QgsDateTimeStatisticalSummary::Statistic> Statistics;
+
+    /** Constructor for QgsDateTimeStatisticalSummary
+     * @param stats flags for statistics to calculate
+     */
+    QgsDateTimeStatisticalSummary( const QgsDateTimeStatisticalSummary::Statistics& stats = All );
+
+    /** Returns flags which specify which statistics will be calculated. Some statistics
+     * are always calculated (eg count).
+     * @see setStatistics
+     */
+    Statistics statistics() const;
+
+    /** Sets flags which specify which statistics will be calculated. Some statistics
+     * are always calculated (eg count).
+     * @param stats flags for statistics to calculate
+     * @see statistics
+     */
+    void setStatistics( const Statistics& stats );
+
+    /** Resets the calculated values
+     */
+    void reset();
+
+    /** Calculates summary statistics for a list of variants. Any non-string variants will be
+     * ignored.
+     * @param values list of variants
+     */
+    void calculate( const QVariantList& values );
+
+    /** Returns the value of a specified statistic
+     * @param stat statistic to return
+     * @returns calculated value of statistic
+     */
+    QVariant statistic( Statistic stat ) const;
+
+    /** Returns the calculated count of values.
+     */
+    int count() const;
+
+    /** Returns the number of distinct datetime values.
+     */
+    int countDistinct() const;
+
+    /** Returns the set of distinct datetime values.
+     */
+    QSet< QDateTime > distinctValues() const;
+
+    /** Returns the number of missing (null) datetime values.
+     */
+    int countMissing() const;
+
+    /** Returns the minimum (earliest) non-null datetime value.
+     */
+    QDateTime min() const;
+
+    /** Returns the maximum (latest) non-null datetime value.
+     */
+    QDateTime max() const;
+
+    /** Returns the range (interval between earliest and latest non-null datetime values).
+     */
+    QgsInterval range() const;
+
+    /** Returns the friendly display name for a statistic
+     * @param statistic statistic to return name for
+     */
+    static QString displayName( Statistic statistic );
+
+};
+
+QFlags<QgsDateTimeStatisticalSummary::Statistic> operator|(QgsDateTimeStatisticalSummary::Statistic f1, QFlags<QgsDateTimeStatisticalSummary::Statistic> f2);
+
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
@@ -100,6 +100,7 @@ SET(QGIS_CORE_SRCS
   qgsdataitemprovider.cpp
   qgsdataitemproviderregistry.cpp
   qgsdatasourceuri.cpp
+  qgsdatetimestatisticalsummary.cpp
   qgsdatumtransformstore.cpp
   qgsdbfilterproxymodel.cpp
   qgsdiagramrendererv2.cpp
@@ -617,6 +618,7 @@ SET(QGIS_CORE_HDRS
   qgsdataitemprovider.h
   qgsdataitemproviderregistry.h
   qgsdatasourceuri.h
+  qgsdatetimestatisticalsummary.h
   qgsdatumtransformstore.h
   qgsdbfilterproxymodel.h
   qgsdiagramrendererv2.h

diff --git a/src/core/qgsdatetimestatisticalsummary.cpp b/src/core/qgsdatetimestatisticalsummary.cpp
@@ -0,0 +1,144 @@
+/***************************************************************************
+  qgsdatetimestatisticalsummary.cpp
+  ---------------------------------
+  Date                 : May 2016
+  Copyright            : (C) 2016 by Nyall Dawson
+  Email                : nyall dot dawson at gmail dot com
+ ***************************************************************************
+ *                                                                         *
+ *   This program is free software; you can redistribute it and/or modify  *
+ *   it under the terms of the GNU General Public License as published by  *
+ *   the Free Software Foundation; either version 2 of the License, or     *
+ *   (at your option) any later version.                                   *
+ *                                                                         *
+ ***************************************************************************/
+
+#include "qgsdatetimestatisticalsummary.h"
+#include <QString>
+#include <QDateTime>
+#include <QStringList>
+#include <QObject>
+#include <QVariant>
+#include <QVariantList>
+#include "limits.h"
+
+/***************************************************************************
+ * This class is considered CRITICAL and any change MUST be accompanied with
+ * full unit tests in test_qgsdatetimestatisticalsummary.py.
+ * See details in QEP #17
+ ****************************************************************************/
+
+QgsDateTimeStatisticalSummary::QgsDateTimeStatisticalSummary( const QgsDateTimeStatisticalSummary::Statistics& stats )
+    : mStatistics( stats )
+{
+  reset();
+}
+
+void QgsDateTimeStatisticalSummary::reset()
+{
+  mCount = 0;
+  mValues.clear();
+  mCountMissing = 0;
+  mMin = QDateTime();
+  mMax = QDateTime();
+}
+
+void QgsDateTimeStatisticalSummary::calculate( const QVariantList& values )
+{
+  reset();
+
+  Q_FOREACH ( const QVariant& variant, values )
+  {
+    if ( variant.type() == QVariant::DateTime )
+    {
+      testDateTime( variant.toDateTime() );
+    }
+    else if ( variant.type() == QVariant::Date )
+    {
+      QDate date = variant.toDate();
+      testDateTime( date.isValid() ? QDateTime( date, QTime( 0, 0, 0 ) )
+                    : QDateTime() );
+    }
+    // QTime?
+  }
+}
+
+void QgsDateTimeStatisticalSummary::testDateTime( const QDateTime& dateTime )
+{
+  mCount++;
+
+  if ( !dateTime.isValid() )
+    mCountMissing++;
+
+  if ( mStatistics & CountDistinct )
+  {
+    mValues << dateTime;
+  }
+  if ( mStatistics & Min || mStatistics & Range )
+  {
+    if ( mMin.isValid() && dateTime.isValid() )
+    {
+      mMin = qMin( mMin, dateTime );
+    }
+    else if ( !mMin.isValid() && dateTime.isValid() )
+    {
+      mMin = dateTime;
+    }
+  }
+  if ( mStatistics & Max || mStatistics & Range )
+  {
+    if ( mMax.isValid() && dateTime.isValid() )
+    {
+      mMax = qMax( mMax, dateTime );
+    }
+    else if ( !mMax.isValid() && dateTime.isValid() )
+    {
+      mMax = dateTime;
+    }
+  }
+}
+
+QVariant QgsDateTimeStatisticalSummary::statistic( QgsDateTimeStatisticalSummary::Statistic stat ) const
+{
+  switch ( stat )
+  {
+    case Count:
+      return mCount;
+    case CountDistinct:
+      return mValues.count();
+    case CountMissing:
+      return mCountMissing;
+    case Min:
+      return mMin;
+    case Max:
+      return mMax;
+    case Range:
+      return QVariant::fromValue( mMax - mMin );
+    case All:
+      return 0;
+  }
+  return 0;
+}
+
+QString QgsDateTimeStatisticalSummary::displayName( QgsDateTimeStatisticalSummary::Statistic statistic )
+{
+  switch ( statistic )
+  {
+    case Count:
+      return QObject::tr( "Count" );
+    case CountDistinct:
+      return QObject::tr( "Count (distinct)" );
+    case CountMissing:
+      return QObject::tr( "Count (missing)" );
+    case Min:
+      return QObject::tr( "Minimum (earliest)" );
+    case Max:
+      return QObject::tr( "Maximum (latest)" );
+    case Range:
+      return QObject::tr( "Range (interval)" );
+    case All:
+      return QString();
+  }
+  return QString();
+}
+