byteify

pitfall/pitfall/minimal-pdf-source.rkt

@@ -1,7 +1,6 @@
 #lang s-exp pitfall/render
 
-(co-version 1.1)
-(co-comment "%¥±ë")
+(co-header "%PDF-1.1\n%¥±ë")
 (co-io
  1
  0

pitfall/pitfall/parse.rkt

@@ -7,7 +7,8 @@
 (module+ reader (provide read-syntax))
 
 (define (read-syntax src port)
-  (define parse-tree (parse (make-tokenizer port src)))
+  ;; use latin-1 reencoding to make one char = one byte
+  (define parse-tree (parse (make-tokenizer (open-input-string (bytes->string/latin-1 (port->bytes port))) src)))
   (strip-bindings
    #`(module pitfall-parse-mod pitfall/parse
        #,parse-tree)))
@@ -84,6 +85,6 @@
 (define-macro (pf-indirect-object-ref (OBJ GEN _))
   #'(co-io-ref OBJ GEN))
 
-(define (pf-version num) (co-version num))
+(define (pf-header num) (co-header num))
 
 (define (pf-comment text) (co-comment text))

pitfall/pitfall/parser.rkt

@@ -1,7 +1,7 @@
 #lang brag
 
 pf-program : pf-object*
-@pf-object : pf-null | CHAR | BOOLEAN | INT | REAL | pf-name | pf-string | pf-array | pf-dict | pf-stream | pf-indirect-object | pf-indirect-object-ref | pf-version | pf-comment
+@pf-object : pf-null | CHAR | BOOLEAN | INT | REAL | pf-name | pf-string | pf-array | pf-dict | pf-stream | pf-indirect-object | pf-indirect-object-ref | pf-header | pf-comment
 @pf-null : NULL
 pf-name : NAME
 pf-string : STRING-TOK | /"<" HEX-DIGIT-PAIR* /">"
@@ -12,5 +12,5 @@ pf-dict : /"<" /"<" (pf-dict-key pf-dict-value)* /">" /">"
 pf-stream : pf-dict STREAM-DATA
 pf-indirect-object : INT INT /"obj" pf-object /"endobj"
 pf-indirect-object-ref : INDIRECT-OBJECT-REF-TOK
-pf-version : PDF-VERSION
+pf-header : PDF-HEADER
 pf-comment : COMMENT

pitfall/pitfall/render.rkt

@@ -21,6 +21,7 @@
     (cond
       [(co-version? x)
        @string-append{%%PDF-@(number->string (co-version-num x))}]
+      [(co-header? x) (co-header-string x)]
       [(co-array? x)
        @string-append{[ @(string-join (map loop (co-array-items x)) " ") ]}]
       [(co-io? x)

pitfall/pitfall/struct.rkt

@@ -5,6 +5,7 @@
 (struct co-array (items) #:transparent)
 (struct co-stream (dict data) #:transparent)
 (struct co-version (num) #:transparent)
+(struct co-header (string) #:transparent)
 (struct co-io (idx rev thing) #:transparent)
 (struct co-io-ref (idx rev) #:transparent)
 (struct co-comment (text) #:transparent)

pitfall/pitfall/tokenizer.rkt

@@ -12,8 +12,8 @@
 (define-lex-abbrev blackspace (:~ pdf-whitespace))
 (define-lex-abbrev not-right-paren (:~ ")"))
 (define-lex-abbrev substring (:seq "(" (:* not-right-paren) ")"))
-
 (define-lex-abbrev nonreg-char (:seq "#" hex-digit hex-digit))
+(define-lex-abbrev pdf-eol (:or #\return #\newline (:: #\return #\newline)))
 
 (define (make-tokenizer port [src #f])
   (port-count-lines! port)
@@ -24,7 +24,8 @@
      [(:seq "%%EOF" any-string) eof]
      [(:seq digits (:+ pdf-whitespace) digits (:+ pdf-whitespace) "R")
       (token 'INDIRECT-OBJECT-REF-TOK (map string->number (string-split lexeme)))]
-     [(:seq "%PDF-" digits "." digits) (token 'PDF-VERSION (string->number (trim-ends "%PDF-" lexeme "")))]
+     [(:seq "%PDF-" digits "." digits
+            (:? (:: pdf-eol "%" (:>= 4 (:~ (union #\return #\newline)))))) (token 'PDF-HEADER lexeme)]
      [pdf-whitespace (token 'IGNORE lexeme #:skip? #t)]
      [(from/stop-before "%" #\newline) (token 'COMMENT lexeme)]
      [(:or "true" "false") (token 'BOOLEAN (equal? lexeme "true"))]