Parser.cpp

#include <parser/Parser.h>

#include <ast/statements/CastStatement.h>
#include <ast/statements/SingleStatement.h>
#include <ast/statements/ContainerStatement.h>
#include <ast/statements/ResultFileStatement.h>
#include <ast/statements/ResultVariableStatement.h>
#include <ast/statements/VariableStatement.h>
#include <ast/statements/ValueStatement.h>
#include <ast/statements/FileStatement.h>
#include <ast/statements/ImmediateStatement.h>
#include <ast/statements/PreviousResultStatement.h>

#include <ast/args/BindedArg.h>
#include <ast/args/ImmediateArg.h>

#include <ast/command/PrintCommand.h>
#include <ast/command/ExecuteCommand.h>
#include <ast/command/QuitCommand.h>
#include <ast/command/EOTCommand.h>
#include <ast/command/HelpCommand.h>
#include <ast/command/AlgorithmsIntrospectionCommand.h>
#include <ast/command/OperatorsIntrospectionCommand.h>
#include <ast/command/OverloadsIntrospectionCommand.h>
#include <ast/command/DataTypesIntrospectionCommand.h>
#include <ast/command/CastsIntrospectionCommand.h>
#include <ast/command/BindingsIntrospectionCommand.h>
#include <ast/command/VariablesIntrospectionCommand.h>
#include <ast/command/SetCommand.h>
#include <ast/command/LoadCommand.h>
#include <ast/command/UnloadCommand.h>

namespace cli {

std::unique_ptr < CategoryOption > Parser::category_option ( ) {
	if ( check ( cli::Lexer::TokenType::COLON_SIGN ) ) {
		match ( cli::Lexer::TokenType::COLON_SIGN );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		return std::make_unique < CategoryOption > ( value );
	} else {
		return nullptr;
	}
}

std::unique_ptr < TypeOption > Parser::type_option ( ) {
	if ( check ( cli::Lexer::TokenType::COLON_SIGN ) ) {
		match ( cli::Lexer::TokenType::COLON_SIGN );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		return std::make_unique < TypeOption > ( std::move ( value ) );
	} else {
		throw exception::CommonException ( "Mismatched colon sign while expanding type_option rule. Token is " + Lexer::tokenTypeToString ( m_current.m_type ) + "." );
	}
}

std::unique_ptr < TypeOption > Parser::optional_type_option ( ) {
	if ( check ( cli::Lexer::TokenType::COLON_SIGN ) ) {
		return type_option ( );
	} else {
		return nullptr;
	}
}

std::unique_ptr < Arg > Parser::file ( ) {
	setHint ( Lexer::Hint::FILE );
	if ( check ( cli::Lexer::TokenType::HASH_SIGN ) ) {
		match ( cli::Lexer::TokenType::HASH_SIGN );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		return std::make_unique < BindedArg > ( std::move ( value ) );
	} else if ( check ( cli::Lexer::TokenType::STRING ) ) {
		return std::make_unique < ImmediateArg > ( matchString ( ) );
	} else {
		return std::make_unique < ImmediateArg > ( matchFile ( ) );
	}
}

std::unique_ptr < Arg > Parser::type ( ) {
	setHint ( Lexer::Hint::TYPE );
	if ( check ( cli::Lexer::TokenType::STRING ) )
		return std::make_unique < ImmediateArg > ( matchString ( ) );
	else {
		return std::make_unique < ImmediateArg > ( matchType ( ) );
	}
}

std::unique_ptr < Arg > Parser::arg ( ) {
	if ( check ( cli::Lexer::TokenType::HASH_SIGN ) ) {
		match ( cli::Lexer::TokenType::HASH_SIGN );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		return std::make_unique < BindedArg > ( std::move ( value ) );
	} else {
		std::string value = matchIdentifier ( );
		return std::make_unique < ImmediateArg > ( value );
	}
}

std::unique_ptr < Arg > Parser::optional_arg ( ) {
	if ( check ( cli::Lexer::TokenType::HASH_SIGN, cli::Lexer::TokenType::IDENTIFIER ) ) {
		return arg ( );
	} else {
		return nullptr;
	}
}

std::unique_ptr < Arg > Parser::template_arg ( ) {
	match ( cli::Lexer::TokenType::AT_SIGN );
	return arg ( );
}

std::unique_ptr < Arg > Parser::optional_variable ( ) {
	if ( check ( cli::Lexer::TokenType::DOLAR_SIGN ) ) {
		match ( cli::Lexer::TokenType::DOLAR_SIGN );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		return std::make_unique < ImmediateArg > ( value );
	} else {
		return nullptr;
	}
}

std::unique_ptr < Arg > Parser::optional_binding ( ) {
	if ( check ( cli::Lexer::TokenType::HASH_SIGN ) ) {
		match ( cli::Lexer::TokenType::HASH_SIGN );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		return std::make_unique < ImmediateArg > ( value );
	} else {
		return nullptr;
	}
}

std::shared_ptr < Statement > Parser::in_redirect ( ) {
	if ( check ( cli::Lexer::TokenType::LEFT_PAREN ) ) {
		match ( cli::Lexer::TokenType::LEFT_PAREN );
		std::shared_ptr < StatementList > res = statement_list ( );
		match ( cli::Lexer::TokenType::RIGHT_PAREN );
		return res;
	} else {
		std::unique_ptr < Arg > fileType;

		if ( check ( cli::Lexer::TokenType::LEFT_BRACKET ) ) {
			match ( cli::Lexer::TokenType::LEFT_BRACKET );
			fileType = arg ( );
			match ( cli::Lexer::TokenType::RIGHT_BRACKET );
		}

		std::unique_ptr < TypeOption > type = optional_type_option ( );
		ext::vector < std::unique_ptr < cli::Arg > > templateArgs;
		while ( check ( cli::Lexer::TokenType::AT_SIGN ) ) {
			templateArgs.emplace_back ( template_arg ( ) );
		}

		return std::make_shared < FileStatement > ( file ( ), std::move ( fileType ), std::move ( type ), std::move ( templateArgs ) );
	}
}

std::unique_ptr < Statement > Parser::out_redirect ( ) {
	if ( check ( cli::Lexer::TokenType::DOLAR_SIGN ) ) {
		match ( cli::Lexer::TokenType::DOLAR_SIGN );
		std::unique_ptr < Arg > name = arg ( );
		return std::make_unique < ResultVariableStatement > ( std::move ( name ) );
	} else {
		std::unique_ptr < Arg > fileType;

		if ( check ( cli::Lexer::TokenType::LEFT_BRACKET ) ) {
			match ( cli::Lexer::TokenType::LEFT_BRACKET );
			fileType = arg ( );
			match ( cli::Lexer::TokenType::RIGHT_BRACKET );
		}

		return std::make_unique < ResultFileStatement > ( file ( ), std::move ( fileType ) );
	}
}

std::shared_ptr < Statement > Parser::common ( ) {
	clearCheckOptions ( );
	if ( check ( cli::Lexer::TokenType::DOLAR_SIGN ) ) {
		match ( cli::Lexer::TokenType::DOLAR_SIGN );
		std::unique_ptr < Arg > name = arg ( );
		return std::make_shared < VariableStatement > ( std::move ( name ) );
	} else if ( check ( cli::Lexer::TokenType::LESS_SIGN ) ) {
		match ( cli::Lexer::TokenType::LESS_SIGN );
		return in_redirect ( );
	} else if ( check ( cli::Lexer::TokenType::STRING ) ) {
		std::string value = matchString ( );
		return std::make_shared < ImmediateStatement < std::string > > ( value );
	} else if ( check ( cli::Lexer::TokenType::UNSIGNED ) ) {
		int value = matchInteger ( );
		return std::make_shared < ImmediateStatement < int > > ( value );
	} else if ( check ( cli::Lexer::TokenType::HASH_SIGN ) ) {
		match ( cli::Lexer::TokenType::HASH_SIGN );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		return std::make_shared < ValueStatement > ( std::make_unique < BindedArg > ( std::move ( value ) ) );
	} else if ( check ( cli::Lexer::TokenType::LEFT_BRACE ) ) {
		match ( cli::Lexer::TokenType::LEFT_BRACE );
		std::unique_ptr < TypeOption > type = type_option ( );

		ext::vector < std::shared_ptr < Statement > > params;
		while ( ! check ( cli::Lexer::TokenType::RIGHT_BRACE ) ) {
			params.emplace_back ( param ( ) );
		}

		std::shared_ptr < Statement > res = std::make_shared < ContainerStatement > ( "Set", std::move ( params ), std::move ( type ) );
		match ( cli::Lexer::TokenType::RIGHT_BRACE );
		return res;
	} else {
		throw exception::CommonException ( "Mismatched set " + ext::to_string ( getCheckOptions ( ) ) + " while expanding common rule. Token is " + ( ( std::string ) m_current ) + "." );
	}
}

std::shared_ptr < Statement > Parser::param ( ) {
	clearCheckOptions ( );
	if ( check ( cli::Lexer::TokenType::DOLAR_SIGN, cli::Lexer::TokenType::LESS_SIGN, cli::Lexer::TokenType::STRING, cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::HASH_SIGN, cli::Lexer::TokenType::LEFT_BRACE ) ) {
		return common ( );
	} else if ( check ( cli::Lexer::TokenType::MINUS_SIGN ) ) {
		match ( cli::Lexer::TokenType::MINUS_SIGN );
		return std::make_shared < PreviousResultStatement > ( );
	} else if ( check ( cli::Lexer::TokenType::IDENTIFIER ) ) {
		std::string value = matchIdentifier ( );
		return std::make_shared < ImmediateStatement < std::string > > ( value );
	} else if ( check ( cli::Lexer::TokenType::LEFT_PAREN ) ) {
		match ( cli::Lexer::TokenType::LEFT_PAREN );
		std::unique_ptr < Arg > result_type = type ( );
		match ( cli::Lexer::TokenType::RIGHT_PAREN );
		std::shared_ptr < Statement > castedParam = param ( );
		return std::make_shared < CastStatement > ( std::move ( result_type ), castedParam );
	} else {
		throw exception::CommonException ( "Mismatched set " + ext::to_string ( getCheckOptions ( ) ) + " while expanding param rule. Token is " + ( ( std::string ) m_current ) + "." );
	}
}

std::shared_ptr < Statement > Parser::statement ( ) {
	clearCheckOptions ( );
	if ( check ( cli::Lexer::TokenType::DOLAR_SIGN, cli::Lexer::TokenType::LESS_SIGN, cli::Lexer::TokenType::STRING, cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::HASH_SIGN, cli::Lexer::TokenType::LEFT_BRACE ) ) {
		return common ( );
	} else if ( check ( cli::Lexer::TokenType::IDENTIFIER ) ) {
		std::unique_ptr < Arg > name = std::make_unique < ImmediateArg > ( matchIdentifier ( ) );
		ext::vector < std::unique_ptr < cli::Arg > > templateArgs;
		while ( check ( cli::Lexer::TokenType::AT_SIGN ) ) {
			templateArgs.emplace_back ( template_arg ( ) );
		}

		std::unique_ptr < CategoryOption > category = category_option ( );
		ext::vector < std::shared_ptr < Statement > > params;
		while ( ! check ( cli::Lexer::TokenType::MORE_SIGN, cli::Lexer::TokenType::PIPE_SIGN, cli::Lexer::TokenType::EOS, cli::Lexer::TokenType::EOT, cli::Lexer::TokenType::RIGHT_PAREN, cli::Lexer::TokenType::SEMICOLON_SIGN ) ) {
			params.emplace_back ( param ( ) );
		}

		return std::make_shared < SingleStatement > ( std::move ( name ), std::move ( templateArgs ), std::move ( params ), std::move ( category ) );
	} else if ( check ( cli::Lexer::TokenType::LEFT_PAREN ) ) {
		match ( cli::Lexer::TokenType::LEFT_PAREN );
		std::unique_ptr < Arg > result_type = type ( );
		match ( cli::Lexer::TokenType::RIGHT_PAREN );
		std::shared_ptr < Statement > castedStatement = statement ( );
		return std::make_shared < CastStatement > ( std::move ( result_type ), castedStatement );
	} else {
		throw exception::CommonException ( "Mismatched set " + ext::to_string ( getCheckOptions ( ) ) + " while expanding statement rule. Token is " + ( ( std::string ) m_current ) + "." );
	}
}

std::shared_ptr < StatementList > Parser::statement_list ( ) {
	ext::vector < std::shared_ptr < Statement > > list;
	list.emplace_back ( statement ( ) );
	while ( check ( cli::Lexer::TokenType::PIPE_SIGN, cli::Lexer::TokenType::MORE_SIGN ) ) {
		if ( check ( cli::Lexer::TokenType::PIPE_SIGN ) ) {
			match ( cli::Lexer::TokenType::PIPE_SIGN );
			list.emplace_back ( statement ( ) );
		} else {
			match ( cli::Lexer::TokenType::MORE_SIGN );
			list.emplace_back ( out_redirect ( ) );
		}
	}
	return std::make_shared < StatementList > ( std::move ( list ) );
}

std::pair < bool, bool > Parser::introspect_cast_from_to ( ) {
	bool from = false;
	bool to = false;
	while ( check ( cli::Lexer::TokenType::COLON_SIGN ) ) {
		match ( cli::Lexer::TokenType::COLON_SIGN );

		clearCheckOptions ( );
		if ( check_nonreserved_kw ( "from" ) ) {
			match_nonreserved_kw ( "from" );
			from = true;
		} else if ( check_nonreserved_kw ( "to" ) ) {
			match_nonreserved_kw ( "to" );
			to = true;
		} else {
			throw exception::CommonException ( "Mismatched set " + ext::to_string ( getCheckOptions ( ) ) + " while expanding introspect_cast_type. Token is " + ( ( std::string ) m_current ) + "." );
		}
	}
	return std::make_pair ( from, to );
}

std::unique_ptr < Command > Parser::introspect_command ( ) {
	clearCheckOptions ( );
	if ( check_nonreserved_kw ( "algorithms" ) ) {
		match_nonreserved_kw ( "algorithms" );
		std::unique_ptr < cli::Arg > param = optional_arg ( );
		return std::make_unique < AlgorithmsIntrospectionCommand > ( std::move ( param ) );
	} else if ( check_nonreserved_kw ( "overloads" ) ) {
		match_nonreserved_kw ( "overloads" );
		std::unique_ptr < cli::Arg > param = arg ( );

		ext::vector < std::unique_ptr < cli::Arg > > templateArgs;
		while ( check ( cli::Lexer::TokenType::AT_SIGN ) ) {
			templateArgs.emplace_back ( template_arg ( ) );
		}

		return std::make_unique < OverloadsIntrospectionCommand > ( std::move ( param ), std::move ( templateArgs ) );
	} else if ( check_nonreserved_kw ( "operators" ) ) {
		match_nonreserved_kw ( "operators" );
		return std::make_unique < OperatorsIntrospectionCommand > ( );
	} else if ( check_nonreserved_kw ( "datatypes" ) ) {
		match_nonreserved_kw ( "datatypes" );
		std::unique_ptr < cli::Arg > param = optional_arg ( );
		return std::make_unique < DataTypesIntrospectionCommand > ( std::move ( param ) );
	} else if ( check_nonreserved_kw ( "casts" ) ) {
		match_nonreserved_kw ( "casts" );
		std::pair < bool, bool > from_to = introspect_cast_from_to ( );
		std::unique_ptr < cli::Arg > param = optional_arg ( );
		return std::make_unique < CastsIntrospectionCommand > ( std::move ( param ), from_to.first, from_to.second );
	} else if ( check_nonreserved_kw ( "variables" ) ) {
		match_nonreserved_kw ( "variables" );
		std::unique_ptr < cli::Arg > param = optional_variable ( );
		return std::make_unique < VariablesIntrospectionCommand > ( std::move ( param ) );
	} else if ( check_nonreserved_kw ( "bindings" ) ) {
		match_nonreserved_kw ( "bindings" );
		std::unique_ptr < cli::Arg > param = optional_arg ( );
		return std::make_unique < BindingsIntrospectionCommand > ( std::move ( param ) );
	} else {
		throw exception::CommonException ( "Mismatched set " + ext::to_string ( getCheckOptions ( ) ) + " while expanding introspection_command rule. Token is " + ( ( std::string ) m_current ) + "." );
	}
}

std::unique_ptr < CommandList > Parser::parse ( ) {
	ext::vector < std::unique_ptr < Command > > list;

	if ( check ( cli::Lexer::TokenType::EOS, cli::Lexer::TokenType::EOT ) ) {
		list.emplace_back ( std::make_unique < EOTCommand > ( ) );
		return std::make_unique < CommandList > ( std::move ( list ) );
	}

	list.emplace_back ( command ( ) );
	while ( check ( cli::Lexer::TokenType::SEMICOLON_SIGN ) ) {
		match ( cli::Lexer::TokenType::SEMICOLON_SIGN );
		list.emplace_back ( command ( ) );
	}
	match ( cli::Lexer::TokenType::EOS, cli::Lexer::TokenType::EOT );
	return std::make_unique < CommandList > ( std::move ( list ) );
}

std::unique_ptr < Command > Parser::command ( ) {
	clearCheckOptions ( );
	if ( check_nonreserved_kw ( "execute" ) ) {
		match_nonreserved_kw ( "execute" );
		std::shared_ptr < StatementList > res = statement_list ( );
		return std::make_unique < ExecuteCommand > ( std::move ( res ) );
	} else if ( check_nonreserved_kw ( "print" ) ) {
		match_nonreserved_kw ( "print" );
		std::shared_ptr < StatementList > res = statement_list ( );
		return std::make_unique < PrintCommand > ( std::move ( res ) );
	} else if ( check_nonreserved_kw ( "quit" ) ) {
		match_nonreserved_kw ( "quit" );

		std::shared_ptr < StatementList > res;
		if ( ! check ( cli::Lexer::TokenType::EOS, cli::Lexer::TokenType::EOT, cli::Lexer::TokenType::SEMICOLON_SIGN ) )
			res = statement_list ( );

		return std::make_unique < QuitCommand > ( res );
	} else if ( check_nonreserved_kw ( "exit" ) ) {
		match_nonreserved_kw ( "exit" );

		std::shared_ptr < StatementList > res;
		if ( ! check ( cli::Lexer::TokenType::EOS, cli::Lexer::TokenType::EOT, cli::Lexer::TokenType::SEMICOLON_SIGN ) )
			res = statement_list ( );

		return std::make_unique < QuitCommand > ( res );
	} else if ( check_nonreserved_kw ( "help" ) ) {
		match_nonreserved_kw ( "help" );
		std::unique_ptr < cli::Arg > command = optional_arg ( );
		return std::make_unique < HelpCommand > ( std::move ( command ) );
	} else if ( check_nonreserved_kw ( "introspect" ) ) {
		match_nonreserved_kw ( "introspect" );
		std::unique_ptr < Command > command = introspect_command ( );
		return command;
	} else if ( check_nonreserved_kw ( "set" ) ) {
		match_nonreserved_kw ( "set" );
		std::string param = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER );
		std::string value = getTokenValue ( );
		match ( cli::Lexer::TokenType::UNSIGNED, cli::Lexer::TokenType::IDENTIFIER, cli::Lexer::TokenType::STRING );
		return std::make_unique < SetCommand > ( std::move ( param ), std::move ( value ) );
	} else if ( check_nonreserved_kw ( "load" ) ) {
		match_nonreserved_kw ( "load" );

		setHint ( Lexer::Hint::FILE );
		std::string libraryName = getTokenValue ( );
		match ( cli::Lexer::TokenType::FILE, cli::Lexer::TokenType::STRING );

		return std::make_unique < LoadCommand > ( std::move ( libraryName ) );
	} else if ( check_nonreserved_kw ( "unload" ) ) {
		match_nonreserved_kw ( "unload" );

		setHint ( Lexer::Hint::FILE );
		std::string libraryName = getTokenValue ( );
		match ( cli::Lexer::TokenType::FILE, cli::Lexer::TokenType::STRING );

		return std::make_unique < UnloadCommand > ( std::move ( libraryName ) );
	} else {
		throw exception::CommonException ( "Mismatched set " + ext::to_string ( getCheckOptions ( ) ) + " while expanding parse rule. Token is " + ( ( std::string ) m_current ) + "." );
	}
}

} /* namespace cli */