Added Encoder.globalUseJavaString option

sksamuel · Apr 15, 2024 · 959f034 · 959f034
1 parent 6baf72c
commit 959f034
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 14 deletions.
diff --git a/centurion-avro/src/main/kotlin/com/sksamuel/centurion/avro/encoders/Encoder.kt b/centurion-avro/src/main/kotlin/com/sksamuel/centurion/avro/encoders/Encoder.kt
@@ -24,14 +24,16 @@ fun interface Encoder<T> {
 
    companion object {
 
+      var globalUseJavaString: Boolean = false
+
       /**
        * Returns an [Encoder] that encodes by simply returning the input value.
        */
       fun <T : Any> identity(): Encoder<T> = Encoder { _, value -> value }
 
       fun encoderFor(type: KType): Encoder<*> {
          val encoder: Encoder<*> = when (val classifier = type.classifier) {
-            String::class -> StringEncoder
+            String::class -> if (globalUseJavaString) JavaStringEncoder else StringEncoder
             Boolean::class -> BooleanEncoder
             Float::class -> FloatEncoder
             Double::class -> DoubleEncoder

diff --git a/centurion-avro/src/main/kotlin/com/sksamuel/centurion/avro/encoders/strings.kt b/centurion-avro/src/main/kotlin/com/sksamuel/centurion/avro/encoders/strings.kt
@@ -21,6 +21,14 @@ object StringEncoder : Encoder<String> {
    }
 }
 
+/**
+ * An [Encoder] for strings which always uses the JVM String object regardless
+ * of any [GenericData.STRING_PROP] settings on the schema.
+ */
+object JavaStringEncoder : Encoder<String> {
+   override fun encode(schema: Schema, value: String) = value
+}
+
 object UUIDEncoder : Encoder<UUID> {
    override fun encode(schema: Schema, value: UUID): Any {
       return Utf8(value.toString())

diff --git a/centurion-avro/src/test/kotlin/com/sksamuel/centurion/avro/encoders/Benchmarks.kt b/centurion-avro/src/test/kotlin/com/sksamuel/centurion/avro/encoders/Benchmarks.kt
@@ -26,26 +26,37 @@ fun main() {
 
    GenericData.setStringType(schema, GenericData.StringType.String)
 
-//   repeat(5) {
-//      val time = measureTime {
-//         repeat(5_000_000) {
-//            ReflectionRecordEncoder().encode(schema, user)
-//         }
-//      }
-//      println("ReflectionRecordEncoder: $time")
-//   }
+   repeat(2) {
+      val time = measureTime {
+         repeat(2_000_000) {
+            ReflectionRecordEncoder().encode(schema, user)
+         }
+      }
+      println("ReflectionRecordEncoder: $time")
+   }
+
+   repeat(10) {
+      val encoder = SpecificRecordEncoder(User::class, schema)
+      val time = measureTime {
+         repeat(2_000_000) {
+            encoder.encode(schema, user)
+         }
+      }
+      println("SpecificRecordEncoder globalUseJavaString=false: $time")
+   }
 
-   repeat(5) {
+   repeat(10) {
+      Encoder.globalUseJavaString = true
       val encoder = SpecificRecordEncoder(User::class, schema)
       val time = measureTime {
-         repeat(5_000_000) {
+         repeat(2_000_000) {
             encoder.encode(schema, user)
          }
       }
-      println("SpecificRecordEncoder: $time")
+      println("SpecificRecordEncoder globalUseJavaString=true: $time")
    }
 
-   repeat(5) {
+   repeat(10) {
 
       val encoder = Encoder<User> { schema, value ->
          val record = GenericData.Record(schema)
@@ -58,7 +69,7 @@ fun main() {
       }
 
       val time = measureTime {
-         repeat(5_000_000) {
+         repeat(2_000_000) {
             encoder.encode(schema, user)
          }
       }