pl/include/parser/ast.hpp

#pragma once
#include <cassert>
#include "parser/lexer.hpp"
#include "parser/nodes.hpp"
#include "prelude/error.hpp"

class AstParser
{
public:
    AstParser(Lexer* lexer)
        : m_lexer(lexer) {}
public:
    ExternNode* ParseExtern()
    {
        m_lexer->NextExpect(TokenType::Id);
        return new ExternNode(m_lexer->token().string);
    }

    FnDeclNode* ParseFnDecl()
    {
        // Function Declaration
        m_lexer->NextExpect(TokenType::Id);
        StringView name = m_lexer->token().string;
        m_lexer->NextExpect('(');
        Builder<StringView> params;
        // TODO: support multiple params
        if (m_lexer->seek_token()->token != ')')
        {
            m_lexer->NextExpect(TokenType::Id);
            params.Push(m_lexer->token().string);
        }
        m_lexer->NextExpect(')');
        m_lexer->NextExpect('{');
        auto compound = new CompoundNode();
        while (m_lexer->seek_token()->token != '}')
        {
            compound->addNode(ParseStatement());
        }
        m_lexer->NextExpect('}');
        return new FnDeclNode(name, compound, params.view());
    }

    FnCallNode* ParseFnCall(const StringView& name)
    {
        // m_lexer->NextExpect(TokenType::Id);
        // char* name = strdup(m_lexer->token().string);
        m_lexer->NextExpect('(');
        Builder<Node*> args;
        while (m_lexer->seek_token()->token != ')')
        {
            auto arg = ParseExpression();
            args.Push(arg);
            if (m_lexer->seek_token()->token == ',') assert(m_lexer->NextToken());
        }
        m_lexer->NextExpect(')');
        return new FnCallNode(name, std::move(args.view()));
    }

    Node* ParseFactor()
    {
        auto token = m_lexer->seek_token();

        switch (token->token)
        {
        case TokenType::IntLiteral: // integer
        {
            m_lexer->NextExpect(TokenType::IntLiteral);
            auto node = new IntLiteralNode(m_lexer->token().int_number);
            return node;
        }
        case TokenType::Id: // variable name or function call
        {
            m_lexer->NextExpect(TokenType::Id);
            auto name = m_lexer->token().string;
            token = m_lexer->seek_token();
            if (token->token == '(')
            {
                return ParseFnCall(name);
            }
            return new VariableNode(name);
        }
        default:
            // fprintf(stderr, "%s:%d:%d: ERROR: unexpected token while parsing %ld\n", m_lexer->filename(), token->line_number, token->offset_start, token->token);
            ErrorLogger::Raise(Error::ParseError(m_lexer->filename(), StringView::FromFormat("unexpected token while parsing '%d'", token->token), token->line_number, token->offset_start));
            break;
        }

        assert(0 && "unreachable");
    }

    Node* ParseTerm()
    {
        auto t = ParseFactor();

        for (auto op = m_lexer->seek_token(); is_one_of(op->token, '/', '*'); op = m_lexer->seek_token())
        {
            m_lexer->NextToken();
            ExpressionNode::Operator eop;
            assert(static_cast<size_t>(ExpressionNode::Operator::COUNT_OPERATORS) == 4 && "some operators may not be handled");
            switch((char)op->token)
            {
            case '/':
                eop = ExpressionNode::Operator::Divide;
                break;
            case '*':
                eop = ExpressionNode::Operator::Multiply;
                break;
            default:
                assert(false && "should be unreachable");
                break;
            }
            auto expr = new ExpressionNode(t, ParseTerm(), eop);
            t = expr;
        }

        return t;
    }

    Node* ParseExpression()
    {
        auto t = ParseTerm();

        for (auto op = m_lexer->seek_token(); is_one_of(op->token, '+', '-'); op = m_lexer->seek_token())
        {
            m_lexer->NextToken();
            ExpressionNode::Operator eop;
            assert(static_cast<size_t>(ExpressionNode::Operator::COUNT_OPERATORS) == 4 && "some operators may not be handled");
            switch((char)op->token)
            {
            case '+':
                eop = ExpressionNode::Operator::Plus;
                break;
            case '-':
                eop = ExpressionNode::Operator::Minus;
                break;
            default:
                assert(false && "should be unreachable");
                break;
            }
            auto expr = new ExpressionNode(t, ParseTerm(), eop);
            t = expr;
        }

        return t;
    }

    VarDeclNode* ParseVarDecl()
    {
        m_lexer->NextExpect(TokenType::Local);
        m_lexer->NextExpect(TokenType::Id);
        auto name = m_lexer->token().string;
        m_lexer->NextExpect('=');
        Node* value = ParseExpression();
        return new VarDeclNode(name, value);
    }

    Node* ParseStatement()
    {
        auto token = m_lexer->seek_token();
        assert(token != nullptr);
        switch(token->token)
        {
        case TokenType::Local: return ParseVarDecl();
        default: return ParseExpression();
        }

        assert(0 && "unreachable");
        return nullptr;
    }

    ProgramNode* Parse()
    {
        auto program = new ProgramNode;

        while (m_lexer->NextToken())
        {
            auto token = m_lexer->token();
            switch(token.token)
            {
            case TokenType::Fn: program->PushFunction(ParseFnDecl()); break;
            case TokenType::Extern: program->PushExtern(ParseExtern()); break;
            default: {
                ErrorLogger::Raise(Error::ParseError(m_lexer->filename(), StringView::FromFormat("unexpected token while parsing '%d'", token.token), token.line_number, token.offset_start));
                assert(false);
            }
            break;
            }
        }

        return program;
    }
private:
    void Exit(int status)
    {
        std::exit(status);
    }
private:
    Lexer* m_lexer;
};