%header%

GRAMMARTYPE						= "LL"

DESCRIPTION						= "A grammar for SunScript."

AUTHOR							= "arookas"
VERSION							= "1.1"
DATE							= "2015/11/29"

LICENSE							= "No license."

COPYRIGHT						= "Copyright (c) 2015 arookas"

%tokens%

// whitespace
WHITESPACE						= <<[ \t\r\n]+>> %ignore%

// comments
SINGLE_LINE_COMMENT				= <<//.*>> %ignore%
MULTI_LINE_COMMENT				= <</\*([^*]|\*+[^*/])*\*+/>> %ignore%

// keywords
IMPORT							= "import"

BUILTIN							= "builtin"
FUNCTION						= "function"
VAR								= "var"
CONST							= "const"

IF								= "if"
ELSE							= "else"
DO								= "do"
WHILE							= "while"
FOR								= "for"

RETURN							= "return"
BREAK							= "break"
CONTINUE						= "continue"

YIELD							= "yield"
EXIT							= "exit"
DUMP							= "dump"
LOCK							= "lock"
UNLOCK							= "unlock"
INT								= "int"
FLOAT							= "float"
TYPEOF							= "typeof"
PRINT							= "print"

TRUE							= "true"
FALSE							= "false"

// punctuation
L_BRACE							= "{"
R_BRACE							= "}"
L_PAREN							= "("
R_PAREN							= ")"
L_BRACKET						= "["
R_BRACKET						= "]"
COLON							= ":"
SEMICOLON						= ";"
COMMA							= ","
DOT								= "."
ELLIPSIS						= "..."
QMARK							= "?"

// operators
ADD								= "+"
SUB								= "-" // doubles as the negation operator
MUL								= "*"
DIV								= "/"
MOD								= "%"

BIT_AND							= "&"
BIT_OR							= "|"
BIT_LSH							= "<<"
BIT_RSH							= ">>"

LOG_NOT							= "!"
LOG_AND							= "&&"
LOG_OR							= "||"

EQ								= "=="
NEQ								= "!="
LT								= "<"
LTEQ							= "<="
GT								= ">"
GTEQ							= ">="

ASSIGN							= "="
ASSIGN_ADD						= "+="
ASSIGN_SUB						= "-="
ASSIGN_MUL						= "*="
ASSIGN_DIV						= "/="
ASSIGN_MOD						= "%="
ASSIGN_BIT_AND					= "&="
ASSIGN_BIT_OR					= "|="
ASSIGN_BIT_LSH					= "<<="
ASSIGN_BIT_RSH					= ">>="

// literals
IDENTIFIER						= <<[_A-Za-z][_A-Za-z0-9]*>>
DEC_NUMBER						= <<-?[0-9]+\.[0-9]+>>
HEX_NUMBER						= <<-?0x[0-9A-Fa-f]+>>
INT_NUMBER						= <<-?[0-9]+>>
STRING							= <<"(\\.|[^"])*">>

%productions%

script							= root_statement+;

// statements
root_statement					=
								import_statement SEMICOLON |
								function_definition |
								builtin_declaration SEMICOLON |
								statement;
statement						=
								compound_statement SEMICOLON |
								if_statement |
								while_statement |
								do_statement SEMICOLON |
								for_statement |
								return_statement SEMICOLON |
								break_statement SEMICOLON |
								continue_statement SEMICOLON |
								yield_statement SEMICOLON |
								exit_statement SEMICOLON |
								dump_statement SEMICOLON |
								lock_statement SEMICOLON |
								unlock_statement SEMICOLON |
								statement_block;
compound_statement				= compound_statement_item {COMMA compound_statement_item};
compound_statement_item			=
								const_definition |
								variable_definition |
								variable_declaration |
								variable_assignment |
								print_statement |
								function_call;
statement_block					= L_BRACE {statement} R_BRACE;

import_statement				= IMPORT STRING;
yield_statement					= YIELD;
exit_statement					= EXIT;
dump_statement					= DUMP;
lock_statement					= LOCK;
unlock_statement				= UNLOCK;
print_statement					= PRINT argument_list;

name_label						= IDENTIFIER COLON;

// operators
assignment_operator				= ASSIGN | ASSIGN_ADD | ASSIGN_SUB | ASSIGN_MUL | ASSIGN_DIV | ASSIGN_MOD | ASSIGN_BIT_AND | ASSIGN_BIT_OR | ASSIGN_BIT_LSH | ASSIGN_BIT_RSH;
ternary_operator				= expression QMARK expression COLON expression;
binary_operator					=
								ADD | SUB | MUL | DIV | MOD | // arithmetic
								BIT_AND | BIT_OR | BIT_LSH | BIT_RSH | // bitwise
								EQ | NEQ | LT | LTEQ | GT | GTEQ | // comparison
								LOG_AND | LOG_OR; // logical
unary_operator					= LOG_NOT | SUB;

// expressions
expression						= operand {binary_operator operand};
operand							= [unary_operator_list] term;
term							=
								int_cast |
								float_cast |
								typeof_cast |
								function_call |
								variable_reference |
								STRING |
								DEC_NUMBER |
								HEX_NUMBER |
								INT_NUMBER |
								TRUE |
								FALSE |
								L_PAREN expression R_PAREN |
								L_BRACKET ternary_operator R_BRACKET; // HACK: the brackets remove ambiguity

unary_operator_list				= unary_operator+;

int_cast						= INT L_PAREN expression R_PAREN;
float_cast						= FLOAT L_PAREN expression R_PAREN;
typeof_cast						= TYPEOF L_PAREN expression R_PAREN;

// constants
const_definition				= CONST IDENTIFIER ASSIGN expression;

// variables
variable_reference				= IDENTIFIER; // used in expressions
variable_declaration			= VAR IDENTIFIER;
variable_definition				= VAR IDENTIFIER assignment_operator expression;
variable_assignment				= IDENTIFIER assignment_operator expression;

// functions
function_definition				= FUNCTION IDENTIFIER parameter_list statement_block;
function_call					= IDENTIFIER argument_list;

parameter_list					= L_PAREN [parameter {COMMA parameter}] R_PAREN; // e.g. (a, b, ...)
parameter						= IDENTIFIER | ELLIPSIS;
argument_list					= L_PAREN [argument {COMMA argument}] R_PAREN;
argument						= expression;

// builtins
builtin_declaration				= BUILTIN IDENTIFIER parameter_list;

// flow control
if_statement					= [name_label] IF expression statement [ELSE statement];
while_statement					= [name_label] WHILE expression statement;
do_statement					= [name_label] DO statement WHILE expression;
for_statement					= [name_label] FOR L_PAREN [for_declaration] SEMICOLON [for_condition] SEMICOLON [for_iteration] R_PAREN statement;
for_declaration					= compound_statement;
for_condition					= expression;
for_iteration					= compound_statement;

return_statement				= RETURN [expression];
break_statement					= BREAK [IDENTIFIER];
continue_statement				= CONTINUE [IDENTIFIER];