grammar.ebnf

(*
the indentation level is only recorded for the first token on each line,
for example:

  if true:
    print "test"

would produce the token stream:

  Token { kind: If,     indent: Some(0), lexeme: "if" }
  Token { kind: True,   indent: None,    lexeme: "true" }
  Token { kind: Colon,  indent: None,    lexeme: ":" }
  Token { kind: Print,  indent: Some(2), lexeme: "print" }
  Token { kind: String, indent: None,    lexeme: "\"test\"" }

the parser maintains a stack of indentation levels,
and supports the following operations on that stack:

  {+} -> expects indentation == Some(indent_stack[0] + N), and pushes N onto the indent_stack
  {-} -> expects indentation == Some(indent_stack[1]), and pops the top of the indent_stack
  {=} -> expects indentation == Some(indent_stack[0])
  {0} -> expects indentation == Some(0)
  {_} -> expects indentation == None

in certain contexts (such as a grouping expression, or call args), the indentation
is ignored, meaning that any indentation checks always pass, and the indentation levels
don't change.

*)


module = ({0} top_level_stmt)* ;

top_level_stmt = {=} stmt ;

stmt = scoped_stmt | simple_stmt ;

simple_stmt =
  | pass_stmt
  | return_stmt
  | continue_stmt
  | break_stmt
  | yield_stmt
  | print_stmt
  | assign_stmt
  ;

scoped_stmt =
  | import_stmt
  | if_stmt
  | for_stmt
  | while_stmt
  | loop_stmt
  | fn_stmt
  | class_stmt
  ;

pass_stmt = "pass" ;

return_stmt = "return" ({_} expr)? ;

continue_stmt = "continue" ;

break_stmt = "break" ;

yield_stmt = "yield" ({_} expr)? ;

print_stmt = "print" {_} expr ({_} "," {_} expr)? ;

assign_stmt = assign_target {_} assign_op {_} expr ;

assign_target =
  | var_expr
  | field_expr
  | index_expr
  ;

assign_op =
  | ":="
  | "="
  | "+="
  | "-="
  | "*="
  | "/="
  | "%="
  | "**="
  | "??="
  ;

import_stmt =
  | "import" {_} import_path ({_} "as" identifier)?
  | "from" {_} import_path {_} "import" {_} import_symbol_list
  ;

import_path = identifier ({_} "." {_} identifier)* ;
import_symbol_list = indentifier ({_} "," {_} identifier)* ;

if_stmt =
  "if" {_} expr {_} ":" block
  ({=} "elif" {_} expr {_} ":" block)*
  ({=} "else" ":" block)?
  ;

for_stmt = "for" {_} identifier {_} "in" {_} for_iter {_} ":" block ;

for_iter =
  | expr {_} ".." {_} expr (* range *)
  | expr                   (* iterable *)
  ;

while_stmt = "while" {_} expr {_} ":" block ;

loop_stmt = "loop" {_} ":" block ;

fn_stmt = "fn" {_} identifier {_} "(" (param ("," param)*)? ")" {_} ":" block ;

param = identifier ({_} "=" {_} expr)? ;

class_stmt = "class" {_} identifier ({_} "(" identifier ")") {_} ":" class_members ;

class_members =
  | {_} pass_stmt
  | {+} class_member_list {-}
  ;

class_member_list =
  (class_field ({=} class_field)*)?
  (class_method ({=} class_method)*)?
  ;

class_field = identifier "=" expr ;

class_method = fn_stmt ;


block =
  | {_} simple_stmt 
  | {+} stmt ({=} stmt)*
  ;

expr = maybe_expr ;

maybe_expr = or_expr ({_} "??" {_} or_expr)* ;

or_expr = and_expr ({_} "||" {_} and_expr)* ;

and_expr = eq_expr ({_} "&&" {_} eq_expr)* ;

eq_expr = comp_expr ({_} ("==" | "!=") {_} comp_expr)* ;

comp_expr = add_expr ({_} ("<" | "<=" | ">" | ">=") {_} add_expr)* ;

add_expr = mul_expr ({_} ("+" | "-") {_} mul_expr)* ;

mul_expr = pow_expr ({_} ("*" | "/" | "%") {_} pow_expr)* ;

pow_expr = unary_expr ({_} "**" {_} unary_expr)* ;

unary_expr = ("-" | "+" | "!" | "?") {_} (unary_expr | postfix_expr) ;

postfix_expr = call_expr | index_expr | field_expr | primary_expr ;

call_expr = postfix_expr {_} "(" (expr ("," expr)*)? ")" ;

index_expr = postfix_expr {_} "[" expr "]" ;

field_expr = postfix_expr {_} "." {_} identifier ;

primary_expr =
  | none_expr
  | bool_expr
  | int_expr
  | float_expr
  | string_expr
  | list_expr
  | table_expr
  | self_expr
  | super_expr
  | var_expr
  | group_expr
  ;

none_expr = "none" ;

bool_expr = "true" | "false" ;

(* NOTE: `int_expr` takes precedence over `float_expr` *)

int_expr = (* regex *) "[0-9]([0-9_]*[0-9])?" ;

float_expr = (* regex *) "[0-9]+(\.[0-9]+)?([Ee][+-]?[0-9]+)?" ;

string_expr = "\"" (* regex *) "([^\"\\]|\\.)*" "\"" ;

list_expr = "[" (expr ("," expr)*)? "]" ;

table_expr = "{" (table_field ("," table_field)*)? "}" ;

table_field = table_key ":" expr ;

table_key =
  | identifier
  | "[" expr "]"
  ;

self_expr = "self" ;

super_expr = "super" ;

var_expr = identifier ;

group_expr = "(" expr ")" ;

identifier = (* regex *) "[a-zA-Z_][a-zA-Z0-9_]*" ;