uahgi2/lexer.rkt

#lang br
(require brag/support)

(define-lex-abbrev digits (:+ (char-set "0123456789")))
(define-lex-abbrev raw-id (:: (:or "_" (:/ "a" "z") (:/ "A" "Z")) (:* (:or "_" (:/ "0" "9") (:/ "a" "z") (:/ "A" "Z")))))
(define-lex-abbrev any-other-char (:~ (char-set "0123456789@\\{}%|")))

(define uahgi2-lexer
  (lexer-srcloc
   [(:or "\r\n" "\n") (token 'NEWLINE lexeme)] ; newline
   [whitespace (token lexeme lexeme)] ; whitespace
   [(:: "\\"(char-set "@\\{}%|")) (token 'CHAR (substring lexeme
                      1 (string-length lexeme)))]
  [(:: "@" raw-id) (token 'ID (substring lexeme
                      1 (string-length lexeme)))] ; @id 
  ["{" (token 'L_PAREN lexeme)] ; {
  ["}" (token 'R_PAREN lexeme)] ; }
  ["|" (token 'SEPERATOR lexeme)] ; | (general) seperator
  ["@@" (token 'G_SEPERATOR lexeme)] ; @@ great seperator
  ["`" (token 'NUM_CONVERTER lexeme)] ; ` convert to number
   [digits (token 'INTEGER (string->number lexeme))] ;123
   [(:: (:? (char-set "-")) digits "." digits) ; 123.456
    (token 'DECIMAL (string->number lexeme))]
   [ (from/to "%" "%") ; %COMMENT%
    (token 'COMMENT #:skip? #t)]
  [any-other-char (token 'CHAR lexeme)])) ; any char into char
(provide uahgi2-lexer)