feat(client): accept InputStream and Path for file params (#277)

Author: stainless-app[bot]

openai-java-core/src/main/kotlin/com/openai/core/ObjectMappers.kt

@@ -3,23 +3,43 @@
 package com.openai.core
 
 import com.fasterxml.jackson.annotation.JsonInclude
+import com.fasterxml.jackson.core.JsonGenerator
 import com.fasterxml.jackson.databind.DeserializationFeature
 import com.fasterxml.jackson.databind.SerializationFeature
+import com.fasterxml.jackson.databind.SerializerProvider
 import com.fasterxml.jackson.databind.cfg.CoercionAction.Fail
 import com.fasterxml.jackson.databind.cfg.CoercionInputShape.Integer
 import com.fasterxml.jackson.databind.json.JsonMapper
+import com.fasterxml.jackson.databind.module.SimpleModule
 import com.fasterxml.jackson.datatype.jdk8.Jdk8Module
 import com.fasterxml.jackson.datatype.jsr310.JavaTimeModule
 import com.fasterxml.jackson.module.kotlin.jacksonMapperBuilder
+import java.io.InputStream
 
 fun jsonMapper(): JsonMapper =
     jacksonMapperBuilder()
         .addModule(Jdk8Module())
         .addModule(JavaTimeModule())
+        .addModule(SimpleModule().addSerializer(InputStreamJsonSerializer))
         .serializationInclusion(JsonInclude.Include.NON_ABSENT)
         .disable(DeserializationFeature.ADJUST_DATES_TO_CONTEXT_TIME_ZONE)
         .disable(SerializationFeature.FLUSH_AFTER_WRITE_VALUE)
         .disable(SerializationFeature.WRITE_DATES_AS_TIMESTAMPS)
         .disable(SerializationFeature.WRITE_DURATIONS_AS_TIMESTAMPS)
         .withCoercionConfig(String::class.java) { it.setCoercion(Integer, Fail) }
         .build()
+
+private object InputStreamJsonSerializer : BaseSerializer<InputStream>(InputStream::class) {
+
+    override fun serialize(
+        value: InputStream?,
+        gen: JsonGenerator?,
+        serializers: SerializerProvider?,
+    ) {
+        if (value == null) {
+            gen?.writeNull()
+        } else {
+            value.use { gen?.writeBinary(it.readBytes()) }
+        }
+    }
+}

openai-java-core/src/main/kotlin/com/openai/core/Values.kt

@@ -27,6 +27,7 @@ import com.fasterxml.jackson.databind.node.JsonNodeType.POJO
 import com.fasterxml.jackson.databind.node.JsonNodeType.STRING
 import com.fasterxml.jackson.databind.ser.std.NullSerializer
 import com.openai.errors.OpenAIInvalidDataException
+import java.io.InputStream
 import java.util.Objects
 import java.util.Optional
 
@@ -508,7 +509,10 @@ private constructor(
             return MultipartField(
                 value,
                 contentType
-                    ?: if (value is KnownValue && value.value is ByteArray)
+                    ?: if (
+                        value is KnownValue &&
+                            (value.value is InputStream || value.value is ByteArray)
+                    )
                         "application/octet-stream"
                     else "text/plain; charset=utf-8",
                 filename,

openai-java-core/src/main/kotlin/com/openai/core/http/HttpRequestBodies.kt

@@ -9,6 +9,8 @@ import com.fasterxml.jackson.databind.json.JsonMapper
 import com.fasterxml.jackson.databind.node.JsonNodeType
 import com.openai.core.MultipartField
 import com.openai.errors.OpenAIInvalidDataException
+import java.io.ByteArrayInputStream
+import java.io.InputStream
 import java.io.OutputStream
 import kotlin.jvm.optionals.getOrNull
 import org.apache.hc.client5.http.entity.mime.MultipartEntityBuilder
@@ -41,8 +43,18 @@ internal fun multipartFormData(
             MultipartEntityBuilder.create()
                 .apply {
                     fields.forEach { (name, field) ->
-                        val node = jsonMapper.valueToTree<JsonNode>(field.value)
-                        serializePart(name, node).forEach { (name, bytes) ->
+                        val knownValue = field.value.asKnown().getOrNull()
+                        val parts =
+                            if (knownValue is InputStream) {
+                                // Read directly from the `InputStream` instead of reading it all
+                                // into memory due to the `jsonMapper` serialization below.
+                                sequenceOf(name to knownValue)
+                            } else {
+                                val node = jsonMapper.valueToTree<JsonNode>(field.value)
+                                serializePart(name, node)
+                            }
+
+                        parts.forEach { (name, bytes) ->
                             addBinaryBody(
                                 name,
                                 bytes,
@@ -55,16 +67,19 @@ internal fun multipartFormData(
                 .build()
         }
 
-        private fun serializePart(name: String, node: JsonNode): Sequence<Pair<String, ByteArray>> =
+        private fun serializePart(
+            name: String,
+            node: JsonNode,
+        ): Sequence<Pair<String, InputStream>> =
             when (node.nodeType) {
                 JsonNodeType.MISSING,
                 JsonNodeType.NULL -> emptySequence()
-                JsonNodeType.BINARY -> sequenceOf(name to node.binaryValue())
-                JsonNodeType.STRING -> sequenceOf(name to node.textValue().toByteArray())
+                JsonNodeType.BINARY -> sequenceOf(name to ByteArrayInputStream(node.binaryValue()))
+                JsonNodeType.STRING -> sequenceOf(name to node.textValue().toInputStream())
                 JsonNodeType.BOOLEAN ->
-                    sequenceOf(name to node.booleanValue().toString().toByteArray())
+                    sequenceOf(name to node.booleanValue().toString().toInputStream())
                 JsonNodeType.NUMBER ->
-                    sequenceOf(name to node.numberValue().toString().toByteArray())
+                    sequenceOf(name to node.numberValue().toString().toInputStream())
                 JsonNodeType.ARRAY ->
                     node.elements().asSequence().flatMap { element ->
                         serializePart("$name[]", element)
@@ -78,6 +93,8 @@ internal fun multipartFormData(
                     throw OpenAIInvalidDataException("Unexpected JsonNode type: ${node.nodeType}")
             }
 
+        private fun String.toInputStream(): InputStream = ByteArrayInputStream(toByteArray())
+
         override fun writeTo(outputStream: OutputStream) = entity.writeTo(outputStream)
 
         override fun contentType(): String = entity.contentType

openai-java-core/src/main/kotlin/com/openai/models/AudioTranscriptionCreateParams.kt

@@ -14,8 +14,13 @@ import com.openai.core.http.Headers
 import com.openai.core.http.QueryParams
 import com.openai.core.toImmutable
 import com.openai.errors.OpenAIInvalidDataException
+import java.io.ByteArrayInputStream
+import java.io.InputStream
+import java.nio.file.Path
 import java.util.Objects
 import java.util.Optional
+import kotlin.io.path.inputStream
+import kotlin.io.path.name
 
 /** Transcribes audio into the input language. */
 class AudioTranscriptionCreateParams
@@ -29,7 +34,7 @@ private constructor(
      * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3, mp4,
      * mpeg, mpga, m4a, ogg, wav, or webm.
      */
-    fun file(): ByteArray = body.file()
+    fun file(): InputStream = body.file()
 
     /**
      * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -78,7 +83,7 @@ private constructor(
      * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3, mp4,
      * mpeg, mpga, m4a, ogg, wav, or webm.
      */
-    fun _file(): MultipartField<ByteArray> = body._file()
+    fun _file(): MultipartField<InputStream> = body._file()
 
     /**
      * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -148,7 +153,7 @@ private constructor(
     class Body
     @JsonCreator
     private constructor(
-        private val file: MultipartField<ByteArray>,
+        private val file: MultipartField<InputStream>,
         private val model: MultipartField<AudioModel>,
         private val language: MultipartField<String>,
         private val prompt: MultipartField<String>,
@@ -161,7 +166,7 @@ private constructor(
          * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3,
          * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
          */
-        fun file(): ByteArray = file.value.getRequired("file")
+        fun file(): InputStream = file.value.getRequired("file")
 
         /**
          * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -214,7 +219,7 @@ private constructor(
          * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3,
          * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
          */
-        fun _file(): MultipartField<ByteArray> = file
+        fun _file(): MultipartField<InputStream> = file
 
         /**
          * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -296,7 +301,7 @@ private constructor(
         /** A builder for [Body]. */
         class Builder internal constructor() {
 
-            private var file: MultipartField<ByteArray>? = null
+            private var file: MultipartField<InputStream>? = null
             private var model: MultipartField<AudioModel>? = null
             private var language: MultipartField<String> = MultipartField.of(null)
             private var prompt: MultipartField<String> = MultipartField.of(null)
@@ -321,13 +326,31 @@ private constructor(
              * The audio file object (not file name) to transcribe, in one of these formats: flac,
              * mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
              */
-            fun file(file: ByteArray) = file(MultipartField.of(file))
+            fun file(file: InputStream) = file(MultipartField.of(file))
 
             /**
              * The audio file object (not file name) to transcribe, in one of these formats: flac,
              * mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
              */
-            fun file(file: MultipartField<ByteArray>) = apply { this.file = file }
+            fun file(file: MultipartField<InputStream>) = apply { this.file = file }
+
+            /**
+             * The audio file object (not file name) to transcribe, in one of these formats: flac,
+             * mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
+             */
+            fun file(file: ByteArray) = file(ByteArrayInputStream(file))
+
+            /**
+             * The audio file object (not file name) to transcribe, in one of these formats: flac,
+             * mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
+             */
+            fun file(file: Path) =
+                file(
+                    MultipartField.builder<InputStream>()
+                        .value(file.inputStream())
+                        .filename(file.name)
+                        .build()
+                )
 
             /**
              * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper
@@ -506,6 +529,18 @@ private constructor(
             additionalQueryParams = audioTranscriptionCreateParams.additionalQueryParams.toBuilder()
         }
 
+        /**
+         * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3,
+         * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
+         */
+        fun file(file: InputStream) = apply { body.file(file) }
+
+        /**
+         * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3,
+         * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
+         */
+        fun file(file: MultipartField<InputStream>) = apply { body.file(file) }
+
         /**
          * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3,
          * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
@@ -516,7 +551,7 @@ private constructor(
          * The audio file object (not file name) to transcribe, in one of these formats: flac, mp3,
          * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
          */
-        fun file(file: MultipartField<ByteArray>) = apply { body.file(file) }
+        fun file(file: Path) = apply { body.file(file) }
 
         /**
          * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2

openai-java-core/src/main/kotlin/com/openai/models/AudioTranslationCreateParams.kt

@@ -10,8 +10,13 @@ import com.openai.core.checkRequired
 import com.openai.core.http.Headers
 import com.openai.core.http.QueryParams
 import com.openai.core.toImmutable
+import java.io.ByteArrayInputStream
+import java.io.InputStream
+import java.nio.file.Path
 import java.util.Objects
 import java.util.Optional
+import kotlin.io.path.inputStream
+import kotlin.io.path.name
 
 /** Translates audio into English. */
 class AudioTranslationCreateParams
@@ -25,7 +30,7 @@ private constructor(
      * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
      * mpeg, mpga, m4a, ogg, wav, or webm.
      */
-    fun file(): ByteArray = body.file()
+    fun file(): InputStream = body.file()
 
     /**
      * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -58,7 +63,7 @@ private constructor(
      * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
      * mpeg, mpga, m4a, ogg, wav, or webm.
      */
-    fun _file(): MultipartField<ByteArray> = body._file()
+    fun _file(): MultipartField<InputStream> = body._file()
 
     /**
      * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -110,7 +115,7 @@ private constructor(
     class Body
     @JsonCreator
     private constructor(
-        private val file: MultipartField<ByteArray>,
+        private val file: MultipartField<InputStream>,
         private val model: MultipartField<AudioModel>,
         private val prompt: MultipartField<String>,
         private val responseFormat: MultipartField<AudioResponseFormat>,
@@ -121,7 +126,7 @@ private constructor(
          * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
          * mpeg, mpga, m4a, ogg, wav, or webm.
          */
-        fun file(): ByteArray = file.value.getRequired("file")
+        fun file(): InputStream = file.value.getRequired("file")
 
         /**
          * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -157,7 +162,7 @@ private constructor(
          * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
          * mpeg, mpga, m4a, ogg, wav, or webm.
          */
-        fun _file(): MultipartField<ByteArray> = file
+        fun _file(): MultipartField<InputStream> = file
 
         /**
          * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2
@@ -221,7 +226,7 @@ private constructor(
         /** A builder for [Body]. */
         class Builder internal constructor() {
 
-            private var file: MultipartField<ByteArray>? = null
+            private var file: MultipartField<InputStream>? = null
             private var model: MultipartField<AudioModel>? = null
             private var prompt: MultipartField<String> = MultipartField.of(null)
             private var responseFormat: MultipartField<AudioResponseFormat> =
@@ -241,13 +246,31 @@ private constructor(
              * The audio file object (not file name) translate, in one of these formats: flac, mp3,
              * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
              */
-            fun file(file: ByteArray) = file(MultipartField.of(file))
+            fun file(file: InputStream) = file(MultipartField.of(file))
 
             /**
              * The audio file object (not file name) translate, in one of these formats: flac, mp3,
              * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
              */
-            fun file(file: MultipartField<ByteArray>) = apply { this.file = file }
+            fun file(file: MultipartField<InputStream>) = apply { this.file = file }
+
+            /**
+             * The audio file object (not file name) translate, in one of these formats: flac, mp3,
+             * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
+             */
+            fun file(file: ByteArray) = file(ByteArrayInputStream(file))
+
+            /**
+             * The audio file object (not file name) translate, in one of these formats: flac, mp3,
+             * mp4, mpeg, mpga, m4a, ogg, wav, or webm.
+             */
+            fun file(file: Path) =
+                file(
+                    MultipartField.builder<InputStream>()
+                        .value(file.inputStream())
+                        .filename(file.name)
+                        .build()
+                )
 
             /**
              * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper
@@ -375,6 +398,18 @@ private constructor(
             additionalQueryParams = audioTranslationCreateParams.additionalQueryParams.toBuilder()
         }
 
+        /**
+         * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
+         * mpeg, mpga, m4a, ogg, wav, or webm.
+         */
+        fun file(file: InputStream) = apply { body.file(file) }
+
+        /**
+         * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
+         * mpeg, mpga, m4a, ogg, wav, or webm.
+         */
+        fun file(file: MultipartField<InputStream>) = apply { body.file(file) }
+
         /**
          * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
          * mpeg, mpga, m4a, ogg, wav, or webm.
@@ -385,7 +420,7 @@ private constructor(
          * The audio file object (not file name) translate, in one of these formats: flac, mp3, mp4,
          * mpeg, mpga, m4a, ogg, wav, or webm.
          */
-        fun file(file: MultipartField<ByteArray>) = apply { body.file(file) }
+        fun file(file: Path) = apply { body.file(file) }
 
         /**
          * ID of the model to use. Only `whisper-1` (which is powered by our open source Whisper V2