highground-fork/src/main.cpp

#include <cctype>
#include <cstdint>
#include <groundvm.h>
#include <vector>
#include <string>
#include <fstream>
#include <iostream>
#include <sstream>
#include <optional>
#include <variant>

#define parseOneToken(token) Parser({token.value()}).parse().children[0]

namespace HighGround {

    int tmpIdIterator = 0;
    int labelIterator = 0;

    namespace Parser {

        enum class HGNodeType {
            Add, Subtract, Equal, Set, While, If, Value, Identifier, None, Root, CodeBlock, CodeBlockStart, CodeBlockEnd, Puts
        };

        enum class HGDataType {
            Int, String, Double, Bool, Char, None
        };

        class HGNode;

        class HGGroundCodeBlock {
            public:
            std::vector<GroundInstruction> code;
            HGGroundCodeBlock() = default;
        };

        class HGData {
            typedef std::variant<int64_t, std::string, double, bool, char> varData; 
            varData data;
            public:
            HGDataType type = HGDataType::Int;
            HGData() = default;
            HGData(int64_t in) : data(in), type(HGDataType::Int) {}
            HGData(double in) : data(in), type(HGDataType::Double) {}
            HGData(std::string in) : data(in), type(HGDataType::String) {}
            HGData(char in) : data(in), type(HGDataType::Char) {}
            HGData(bool in) : data(in), type(HGDataType::Bool) {}
            std::optional<int64_t> getInt() {
                if (type == HGDataType::Int) {
                    return std::get<int64_t>(data);
                } else {
                    return {};
                }
            }
            std::optional<double> getDouble() {
                if (type == HGDataType::Double) {
                    return std::get<double>(data);
                } else {
                    return {};
                }
            }
            std::optional<std::string> getString() {
                if (type == HGDataType::String) {
                    return std::get<std::string>(data);
                } else {
                    return {};
                }
            }
            std::optional<char> getChar() {
                if (type == HGDataType::Char) {
                    return std::get<char>(data);
                } else {
                    return {};
                }
            }
            std::optional<bool> getBool() {
                if (type == HGDataType::Bool) {
                    return std::get<bool>(data);
                } else {
                    return {};
                }
            }
        };

        class HGNode {
            HGNodeType nodeType = HGNodeType::None;
            HGData data;
            public:
            std::vector<HGNode> children;
            std::string outputId;
            HGNode(HGNodeType nodeType) : nodeType(nodeType) {}
            HGNode(HGNodeType nodeType, HGData data) : nodeType(nodeType), data(data) {}
            HGNode() = default;
            void addNode(HGNode in) {
                children.push_back(in);
            }
            void setValue(HGData in) {
                data = in;
            }
            const std::vector<HGGroundCodeBlock> generateCode() {
                std::vector<HGGroundCodeBlock> code;
                if (nodeType != HGNodeType::If) for (auto& child : children) {
                    auto childCode = child.generateCode();
                    code.insert(code.end(), childCode.begin(), childCode.end());
                }
                switch (nodeType) {
                    case HGNodeType::Value: {
                        outputId = "tmp_" + std::to_string(tmpIdIterator++);
                        HGGroundCodeBlock codeBlock;
                        GroundInstruction gi = groundCreateInstruction(SET);
                        groundAddReferenceToInstruction(&gi, groundCreateReference(DIRREF, outputId.data()));
                        switch (data.type) {
                            case HGDataType::Int: {
                                auto dataopt = data.getInt();
                                if (dataopt) {
                                    groundAddValueToInstruction(&gi, groundCreateValue(INT, dataopt.value()));
                                }
                                break;
                            }
                            case HGDataType::Double: {
                                auto dataopt = data.getDouble();
                                if (dataopt) {
                                    groundAddValueToInstruction(&gi, groundCreateValue(DOUBLE, dataopt.value()));
                                }
                                break;
                            }
                            case HGDataType::String: {
                                auto dataopt = data.getString();
                                if (dataopt) {
                                    groundAddValueToInstruction(&gi, groundCreateValue(STRING, dataopt.value().c_str()));
                                }
                                break;
                            }
                            case HGDataType::Char: {
                                auto dataopt = data.getChar();
                                if (dataopt) {
                                    groundAddValueToInstruction(&gi, groundCreateValue(CHAR, dataopt.value()));
                                }
                                break;
                            }
                            case HGDataType::Bool: {
                                auto dataopt = data.getBool();
                                if (dataopt) {
                                    groundAddValueToInstruction(&gi, groundCreateValue(BOOL, dataopt.value()));
                                }
                                break;
                            }
                        }
                        codeBlock.code.push_back(gi);
                        code.push_back(codeBlock);
                        break;
                    }
                    case HGNodeType::Add: {
                        HGGroundCodeBlock codeBlock;
                        outputId = "tmp_" + std::to_string(tmpIdIterator++);
                        GroundInstruction gi = groundCreateInstruction(ADD); 
                        if (children.size() < 2) {
                            std::cout << "Need more stuff to add\n";
                        }
                        groundAddReferenceToInstruction(&gi, groundCreateReference(VALREF, children[0].outputId.data()));
                        groundAddReferenceToInstruction(&gi, groundCreateReference(VALREF, children[1].outputId.data()));
                        groundAddReferenceToInstruction(&gi, groundCreateReference(DIRREF, outputId.data()));
                        codeBlock.code.push_back(gi);
                        code.push_back(codeBlock);
                        break;
                    }
                    case HGNodeType::Equal: {
                        std::cout << "Equal node children: " << children.size() << "\n";
                        std::cout << "Left outputId: '" << children[0].outputId << "'\n";
                        std::cout << "Right outputId: '" << children[1].outputId << "'\n";
                        HGGroundCodeBlock codeBlock;
                        outputId = "tmp_" + std::to_string(tmpIdIterator++);
                        GroundInstruction gi = groundCreateInstruction(EQUAL); 
                        if (children.size() < 2) {
                            std::cout << "Need more stuff to equal\n";
                        }
                        groundAddReferenceToInstruction(&gi, groundCreateReference(VALREF, children[0].outputId.data()));
                        groundAddReferenceToInstruction(&gi, groundCreateReference(VALREF, children[1].outputId.data()));
                        groundAddReferenceToInstruction(&gi, groundCreateReference(DIRREF, outputId.data()));
                        codeBlock.code.push_back(gi);
                        code.push_back(codeBlock);
                        break;
                    }
                    case HGNodeType::Puts: {
                        HGGroundCodeBlock codeBlock;
                        GroundInstruction gi = groundCreateInstruction(PRINTLN);                   
                        if (children.size() < 1) {
                            std::cout << "Need more stuff to puts\n";
                        }
                        groundAddReferenceToInstruction(&gi, groundCreateReference(VALREF, children[0].outputId.data()));
                        codeBlock.code.push_back(gi);
                        code.push_back(codeBlock);
                        break;
                    }
                    case HGNodeType::If: {
                        auto conditionCode = children[0].generateCode();
                        code.insert(code.end(), conditionCode.begin(), conditionCode.end());
                        outputId = "tmp_" + std::to_string(tmpIdIterator++);
                        HGGroundCodeBlock codeBlock;
                        GroundInstruction gi = groundCreateInstruction(NOT);
                        groundAddReferenceToInstruction(&gi, groundCreateReference(VALREF, children[0].outputId.data()));
                        groundAddReferenceToInstruction(&gi, groundCreateReference(DIRREF, outputId.data()));
                        codeBlock.code.push_back(gi);
                        std::string labelId = "if_" + std::to_string(labelIterator++);
                        GroundInstruction gi2 = groundCreateInstruction(IF);
                        groundAddReferenceToInstruction(&gi2, groundCreateReference(VALREF, outputId.data()));
                        groundAddReferenceToInstruction(&gi2, groundCreateReference(LINEREF, labelId.data()));
                        codeBlock.code.push_back(gi2);
                        code.push_back(codeBlock);
                        for (size_t i = 1; i < children.size(); i++) {
                            auto childCode = children[i].generateCode();
                            code.insert(code.end(), childCode.begin(), childCode.end());
                        }
                        codeBlock.code.clear();
                        GroundInstruction gi3 = groundCreateInstruction(CREATELABEL);
                        groundAddReferenceToInstruction(&gi3, groundCreateReference(LABEL, labelId.data()));
                        codeBlock.code.push_back(gi3);
                        codeBlock.code.push_back(groundCreateInstruction(PAUSE));
                        code.push_back(codeBlock);
                        break;
                    }
                    default: {}
                }
                return code;
            }
            
        };

        class Parser {
            std::vector<std::string> tokensToParse;
            size_t current;
            size_t size;

            std::optional<std::string> peek(int ahead = 1) {
                if (current + ahead < size) {
                    return tokensToParse[current + ahead];
                } else {
                    return {};
                }
            }

            std::optional<std::string> consume() {
                if (current < size) {
                    return tokensToParse[current++];
                } else {
                    return {};
                }
            }

            bool isInt(std::string in) {
                for (const char& c : in) {
                    if (!std::isdigit(c)) {
                        return false;
                    }
                }
                return true;
            }
            bool isDouble(std::string in) {
                bool foundDot = false;
                for (const char& c : in) {
                    if (!std::isdigit(c)) {
                        if (!foundDot && c == '.') {
                            foundDot = true;
                            continue;
                        }
                        return false;
                    }
                }
                return true;
            }
            bool isString(std::string in) {
                if (in.size() > 1 && in[0] == '"' && in.back() == '"') {
                    return true;
                }
                return false;
            }
            bool isChar(std::string in) {
                if (in.size() == 3 && in[0] == '\'' && in.back() == '\'') {
                    return true;
                }
                return false;
            }
            bool isBool(std::string in) {
                if (in == "true" || in == "false") {
                    return true;
                }
                return false;
            }

            HGDataType getDataType(std::string in) {
                if (isInt(in)) {
                    return HGDataType::Int;
                }
                if (isDouble(in)) {
                    return HGDataType::Double;
                }
                if (isString(in)) {
                    return HGDataType::String;
                }
                if (isChar(in)) {
                    return HGDataType::Char;
                }
                if (isBool(in)) {
                    return HGDataType::Bool;
                }
                return HGDataType::None;
            }

            HGNodeType getNodeType(std::string in) {
                if (getDataType(in) != HGDataType::None) {
                    return HGNodeType::Value;
                }
                if (in == "+") {
                    return HGNodeType::Add;
                }
                if (in == "==") {
                    return HGNodeType::Equal;
                }
                if (in == "puts") {
                    return HGNodeType::Puts;
                }
                if (in == "if") {
                    return HGNodeType::If;
                }
                if (in == "{") {
                    return HGNodeType::CodeBlockStart;
                }
                if (in == "}") {
                    return HGNodeType::CodeBlockEnd;
                }
                return HGNodeType::None;
            }

            public:
            Parser(std::vector<std::string> in) : tokensToParse(in) {}

            HGNode parse() {
                current = 0;
                size = tokensToParse.size();
                HGNode rootNode(HGNodeType::Root);
                while (auto tokenopt = consume()) {
                    std::string token = tokenopt.value();
                    switch (getNodeType(token)) {
                        case HGNodeType::Value: {
                            switch (getDataType(token)) {
                                case HGDataType::Int: {
                                    HGNode intNode(HGNodeType::Value);
                                    intNode.setValue((int64_t) std::stoll(token));
                                    rootNode.addNode(intNode);
                                    break;
                                }
                                case HGDataType::Double: {
                                    HGNode doubleNode(HGNodeType::Value);
                                    doubleNode.setValue(std::stod(token));
                                    rootNode.addNode(doubleNode);
                                    break;
                                }
                                case HGDataType::String: {
                                    HGNode stringNode(HGNodeType::Value);
                                    stringNode.setValue(token.substr(1, token.size() - 2));
                                    rootNode.addNode(stringNode);
                                    break;
                                }
                                case HGDataType::Char: {
                                    HGNode charNode(HGNodeType::Value);
                                    charNode.setValue(token[1]);
                                    rootNode.addNode(charNode);
                                    break;
                                }
                                case HGDataType::Bool: {
                                    HGNode boolNode(HGNodeType::Value);
                                    boolNode.setValue(token == "true");
                                    rootNode.addNode(boolNode);
                                    break;
                                }
                            }
                            break;
                        }
                        case HGNodeType::Add: {
                            HGNode addNode(HGNodeType::Add);
                            addNode.addNode(rootNode.children.back());
                            rootNode.children.pop_back();
                            auto tokenopt = consume();
                            if (tokenopt) {
                                addNode.addNode(parseOneToken(tokenopt));
                            } else {
                                std::cout << "FEED ME MORE TOKENS\n";
                                exit(1);
                            }
                            rootNode.addNode(addNode);
                            break;
                        }
                        case HGNodeType::Equal: {
                            HGNode addNode(HGNodeType::Equal);
                            addNode.addNode(rootNode.children.back());
                            rootNode.children.pop_back();
                            auto tokenopt = consume();
                            if (tokenopt) {
                                addNode.addNode(parseOneToken(tokenopt));
                            } else {
                                std::cout << "FEED ME MORE TOKENS\n";
                                exit(1);
                            }
                            rootNode.addNode(addNode);
                            break;
                        }
                        case HGNodeType::Puts: {
                            HGNode putsNode(HGNodeType::Puts);
                            std::vector<std::string> tokens;
                            while (auto tokenopt = consume()) {
                                if (tokenopt.value() == "\n") {
                                    break;
                                }
                                tokens.push_back(tokenopt.value());
                            }
                            auto children = Parser(tokens).parse();
                            for (auto& child : children.children) {
                                putsNode.addNode(child);
                            } 
                            rootNode.addNode(putsNode);
                            break;
                        }
                        case HGNodeType::If: {
                            HGNode ifNode(HGNodeType::If);
                            std::vector<std::string> tokens;
                            while (auto tokenopt = consume()) {
                                if (tokenopt.value() == "\n") {
                                    break;
                                }
                                tokens.push_back(tokenopt.value());
                            }
                            auto children = Parser(tokens).parse();
                            ifNode.addNode(children.children[0]);
                            tokens.clear();
                            size_t brackets = 1;
                            auto tokenopt = consume();
                            if (tokenopt) {
                                if (tokenopt.value() == "{") {
                                    tokens.push_back(tokenopt.value());
                                    while (auto tokenopt = consume()) {
                                        tokens.push_back(tokenopt.value());
                                        if (tokenopt.value() == "{") {
                                            brackets++;
                                        }
                                        if (tokenopt.value() == "}") {
                                            brackets--;
                                        }
                                        if (brackets == 0) {
                                            break;
                                        }
                                    }
                                } else {
                                    std::cout << "I want a code block instead of a " + tokenopt.value() + "\n";
                                    exit(1);
                                }
                            } else {
                                std::cout << "FEED ME MORE TOKENSSSSS\n";
                                exit(1);
                            }
                            auto childCodeBlock = Parser(tokens).parse();
                            ifNode.addNode(childCodeBlock.children[0]); 
                            rootNode.addNode(ifNode);
                            break;
                        }
                        case HGNodeType::CodeBlockStart: {
                            HGNode codeBlockNode(HGNodeType::CodeBlock);
                            size_t brackets = 1;
                            std::vector<std::string> tokens;
                            while (auto tokenopt = consume()) {
                                if (tokenopt.value() == "{") {
                                    brackets++;
                                }
                                if (tokenopt.value() == "}") {
                                    brackets--;
                                }
                                if (brackets == 0) {
                                    break;
                                }
                                tokens.push_back(tokenopt.value());
                            }
                            codeBlockNode.children = Parser(tokens).parse().children;
                            rootNode.addNode(codeBlockNode);
                            // WIP
                            break;
                        }
                    }
                }
                return rootNode;
            }
        };

        GroundProgram assembleProgram(HGNode& rootNode) {
            GroundProgram gp = groundCreateProgram();
            auto code = rootNode.generateCode();
            for (int i = 0; i < code.size(); i++) {
                for (const auto& inst : code[i].code) {
                    groundAddInstructionToProgram(&gp, inst);
                }
            }
            return gp;
        }

    } // namespace Parser
    
    class Lexer {
        std::string input;
        size_t size;
        size_t current;

        std::optional<char> peek(int ahead = 1) {
            if (current + ahead < size) {
                return input[current + ahead];
            } else {
                return {};
            }
        }

        std::optional<char> consume() {
            if (current < size) {
                return input[current++];
            } else {
                return {};
            }
        }

        public:
        Lexer(std::string in) : input(in), size(in.size()) {};
        std::vector<std::string> lex() {
            current = 0;
            std::vector<std::string> tokens;
            std::string buf;
            while (auto copt = consume()) {
                char c = copt.value();
                switch (c) {
                    // tokens which are not followed by anything
                    case '\n':
                    case '(':
                    case ')':
                    case '}': 
                    {
                        if (!buf.empty()) {
                            tokens.push_back(buf);
                            buf.clear();
                        }
                        tokens.push_back(std::string(1, c));
                        break;
                    }
                    // tokens which may be followed by either themselves
                    // or an equals sign
                    case '+':
                    case '-':
                    {
                        std::string newToken(1, c);
                        auto tokenopt = peek();
                        if (tokenopt) {
                            char token = tokenopt.value();
                            if (token == c || token == '=') {
                                newToken += token;
                                consume();
                            }
                        }
                        if (!buf.empty()) {
                            tokens.push_back(buf);
                            buf.clear();
                        }
                        tokens.push_back(newToken);
                        break;
                    }
                    // tokens which may be followed by an equals sign
                    case '*':
                    case '/':
                    case '=':
                    {
                        std::string newToken(1, c);
                        auto tokenopt = peek();
                        if (tokenopt) {
                            char token = tokenopt.value();
                            if (token == '=') {
                                newToken += token;
                                consume();
                            }
                        }
                        if (!buf.empty()) {
                            tokens.push_back(buf);
                            buf.clear();
                        }
                        tokens.push_back(newToken);
                        break;
                    }
                    // tokens which need a newline inserted for them
                    case '{':
                    {
                        if (!buf.empty()) {
                            tokens.push_back(buf);
                            buf.clear();
                        }
                        tokens.push_back("\n");
                        tokens.push_back(std::string(1, c));
                    }
                    // tokens which do not need to be included
                    case ' ':
                    {
                        if (!buf.empty()) {
                            tokens.push_back(buf);
                            buf.clear();
                        }
                        break;
                    }
                    default: 
                    {
                        buf += c;
                    }
                }
            }

            if (!buf.empty()) {
                tokens.push_back(buf);
            }
            return tokens;
        }

    };


} // namespace HighGround


int main(int argc, char** argv) {
    if (argc < 2) {
        std::cout << "Usage: " << argv[0] << " (file)\n"; 
        exit(1);
    }
    std::ifstream file(argv[1]);
    std::ostringstream ss;
    ss << file.rdbuf();
    auto lexed = HighGround::Lexer(ss.str()).lex();
    auto parsed = HighGround::Parser::Parser(lexed).parse();
    GroundProgram program = HighGround::Parser::assembleProgram(parsed);
    groundRunProgram(&program);
}