Add list, symbol/annotation SID, single-char symbol opcode handlers

src/main/java/com/amazon/ion/bytecode/BytecodeEmitter.kt

@@ -87,4 +87,19 @@ internal object BytecodeEmitter {
     fun emitShortTimestampReference(destination: BytecodeBuffer, opcode: Int, dataPosition: Int) {
         destination.add2(Instructions.I_SHORT_TIMESTAMP_REF.packInstructionData(opcode), dataPosition)
     }
+
+    /**
+     * Writes a list to the destination, automatically handling the calculation of the generated bytecode
+     * length and reserving the space in the buffer for the list start instruction.
+     *
+     * @param contentWriter Callback function that should write children's bytecode to `destination`
+     */
+    @JvmStatic
+    inline fun emitList(destination: BytecodeBuffer, contentWriter: () -> Unit) {
+        val containerStartIndex = destination.reserve()
+        contentWriter()
+        destination.add(Instructions.I_END_CONTAINER)
+        val containerBytecodeSize = destination.size() - containerStartIndex - 1 // excludes the container start instruction
+        destination[containerStartIndex] = Instructions.I_LIST_START.packInstructionData(containerBytecodeSize)
+    }
 }

src/main/java/com/amazon/ion/bytecode/bin11/ByteArrayBytecodeGenerator11.kt

@@ -38,16 +38,21 @@ internal class ByteArrayBytecodeGenerator11(
         while (currentPosition < source.size && !isSystemValue(opcode)) {
             opcode = source[currentPosition++].unsignedToInt()
             val handler = OpcodeHandlerTable.handler(opcode)
-            currentPosition += handler.convertOpcodeToBytecode(
-                opcode,
-                source,
-                currentPosition,
-                destination,
-                constantPool,
-                macroSrc,
-                macroIndices,
-                symTab
-            )
+            try {
+                currentPosition += handler.convertOpcodeToBytecode(
+                    opcode,
+                    source,
+                    currentPosition,
+                    destination,
+                    constantPool,
+                    macroSrc,
+                    macroIndices,
+                    symTab
+                )
+            } catch (e: StackOverflowError) {
+                // TODO: implement recursion limit instead of catching StackOverflowError
+                throw IonException("Ion data nested too deeply", e)
+            }
         }
 
         if (currentPosition >= source.size) {

src/main/java/com/amazon/ion/bytecode/bin11/bytearray/AnnotationSIDOpcodeHandler.kt

@@ -0,0 +1,31 @@
+// Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+// SPDX-License-Identifier: Apache-2.0
+package com.amazon.ion.bytecode.bin11.bytearray
+
+import com.amazon.ion.bytecode.ir.Instructions
+import com.amazon.ion.bytecode.ir.Instructions.packInstructionData
+import com.amazon.ion.bytecode.util.AppendableConstantPoolView
+import com.amazon.ion.bytecode.util.BytecodeBuffer
+
+/**
+ * Writes an annotation with symbol address to the bytecode buffer. Handles opcode `0x58`.
+ */
+internal object AnnotationSIDOpcodeHandler : OpcodeToBytecodeHandler {
+    @OptIn(ExperimentalStdlibApi::class)
+    override fun convertOpcodeToBytecode(
+        opcode: Int,
+        source: ByteArray,
+        position: Int,
+        destination: BytecodeBuffer,
+        constantPool: AppendableConstantPoolView,
+        macroSrc: IntArray,
+        macroIndices: IntArray,
+        symbolTable: Array<String?>
+    ): Int {
+        val sidValueAndLength = PrimitiveDecoder.readFlexUIntValueAndLength(source, position)
+        val sid = sidValueAndLength.toInt()
+        val length = sidValueAndLength.shr(Int.SIZE_BITS).toInt()
+        destination.add(Instructions.I_ANNOTATION_SID.packInstructionData(sid))
+        return length
+    }
+}

src/main/java/com/amazon/ion/bytecode/bin11/bytearray/IntOpcodeHandlers.kt

@@ -8,6 +8,8 @@ import com.amazon.ion.bytecode.bin11.bytearray.PrimitiveDecoder.readFixedInt24As
 import com.amazon.ion.bytecode.bin11.bytearray.PrimitiveDecoder.readFixedInt32
 import com.amazon.ion.bytecode.bin11.bytearray.PrimitiveDecoder.readFixedInt8AsShort
 import com.amazon.ion.bytecode.bin11.bytearray.PrimitiveDecoder.readFixedIntAsLong
+import com.amazon.ion.bytecode.ir.Instructions
+import com.amazon.ion.bytecode.ir.Instructions.packInstructionData
 import com.amazon.ion.bytecode.util.AppendableConstantPoolView
 import com.amazon.ion.bytecode.util.BytecodeBuffer
 
@@ -143,3 +145,31 @@ internal object LongIntOpcodeHandler : OpcodeToBytecodeHandler {
         return fixedIntLength
     }
 }
+
+/** Writes a variable-length integer in a tagless context to the bytecode buffer. Handles tagless opcode `0x60`.
+ * For simplicity this only ever emits `I_INT_I32`, `I_INT_I64`, and `I_INT_CP` bytecode, even if the integer could fit
+ * in `I_INT_I16`.
+ * */
+@OptIn(ExperimentalStdlibApi::class)
+internal val TAGLESS_FLEX_INT = OpcodeToBytecodeHandler { opcode, src, pos, dest, cp, _, _, _ ->
+    val flexIntLength = PrimitiveDecoder.lengthOfFlexIntOrUIntAt(src, pos)
+    when (flexIntLength) {
+        // TODO(perf): See if there's any performance benefit to having a separate case for length=1|2 and using INT_I16 instruction
+        1, 2, 3, 4 -> {
+            val valueAndLength = PrimitiveDecoder.readFlexIntValueAndLength(src, pos)
+            val value = valueAndLength.toInt()
+            dest.add2(Instructions.I_INT_I32, value)
+        }
+        5, 6, 7, 8, 9 -> {
+            val longValue = PrimitiveDecoder.readFlexIntAsLong(src, pos)
+            BytecodeEmitter.emitInt64Value(dest, longValue)
+        }
+        else -> {
+            val bigInt = PrimitiveDecoder.readFlexIntAsBigInteger(src, pos)
+            val cpIndex = cp.size
+            cp.add(bigInt)
+            dest.add(Instructions.I_INT_CP.packInstructionData(cpIndex))
+        }
+    }
+    flexIntLength
+}

src/main/java/com/amazon/ion/bytecode/bin11/bytearray/ListOpcodeHandlers.kt

@@ -0,0 +1,197 @@
+// Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+// SPDX-License-Identifier: Apache-2.0
+package com.amazon.ion.bytecode.bin11.bytearray
+
+import com.amazon.ion.bytecode.BytecodeEmitter
+import com.amazon.ion.bytecode.bin11.OpCode
+import com.amazon.ion.bytecode.util.AppendableConstantPoolView
+import com.amazon.ion.bytecode.util.BytecodeBuffer
+import com.amazon.ion.bytecode.util.unsignedToInt
+
+// TODO: much of the logic here is shared between lists and sexps. It might be worthwhile to do something like
+//   "SequenceOpcodeHandlers" and pass the start instruction (Instructions.I_LIST_START vs .I_SEXP_START) to a helper
+//   BytecodeEmitter.emitSequence() or similar so this logic is not duplicated in a set of `*SexpOpcodeHandler`s.
+
+/**
+ * Writes a length prefixed list to the bytecode buffer. Handles opcode `0xB0`-`0xBF`.
+ */
+internal object ShortLengthPrefixedListOpcodeHandler : OpcodeToBytecodeHandler {
+    @OptIn(ExperimentalStdlibApi::class)
+    override fun convertOpcodeToBytecode(
+        opcode: Int,
+        source: ByteArray,
+        position: Int,
+        destination: BytecodeBuffer,
+        constantPool: AppendableConstantPoolView,
+        macroSrc: IntArray,
+        macroIndices: IntArray,
+        symbolTable: Array<String?>
+    ): Int {
+        val length = opcode and 0xF
+        BytecodeEmitter.emitList(destination) {
+            var p = position
+            val end = p + length
+            while (p < end) {
+                val opcode = source[p++].unsignedToInt()
+                p += OpcodeHandlerTable.handler(opcode).convertOpcodeToBytecode(
+                    opcode,
+                    source,
+                    p,
+                    destination,
+                    constantPool,
+                    macroSrc,
+                    macroIndices,
+                    symbolTable,
+                )
+            }
+        }
+        return length
+    }
+}
+
+/**
+ * Writes a length prefixed list to the bytecode buffer. Handles opcode `0xFA`.
+ */
+internal object LongLengthPrefixedListOpcodeHandler : OpcodeToBytecodeHandler {
+    @OptIn(ExperimentalStdlibApi::class)
+    override fun convertOpcodeToBytecode(
+        opcode: Int,
+        source: ByteArray,
+        position: Int,
+        destination: BytecodeBuffer,
+        constantPool: AppendableConstantPoolView,
+        macroSrc: IntArray,
+        macroIndices: IntArray,
+        symbolTable: Array<String?>
+    ): Int {
+        val containerSizeUIntValueAndLength = PrimitiveDecoder.readFlexUIntValueAndLength(source, position)
+        val containerLength = containerSizeUIntValueAndLength.toInt()
+        val prefixLength = containerSizeUIntValueAndLength.shr(Int.SIZE_BITS).toInt()
+        BytecodeEmitter.emitList(destination) {
+            var p = position + prefixLength
+            val end = p + containerLength
+            while (p < end) {
+                val opcode = source[p++].unsignedToInt()
+                p += OpcodeHandlerTable.handler(opcode).convertOpcodeToBytecode(
+                    opcode,
+                    source,
+                    p,
+                    destination,
+                    constantPool,
+                    macroSrc,
+                    macroIndices,
+                    symbolTable,
+                )
+            }
+        }
+        return containerLength + prefixLength
+    }
+}
+
+/**
+ * Writes a delimited list to the bytecode buffer. Handles opcode `0xF0`.
+ */
+internal object DelimitedListOpcodeHandler : OpcodeToBytecodeHandler {
+    @OptIn(ExperimentalStdlibApi::class)
+    override fun convertOpcodeToBytecode(
+        opcode: Int,
+        source: ByteArray,
+        position: Int,
+        destination: BytecodeBuffer,
+        constantPool: AppendableConstantPoolView,
+        macroSrc: IntArray,
+        macroIndices: IntArray,
+        symbolTable: Array<String?>
+    ): Int {
+        var p = position
+        BytecodeEmitter.emitList(destination) {
+            while (true) {
+                val opcode = source[p++].unsignedToInt()
+                if (opcode == OpCode.DELIMITED_CONTAINER_END) {
+                    break
+                }
+                p += OpcodeHandlerTable.handler(opcode).convertOpcodeToBytecode(
+                    opcode,
+                    source,
+                    p,
+                    destination,
+                    constantPool,
+                    macroSrc,
+                    macroIndices,
+                    symbolTable,
+                )
+            }
+        }
+        val bytesRead = p - position
+        return bytesRead
+    }
+}
+
+/**
+ * Writes a tagless-element list to the bytecode buffer. Handles opcode `0x5B`.
+ */
+internal object TaglessElementListOpcodeHandler : OpcodeToBytecodeHandler {
+    @OptIn(ExperimentalStdlibApi::class)
+    override fun convertOpcodeToBytecode(
+        opcode: Int,
+        source: ByteArray,
+        position: Int,
+        destination: BytecodeBuffer,
+        constantPool: AppendableConstantPoolView,
+        macroSrc: IntArray,
+        macroIndices: IntArray,
+        symbolTable: Array<String?>
+    ): Int {
+        var p = position
+        val childOpcode = source[p++].unsignedToInt()
+        val macroAddress = when (childOpcode) {
+            in 0x00..0x47 -> childOpcode
+            in 0x48..0x4f -> {
+                val flexUIntValueAndLength = PrimitiveDecoder.readFlexUIntValueAndLength(source, p)
+                val addressLength = flexUIntValueAndLength.shr(Int.SIZE_BITS).toInt()
+                p += addressLength
+                val lsb = childOpcode - 0x48
+                val msb = flexUIntValueAndLength.toInt() * 8
+                msb + lsb + 72
+            }
+            0xf4 -> {
+                val addressValueAndLength = PrimitiveDecoder.readFlexUIntValueAndLength(source, p)
+                val addressValue = addressValueAndLength.toInt()
+                val addressLength = addressValueAndLength.shr(Int.SIZE_BITS).toInt()
+                p += addressLength
+                addressValue
+            }
+            else -> -1
+        }
+
+        val childCountValueAndLength = PrimitiveDecoder.readFlexUIntValueAndLength(source, p)
+        val childCount = childCountValueAndLength.toInt()
+        val prefixSize = childCountValueAndLength.shr(Int.SIZE_BITS).toInt()
+        p += prefixSize
+
+        // If macroAddress > -1, then it is the address of the macro-shaped values,
+        // and childOpcode should be ignored.
+        // If macroAddress is -1, then childOpcode is the opcode of the values.
+        if (macroAddress < 0) {
+            val handler = TaglessOpcodeHandlerTable.handler(childOpcode)
+            BytecodeEmitter.emitList(destination) {
+                for (i in 0 until childCount) {
+                    p += handler.convertOpcodeToBytecode(
+                        childOpcode,
+                        source,
+                        p,
+                        destination,
+                        constantPool,
+                        macroSrc,
+                        macroIndices,
+                        symbolTable,
+                    )
+                }
+            }
+        } else {
+            TODO("Macro evaluation not yet implemented")
+        }
+
+        return p - position
+    }
+}