Add benchmark for matching utf8 bytes

Found no meaningful difference between binary vs string.
google · Mar 8, 2021 · 34f3041 · 34f3041
1 parent c4e8120
commit 34f3041
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 5 deletions.
diff --git a/benchmarks/src/main/java/com/google/re2j/benchmark/BenchmarkFullMatch.java b/benchmarks/src/main/java/com/google/re2j/benchmark/BenchmarkFullMatch.java
@@ -6,6 +6,8 @@
  */
 package com.google.re2j.benchmark;
 
+import static java.nio.charset.StandardCharsets.UTF_8;
+
 import org.openjdk.jmh.annotations.Benchmark;
 import org.openjdk.jmh.annotations.OutputTimeUnit;
 import org.openjdk.jmh.annotations.Param;
@@ -23,8 +25,17 @@ public class BenchmarkFullMatch {
   @Param({"JDK", "RE2J"})
   private Implementations impl;
 
+  @Param({"true", "false"})
+  private boolean binary;
+
   private Implementations.Pattern pattern;
 
+  private String password = "password";
+  private byte[] password_bytes = password.getBytes(UTF_8);
+
+  private String l0ngpassword = "l0ngpassword";
+  private byte[] l0ngpassword_bytes = "l0ngpassword".getBytes(UTF_8);
+
   @Setup
   public void setup() {
     pattern =
@@ -34,7 +45,8 @@ public void setup() {
 
   @Benchmark
   public void matched(Blackhole bh) {
-    Implementations.Matcher matcher = pattern.matcher("password");
+    Implementations.Matcher matcher =
+        binary ? pattern.matcher(password_bytes) : pattern.matcher(password);
     boolean matches = matcher.matches();
     if (!matches) {
       throw new AssertionError();
@@ -44,7 +56,8 @@ public void matched(Blackhole bh) {
 
   @Benchmark
   public void notMatched(Blackhole bh) {
-    Implementations.Matcher matcher = pattern.matcher("l0ngpassword");
+    Implementations.Matcher matcher =
+        binary ? pattern.matcher(l0ngpassword_bytes) : pattern.matcher(l0ngpassword);
     boolean matches = matcher.matches();
     if (matches) {
       throw new AssertionError();

diff --git a/benchmarks/src/main/java/com/google/re2j/benchmark/BenchmarkSubMatch.java b/benchmarks/src/main/java/com/google/re2j/benchmark/BenchmarkSubMatch.java
@@ -27,8 +27,12 @@ public class BenchmarkSubMatch {
   @Param({"JDK", "RE2J"})
   private Implementations impl;
 
-  private String html =
-      new String(readFile("google-maps-contact-info.html"), StandardCharsets.UTF_8);
+  @Param({"true", "false"})
+  private boolean binary;
+
+  byte[] bytes = readFile("google-maps-contact-info.html");
+  private String html = new String(bytes, StandardCharsets.UTF_8);
+
   private Implementations.Pattern pattern;
 
   @Setup
@@ -38,7 +42,7 @@ public void setup() {
 
   @Benchmark
   public void findPhoneNumbers(Blackhole bh) {
-    Implementations.Matcher matcher = pattern.matcher(html);
+    Implementations.Matcher matcher = binary ? pattern.matcher(bytes) : pattern.matcher(html);
     int count = 0;
     while (matcher.find()) {
       bh.consume(matcher.group());

diff --git a/benchmarks/src/main/java/com/google/re2j/benchmark/Implementations.java b/benchmarks/src/main/java/com/google/re2j/benchmark/Implementations.java
@@ -80,6 +80,8 @@ public static Pattern compile(Implementations impl, String pattern) {
 
     public abstract Matcher matcher(String str);
 
+    public abstract Matcher matcher(byte[] bytes);
+
     public static class JdkPattern extends Pattern {
 
       private final java.util.regex.Pattern pattern;
@@ -92,6 +94,11 @@ public JdkPattern(String pattern) {
       public Matcher matcher(String str) {
         return new Matcher.JdkMatcher(pattern.matcher(str));
       }
+
+      @Override
+      public Matcher matcher(byte[] bytes) {
+        return new Matcher.JdkMatcher(pattern.matcher(new String(bytes)));
+      }
     }
 
     public static class Re2Pattern extends Pattern {
@@ -106,6 +113,11 @@ public Re2Pattern(String pattern) {
       public Matcher matcher(String str) {
         return new Matcher.Re2Matcher(pattern.matcher(str));
       }
+
+      @Override
+      public Matcher matcher(byte[] bytes) {
+        return new Matcher.Re2Matcher(pattern.matcher(bytes));
+      }
     }
   }
 }