Mako/compiler/compiler.go
2025-06-11 17:43:17 -05:00

539 lines
14 KiB
Go

package compiler
import (
"fmt"
"git.sharkk.net/Sharkk/Mako/parser"
)
// Compiler holds the compilation state and compiles AST to bytecode
type Compiler struct {
current *CompilerState // Current compilation state
enclosing *CompilerState // Enclosing function state for closures
errors []CompileError // Compilation errors
}
// NewCompiler creates a new compiler instance
func NewCompiler() *Compiler {
return &Compiler{
current: NewCompilerState(FunctionTypeScript),
errors: make([]CompileError, 0),
}
}
// Compile compiles a program AST to bytecode
func (c *Compiler) Compile(program *parser.Program) (*Chunk, []CompileError) {
for _, stmt := range program.Statements {
c.compileStatement(stmt)
}
c.current.EmitInstruction(OpReturnNil)
if len(c.errors) > 0 {
return nil, c.errors
}
return c.current.Chunk, nil
}
// Statement compilation
func (c *Compiler) compileStatement(stmt parser.Statement) {
switch s := stmt.(type) {
case *parser.StructStatement:
c.compileStructStatement(s)
case *parser.MethodDefinition:
c.compileMethodDefinition(s)
case *parser.Assignment:
c.compileAssignment(s)
case *parser.ExpressionStatement:
c.compileExpression(s.Expression)
c.current.EmitInstruction(OpPop) // Discard result
case *parser.EchoStatement:
c.compileExpression(s.Value)
c.current.EmitInstruction(OpEcho)
case *parser.IfStatement:
c.compileIfStatement(s)
case *parser.WhileStatement:
c.compileWhileStatement(s)
case *parser.ForStatement:
c.compileForStatement(s)
case *parser.ForInStatement:
c.compileForInStatement(s)
case *parser.ReturnStatement:
c.compileReturnStatement(s)
case *parser.ExitStatement:
c.compileExitStatement(s)
case *parser.BreakStatement:
c.current.EmitBreak()
default:
c.addError(fmt.Sprintf("unknown statement type: %T", stmt))
}
}
// Expression compilation
func (c *Compiler) compileExpression(expr parser.Expression) {
switch e := expr.(type) {
case *parser.Identifier:
c.compileIdentifier(e)
case *parser.NumberLiteral:
c.compileNumberLiteral(e)
case *parser.StringLiteral:
c.compileStringLiteral(e)
case *parser.BooleanLiteral:
c.compileBooleanLiteral(e)
case *parser.NilLiteral:
c.compileNilLiteral(e)
case *parser.TableLiteral:
c.compileTableLiteral(e)
case *parser.StructConstructor:
c.compileStructConstructor(e)
case *parser.FunctionLiteral:
c.compileFunctionLiteral(e)
case *parser.CallExpression:
c.compileCallExpression(e)
case *parser.PrefixExpression:
c.compilePrefixExpression(e)
case *parser.InfixExpression:
c.compileInfixExpression(e)
case *parser.IndexExpression:
c.compileIndexExpression(e)
case *parser.DotExpression:
c.compileDotExpression(e)
case *parser.Assignment:
c.compileAssignmentExpression(e)
default:
c.addError(fmt.Sprintf("unknown expression type: %T", expr))
}
}
// Literal compilation
func (c *Compiler) compileNumberLiteral(node *parser.NumberLiteral) {
value := Value{Type: ValueNumber, Data: node.Value}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpLoadConst, uint16(index))
}
func (c *Compiler) compileStringLiteral(node *parser.StringLiteral) {
value := Value{Type: ValueString, Data: node.Value}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpLoadConst, uint16(index))
}
func (c *Compiler) compileBooleanLiteral(node *parser.BooleanLiteral) {
value := Value{Type: ValueBool, Data: node.Value}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpLoadConst, uint16(index))
}
func (c *Compiler) compileNilLiteral(node *parser.NilLiteral) {
value := Value{Type: ValueNil, Data: nil}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpLoadConst, uint16(index))
}
// Identifier compilation
func (c *Compiler) compileIdentifier(node *parser.Identifier) {
// Try local variables first
slot := c.current.ResolveLocal(node.Value)
if slot != -1 {
if slot == -2 {
c.addError("can't read local variable in its own initializer")
return
}
c.current.EmitInstruction(OpLoadLocal, uint16(slot))
return
}
// Try upvalues
upvalue := c.resolveUpvalue(node.Value)
if upvalue != -1 {
c.current.EmitInstruction(OpGetUpvalue, uint16(upvalue))
return
}
// Must be global
value := Value{Type: ValueString, Data: node.Value}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpLoadGlobal, uint16(index))
}
// Assignment compilation
func (c *Compiler) compileAssignment(node *parser.Assignment) {
c.compileExpression(node.Value)
switch target := node.Target.(type) {
case *parser.Identifier:
if node.IsDeclaration {
// Check if we're at global scope
if c.current.FunctionType == FunctionTypeScript && c.current.ScopeDepth == 0 {
// Global variable declaration - treat as global assignment
value := Value{Type: ValueString, Data: target.Value}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpStoreGlobal, uint16(index))
} else {
// Local variable declaration
if err := c.current.AddLocal(target.Value); err != nil {
c.addError(err.Error())
return
}
c.current.MarkInitialized()
}
} else {
// Assignment to existing variable
slot := c.current.ResolveLocal(target.Value)
if slot != -1 {
c.current.EmitInstruction(OpStoreLocal, uint16(slot))
} else {
upvalue := c.resolveUpvalue(target.Value)
if upvalue != -1 {
c.current.EmitInstruction(OpSetUpvalue, uint16(upvalue))
} else {
// Global assignment
value := Value{Type: ValueString, Data: target.Value}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpStoreGlobal, uint16(index))
}
}
}
case *parser.DotExpression:
// table.field = value
c.compileExpression(target.Left)
value := Value{Type: ValueString, Data: target.Key}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpSetField, uint16(index))
case *parser.IndexExpression:
// table[key] = value
c.compileExpression(target.Left)
c.compileExpression(target.Index)
c.current.EmitInstruction(OpSetIndex)
default:
c.addError("invalid assignment target")
}
}
func (c *Compiler) compileAssignmentExpression(node *parser.Assignment) {
c.compileAssignment(node)
// Assignment expressions leave the assigned value on stack
}
// Operator compilation
func (c *Compiler) compilePrefixExpression(node *parser.PrefixExpression) {
c.compileExpression(node.Right)
switch node.Operator {
case "-":
c.current.EmitInstruction(OpNeg)
case "not":
c.current.EmitInstruction(OpNot)
default:
c.addError(fmt.Sprintf("unknown prefix operator: %s", node.Operator))
}
}
func (c *Compiler) compileInfixExpression(node *parser.InfixExpression) {
// Handle short-circuit operators specially
if node.Operator == "and" {
c.compileExpression(node.Left)
jump := c.current.EmitJump(OpJumpIfFalse)
c.current.EmitInstruction(OpPop)
c.compileExpression(node.Right)
c.current.PatchJump(jump)
return
}
if node.Operator == "or" {
c.compileExpression(node.Left)
elseJump := c.current.EmitJump(OpJumpIfFalse)
endJump := c.current.EmitJump(OpJump)
c.current.PatchJump(elseJump)
c.current.EmitInstruction(OpPop)
c.compileExpression(node.Right)
c.current.PatchJump(endJump)
return
}
// Regular binary operators
c.compileExpression(node.Left)
c.compileExpression(node.Right)
switch node.Operator {
case "+":
c.current.EmitInstruction(OpAdd)
case "-":
c.current.EmitInstruction(OpSub)
case "*":
c.current.EmitInstruction(OpMul)
case "/":
c.current.EmitInstruction(OpDiv)
case "==":
c.current.EmitInstruction(OpEq)
case "!=":
c.current.EmitInstruction(OpNeq)
case "<":
c.current.EmitInstruction(OpLt)
case "<=":
c.current.EmitInstruction(OpLte)
case ">":
c.current.EmitInstruction(OpGt)
case ">=":
c.current.EmitInstruction(OpGte)
default:
c.addError(fmt.Sprintf("unknown infix operator: %s", node.Operator))
}
}
// Control flow compilation
func (c *Compiler) compileIfStatement(node *parser.IfStatement) {
c.compileExpression(node.Condition)
// Jump over then branch if condition is false
thenJump := c.current.EmitJump(OpJumpIfFalse)
c.current.EmitInstruction(OpPop)
// Compile then branch
c.current.BeginScope()
for _, stmt := range node.Body {
c.compileStatement(stmt)
}
c.current.EndScope()
// Jump over else branches
elseJump := c.current.EmitJump(OpJump)
c.current.PatchJump(thenJump)
c.current.EmitInstruction(OpPop)
// Compile elseif branches
var elseifJumps []int
for _, elseif := range node.ElseIfs {
c.compileExpression(elseif.Condition)
nextJump := c.current.EmitJump(OpJumpIfFalse)
c.current.EmitInstruction(OpPop)
c.current.BeginScope()
for _, stmt := range elseif.Body {
c.compileStatement(stmt)
}
c.current.EndScope()
elseifJumps = append(elseifJumps, c.current.EmitJump(OpJump))
c.current.PatchJump(nextJump)
c.current.EmitInstruction(OpPop)
}
// Compile else branch
if len(node.Else) > 0 {
c.current.BeginScope()
for _, stmt := range node.Else {
c.compileStatement(stmt)
}
c.current.EndScope()
}
// Patch all jumps to end
c.current.PatchJump(elseJump)
for _, jump := range elseifJumps {
c.current.PatchJump(jump)
}
}
func (c *Compiler) compileWhileStatement(node *parser.WhileStatement) {
c.current.EnterLoop()
c.compileExpression(node.Condition)
exitJump := c.current.EmitJump(OpJumpIfFalse)
c.current.EmitInstruction(OpPop)
c.current.BeginScope()
for _, stmt := range node.Body {
c.compileStatement(stmt)
}
c.current.EndScope()
// Jump back to condition
jump := len(c.current.Chunk.Code) - c.current.LoopStart + 2
c.current.EmitInstruction(OpJump, uint16(jump))
c.current.PatchJump(exitJump)
c.current.EmitInstruction(OpPop)
c.current.ExitLoop()
}
// Table operations
func (c *Compiler) compileTableLiteral(node *parser.TableLiteral) {
c.current.EmitInstruction(OpNewTable)
for _, pair := range node.Pairs {
if pair.Key == nil {
// Array-style element
c.compileExpression(pair.Value)
c.current.EmitInstruction(OpTableInsert)
} else {
// Key-value pair
c.current.EmitInstruction(OpDup) // Duplicate table reference
c.compileExpression(pair.Key)
c.compileExpression(pair.Value)
c.current.EmitInstruction(OpSetIndex)
}
}
}
func (c *Compiler) compileDotExpression(node *parser.DotExpression) {
c.compileExpression(node.Left)
value := Value{Type: ValueString, Data: node.Key}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpGetField, uint16(index))
}
func (c *Compiler) compileIndexExpression(node *parser.IndexExpression) {
c.compileExpression(node.Left)
c.compileExpression(node.Index)
c.current.EmitInstruction(OpGetIndex)
}
// Function compilation
func (c *Compiler) compileCallExpression(node *parser.CallExpression) {
c.compileExpression(node.Function)
// Compile arguments
for _, arg := range node.Arguments {
c.compileExpression(arg)
}
c.current.EmitInstruction(OpCall, uint16(len(node.Arguments)))
}
func (c *Compiler) compileReturnStatement(node *parser.ReturnStatement) {
if node.Value != nil {
c.compileExpression(node.Value)
c.current.EmitInstruction(OpReturn)
} else {
c.current.EmitInstruction(OpReturnNil)
}
}
func (c *Compiler) compileExitStatement(node *parser.ExitStatement) {
if node.Value != nil {
c.compileExpression(node.Value)
} else {
// Default exit code 0
value := Value{Type: ValueNumber, Data: float64(0)}
index := c.current.AddConstant(value)
if index == -1 {
c.addError("too many constants")
return
}
c.current.EmitInstruction(OpLoadConst, uint16(index))
}
c.current.EmitInstruction(OpExit)
}
// Placeholder implementations for complex features
func (c *Compiler) compileStructStatement(node *parser.StructStatement) {
// TODO: Implement struct compilation
c.addError("struct compilation not yet implemented")
}
func (c *Compiler) compileMethodDefinition(node *parser.MethodDefinition) {
// TODO: Implement method compilation
c.addError("method compilation not yet implemented")
}
func (c *Compiler) compileStructConstructor(node *parser.StructConstructor) {
// TODO: Implement struct constructor compilation
c.addError("struct constructor compilation not yet implemented")
}
func (c *Compiler) compileFunctionLiteral(node *parser.FunctionLiteral) {
// TODO: Implement function literal compilation
c.addError("function literal compilation not yet implemented")
}
func (c *Compiler) compileForStatement(node *parser.ForStatement) {
// TODO: Implement numeric for loop compilation
c.addError("for statement compilation not yet implemented")
}
func (c *Compiler) compileForInStatement(node *parser.ForInStatement) {
// TODO: Implement for-in loop compilation
c.addError("for-in statement compilation not yet implemented")
}
// Helper methods
func (c *Compiler) resolveUpvalue(name string) int {
if c.enclosing == nil {
return -1
}
local := c.enclosing.ResolveLocal(name)
if local != -1 {
c.enclosing.Locals[local].IsCaptured = true
return c.current.AddUpvalue(uint8(local), true)
}
upvalue := c.resolveUpvalueInEnclosing(name)
if upvalue != -1 {
return c.current.AddUpvalue(uint8(upvalue), false)
}
return -1
}
func (c *Compiler) resolveUpvalueInEnclosing(name string) int {
if c.enclosing == nil {
return -1
}
// This would recursively check enclosing scopes
// Simplified for now
return -1
}
func (c *Compiler) addError(message string) {
c.errors = append(c.errors, CompileError{
Message: message,
Line: 0, // TODO: Add line tracking
Column: 0, // TODO: Add column tracking
})
}
// Error reporting
func (c *Compiler) Errors() []CompileError { return c.errors }
func (c *Compiler) HasErrors() bool { return len(c.errors) > 0 }