simple secd parser

.idea/vcs.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="VcsDirectoryMappings">
+    <mapping directory="$PROJECT_DIR$" vcs="Git" />
+  </component>
+</project>

src/vm/CMakeLists.txt

@@ -1,3 +1,7 @@
-add_library(vm src/VM.cpp src/Cell.cpp)
+set(CMAKE_CXX_STANDARD 20)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+
+add_library(vm src/VM.cpp src/Cell.cpp
+        src/Parser.cpp)
 
 target_include_directories(vm PUBLIC includes)

src/vm/includes/Parser.h

@@ -0,0 +1,46 @@
+//
+// Created by Stepan Usatiuk on 23.12.2023.
+//
+
+#ifndef PSIL_PARSER_H
+#define PSIL_PARSER_H
+
+#include <string_view>
+#include <string>
+#include <queue>
+
+class VM;
+
+struct Cell;
+
+class Parser {
+public:
+    explicit Parser(VM &vm);
+
+    void loadSecd(std::string_view input);
+
+private:
+
+    void compileBody(const std::function<void(Cell *)> &sink);
+
+    VM &_vm;
+
+    class Tokenizer {
+    public:
+        void load(std::string_view input);
+
+        std::string getNext();
+
+        std::string_view peek() const;
+
+        bool empty() const;
+
+    private:
+        std::queue<std::string> _tokens;
+    };
+
+    Tokenizer _tokenizer;
+};
+
+
+#endif //PSIL_PARSER_H

src/vm/src/Parser.cpp

@@ -0,0 +1,105 @@
+//
+// Created by Stepan Usatiuk on 23.12.2023.
+//
+
+#include "Parser.h"
+
+#include <ranges>
+#include <stack>
+
+#include "VM.h"
+
+Parser::Parser(VM &vm) : _vm(vm) {}
+
+void Parser::loadSecd(std::string_view input) {
+    _tokenizer.load(input);
+    std::stack<Cell *> out;
+
+    compileBody([&out](Cell *cmd) { out.emplace(cmd); });
+
+    while (!out.empty()) {
+        this->_vm.appendCommand(out.top());
+        out.pop();
+    }
+}
+
+void Parser::compileBody(const std::function<void(Cell *)> &sink) {
+    auto token = _tokenizer.getNext();
+    if (token != "(") throw std::invalid_argument("Expected (");
+
+    while (token != ")") {
+        token = _tokenizer.getNext();
+
+        if (token == "NIL") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::NIL));
+        } else if (token == "LDC") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::LDC));
+            sink(_vm.makeCell<IntCell>(std::stoi(_tokenizer.getNext())));
+        } else if (token == "LD") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::LD));
+            if (_tokenizer.getNext() != "(") throw std::invalid_argument("Expected (");
+            int64_t frame = std::stoi(_tokenizer.getNext());
+            if (_tokenizer.getNext() != ".") throw std::invalid_argument("Expected .");
+            int64_t loc = std::stoi(_tokenizer.getNext());
+            if (_tokenizer.getNext() != ")") throw std::invalid_argument("Expected )");
+            sink(_vm.makeCell<ConsCell>(_vm.makeCell<IntCell>(frame), _vm.makeCell<IntCell>(loc)));
+        } else if (token == "LDF") {
+            std::stack<Cell *> out;
+
+            compileBody([&out](Cell *cmd) { out.emplace(cmd); });
+
+            if (out.empty()) throw std::invalid_argument("Function body empty");
+
+            ConsCell *fntop = _vm.makeCell<ConsCell>(out.top());
+            out.pop();
+
+            while (!out.empty()) {
+                _vm.push(fntop, out.top());
+                out.pop();
+            }
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::LDF));
+            sink(fntop);
+        } else if (token == "CONS") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::CONS));
+        } else if (token == "AP") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::AP));
+        } else if (token == "ADD") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::ADD));
+        } else if (token == "RET") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::RET));
+        } else if (token == "PUTCHAR") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::PUTCHAR));
+        } else if (token == "PUTNUM") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::PUTNUM));
+        } else if (token == "STOP") {
+            sink(_vm.makeCell<CommandCell>(CommandCell::CommandNum::STOP));
+        } else {
+            if (token != ")")throw std::invalid_argument("Unknown token " + token);
+        }
+    }
+}
+
+std::string Parser::Tokenizer::getNext() {
+    std::string ret = std::move(_tokens.front());
+    _tokens.pop();
+    return ret;
+}
+
+std::string_view Parser::Tokenizer::peek() const {
+    return _tokens.back();
+}
+
+void Parser::Tokenizer::load(std::string_view input) {
+    for (const auto &w: input
+                        | std::views::split(' ')
+                        | std::views::transform([](auto &&rng) {
+        return std::string_view(&*rng.begin(), std::ranges::distance(rng));
+    })) {
+        _tokens.emplace(w);
+    }
+}
+
+bool Parser::Tokenizer::empty() const {
+    return _tokens.empty();
+}
+

test/vm/CMakeLists.txt

@@ -8,5 +8,16 @@ target_link_libraries(
         GTest::gtest_main
 )
 
+add_executable(
+        VMWithParserTest
+        VMWithParserTest.cpp
+)
+target_link_libraries(
+        VMWithParserTest
+        vm
+        GTest::gtest_main
+)
+
 include(GoogleTest)
-gtest_discover_tests(VMTest)
+gtest_discover_tests(VMTest)
+gtest_discover_tests(VMWithParserTest)

test/vm/VMWithParserTest.cpp

@@ -0,0 +1,30 @@
+#include <gtest/gtest.h>
+
+#include "VM.h"
+#include "Parser.h"
+
+TEST(VMWithParserTest, BasicHello) {
+    std::stringstream ssin;
+    std::stringstream ssout;
+    {
+        VM vm(ssin, ssout);
+        Parser parser(vm);
+        parser.loadSecd("( LDC 104 PUTCHAR STOP )");
+        vm.run();
+    }
+    ssout.flush();
+    EXPECT_EQ(ssout.str(), "h");
+}
+
+TEST(VMWithParserTest, BasicFunction) {
+    std::stringstream ssin;
+    std::stringstream ssout;
+    {
+        VM vm(ssin, ssout);
+        Parser parser(vm);
+        parser.loadSecd("( NIL LDC 1 CONS LDC 2 CONS LDF ( LD ( 1 . 1 ) LD ( 1 . 2 ) ADD RET ) AP PUTNUM STOP )");
+        vm.run();
+    }
+    ssout.flush();
+    EXPECT_EQ(ssout.str(), "3");
+}