Adds an option to control utf8Encoding of strings.

avro-builder/builder/src/main/java/com/linkedin/avroutil1/builder/SchemaBuilder.java

@@ -51,6 +51,7 @@ public static void main(String[] args) throws Exception {
 
     long optionParseStart = System.currentTimeMillis();
     OptionParser parser = new OptionParser();
+
     OptionSpec<String> inputOpt = parser.accepts("input", "Schema or directory of schemas to compile [REQUIRED]")
         .withRequiredArg().required()
         .describedAs("file");
@@ -114,6 +115,11 @@ public static void main(String[] args) throws Exception {
         .defaultsTo("false")
         .describedAs("true/false");
 
+    OptionSpec<String> enableUtf8Encoding = parser.accepts("enableUtf8Encoding", "enable encoding strings to their utf8 values throughout generated code.")
+        .withOptionalArg()
+        .defaultsTo("true")
+        .describedAs("true/false");
+
     //allow plugins to add CLI options
     for (BuilderPlugin plugin : plugins) {
       plugin.customizeCLI(parser);
@@ -238,6 +244,15 @@ public static void main(String[] args) throws Exception {
       skipCodegenIfSchemaOnClasspath = Boolean.TRUE.equals(Boolean.parseBoolean(value));
     }
 
+    boolean handleUtf8Encoding = true;
+    if (options.has(enableUtf8Encoding)) {
+      String value = options.valueOf(enableUtf8Encoding);
+      handleUtf8Encoding = Boolean.TRUE.equals(Boolean.parseBoolean(value));
+      if (methodStringRepresentation.equals(StringRepresentation.CharSeq) && stringRepresentation.equals(StringRepresentation.CharSeq)) {
+        handleUtf8EncodingInPutByIndex = handleUtf8Encoding;
+      }
+    }
+
     //allow plugins to parse and validate their own added options
     for (BuilderPlugin plugin : plugins) {
       plugin.parseAndValidateOptions(options);
@@ -258,7 +273,8 @@ public static void main(String[] args) throws Exception {
         minAvroVer,
         handleAvro702,
         handleUtf8EncodingInPutByIndex,
-        skipCodegenIfSchemaOnClasspath
+        skipCodegenIfSchemaOnClasspath,
+        handleUtf8Encoding
     );
 
     opConfig.validateParameters();

avro-builder/builder/src/main/java/com/linkedin/avroutil1/builder/operations/codegen/CodeGenOpConfig.java

@@ -49,6 +49,7 @@ public class CodeGenOpConfig {
   boolean avro702Handling;
   boolean utf8EncodingPutByIndex;
   boolean skipCodegenIfSchemaOnClasspath;
+  boolean enableUtf8Encoding;
 
   @Deprecated
   public CodeGenOpConfig(
@@ -77,6 +78,7 @@ public CodeGenOpConfig(
     this.minAvroVersion = minAvroVersion;
     this.avro702Handling = avro702Handling;
     this.utf8EncodingPutByIndex = true;
+    this.enableUtf8Encoding = true;
   }
 
   @Deprecated
@@ -107,6 +109,7 @@ public CodeGenOpConfig(
     this.minAvroVersion = minAvroVersion;
     this.avro702Handling = avro702Handling;
     this.utf8EncodingPutByIndex = true;
+    this.enableUtf8Encoding = true;
   }
 
   @Deprecated
@@ -136,6 +139,7 @@ public CodeGenOpConfig(List<File> inputRoots,
     this.minAvroVersion = minAvroVersion;
     this.avro702Handling = avro702Handling;
     this.utf8EncodingPutByIndex = handleUtf8EncodingInPutByIndex;
+    this.enableUtf8Encoding = true;
   }
 
   @Deprecated
@@ -167,8 +171,10 @@ public CodeGenOpConfig(List<File> inputRoots,
     this.avro702Handling = avro702Handling;
     this.utf8EncodingPutByIndex = handleUtf8EncodingInPutByIndex;
     this.skipCodegenIfSchemaOnClasspath = skipCodegenIfSchemaOnClasspath;
+    this.enableUtf8Encoding = true;
   }
 
+  @Deprecated
   public CodeGenOpConfig(List<File> inputRoots,
       List<File> nonImportableSourceRoots,
       boolean includeClasspath,
@@ -199,6 +205,41 @@ public CodeGenOpConfig(List<File> inputRoots,
     this.avro702Handling = avro702Handling;
     this.utf8EncodingPutByIndex = handleUtf8EncodingInPutByIndex;
     this.skipCodegenIfSchemaOnClasspath = skipCodegenIfSchemaOnClasspath;
+    this.enableUtf8Encoding = true;
+  }
+
+  public CodeGenOpConfig(List<File> inputRoots,
+      List<File> nonImportableSourceRoots,
+      boolean includeClasspath,
+      File outputSpecificRecordClassesRoot,
+      File outputExpandedSchemasRoot,
+      List<File> resolverPath,
+      CodeGenerator generatorType,
+      DuplicateSchemaBehaviour dupBehaviour,
+      List<String> duplicateSchemasToIgnore,
+      StringRepresentation stringRepresentation,
+      StringRepresentation methodStringRepresentation,
+      AvroVersion minAvroVersion,
+      boolean avro702Handling,
+      boolean handleUtf8EncodingInPutByIndex,
+      boolean skipCodegenIfSchemaOnClasspath,
+      boolean handleUtf8Encoding) {
+    this.inputRoots = inputRoots;
+    this.nonImportableSourceRoots = nonImportableSourceRoots;
+    this.includeClasspath = includeClasspath;
+    this.outputSpecificRecordClassesRoot = outputSpecificRecordClassesRoot;
+    this.outputExpandedSchemasRoot = outputExpandedSchemasRoot;
+    this.resolverPath = resolverPath;
+    this.generatorType = generatorType;
+    this.dupBehaviour = dupBehaviour;
+    this.duplicateSchemasToIgnore = duplicateSchemasToIgnore;
+    this.stringRepresentation = stringRepresentation;
+    this.methodStringRepresentation = methodStringRepresentation;
+    this.minAvroVersion = minAvroVersion;
+    this.avro702Handling = avro702Handling;
+    this.utf8EncodingPutByIndex = handleUtf8EncodingInPutByIndex;
+    this.skipCodegenIfSchemaOnClasspath = skipCodegenIfSchemaOnClasspath;
+    this.enableUtf8Encoding = handleUtf8Encoding;
   }
 
   /**
@@ -312,6 +353,10 @@ public boolean shouldSkipCodegenIfSchemaOnClasspath() {
     return skipCodegenIfSchemaOnClasspath;
   }
 
+  public boolean isUtf8EncodingEnabled() {
+    return enableUtf8Encoding;
+  }
+
   private void validateInput(Collection<File> files, String desc) {
     for (File f : files) {
       if (!f.exists()) {

avro-builder/builder/src/main/java/com/linkedin/avroutil1/builder/operations/codegen/own/AvroUtilCodeGenPlugin.java

@@ -102,7 +102,7 @@ private void generateCode(OperationContext opContext) {
         SpecificRecordGenerationConfig.getBroadCompatibilitySpecificRecordGenerationConfig(
             AvroJavaStringRepresentation.fromJson(config.getStringRepresentation().toString()),
             AvroJavaStringRepresentation.fromJson(config.getMethodStringRepresentation().toString()),
-            config.getMinAvroVersion(), config.isUtf8EncodingPutByIndexEnabled());
+            config.getMinAvroVersion(), config.isUtf8EncodingPutByIndexEnabled(), config.isUtf8EncodingEnabled());
     final SpecificRecordClassGenerator generator = new SpecificRecordClassGenerator();
     List<JavaFile> generatedClasses = allNamedSchemas.stream().collect(StreamUtil.toParallelStream(namedSchema -> {
       try {

avro-builder/tests/codegen-no-utf8-encoding/build.gradle

@@ -0,0 +1,68 @@
+/*
+ * Copyright 2022 LinkedIn Corp.
+ * Licensed under the BSD 2-Clause License (the "License").
+ * See License in the project root for license information.
+ */
+
+plugins {
+  id "java-library"
+}
+
+configurations {
+  codegen
+}
+
+sourceSets {
+  main {
+    java {
+      srcDir 'src/main/java'
+      srcDir "$buildDir/generated/sources/avro/java/main"
+    }
+    resources {
+      srcDirs = [
+        "src/main/avro"
+      ]
+    }
+  }
+}
+
+dependencies {
+  codegen project(":avro-builder:builder")
+}
+
+task runOwnCodegen {
+  description = 'generate specific classes using own codegen utility'
+
+  dependsOn configurations.codegen
+
+  doLast {
+    javaexec {
+      classpath configurations.codegen
+      main = 'com.linkedin.avroutil1.builder.SchemaBuilder'
+      args = [
+          "--input", "$projectDir/src/main/avro",
+          "--output", "$buildDir/generated/sources/avro/java/main",
+          "--generator", "AVRO_UTIL",
+          "--stringRepresentation", "CharSequence",
+          "--methodStringRepresentation", "CharSequence",
+          "--enableUtf8Encoding", "false"
+      ]
+    }
+  }
+}
+
+compileJava.dependsOn runOwnCodegen
+
+dependencies {
+  codegen project(":avro-builder:builder")
+
+  implementation("org.apache.avro:avro:1.4.1") {
+    exclude group: "org.mortbay.jetty"
+    exclude group: "org.apache.velocity"
+    exclude group: "commons-lang"
+    exclude group: "org.jboss.netty"
+    exclude group: "com.thoughtworks.paranamer", module: "paranamer-ant"
+  }
+  //required because generated code depends on the helper
+  implementation project(":helper:helper")
+}

avro-builder/tests/codegen-no-utf8-encoding/src/main/avro/noUtf8Encoding/ArrayOfStringRecord.avsc

@@ -0,0 +1,15 @@
+{
+  "type": "record",
+  "namespace": "noutf8encoding",
+  "name": "ArrayOfStringRecord",
+  "doc": "Array of String Record",
+  "fields": [
+    {
+      "name": "arOfRec",
+      "type": {
+        "type":"array",
+        "items": "string"
+      }
+    }
+  ]
+}

avro-builder/tests/codegen-no-utf8-encoding/src/main/avro/noUtf8Encoding/HasNoSimpleString.avsc

@@ -0,0 +1,23 @@
+{
+  "type": "record",
+  "namespace": "noutf8encoding",
+  "name": "HasNoSimpleString",
+  "fields": [
+    {
+      "name": "one",
+      "type": "float"
+    },
+    {
+      "name": "two",
+      "type": ["string"]
+    },
+    {
+      "name": "three",
+      "type": "int"
+    },
+    {
+      "name": "four",
+      "type": "boolean"
+    }
+  ]
+}

avro-builder/tests/codegen-no-utf8-encoding/src/main/avro/noUtf8Encoding/TestCollections.avsc

@@ -0,0 +1,95 @@
+{
+  "name": "TestCollections",
+  "namespace": "noutf8encoding",
+  "fields": [
+    {
+      "name": "str",
+      "type": "string"
+    },
+    {
+      "name": "strAr",
+      "type": {
+        "type": "array",
+        "items": "string"
+      }
+    },
+    {
+      "name": "strArAr",
+      "type": {
+        "type": "array",
+        "items": {
+          "type": "array",
+          "items": "string"
+        }
+      }
+    },
+    {
+      "name": "unionOfArray",
+      "type": [
+        "null",
+        {
+          "type": "array",
+          "items": "string"
+        }
+      ]
+    },
+    {
+      "name": "arOfMap",
+      "type": {
+        "type": "array",
+        "items": {
+          "type": "map",
+          "values": "string"
+        }
+      }
+    },
+    {
+      "name": "unionOfMap",
+      "type": [
+        "null",
+        {
+          "type": "map",
+          "values": "string"
+        }
+      ]
+    },
+    {
+      "name": "arOfUnionOfStr",
+      "type": {
+        "type": "array",
+        "items": ["null", "string"]
+      }
+    },
+    {
+      "name": "arOfMapOfUnionOfArray",
+      "type": {
+        "type": "array",
+        "items": {
+          "type": "map",
+          "values": ["null", {
+            "type": "array",
+            "items": "string"
+          }]
+        }
+      }
+    },
+    {
+      "name": "intAr",
+      "type": {
+        "type": "array",
+        "items": "int"
+      }
+    },
+    {
+      "name": "unionOfIntMap",
+      "type": [
+        "null",
+        {
+          "type": "map",
+          "values": "int"
+        }
+      ]
+    }
+  ],
+  "type": "record"
+}

avro-builder/tests/tests-allavro/build.gradle

@@ -80,6 +80,9 @@ dependencies {
   testImplementation (project(":avro-builder:tests:codegen-no-utf8-in-putbyindex")) {
     exclude group: "org.apache.avro"
   }
+  testImplementation (project(":avro-builder:tests:codegen-no-utf8-encoding")) {
+    exclude group: "org.apache.avro"
+  }
 
   testImplementation "com.google.guava:guava:28.2-jre"
   testImplementation "org.mockito:mockito-core:3.2.4"