From 80dc4ea49a14c90c21d6b45c08d8681c06d4db37 Mon Sep 17 00:00:00 2001
From: Mathew Sutcliffe <oktal3700@gmail.com>
Date: Thu, 1 Sep 2016 21:06:18 +0100
Subject: [PATCH] refs #710 Sample to compare performance of string utilities
 versus regular expressions.

---
 samples/blackmisc/main.cpp               |  2 +
 samples/blackmisc/samplesperformance.cpp | 83 ++++++++++++++++++++++++
 samples/blackmisc/samplesperformance.h   |  3 +
 3 files changed, 88 insertions(+)

diff --git a/samples/blackmisc/main.cpp b/samples/blackmisc/main.cpp
index dc85c2382..664393763 100644
--- a/samples/blackmisc/main.cpp
+++ b/samples/blackmisc/main.cpp
@@ -49,6 +49,7 @@ int main(int argc, char *argv[])
         qtout << "6c .. 25/20 Performance impl. type" << endl;
         qtout << "6d .. 40/20 Interpolator scenario" << endl;
         qtout << "6e .. JSON performance" << endl;
+        qtout << "6f .. string utils vs. regex" << endl;
         qtout << "7 .. Algorithms" << endl;
         qtout << "-----" << endl;
         qtout << "x .. Bye" << endl;
@@ -63,6 +64,7 @@ int main(int argc, char *argv[])
         else if (s.startsWith("6c")) { CSamplesPerformance::samplesImplementationType(qtout, 25, 20); }
         else if (s.startsWith("6d")) { CSamplesPerformance::interpolatorScenario(qtout, 40, 20); }
         else if (s.startsWith("6e")) { CSamplesPerformance::samplesJson(qtout); }
+        else if (s.startsWith("6f")) { CSamplesPerformance::samplesString(qtout); }
         else if (s.startsWith("7")) { CSamplesAlgorithm::samples(); }
         else if (s.startsWith("x")) { break; }
     }
diff --git a/samples/blackmisc/samplesperformance.cpp b/samples/blackmisc/samplesperformance.cpp
index c5a65c51d..47436a92d 100644
--- a/samples/blackmisc/samplesperformance.cpp
+++ b/samples/blackmisc/samplesperformance.cpp
@@ -19,6 +19,7 @@
 #include "blackmisc/aviation/callsign.h"
 #include "blackmisc/geo/coordinategeodetic.h"
 #include "blackmisc/pq/units.h"
+#include "blackmisc/stringutils.h"
 #include "blackmisc/testing.h"
 
 #include <QDateTime>
@@ -36,6 +37,7 @@
 #include <algorithm>
 #include <iterator>
 
+using namespace BlackMisc;
 using namespace BlackMisc::Aviation;
 using namespace BlackMisc::Geo;
 using namespace BlackMisc::PhysicalQuantities;
@@ -340,6 +342,87 @@ namespace BlackSample
         return 0;
     }
 
+    int CSamplesPerformance::samplesString(QTextStream &out)
+    {
+        QTime timer;
+        static const QString chars = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789~_-=+!\"@#$%^&*()[]{} \t;:\\/?,.<>";
+        QStringList strings;
+        std::generate_n(std::back_inserter(strings), 100000, []
+        {
+            QString s;
+            std::generate_n(std::back_inserter(s), 10, [] { return chars[qrand() % chars.size()]; });
+            return s;
+        });
+        QString bigString = strings.join("\n");
+
+        QRegularExpression upperRegex("[A-Z]");
+        upperRegex.optimize();
+
+        timer.start();
+        for (const QString &s : strings)
+        {
+            auto c = containsChar(s, [](QChar c) { return c.isUpper(); });
+            Q_UNUSED(c);
+        }
+        out << "Check 100,000 strings for containing uppercase letter: (utility)     " << timer.elapsed() << "ms" << endl;
+
+        timer.start();
+        for (const QString &s : strings)
+        {
+            auto c = s.contains(upperRegex);
+            Q_UNUSED(c);
+        }
+        out << "Check 100,000 strings for containing uppercase letter: (regex)       " << timer.elapsed() << "ms" << endl << endl;
+
+        timer.start();
+        for (const QString &s : strings)
+        {
+            auto i = indexOfChar(s, [](QChar c) { return c.isUpper(); });
+            Q_UNUSED(i);
+        }
+        out << "Check 100,000 strings for index of first uppercase letter: (utility) " << timer.elapsed() << "ms" << endl;
+
+        timer.start();
+        for (const QString &s : strings)
+        {
+            auto i = s.indexOf(upperRegex);
+            Q_UNUSED(i);
+        }
+        out << "Check 100,000 strings for index of first uppercase letter: (regex)   " << timer.elapsed() << "ms" << endl << endl;
+
+        auto temp = strings;
+        timer.start();
+        for (QString &s : strings)
+        {
+            removeChars(s, [](QChar c) { return c.isUpper(); });
+        }
+        out << "Remove from 100,000 strings all uppercase letters: (utility)         " << timer.elapsed() << "ms" << endl;
+        strings = temp;
+
+        timer.start();
+        for (QString &s : strings)
+        {
+            s.remove(upperRegex);
+        }
+        out << "Remove from 100,000 strings all uppercase letters: (regex)           " << timer.elapsed() << "ms" << endl << endl;
+
+        timer.start();
+        {
+            auto lines = splitLines(bigString);
+            Q_UNUSED(lines);
+        }
+        out << "Split 100,000 line string into list of lines: (QStringList)          " << timer.elapsed() << "ms" << endl;
+
+        timer.start();
+        {
+            auto lines = splitLinesRefs(bigString);
+            Q_UNUSED(lines);
+        }
+        out << "Split 100,000 line string into list of lines: (QList<QStringRef>)    " << timer.elapsed() << "ms" << endl;
+
+        return 0;
+    }
+
     CAircraftSituationList CSamplesPerformance::createSituations(qint64 baseTimeEpoch, int numberOfCallsigns, int numberOfTimes)
     {
         CAircraftSituationList situations;
diff --git a/samples/blackmisc/samplesperformance.h b/samples/blackmisc/samplesperformance.h
index f1d5e5a2c..d8960e2de 100644
--- a/samples/blackmisc/samplesperformance.h
+++ b/samples/blackmisc/samplesperformance.h
@@ -38,6 +38,9 @@ namespace BlackSample
         //! JSON (de)serialization
         static int samplesJson(QTextStream &out);
 
+        //! String manipulation
+        static int samplesString(QTextStream &out);
+
         //! Situation values for testing
         static BlackMisc::Aviation::CAircraftSituationList createSituations(qint64 baseTimeEpoch, int numberOfCallsigns, int numberOfTimes);