326 lines
16 KiB
C#
326 lines
16 KiB
C#
namespace STCompiler.Compiler;
|
|
|
|
using System;
|
|
using System.IO;
|
|
using System.Text;
|
|
using System.Collections.Generic;
|
|
using STCompiler.Common;
|
|
|
|
// === ENTRY POINT ===
|
|
class Program {
|
|
static int Main(string[] args) {
|
|
if (args.Length < 2) {
|
|
Console.WriteLine("Usage: StEmitter <input.st> <output.bin>");
|
|
return 1;
|
|
}
|
|
|
|
var input = File.ReadAllText(args[0]);
|
|
var parser = new StParser(input);
|
|
var prog = parser.ParseProgram();
|
|
|
|
if (parser.HasErrors) {
|
|
Console.WriteLine($"Compilation failed with {parser.Errors.Count} errors:");
|
|
foreach (var error in parser.Errors) {
|
|
Console.WriteLine($"Error at line {error.Line}: {error.Message}");
|
|
}
|
|
return 1;
|
|
}
|
|
|
|
if (prog == null) {
|
|
Console.WriteLine("Compilation failed: invalid program structure");
|
|
return 1;
|
|
}
|
|
|
|
try {
|
|
var emitter = new BytecodeEmitter();
|
|
emitter.Compile(prog);
|
|
File.WriteAllBytes(args[1], emitter.BuildBinary());
|
|
Console.WriteLine($"Wrote {args[1]}: consts={emitter.ConstantsCount}, vars={emitter.VarCount}, code={emitter.CodeLength}");
|
|
return 0;
|
|
} catch (Exception ex) {
|
|
Console.WriteLine($"Internal compiler error: {ex.Message}");
|
|
return 1;
|
|
}
|
|
}
|
|
}
|
|
|
|
// Parser, Lexer und AST wurden ausgelagert nach Parser.cs, Lexer.cs und Ast.cs
|
|
// Die Implementierungen dort werden vom Entry-Point genutzt.
|
|
|
|
// === BYTECODE ===
|
|
public class BytecodeEmitter {
|
|
List<object> consts = new(); // Kann nun int, long, float oder double speichern
|
|
Dictionary<string,Symbol> syms = new();
|
|
List<byte> code = new();
|
|
|
|
class Symbol {
|
|
required public string Name;
|
|
public VarType Type;
|
|
public int Index;
|
|
}
|
|
|
|
public int ConstantsCount => consts.Count;
|
|
public int VarCount => syms.Count;
|
|
public int CodeLength => code.Count;
|
|
|
|
public void Compile(ProgramNode p){
|
|
int idx=0;
|
|
foreach(var v in p.Vars)
|
|
syms[v.Name]=new Symbol{Name=v.Name,Type=v.Type,Index=idx++};
|
|
|
|
foreach(var v in p.Vars)
|
|
if(v.Init!=null){EmitExpr(v.Init);EmitByte(Bytecode.OpCodes.STORE_VAR);EmitU16((ushort)syms[v.Name].Index);}
|
|
|
|
foreach(var s in p.Stmts)
|
|
EmitStmt(s);
|
|
|
|
EmitByte(Bytecode.OpCodes.HALT); // Program End
|
|
}
|
|
|
|
// Operationscodes für verschiedene Datentypen
|
|
Dictionary<(TokType, VarType), byte> opCodes = new() {
|
|
// Signed integer arithmetic
|
|
{(TokType.PLUS, VarType.SINT), Bytecode.OpCodes.ADD_SINT}, {(TokType.MINUS, VarType.SINT), Bytecode.OpCodes.SUB_SINT},
|
|
{(TokType.MUL, VarType.SINT), Bytecode.OpCodes.MUL_SINT}, {(TokType.DIV, VarType.SINT), Bytecode.OpCodes.DIV_SINT},
|
|
{(TokType.PLUS, VarType.INT), Bytecode.OpCodes.ADD_INT}, {(TokType.MINUS, VarType.INT), Bytecode.OpCodes.SUB_INT},
|
|
{(TokType.MUL, VarType.INT), Bytecode.OpCodes.MUL_INT}, {(TokType.DIV, VarType.INT), Bytecode.OpCodes.DIV_INT},
|
|
{(TokType.PLUS, VarType.DINT), Bytecode.OpCodes.ADD_DINT}, {(TokType.MINUS, VarType.DINT), Bytecode.OpCodes.SUB_DINT},
|
|
{(TokType.MUL, VarType.DINT), Bytecode.OpCodes.MUL_DINT}, {(TokType.DIV, VarType.DINT), Bytecode.OpCodes.DIV_DINT},
|
|
{(TokType.PLUS, VarType.LINT), Bytecode.OpCodes.ADD_LINT}, {(TokType.MINUS, VarType.LINT), Bytecode.OpCodes.SUB_LINT},
|
|
{(TokType.MUL, VarType.LINT), Bytecode.OpCodes.MUL_LINT}, {(TokType.DIV, VarType.LINT), Bytecode.OpCodes.DIV_LINT},
|
|
|
|
// Unsigned integer arithmetic
|
|
{(TokType.PLUS, VarType.USINT), Bytecode.OpCodes.ADD_USINT}, {(TokType.MINUS, VarType.USINT), Bytecode.OpCodes.SUB_USINT},
|
|
{(TokType.MUL, VarType.USINT), Bytecode.OpCodes.MUL_USINT}, {(TokType.DIV, VarType.USINT), Bytecode.OpCodes.DIV_USINT},
|
|
{(TokType.PLUS, VarType.UINT), Bytecode.OpCodes.ADD_UINT}, {(TokType.MINUS, VarType.UINT), Bytecode.OpCodes.SUB_UINT},
|
|
{(TokType.MUL, VarType.UINT), Bytecode.OpCodes.MUL_UINT}, {(TokType.DIV, VarType.UINT), Bytecode.OpCodes.DIV_UINT},
|
|
{(TokType.PLUS, VarType.UDINT), Bytecode.OpCodes.ADD_UDINT}, {(TokType.MINUS, VarType.UDINT), Bytecode.OpCodes.SUB_UDINT},
|
|
{(TokType.MUL, VarType.UDINT), Bytecode.OpCodes.MUL_UDINT}, {(TokType.DIV, VarType.UDINT), Bytecode.OpCodes.DIV_UDINT},
|
|
{(TokType.PLUS, VarType.ULINT), Bytecode.OpCodes.ADD_ULINT}, {(TokType.MINUS, VarType.ULINT), Bytecode.OpCodes.SUB_ULINT},
|
|
{(TokType.MUL, VarType.ULINT), Bytecode.OpCodes.MUL_ULINT}, {(TokType.DIV, VarType.ULINT), Bytecode.OpCodes.DIV_ULINT},
|
|
|
|
// Floating point arithmetic
|
|
{(TokType.PLUS, VarType.REAL), Bytecode.OpCodes.ADD_REAL}, {(TokType.MINUS, VarType.REAL), Bytecode.OpCodes.SUB_REAL},
|
|
{(TokType.MUL, VarType.REAL), Bytecode.OpCodes.MUL_REAL}, {(TokType.DIV, VarType.REAL), Bytecode.OpCodes.DIV_REAL},
|
|
{(TokType.PLUS, VarType.LREAL), Bytecode.OpCodes.ADD_LREAL}, {(TokType.MINUS, VarType.LREAL), Bytecode.OpCodes.SUB_LREAL},
|
|
{(TokType.MUL, VarType.LREAL), Bytecode.OpCodes.MUL_LREAL}, {(TokType.DIV, VarType.LREAL), Bytecode.OpCodes.DIV_LREAL},
|
|
|
|
// Comparisons signed
|
|
{(TokType.LT, VarType.SINT), Bytecode.OpCodes.LT_S}, {(TokType.GT, VarType.SINT), Bytecode.OpCodes.GT_S},
|
|
{(TokType.LE, VarType.SINT), Bytecode.OpCodes.LE_S}, {(TokType.GE, VarType.SINT), Bytecode.OpCodes.GE_S},
|
|
{(TokType.EQ, VarType.SINT), Bytecode.OpCodes.EQ_S}, {(TokType.NEQ, VarType.SINT), Bytecode.OpCodes.NEQ_S},
|
|
{(TokType.LT, VarType.INT), Bytecode.OpCodes.LT_S}, {(TokType.GT, VarType.INT), Bytecode.OpCodes.GT_S},
|
|
{(TokType.LE, VarType.INT), Bytecode.OpCodes.LE_S}, {(TokType.GE, VarType.INT), Bytecode.OpCodes.GE_S},
|
|
{(TokType.EQ, VarType.INT), Bytecode.OpCodes.EQ_S}, {(TokType.NEQ, VarType.INT), Bytecode.OpCodes.NEQ_S},
|
|
{(TokType.LT, VarType.DINT), Bytecode.OpCodes.LT_S}, {(TokType.GT, VarType.DINT), Bytecode.OpCodes.GT_S},
|
|
{(TokType.LE, VarType.DINT), Bytecode.OpCodes.LE_S}, {(TokType.GE, VarType.DINT), Bytecode.OpCodes.GE_S},
|
|
{(TokType.EQ, VarType.DINT), Bytecode.OpCodes.EQ_S}, {(TokType.NEQ, VarType.DINT), Bytecode.OpCodes.NEQ_S},
|
|
{(TokType.LT, VarType.LINT), Bytecode.OpCodes.LT_S}, {(TokType.GT, VarType.LINT), Bytecode.OpCodes.GT_S},
|
|
{(TokType.LE, VarType.LINT), Bytecode.OpCodes.LE_S}, {(TokType.GE, VarType.LINT), Bytecode.OpCodes.GE_S},
|
|
{(TokType.EQ, VarType.LINT), Bytecode.OpCodes.EQ_S}, {(TokType.NEQ, VarType.LINT), Bytecode.OpCodes.NEQ_S},
|
|
|
|
// Comparisons unsigned
|
|
{(TokType.LT, VarType.USINT), Bytecode.OpCodes.LT_U}, {(TokType.GT, VarType.USINT), Bytecode.OpCodes.GT_U},
|
|
{(TokType.LE, VarType.USINT), Bytecode.OpCodes.LE_U}, {(TokType.GE, VarType.USINT), Bytecode.OpCodes.GE_U},
|
|
{(TokType.EQ, VarType.USINT), Bytecode.OpCodes.EQ_U}, {(TokType.NEQ, VarType.USINT), Bytecode.OpCodes.NEQ_U},
|
|
{(TokType.LT, VarType.UINT), Bytecode.OpCodes.LT_U}, {(TokType.GT, VarType.UINT), Bytecode.OpCodes.GT_U},
|
|
{(TokType.LE, VarType.UINT), Bytecode.OpCodes.LE_U}, {(TokType.GE, VarType.UINT), Bytecode.OpCodes.GE_U},
|
|
{(TokType.EQ, VarType.UINT), Bytecode.OpCodes.EQ_U}, {(TokType.NEQ, VarType.UINT), Bytecode.OpCodes.NEQ_U},
|
|
{(TokType.LT, VarType.UDINT), Bytecode.OpCodes.LT_U}, {(TokType.GT, VarType.UDINT), Bytecode.OpCodes.GT_U},
|
|
{(TokType.LE, VarType.UDINT), Bytecode.OpCodes.LE_U}, {(TokType.GE, VarType.UDINT), Bytecode.OpCodes.GE_U},
|
|
{(TokType.EQ, VarType.UDINT), Bytecode.OpCodes.EQ_U}, {(TokType.NEQ, VarType.UDINT), Bytecode.OpCodes.NEQ_U},
|
|
{(TokType.LT, VarType.ULINT), Bytecode.OpCodes.LT_U}, {(TokType.GT, VarType.ULINT), Bytecode.OpCodes.GT_U},
|
|
{(TokType.LE, VarType.ULINT), Bytecode.OpCodes.LE_U}, {(TokType.GE, VarType.ULINT), Bytecode.OpCodes.GE_U},
|
|
{(TokType.EQ, VarType.ULINT), Bytecode.OpCodes.EQ_U}, {(TokType.NEQ, VarType.ULINT), Bytecode.OpCodes.NEQ_U},
|
|
|
|
// Comparisons floating
|
|
{(TokType.LT, VarType.REAL), Bytecode.OpCodes.LT_F}, {(TokType.GT, VarType.REAL), Bytecode.OpCodes.GT_F},
|
|
{(TokType.LE, VarType.REAL), Bytecode.OpCodes.LE_F}, {(TokType.GE, VarType.REAL), Bytecode.OpCodes.GE_F},
|
|
{(TokType.EQ, VarType.REAL), Bytecode.OpCodes.EQ_F}, {(TokType.NEQ, VarType.REAL), Bytecode.OpCodes.NEQ_F},
|
|
{(TokType.LT, VarType.LREAL), Bytecode.OpCodes.LT_F}, {(TokType.GT, VarType.LREAL), Bytecode.OpCodes.GT_F},
|
|
{(TokType.LE, VarType.LREAL), Bytecode.OpCodes.LE_F}, {(TokType.GE, VarType.LREAL), Bytecode.OpCodes.GE_F},
|
|
{(TokType.EQ, VarType.LREAL), Bytecode.OpCodes.EQ_F}, {(TokType.NEQ, VarType.LREAL), Bytecode.OpCodes.NEQ_F}
|
|
};
|
|
|
|
void EmitStmt(Stmt s){
|
|
switch(s){
|
|
case AssignStmt a:
|
|
Symbol? symbol = null;
|
|
if (!syms.TryGetValue(a.Target, out symbol)) {
|
|
throw new Exception($"Undeclared variable '{a.Target}'");
|
|
}
|
|
EmitExpr(a.Expr);
|
|
EmitByte(Bytecode.OpCodes.STORE_VAR); EmitU16((ushort)symbol.Index);
|
|
break;
|
|
|
|
case IfStmt iff:
|
|
EmitExpr(iff.Cond);
|
|
EmitByte(Bytecode.OpCodes.JZ); int jz=code.Count; EmitU16(0);
|
|
foreach(var st in iff.ThenStmts) EmitStmt(st);
|
|
if(iff.ElseStmts.Count>0){
|
|
EmitByte(Bytecode.OpCodes.JMP); int jmp=code.Count; EmitU16(0);
|
|
PatchJump(jz,code.Count);
|
|
foreach(var st in iff.ElseStmts) EmitStmt(st);
|
|
PatchJump(jmp,code.Count);
|
|
} else PatchJump(jz,code.Count);
|
|
break;
|
|
|
|
case WhileStmt w:
|
|
int loopStart=code.Count;
|
|
EmitExpr(w.Cond);
|
|
EmitByte(Bytecode.OpCodes.JZ); int jzpos=code.Count; EmitU16(0);
|
|
foreach(var st in w.Body) EmitStmt(st);
|
|
EmitByte(Bytecode.OpCodes.JMP); EmitU16((ushort)loopStart);
|
|
PatchJump(jzpos,code.Count);
|
|
break;
|
|
|
|
case ForStmt f:
|
|
Symbol? forSymbol = null;
|
|
if (!syms.TryGetValue(f.Var, out forSymbol)) {
|
|
throw new Exception($"Undeclared variable '{f.Var}'");
|
|
}
|
|
// Initialisierung: var := start
|
|
EmitExpr(f.Start);
|
|
EmitByte(Bytecode.OpCodes.STORE_VAR); EmitU16((ushort)forSymbol.Index);
|
|
|
|
int cmpPos = code.Count; // Position des Vergleichs
|
|
EmitExpr(new VarExpr(f.Var, forSymbol.Type));
|
|
EmitExpr(f.End);
|
|
var key = (TokType.LE, forSymbol.Type);
|
|
if (!opCodes.ContainsKey(key)) {
|
|
throw new Exception($"Comparison not supported for type {forSymbol.Type}");
|
|
}
|
|
EmitByte(opCodes[key]);
|
|
EmitByte(Bytecode.OpCodes.JZ); int jzFor = code.Count; EmitU16(0); // Jump zum Ende
|
|
|
|
// Body
|
|
foreach(var st in f.Body) EmitStmt(st);
|
|
|
|
// Inkrement: i := i + step
|
|
EmitExpr(new VarExpr(f.Var, forSymbol.Type));
|
|
EmitExpr(f.Step);
|
|
var addKey = (TokType.PLUS, forSymbol.Type);
|
|
if (!opCodes.ContainsKey(addKey)) {
|
|
throw new Exception($"Addition not supported for type {forSymbol.Type}");
|
|
}
|
|
EmitByte(opCodes[addKey]);
|
|
EmitByte(Bytecode.OpCodes.STORE_VAR); EmitU16((ushort)forSymbol.Index);
|
|
|
|
// Vergleich erneut, aber wir merken uns den Sprung zum Vergleich
|
|
EmitByte(Bytecode.OpCodes.JMP); EmitU16((ushort)cmpPos);
|
|
|
|
// Patch Jump: Ende der Schleife springt hierhin
|
|
PatchJump(jzFor, code.Count);
|
|
|
|
// Korrektur: Schleifenvariable auf Endwert, falls sie überläuft
|
|
EmitExpr(f.End);
|
|
EmitByte(Bytecode.OpCodes.STORE_VAR); EmitU16((ushort)syms[f.Var].Index);
|
|
break;
|
|
}
|
|
}
|
|
|
|
void PatchJump(int pos,int target){code[pos]=(byte)(target&0xFF);code[pos+1]=(byte)(target>>8);}
|
|
void EmitExpr(Expr e){
|
|
switch(e){
|
|
case IntExpr ie:
|
|
int ci = AddConst(ie.Value);
|
|
EmitByte(Bytecode.OpCodes.PUSH_CONST); // PUSH_CONST
|
|
EmitU16((ushort)ci);
|
|
// removed emitting the type byte into code stream; type is stored with constant table
|
|
break;
|
|
|
|
case RealExpr re:
|
|
int cri = AddConst(re.Value);
|
|
EmitByte(Bytecode.OpCodes.PUSH_REAL_CONST); // PUSH_REAL_CONST
|
|
EmitU16((ushort)cri);
|
|
// removed emitting the type byte into code stream; type is stored with constant table
|
|
break;
|
|
|
|
case VarExpr ve:
|
|
Symbol? symbol = null;
|
|
if (!syms.TryGetValue(ve.Name, out symbol)) {
|
|
throw new Exception($"Undeclared variable '{ve.Name}'");
|
|
}
|
|
EmitByte(Bytecode.OpCodes.LOAD_VAR); // LOAD_VAR
|
|
EmitU16((ushort)symbol.Index);
|
|
break;
|
|
|
|
case BinaryExpr be:
|
|
EmitExpr(be.L);
|
|
EmitExpr(be.R);
|
|
|
|
var key = (be.Op, be.Type);
|
|
if (!opCodes.ContainsKey(key)) {
|
|
throw new Exception($"Unknown operator '{be.Op}' for type {be.Type}");
|
|
}
|
|
|
|
// Wenn nötig, Typenkonvertierung vor der Operation
|
|
// Conversion opcodes removed - runtime will handle type differences using constant type markers and operation opcodes
|
|
// if (be.L.Type != be.Type)
|
|
// EmitByte((byte)(0x50 + (int)be.Type)); // CONVERT_TO_*
|
|
// if (be.R.Type != be.Type)
|
|
// EmitByte((byte)(0x50 + (int)be.Type)); // CONVERT_TO_*
|
|
|
|
EmitByte(opCodes[key]);
|
|
break;
|
|
}
|
|
}
|
|
|
|
int AddConst(object v) {
|
|
int i = consts.FindIndex(c => c.Equals(v));
|
|
if (i >= 0) return i;
|
|
consts.Add(v);
|
|
return consts.Count - 1;
|
|
}
|
|
|
|
void EmitByte(byte b) => code.Add(b);
|
|
void EmitU16(ushort v) {code.Add((byte)(v&0xFF));code.Add((byte)(v>>8));}
|
|
|
|
public byte[] BuildBinary() {
|
|
using var ms = new MemoryStream();
|
|
var w = new BinaryWriter(ms);
|
|
|
|
// Header
|
|
w.Write(Encoding.ASCII.GetBytes(Bytecode.Magic));
|
|
w.Write(Bytecode.Version);
|
|
|
|
// Konstanten
|
|
w.Write((ushort)consts.Count);
|
|
foreach(var c in consts) {
|
|
if (c is long l) {
|
|
w.Write((byte)1); // Long type marker
|
|
w.Write(l);
|
|
}
|
|
else if (c is double d) {
|
|
w.Write((byte)2); // Double type marker
|
|
w.Write(d);
|
|
}
|
|
else if (c is float f) {
|
|
w.Write((byte)3); // Float type marker
|
|
w.Write(f);
|
|
}
|
|
else if (c is int i) {
|
|
w.Write((byte)4); // Int type marker
|
|
w.Write(i);
|
|
}
|
|
else {
|
|
throw new Exception($"Unsupported constant type: {c.GetType()}");
|
|
}
|
|
}
|
|
|
|
// Variablen
|
|
w.Write((ushort)syms.Count);
|
|
var types = new byte[syms.Count];
|
|
foreach(var kv in syms)
|
|
types[kv.Value.Index] = (byte)kv.Value.Type;
|
|
foreach(var b in types)
|
|
w.Write(b);
|
|
|
|
// Code
|
|
w.Write((ushort)code.Count);
|
|
w.Write(code.ToArray());
|
|
|
|
return ms.ToArray();
|
|
}
|
|
}
|