kraken/krakenGrammer.kgm

Goal = translation_unit ;
cast_expression = "\(" WS boolean_expression WS "\)" WS "cast" WS type ;
translation_unit = WS unorderd_list_part WS ;
unorderd_list_part = import WS unorderd_list_part | function WS unorderd_list_part | type_def line_end WS unorderd_list_part | adt_def line_end WS unorderd_list_part | if_comp WS unorderd_list_part | simple_passthrough WS unorderd_list_part | declaration_statement line_end WS unorderd_list_part | compiler_intrinsic line_end WS unorderd_list_part | import | function | type_def line_end | adt_def line_end | if_comp | simple_passthrough | declaration_statement line_end | compiler_intrinsic line_end ;

type = "ref" WS pre_reffed | pre_reffed ;
pre_reffed = "\*" WS pre_reffed | "void" | "char" | "uchar" | "short" | "ushort" | "int" | "uint" | "long" | "ulong" | "float" | "double" | scoped_identifier | scoped_identifier WS template_inst | function_type ;
function_type = "fun" WS "\(" WS opt_type_list WS "\)" WS ":" WS type ;
dec_type = ":" WS type ;

opt_type_list = type_list | ;
template_inst = "<" WS type_list WS ">" ;
type_list = type_list WS "," WS type | type ;

template_dec = "<" WS template_param_list WS ">" ;
template_param_list = template_param_list WS "," WS template_param | template_param ;
template_param = identifier WS traits | identifier ;

import = "import" WS identifier line_end | "import" WS identifier WS ":" WS "\*" line_end | "import" WS identifier WS ":" WS identifier_list line_end ;
identifier_list = identifier | identifier WS "," WS identifier_list ;

# all for optional semicolons      
line_break = "
+" ;
# why use line_white here but not below? who knows. It's wayy faster this way. Or maybe when I changed it there was a typing mistake. Noone knows.
line_white = "( |	)+" ;
actual_white = line_white | line_break | line_break actual_white | line_white actual_white ;
# Why is WS comment necessary? The null case SHOULD handle it, I think. I'm just a tad worred......
WS = actual_white | WS comment WS | WS comment | ;
# cpp_comment lets us do stuff like ending a statement with a cpp comment - c comments already work as they don't eat the return

maybe_line_white = "( |	)+" | c_comment | maybe_line_white c_comment | maybe_line_white "( |	)+" | ;
line_end = maybe_line_white ";" | maybe_line_white line_break | maybe_line_white cpp_comment ;
#line_end = maybe_line_white ";" | maybe_line_white line_break | maybe_line_white cpp_comment | WS c_comment line_end ;

# line_end = "( |	)+" ";" | "( |	)+" line_break | "( |	)+" cpp_comment | ";" | line_break | cpp_comment ;
# line_end = WS ";" | WS line_break | WS cpp_comment ;
# line_end = "( |	)+" ending | ending ;
# ending = ";" | line_break | cpp_comment ;

if_comp = "__if_comp__" WS identifier WS statement ;
#if_comp = "__if_comp__" WS identifier WS if_comp_pred ;
#if_comp_pred = code_block | simple_passthrough ;

simple_passthrough = "simple_passthrough" WS passthrough_params WS triple_quoted_string ;
passthrough_params = "\(" WS in_passthrough_params WS ":" WS out_passthrough_params WS ":" WS opt_string WS "\)" | ;
in_passthrough_params = opt_param_assign_list ;
out_passthrough_params = opt_param_assign_list ;
opt_param_assign_list = param_assign_list | ;
param_assign_list = param_assign WS "," WS param_assign_list | param_assign ;
param_assign = identifier WS "=" WS identifier | identifier ;
opt_string = string | ;

triple_quoted_string = "\"\"\"((\"\"(`|[0-9]|-|=|	|[a-z]|\[|]|\\|;|'|
|,|.|/|~|!|@|#|$|%|^|&|\*|\(|\)|_|\+|[A-Z]|{|}|\||:|<|>|\?| )+)|(\"(`|[0-9]|-|=|	|[a-z]|\[|]|\\|'|
|,|.|/|~|!|@|#|$|%|^|&|\*|\(|\)|_|\+|[A-Z]|{|}|\||:|<|>|\?| )+))*(`|[0-9]|-|=|	|[a-z]|\[|]|\\|;|'|
|,|.|/|~|!|@|#|$|%|^|&|\*|\(|\)|_|\+|[A-Z]|{|}|\||:|<|>|\?| )*(((`|[0-9]|-|=|	|[a-z]|\[|]|\\|;|'|
|,|.|/|~|!|@|#|$|%|^|&|\*|\(|\)|_|\+|[A-Z]|{|}|\||:|<|>|\?| )+\")|((`|[0-9]|-|=|	|[a-z]|\[|]|\\|;|'|
|,|.|/|~|!|@|#|$|%|^|&|\*|\(|\)|_|\+|[A-Z]|{|}|\||:|<|>|\?| )+\"\")|((`|[0-9]|-|=|	|[a-z]|\[|]|\\|;|'|
|,|.|/|~|!|@|#|$|%|^|&|\*|\(|\)|_|\+|[A-Z]|{|}|\||:|<|>|\?| )+))*\"\"\"" ;

#identifier = alpha_alphanumeric ;
identifier = augmented_alpha_alphanumeric ;
scope_op = ":" ":" ;
scoped_identifier = scoped_identifier WS scope_op WS identifier | identifier ;

#Note that to prevent confilct with nested templates (T<A<B>>) right_shift is a nonterminal contructed as follows
right_shift = ">" ">" ;
overloadable_operator = "\+" | "-" | "\*" | "/" | "%" | "^" | "&" | "\|" | "~" | "!" | "," | "=" | "\+\+" | "--" | "<<" | right_shift | "==" | "!=" | "&&" | "\|\|" | "\+=" | "-=" | "/=" | "%=" | "^=" | "&=" | "\|=" | "\*=" | "<<=" | ">>=" | "->" | "\(" "\)" | "\[]" | "\[]=" ;
func_identifier = identifier | identifier overloadable_operator ;
# allow omitting of return type (automatic void)

# HACKY - typed_return has it's own internal whitespace as to not make WS typed_return-reduces to null WS ambigious
typed_return = WS dec_type | ;
function = "ext" WS "fun" WS func_identifier WS "\(" WS opt_typed_parameter_list WS "\)" typed_return | "fun" WS func_identifier WS template_dec WS "\(" WS opt_typed_parameter_list WS "\)" typed_return WS statement | "fun" WS func_identifier WS "\(" WS opt_typed_parameter_list WS "\)" typed_return WS statement ;
lambda = "fun" WS "\(" WS opt_typed_parameter_list WS "\)" typed_return WS statement ;

opt_typed_parameter_list = typed_parameter_list | typed_parameter_list WS "," WS "..." | ;
typed_parameter_list = typed_parameter_list WS "," WS typed_parameter | typed_parameter ;
typed_parameter = identifier WS dec_type ;

opt_parameter_list = parameter_list | ;
parameter_list = parameter_list WS "," WS parameter | parameter ;
parameter = boolean_expression ;

def_nonterm = "def" ;
obj_nonterm = "obj" ;
type_def = def_nonterm WS identifier WS type | obj_nonterm WS identifier WS template_dec WS "{" WS declaration_block WS "}" | obj_nonterm WS identifier WS "{" WS declaration_block WS "}" | obj_nonterm WS identifier WS template_dec WS traits WS "{" WS declaration_block WS "}" | obj_nonterm WS identifier WS traits WS "{" WS declaration_block WS "}" ;

declaration_block = declaration_statement line_end WS declaration_block | function WS declaration_block | declaration_statement line_end | function | ;
traits = "\(" WS trait_list WS "\)" ;
trait_list = trait_list WS "," WS scoped_identifier | scoped_identifier ;

adt_nonterm = "adt" ;
adt_def = adt_nonterm WS identifier WS "{" WS adt_option_list WS "}" ;
adt_option_list = adt_option | adt_option WS "," WS adt_option_list ;
adt_option = identifier | identifier WS dec_type ;

if_statement = "if" WS "\(" WS boolean_expression WS "\)" WS statement | "if" WS "\(" WS boolean_expression WS "\)" WS statement WS "else" WS statement ;

match_statement = "match" WS "\(" WS boolean_expression WS "\)" WS "{" WS case_statement_list WS "}" ;
case_statement_list = case_statement WS case_statement_list | case_statement ;
case_statement = scoped_identifier WS "\(" WS identifier WS "\)" WS statement | scoped_identifier WS "\(" WS "\)" WS statement ;

while_loop = "while" WS boolean_expression WS statement ;

for_loop = "for" WS "\(" WS statement WS boolean_expression line_end WS statement WS "\)"  WS statement ;

return_statement = "return" | "return" WS boolean_expression ;

code_block = "{" WS statement_list WS "}" | "{" WS "}" ;

statement_list = statement_list WS statement | statement ;
statement = if_statement | match_statement | while_loop | for_loop | return_statement line_end | boolean_expression line_end | assignment_statement line_end | declaration_statement line_end | code_block | if_comp | simple_passthrough | break_statement | continue_statement | defer_statement ; 
break_statement = "break" ;
continue_statement = "continue" ;
defer_statement = "defer" WS statement ;
function_call = unarad WS "\(" WS opt_parameter_list WS "\)" ;

compiler_intrinsic = "#" identifier WS "\(" WS intrinsic_parameter_list WS "\)" | "#" identifier WS "<" WS type_list WS ">" ;
intrinsic_parameter_list = intrinsic_parameter_list WS "," WS intrinsic_parameter | intrinsic_parameter ;
intrinsic_parameter = identifier ;

boolean_expression = boolean_expression WS "\|\|" WS and_boolean_expression | and_boolean_expression ;
and_boolean_expression = and_boolean_expression WS "&&" WS bitwise_or | bitwise_or ;
bitwise_or = bitwise_or WS "\|" WS bitwise_xor | bitwise_xor ;
bitwise_xor = bitwise_xor WS "^" WS bitwise_and | bitwise_and ;
bitwise_and = bitwise_and WS "&" WS bool_exp | bool_exp ;
bool_exp = expression WS comparator WS expression | expression ;
comparator = "==" | "<=" | ">=" | "!=" | "<" | ">" ;

expression = expression WS "<<" WS term | expression WS right_shift WS shiftand | shiftand ;
shiftand = shiftand WS "-" WS term | shiftand WS "\+" WS term | term ;
term = term WS "/" WS factor | term WS "\*" WS factor | term WS "%" WS factor | factor ;
factor = "\+\+" WS unarad | unarad WS "\+\+" | "--" WS unarad | unarad WS "--" | "\+" WS unarad | "-" WS unarad | "!" WS unarad | "~" WS unarad | "\*" WS unarad | "&" WS unarad | unarad ;
unarad = number | scoped_identifier | scoped_identifier WS template_inst | access_operation | function_call | compiler_intrinsic | bool | string | character | "\(" WS boolean_expression WS "\)" | unarad WS "\[" WS expression WS "]" | lambda | cast_expression ; 
cast_expression = "\(" WS boolean_expression WS "\)" WS "cast" WS type ;
number = integer | floating_literal ;
access_operation = unarad WS "." WS identifier | unarad WS "->" WS identifier | unarad WS "." WS identifier WS template_inst | unarad WS "->" WS identifier WS template_inst ;

assignment_statement = factor WS "=" WS boolean_expression | factor WS "\+=" WS boolean_expression | factor WS "-=" WS boolean_expression | factor WS "\*=" WS boolean_expression | factor WS "/=" WS boolean_expression ;
# if it's being assigned to, we allow type inferencing
declaration_statement = "var" WS identifier WS "=" WS boolean_expression | "var" WS identifier WS dec_type WS "=" WS boolean_expression | "var" WS identifier WS dec_type | "var" WS identifier WS "." WS identifier WS "\(" WS opt_parameter_list WS "\)" WS dec_type ;
hexadecimal = "0x([0-9]|[a-f])+" ;
integer = numeric | hexadecimal ;
floating_literal = numeric "." float_end ;
float_end = "[0-9]+" | "[0-9]+f" | "[0-9]+d" ;
bool = "true" | "false" ;
character = "'(`|[0-9]|-|=|(\\t)|[a-z]|\[|]|(\\\\)|;|(\\')|(\\n)|,|.|/|~|!|@|#|$|%|^|&|\*|\(|\)|_|\+|[A-Z]|{|}|\||:|\"|<|>|\?| |(\\0))'" ;

keywords_also_identifiers = "obj" | "def" | "fun" | "var" | "ref" | "adt" | "cast" | "import" | "simple_passthrough" ;
alpha_alphanumeric = "([a-z]|[A-Z]|_)([a-z]|[A-Z]|_|[0-9])*" ;
augmented_alpha_alphanumeric = alpha_alphanumeric augmented_alpha_alphanumeric | keywords_also_identifiers augmented_alpha_alphanumeric | alpha_alphanumeric | keywords_also_identifiers ;

numeric = "[0-9]+" ;
# note the hacks around \things. Hmm, I feel like it actually shouldn't be like this. Added \\\* because I want to come back later
string = triple_quoted_string | "\"([#-[]|	|[]-~]|(\\\\)|(\\n)|(\\t)|(\\\*)|(\\0)|
|[ -!]|(\\\"))*\"" ;
comment = cpp_comment | c_comment ;
cpp_comment = "//[ -~]*
" ;
c_comment = "(/\*/*\**(([ -)]|[0-~]|[+-.]|	|
)/*\**)+\*/)|(/\*\*/)" ;