implement expr lexer

kaitai/expr.rkt

@@ -0,0 +1,76 @@
+#lang racket/base
+
+(require racket/function racket/match racket/set
+         parser-tools/lex
+         (prefix-in : parser-tools/lex-sre)
+         parser-tools/yacc)
+
+(define-tokens kaitai-expr [boolean number string identifier])
+(define-empty-tokens kaitai-empty [eof + - * / % < <= > >= == != << >> & pipe ^ not and or ? :
+                                   lparen rparen lbracket comma dot rbracket])
+
+(define (kaitai-numstr->number str)
+  (match (regexp-replace* #px"_" str "")
+    [(pregexp #px"^0x(.*?)$" (list _ num)) (string->number num 16)]
+    [(pregexp #px"^0b(.*?)$" (list _ num)) (string->number num 2)]
+    [(pregexp #px"^0o(.*?)$" (list _ num)) (string->number num 8)]
+    [num (string->number num)]))
+
+(define (kaitai-subst-string str)
+  ;; racket's built in string translation works fine
+  (printf "uwu ~s\n" str)
+  (read (open-input-string
+          (string-append "\"" (substring str 1 (sub1 (string-length str))) "\""))))
+
+(define kaitai-lexer
+  (lexer
+    [(:or "true" "false") (token-boolean (string=? "true" lexeme))]
+    [(:or (:: "0x" (:+ (:or numeric (char-set "abcdefABCDEF_"))))
+          (:: "0b" (:+ (char-set "01_")))
+          (:: "0o" (:+ (char-set "01234567_")))
+          (:: (:? (char-set "+-"))
+              (:+ (:or numeric "_"))
+              (:? (:: "." (:* (:or numeric "_"))))
+              (:? (:: (char-set "eE") (:? (char-set "+-")) (:+ (:or numeric "_"))))))
+     (token-number (kaitai-numstr->number lexeme))]
+    [(:or (:: "'" (:* (:- any-char "'") "\\'") "'")
+          (:: "\"" (:* (:- any-char "\"") "\\\"") "\""))
+     (token-string (kaitai-subst-string lexeme))]
+    [whitespace (kaitai-lexer input-port)]
+    ["+" (token-+)]
+    ["-" (token--)]
+    ["*" (token-*)]
+    ["/" (token-/)]
+    ["%" (token-%)]
+    ["<" (token-<)]
+    ["<=" (token-<=)]
+    [">" (token->)]
+    [">=" (token->=)]
+    ["==" (token-==)]
+    ["!=" (token-!=)]
+    [">>" (token->>)]
+    ["<<" (token-<<)]
+    ["&" (token-&)]
+    ["|" (token-pipe)]
+    ["^" (token-^)]
+    ["not" (token-not)]
+    ["and" (token-and)]
+    ["or" (token-or)]
+    ["?" (token-?)]
+    [":" (token-:)]
+    ["(" (token-lparen)]
+    [")" (token-rparen)]
+    ["[" (token-lbracket)]
+    ["," (token-comma)]
+    ["." (token-dot)]
+    ["]" (token-rbracket)]
+    [(:: (:+ (:or alphabetic "_"))
+         (:* (:or alphabetic numeric "_" "::")))
+     (token-identifier lexeme)]
+    [(eof) (token-eof)]))
+
+(let ([input (open-input-string "_root._io.size.to_i & 0x08000000 != 0 + code == block_type::int32 ? 4 : 8")])
+  (let loop ()
+    (match (kaitai-lexer input)
+      [(? (curry equal? (token-eof))) (void)]
+      [x (printf "~v\n" x) (loop)])))

kaitai/info.rkt

@@ -2,6 +2,8 @@
 
 (define version "0.0.1")
 (define collection "kaitai")
-(define deps '("base" "yaml" "binaryio-lib"))
+(define deps '("base"
+               "binaryio-lib"
+               "yaml"))
 (define test-omit-paths '("info.rkt"))
 ; (define scribblings '(("kaitai.scrbl"))')

kaitai/main.rkt

@@ -1,6 +1,6 @@
 #lang racket/base
 
-(require (for-syntax racket/base racket/list))
+(require (for-syntax racket/base racket/list racket/match))
 (provide (rename-out [kaitai:module-begin #%module-begin])
          (except-out (all-from-out racket/base) #%module-begin))
 
@@ -42,13 +42,6 @@
       [#,@(map (lambda (a) (kaitai-str->sym (hash-ref a "id" (make-rse "seq element has no id"))))
                seq)] #:transparent)))
 
-;; generates parser function
-(define-for-syntax (gen-parser top [given-id #f])
-  (define-values [id-sym seq] (get-id-seq top given-id))
-  #`((define (,(sym+ "parse-" id-sym) input)
-       (void))
-     (provide ,(sym+ "parse-" id-sym))))
-
 ;; runtime lib
 
 (module runtime racket/base
@@ -66,10 +59,8 @@
 
 (define-syntax (kaitai:module-begin stx)
   (define body (second (syntax-e stx)))
-  (define top-struct (gen-struct (syntax->datum body)))
-  ; (displayln top-struct)
+  (displayln (syntax->datum body))
   #`(#%module-begin
-     #,@top-struct
      (module+ main
        (kaitai:uwu))))