chore(disfuras): implement basic disfuras

Add a basic view (span from C++20).

CMakeLists.txt

@@ -15,6 +15,7 @@ add_subdirectory(furlang)
 add_subdirectory(furvm)
 add_subdirectory(furc)
 add_subdirectory(furas)
+add_subdirectory(disfuras)
 
 if(DOXYGEN_FOUND)
     set(DOXYGEN_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/docs)

disfuras/CMakeLists.txt

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

disfuras/src/main.cpp

@@ -0,0 +1,238 @@
+#include "furvm/instruction.hpp"
+#include "furvm/module.hpp"
+
+#include <cassert>
+#include <exception>
+#include <fstream>
+#include <iostream>
+#include <stdexcept>
+#include <string>
+#include <unordered_map>
+#include <unordered_set>
+
+using namespace std::string_literals;
+
+void print_type(const furvm::mod_type& type, const furvm::mod& mod) {
+    switch (type.type) {
+    case furvm::mod_type::S8: std::cout << "$s8"; return;
+    case furvm::mod_type::S16: std::cout << "$s16"; return;
+    case furvm::mod_type::S32: std::cout << "$s32"; return;
+    case furvm::mod_type::S64: std::cout << "$s64"; return;
+    case furvm::mod_type::U8: std::cout << "$u8"; return;
+    case furvm::mod_type::U16: std::cout << "$u16"; return;
+    case furvm::mod_type::U32: std::cout << "$u32"; return;
+    case furvm::mod_type::U64: std::cout << "$u64"; return;
+    case furvm::mod_type::Ptr:
+        std::cout << "ptr ";
+        print_type(*mod.type_at(type.value.typeRef), mod);
+        return;
+    case furvm::mod_type::Ref:
+        std::cout << "ref ";
+        print_type(*mod.type_at(type.value.typeRef), mod);
+        return;
+    case furvm::mod_type::Array:
+        std::cout << "array ";
+        print_type(*mod.type_at(type.value.array.typeId), mod);
+        if (type.value.array.size == 0) {
+            std::cout << " dynamic";
+        } else {
+            std::cout << ' ' << type.value.array.size;
+        }
+        return;
+    case furvm::mod_type::Import: std::cout << "import " << type.value.imprt.modId << "::t" << type.value.imprt.typeId;
+    case furvm::mod_type::Count: break;
+    }
+    assert(false);
+}
+
+static const char* typeNames[furvm::instruction::Count] = {
+    // NoOperation:
+    "nop",
+    // PushS8:
+    "push $__t0",
+    // PushU8:
+    "push $__t1",
+    // PushS16:
+    "push $__t2",
+    // PushU16:
+    "push $__t3",
+    // PushS32:
+    "push $__t4",
+    // PushU32:
+    "push $__t5",
+    // PushConstant:
+    "push",
+    // Array:
+    "array",
+    // Get:
+    "get",
+    // Set:
+    "set",
+    // Drop:
+    "drop",
+    // Duplicate:
+    "dup",
+    // Swap:
+    "swap",
+    // Clone:
+    "clone",
+    // Reference:
+    "ref",
+    // Add:
+    "add",
+    // Sub:
+    "sub",
+    // Mul:
+    "mul",
+    // Div:
+    "div",
+    // Mod:
+    "mod",
+    // Equals:
+    "eq",
+    // NotEquals:
+    "ne",
+    // LessThan:
+    "lt",
+    // GreaterThan:
+    "gt",
+    // LessEqual:
+    "le",
+    // GreaterEqual:
+    "ge",
+    // Pointerof:
+    "pointerof",
+    // Sizeof:
+    "sizeof",
+    // Lengthof:
+    "lengthof",
+    // Load:
+    "load",
+    // Store:
+    "store",
+    // Call:
+    "call",
+    // Jump:
+    "jmp",
+    // JumpNotZero:
+    "jnz",
+    // Return:
+    "ret",
+};
+
+int main(int argc, char** argv) {
+    if (argc != 2) {
+        std::cerr << "Usage: " << argv[0] << " <module.fmod>\n";
+        return 1;
+    }
+
+    try {
+        std::ifstream file(argv[1], std::ios::binary | std::ios::in);
+
+        furvm::mod mod;
+        try {
+            mod = furvm::mod::load(file);
+        } catch (std::exception ex) {
+            std::cerr << "Failed to load module " << argv[1] << ": " << ex.what() << '\n';
+            return 1;
+        }
+
+        std::cout << "; Generated with disfuras\n";
+
+        std::cout << "; Types:\n";
+        for (const auto& ptr : mod.types()) {
+            const auto& [header, type] = *ptr;
+            std::cout << "type __t" << header.id() << " = ";
+            print_type(type, mod);
+            std::cout << '\n';
+        }
+
+        std::unordered_map<furvm::function_id, std::string> funcNames;
+        std::size_t                                         unnamedCounter = 0;
+
+        std::unordered_map<std::size_t, std::string> labels;
+        std::unordered_set<std::size_t>              deadLabels;
+        std::size_t                                  labelCounter = 0;
+
+        std::cout << "; Functions:\n";
+        for (const auto& ptr : mod.functions()) {
+            const auto& [header, func] = *ptr;
+            auto it                    = mod.function_map().find(header.id());
+            if (it != mod.function_map().end()) {
+                std::cout << "public func " << (funcNames[header.id()] = it->second.first);
+            } else {
+                std::cout << "func " << (funcNames[header.id()] = "__f"s + std::to_string(unnamedCounter++));
+            }
+            for (const auto& param : func.signature().params) {
+                std::cout << " $__t" << param.id();
+            }
+
+            std::cout << " = ";
+            if (func.signature().returnType.has_value())
+                std::cout << "$__t" << func.signature().returnType->id() << ' '; // NOLINT
+            switch (func.type()) {
+            case furvm::function_t::Normal: {
+                auto it = labels.find(func.position());
+                if (it == labels.end()) it = labels.emplace(func.position(), funcNames[header.id()]).first;
+                deadLabels.insert(func.position());
+                std::cout << '#' << it->second;
+            } break;
+            case furvm::function_t::Native: {
+                std::cout << "native " << func.native();
+            } break;
+            case furvm::function_t::Import: {
+                std::cout << "import " << func.imp().mod << "::" << func.imp().function;
+            } break;
+            }
+            std::cout << '\n';
+        }
+
+        std::cout << "; Bytecode:\n";
+        // Label pass:
+        for (std::size_t off = 0; off < mod.bytecode().size();) {
+            furvm::instruction instr{};
+            off += instr.read(mod.bytecode_view().subview(off));
+            if (instr.type != furvm::instruction::Jump && instr.type != furvm::instruction::JumpNotZero) continue;
+            if (labels.find(off + instr.arg.s8) == labels.end())
+                labels[off + instr.arg.s8] = "__l"s + std::to_string(labelCounter++);
+            deadLabels.insert(off + instr.arg.s8);
+        }
+        // Actual printing pass:
+        for (std::size_t off = 0; off < mod.bytecode().size();) {
+            if (auto it = labels.find(off); it != labels.end()) {
+                std::cout << it->second << ":\n";
+                deadLabels.erase(off);
+            }
+            furvm::instruction instr{};
+            off += instr.read(mod.bytecode_view().subview(off));
+            std::cout << "    " << typeNames[instr.type];
+            switch (instr.arg.type) {
+            case furvm::instruction_argument::None: break;
+            case furvm::instruction_argument::S8: std::cout << ' ' << std::to_string(instr.arg.s8); break;
+            case furvm::instruction_argument::U8: std::cout << ' ' << std::to_string(instr.arg.u8); break;
+            case furvm::instruction_argument::S16: std::cout << ' ' << instr.arg.s16; break;
+            case furvm::instruction_argument::U16: std::cout << ' ' << instr.arg.u16; break;
+            case furvm::instruction_argument::S32: std::cout << ' ' << std::to_string(instr.arg.s8); break;
+            case furvm::instruction_argument::U32: std::cout << ' ' << std::to_string(instr.arg.u8); break;
+            case furvm::instruction_argument::Constant: throw std::runtime_error("unimplemented");
+            case furvm::instruction_argument::Type: std::cout << " $__t" << instr.arg.u16; break;
+            case furvm::instruction_argument::Variable: std::cout << " %" << instr.arg.u16; break;
+            case furvm::instruction_argument::Function: std::cout << ' ' << funcNames[instr.arg.s16]; break;
+            case furvm::instruction_argument::Offset: std::cout << " #" << labels[instr.arg.s16]; break;
+            case furvm::instruction_argument::Count: break;
+            }
+            std::cout << '\n';
+        }
+        if (!deadLabels.empty()) {
+            std::cerr << "Malformed module: invalid jumps\n";
+            return 1;
+        }
+    } catch (const std::exception& ex) {
+        std::cerr << "Exception uncaught: " << ex.what() << '\n';
+        return 1;
+    }
+
+    return 0;
+}
+
+// TODO: Disassemble into FIR (furlang's IR)

furas/examples/simple.furas

@@ -0,0 +1,7 @@
+func println $s32 = native println
+
+public func main = #main
+main:
+    push $s32 69
+    call $s32 println
+    ret

furas/src/gen.cpp

@@ -21,48 +21,36 @@ 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 = {
+std::unordered_map<enum token::type, furvm::instruction_t> instructions = {
-    { token::Array, { furvm::instruction_t::Array, instruction::Type } },
+    { token::Array, furvm::instruction_t::Array },
-    { token::Get, { furvm::instruction_t::Get } },
+    { token::Get, furvm::instruction_t::Get },
-    { token::Set, { furvm::instruction_t::Set } },
+    { token::Set, furvm::instruction_t::Set },
-    { token::Drop, { furvm::instruction_t::Drop } },
+    { token::Drop, furvm::instruction_t::Drop },
-    { token::Dup, { furvm::instruction_t::Duplicate } },
+    { token::Dup, furvm::instruction_t::Duplicate },
-    { token::Swap, { furvm::instruction_t::Swap } },
+    { token::Swap, furvm::instruction_t::Swap },
-    { token::Clone, { furvm::instruction_t::Clone } },
+    { token::Clone, furvm::instruction_t::Clone },
-    { token::Ref, { furvm::instruction_t::Reference } },
+    { token::Ref, furvm::instruction_t::Reference },
-    { token::Add, { furvm::instruction_t::Add } },
+    { token::Add, furvm::instruction_t::Add },
-    { token::Sub, { furvm::instruction_t::Sub } },
+    { token::Sub, furvm::instruction_t::Sub },
-    { token::Mul, { furvm::instruction_t::Mul } },
+    { token::Mul, furvm::instruction_t::Mul },
-    { token::Div, { furvm::instruction_t::Div } },
+    { token::Div, furvm::instruction_t::Div },
-    { token::Mod, { furvm::instruction_t::Mod } },
+    { token::Mod, furvm::instruction_t::Mod },
-    { token::Eq, { furvm::instruction_t::Equals } },
+    { token::Eq, furvm::instruction_t::Equals },
-    { token::Neq, { furvm::instruction_t::NotEquals } },
+    { token::Neq, furvm::instruction_t::NotEquals },
-    { token::Lt, { furvm::instruction_t::LessThan } },
+    { token::Lt, furvm::instruction_t::LessThan },
-    { token::Gt, { furvm::instruction_t::GreaterThan } },
+    { token::Gt, furvm::instruction_t::GreaterThan },
-    { token::Le, { furvm::instruction_t::LessEqual } },
+    { token::Le, furvm::instruction_t::LessEqual },
-    { token::Ge, { furvm::instruction_t::GreaterEqual } },
+    { token::Ge, furvm::instruction_t::GreaterEqual },
-    { token::Ptrof, { furvm::instruction_t::Pointerof } },
+    { token::Ptrof, furvm::instruction_t::Pointerof },
-    { token::Sizeof, { furvm::instruction_t::Sizeof } },
+    { token::Sizeof, furvm::instruction_t::Sizeof },
-    { token::Lenof, { furvm::instruction_t::Lengthof } },
+    { token::Lenof, furvm::instruction_t::Lengthof },
-    { token::Load, { furvm::instruction_t::Load, instruction::Variable } },
+    { token::Load, furvm::instruction_t::Load },
-    { token::Store, { furvm::instruction_t::Store, instruction::Variable } },
+    { token::Store, furvm::instruction_t::Store },
-    { token::Call, { furvm::instruction_t::Call, instruction::Function } },
+    { token::Call, furvm::instruction_t::Call },
-    { token::Jmp, { furvm::instruction_t::Jump, instruction::Label } },
+    { token::Jmp, furvm::instruction_t::Jump },
-    { token::Jnz, { furvm::instruction_t::JumpNotZero, instruction::Label } },
+    { token::Jnz, furvm::instruction_t::JumpNotZero },
-    { token::Ret, { furvm::instruction_t::Return } },
+    { token::Ret, furvm::instruction_t::Return },
 };
 // NOLINTEND
 
@@ -214,12 +202,12 @@ struct mod_context {
                 handle.dispatch();
             }
             for (auto unknown : label.unknowns) {
-                std::ptrdiff_t jmpOff = static_cast<std::ptrdiff_t>(label.offset - unknown);
+                const auto jmpOff = static_cast<std::ptrdiff_t>(label.offset) - static_cast<std::ptrdiff_t>(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;
+                mod.bytecode()[unknown - 1] = static_cast<std::int8_t>(jmpOff);
             }
             label.functions = {};
             label.unknowns  = {};
@@ -453,10 +441,11 @@ struct mod_context {
         case token::Ret: {
             auto it = instructions.find(result->type);
             assert(it != instructions.end());
-            mod.bytecode().push_back(static_cast<furvm::byte>(it->second.fur));
+            furvm::instruction instr{ it->second };
-            switch (it->second.arg) {
+            instr.arg.type = furvm::instruction::s_arguments[instr.type];
-            case instruction::None: break;
+            switch (instr.arg.type) {
-            case instruction::Type: {
+            case furvm::instruction_argument::None: break;
+            case furvm::instruction_argument::Type: {
                 result = eat_token(lexer, token::Dolar);
                 if (!result) return result.error;
                 result = eat_token(lexer, token::Identifier);
@@ -464,25 +453,19 @@ struct mod_context {
                 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();
+                instr.arg.u32 = 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: {
+            case furvm::instruction_argument::Constant: {
                 assert(false); // TODO: Unimplemented
             } break;
-            case instruction::Variable: {
+            case furvm::instruction_argument::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;
+                instr.arg.u16 = result->value.uint;
-                mod.bytecode().push_back((var >> 0) & 0xFF);
-                mod.bytecode().push_back((var >> 8) & 0xFF);
             } break;
-            case instruction::Function: {
+            case furvm::instruction_argument::Function: {
                 furvm::function_sig signature;
                 while ((result = next_token(lexer)).error.type == generator_error::Success &&
                        result->type == token::Dolar) {
@@ -499,11 +482,10 @@ struct mod_context {
                 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();
+                auto id       = func->second.id();
-                mod.bytecode().push_back((id >> 0) & 0xFF);
+                instr.arg.u16 = id;
-                mod.bytecode().push_back((id >> 8) & 0xFF);
             } break;
-            case instruction::Label: {
+            case furvm::instruction_argument::Offset: {
                 result = eat_token(lexer, token::Sha256);
                 if (!result) return result.error;
                 result = eat_token(lexer, token::Identifier);
@@ -511,18 +493,28 @@ struct mod_context {
                 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);
+                    label.unknowns.push_back(mod.bytecode().size() + 2);
-                    mod.bytecode().push_back(0);
+                    instr.arg.s8 = 0;
-                    return { generator_error::Success };
+                    break;
                 }
-                std::ptrdiff_t jmpOff = static_cast<std::ptrdiff_t>(offset - mod.bytecode().size() - 1);
+                const auto jmpOff =
+                    static_cast<std::ptrdiff_t>(offset) - static_cast<std::ptrdiff_t>(mod.bytecode().size()) - 2;
                 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);
+                instr.arg.s8 = static_cast<std::int8_t>(jmpOff);
             } break;
+            case furvm::instruction_argument::S8:
+            case furvm::instruction_argument::U8:
+            case furvm::instruction_argument::S16:
+            case furvm::instruction_argument::U16:
+            case furvm::instruction_argument::S32:
+            case furvm::instruction_argument::U32:
+            case furvm::instruction_argument::Count:
+            default: throw std::runtime_error("unreachable");
             }
+            instr.write(mod.bytecode());
             return { generator_error::Success };
         }
 

furc/include/furc/middle/ir.hpp

@@ -1,5 +1,5 @@
-#ifndef FURC_BACK_IR_HPP
+#ifndef FURC_MIDDLE_IR_HPP
-#define FURC_BACK_IR_HPP
+#define FURC_MIDDLE_IR_HPP
 
 #include "furc/front/ast.hpp"
 #include "furlang/arena.hpp"
@@ -403,4 +403,4 @@ private:
 
 } // namespace furc
 
-#endif // FURC_BACK_IR_HPP
+#endif // FURC_MIDDLE_IR_HPP

furc/src/main.cpp

@@ -1,6 +1,6 @@
-#include "furc/back/ir.hpp"
 #include "furc/front/lexer.hpp"
 #include "furc/front/parser.hpp"
+#include "furc/middle/ir.hpp"
 #include "furlang/arena.hpp"
 
 int main(void) {

furc/src/middle/ir.cpp

@@ -1,4 +1,4 @@
-#include "furc/back/ir.hpp"
+#include "furc/middle/ir.hpp"
 
 #include "furc/front/ast.hpp"
 

furlang/include/furlang/view.hpp

@@ -0,0 +1,38 @@
+#ifndef FURLANG_VIEW_HPP
+#define FURLANG_VIEW_HPP
+
+#include <algorithm>
+#include <cstddef>
+#include <limits>
+#include <stdexcept>
+
+namespace furlang {
+
+template <typename T>
+class view {
+public:
+    constexpr view() noexcept = default;
+
+    constexpr view(const T* data, std::size_t size) noexcept
+      : m_data(data), m_size(size) {}
+public:
+    constexpr view subview(std::size_t offset, std::size_t count = std::numeric_limits<std::size_t>::max()) {
+        if (count > 0 && offset >= m_size) throw std::runtime_error("offset too large");
+        return { m_data + offset, std::min(m_size - offset, count) };
+    }
+
+    const T& operator[](std::size_t offset) const {
+        if (offset >= m_size) throw std::runtime_error("out of bounds");
+        return m_data[offset];
+    }
+
+    constexpr const T*    data() const { return m_data; }
+    constexpr std::size_t size() const { return m_size; }
+private:
+    const T*    m_data = nullptr;
+    std::size_t m_size = 0;
+};
+
+} // namespace furlang
+
+#endif // FURLANG_VIEW_HPP

furvm/include/furvm/fwd.hpp

@@ -75,11 +75,7 @@ class constant;
 
 // instruction.hpp
 
-/**
+struct instruction_argument;
- * @enum instruction_t
- * @brief Furvm's instruction type.
- */
-enum class instruction_t : byte;
 
 /**
  * @struct instruction

furvm/include/furvm/instruction.hpp

@@ -1,221 +1,100 @@
 #ifndef FURVM_INSTRUCTION_HPP
 #define FURVM_INSTRUCTION_HPP
 
+#include "furlang/view.hpp"
 #include "furvm/fwd.hpp"
 
+#include <cstddef>
+#include <vector>
+
 namespace furvm {
 
-enum class instruction_t : byte {
+struct instruction_argument {
-    /**
+    enum type_e {
-     * @brief No operation.
+        None = 0,
-     */
+        S8,
-    NoOperation = 0,
+        U8,
+        S16,
+        U16,
+        S32,
+        U32,
+        Constant,
+        Type,
+        Variable,
+        Function,
+        Offset,
 
-    /**
+        Count,
-     * @brief Pushes an s8 integer from a byte onto the stack.
+    } type;
-     */
+    union {
-    PushS8,
+        std::int8_t   s8;
+        std::uint8_t  u8;
+        std::int16_t  s16;
+        std::uint16_t u16;
+        std::int32_t  s32;
+        std::uint32_t u32;
+    };
 
-    /**
+    static const std::size_t s_sizes[Count];
-     * @brief Pushes an u8 integer from a byte onto the stack.
+    static const bool        s_signedness[Count];
-     */
-    PushU8,
 
-    /**
+    std::size_t size() const { return s_sizes[type]; }
-     * @brief Pushes an s16 integer from two byte onto the stack.
+    bool        is_signed() const { return s_signedness[type]; }
-     */
-    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.
-     *
-     * Pushes a constant from the constant pool denoted by two next bytes in little-endian onto the stack.
-     */
-    PushConstant,
-
-    /**
-     * @brief Pushes a new array onto the stack.
-     *
-     * Type is the next 4 bytes in little-endian.
-     * If the type is dynamic the array's size will be popped off of the stack.
-     */
-    Array,
-
-    /**
-     * @brief Pushes an element from an array onto the stack.
-     */
-    Get,
-
-    /**
-     * @brief Sets an array element.
-     */
-    Set,
-
-    /**
-     * @brief Pops top element from the stack.
-     */
-    Drop,
-
-    /**
-     * @brief Duplicates top element on the stack.
-     */
-    Duplicate,
-
-    /**
-     * @brief Swaps two top elements of the stack.
-     */
-    Swap,
-
-    /**
-     * @brief Clones top element on the stack.
-     */
-    Clone,
-
-    /**
-     * @brief Pushes a new reference onto the stack.
-     *
-     * Pops the top thing from the stack and pushes its reference.
-     */
-    Reference,
-
-    /**
-     * @brief Adds two things together on the stack.
-     */
-    Add,
-
-    /**
-     * @brief Subtracts two things together on the stack.
-     */
-    Sub,
-
-    /**
-     * @brief Multiplies two things together on the stack.
-     */
-    Mul,
-
-    /**
-     * @brief Divides two things together on the stack.
-     */
-    Div,
-
-    /**
-     * @brief Modulos two things together on the stack.
-     */
-    Mod,
-
-    /**
-     * @brief Compares two top-most things from the stack for equality.
-     */
-    Equals,
-
-    /**
-     * @brief Compares two top-most things from the stack for inequality.
-     */
-    NotEquals,
-
-    /**
-     * @brief Compares if the first top-most thing is less than the second top-most thing.
-     */
-    LessThan,
-
-    /**
-     * @brief Compares if the first top-most thing is greater than the second top-most thing.
-     */
-    GreaterThan,
-
-    /**
-     * @brief Compares if the first top-most thing is less than or equal to the second top-most thing.
-     */
-    LessEqual,
-
-    /**
-     * @brief Compares if the first top-most thing is greater than or equal to the second top-most thing.
-     */
-    GreaterEqual,
-
-    /**
-     * @brief Pushes a pointer of popped-off thing onto the stack.
-     */
-    Pointerof,
-
-    /**
-     * @brief Pushes a size of popped-off thing onto the stack.
-     */
-    Sizeof,
-
-    /**
-     * @brief Pushes a length of popped-off thing onto the stack.
-     */
-    Lengthof,
-
-    /**
-     * @brief Pushes a variable onto the stack.
-     *
-     * Fetches a variable denoted by next two bytes in little-endian and pushes it onto the stack.
-     */
-    Load,
-
-    /**
-     * @brief Stores an element from the stack in a variable.
-     *
-     * Pops a thing from the stack and stores it in a variable denoted by next two bytes in little-endian.
-     */
-    Store,
-
-    /**
-     * @brief Calls a function.
-     *
-     * Calls a function denoted by next two bytes in little-endian from current frame's module.
-     */
-    Call,
-
-    /**
-     * @brief Jumps to an instruction relative to the current instruction.
-     *
-     * Jumps to an instruction relative to the current instruction with offset denoted by next byte.
-     */
-    Jump,
-
-    /**
-     * @brief Jumps to an instruction relative to the current instruction if top thing on the stack is not zero.
-     *
-     * Jumps to an instruction relative to the current instruction with offset denoted by next byte if the top thing on
-     * the stack is not zero (is true).
-     */
-    JumpNotZero,
-
-    /**
-     * @brief Pops the current call frame.
-     */
-    Return,
 };
 
+using instruction_argument_t = instruction_argument::type_e;
+
 struct instruction {
-    instruction_t type; /**< Instruction type. */
+    enum type_e : byte {
+        NoOperation = 0,
+        PushS8,
+        PushU8,
+        PushS16,
+        PushU16,
+        PushS32,
+        PushU32,
+        PushConstant,
+        Array,
+        Get,
+        Set,
+        Drop,
+        Duplicate,
+        Swap,
+        Clone,
+        Reference,
+        Add,
+        Sub,
+        Mul,
+        Div,
+        Mod,
+        Equals,
+        NotEquals,
+        LessThan,
+        GreaterThan,
+        LessEqual,
+        GreaterEqual,
+        Pointerof,
+        Sizeof,
+        Lengthof,
+        Load,
+        Store,
+        Call,
+        Jump,
+        JumpNotZero,
+        Return,
 
-    /**
+        Count,
-     * @brief Instruction value.
+    } type;
-     */
+    instruction_argument arg;
-    union value {
+
-        constant_index constant; /**< Constant instruction argument. */
+    static const instruction_argument_t s_arguments[Count];
-    } value;                     /**< Instruction value. */
+
+    std::size_t read(furlang::view<std::uint8_t> in);
+    std::size_t write(std::vector<std::uint8_t>& out) const;
 };
 
+using instruction_t = instruction::type_e;
+
 } // namespace furvm
 
 #endif // FURVM_INSTRUCTION_HPP

furvm/include/furvm/module.hpp

@@ -2,6 +2,7 @@
 #define FURVM_MODULE_HPP
 
 #include "furlang/utility/hash.hpp"
+#include "furlang/view.hpp"
 #include "furvm/function.hpp"
 #include "furvm/fwd.hpp"
 #include "furvm/handle.hpp"
@@ -195,7 +196,7 @@ public:
      *
      * @return A constant reference to the bytecode.
      */
-    constexpr const bytecode_t& bytecode() const { return m_bytecode; }
+    furlang::view<std::uint8_t> bytecode_view() const { return { m_bytecode.data(), m_bytecode.size() }; }
 public:
     /**
      * @brief Emplaces a function in the module's function container.
@@ -284,6 +285,10 @@ public:
             m_functionMap.erase(it);
         }
     }
+
+    const handle_container<function_h>& functions() const { return m_functions; }
+
+    const auto& function_map() const { return m_functionMap; }
 public:
     template <typename NameFwd, typename Func>
     void set_native_function(NameFwd&& name, Func&& func) {
@@ -341,6 +346,8 @@ public:
     void erase_type(Args&&... args) {
         m_types.erase(std::forward<Args>(args)...);
     }
+
+    const handle_container<mod_type_h>& types() const { return m_types; }
 public:
     /**
      * @brief Prints the module in a bytecode form to an output stream.

furvm/src/executor.cpp

@@ -1,5 +1,6 @@
 #include "furvm/executor.hpp"
 
+#include "furlang/view.hpp"
 #include "furvm/context.hpp" // IWYU pragma: keep
 #include "furvm/exceptions.hpp"
 #include "furvm/function.hpp" // IWYU pragma: keep
@@ -131,45 +132,36 @@ void executor::step() {
 
     struct frame& frame = m_frames.top();
 
-    instruction_t instr = static_cast<instruction_t>((*frame.mod).byte_at(frame.position++));
+    instruction instr{};
-    switch (instr) {
+    frame.position += instr.read(frame.mod->bytecode_view().subview(frame.position));
+    switch (instr.type) {
     case instruction_t::NoOperation: break;
     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_at(frame.position++));
+            instr.arg.s8;
     } 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_at(frame.position++));
+            instr.arg.u8;
     } break;
     case instruction_t::PushS16: {
-        thing_type::u16 value = frame.mod->byte_at(frame.position++);
-        value                |= static_cast<thing_type::u16>(frame.mod->byte_at(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);
+            instr.arg.s16;
     } break;
     case instruction_t::PushU16: {
-        thing_type::u16 value = frame.mod->byte_at(frame.position++);
-        value                |= static_cast<thing_type::u16>(frame.mod->byte_at(frame.position++) << 8);
         push_thing({ (struct thing_type){ thing_type::U16 }, m_context->thing_alloc() })->get<thing_type::u16>() =
-            value;
+            instr.arg.u16;
     } 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_at(frame.position++));
+            instr.arg.s8; // NOLINT
     } 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_at(frame.position++));
+            static_cast<thing_type::u32>(instr.arg.u8);
     } break;
     case instruction_t::Array: {
-        mod_type_id typeId = static_cast<mod_type_id>(frame.mod->byte_at(frame.position)) |
+        const auto& type = *mod_to_thing_type(frame.mod, *frame.mod->type_at(instr.arg.u32));
-                             (static_cast<mod_type_id>(frame.mod->byte_at(frame.position + 1)) << 8) |
-                             (static_cast<mod_type_id>(frame.mod->byte_at(frame.position + 2)) << 16) |
-                             (static_cast<mod_type_id>(frame.mod->byte_at(frame.position + 3)) << 24);
-        frame.position    += 4;
-
-        const auto& type = *mod_to_thing_type(frame.mod, *frame.mod->type_at(typeId));
         if (type.type != thing_type::Array || type.value.array.type == nullptr || type.value.array.type == &type)
             throw std::runtime_error("invalid array type");
 
@@ -304,31 +296,19 @@ void executor::step() {
         length->get<thing_type::u64>() = thing->length();
     } break;
     case instruction_t::Load: {
-        variable_t variable = static_cast<std::uint16_t>(frame.mod->byte_at(frame.position)) |
+        push_thing(load_thing(instr.arg.u16));
-                              (static_cast<std::uint16_t>(frame.mod->byte_at(frame.position + 1)) << 8);
-        frame.position     += 2;
-        push_thing(load_thing(variable));
     } break;
     case instruction_t::Store: {
-        variable_t variable = static_cast<std::uint16_t>(frame.mod->byte_at(frame.position)) |
+        store_thing(instr.arg.u16, std::move(pop_thing()));
-                              (static_cast<std::uint16_t>(frame.mod->byte_at(frame.position + 1)) << 8);
-        frame.position     += 2;
-        store_thing(variable, std::move(pop_thing()));
     } break;
     case instruction_t::Call: {
-        function_id funcId = static_cast<std::uint16_t>(frame.mod->byte_at(frame.position)) |
+        push_frame(frame.mod, *frame.mod->function_at(instr.arg.u16));
-                             (static_cast<std::uint16_t>(frame.mod->byte_at(frame.position + 1)) << 8);
-        frame.position    += 2;
-        push_frame(frame.mod, *frame.mod->function_at(funcId));
     } break;
     case instruction_t::Jump: {
-        std::int8_t offset = static_cast<std::int8_t>(frame.mod->byte_at(frame.position++));
+        frame.position += instr.arg.s8;
-        frame.position    += offset;
     } break;
     case instruction_t::JumpNotZero: {
-        byte offset = frame.mod->byte_at(frame.position++);
+        if (pop_thing()->integer() != 0) frame.position += instr.arg.s8;
-        auto cond   = pop_thing();
-        if (cond->integer() != 0) frame.position += (std::int8_t)offset;
     } break;
     case instruction_t::Return: {
         pop_frame();

furvm/src/instruction.cpp

@@ -0,0 +1,178 @@
+#include "furvm/instruction.hpp"
+
+#include "furlang/view.hpp"
+
+#include <stdexcept>
+
+namespace furvm {
+
+const std::size_t instruction_argument::s_sizes[instruction_argument::Count] = {
+    // None:
+    0,
+    // S8:
+    1,
+    // U8:
+    1,
+    // S16:
+    2,
+    // U16:
+    2,
+    // S32
+    1,
+    // U32
+    1,
+    // Constant:
+    4,
+    // Type:
+    4,
+    // Variable:
+    2,
+    // Function:
+    4,
+    // Offset:
+    1,
+};
+
+const bool instruction_argument::s_signedness[instruction_argument::Count] = {
+    // None:
+    false,
+    // S8:
+    true,
+    // U8:
+    false,
+    // S16:
+    true,
+    // U16:
+    false,
+    // S32
+    true,
+    // U32:
+    false,
+    // Constant:
+    false,
+    // Type:
+    false,
+    // Variable:
+    false,
+    // Function:
+    false,
+    // Offset:
+    true,
+};
+
+const instruction_argument_t instruction::s_arguments[instruction::Count] = {
+    // NoOperation:
+    instruction_argument::None,
+    // PushS8:
+    instruction_argument::S8,
+    // PushU8:
+    instruction_argument::U8,
+    // PushS16:
+    instruction_argument::S16,
+    // PushU16:
+    instruction_argument::U16,
+    // PushS32:
+    instruction_argument::S32,
+    // PushU32:
+    instruction_argument::U32,
+    // PushConstant:
+    instruction_argument::Constant,
+    // Array:
+    instruction_argument::Type,
+    // Get:
+    instruction_argument::None,
+    // Set:
+    instruction_argument::None,
+    // Drop:
+    instruction_argument::None,
+    // Duplicate:
+    instruction_argument::None,
+    // Swap:
+    instruction_argument::None,
+    // Clone:
+    instruction_argument::None,
+    // Reference:
+    instruction_argument::None,
+    // Add:
+    instruction_argument::None,
+    // Sub:
+    instruction_argument::None,
+    // Mul:
+    instruction_argument::None,
+    // Div:
+    instruction_argument::None,
+    // Mod:
+    instruction_argument::None,
+    // Equals:
+    instruction_argument::None,
+    // NotEquals:
+    instruction_argument::None,
+    // LessThan:
+    instruction_argument::None,
+    // GreaterThan:
+    instruction_argument::None,
+    // LessEqual:
+    instruction_argument::None,
+    // GreaterEqual:
+    instruction_argument::None,
+    // Pointerof:
+    instruction_argument::None,
+    // Sizeof:
+    instruction_argument::None,
+    // Lengthof:
+    instruction_argument::None,
+    // Load:
+    instruction_argument::Variable,
+    // Store:
+    instruction_argument::Variable,
+    // Call:
+    instruction_argument::Function,
+    // Jump:
+    instruction_argument::Offset,
+    // JumpNotZero:
+    instruction_argument::Offset,
+    // Return:
+    instruction_argument::None,
+};
+
+std::size_t instruction::read(furlang::view<std::uint8_t> in) {
+    type = static_cast<type_e>(in[0]);
+    if (type >= Count) throw std::runtime_error("invalid instruction");
+    arg.type = s_arguments[type];
+
+    furlang::view argView = in.subview(1, arg.size());
+    if (argView.size() != arg.size()) throw std::runtime_error("invalid bytecode");
+    switch (argView.size()) {
+    case 0: break;
+    case 1: arg.u8 = argView[0]; break;
+    case 2: arg.u16 = argView[0] | (argView[1] << 8); break;
+    case 4: arg.u32 = argView[0] | (argView[1] << 8) | (argView[2] << 16) | (argView[3] << 24); break;
+    default: throw std::runtime_error("unreachable");
+    }
+
+    return 1 + argView.size();
+}
+
+std::size_t instruction::write(std::vector<std::uint8_t>& out) const {
+    if (type >= Count) throw std::runtime_error("invalid instruction");
+    if (s_arguments[type] != arg.type) throw std::runtime_error("malformed instruction");
+    out.push_back(static_cast<char>(type));
+    switch (arg.size()) {
+    case 0: break;
+    case 1: out.push_back(arg.u8); break;
+    case 2:
+        out.push_back(arg.u16 & 0xFF);
+        out.push_back((arg.u16 >> 8) & 0xFF);
+        break;
+    case 4:
+        out.push_back(arg.u32 & 0xFF);
+        out.push_back((arg.u32 >> 8) & 0xFF);
+        out.push_back((arg.u32 >> 16) & 0xFF);
+        out.push_back((arg.u32 >> 24) & 0xFF);
+        break;
+    default: throw std::runtime_error("unreachable");
+    }
+    return 0;
+}
+
+} // namespace furvm