Compare commits

...

10 Commits

Author SHA1 Message Date
45df4fdf5f Fix array to ptr cast 2026-04-14 18:26:33 +03:00
d3f0a730fd Make strings be char[n] 2026-04-14 17:01:08 +03:00
26779414a7 Implement arrays 2026-04-14 15:55:13 +03:00
c0a2c41c33 Fix parsing of empty arrays 2026-04-14 15:24:49 +03:00
0400aa1d99 Make empty arrays into just pointers 2026-04-14 15:20:14 +03:00
d70f6fffa0 Parse array type 2026-04-14 15:05:44 +03:00
3370e76c48 Add ArrayType 2026-04-14 14:57:25 +03:00
e797cab50f Remove some unused code 2026-04-14 14:49:07 +03:00
fa8c4b0e74 Allow underscores in idents 2026-04-14 14:19:08 +03:00
0cdf1abf82 Tweak parsing 2026-04-14 14:17:17 +03:00
11 changed files with 285 additions and 104 deletions

View File

@ -61,6 +61,13 @@ namespace AST {
return out.str();
}
std::string IndexAccessExpression::formatted() {
std::stringstream out{ "" };
out << this->m_expr->formatted();
out << "[" << this->m_num << "]";
return out.str();
}
std::string ExpressionStatement::formatted() {
std::stringstream out{ "" };
out << this->m_expr->formatted();

View File

@ -211,6 +211,30 @@ namespace AST {
) override;
};
/// @brief Same as value[num]
class IndexAccessExpression : public Expression {
private:
std::unique_ptr<Expression> m_expr;
uint32_t m_num;
public:
IndexAccessExpression(
token::Metadata meta,
std::unique_ptr<Expression> expr,
uint32_t num)
: Expression{ meta }
, m_expr{ std::move(expr) }
, m_num{ num } {
}
virtual ~IndexAccessExpression() override = default;
virtual std::string formatted() override;
virtual codegen::StackValue codegen(codegen::Builder& builder, codegen::Scope& scope) override;
virtual std::shared_ptr<types::Type> typecheck(
typecheck::State& state,
typecheck::Scope& scope,
std::optional<std::shared_ptr<types::Type>> expected_ty
) override;
};
class ReturnStatement : public Statement {
private:

View File

@ -53,6 +53,25 @@ namespace types {
std::shared_ptr<Type> casted_ty,
std::shared_ptr<Type> target_ty) {
if (casted_ty->m_kind == types::TypeKind::Array && target_ty->m_kind == types::TypeKind::Pointer) {
auto array_ty = dynamic_cast<types::ArrayType*>(casted_ty.get());
auto ptr_ty = dynamic_cast<types::PointerType*>(target_ty.get());
if (!types_equal(array_ty->m_inner, ptr_ty->m_inner))
return {};
return CastDefinition{ casted_ty, target_ty, true,
[](codegen::Builder&, std::shared_ptr<Type>, llvm::Value* value) {
// auto ptr_ty = dynamic_cast<types::PointerType*>(target.get());
// std::vector<llvm::Value*> indices {};
// indices.push_back(llvm::ConstantInt::get(
// builder.builder->getInt32Ty(),
// 0
// ));
// return builder.builder->CreateGEP(ptr_ty->m_inner->codegen(builder), value, indices, "arraydecay");
return value;
} };
}
for (auto& cast : casts) {
if (types_equal(cast.casted_ty, casted_ty) && types_equal(cast.target_ty, target_ty)) {
return cast;

View File

@ -26,13 +26,16 @@ namespace AST {
}
codegen::StackValue StringLiteralExpression::codegen(codegen::Builder& builder, codegen::Scope&) {
auto stack_type = new types::PointerType{ std::make_unique<types::FundamentalType>(types::FundamentalTypeKind::Char) };
auto stack_type = new types::ArrayType{
std::make_unique<types::FundamentalType>(types::FundamentalTypeKind::Char),
static_cast<uint32_t>(this->m_value.size()) + 1
};
auto str = llvm::StringRef{ this->m_value.c_str() };
auto global_str = builder.builder->CreateGlobalString(str);
return codegen::StackValue{
builder.builder->CreateGlobalString(str),
builder.builder->CreateLoad(stack_type->codegen(builder), global_str, "literal"),
std::unique_ptr<types::Type>{stack_type},
};
}
@ -103,6 +106,14 @@ namespace AST {
auto expr = this->m_expr->codegen(builder, scope);
auto cast = types::find_cast(scope.casts, expr.ty, this->m_ty);
if (cast) {
if (cast->target_ty->m_kind == types::TypeKind::Pointer && cast->casted_ty->m_kind == types::TypeKind::Array) {
auto allocated = builder.builder->CreateAlloca(cast->casted_ty->codegen(builder));
builder.builder->CreateStore(expr.value, allocated);
return codegen::StackValue{
cast->codegen(builder, cast->target_ty, allocated),
cast->target_ty
};
}
return codegen::StackValue{
cast->codegen(builder, cast->target_ty, expr.value),
cast->target_ty
@ -131,6 +142,59 @@ namespace AST {
}
codegen::StackValue IndexAccessExpression::codegen(codegen::Builder& builder, codegen::Scope& scope) {
auto lvalue_scope = scope.with_lvalue();
auto value = this->m_expr->codegen(builder, lvalue_scope);
std::shared_ptr<types::Type> inner_ty{};
std::vector<llvm::Value*> idx_list{ };
idx_list.push_back(llvm::ConstantInt::get(builder.builder->getInt32Ty(), this->m_num));
if (value.ty->m_kind == types::TypeKind::Pointer) {
auto ptr_ty = dynamic_cast<types::PointerType*>(value.ty.get());
if (ptr_ty->m_inner->m_kind == types::TypeKind::Pointer) {
auto inner_ptr_ty = dynamic_cast<types::PointerType*>(ptr_ty->m_inner.get());
auto gep_value = builder.builder->CreateGEP(inner_ptr_ty->m_inner->codegen(builder), value.value, idx_list, "GEP");
if (scope.is_lvalue) {
return codegen::StackValue{
gep_value,
value.ty,
};
}
else {
auto loaded = value.ty->load(builder, gep_value);
return codegen::StackValue{
loaded.first,
loaded.second
};
}
}
// Must be an array otherwise
auto arr_ty = dynamic_cast<types::ArrayType*>(ptr_ty->m_inner.get());
auto gep_value = builder.builder->CreateGEP(arr_ty->m_inner->codegen(builder), value.value, idx_list, "GEP");
if (scope.is_lvalue) {
return codegen::StackValue{
gep_value,
value.ty,
};
}
else {
auto inner_ptr_ty = types::PointerType{ arr_ty->m_inner };
auto loaded = inner_ptr_ty.load(builder, gep_value);
return codegen::StackValue{
loaded.first,
loaded.second
};
}
}
throw CompileError("Tried indexing a non-pointer", this->m_meta);
}
void ReturnStatement::codegen(codegen::Builder& builder, codegen::Scope& scope) {
if (!builder.block)
return;
@ -291,4 +355,8 @@ namespace types {
llvm::Type* PointerType::codegen(codegen::Builder& builder) {
return llvm::PointerType::get(*builder.context, 0);
}
llvm::Type* ArrayType::codegen(codegen::Builder& builder) {
return llvm::ArrayType::get(this->m_inner->codegen(builder), this->m_size);
}
}

View File

@ -162,6 +162,8 @@ std::optional<CompileOutput> compile(std::string_view in_filename) {
builder.mod->print(llvm_ir_dest, nullptr);
llvm_ir_dest.flush();
std::cout << llvm_ir_string << std::endl;
// Print output to obj-file
std::error_code EC;
std::string obj_string;

View File

@ -49,6 +49,28 @@ namespace parsing {
}
}
Result<std::optional<uint32_t>, std::string> parse_array_postfix(token::TokenStream& stream, bool allow_empty) {
token::TokenStream inner{ stream };
try {
std::optional<uint32_t> returned{};
inner.expect(token::Type::Symbol, "[");
if (inner.peek().type == token::Type::LiteralInt) {
returned = std::stoi(inner.next().content);
}
if (!allow_empty && !returned.has_value()) {
throw std::runtime_error("Expected array size");
}
inner.expect(token::Type::Symbol, "]");
stream.m_position = inner.m_position;
return returned;
}
catch (std::runtime_error& error) {
return std::string{ error.what() };
}
}
Result<std::unique_ptr<AST::Expression>, std::string> parse_plain_expression(token::TokenStream& stream) {
token::TokenStream inner{ stream };
try {
@ -101,16 +123,9 @@ namespace parsing {
try {
auto before_meta = inner.metadata();
if (inner.peek().content == "(") {
inner.next();
auto expr = parse_expression(inner).unwrap();
inner.expect(token::Type::Symbol, ")");
stream.m_position = inner.m_position;
return expr;
}
if (auto cast = parse_cast(inner); cast.ok()) {
auto expr = parse_expression(inner).unwrap();
auto expr = parse_primary_expression(inner).unwrap();
stream.m_position = inner.m_position;
return std::unique_ptr<AST::Expression>{
new AST::CastExpression{
@ -120,6 +135,13 @@ namespace parsing {
}
};
}
else if (inner.peek().content == "(") {
inner.next();
auto expr = parse_expression(inner).unwrap();
inner.expect(token::Type::Symbol, ")");
stream.m_position = inner.m_position;
return expr;
}
else if (inner.peek().content == "&") {
inner.next();
auto expr = parse_primary_expression(inner).unwrap();
@ -138,22 +160,29 @@ namespace parsing {
}
auto plain_expr = parse_plain_expression(inner);
while (inner.peek().content == "(") {
inner.next();
while (inner.peek().content == "(" || inner.peek().content == "[") {
if (inner.peek().content == "(") {
inner.next();
std::vector<std::unique_ptr<AST::Expression>> args{};
std::vector<std::unique_ptr<AST::Expression>> args{};
int counter = 0;
while (inner.peek().content != ")") {
if (counter++ > 0)
inner.expect(token::Type::Symbol, ",");
args.push_back(parse_expression(inner).unwrap());
int counter = 0;
while (inner.peek().content != ")") {
if (counter++ > 0)
inner.expect(token::Type::Symbol, ",");
args.push_back(parse_expression(inner).unwrap());
}
inner.expect(token::Type::Symbol, ")");
auto fn_call = new AST::FunctionCallExpression{ before_meta + inner.metadata(), plain_expr.unwrap(), std::move(args) };
plain_expr = std::unique_ptr<AST::Expression>{ fn_call };
}
else if (auto postfix = parse_array_postfix(inner, false); postfix.ok()) {
auto idx_expr = new AST::IndexAccessExpression{
before_meta + inner.metadata(), plain_expr.unwrap(), *postfix.unwrap() };
plain_expr = std::unique_ptr<AST::Expression>{ idx_expr };
}
inner.expect(token::Type::Symbol, ")");
auto fn_call = new AST::FunctionCallExpression{ before_meta + inner.metadata(), plain_expr.unwrap(), std::move(args) };
plain_expr = std::unique_ptr<AST::Expression>{ fn_call };
}
@ -244,6 +273,22 @@ namespace parsing {
auto ty = parse_type(inner).unwrap();
auto name = inner.expect(token::Type::Ident);
auto array_postfix = parse_array_postfix(inner, false);
while (array_postfix.ok()) {
auto postfix = array_postfix.unwrap();
if (postfix) {
ty = std::shared_ptr<types::Type>{
new types::ArrayType(ty, *postfix)
};
}
else {
ty = std::shared_ptr<types::Type>{
new types::PointerType(ty)
};
}
array_postfix = parse_array_postfix(inner, false);
}
std::optional<std::unique_ptr<AST::Expression>> expr = {};
if (inner.peek().type == token::Type::Symbol && inner.peek().content == "=") {
inner.expect(token::Type::Symbol, "=");

View File

@ -159,13 +159,13 @@ namespace token {
i++; // Skip second "
tokens.push_back(token::Token{ token::Type::LiteralStr, content, meta + (content.size() + 2) });
}
else if (std::isalpha(c)) {
else if (std::isalpha(c) || c == '_') {
std::string content{};
do {
content += c;
if ((i + 1) >= text_length) break;
c = text[++i];
} while (std::isalnum(c));
} while (std::isalnum(c) || c == '_');
token::Type type = token::Type::Ident;
if (content == "return") {

View File

@ -89,7 +89,7 @@ namespace AST {
auto char_ty = std::shared_ptr<types::Type>{
new types::FundamentalType{ types::FundamentalTypeKind::Char }
};
auto ptr_ty = new types::PointerType{ char_ty };
auto ptr_ty = new types::ArrayType{ char_ty, static_cast<uint32_t>(this->m_value.size()) + 1 };
return std::shared_ptr<types::Type>{ptr_ty};
}
@ -281,6 +281,34 @@ namespace AST {
return ptr_ty->m_inner;
}
std::shared_ptr<types::Type> IndexAccessExpression::typecheck(
typecheck::State& state,
typecheck::Scope& scope,
std::optional<std::shared_ptr<types::Type>>
) {
auto expr_ty = this->m_expr->typecheck(state, scope, {});
if (expr_ty->m_kind != types::TypeKind::Pointer && expr_ty->m_kind != types::TypeKind::Array) {
state.errors.push_back(
CompileError("Tried to index " + expr_ty->formatted(), this->m_meta));
return std::shared_ptr<types::Type> {
new types::FundamentalType{ types::FundamentalTypeKind::Void }
};
}
if (expr_ty->m_kind == types::TypeKind::Pointer) {
auto ptr_ty = dynamic_cast<types::PointerType*>(expr_ty.get());
return ptr_ty->m_inner;
}
else if (expr_ty->m_kind == types::TypeKind::Array) {
auto ptr_ty = dynamic_cast<types::ArrayType*>(expr_ty.get());
return ptr_ty->m_inner;
}
// Default return type
return std::shared_ptr<types::Type> {
new types::FundamentalType{ types::FundamentalTypeKind::Void }
};
}
void ReturnStatement::typecheck(typecheck::State& state, typecheck::Scope& scope) {
auto res_ty = this->m_expr->typecheck(state, scope, scope.return_ty);
@ -292,7 +320,9 @@ namespace AST {
void InitializationStatement::typecheck(typecheck::State& state, typecheck::Scope& scope) {
if (this->m_expr) {
(*this->m_expr)->typecheck(state, scope, this->m_type);
auto expr_ty = (*this->m_expr)->typecheck(state, scope, this->m_type);
auto check_res = check_type(state, expr_ty, this->m_type);
this->m_expr = handle_res(std::move(*this->m_expr), check_res, state);
}
scope.symbols[this->m_name] = this->m_type;
}

View File

@ -1,5 +1,6 @@
#include <sstream>
#include <iostream>
#include "types.h"
#include "binops.h"
@ -54,23 +55,6 @@ namespace types {
}
}
llvm::Value* Type::add(codegen::Builder&, llvm::Value*, llvm::Value*) {
throw std::runtime_error("Invalid operation for this type");
}
llvm::Value* Type::sub(codegen::Builder&, llvm::Value*, llvm::Value*) {
throw std::runtime_error("Invalid operation for this type");
}
llvm::Value* Type::lt(codegen::Builder&, llvm::Value*, llvm::Value*) {
throw std::runtime_error("Invalid operation for this type");
}
llvm::Value* Type::gt(codegen::Builder&, llvm::Value*, llvm::Value*) {
throw std::runtime_error("Invalid operation for this type");
}
std::optional<std::shared_ptr<Type>> Type::return_type() {
return {};
}
@ -84,50 +68,6 @@ namespace types {
return std::pair(ptr, self);
}
llvm::Value* FundamentalType::add(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) {
switch (this->m_ty) {
case FundamentalTypeKind::Int:
case FundamentalTypeKind::Bool:
case FundamentalTypeKind::Char:
return builder.builder->CreateAdd(lhs, rhs, "add");
default:
throw std::runtime_error("Invalid type for add");
}
}
llvm::Value* FundamentalType::sub(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) {
switch (this->m_ty) {
case FundamentalTypeKind::Int:
case FundamentalTypeKind::Bool:
case FundamentalTypeKind::Char:
return builder.builder->CreateSub(lhs, rhs, "sub");
default:
throw std::runtime_error("Invalid type");
}
}
llvm::Value* FundamentalType::lt(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) {
switch (this->m_ty) {
case FundamentalTypeKind::Int:
case FundamentalTypeKind::Bool:
case FundamentalTypeKind::Char:
return builder.builder->CreateCmp(llvm::CmpInst::Predicate::ICMP_SLT, lhs, rhs, "cmp");
default:
throw std::runtime_error("Invalid type");
}
}
llvm::Value* FundamentalType::gt(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) {
switch (this->m_ty) {
case FundamentalTypeKind::Int:
case FundamentalTypeKind::Bool:
case FundamentalTypeKind::Char:
return builder.builder->CreateCmp(llvm::CmpInst::Predicate::ICMP_SGT, lhs, rhs);
default:
throw std::runtime_error("Invalid type");
}
}
bool FundamentalType::is_signed() {
switch (this->m_ty) {
case FundamentalTypeKind::Int:
@ -204,7 +144,34 @@ namespace types {
return 64;
}
std::string ArrayType::formatted() {
std::stringstream out{ "" };
out << this->m_inner->formatted();
out << "[" << this->m_size << "]";
return out.str();
}
std::pair<llvm::Value*, std::shared_ptr<Type>> ArrayType::load(codegen::Builder&, llvm::Value* ptr) {
auto self = std::make_shared<ArrayType>(*this);
return std::pair(ptr, self);
}
uint32_t ArrayType::size() {
return this->m_size * this->m_inner->size();
}
bool types_equal(std::shared_ptr<types::Type> type1, std::shared_ptr<types::Type> type2) {
// if (type1->m_kind == TypeKind::Array && type2->m_kind == TypeKind::Pointer) {
// auto ty1 = dynamic_cast<ArrayType*>(type1.get());
// auto ty2 = dynamic_cast<PointerType*>(type2.get());
// return types_equal(ty1->m_inner, ty2->m_inner);
// }
// else if (type1->m_kind == TypeKind::Pointer && type2->m_kind == TypeKind::Array) {
// auto ty1 = dynamic_cast<PointerType*>(type1.get());
// auto ty2 = dynamic_cast<ArrayType*>(type2.get());
// return types_equal(ty1->m_inner, ty2->m_inner);
// }
if (type1->m_kind != type2->m_kind)
return false;
@ -239,6 +206,12 @@ namespace types {
return types_equal(ty1->m_inner, ty2->m_inner);
}
else if (type1->m_kind == TypeKind::Array) {
auto ty1 = dynamic_cast<ArrayType*>(type1.get());
auto ty2 = dynamic_cast<ArrayType*>(type2.get());
return types_equal(ty1->m_inner, ty2->m_inner) && ty1->m_size == ty2->m_size;
}
else {
return false;
}

View File

@ -11,6 +11,7 @@ namespace types {
Fundamental,
Function,
Pointer,
Array,
};
enum FundamentalTypeKind {
@ -31,10 +32,6 @@ namespace types {
virtual llvm::Type* codegen(codegen::Builder& builder) = 0;
virtual std::pair<llvm::Value*, std::shared_ptr<Type>> load(codegen::Builder& builder, llvm::Value* ptr) = 0;
virtual std::optional<std::shared_ptr<Type>> return_type();
virtual llvm::Value* add(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs);
virtual llvm::Value* sub(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs);
virtual llvm::Value* lt(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs);
virtual llvm::Value* gt(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs);
virtual bool is_signed();
virtual uint32_t size() = 0;
};
@ -47,10 +44,6 @@ namespace types {
virtual std::string formatted() override;
virtual llvm::Type* codegen(codegen::Builder& builder) override;
virtual std::pair<llvm::Value*, std::shared_ptr<Type>> load(codegen::Builder& builder, llvm::Value* ptr) override;
virtual llvm::Value* add(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) override;
virtual llvm::Value* sub(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) override;
virtual llvm::Value* lt(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) override;
virtual llvm::Value* gt(codegen::Builder& builder, llvm::Value* lhs, llvm::Value* rhs) override;
virtual bool is_signed() override;
virtual uint32_t size() override;
};
@ -75,11 +68,10 @@ namespace types {
virtual uint32_t size() override;
};
class PointerType : public Type {
public:
std::shared_ptr<Type> m_inner;
PointerType(std::shared_ptr<Type> inner)
: Type(TypeKind::Pointer), m_inner{ std::move(inner) } {
}
@ -90,6 +82,21 @@ namespace types {
virtual uint32_t size() override;
};
class ArrayType : public Type {
public:
std::shared_ptr<Type> m_inner;
uint32_t m_size;
ArrayType(std::shared_ptr<Type> inner, uint32_t size)
: Type(TypeKind::Array), m_inner{ std::move(inner) }, m_size{ size } {
}
virtual ~ArrayType() override = default;
virtual std::string formatted() override;
virtual llvm::Type* codegen(codegen::Builder& builder) override;
virtual std::pair<llvm::Value*, std::shared_ptr<Type>> load(codegen::Builder& builder, llvm::Value* ptr) override;
virtual uint32_t size() override;
};
bool types_equal(std::shared_ptr<types::Type> type1, std::shared_ptr<types::Type> type2);
}

16
test.c
View File

@ -6,13 +6,19 @@ int fibonacci(int n) {
return fibonacci(n - 1) + fibonacci(n - 2);
}
void modifyvalue(char* otus) {
void modify_value(char* otus) {
*otus = 20;
}
int main() {
printf("10th fibonacci number is %d!", fibonacci(10));
char res = 10;
modifyvalue(&res);
return res;
char text[29] = "10th fibonacci number is %d!";
printf(text, fibonacci(10));
char somelist[5];
somelist[0] = 15;
somelist[1] = 20;
somelist[2] = 25;
return somelist[0];
}