fix: fix this mess

I am stupid
I am stupid
I am stupid
I can't believe how f- ricking stupid I am

CMakeLists.txt

@@ -14,6 +14,7 @@ set(CMAKE_CXX_STANDARD_REQUIRED ON)
 add_subdirectory(furlang)
 add_subdirectory(furvm)
 add_subdirectory(furc)
+add_subdirectory(furas)
 
 if(DOXYGEN_FOUND)
     set(DOXYGEN_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/docs)
@@ -24,4 +25,4 @@ if(DOXYGEN_FOUND)
     doxygen_add_docs(doxygen_doc ${CMAKE_SOURCE_DIR}/furlang/ ${CMAKE_SOURCE_DIR}/furvm/ ${CMAKE_SOURCE_DIR}/furc/
         ALL WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
         COMMENT "Generating API documentation with Doxygen")
-endif()
+endif()

furas/CMakeLists.txt

@@ -0,0 +1,5 @@
+file(GLOB_RECURSE FURAS_SRCS "src/**.cpp")
+file(GLOB_RECURSE FURAS_HDRS "include/**.hpp")
+add_executable(furas ${FURAS_SRCS} ${FURAS_HDRS})
+target_include_directories(furas PUBLIC include/)
+target_link_libraries(furas PRIVATE furlang libfurvm)

furas/examples/array.furas

@@ -0,0 +1,35 @@
+type arr = array $s8 10
+
+func println $arr = native println
+
+public func main = #main
+main:
+    array $arr
+    dup
+    store %0
+
+    lenof
+    store %1
+
+    push $s32 0
+loop:
+    dup
+    load %1
+    ge
+    jnz #end
+body:
+    dup
+    load %0
+    swap
+    push $s8 69
+    set
+
+    push $s32 1
+    add
+
+    jmp #loop
+end:
+    drop
+    load %0
+    call $arr println
+    ret

furas/examples/array2.furas

@@ -0,0 +1,34 @@
+type arr = array $s8 10
+
+func println $arr = native println
+
+public func main = #main
+main:
+    array $arr
+    store %0
+
+    push $s32 2
+    store %1
+
+    push $s32 0
+loop:
+    dup
+    load %1
+    ge
+    jnz #end
+body:
+    dup
+    load %0
+    swap
+    push $s8 69
+    set
+
+    push $s32 1
+    add
+
+    jmp #loop
+end:
+    drop
+    load %0
+    call $arr println
+    ret

furas/examples/ops.furas

@@ -0,0 +1,23 @@
+func println $u32 = native println
+
+public func main = #main
+main:
+    push $u32 0
+    store %0
+loop_header:
+    load %0
+    push $u32 10
+    ge
+    jnz #loop_end
+loop_body:
+    load %0
+    call $u32 println
+
+    load %0
+    push $u32 1
+    add
+    store %0
+
+    jmp #loop_header
+loop_end:
+    ret

furas/include/furas/gen.hpp

@@ -0,0 +1,36 @@
+#ifndef FURAS_GEN_HPP
+#define FURAS_GEN_HPP
+
+#include "furas/lexer.hpp"
+#include "furvm/module.hpp"
+
+#include <string>
+
+namespace furas {
+
+struct generator_error {
+    enum type {
+        Success = 0,
+        Eof     = 1,
+
+        UnexpectedEof    = -1,
+        UnexpectedToken  = -2,
+        UnknownCharacter = -3,
+        UnknownType      = -4,
+    } type = Success;
+    std::string message;
+};
+
+class generator {
+public:
+    struct result {
+        generator_error error;
+        furvm::mod      mod;
+    };
+public:
+    static result generate(lexer lexer);
+};
+
+} // namespace furas
+
+#endif // FURAS_GEN_HPP

furas/include/furas/lexer.hpp

@@ -0,0 +1,47 @@
+#ifndef FURAS_LEXER_HPP
+#define FURAS_LEXER_HPP
+
+#include "furas/token.hpp"
+#include "furlang/result.hpp"
+
+#include <cstddef>
+#include <string_view>
+
+namespace furas {
+
+struct lexer_location {
+    std::string_view filename;
+    std::size_t      row, col;
+};
+
+struct lexer_error {
+    enum type {
+        EndOfFile = 0,
+        UnknownCharacter,
+    } type;
+
+    lexer_location location;
+    std::string    message;
+};
+
+using token_r = furlang::result<token, lexer_error>;
+
+class lexer {
+public:
+    lexer(std::string_view filename, std::string_view content)
+      : m_filename(filename), m_content(content) {}
+
+    token_r next_token();
+private:
+    constexpr lexer_location location() const { return { m_filename, m_cursor - m_lineStart, m_column }; }
+private:
+    std::string_view m_filename;
+    std::string_view m_content;
+    std::size_t      m_cursor    = 0;
+    std::size_t      m_lineStart = 0;
+    std::size_t      m_column    = 0;
+};
+
+} // namespace furas
+
+#endif // FURAS_LEXER_HPP

furas/include/furas/token.hpp

@@ -0,0 +1,94 @@
+#ifndef FURAS_TOKEN_HPP
+#define FURAS_TOKEN_HPP
+
+#include <cstdint>
+#include <string_view>
+
+namespace furas {
+
+struct token {
+    enum type {
+        Identifier = 0, /**< An identifier. */
+        Signed,         /**< A signed integer. */
+        Unsigned,       /**< An unsigned integer. */
+
+        // Markers
+        Monkey,  /**< Constant marker (`@`). */
+        Dolar,   /**< Type marker(`$`). */
+        Sha256,  /**< Label marker(`#`). */
+        Percent, /**< Variable marker(`%`). The more the better. */
+
+        EqSign, /**< `=` */
+        Dot,    /**< . */
+        Colon,  /**< `:` */
+
+        // Keywords
+        Func,    /**< `func` keyword for defining functions. */
+        Type,    /**< `type` keyword for defining types. */
+        Native,  /**< `native` keyword for native functions. :v: */
+        Import,  /**< `import` keyword for importing functions and types. */
+        Public,  /**< `public` access specifier. */
+        Private, /**< `private` access specifier. */
+
+        // Instructions
+        Push,
+        Array,
+        Get,
+        Set,
+        Drop,
+        Dup,
+        Swap,
+        Clone,
+        Ref,
+        Add,
+        Sub,
+        Mul,
+        Div,
+        Mod,
+        Eq,
+        Neq,
+        Lt,
+        Gt,
+        Le,
+        Ge,
+        Ptrof,
+        Sizeof,
+        Lenof,
+        Load,
+        Store,
+        Call,
+        Jmp,
+        Jnz,
+        Ret,
+
+        Count
+    } type = Count;
+    union value {
+        std::nullptr_t   null = nullptr;
+        std::string_view string;
+        std::int64_t     integer;
+        std::uint64_t    uint;
+    } value;
+
+    token(enum type type)
+      : type(type) {}
+
+    token(enum type type, std::string_view string)
+      : type(type) {
+        value.string = string;
+    }
+
+    token(std::uint64_t num)
+      : type(Unsigned) {
+        value.uint = num;
+    }
+
+    token(std::int64_t num)
+      : type(Signed) {
+        value.integer = num;
+    }
+};
+
+} // namespace furas
+
+#endif // FURAS_TOKEN_HPP

furas/src/gen.cpp

@@ -0,0 +1,558 @@
+#include "furas/gen.hpp"
+
+#include "furas/token.hpp"
+#include "furvm/function.hpp"
+#include "furvm/fwd.hpp"
+#include "furvm/instruction.hpp"
+
+#include <cassert>
+#include <cstdint>
+#include <cstdlib>
+#include <limits>
+#include <optional>
+#include <stdexcept>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+
+namespace furas {
+
+using namespace std::string_literals;
+
+namespace {
+
+struct instruction {
+    furvm::instruction_t fur{};
+    enum arg_type {
+        None = 0,
+        Type,
+        Constant,
+        Variable,
+        Function,
+        Label,
+    } arg = None;
+};
+
+// NOLINTBEGIN
+std::unordered_map<enum token::type, instruction> instructions = {
+    { token::Array, { furvm::instruction_t::Array, instruction::Type } },
+    { token::Get, { furvm::instruction_t::Get } },
+    { token::Set, { furvm::instruction_t::Set } },
+    { token::Drop, { furvm::instruction_t::Drop } },
+    { token::Dup, { furvm::instruction_t::Duplicate } },
+    { token::Swap, { furvm::instruction_t::Swap } },
+    { token::Clone, { furvm::instruction_t::Clone } },
+    { token::Ref, { furvm::instruction_t::Reference } },
+    { token::Add, { furvm::instruction_t::Add } },
+    { token::Sub, { furvm::instruction_t::Sub } },
+    { token::Mul, { furvm::instruction_t::Mul } },
+    { token::Div, { furvm::instruction_t::Div } },
+    { token::Mod, { furvm::instruction_t::Mod } },
+    { token::Eq, { furvm::instruction_t::Equals } },
+    { token::Neq, { furvm::instruction_t::NotEquals } },
+    { token::Lt, { furvm::instruction_t::LessThan } },
+    { token::Gt, { furvm::instruction_t::GreaterThan } },
+    { token::Le, { furvm::instruction_t::LessEqual } },
+    { token::Ge, { furvm::instruction_t::GreaterEqual } },
+    { token::Ptrof, { furvm::instruction_t::Pointerof } },
+    { token::Sizeof, { furvm::instruction_t::Sizeof } },
+    { token::Lenof, { furvm::instruction_t::Lengthof } },
+    { token::Load, { furvm::instruction_t::Load, instruction::Variable } },
+    { token::Store, { furvm::instruction_t::Store, instruction::Variable } },
+    { token::Call, { furvm::instruction_t::Call, instruction::Function } },
+    { token::Jmp, { furvm::instruction_t::Jump, instruction::Label } },
+    { token::Jnz, { furvm::instruction_t::JumpNotZero, instruction::Label } },
+    { token::Ret, { furvm::instruction_t::Return } },
+};
+// NOLINTEND
+
+const char* token_type(enum token::type type) {
+    switch (type) {
+    case token::Identifier: return "identifier";
+    case token::Signed: return "signed";
+    case token::Unsigned: return "unsigned";
+    case token::Monkey: return "'@'";
+    case token::Dolar: return "'$'";
+    case token::Sha256: return "'#'";
+    case token::Percent: return "'%'";
+    case token::EqSign: return "'='";
+    case token::Dot: return "'.'";
+    case token::Colon: return "':'";
+    case token::Func: return "func";
+    case token::Type: return "type";
+    case token::Native: return "native";
+    case token::Import: return "import";
+    case token::Public: return "public";
+    case token::Private: return "private";
+    case token::Push: return "push";
+    case token::Array: return "array";
+    case token::Get: return "get";
+    case token::Set: return "set";
+    case token::Drop: return "drop";
+    case token::Dup: return "dup";
+    case token::Swap: return "swap";
+    case token::Clone: return "clone";
+    case token::Ref: return "ref";
+    case token::Add: return "add";
+    case token::Sub: return "sub";
+    case token::Mul: return "mul";
+    case token::Div: return "div";
+    case token::Mod: return "mod";
+    case token::Eq: return "eq";
+    case token::Neq: return "neq";
+    case token::Lt: return "lt";
+    case token::Gt: return "gt";
+    case token::Le: return "le";
+    case token::Ge: return "ge";
+    case token::Ptrof: return "ptrof";
+    case token::Sizeof: return "sizeof";
+    case token::Lenof: return "lenof";
+    case token::Load: return "load";
+    case token::Store: return "store";
+    case token::Call: return "call";
+    case token::Jmp: return "jmp";
+    case token::Jnz: return "jnz";
+    case token::Ret: return "ret";
+    case token::Count: break;
+    }
+    throw std::runtime_error("unreachable");
+}
+
+struct mod_context {
+    struct label_context {
+        struct function_info {
+            std::string         name;
+            furvm::function_sig signature;
+            bool                pub;
+        };
+
+        static constexpr std::size_t INVALID = std::numeric_limits<std::size_t>::max();
+
+        std::size_t                offset = INVALID;
+        std::vector<function_info> functions;
+        std::vector<std::size_t>   unknowns;
+    };
+
+    furvm::mod mod;
+    std::unordered_map<std::pair<std::string, furvm::function_sig>,
+        furvm::function_h,
+        furlang::utility::
+            pair_hash<std::string, furvm::function_sig, std::hash<std::string>, furvm::detail::function_sig_hash>>
+                                                       functions;
+    std::unordered_map<std::string, furvm::mod_type_h> types;
+
+    std::unordered_map<std::string, label_context> labels;
+
+    struct token_result {
+        generator_error error;
+        token           token;
+
+        struct token* operator->() { return &token; }
+        struct token& operator*() { return token; }
+
+        bool operator!() const { return error.type != generator_error::Success; }
+    };
+
+    mod_context() {
+        types.emplace("s8", mod.emplace_type(furvm::mod_type::S8));
+        types.emplace("u8", mod.emplace_type(furvm::mod_type::U8));
+        types.emplace("s16", mod.emplace_type(furvm::mod_type::S16));
+        types.emplace("u16", mod.emplace_type(furvm::mod_type::U16));
+        types.emplace("s32", mod.emplace_type(furvm::mod_type::S32));
+        types.emplace("u32", mod.emplace_type(furvm::mod_type::U32));
+        types.emplace("s64", mod.emplace_type(furvm::mod_type::S64));
+        types.emplace("u64", mod.emplace_type(furvm::mod_type::U64));
+    }
+
+    static token_result next_token(lexer& lexer) {
+        auto token = lexer.next_token();
+        if (token.has_error()) {
+            switch (token.error().type) {
+            case lexer_error::EndOfFile:
+                return { { generator_error::UnexpectedEof, "Unexpected end of file" }, { token::Count } };
+            case lexer_error::UnknownCharacter:
+                return { { generator_error::UnknownCharacter, token.error().message }, { token::Count } };
+            }
+        }
+        return { { generator_error::Success }, *token };
+    }
+
+    static token_result eat_token(lexer& lexer, enum token::type type) {
+        auto token = next_token(lexer);
+        if (!token) return token;
+        if (token.token.type != type) {
+            return { { generator_error::UnexpectedToken,
+                         "Expected "s + token_type(type) + ", but got " + token_type(token->type) },
+                { token::Count } };
+        }
+        return { { generator_error::Success }, token.token };
+    }
+
+    generator_error generate(lexer& lexer) {
+        auto result = next_token(lexer);
+        if (result.error.type == generator_error::UnexpectedEof) return { generator_error::Eof };
+        if (!result) return result.error;
+        switch (result->type) {
+        case token::Identifier: {
+            std::string labelName = std::string(result->value.string);
+
+            result = eat_token(lexer, token::Colon);
+            if (!result) return result.error;
+
+            auto& label = labels[labelName];
+            if (label.offset != label_context::INVALID)
+                return { generator_error::UnexpectedToken, "Label "s + labelName + " already exists" };
+            label.offset = mod.bytecode().size();
+
+            for (auto& func : label.functions) {
+                furvm::function_h handle =
+                    (func.pub ? mod.emplace_function(func.name, std::move(func.signature), label.offset)
+                              : mod.emplace_function(std::move(func.signature), label.offset));
+
+                functions.emplace(std::make_pair(std::move(func.name), std::move(func.signature)), handle);
+
+                handle.dispatch();
+            }
+            for (auto unknown : label.unknowns) {
+                std::ptrdiff_t jmpOff = static_cast<std::ptrdiff_t>(label.offset - unknown);
+                if (jmpOff < std::numeric_limits<std::int8_t>::min() ||
+                    jmpOff > std::numeric_limits<std::int8_t>::max()) {
+                    assert(false); // TODO: Further jumps are not implemented
+                }
+                mod.bytecode()[unknown - 1] = jmpOff;
+            }
+            label.functions = {};
+            label.unknowns  = {};
+            return { generator_error::Success };
+        }
+
+        case token::Func:
+        case token::Type:
+        case token::Public:
+        case token::Private: {
+            bool pub = false;
+            if (result->type == token::Public) {
+                pub    = true;
+                result = next_token(lexer);
+                if (!result) return { result.error };
+            } else if (result->type == token::Private) {
+                result = next_token(lexer);
+                if (!result) return { result.error };
+            }
+
+            if (result->type == token::Func) {
+                auto nameRes = next_token(lexer);
+                if (!nameRes) return nameRes.error;
+
+                furvm::function_sig signature;
+
+                result = next_token(lexer);
+                if (!result) return result.error;
+                while (result->type == token::Dolar) {
+                    result = eat_token(lexer, token::Identifier);
+                    if (!result) return result.error;
+                    if (auto it = types.find(std::string(result->value.string)); it != types.end()) {
+                        signature.params.push_back(it->second);
+                    } else {
+                        return { generator_error::UnknownType, "Unknown type "s + std::string(result->value.string) };
+                    }
+
+                    result = next_token(lexer);
+                    if (!result) return result.error;
+                }
+                if (result->type != token::EqSign) return result.error;
+
+                result = next_token(lexer);
+                if (!result) return result.error;
+                if (result->type == token::Dolar) {
+                    result = eat_token(lexer, token::Identifier);
+                    if (!result) return result.error;
+                    if (auto it = types.find(std::string(result->value.string)); it != types.end()) {
+                        signature.returnType = it->second;
+                    } else {
+                        return { generator_error::UnknownType, "Unknown type "s + std::string(result->value.string) };
+                    }
+
+                    result = next_token(lexer);
+                    if (!result) return result.error;
+                }
+
+                switch (result->type) {
+                case token::Sha256: {
+                    result = eat_token(lexer, token::Identifier);
+                    if (!result) return result.error;
+
+                    std::string name = std::string(nameRes->value.string);
+
+                    std::string labelName = std::string(result->value.string);
+                    std::size_t offset    = 0;
+                    if (auto it = labels.find(labelName); it != labels.end()) {
+                        offset = it->second.offset;
+                    } else {
+                        labels[labelName].functions.emplace_back(
+                            label_context::function_info{ std::move(name), std::move(signature), pub });
+                        return { generator_error::Success };
+                    }
+
+                    furvm::function_h handle =
+                        (pub ? mod.emplace_function(name, signature, offset) : mod.emplace_function(signature, offset));
+
+                    functions.emplace(std::make_pair(name, std::move(signature)), handle);
+
+                    handle.dispatch();
+                    return { generator_error::Success };
+                }
+                case token::Native: {
+                    result = eat_token(lexer, token::Identifier);
+                    if (!result) return result.error;
+
+                    std::string nativeName = std::string(result->value.string);
+
+                    std::string name = std::string(nameRes->value.string);
+
+                    furvm::function_h handle = (pub ? mod.emplace_function(name, signature, std::move(nativeName))
+                                                    : mod.emplace_function(signature, std::move(nativeName)));
+
+                    functions.emplace(std::make_pair(name, std::move(signature)), handle);
+
+                    handle.dispatch();
+                    return { generator_error::Success };
+                }
+                case token::Import: {
+                    throw std::runtime_error("unimplemented");
+                }
+                default:
+                    return { generator_error::UnexpectedToken,
+                        "Unexpected token "s + token_type(result->type) +
+                            ", expected label name, `native`, or `import`" };
+                }
+            }
+
+            if (result->type == token::Type) {
+                auto nameRes = eat_token(lexer, token::Identifier);
+                if (!nameRes) return nameRes.error;
+
+                result = eat_token(lexer, token::EqSign);
+                if (!result) return result.error;
+
+                result = next_token(lexer);
+                if (!result) return result.error;
+                switch (result->type) {
+                case token::Import: {
+                    auto typeNameRes = eat_token(lexer, token::Identifier);
+                    if (!typeNameRes) return typeNameRes.error;
+                    return { generator_error::Success };
+                }
+                case token::Array: {
+                    result = eat_token(lexer, token::Dolar);
+                    if (!result) return result.error;
+
+                    auto typeNameRes = eat_token(lexer, token::Identifier);
+                    if (!typeNameRes) return typeNameRes.error;
+
+                    auto size = eat_token(lexer, token::Unsigned);
+                    if (!size) return size.error;
+
+                    furvm::mod_type_id innerId = 0;
+                    if (auto it = types.find(std::string(typeNameRes->value.string)); it != types.end()) {
+                        innerId = it->second.id();
+                    } else {
+                        return { generator_error::UnknownType,
+                            "Unknown type "s + std::string(typeNameRes->value.string) };
+                    }
+
+                    auto type = mod.emplace_type(innerId, size->value.uint);
+                    types.emplace(std::string(nameRes->value.string), type);
+                    type.dispatch();
+
+                    return { generator_error::Success };
+                }
+                default:
+                    return { generator_error::UnexpectedToken,
+                        "Unexpected token "s + token_type(result->type) +
+                            ", expected either type, `import`, or `array`" };
+                }
+            }
+
+            return { generator_error::UnexpectedToken,
+                "Unexpected token "s + token_type(result->type) + ", expected either `func` or `type`" };
+        }
+
+        case token::Push: {
+            auto result = eat_token(lexer, token::Dolar);
+            if (!result) return result.error;
+            auto typeName = eat_token(lexer, token::Identifier);
+            if (!typeName) return result.error;
+            auto it = types.find(std::string(typeName->value.string));
+            if (it == types.end())
+                return { generator_error::UnexpectedToken, "Unknown type "s + std::string(typeName->value.string) };
+            // TODO: Add support for signed integers
+            auto value = eat_token(lexer, token::Unsigned);
+            if (!value) return result.error;
+            auto type = it->second;
+            switch (type->type) {
+            case furvm::mod_type::S8: {
+                mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushS8));
+                mod.bytecode().push_back(value->value.uint & 0xFF);
+            } break;
+            case furvm::mod_type::U8: {
+                mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushU8));
+                mod.bytecode().push_back(value->value.uint & 0xFF);
+            } break;
+            case furvm::mod_type::S16: {
+                mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushS16));
+                mod.bytecode().push_back(value->value.uint & 0xFF);
+                mod.bytecode().push_back((value->value.uint >> 8) & 0xFF);
+            } break;
+            case furvm::mod_type::U16: {
+                mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushU16));
+                mod.bytecode().push_back(value->value.uint & 0xFF);
+                mod.bytecode().push_back((value->value.uint >> 8) & 0xFF);
+            } break;
+            case furvm::mod_type::S32: {
+                mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushS32));
+                mod.bytecode().push_back(value->value.uint & 0xFF);
+            } break;
+            case furvm::mod_type::U32: {
+                mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushU32));
+                mod.bytecode().push_back(value->value.uint & 0xFF);
+            } break;
+            default:
+                return { generator_error::UnexpectedToken, "Unexpected type "s + std::string(typeName->value.string) };
+            }
+            return { generator_error::Success };
+        }
+
+        case token::Array:
+        case token::Get:
+        case token::Set:
+        case token::Drop:
+        case token::Dup:
+        case token::Swap:
+        case token::Clone:
+        case token::Ref:
+        case token::Add:
+        case token::Sub:
+        case token::Mul:
+        case token::Div:
+        case token::Mod:
+        case token::Eq:
+        case token::Neq:
+        case token::Lt:
+        case token::Gt:
+        case token::Le:
+        case token::Ge:
+        case token::Ptrof:
+        case token::Sizeof:
+        case token::Lenof:
+        case token::Load:
+        case token::Store:
+        case token::Call:
+        case token::Jmp:
+        case token::Jnz:
+        case token::Ret: {
+            auto it = instructions.find(result->type);
+            assert(it != instructions.end());
+            mod.bytecode().push_back(static_cast<furvm::byte>(it->second.fur));
+            switch (it->second.arg) {
+            case instruction::None: break;
+            case instruction::Type: {
+                result = eat_token(lexer, token::Dolar);
+                if (!result) return result.error;
+                result = eat_token(lexer, token::Identifier);
+                if (!result) return result.error;
+                auto type = types.find(std::string(result->value.string));
+                if (type == types.end())
+                    return { generator_error::UnknownType, "Unknown type "s + std::string(result->value.string) };
+                auto id = type->second.id();
+                mod.bytecode().push_back((id >> 0) & 0xFF);
+                mod.bytecode().push_back((id >> 8) & 0xFF);
+                mod.bytecode().push_back((id >> 16) & 0xFF);
+                mod.bytecode().push_back((id >> 24) & 0xFF);
+            } break;
+            case instruction::Constant: {
+                assert(false); // TODO: Unimplemented
+            } break;
+            case instruction::Variable: {
+                result = eat_token(lexer, token::Percent);
+                if (!result) return result.error;
+                result = eat_token(lexer, token::Unsigned);
+                if (!result) return result.error;
+                std::uint16_t var = result->value.uint;
+                mod.bytecode().push_back((var >> 0) & 0xFF);
+                mod.bytecode().push_back((var >> 8) & 0xFF);
+            } break;
+            case instruction::Function: {
+                furvm::function_sig signature;
+                while ((result = next_token(lexer)).error.type == generator_error::Success &&
+                       result->type == token::Dolar) {
+                    result = eat_token(lexer, token::Identifier);
+                    if (!result) return result.error;
+                    auto type = types.find(std::string(result->value.string));
+                    if (type == types.end())
+                        return { generator_error::UnknownType, "Unknown type "s + std::string(result->value.string) };
+                    signature.params.push_back(type->second);
+                }
+                if (!result || result->type != token::Identifier) return result.error;
+                std::string name(result->value.string);
+
+                auto func = functions.find(std::make_pair(name, signature));
+                if (func == functions.end())
+                    return { generator_error::UnknownType, "Unknown type "s + std::string(result->value.string) };
+                auto id = func->second.id();
+                mod.bytecode().push_back((id >> 0) & 0xFF);
+                mod.bytecode().push_back((id >> 8) & 0xFF);
+            } break;
+            case instruction::Label: {
+                result = eat_token(lexer, token::Sha256);
+                if (!result) return result.error;
+                result = eat_token(lexer, token::Identifier);
+                if (!result) return result.error;
+                auto& label  = labels[std::string(result->value.string)];
+                auto  offset = label.offset;
+                if (offset == label_context::INVALID) {
+                    label.unknowns.push_back(mod.bytecode().size() + 1);
+                    mod.bytecode().push_back(0);
+                    return { generator_error::Success };
+                }
+                std::ptrdiff_t jmpOff = static_cast<std::ptrdiff_t>(offset - mod.bytecode().size() - 1);
+                if (jmpOff < std::numeric_limits<std::int8_t>::min() ||
+                    jmpOff > std::numeric_limits<std::int8_t>::max()) {
+                    assert(false); // TODO: Further jumps are not implemented
+                }
+                mod.bytecode().push_back(jmpOff);
+            } break;
+            }
+            return { generator_error::Success };
+        }
+
+        case token::Signed:
+        case token::Unsigned:
+        case token::Colon:
+        case token::Monkey:
+        case token::Dolar:
+        case token::Sha256:
+        case token::Percent:
+        case token::EqSign:
+        case token::Dot:
+        case token::Native:
+        case token::Import:
+        case token::Count: break;
+        }
+        return { generator_error::UnexpectedToken, "Unexpected token "s + token_type(result->type) };
+    }
+};
+
+} // namespace
+
+generator::result generator::generate(lexer lexer) {
+    mod_context     context;
+    generator_error error;
+    do {
+        error = context.generate(lexer);
+    } while (error.type == generator_error::Success);
+    if (error.type == generator_error::Eof) return { { generator_error::Success }, std::move(context.mod) };
+    return { error };
+}
+
+} // namespace furas

furas/src/lexer.cpp

@@ -0,0 +1,98 @@
+#include "furas/lexer.hpp"
+
+#include <cctype>
+#include <unordered_map>
+
+namespace furas {
+
+using namespace std::string_literals;
+
+token_r lexer::next_token() {
+    while (m_cursor < m_content.size() && std::isspace(m_content[m_cursor]) != 0) {
+        if (m_content[m_cursor] == '\n') {
+            m_lineStart = m_cursor + 1;
+            ++m_column;
+        }
+        ++m_cursor;
+    }
+
+    // TODO: Add support for single-line comments
+    // TODO: Add support for multi-line comments
+
+    if (m_cursor >= m_content.size()) return token_r{ lexer_error{ lexer_error::EndOfFile, location() } };
+
+    // TODO: Add support for negative integers (I am positive thanks to stasiu :v:)
+    // TODO: Add support for hexadecimal and binary numeric literals
+    if (std::isdigit(m_content[m_cursor]) != 0) {
+        std::uint64_t num = 0;
+        while (m_cursor < m_content.size() && std::isdigit(m_content[m_cursor]) != 0) {
+            num *= 10;
+            num += m_content[m_cursor++] - '0';
+        }
+        return { num };
+    }
+
+    static std::unordered_map<std::string_view, enum token::type> s_tokens = {
+        { "func", token::Func },
+        { "type", token::Type },
+        { "native", token::Native },
+        { "import", token::Import },
+        { "public", token::Public },
+        { "private", token::Private },
+
+        { "push", token::Push },
+        { "array", token::Array },
+        { "get", token::Get },
+        { "set", token::Set },
+        { "drop", token::Drop },
+        { "dup", token::Dup },
+        { "swap", token::Swap },
+        { "clone", token::Clone },
+        { "ref", token::Ref },
+        { "add", token::Add },
+        { "sub", token::Sub },
+        { "mul", token::Mul },
+        { "div", token::Div },
+        { "mod", token::Mod },
+        { "eq", token::Eq },
+        { "neq", token::Neq },
+        { "lt", token::Lt },
+        { "gt", token::Gt },
+        { "le", token::Le },
+        { "ge", token::Ge },
+        { "ptrof", token::Ptrof },
+        { "sizeof", token::Sizeof },
+        { "lenof", token::Lenof },
+        { "load", token::Load },
+        { "store", token::Store },
+        { "call", token::Call },
+        { "jmp", token::Jmp },
+        { "jnz", token::Jnz },
+        { "ret", token::Ret },
+    };
+
+    if (std::isalnum(m_content[m_cursor]) != 0) {
+        std::size_t begin = m_cursor++;
+        while (m_cursor < m_content.size() && (std::isalnum(m_content[m_cursor]) != 0 || m_content[m_cursor] == '_'))
+            ++m_cursor;
+        std::string_view str = m_content.substr(begin, m_cursor - begin);
+        if (auto it = s_tokens.find(str); it != s_tokens.end()) return { it->second };
+        return { token::Identifier, str };
+    }
+
+    switch (m_content[m_cursor]) {
+    case '@': ++m_cursor; return { token::Monkey };
+    case '$': ++m_cursor; return { token::Dolar };
+    case '#': ++m_cursor; return { token::Sha256 };
+    case '%': ++m_cursor; return { token::Percent };
+    case '=': ++m_cursor; return { token::EqSign };
+    case '.': ++m_cursor; return { token::Dot };
+    case ':': ++m_cursor; return { token::Colon };
+    default:
+        return token_r{
+            lexer_error{ lexer_error::UnknownCharacter, location(), "Unknown character '"s + m_content[m_cursor] + "'" }
+        };
+    }
+}
+
+} // namespace furas

furas/src/main.cpp

@@ -0,0 +1,39 @@
+#include "furas/gen.hpp"
+#include "furas/lexer.hpp"
+
+#include <fstream>
+#include <furvm/module.hpp>
+#include <iostream>
+
+int main(int argc, char** argv) { // NOLINT
+    if (argc < 2) {
+        std::cerr << "feed me more arguments daddy >_<\n";
+        return 1;
+    }
+    if (argc > 2) {
+        std::cerr << "too much O_O\n";
+        return 1;
+    }
+
+    const char* filepath = argv[1];
+
+    std::ifstream file(filepath, std::ios::binary | std::ios::ate | std::ios::in);
+    if (!file.is_open()) {
+        std::cerr << "file won't open >~<\n";
+        return 1;
+    }
+    std::string content;
+    content.resize(file.tellg());
+    file.seekg(0);
+    file.read(content.data(), static_cast<std::streamsize>(content.size()));
+
+    auto result = furas::generator::generate(furas::lexer(filepath, content));
+    if (result.error.type != furas::generator_error::Success) {
+        std::cerr << result.error.message << '\n';
+        return 1;
+    }
+
+    result.mod.serialize(std::cout);
+
+    return 0;
+}

furc/src/back/furvm.cpp

@@ -170,7 +170,7 @@ void furvm_generator::generate_operand(furvm::mod& mod, function_context& ctx, c
         static_assert(sizeof(var) == 2, "sizeof(furvm::variable_t) has changed");
     } break;
     case furlang::ir::operand_t::Integer: {
-        mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushB2I));
+        mod.bytecode().push_back(static_cast<furvm::byte>(furvm::instruction_t::PushS32));
         mod.bytecode().push_back(operand.integer());
     } break;
     case furlang::ir::operand_t::Variable:

furvm/include/furvm/instruction.hpp

@@ -12,11 +12,34 @@ enum class instruction_t : byte {
     NoOperation = 0,
 
     /**
-     * @brief Pushes an integer from a byte onto the stack.
-     *
-     * Pushes an integer constructed from a next byte onto the stack.
+     * @brief Pushes an s8 integer from a byte onto the stack.
      */
-    PushB2I,
+    PushS8,
+
+    /**
+     * @brief Pushes an u8 integer from a byte onto the stack.
+     */
+    PushU8,
+
+    /**
+     * @brief Pushes an s16 integer from two byte onto the stack.
+     */
+    PushS16,
+
+    /**
+     * @brief Pushes an u16 integer from two byte onto the stack.
+     */
+    PushU16,
+
+    /**
+     * @brief Pushes an s32 integer from a byte onto the stack.
+     */
+    PushS32,
+
+    /**
+     * @brief Pushes an u32 integer from a byte onto the stack.
+     */
+    PushU32,
 
     /**
      * @brief Pushes a constant onto the stack.
@@ -38,6 +61,11 @@ enum class instruction_t : byte {
      */
     Get,
 
+    /**
+     * @brief Sets an array element.
+     */
+    Set,
+
     /**
      * @brief Pops top element from the stack.
      */
@@ -48,6 +76,11 @@ enum class instruction_t : byte {
      */
     Duplicate,
 
+    /**
+     * @brief Swaps two top elements of the stack.
+     */
+    Swap,
+
     /**
      * @brief Clones top element on the stack.
      */

furvm/include/furvm/thing.hpp

@@ -433,7 +433,7 @@ public:
      * @return The integer value.
      */
     thing_type::s64 integer() const {
-        switch (m_type.type) {
+        switch (true_type().type) {
         case thing_type::S8: return get<thing_type::s8>();
         case thing_type::S16: return get<thing_type::s16>();
         case thing_type::S32: return get<thing_type::s32>();
@@ -448,11 +448,11 @@ public:
 
     void resize(thing_type::u64 newSize) {
         if (!is(thing_type::Array)) throw bad_thing_access();
-        if (m_type.value.array.size > 0) throw std::runtime_error("cannot resize a static array");
+        if (true_type().value.array.size > 0) throw std::runtime_error("cannot resize a static array");
 
         auto& array = get<union array>();
         if (newSize < 0 || newSize == array.dynamic.size) return;
-        std::size_t innerSize = compute_size_na(*m_type.value.array.type);
+        std::size_t innerSize = compute_size_na(*true_type().value.array.type);
         std::byte*  newData   = new std::byte[innerSize * newSize];
         std::memcpy(newData,
             array.dynamic.data,
@@ -502,6 +502,32 @@ public:
         if (m_type.type != thing_type::Ref || *m_type.value.typeRef != thing.type()) throw bad_thing_access();
         m_data = thing.m_data;
     }
+
+    /**
+     * @brief Self-explainatory.
+     *
+     * TODO: Document
+     */
+    void assign(thing&& thing) {
+        class thing rhs = std::move(thing);
+        if (true_type() != rhs.true_type()) throw std::runtime_error("thing type mismatch");
+        // TODO: Move this to another function
+        switch (true_type().type) {
+        case thing_type::S8:
+        case thing_type::S16:
+        case thing_type::S32:
+        case thing_type::S64:
+        case thing_type::U8:
+        case thing_type::U16:
+        case thing_type::U32:
+        case thing_type::U64: std::memcpy(m_data, rhs.m_data, m_size); return;
+        case thing_type::Ptr:
+        case thing_type::Ref:
+        case thing_type::Array: throw std::runtime_error("unimplemented");
+        case thing_type::Count: break;
+        }
+        throw std::runtime_error("unreachable");
+    }
 private:
     static void copy_list(const thing_type& arrayType, array& dst, const array& src) {
         if (arrayType.type != thing_type::Array || arrayType.value.array.type == nullptr)
@@ -577,7 +603,7 @@ private:
 private:
     template <typename Func>
     decltype(auto) visit_primitive(Func&& func) const {
-        switch (m_type.type) {
+        switch (true_type().type) {
         case thing_type::S8: return std::forward<Func>(func)(get<thing_type::s8>());
         case thing_type::S16: return std::forward<Func>(func)(get<thing_type::s16>());
         case thing_type::S32: return std::forward<Func>(func)(get<thing_type::s32>());
@@ -592,7 +618,7 @@ private:
 
     template <typename Op>
     thing binary_op(const thing& rhs, const Op& op) const {
-        if (thing_type::is_primitive(m_type.type) && thing_type::is_primitive(rhs.m_type.type)) {
+        if (thing_type::is_primitive(true_type().type) && thing_type::is_primitive(true_type().type)) {
             static constexpr enum thing_type::type promotions[8 * 8] = {
                 // S8
                 thing_type::S8,

furvm/include/furvm/thing_allocator.hpp

@@ -69,8 +69,9 @@ public:
     [[nodiscard]] T* allocate(std::size_t count = 1) {
         for (auto it = m_deadThings->begin(); it != m_deadThings->end(); ++it) {
             if (it->second != count) continue;
+            T* data = it->first;
             m_deadThings->erase(it);
-            return it->first;
+            return data;
         }
         return m_arena->allocate<T>(count);
     }

furvm/src/executor.cpp

@@ -10,6 +10,7 @@
 #include <cassert>
 #include <cstdint>
 #include <stdexcept>
+#include <utility>
 #include <vector>
 
 namespace furvm {
@@ -100,7 +101,7 @@ thing_h executor::pop_thing() {
     if (m_frames.top().stackBase >= m_stack.size()) throw stack_underflow();
     thing_h top = std::move(m_stack.top());
     m_stack.pop();
-    return top;
+    return std::move(top);
 }
 
 thing_h executor::thing() const {
@@ -122,7 +123,7 @@ void executor::store_thing(variable_t variable, thing_h&& thing) {
 
 thing_h executor::load_thing(variable_t variable) const {
     const auto& frame = m_frames.top();
-    return frame.variables[variable];
+    return { frame.variables[variable] };
 }
 
 void executor::step() {
@@ -133,9 +134,33 @@ void executor::step() {
     instruction_t instr = static_cast<instruction_t>((*frame.mod).byte(frame.position++));
     switch (instr) {
     case instruction_t::NoOperation: break;
-    case instruction_t::PushB2I: {
-        push_thing({ (struct thing_type){ thing_type::S32 }, m_context->thing_alloc() })->get<int>() =
-            frame.mod->byte(frame.position++);
+    case instruction_t::PushS8: {
+        push_thing({ (struct thing_type){ thing_type::S8 }, m_context->thing_alloc() })->get<thing_type::s8>() =
+            static_cast<thing_type::s8>(frame.mod->byte(frame.position++));
+    } break;
+    case instruction_t::PushU8: {
+        push_thing({ (struct thing_type){ thing_type::U8 }, m_context->thing_alloc() })->get<thing_type::u8>() =
+            static_cast<thing_type::u8>(frame.mod->byte(frame.position++));
+    } break;
+    case instruction_t::PushS16: {
+        thing_type::u16 value = frame.mod->byte(frame.position++);
+        value                |= static_cast<thing_type::u16>(frame.mod->byte(frame.position++) << 8);
+        push_thing({ (struct thing_type){ thing_type::S16 }, m_context->thing_alloc() })->get<thing_type::s16>() =
+            static_cast<thing_type::s16>(value);
+    } break;
+    case instruction_t::PushU16: {
+        thing_type::u16 value = frame.mod->byte(frame.position++);
+        value                |= static_cast<thing_type::u16>(frame.mod->byte(frame.position++) << 8);
+        push_thing({ (struct thing_type){ thing_type::U16 }, m_context->thing_alloc() })->get<thing_type::u16>() =
+            value;
+    } break;
+    case instruction_t::PushS32: {
+        push_thing({ (struct thing_type){ thing_type::S32 }, m_context->thing_alloc() })->get<thing_type::s32>() =
+            static_cast<thing_type::s32>(frame.mod->byte(frame.position++));
+    } break;
+    case instruction_t::PushU32: {
+        push_thing({ (struct thing_type){ thing_type::U32 }, m_context->thing_alloc() })->get<thing_type::u32>() =
+            static_cast<thing_type::u32>(frame.mod->byte(frame.position++));
     } break;
     case instruction_t::Array: {
         mod_type_id typeId = static_cast<mod_type_id>(frame.mod->byte(frame.position)) |
@@ -161,12 +186,24 @@ void executor::step() {
         auto array = pop_thing();
         push_thing(array->at(index->integer()));
     } break;
+    case instruction_t::Set: {
+        auto element = pop_thing();
+        auto index   = pop_thing();
+        auto array   = pop_thing();
+        array->at(index->integer()).assign(std::move(element->clone()));
+    } break;
     case instruction_t::Drop: {
         pop_thing();
     } break;
     case instruction_t::Duplicate: {
         push_thing(thing());
     } break;
+    case instruction_t::Swap: {
+        auto thing1 = pop_thing();
+        auto thing2 = pop_thing();
+        push_thing(std::move(thing1));
+        push_thing(std::move(thing2));
+    } break;
     case instruction_t::Clone: {
         push_thing(std::move(thing()->clone()));
     } break;
@@ -286,7 +323,8 @@ void executor::step() {
         push_frame(frame.mod, *frame.mod->function_at(funcId));
     } break;
     case instruction_t::Jump: {
-        frame.position += ((std::int8_t)frame.mod->byte(frame.position)) + 1;
+        std::int8_t offset = static_cast<std::int8_t>(frame.mod->byte(frame.position++));
+        frame.position    += offset;
     } break;
     case instruction_t::JumpNotZero: {
         byte offset = frame.mod->byte(frame.position++);

furvm/src/main.cpp

@@ -13,7 +13,7 @@
 static void print_thing(const furvm::thing<furvm::thing_allocator>& thing) {
     using namespace furvm;
 
-    switch (thing.type().type) {
+    switch (thing.true_type().type) {
     case thing_type::S8: std::cout << thing.cast_to<thing_type::s16>(); break;
     case thing_type::S16: std::cout << thing.get<thing_type::s16>(); break;
     case thing_type::S32: std::cout << thing.get<thing_type::s32>(); break;
@@ -22,7 +22,7 @@ static void print_thing(const furvm::thing<furvm::thing_allocator>& thing) {
     case thing_type::U16: std::cout << thing.get<thing_type::u16>(); break;
     case thing_type::U32: std::cout << thing.get<thing_type::u32>(); break;
     case thing_type::U64: std::cout << thing.get<thing_type::u64>(); break;
-    case furvm::thing_type::Array:
+    case thing_type::Array:
         std::cout << "{ ";
         for (thing_type::u64 i = 0; i < thing.length(); ++i) {
             if (i > 0) std::cout << ", ";
@@ -37,20 +37,20 @@ static void print_thing(const furvm::thing<furvm::thing_allocator>& thing) {
 int main(int argc, char** argv) {
     auto context = std::make_shared<furvm::context>();
 
-#if 0 // NOLINT
+#if 1 // NOLINT
     if (argc != 2) {
         std::cerr << "Usage: " << argv[0] << " <mod.fmod>\n";
         return 1;
     }
 
-
     std::ifstream file(argv[1]);
     if (!file.is_open()) {
         std::cerr << "Failed to open " << argv[1] << '\n';
         return 1;
     }
 
-    furvm::mod_h mod = context->emplace("main", std::move(furvm::mod::load(file)));
+    furvm::mod_h mod      = context->emplace("main", std::move(furvm::mod::load(file)));
+    auto         mainFunc = mod->function_at("main", furvm::function_sig{});
 #else
     static const furvm::byte s_bytecode[] = {
         static_cast<furvm::byte>(furvm::instruction_t::Array),
@@ -74,11 +74,10 @@ int main(int argc, char** argv) {
     auto mainFunc  = mod->emplace_function("main", furvm::function_sig{}, 0);
     mod->emplace_function(furvm::function_sig{ { u64Type }, u64Type }, "print").dispatch();
 #endif
-    mod->set_native_function("print", [](furvm::executor& executor) {
-        auto arg = std::move(*executor.load_thing(0));
-        print_thing(arg);
+    mod->set_native_function("println", [](furvm::executor& executor) {
+        auto arg = executor.load_thing(0);
+        print_thing(*arg);
         std::cout << '\n';
-        executor.push_thing(std::move(arg));
     });
 
     furvm::executor_h executor = context->emplace_executor(context);

furvm/src/module.cpp

@@ -25,6 +25,7 @@ std::ostream& mod::serialize(std::ostream& os) const {
         }
 
         auto type = *m_types.at(id);
+        detail::serialize(os, static_cast<std::uint8_t>(type.type));
         switch (type.type) {
         case mod_type::S8:
         case mod_type::S16:
@@ -67,7 +68,7 @@ std::ostream& mod::serialize(std::ostream& os) const {
         }
 
         // Function signature
-        detail::serialize(os, func->signature().params.size());
+        detail::serialize(os, static_cast<std::uint32_t>(func->signature().params.size()));
         for (std::uint32_t i = 0; i < func->signature().params.size(); ++i)
             detail::serialize(os, func->signature().params[i].id());
 
@@ -116,7 +117,10 @@ mod mod::load(std::istream& is) {
         case mod_type::U8:
         case mod_type::U16:
         case mod_type::U32:
-        case mod_type::U64: break;
+        case mod_type::U64: {
+            mod_type theType = { (enum mod_type::type)type };
+            mod.emplace_type(id, theType).dispatch();
+        } break;
         case mod_type::Ptr: {
             mod_type_id typeId = 0;
             detail::load(is, typeId);
@@ -165,17 +169,17 @@ mod mod::load(std::istream& is) {
             decltype(std::declval<function>().position()) offset = 0;
             detail::load(is, offset);
             if (name.empty())
-                mod.emplace_function(std::move(name), id, std::move(signature), offset).dispatch();
-            else
                 mod.emplace_function(id, std::move(signature), offset).dispatch();
+            else
+                mod.emplace_function(std::move(name), id, std::move(signature), offset).dispatch();
         } break;
         case function_t::Native: {
             std::string native;
             detail::load(is, native);
             if (name.empty())
-                mod.emplace_function(std::move(name), id, std::move(signature), std::move(native)).dispatch();
-            else
                 mod.emplace_function(id, std::move(signature), std::move(native)).dispatch();
+            else
+                mod.emplace_function(std::move(name), id, std::move(signature), std::move(native)).dispatch();
         } break;
         case function_t::Import: {
             std::string modName;

furvm/test/test.cpp

@@ -1,3 +1,28 @@
+#include "furlang/arena.hpp"
+#include "furvm/furvm.hpp"
+#include "furvm/thing.hpp"
+#include "furvm/thing_allocator.hpp"
+
 #include "gtest/gtest.h" // IWYU pragma: keep
 
-namespace {}
+namespace {
+
+// TODO: Basic program tests (e.g. for loops)
+
+TEST(Things, Ops) {
+    furlang::arena                    arena;
+    furvm::thing_allocator<std::byte> alloc{ arena };
+
+    furvm::thing lhs{ furvm::thing_type{ furvm::thing_type::U32 }, alloc };
+    lhs.get<furvm::thing_type::u32>() = 6;
+    furvm::thing rhs{ furvm::thing_type{ furvm::thing_type::U32 }, alloc };
+    rhs.get<furvm::thing_type::u32>() = 7;
+
+    auto res = lhs.add(rhs);
+    ASSERT_EQ(res.type().type, furvm::thing_type::U32);
+    EXPECT_EQ(lhs.get<furvm::thing_type::u32>(), 6);
+    EXPECT_EQ(rhs.get<furvm::thing_type::u32>(), 7);
+    EXPECT_EQ(res.get<furvm::thing_type::u32>(), 6 + 7);
+}
+
+} // namespace