Add simple wxRegEx benchmark

Allow testing speed of some simple regex operations to compare the speed
of the currently used regex library with some other alternatives.

tests/benchmarks/Makefile.in

@@ -57,6 +57,7 @@ BENCH_OBJECTS =  \
 	bench_ipcclient.o \
 	bench_log.o \
 	bench_mbconv.o \
+	bench_regex.o \
 	bench_strings.o \
 	bench_tls.o \
 	bench_printfbench.o
@@ -299,6 +300,9 @@ bench_log.o: $(srcdir)/log.cpp
 bench_mbconv.o: $(srcdir)/mbconv.cpp
 	$(CXXC) -c -o $@ $(BENCH_CXXFLAGS) $(srcdir)/mbconv.cpp
 
+bench_regex.o: $(srcdir)/regex.cpp
+	$(CXXC) -c -o $@ $(BENCH_CXXFLAGS) $(srcdir)/regex.cpp
+
 bench_strings.o: $(srcdir)/strings.cpp
 	$(CXXC) -c -o $@ $(BENCH_CXXFLAGS) $(srcdir)/strings.cpp
 

tests/benchmarks/bench.bkl

@@ -16,6 +16,7 @@
             ipcclient.cpp
             log.cpp
             mbconv.cpp
+            regex.cpp
             strings.cpp
             tls.cpp
             printfbench.cpp

tests/benchmarks/bench_vc8_bench.vcproj

@@ -838,6 +838,10 @@
 				RelativePath=".\printfbench.cpp"
 				>
 			</File>
+			<File
+				RelativePath=".\regex.cpp"
+				>
+			</File>
 			<File
 				RelativePath=".\strings.cpp"
 				>

tests/benchmarks/bench_vc9_bench.vcproj

@@ -810,6 +810,10 @@
 				RelativePath=".\printfbench.cpp"
 				>
 			</File>
+			<File
+				RelativePath=".\regex.cpp"
+				>
+			</File>
 			<File
 				RelativePath=".\strings.cpp"
 				>

tests/benchmarks/makefile.gcc

@@ -36,6 +36,7 @@ BENCH_OBJECTS =  \
 	$(OBJS)\bench_ipcclient.o \
 	$(OBJS)\bench_log.o \
 	$(OBJS)\bench_mbconv.o \
+	$(OBJS)\bench_regex.o \
 	$(OBJS)\bench_strings.o \
 	$(OBJS)\bench_tls.o \
 	$(OBJS)\bench_printfbench.o
@@ -310,6 +311,9 @@ $(OBJS)\bench_log.o: ./log.cpp
 $(OBJS)\bench_mbconv.o: ./mbconv.cpp
 	$(CXX) -c -o $@ $(BENCH_CXXFLAGS) $(CPPDEPS) $<
 
+$(OBJS)\bench_regex.o: ./regex.cpp
+	$(CXX) -c -o $@ $(BENCH_CXXFLAGS) $(CPPDEPS) $<
+
 $(OBJS)\bench_strings.o: ./strings.cpp
 	$(CXX) -c -o $@ $(BENCH_CXXFLAGS) $(CPPDEPS) $<
 

tests/benchmarks/makefile.vc

@@ -37,6 +37,7 @@ BENCH_OBJECTS =  \
 	$(OBJS)\bench_ipcclient.obj \
 	$(OBJS)\bench_log.obj \
 	$(OBJS)\bench_mbconv.obj \
+	$(OBJS)\bench_regex.obj \
 	$(OBJS)\bench_strings.obj \
 	$(OBJS)\bench_tls.obj \
 	$(OBJS)\bench_printfbench.obj
@@ -698,6 +699,9 @@ $(OBJS)\bench_log.obj: .\log.cpp
 $(OBJS)\bench_mbconv.obj: .\mbconv.cpp
 	$(CXX) /c /nologo /TP /Fo$@ $(BENCH_CXXFLAGS) .\mbconv.cpp
 
+$(OBJS)\bench_regex.obj: .\regex.cpp
+	$(CXX) /c /nologo /TP /Fo$@ $(BENCH_CXXFLAGS) .\regex.cpp
+
 $(OBJS)\bench_strings.obj: .\strings.cpp
 	$(CXX) /c /nologo /TP /Fo$@ $(BENCH_CXXFLAGS) .\strings.cpp
 

tests/benchmarks/regex.cpp

@@ -0,0 +1,74 @@
+/////////////////////////////////////////////////////////////////////////////
+// Name:        tests/benchmarks/regex.cpp
+// Purpose:     wxRegEx benchmarks
+// Author:      Vadim Zeitlin
+// Created:     2018-11-15
+// Copyright:   (c) 2018 Vadim Zeitlin <vadim@wxwidgets.org>
+// Licence:     wxWindows licence
+/////////////////////////////////////////////////////////////////////////////
+
+#include "wx/ffile.h"
+#include "wx/regex.h"
+
+#include "bench.h"
+
+// ----------------------------------------------------------------------------
+// Benchmark relative costs of compiling and matching for a simple regex
+// ----------------------------------------------------------------------------
+
+static const char* const RE_SIMPLE = ".";
+
+BENCHMARK_FUNC(RECompile)
+{
+    return wxRegEx(RE_SIMPLE).IsValid();
+}
+
+BENCHMARK_FUNC(REMatch)
+{
+    static wxRegEx re(RE_SIMPLE);
+    return re.Matches("foo");
+}
+
+BENCHMARK_FUNC(RECompileAndMatch)
+{
+    return wxRegEx(RE_SIMPLE).Matches("foo");
+}
+
+// ----------------------------------------------------------------------------
+// Benchmark the cost of using a more complicated regex
+// ----------------------------------------------------------------------------
+
+namespace
+{
+
+// Use the contents of an already existing test file.
+const wxString& GetTestText()
+{
+    static wxString text;
+    if ( text.empty() )
+    {
+        wxFFile("htmltest.html").ReadAll(&text);
+    }
+
+    return text;
+}
+
+} // anonymous namespace
+
+BENCHMARK_FUNC(REFindTD)
+{
+    // This is too simplistic, but good enough for benchmarking.
+    static wxRegEx re("<td>[^<]*</td>", wxRE_ICASE | wxRE_NEWLINE);
+
+    int matches = 0;
+    for ( const wxChar* p = GetTestText().c_str(); re.Matches(p); ++matches )
+    {
+        size_t start, len;
+        if ( !re.GetMatch(&start, &len) )
+            return false;
+
+        p += start + len;
+    }
+
+    return matches == 21; // result of "grep -c"
+}