more assembler

src/assembler.cpp

@@ -13,6 +13,7 @@
 #include <memory>
 #include <string_view>
 #include <unordered_map>
+#include <utility>
 #include <vector>
 
 using namespace vc5::tools;
@@ -22,6 +23,9 @@ namespace fs = std::filesystem;
 
 using TT = Tok::Ty;
 
+using EO = EvaledOperand;
+using EOT = EO::Ty;
+
 auto vc5::tools::assemble_file(fs::path input_path, fs::path output_path)
     -> std::expected<void, std::string>
 {
@@ -49,6 +53,8 @@ auto vc5::tools::assemble_file(fs::path input_path, fs::path output_path)
 
     auto assembler = Assembler(text, builder);
 
+    auto lines = std::vector<Line>();
+
     while (true) {
         auto ins = parser.parse_ins();
         if (not ins) {
@@ -75,10 +81,35 @@ auto vc5::tools::assemble_file(fs::path input_path, fs::path output_path)
     return {};
 }
 
-void Assembler::assemble_ins(Ins& ins)
+auto Assembler::assemble_file(std::string_view text)
+    -> std::expected<std::vector<uint8_t>, std::string>
 {
-    if (ins.labels) {
+    auto parser = Parser(text);
-        for (const auto& label : *ins.labels) {
+
+    auto program = std::vector<uint8_t>(65536);
+    auto builder = Builder(program.data());
+
+    auto assembler = Assembler(text, builder);
+
+    auto lines = std::vector<Line>();
+
+    while (true) {
+        auto ins = parser.parse_ins();
+        if (not ins) {
+            break;
+        }
+    }
+
+    if (not parser.ok()) {
+        return std::unexpected("parsing failed");
+    }
+
+    return { std::move(program) };
+}
+
+void Assembler::assemble_define_labels(const std::vector<Label>& labels)
+{
+    for (const auto& label : labels) {
         if (label.is_local) {
             m_local_labels[label.ident] = m_builder->ip();
         } else {
@@ -86,13 +117,13 @@ void Assembler::assemble_ins(Ins& ins)
             m_global_labels[label.ident] = m_builder->ip();
         }
     }
-    }
-
-    assemble_line(ins);
 }
 
-void Assembler::assemble_line(Ins& ins)
+void Assembler::assemble_line(const Line& ins)
 {
+    if (ins.ident == "db") {
+        for (const auto& arg : ins.args) { }
+    }
     if (ins.ident == "mov") {
         if (arg_count_wrong(ins, 2))
             return;
@@ -102,7 +133,88 @@ void Assembler::assemble_line(Ins& ins)
         std::format("instruction '{}' not supported/implemented", ins.ident));
 }
 
-bool Assembler::arg_count_wrong(Ins& ins, size_t count)
+auto Assembler::eval_operand(const Expr& expr) -> std::unique_ptr<EvaledOperand>
+{
+    auto loc = expr.loc;
+    switch (expr.ty) {
+        case Expr::Ty::Ident:
+        case Expr::Ty::SubLabel:
+        case Expr::Ty::Reg:
+            return std::make_unique<EO>(loc, EOT::Reg, expr.as_reg());
+        case Expr::Ty::Int:
+        case Expr::Ty::Str:
+            return std::make_unique<EO>(loc, EOT::Str, expr.as_str());
+        case Expr::Ty::Mem:
+        case Expr::Ty::MemByte:
+        case Expr::Ty::MemWord:
+        case Expr::Ty::Negate:
+        case Expr::Ty::Not:
+        case Expr::Ty::Or:
+        case Expr::Ty::Xor:
+        case Expr::Ty::And:
+        case Expr::Ty::Add:
+        case Expr::Ty::Sub:
+            break;
+    }
+    std::unreachable();
+}
+
+auto Assembler::eval_operand_mem(const Expr& expr)
+    -> std::unique_ptr<EvaledOperand>
+{
+    switch (expr.ty) {
+        case Expr::Ty::Ident:
+        case Expr::Ty::SubLabel:
+        case Expr::Ty::Reg:
+        case Expr::Ty::Int:
+        case Expr::Ty::Str:
+        case Expr::Ty::Mem:
+        case Expr::Ty::MemByte:
+        case Expr::Ty::MemWord:
+        case Expr::Ty::Negate:
+        case Expr::Ty::Not:
+        case Expr::Ty::Or:
+        case Expr::Ty::Xor:
+        case Expr::Ty::And:
+        case Expr::Ty::Add:
+        case Expr::Ty::Sub:
+            break;
+    }
+    std::unreachable();
+}
+
+auto Assembler::eval_operand_to_imm(const Expr& expr)
+    -> std::unique_ptr<EvaledOperand>
+{
+    switch (expr.ty) {
+        case Expr::Ty::Ident:
+        case Expr::Ty::SubLabel:
+        case Expr::Ty::Reg:
+            error(expr.loc, "registers cannot be part of an expression");
+        case Expr::Ty::Int:
+            return std::make_unique<EO>(
+                expr.loc, EOT::Imm, static_cast<uint16_t>(expr.as_int()));
+        case Expr::Ty::Str:
+            error(expr.loc, "strings cannot be part of an expression");
+            return nullptr;
+        case Expr::Ty::Mem:
+        case Expr::Ty::MemByte:
+        case Expr::Ty::MemWord:
+            error(expr.loc, "indirections cannot be part of an expression");
+            return nullptr;
+        case Expr::Ty::Negate:
+        case Expr::Ty::Not:
+        case Expr::Ty::Or:
+        case Expr::Ty::Xor:
+        case Expr::Ty::And:
+        case Expr::Ty::Add:
+        case Expr::Ty::Sub:
+            break;
+    }
+    std::unreachable();
+}
+
+bool Assembler::arg_count_wrong(const Line& ins, size_t count)
 {
     if (ins.args.size() != count) {
         error(ins.loc, std::format("expected {} operands", count));
@@ -117,11 +229,11 @@ void Assembler::error(Loc loc, std::string_view message)
     loc.print_error(m_text, message);
 }
 
-auto Parser::parse_ins() -> std::unique_ptr<Ins>
+auto Parser::parse_ins() -> std::unique_ptr<Line>
 {
     auto loc = current_loc();
 
-    auto labels = Ins::Labels(nullptr);
+    auto labels = Line::Labels(nullptr);
 
     auto ident = std::string_view();
 
@@ -158,7 +270,7 @@ auto Parser::parse_ins() -> std::unique_ptr<Ins>
         }
     }
 
-    auto args = Ins::Args();
+    auto args = Line::Args();
 
     auto first = true;
     while (not test(TT::Eof) and not eat('\n')) {
@@ -179,7 +291,7 @@ auto Parser::parse_ins() -> std::unique_ptr<Ins>
         error(current_loc(), "expected line ending");
     }
 
-    return std::make_unique<Ins>(
+    return std::make_unique<Line>(
         loc, std::move(labels), ident, std::move(args));
 }
 
@@ -238,7 +350,9 @@ auto Parser::parse_prefix() -> std::unique_ptr<Expr>
     }
 }
 
-static const auto reg_idents = std::unordered_map<std::string_view, vc5::Reg> {
+namespace {
+
+const auto reg_idents = std::unordered_map<std::string_view, vc5::Reg> {
     { "R0", vc5::Reg::R0 },
     { "R1", vc5::Reg::R1 },
     { "R2", vc5::Reg::R2 },
@@ -251,6 +365,25 @@ static const auto reg_idents = std::unordered_map<std::string_view, vc5::Reg> {
     { "Rip", vc5::Reg::Rip },
 };
 
+char escape_char(char ch)
+{
+
+    switch (ch) {
+        case 'n':
+            return '\n';
+        case 'r':
+            return '\r';
+        case 't':
+            return '\t';
+        case '0':
+            return '\0';
+        default:
+            return ch;
+    }
+}
+
+}
+
 auto Parser::parse_operand() -> std::unique_ptr<Expr>
 {
     auto loc = current_loc();
@@ -263,21 +396,56 @@ auto Parser::parse_operand() -> std::unique_ptr<Expr>
         } else {
             return std::make_unique<Expr>(loc, Expr::Ty::Ident, ident);
         }
+    } else if (eat('.')) {
+        if (!test(TT::Ident)) {
+            error(current_loc(), "expected ')'");
+            return nullptr;
+        }
+        auto value
+            = std::string_view(m_tok.text.data() - 1, m_tok.text.size() + 1);
     } else if (test(TT::Int)) {
         auto text = std::string(m_tok.text);
+        step();
         auto value = std::strtol(text.c_str(), nullptr, 10);
         return std::make_unique<Expr>(
             loc, Expr::Ty::Int, static_cast<int>(value));
     } else if (test(TT::Bin)) {
         auto text = std::string(m_tok.text);
+        step();
         auto value = std::strtol(&text[2], nullptr, 2);
         return std::make_unique<Expr>(
             loc, Expr::Ty::Int, static_cast<int>(value));
     } else if (test(TT::Hex)) {
         auto text = std::string(m_tok.text);
+        step();
         auto value = std::strtol(&text[2], nullptr, 16);
         return std::make_unique<Expr>(
             loc, Expr::Ty::Int, static_cast<int>(value));
+    } else if (test(TT::Char)) {
+        auto text = std::string(m_tok.text);
+        step();
+
+        int value = text.at(1) == '\\' ? escape_char(text.at(2)) : text.at(1);
+
+        return std::make_unique<Expr>(
+            loc, Expr::Ty::Int, static_cast<int>(value));
+    } else if (test(TT::Str)) {
+        auto text = std::string(m_tok.text);
+        step();
+
+        auto value = std::string();
+        size_t i = 1;
+        while (i < text.size() - 1) {
+            if (text.at(i) == '\\') {
+                i += 1;
+                value.push_back(escape_char(text.at(i)));
+            } else {
+                value.push_back(text.at(i));
+            }
+            i += 1;
+        }
+
+        return std::make_unique<Expr>(loc, Expr::Ty::Str, std::move(value));
     } else if (eat('(')) {
         auto expr = parse_expr();
         if (not eat(')')) {
@@ -291,8 +459,31 @@ auto Parser::parse_operand() -> std::unique_ptr<Expr>
             error(current_loc(), "expected ']'");
             return nullptr;
         }
-        return std::make_unique<Expr>(
+        return std::make_unique<Expr>(loc, Expr::Ty::Mem, std::move(expr));
-            loc, Expr::Ty::Indirection, std::move(expr));
+    } else if (test(TT::Ident) and m_tok.text == "byte") {
+        step();
+        if (not eat('[')) {
+            error(current_loc(), "expected '['");
+            return nullptr;
+        }
+        auto expr = parse_expr();
+        if (not eat(']')) {
+            error(current_loc(), "expected ']'");
+            return nullptr;
+        }
+        return std::make_unique<Expr>(loc, Expr::Ty::MemByte, std::move(expr));
+    } else if (test(TT::Ident) and m_tok.text == "word") {
+        step();
+        if (not eat('[')) {
+            error(current_loc(), "expected '['");
+            return nullptr;
+        }
+        auto expr = parse_expr();
+        if (not eat(']')) {
+            error(current_loc(), "expected ']'");
+            return nullptr;
+        }
+        return std::make_unique<Expr>(loc, Expr::Ty::MemWord, std::move(expr));
     } else {
         error(current_loc(), "expected expression");
         return nullptr;
@@ -300,6 +491,11 @@ auto Parser::parse_operand() -> std::unique_ptr<Expr>
 }
 
 auto Parser::eat(int ty) -> bool
+{
+    return eat(static_cast<Tok::Ty>(ty));
+}
+
+auto Parser::eat(Tok::Ty ty) -> bool
 {
     if (test(ty)) {
         step();
@@ -314,6 +510,11 @@ void Parser::step()
 }
 
 auto Parser::test(int ty) const -> bool
+{
+    return m_tok.ty == static_cast<Tok::Ty>(ty);
+}
+
+auto Parser::test(Tok::Ty ty) const -> bool
 {
     return m_tok.ty == ty;
 }

src/assembler.hpp

@@ -28,9 +28,13 @@ namespace asmer {
     struct Expr {
         enum class Ty {
             Ident,
+            SubLabel,
             Reg,
             Int,
-            Indirection,
+            Str,
+            Mem,
+            MemByte,
+            MemWord,
             Negate,
             Not,
             Or,
@@ -42,22 +46,40 @@ namespace asmer {
 
         using Ptr = std::unique_ptr<Expr>;
         using Binary = std::tuple<Ptr, Ptr>;
-        using Data = std::variant<std::string_view, Reg, int, Ptr, Binary>;
+        using Data = std::
+            variant<std::string_view, std::string, Reg, int, Ptr, Binary>;
 
-        // clang-format off
+        auto as_ident() const -> const std::string_view&
-    auto as_ident() -> std::string_view& { return std::get<std::string_view>(data); }
+        {
-    auto as_reg() -> Reg& { return std::get<Reg>(data); }
+            return std::get<std::string_view>(data);
-    auto as_int() -> int& { return std::get<int>(data); }
+        }
-    auto as_unary() -> Ptr& { return std::get<Ptr>(data); }
+        auto as_str() const -> const std::string&
-    auto as_binary() -> Binary& { return std::get<Binary>(data); }
+        {
-        // clang-format on
+            return std::get<std::string>(data);
+        }
+        auto as_reg() const -> Reg
+        {
+            return std::get<Reg>(data);
+        }
+        auto as_int() const -> int
+        {
+            return std::get<int>(data);
+        }
+        auto as_unary() const -> const Ptr&
+        {
+            return std::get<Ptr>(data);
+        }
+        auto as_binary() const -> const Binary&
+        {
+            return std::get<Binary>(data);
+        }
 
         Loc loc;
         Ty ty;
         Data data;
     };
 
-    struct Ins {
+    struct Line {
         using Labels = std::unique_ptr<std::vector<Label>>;
         using Args = std::vector<std::unique_ptr<Expr>>;
 
@@ -76,7 +98,7 @@ namespace asmer {
         {
         }
 
-        auto parse_ins() -> std::unique_ptr<Ins>;
+        auto parse_ins() -> std::unique_ptr<Line>;
 
         auto ok() const -> bool
         {
@@ -90,9 +112,11 @@ namespace asmer {
         auto parse_operand() -> std::unique_ptr<Expr>;
 
         auto eat(int ty) -> bool;
+        auto eat(Tok::Ty ty) -> bool;
         void step();
 
         auto test(int ty) const -> bool;
+        auto test(Tok::Ty ty) const -> bool;
         auto current_loc() const -> Loc;
 
         void error(Loc loc, std::string_view message);
@@ -103,6 +127,31 @@ namespace asmer {
         bool m_error_occured = false;
     };
 
+    struct EvaledOperand {
+        enum class Ty {
+            Reg,
+            Imm,
+            Str,
+            MemByteImm,
+            MemByteReg,
+            MemWordImm,
+            MemWordReg,
+        };
+
+        auto as_reg() -> Reg&
+        {
+            return std::get<Reg>(data);
+        }
+        auto as_imm() -> uint16_t&
+        {
+            return std::get<uint16_t>(data);
+        }
+
+        Loc loc;
+        Ty ty;
+        std::variant<Reg, uint16_t, std::string> data;
+    };
+
     class Assembler {
     public:
         explicit Assembler(std::string_view text, Builder& builder)
@@ -111,7 +160,11 @@ namespace asmer {
         {
         }
 
-        void assemble_ins(Ins& ins);
+        auto assemble_file(std::string_view text)
+            -> std::expected<std::vector<uint8_t>, std::string>;
+
+        void assemble_define_labels(const std::vector<Label>& labels);
+        void assemble_line(const Line& line);
 
         auto ok() const -> bool
         {
@@ -119,10 +172,14 @@ namespace asmer {
         }
 
     private:
-        void assemble_line(Ins& ins);
+        auto eval_operand(const Expr& expr) -> std::unique_ptr<EvaledOperand>;
+        auto eval_operand_mem(const Expr& expr)
+            -> std::unique_ptr<EvaledOperand>;
+        auto eval_operand_to_imm(const Expr& expr)
+            -> std::unique_ptr<EvaledOperand>;
 
         /// true means fail
-        bool arg_count_wrong(Ins& ins, size_t count);
+        bool arg_count_wrong(const Line& ins, size_t count);
 
         void error(Loc loc, std::string_view message);
 

src/builder.hpp

@@ -71,13 +71,26 @@ public:
         m_ip = ip;
     }
 
-    inline void push(uint16_t v)
+    void push(uint16_t v)
     {
         m_data[m_ip] = v >> 8;
         m_data[m_ip + 1] = v & 0xff;
         m_ip += 2;
     }
 
+    void push_byte(uint8_t v)
+    {
+        m_data[m_ip] = v;
+        m_ip += 1;
+    }
+
+    void align_word()
+    {
+        if (m_ip & 1) {
+            m_ip += 1;
+        }
+    }
+
 private:
     void binary_reg(Reg dst, Reg op1, Reg op2, Op op);
     void binary_imm(Reg dst, Reg op1, uint16_t op2, Op op);

src/scanner.cpp

@@ -6,11 +6,13 @@
 using namespace vc5::tools;
 using namespace std::literals;
 
+using TT = Tok::Ty;
+
 auto Scanner::next() -> Tok
 {
     auto loc = this->current_loc();
     if (done()) {
-        return tok(Tok::Eof, loc);
+        return tok(TT::Eof, loc);
     }
     if (test_in(" \t\r")) {
         while (test_in(" \t\r")) {
@@ -30,13 +32,13 @@ auto Scanner::next() -> Tok
             or test_range('a', 'z')) {
             step();
         }
-        return tok(Tok::Ident, loc);
+        return tok(TT::Ident, loc);
     }
     if (test_range('1', '9')) {
         while (test_range('0', '9')) {
             step();
         }
-        return tok(Tok::Int, loc);
+        return tok(TT::Int, loc);
     }
     if (test('0')) {
         step();
@@ -45,20 +47,66 @@ auto Scanner::next() -> Tok
             while (test_in("01")) {
                 step();
             }
-            return tok(Tok::Bin, loc);
+            return tok(TT::Bin, loc);
         } else if (test('x')) {
             step();
             while (test_range('0', '9') or test_range('a', 'f')
                 or test_range('A', 'F')) {
                 step();
             }
-            return tok(Tok::Hex, loc);
+            return tok(TT::Hex, loc);
         } else {
-            return tok(Tok::Int, loc);
+            return tok(TT::Int, loc);
         }
     }
+    if (test('\'')) {
+        step();
+        if (done()) {
+            error(loc, "malformed literal");
+            return next();
+        }
+        if (test('\\')) {
+            step();
+            if (done()) {
+                error(loc, "malformed literal");
+                return next();
+            }
+            step();
+        } else {
+            if (test('\'')) {
+                error(loc, "malformed literal");
+                return next();
+            }
+            step();
+        }
+        if (done() or not test('\'')) {
+            error(loc, "malformed literal");
+            return next();
+        }
+        step();
+        return tok(TT::Char, loc);
+    }
+    if (test('"')) {
+        step();
+        while (not done() and not test('"')) {
+            if (test('\\')) {
+                step();
+                if (done()) {
+                    error(loc, "malformed literal");
+                    return next();
+                }
+            }
+            step();
+        }
+        if (done() or not test('"')) {
+            error(loc, "malformed literal");
+            return next();
+        }
+        step();
+        return tok(TT::Str, loc);
+    }
     if (test_in("\n()[],:|^+-!")) {
-        auto ty = static_cast<Tok::Ty>(current());
+        auto ty = static_cast<TT>(current());
         step();
         return tok(ty, loc);
     }
@@ -83,7 +131,7 @@ void Scanner::step()
     }
 }
 
-auto Scanner::tok(Tok::Ty ty, Loc loc) const -> Tok
+auto Scanner::tok(TT ty, Loc loc) const -> Tok
 {
     return Tok { m_text.substr(loc.idx, m_idx - loc.idx), loc, ty };
 }

src/scanner.hpp

@@ -14,12 +14,14 @@ struct Loc {
 };
 
 struct Tok {
-    enum Ty {
+    enum class Ty {
         Eof,
         Ident,
         Int,
         Hex,
         Bin,
+        Char,
+        Str,
         Newline = '\n',
         LParen = '(',
         RParen = ')',