Scala Json Parsing (#100)

Co-authored-by: raulraja <raulraja@gmail.com>
Co-authored-by: Yago Cervantes <1420230+Yawolf@users.noreply.github.com>

core/src/commonMain/kotlin/com/xebia/functional/xef/AIError.kt

@@ -29,11 +29,7 @@ sealed interface AIError {
       "Prompt exceeds max token length: $promptTokens + $maxTokens = ${promptTokens + maxTokens}"
   }
 
-  data class JsonParsing(
-    val result: String,
-    val maxAttempts: Int,
-    val cause: IllegalArgumentException
-  ) : AIError {
+  data class JsonParsing(val result: String, val maxAttempts: Int, val cause: Throwable) : AIError {
     override val reason: String
       get() = "Failed to parse the JSON response after $maxAttempts attempts: $result"
   }

core/src/commonMain/kotlin/com/xebia/functional/xef/auto/DeserializerLLMAgent.kt

@@ -1,14 +1,21 @@
+@file:JvmMultifileClass
+@file:JvmName("Agent")
+
 package com.xebia.functional.xef.auto
 
+import arrow.core.nonFatalOrThrow
 import arrow.core.raise.catch
 import arrow.core.raise.ensureNotNull
 import com.xebia.functional.xef.AIError
 import com.xebia.functional.xef.auto.serialization.buildJsonSchema
 import com.xebia.functional.xef.llm.openai.LLMModel
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.append
+import kotlin.jvm.JvmMultifileClass
+import kotlin.jvm.JvmName
 import kotlinx.serialization.KSerializer
 import kotlinx.serialization.SerializationException
+import kotlinx.serialization.descriptors.SerialDescriptor
 import kotlinx.serialization.json.Json
 import kotlinx.serialization.serializer
 
@@ -99,14 +106,37 @@ suspend fun <A> AIScope.prompt(
   temperature: Double = 0.0,
   bringFromContext: Int = 10,
   minResponseTokens: Int = 500,
-): A {
-  val serializationConfig: SerializationConfig<A> =
-    SerializationConfig(
-      jsonSchema = buildJsonSchema(serializer.descriptor, false),
-      descriptor = serializer.descriptor,
-      deserializationStrategy = serializer
-    )
+): A =
+  prompt(
+    prompt,
+    serializer.descriptor,
+    { json.decodeFromString(serializer, it) },
+    maxDeserializationAttempts,
+    model,
+    user,
+    echo,
+    n,
+    temperature,
+    bringFromContext,
+    minResponseTokens
+  )
 
+@AiDsl
+@JvmName("promptWithSerializer")
+suspend fun <A> AIScope.prompt(
+  prompt: Prompt,
+  descriptor: SerialDescriptor,
+  serializer: (json: String) -> A,
+  maxDeserializationAttempts: Int = 5,
+  model: LLMModel = LLMModel.GPT_3_5_TURBO,
+  user: String = "testing",
+  echo: Boolean = false,
+  n: Int = 1,
+  temperature: Double = 0.0,
+  bringFromContext: Int = 10,
+  minResponseTokens: Int = 500,
+): A {
+  val jsonSchema = buildJsonSchema(descriptor, false)
   val responseInstructions =
     """
         |
@@ -116,12 +146,12 @@ suspend fun <A> AIScope.prompt(
         |3. Use the JSON schema to produce the result exclusively in valid JSON format.
         |4. Pay attention to required vs non-required fields in the schema.
         |JSON Schema:
-        |${serializationConfig.jsonSchema}
+        |$jsonSchema
         |Response:
         """
       .trimMargin()
 
-  return tryDeserialize(serializationConfig, json, maxDeserializationAttempts) {
+  return tryDeserialize(serializer, maxDeserializationAttempts) {
     promptMessage(
       prompt.append(responseInstructions),
       model,
@@ -136,8 +166,7 @@ suspend fun <A> AIScope.prompt(
 }
 
 suspend fun <A> AIScope.tryDeserialize(
-  serializationConfig: SerializationConfig<A>,
-  json: Json,
+  serializer: (json: String) -> A,
   maxDeserializationAttempts: Int,
   agent: AI<List<String>>
 ): A {
@@ -146,13 +175,10 @@ suspend fun <A> AIScope.tryDeserialize(
     currentAttempts++
     val result = ensureNotNull(agent().firstOrNull()) { AIError.NoResponse }
     catch({
-      return@tryDeserialize json.decodeFromString(
-        serializationConfig.deserializationStrategy,
-        result
-      )
-    }) { e: IllegalArgumentException ->
+      return@tryDeserialize serializer(result)
+    }) { e: Throwable ->
       if (currentAttempts == maxDeserializationAttempts)
-        raise(AIError.JsonParsing(result, maxDeserializationAttempts, e))
+        raise(AIError.JsonParsing(result, maxDeserializationAttempts, e.nonFatalOrThrow()))
       // else continue with the next attempt
     }
   }

core/src/commonMain/kotlin/com/xebia/functional/xef/auto/ImageGenerationAgent.kt

@@ -1,9 +1,14 @@
+@file:JvmMultifileClass
+@file:JvmName("Agent")
+
 package com.xebia.functional.xef.auto
 
 import com.xebia.functional.xef.AIError
 import com.xebia.functional.xef.llm.openai.ImagesGenerationRequest
 import com.xebia.functional.xef.llm.openai.ImagesGenerationResponse
 import com.xebia.functional.xef.prompt.Prompt
+import kotlin.jvm.JvmMultifileClass
+import kotlin.jvm.JvmName
 
 /**
  * Run a [prompt] describes the images you want to generate within the context of [AIScope].

core/src/commonMain/kotlin/com/xebia/functional/xef/auto/LLMAgent.kt

@@ -1,3 +1,6 @@
+@file:JvmMultifileClass
+@file:JvmName("Agent")
+
 package com.xebia.functional.xef.auto
 
 import arrow.core.raise.Raise
@@ -14,6 +17,8 @@ import com.xebia.functional.xef.llm.openai.Role
 import com.xebia.functional.xef.prompt.Prompt
 import io.github.oshai.KLogger
 import io.github.oshai.KotlinLogging
+import kotlin.jvm.JvmMultifileClass
+import kotlin.jvm.JvmName
 
 private val logger: KLogger by lazy { KotlinLogging.logger {} }
 

gradle/libs.versions.toml

@@ -82,6 +82,7 @@ doobie-postgres = { module = "org.tpolecat:doobie-postgres_3", version.ref = "do
 doobie-hikari = { module = "org.tpolecat:doobie-hikari_3", version.ref = "doobie" }
 doobie-munit = { module = "org.tpolecat:doobie-munit_3", version.ref = "doobie" }
 circe = { module = "io.circe:circe-generic_3", version.ref = "circe" }
+circe-parser = { module = "io.circe:circe-parser_3", version.ref = "circe" }
 cats-effect = { module = "org.typelevel:cats-effect_3", version.ref = "catsEffect" }
 logger = { module = "org.typelevel:log4cats-slf4j_3", version.ref = "log4cats" }
 openai = { module = "com.theokanning.openai-gpt3-java:service", version.ref = "openai" }

scala/build.gradle.kts

@@ -16,6 +16,7 @@ dependencies {
     implementation(libs.http4s.dsl)
     implementation(libs.http4s.client)
     implementation(libs.http4s.circe)
+    implementation(libs.circe.parser)
     implementation(libs.http4s.emberClient)
     implementation(libs.doobie.core)
     implementation(libs.doobie.postgres)

scala/src/main/scala/com/xebia/functional/auto/AI.scala

@@ -1,17 +1,15 @@
 package com.xebia.functional.auto
 
 import com.xebia.functional.loom.LoomAdapter
+import com.xebia.functional.scala.auto.ScalaSerialDescriptor
 import com.xebia.functional.xef.auto.AIScope as KtAIScope
+import com.xebia.functional.xef.auto.Agent as KtAgent
 import com.xebia.functional.xef.auto.AIException
 import com.xebia.functional.xef.auto.AIKt
 import com.xebia.functional.xef.AIError
 import com.xebia.functional.xef.llm.openai.LLMModel
-
-//def example(using AIScope): String =
-//    prompt[String]("What is your name?")
-
-//val example: AIScope ?=> String =
-//  prompt[String]("What is your name?")
+import io.circe.{Decoder, Json}
+import io.circe.parser.parse
 
 object AI:
 
@@ -20,6 +18,7 @@ object AI:
       AIKt.AIScope[A](
         { (coreAIScope, cont) =>
           given AIScope = AIScope.fromCore(coreAIScope)
+
           block
         },
         (e: AIError, cont) => throw AIException(e.getReason),
@@ -29,20 +28,35 @@ object AI:
 
 end AI
 
-final case class AIScope(kt: KtAIScope):
-
-  // TODO: Design signature for Scala3 w/ Json parser (with support for generating Json Schema)?
-  def prompt[A](
-      prompt: String,
-      maxAttempts: Int = 5,
-      llmMode: LLMModel = LLMModel.getGPT_3_5_TURBO
-  ): A = ???
-
-  def promptMessage(
-      prompt: String,
-      maxAttempts: Int = 5,
-      llmMode: LLMModel = LLMModel.getGPT_3_5_TURBO
-  ): String = ???
-
+final case class AIScope(kt: KtAIScope)
 private object AIScope:
   def fromCore(coreAIScope: KtAIScope): AIScope = new AIScope(coreAIScope)
+
+def prompt[A: Decoder: ScalaSerialDescriptor](
+    prompt: String,
+    maxAttempts: Int = 5,
+    llmMode: LLMModel = LLMModel.getGPT_3_5_TURBO,
+    user: String = "testing",
+    echo: Boolean = false,
+    n: Int = 1,
+    temperature: Double = 0.0,
+    bringFromContext: Int = 10,
+    minResponseTokens: Int = 500
+)(using scope: AIScope): A =
+  LoomAdapter.apply((cont) =>
+    KtAgent.promptWithSerializer[A](
+      scope.kt,
+      prompt,
+      ScalaSerialDescriptor[A].serialDescriptor,
+      (json) => parse(json).flatMap(Decoder[A].decodeJson(_)).fold(throw _, identity),
+      maxAttempts,
+      llmMode,
+      user,
+      echo,
+      n,
+      temperature,
+      bringFromContext,
+      minResponseTokens,
+      cont
+    )
+  )