usql/parser.cpp

#include "parser.h"
#include "exception.h"

// TOOD handle premature eof

Parser::Parser() {
	lexer = Lexer{};
}

std::unique_ptr<Node> Parser::parse(const std::string &code) {
	lexer.parse(code);
	// lexer.debugTokens();

	if (lexer.tokenType() == TokenType::keyword_create && lexer.nextTokenType() == TokenType::keyword_table) {
		return parse_create_table();
    } if (lexer.tokenType() == TokenType::keyword_insert) {
        return parse_insert_into_table();
	} if (lexer.tokenType() == TokenType::keyword_select) {
		return parse_select_from_table();
	}

	std::cout << "ERROR, token:" << lexer.currentToken().token_string << std::endl;
	return std::make_unique<Node>(NodeType::error);
}

std::unique_ptr<Node> Parser::parse_create_table() {
	std::vector<ColDefNode> cols_def {};

	lexer.skipToken(TokenType::keyword_create);
	lexer.skipToken(TokenType::keyword_table);

	if (lexer.tokenType() != TokenType::identifier)	{ /* TODO handle error */ }
	std::string table_name = lexer.consumeCurrentToken().token_string;

	lexer.skipToken(TokenType::open_paren);
	int column_order = 0;
	do {
		std::string	column_name;
		ColumnType	column_type;
		int		    column_len {1};
		bool		column_nullable {true};

		// column name
		if (lexer.tokenType() != TokenType::identifier)	{ /* TODO handle error */ }
		column_name = lexer.consumeCurrentToken().token_string;

		// column type and optionaly len
		if (lexer.tokenType() == TokenType::keyword_int) {
			column_type = ColumnType::integer_type;
			lexer.nextToken();
		} else if (lexer.tokenType() == TokenType::keyword_float) {
			column_type = ColumnType::float_type;
			lexer.nextToken();
		} else if (lexer.tokenType() == TokenType::keyword_varchar) {
			column_type = ColumnType::varchar_type;
			lexer.nextToken();
			lexer.skipToken(TokenType::open_paren);
			if (lexer.tokenType() == TokenType::int_number) {
				column_len = std::stoi(lexer.consumeCurrentToken().token_string);
			} else { /* TODO handle error */ }
			lexer.skipToken(TokenType::close_paren);
		} else { /* TODO handle error */ }

		if (lexer.tokenType() == TokenType::keyword_not) {
			lexer.nextToken();
			lexer.skipToken(TokenType::keyword_null);
			column_nullable = false;
		} else if (lexer.tokenType() == TokenType::keyword_null) {
			lexer.nextToken();
		}

		cols_def.push_back(ColDefNode(column_name, column_type, column_order++, column_len, column_nullable));

        lexer.skipTokenOptional(TokenType::comma);

		// TODO in future constraints

	} while (lexer.tokenType() != TokenType::close_paren);


	return std::make_unique<CreateTableNode>(table_name, cols_def);
}


std::unique_ptr<Node> Parser::parse_insert_into_table() {
    std::vector<Node> exec_code {};
    std::vector<ColNameNode> cols_names {};
    std::vector<ColValueNode> cols_values {};

    lexer.skipToken(TokenType::keyword_insert);
    lexer.skipToken(TokenType::keyword_into);

    // table name
    if (lexer.tokenType() != TokenType::identifier)	{ /* TODO handle error */ }
    std::string table_name = lexer.consumeCurrentToken().token_string;

    // column names
    lexer.skipToken(TokenType::open_paren);
    do {
        if (lexer.tokenType() != TokenType::identifier)	{ /* TODO handle error */ }
        cols_names.push_back(lexer.consumeCurrentToken().token_string);

        lexer.skipTokenOptional(TokenType::comma);
    } while (lexer.tokenType() != TokenType::close_paren);
    lexer.skipToken(TokenType::close_paren);

    lexer.skipToken(TokenType::keyword_values);

    // column values
    lexer.skipToken(TokenType::open_paren);
    do {
        cols_values.push_back(lexer.consumeCurrentToken().token_string);

        lexer.skipTokenOptional(TokenType::comma);
    } while (lexer.tokenType() != TokenType::close_paren);
    lexer.skipToken(TokenType::close_paren);

    return std::make_unique<InsertIntoTableNode>(table_name, cols_names, cols_values);
}

std::unique_ptr<Node> Parser::parse_select_from_table() {
    std::vector<ColNameNode> cols_names {};
    std::unique_ptr<Node> where_node;

    lexer.skipToken(TokenType::keyword_select);
    // TODO support also numbers and expressions
    while (lexer.tokenType() != TokenType::keyword_from) {
	// TODO add consumeToken() which returns token and advances to next token
        cols_names.push_back(lexer.consumeCurrentToken().token_string);
        lexer.skipTokenOptional(TokenType::comma);
    }
    lexer.skipToken(TokenType::keyword_from);
    std::string table_name = lexer.consumeCurrentToken().token_string;

    if (lexer.tokenType() == TokenType::keyword_where) {
        lexer.skipToken(TokenType::keyword_where);
        where_node = parse_where_clause();
    } else {
        where_node = std::make_unique<TrueNode>();
    }
//    if (lexer.tokenType() == TokenType::keyword_order_by) {}
//    if (lexer.tokenType() == TokenType::keyword_offset) {}
//    if (lexer.tokenType() == TokenType::keyword_limit) {}

	return std::make_unique<SelectFromTableNode>(table_name, cols_names, std::move(where_node));
}

std::unique_ptr<Node> Parser::parse_where_clause() {
    // TODO add support for multiple filters
    // TODO add support for parenthises

    auto left = parse_operand_node();
    auto operation = parse_operator();
    auto right = parse_operand_node();

    return std::make_unique<RelationalOperatorNode>(operation, std::move(left), std::move(right));
}

std::unique_ptr<Node> Parser::parse_operand_node() {
    // while not end or order or limit
    auto token_type = lexer.tokenType();
    std::string tokenString = lexer.consumeCurrentToken().token_string;
    switch (token_type) {
        case TokenType::int_number:
            return std::make_unique<IntValueNode>(std::stoi(tokenString));
        case TokenType::double_number:
            return std::make_unique<FloatValueNode>(std::stod(tokenString));
        case TokenType::string_literal:
            return std::make_unique<StringValueNode>(tokenString);
        case TokenType::identifier:
            return std::make_unique<DatabaseValueNode>(tokenString);
        default: ;
            // Throw exception
    }
}

RelationalOperatorType Parser::parse_operator() {
    auto op = lexer.consumeCurrentToken();
    switch (op.type) {
        case TokenType::equal:
            return RelationalOperatorType::equal;
        case TokenType::greater:
            return RelationalOperatorType::greater;
        default: ;
            // Throw exception
    }
}