avro-kotlin · Chuckame · Jul 11, 2024 · Jul 10, 2024
diff --git a/README.md b/README.md
diff --git a/api/avro4k-core.api b/api/avro4k-core.api
@@ -228,6 +228,13 @@ public final class com/github/avrokotlin/avro4k/AvroSingleObjectKt {
 	public static final fun encodeToByteArray (Lcom/github/avrokotlin/avro4k/AvroSingleObject;Lorg/apache/avro/Schema;Lkotlinx/serialization/SerializationStrategy;Ljava/lang/Object;)[B
 }
 
+public abstract interface annotation class com/github/avrokotlin/avro4k/AvroStringable : java/lang/annotation/Annotation {
+}
+
+public synthetic class com/github/avrokotlin/avro4k/AvroStringable$Impl : com/github/avrokotlin/avro4k/AvroStringable {
+	public fun <init> ()V
+}
+
 public abstract interface class com/github/avrokotlin/avro4k/BooleanValueDecoder {
 	public abstract fun decodeBoolean (Lcom/github/avrokotlin/avro4k/AvroDecoder;)Z
 }
@@ -361,9 +368,11 @@ public abstract class com/github/avrokotlin/avro4k/serializer/AvroSerializer : c
 }
 
 public final class com/github/avrokotlin/avro4k/serializer/AvroSerializerKt {
-	public static final fun createSchema (Lcom/github/avrokotlin/avro4k/serializer/FoundElementAnnotation;)Lorg/apache/avro/Schema;
-	public static final fun getDecimal (Lcom/github/avrokotlin/avro4k/serializer/SchemaSupplierContext;)Lcom/github/avrokotlin/avro4k/serializer/FoundElementAnnotation;
-	public static final fun getFixed (Lcom/github/avrokotlin/avro4k/serializer/SchemaSupplierContext;)Lcom/github/avrokotlin/avro4k/serializer/FoundElementAnnotation;
+	public static final fun createSchema (Lcom/github/avrokotlin/avro4k/AvroFixed;Lcom/github/avrokotlin/avro4k/serializer/ElementLocation;)Lorg/apache/avro/Schema;
+	public static final fun createSchema (Lcom/github/avrokotlin/avro4k/AvroStringable;)Lorg/apache/avro/Schema;
+	public static final fun getDecimal (Lcom/github/avrokotlin/avro4k/serializer/ElementLocation;)Lcom/github/avrokotlin/avro4k/AvroDecimal;
+	public static final fun getFixed (Lcom/github/avrokotlin/avro4k/serializer/ElementLocation;)Lcom/github/avrokotlin/avro4k/AvroFixed;
+	public static final fun getStringable (Lcom/github/avrokotlin/avro4k/serializer/ElementLocation;)Lcom/github/avrokotlin/avro4k/AvroStringable;
 }
 
 public final class com/github/avrokotlin/avro4k/serializer/BigDecimalAsStringSerializer : com/github/avrokotlin/avro4k/serializer/AvroSerializer {
@@ -418,21 +427,6 @@ public final class com/github/avrokotlin/avro4k/serializer/ElementLocation {
 	public fun toString ()Ljava/lang/String;
 }
 
-public final class com/github/avrokotlin/avro4k/serializer/FoundElementAnnotation {
-	public fun <init> (Lkotlinx/serialization/descriptors/SerialDescriptor;ILjava/lang/annotation/Annotation;)V
-	public final fun component1 ()Lkotlinx/serialization/descriptors/SerialDescriptor;
-	public final fun component2 ()I
-	public final fun component3 ()Ljava/lang/annotation/Annotation;
-	public final fun copy (Lkotlinx/serialization/descriptors/SerialDescriptor;ILjava/lang/annotation/Annotation;)Lcom/github/avrokotlin/avro4k/serializer/FoundElementAnnotation;
-	public static synthetic fun copy$default (Lcom/github/avrokotlin/avro4k/serializer/FoundElementAnnotation;Lkotlinx/serialization/descriptors/SerialDescriptor;ILjava/lang/annotation/Annotation;ILjava/lang/Object;)Lcom/github/avrokotlin/avro4k/serializer/FoundElementAnnotation;
-	public fun equals (Ljava/lang/Object;)Z
-	public final fun getAnnotation ()Ljava/lang/annotation/Annotation;
-	public final fun getDescriptor ()Lkotlinx/serialization/descriptors/SerialDescriptor;
-	public final fun getElementIndex ()I
-	public fun hashCode ()I
-	public fun toString ()Ljava/lang/String;
-}
-
 public final class com/github/avrokotlin/avro4k/serializer/InstantSerializer : com/github/avrokotlin/avro4k/serializer/AvroSerializer {
 	public static final field INSTANCE Lcom/github/avrokotlin/avro4k/serializer/InstantSerializer;
 	public synthetic fun deserializeAvro (Lcom/github/avrokotlin/avro4k/AvroDecoder;)Ljava/lang/Object;

diff --git a/src/main/kotlin/com/github/avrokotlin/avro4k/Annotations.kt b/src/main/kotlin/com/github/avrokotlin/avro4k/Annotations.kt
@@ -1,7 +1,6 @@
-@file:OptIn(ExperimentalSerializationApi::class)
-
 package com.github.avrokotlin.avro4k
 
+import com.github.avrokotlin.avro4k.serializer.AvroSerializer
 import com.github.avrokotlin.avro4k.serializer.BigDecimalSerializer
 import kotlinx.serialization.ExperimentalSerializationApi
 import kotlinx.serialization.SerialInfo
@@ -46,6 +45,29 @@ public annotation class AvroDoc(val value: String)
 @Target(AnnotationTarget.PROPERTY, AnnotationTarget.CLASS)
 public annotation class AvroAlias(vararg val value: String)
 
+/**
+ * Sets the annotated property as a `string` type when inferring the class' schema. Takes precedence over any other schema modifier built-in annotation, except
+ * the fields referring to a custom serializer implementing [AvroSerializer] where you'll need to handle the annotation.
+ *
+ * If the given property is not string-able, the schema will be generated as a string, but it will fail at runtime during serialization or deserialization.
+ * You may need to provide a custom serializer for the given property to handle the string specifications.
+ *
+ * Only works with class properties for the following inferred schemas:
+ * - `string`
+ * - `boolean`
+ * - `int`
+ * - `long`
+ * - `float`
+ * - `double`
+ * - `bytes` & `fixed` (will take the fixed bytes as UTF-8 string, or custom toString/parse for logical types)
+ * - works also for all the nullable types of the above
+ * The rest will fail, except if your custom serializers handle the string type.
+ */
+@SerialInfo
+@ExperimentalSerializationApi
+@Target(AnnotationTarget.PROPERTY)
+public annotation class AvroStringable
+
 /**
  * To be used with [BigDecimalSerializer] to specify the scale, precision, type and rounding mode of the decimal value.
  *

diff --git a/src/main/kotlin/com/github/avrokotlin/avro4k/internal/SerializerLocatorMiddleware.kt b/src/main/kotlin/com/github/avrokotlin/avro4k/internal/SerializerLocatorMiddleware.kt
@@ -10,6 +10,7 @@ import com.github.avrokotlin.avro4k.serializer.SchemaSupplierContext
 import com.github.avrokotlin.avro4k.serializer.SerialDescriptorWithAvroSchemaDelegate
 import com.github.avrokotlin.avro4k.serializer.createSchema
 import com.github.avrokotlin.avro4k.serializer.fixed
+import com.github.avrokotlin.avro4k.serializer.stringable
 import kotlinx.serialization.DeserializationStrategy
 import kotlinx.serialization.InternalSerializationApi
 import kotlinx.serialization.KSerializer
@@ -64,14 +65,17 @@ internal object SerializerLocatorMiddleware {
 
 private val AvroStringSerialDescriptor: SerialDescriptor =
     SerialDescriptorWithAvroSchemaDelegate(String.serializer().descriptor) { context ->
-        context.fixed?.createSchema() ?: Schema.create(Schema.Type.STRING)
+        context.inlinedElements.firstNotNullOfOrNull {
+            it.stringable?.createSchema() ?: it.fixed?.createSchema(it)
+        } ?: Schema.create(Schema.Type.STRING)
     }
 
 private object KotlinDurationSerializer : AvroSerializer<Duration>(Duration::class.qualifiedName!!) {
     private const val MILLIS_PER_DAY = 1000 * 60 * 60 * 24
 
     override fun getSchema(context: SchemaSupplierContext): Schema {
-        return AvroDurationSerializer.DURATION_SCHEMA
+        return context.inlinedElements.firstNotNullOfOrNull { it.stringable?.createSchema() }
+            ?: AvroDurationSerializer.DURATION_SCHEMA
     }
 
     override fun serializeAvro(
@@ -128,7 +132,9 @@ private object KotlinDurationSerializer : AvroSerializer<Duration>(Duration::cla
 
 private object AvroByteArraySerializer : AvroSerializer<ByteArray>(ByteArray::class.qualifiedName!!) {
     override fun getSchema(context: SchemaSupplierContext): Schema {
-        return context.fixed?.createSchema() ?: Schema.create(Schema.Type.BYTES)
+        return context.inlinedElements.firstNotNullOfOrNull {
+            it.stringable?.createSchema() ?: it.fixed?.createSchema(it)
+        } ?: Schema.create(Schema.Type.BYTES)
     }
 
     override fun serializeAvro(

diff --git a/.../kotlin/com/github/avrokotlin/avro4k/internal/encoder/direct/AbstractAvroDirectEncoder.kt b/.../kotlin/com/github/avrokotlin/avro4k/internal/encoder/direct/AbstractAvroDirectEncoder.kt
@@ -17,6 +17,7 @@ import kotlinx.serialization.encoding.CompositeEncoder
 import kotlinx.serialization.modules.SerializersModule
 import org.apache.avro.Schema
 import org.apache.avro.generic.GenericFixed
+import org.apache.avro.util.Utf8
 import java.nio.ByteBuffer
 
 internal class AvroValueDirectEncoder(
@@ -138,12 +139,14 @@ internal sealed class AbstractAvroDirectEncoder(
             { BadEncodedValueError(value, currentWriterSchema, Schema.Type.BYTES, Schema.Type.STRING, Schema.Type.FIXED) }
         ) {
             when (it.type) {
-                Schema.Type.STRING,
-                Schema.Type.BYTES,
-                -> {
+                Schema.Type.BYTES -> {
                     { binaryEncoder.writeBytes(value) }
                 }
 
+                Schema.Type.STRING -> {
+                    { binaryEncoder.writeString(Utf8(value.array())) }
+                }
+
                 Schema.Type.FIXED -> {
                     if (value.remaining() == it.fixedSize) {
                         { binaryEncoder.writeFixed(value.array()) }
@@ -162,12 +165,14 @@ internal sealed class AbstractAvroDirectEncoder(
             { BadEncodedValueError(value, currentWriterSchema, Schema.Type.BYTES, Schema.Type.STRING, Schema.Type.FIXED) }
         ) {
             when (it.type) {
-                Schema.Type.STRING,
-                Schema.Type.BYTES,
-                -> {
+                Schema.Type.BYTES -> {
                     { binaryEncoder.writeBytes(value) }
                 }
 
+                Schema.Type.STRING -> {
+                    { binaryEncoder.writeString(Utf8(value)) }
+                }
+
                 Schema.Type.FIXED -> {
                     if (value.size == it.fixedSize) {
                         { binaryEncoder.writeFixed(value) }
@@ -194,12 +199,14 @@ internal sealed class AbstractAvroDirectEncoder(
                     }
                 }
 
-                Schema.Type.STRING,
-                Schema.Type.BYTES,
-                -> {
+                Schema.Type.BYTES -> {
                     { binaryEncoder.writeBytes(value.bytes()) }
                 }
 
+                Schema.Type.STRING -> {
+                    { binaryEncoder.writeString(Utf8(value.bytes())) }
+                }
+
                 else -> null
             }
         }
@@ -217,12 +224,14 @@ internal sealed class AbstractAvroDirectEncoder(
                         null
                     }
 
-                Schema.Type.STRING,
-                Schema.Type.BYTES,
-                -> {
+                Schema.Type.BYTES -> {
                     { binaryEncoder.writeBytes(value) }
                 }
 
+                Schema.Type.STRING -> {
+                    { binaryEncoder.writeString(Utf8(value)) }
+                }
+
                 else -> null
             }
         }

diff --git a/src/main/kotlin/com/github/avrokotlin/avro4k/internal/schema/ValueVisitor.kt b/src/main/kotlin/com/github/avrokotlin/avro4k/internal/schema/ValueVisitor.kt
@@ -6,6 +6,7 @@ import com.github.avrokotlin.avro4k.internal.jsonNode
 import com.github.avrokotlin.avro4k.internal.nonNullSerialName
 import com.github.avrokotlin.avro4k.internal.nullable
 import com.github.avrokotlin.avro4k.serializer.AvroSchemaSupplier
+import com.github.avrokotlin.avro4k.serializer.stringable
 import kotlinx.serialization.descriptors.PolymorphicKind
 import kotlinx.serialization.descriptors.PrimitiveKind
 import kotlinx.serialization.descriptors.SerialDescriptor
@@ -84,9 +85,17 @@ internal class ValueVisitor internal constructor(
     override fun visitValue(descriptor: SerialDescriptor) {
         val finalDescriptor = SerializerLocatorMiddleware.apply(unwrapNullable(descriptor))
 
-        (finalDescriptor.nonNullOriginal as? AvroSchemaSupplier)
-            ?.getSchema(context)?.let { setSchema(it) }
-            ?: super.visitValue(finalDescriptor)
+        if (finalDescriptor is AvroSchemaSupplier) {
+            setSchema(finalDescriptor.getSchema(context))
+            return
+        }
+
+        if (context.inlinedElements.any { it.stringable != null }) {
+            setSchema(Schema.create(Schema.Type.STRING))
+            return
+        }
+
+        super.visitValue(finalDescriptor)
     }
 
     private fun unwrapNullable(descriptor: SerialDescriptor): SerialDescriptor {

diff --git a/src/main/kotlin/com/github/avrokotlin/avro4k/serializer/AvroDuration.kt b/src/main/kotlin/com/github/avrokotlin/avro4k/serializer/AvroDuration.kt
@@ -44,8 +44,8 @@ public data class AvroDuration(
             if (days != 0u) {
                 append("${days}D")
             }
-            append("T")
             if (millis != 0u) {
+                append("T")
                 append(millis / 1000u)
                 val millisPart = millis % 1000u
                 if (millisPart != 0u) {
@@ -89,12 +89,20 @@ public data class AvroDuration(
             val match = PATTERN.matchEntire(value) ?: return null
             val (years, months, weeks, days, hours, minutes, seconds, millis) = match.destructured
             return AvroDuration(
-                months = years.toUInt() * 12u + months.toUInt(),
-                days = weeks.toUInt() * 7u + days.toUInt(),
-                millis = hours.toUInt() * 60u * 60u * 1000u + minutes.toUInt() * 60u * 1000u + seconds.toUInt() * 1000u + millis.toUInt()
+                months = years * 12u + months.toUIntOrZero(),
+                days = weeks * 7u + days.toUIntOrZero(),
+                millis = hours * 60u * 60u * 1000u + minutes * 60u * 1000u + seconds * 1000u + millis.toUIntOrZero()
             )
         }
 
+        private operator fun String.times(other: UInt): UInt {
+            return toUIntOrNull()?.times(other) ?: 0u
+        }
+
+        private fun String.toUIntOrZero(): UInt {
+            return toUIntOrNull() ?: 0u
+        }
+
         @JvmStatic
         public fun parse(value: String): AvroDuration {
             return tryParse(value) ?: throw AvroDurationParseException(value)
@@ -114,7 +122,9 @@ internal object AvroDurationSerializer : AvroSerializer<AvroDuration>(AvroDurati
         }
 
     override fun getSchema(context: SchemaSupplierContext): Schema {
-        return DURATION_SCHEMA
+        return context.inlinedElements.firstNotNullOfOrNull {
+            it.stringable?.createSchema()
+        } ?: DURATION_SCHEMA
     }
 
     override fun serializeAvro(

diff --git a/src/main/kotlin/com/github/avrokotlin/avro4k/serializer/AvroSerializer.kt b/src/main/kotlin/com/github/avrokotlin/avro4k/serializer/AvroSerializer.kt
@@ -5,6 +5,7 @@ import com.github.avrokotlin.avro4k.AvroDecimal
 import com.github.avrokotlin.avro4k.AvroDecoder
 import com.github.avrokotlin.avro4k.AvroEncoder
 import com.github.avrokotlin.avro4k.AvroFixed
+import com.github.avrokotlin.avro4k.AvroStringable
 import com.github.avrokotlin.avro4k.internal.findElementAnnotation
 import com.github.avrokotlin.avro4k.internal.namespace
 import kotlinx.serialization.ExperimentalSerializationApi
@@ -100,57 +101,53 @@ public interface SchemaSupplierContext {
 }
 
 /**
- * Search for the first annotation of type [T] in the [SchemaSupplierContext.inlinedElements].
- *
- * The top-est matching annotation is returned, that means the one that is closest to the class element.
+ * Search for the first annotation of type [T] in the given [ElementLocation].
  */
 @ExperimentalSerializationApi
-public inline fun <reified T : Annotation> SchemaSupplierContext.findAnnotation(): FoundElementAnnotation<T>? {
-    return inlinedElements.firstNotNullOfOrNull { elementLocation ->
-        elementLocation.descriptor.findElementAnnotation<T>(elementLocation.elementIndex)?.let {
-            FoundElementAnnotation(elementLocation.descriptor, elementLocation.elementIndex, it)
-        }
-    }
+public inline fun <reified T : Annotation> ElementLocation.findAnnotation(): T? {
+    return descriptor.findElementAnnotation(elementIndex)
 }
 
 /**
  * Shorthand for [findAnnotation] with [AvroDecimal] as it is a built-in annotation.
  */
 @ExperimentalSerializationApi
-public val SchemaSupplierContext.decimal: FoundElementAnnotation<AvroDecimal>?
+public val ElementLocation.decimal: AvroDecimal?
     get() = findAnnotation()
 
 /**
- * Shorthand for [findAnnotation] with [AvroFixed] as it is a built-in annotation.
+ * Shorthand for [findAnnotation] with [AvroStringable] as it is a built-in annotation.
  */
 @ExperimentalSerializationApi
-public val SchemaSupplierContext.fixed: FoundElementAnnotation<AvroFixed>?
+public val ElementLocation.stringable: AvroStringable?
     get() = findAnnotation()
 
 /**
- * Creates a fixed schema from the [AvroFixed] annotation.
+ * Creates a string schema from the [AvroStringable] annotation.
  */
 @ExperimentalSerializationApi
-public fun FoundElementAnnotation<AvroFixed>.createSchema(): Schema = Schema.createFixed(descriptor.getElementName(elementIndex), null, descriptor.namespace, annotation.size)
+public fun AvroStringable.createSchema(): Schema = Schema.create(Schema.Type.STRING)
 
+/**
+ * Shorthand for [findAnnotation] with [AvroFixed] as it is a built-in annotation.
+ */
 @ExperimentalSerializationApi
-public data class ElementLocation
-    @PublishedApi
-    internal constructor(
-        val descriptor: SerialDescriptor,
-        val elementIndex: Int,
-    )
+public val ElementLocation.fixed: AvroFixed?
+    get() = findAnnotation()
 
 /**
- * Represents a found annotation on an element, provided by [findAnnotation].
+ * Creates a fixed schema from the [AvroFixed] annotation.
  */
 @ExperimentalSerializationApi
-public data class FoundElementAnnotation<T : Annotation>
+public fun AvroFixed.createSchema(elementLocation: ElementLocation): Schema =
+    Schema.createFixed(elementLocation.descriptor.getElementName(elementLocation.elementIndex), null, elementLocation.descriptor.namespace, size)
+
+@ExperimentalSerializationApi
+public data class ElementLocation
     @PublishedApi
     internal constructor(
         val descriptor: SerialDescriptor,
         val elementIndex: Int,
-        val annotation: T,
     )
 
 internal fun interface AvroSchemaSupplier {

diff --git a/src/main/kotlin/com/github/avrokotlin/avro4k/serializer/JavaStdLibSerializers.kt b/src/main/kotlin/com/github/avrokotlin/avro4k/serializer/JavaStdLibSerializers.kt
@@ -185,10 +185,10 @@ private val defaultAnnotation = AvroDecimal()
 
 public object BigDecimalSerializer : AvroSerializer<BigDecimal>(BigDecimal::class.qualifiedName!!) {
     override fun getSchema(context: SchemaSupplierContext): Schema {
-        val decimalAnnotation = context.decimal?.annotation ?: defaultAnnotation
-        val schema = context.fixed?.createSchema() ?: Schema.create(Schema.Type.BYTES)
-        decimalAnnotation.logicalType.addToSchema(schema)
-        return schema
+        val logicalType = (context.inlinedElements.firstNotNullOfOrNull { it.decimal } ?: defaultAnnotation).logicalType
+        return context.inlinedElements.firstNotNullOfOrNull {
+            it.stringable?.createSchema() ?: it.fixed?.createSchema(it)?.copy(logicalType = logicalType)
+        } ?: Schema.create(Schema.Type.BYTES).copy(logicalType = logicalType)
     }
 
     override fun serializeAvro(