IXML grammar for J

j.ixml

jsource : line++NL .

{ "https://johnlumley.github.io/jwiXML.xhtml" }

{ Number constant }

-number : integer; float; complex; extend; power; radian; radix .
-uptofloat : integer; float .
-uptocomplex : uptofloat; complex .
radix : (uptocomplex; power; radian), -"b", rdxstr .
@rdxstr : ["0"-"9";"a"-"z"]+ .
radian : uptofloat, rtype, uptofloat .
@rtype : "ar"; "ad" .
power : uptocomplex, ptype, uptocomplex .
@ptype : "r"; "p"; "x" .
extend : integer, -"x" .
complex : uptofloat, -'j', uptofloat .
integer : "_"?, digit+ .
float : "_" ; "__" ; ("_"?, digit+,
  (point; (point?, @expnt))) .
-point : ".", digit* .
@expnt : -"e", "_"?, digit+ .

{ Name }
name : alpha, (alpha; digit)**("_"?), locale? .
-locale : locale-name; locale-var .
locale-name : -"_", (alpha; digit)*, -"_" .
locale-var : -"__", alpha, (alpha; digit)*, locale? .

{ Primary }
single : graphic .
trail : (":"; ".")+ .
multip : -single, -trail .
keyword : ["a"-"z"], (alpha; "_"; digit)+, "." .
aword : alpha, -trail .
num_const : "_"?, ("_"; ["0"-"9"]), ":" .
left_cur : "{" .
right_cur : "}" .
left_cur_mul : "{" , -trail .
right_cur_mul : "}" , -trail .

-valid_after_any : valid_single; valid_multip; valid_paren .
-valid_after_const : valid_after_any; valid_direct;
  valid_left_cur; valid_right_cur; valid_left_cur_mul; valid_right_cur_mul .
-valid_after_trail : valid_after_any; valid_direct;
  valid_aword; valid_keyword; valid_number; valid_name; valid_string;
  valid_num_const;
  valid_left_cur; valid_right_cur; valid_left_cur_mul; valid_right_cur_mul .
-valid_after_cur : valid_after_any; valid_right_cur_mul; valid_left_cur_mul;
  valid_aword; valid_keyword; valid_number; valid_name; valid_string;
  valid_num_const .

{ Ends with trail }

-valid_single : single, valid_after_trail? .
-valid_trail  : trail, valid_after_trail? .
-valid_multip : multip, valid_after_trail? .
-valid_aword  : aword, valid_after_trail? .
-valid_keyword : keyword, valid_after_trail? .
-valid_left_cur_mul : left_cur_mul, valid_after_trail? .
-valid_right_cur_mul : right_cur_mul, valid_after_trail? .
-valid_num_const : num_const, valid_after_trail? .

{ Handle curly brackts }

-valid_left_cur : left_cur, (valid_after_cur; valid_right_cur)? .
-valid_right_cur : right_cur, (valid_after_cur; valid_left_cur; valid_direct)? .

-valid_name  : name, (valid_after_const; valid_string)? .
-valid_string  : string, (valid_after_const; valid_name;
  valid_number; valid_trail; valid_aword; valid_keyword; valid_num_const)? .
-valid_number : number, (valid_after_const; valid_string)? .

{ Paren expr }

-valid_paren : paren, group? .
-valid_direct : direct, (valid_after_cur; valid_left_cur; valid_left_cur_mul)? .

{ String }
string : "'", (~["'"; #d; #a ]; "''")*  , "'" .

{ Group }
-group : valid_single; valid_trail; valid_multip; valid_aword; valid_keyword;
 valid_left_cur_mul; valid_right_cur_mul; valid_left_cur; valid_right_cur;
 valid_name; valid_string; valid_number; valid_num_const;
 valid_paren; valid_direct .
line : blank; comment; (WHITE?, group++WHITE, WHITE?, comment?); ')'.
@blank : -" "*, +"blank" .
paren : -"(", WHITE?, group++WHITE, WHITE?, -")" .
direct : -"{{", kind?, (multiline; oneline),  -"}}" .
@kind : -")", ['mdvacn*'] .
-multiline : (line, NL)+, ((WHITE?, group++WHITE, WHITE?); WHITE)? .
-oneline : WHITE?, group++WHITE, WHITE? .

{ Comment }
comment : -"NB.", ~[#d; #a]* .
-NL: #d?, #a .
-WHITE: " "+ .
-alpha : ["a"-"z";"A"-"Z"] .
-digit : ["0"-"9"] .
-graphic : ["=<>+*-%$~|,;#!/\[]`@&?"; #5e; #22] .

{ Left right curly brackts are special }