Merge pull request #106 from onion-lang/codex/bcelからasmへの置き換え

kmizu · web-flow · commit 23689b13c10d · 2025-05-22T05:25:08.000+09:00
Introduce experimental ASM backend
diff --git a/src/main/scala/onion/compiler/AsmCodeGeneration.scala b/src/main/scala/onion/compiler/AsmCodeGeneration.scala
@@ -0,0 +1,81 @@
+package onion.compiler
+
+import org.objectweb.asm.{ClassWriter, Opcodes, Type}
+import org.objectweb.asm.commons.GeneratorAdapter
+import org.objectweb.asm.commons.Method
+
+/**
+ * Experimental ASM-based bytecode generator. Only supports a very small
+ * subset of language features.  Unsupported constructs are ignored or
+ * generate empty stubs.
+ */
+class AsmCodeGeneration(config: CompilerConfig) extends BytecodeGenerator:
+  private def toAsmModifier(mod: Int): Int =
+    var access = 0
+    if Modifier.isPublic(mod) then access |= Opcodes.ACC_PUBLIC
+    if Modifier.isProtected(mod) then access |= Opcodes.ACC_PROTECTED
+    if Modifier.isPrivate(mod) then access |= Opcodes.ACC_PRIVATE
+    if Modifier.isStatic(mod) then access |= Opcodes.ACC_STATIC
+    if Modifier.isFinal(mod) then access |= Opcodes.ACC_FINAL
+    if Modifier.isAbstract(mod) then access |= Opcodes.ACC_ABSTRACT
+    access
+
+  private def asmType(tp: TypedAST.Type): Type = tp match
+    case TypedAST.BasicType.VOID    => Type.VOID_TYPE
+    case TypedAST.BasicType.BOOLEAN => Type.BOOLEAN_TYPE
+    case TypedAST.BasicType.BYTE    => Type.BYTE_TYPE
+    case TypedAST.BasicType.SHORT   => Type.SHORT_TYPE
+    case TypedAST.BasicType.CHAR    => Type.CHAR_TYPE
+    case TypedAST.BasicType.INT     => Type.INT_TYPE
+    case TypedAST.BasicType.LONG    => Type.LONG_TYPE
+    case TypedAST.BasicType.FLOAT   => Type.FLOAT_TYPE
+    case TypedAST.BasicType.DOUBLE  => Type.DOUBLE_TYPE
+    case c: TypedAST.ClassType      => Type.getObjectType(c.name.replace('.', '/'))
+    case a: TypedAST.ArrayType      => Type.getType("[" * a.dimension + asmType(a.component).getDescriptor)
+    case _                          => Type.VOID_TYPE
+
+  def process(classes: Seq[TypedAST.ClassDefinition]): Seq[CompiledClass] =
+    classes.map(codeClass)
+
+  private def codeClass(node: TypedAST.ClassDefinition): CompiledClass =
+    val cw = new ClassWriter(ClassWriter.COMPUTE_FRAMES | ClassWriter.COMPUTE_MAXS)
+    val superName =
+      if node.superClass != null then node.superClass.name.replace('.', '/')
+      else "java/lang/Object"
+    val interfaces =
+      if node.interfaces != null then node.interfaces.map(_.name.replace('.', '/')).toArray
+      else Array.empty[String]
+    cw.visit(Opcodes.V21, toAsmModifier(node.modifier) | (if node.isInterface then Opcodes.ACC_INTERFACE else 0),
+      node.name.replace('.', '/'), null, superName, interfaces)
+
+    for m <- node.methods do m match
+      case md: TypedAST.MethodDefinition => codeMethod(cw, md)
+      case _ =>
+    cw.visitEnd()
+    val bytes = cw.toByteArray
+    val dir = if config.outputDirectory != null then config.outputDirectory else ""
+    CompiledClass(node.name, dir, bytes)
+
+  private def codeMethod(cw: ClassWriter, node: TypedAST.MethodDefinition): Unit =
+    val access = toAsmModifier(node.modifier)
+    val desc = Method(node.name, asmType(node.returnType), node.arguments.map(asmType)).getDescriptor
+    val mv = cw.visitMethod(access, node.name, desc, null, null)
+    val gen = new GeneratorAdapter(mv, access, node.name, desc)
+    mv.visitCode()
+    node.block.statements match
+      case Array(ret: TypedAST.Return) =>
+        emitReturn(gen, ret.term, node.returnType)
+      case _ =>
+        gen.visitInsn(Opcodes.RETURN)
+    gen.endMethod()
+
+  private def emitReturn(gen: GeneratorAdapter, term: TypedAST.Term, tp: TypedAST.Type): Unit = term match
+    case v: TypedAST.IntValue =>
+      gen.push(v.value)
+      gen.returnValue()
+    case v: TypedAST.StringValue =>
+      gen.push(v.value)
+      gen.returnValue()
+    case _ =>
+      gen.visitInsn(Opcodes.RETURN)
+
diff --git a/src/main/scala/onion/compiler/BytecodeGenerator.scala b/src/main/scala/onion/compiler/BytecodeGenerator.scala
@@ -0,0 +1,4 @@
+package onion.compiler
+
+trait BytecodeGenerator:
+  def process(classes: Seq[TypedAST.ClassDefinition]): Seq[CompiledClass]
diff --git a/src/main/scala/onion/compiler/ClassTable.scala b/src/main/scala/onion/compiler/ClassTable.scala
@@ -9,6 +9,7 @@ package onion.compiler
 
 import java.util.{HashMap => JHashMap}
 import onion.compiler.environment.BcelRefs.BcelClassType
+import onion.compiler.environment.AsmRefs.AsmClassType
 import onion.compiler.environment.ClassFileTable
 import onion.compiler.environment.ReflectionRefs.ReflectClassType
 
@@ -39,12 +40,18 @@ class ClassTable(classPath: String) {
         clazz = new BcelClassType(javaClass, this)
         classFiles.put(clazz.name, clazz)
       } else {
-        try {
-          clazz = new ReflectClassType(Class.forName(className, true, Thread.currentThread.getContextClassLoader), this)
+        val bytes = table.loadBytes(className)
+        if (bytes != null) {
+          clazz = new AsmClassType(bytes, this)
           classFiles.put(clazz.name, clazz)
-        }
-        catch {
-          case e: ClassNotFoundException => {}
+        } else {
+          try {
+            clazz = new ReflectClassType(Class.forName(className, true, Thread.currentThread.getContextClassLoader), this)
+            classFiles.put(clazz.name, clazz)
+          }
+          catch {
+            case e: ClassNotFoundException => {}
+          }
         }
       }
     }
diff --git a/src/main/scala/onion/compiler/CodeGeneration.scala b/src/main/scala/onion/compiler/CodeGeneration.scala
@@ -307,10 +307,10 @@ object CodeGeneration {
   }
 }
 
-class CodeGeneration(config: CompilerConfig) {
+class CodeGeneration(config: CompilerConfig) extends BytecodeGenerator {
   import CodeGeneration._
 
-  def process(classes: Seq[IRT.ClassDefinition]): Seq[CompiledClass] = {
+  def process(classes: Seq[TypedAST.ClassDefinition]): Seq[CompiledClass] = {
     compiledClasses.clear
     val base =  (if (config.outputDirectory != null) config.outputDirectory else ".") + Systems.fileSeparator
     for (klass <- classes) codeClass(klass)
diff --git a/src/main/scala/onion/compiler/TypedGenerating.scala b/src/main/scala/onion/compiler/TypedGenerating.scala
@@ -13,7 +13,11 @@ class TypedGenerating(config: CompilerConfig)
   def newEnvironment(source: Seq[TypedAST.ClassDefinition]): TypedGeneratingEnvironment =
     new TypedGeneratingEnvironment
 
-  private val generator = new CodeGeneration(config)
+  private val generator: BytecodeGenerator =
+    if sys.props.get("onion.asm").exists(_.toBoolean) then
+      new AsmCodeGeneration(config)
+    else
+      new CodeGeneration(config)
 
   def processBody(source: Seq[TypedAST.ClassDefinition], environment: TypedGeneratingEnvironment): Seq[CompiledClass] =
-    generator.process(source.asInstanceOf[Seq[IRT.ClassDefinition]])
+    generator.process(source)
diff --git a/src/main/scala/onion/compiler/environment/AsmRefs.scala b/src/main/scala/onion/compiler/environment/AsmRefs.scala
@@ -0,0 +1,101 @@
+package onion.compiler.environment
+
+import onion.compiler.{IRT, Modifier, OnionTypeConversion, MultiTable, OrderedTable, ClassTable}
+import org.objectweb.asm.{ClassReader, Opcodes}
+import org.objectweb.asm.tree.{ClassNode, MethodNode, FieldNode}
+import org.apache.bcel.generic.{Type => BType}
+
+object AsmRefs {
+  private def toOnionModifier(access: Int): Int = {
+    var mod = 0
+    if ((access & Opcodes.ACC_PRIVATE) != 0) mod |= Modifier.PRIVATE
+    if ((access & Opcodes.ACC_PROTECTED) != 0) mod |= Modifier.PROTECTED
+    if ((access & Opcodes.ACC_PUBLIC) != 0) mod |= Modifier.PUBLIC
+    if ((access & Opcodes.ACC_STATIC) != 0) mod |= Modifier.STATIC
+    if ((access & Opcodes.ACC_SYNCHRONIZED) != 0) mod |= Modifier.SYNCHRONIZED
+    if ((access & Opcodes.ACC_ABSTRACT) != 0) mod |= Modifier.ABSTRACT
+    if ((access & Opcodes.ACC_FINAL) != 0) mod |= Modifier.FINAL
+    mod
+  }
+
+  final val CONSTRUCTOR_NAME = "<init>"
+
+  class AsmMethodRef(method: MethodNode, override val affiliation: IRT.ClassType, bridge: OnionTypeConversion) extends IRT.Method {
+    override val modifier: Int = toOnionModifier(method.access)
+    override val name: String = method.name
+    private val argTypes = org.objectweb.asm.Type.getArgumentTypes(method.desc).map(t => bridge.toOnionType(BType.getType(t.getDescriptor)))
+    override def arguments: Array[IRT.Type] = argTypes.clone()
+    override val returnType: IRT.Type = bridge.toOnionType(BType.getType(org.objectweb.asm.Type.getReturnType(method.desc).getDescriptor))
+    val underlying: MethodNode = method
+  }
+
+  class AsmFieldRef(field: FieldNode, override val affiliation: IRT.ClassType, bridge: OnionTypeConversion) extends IRT.FieldRef {
+    override val modifier: Int = toOnionModifier(field.access)
+    override val name: String = field.name
+    override val `type`: IRT.Type = bridge.toOnionType(BType.getType(field.desc))
+    val underlying: FieldNode = field
+  }
+
+  class AsmConstructorRef(method: MethodNode, override val affiliation: IRT.ClassType, bridge: OnionTypeConversion) extends IRT.ConstructorRef {
+    override val modifier: Int = toOnionModifier(method.access)
+    override val name: String = CONSTRUCTOR_NAME
+    private val args0 = org.objectweb.asm.Type.getArgumentTypes(method.desc).map(t => bridge.toOnionType(BType.getType(t.getDescriptor)))
+    override def getArgs: Array[IRT.Type] = args0.clone()
+    val underlying: MethodNode = method
+  }
+
+  class AsmClassType(classBytes: Array[Byte], table: ClassTable) extends IRT.AbstractClassType {
+    private val node = {
+      val cr = new ClassReader(classBytes)
+      val n = new ClassNode()
+      cr.accept(n, ClassReader.SKIP_DEBUG | ClassReader.SKIP_FRAMES)
+      n
+    }
+
+    private val bridge = new OnionTypeConversion(table)
+    private val modifier_ = toOnionModifier(node.access)
+
+    private lazy val methods_ : MultiTable[IRT.Method] = {
+      val m = new MultiTable[IRT.Method]
+      import scala.jdk.CollectionConverters._
+      for (method <- node.methods.asInstanceOf[java.util.List[MethodNode]].asScala if method.name != CONSTRUCTOR_NAME) {
+        m.add(new AsmMethodRef(method, this, bridge))
+      }
+      m
+    }
+
+    private lazy val fields_ : OrderedTable[IRT.FieldRef] = {
+      val f = new OrderedTable[IRT.FieldRef]
+      import scala.jdk.CollectionConverters._
+      for (field <- node.fields.asInstanceOf[java.util.List[FieldNode]].asScala) {
+        f.add(new AsmFieldRef(field, this, bridge))
+      }
+      f
+    }
+
+    private lazy val constructors_ : Seq[IRT.ConstructorRef] = {
+      import scala.jdk.CollectionConverters._
+      node.methods.asInstanceOf[java.util.List[MethodNode]].asScala.collect {
+        case m if m.name == CONSTRUCTOR_NAME => new AsmConstructorRef(m, this, bridge)
+      }.toSeq
+    }
+
+    def isInterface: Boolean = (node.access & Opcodes.ACC_INTERFACE) != 0
+    def modifier: Int = modifier_
+    def name: String = node.name.replace('/', '.')
+    def superClass: IRT.ClassType = {
+      if (node.superName == null) null else table.load(node.superName.replace('/', '.'))
+    }
+    def interfaces: Seq[IRT.ClassType] = {
+      import scala.jdk.CollectionConverters._
+      node.interfaces.asInstanceOf[java.util.List[String]].asScala.map(n => table.load(n.replace('/', '.'))).toIndexedSeq
+    }
+
+    def methods: Seq[IRT.Method] = methods_.values
+    def methods(name: String): Array[IRT.Method] = methods_.get(name).toArray
+    def fields: Array[IRT.FieldRef] = fields_.values.toArray
+    def field(name: String): IRT.FieldRef = fields_.get(name).orNull
+    def constructors: Array[IRT.ConstructorRef] = constructors_.toArray
+  }
+}
+
diff --git a/src/main/scala/onion/compiler/environment/ClassFileTable.scala b/src/main/scala/onion/compiler/environment/ClassFileTable.scala
@@ -36,6 +36,24 @@ class ClassFileTable(classPathString: String) {
     }
   }
 
+  def loadBytes(className: String): Array[Byte] = {
+    try {
+      val classFile: ClassPath.ClassFile = classPath.getClassFile(className)
+      val in = classFile.getInputStream
+      val out = new ByteArrayOutputStream()
+      val buf = new Array[Byte](8192)
+      var len = in.read(buf)
+      while (len != -1) {
+        out.write(buf, 0, len)
+        len = in.read(buf)
+      }
+      in.close()
+      out.toByteArray
+    } catch {
+      case _: IOException => null
+    }
+  }
+
   private def add(className: String): JavaClass = {
     try {
       val classFile: ClassPath.ClassFile = classPath.getClassFile(className)

Original file line number	Diff line number	Diff line change
`@@ -307,10 +307,10 @@ object CodeGeneration {`
`307`	`307`	`}`
`308`	`308`	`}`
`309`	`309`
`310`		`-class CodeGeneration(config: CompilerConfig) {`
	`310`	`+class CodeGeneration(config: CompilerConfig) extends BytecodeGenerator {`
`311`	`311`	`import CodeGeneration._`
`312`	`312`
`313`		`- def process(classes: Seq[IRT.ClassDefinition]): Seq[CompiledClass] = {`
	`313`	`+ def process(classes: Seq[TypedAST.ClassDefinition]): Seq[CompiledClass] = {`
`314`	`314`	`compiledClasses.clear`
`315`	`315`	`val base = (if (config.outputDirectory != null) config.outputDirectory else ".") + Systems.fileSeparator`
`316`	`316`	`for (klass <- classes) codeClass(klass)`