More code, added unit tests. Doesn't compile yet!

fudge-chez.ss

@@ -5,24 +5,63 @@
     (let ([c (lookahead-char file)])
       (cond
         [(char-alphabetic? c)
-          (get-char file)
+         (get-char file)
-          (string-append (string c) (get-word file))]
+         (string-append (string c) (get-word file))]
         [else ""]))))
 
-(define get-words
+(define char->digit
+  (lambda (c)
+    (- (char->integer c) (char->integer #\0))))
+
+; get-number supports only nonnegative integers.
+(define get-number
+  (lambda (file)
+    (let f ([total 0])
+      (let ([c (lookahead-char file)])
+        (if (char-numeric? c)
+          (begin
+            (get-char file)
+            (f (+ (* total 10) (char->digit c))))
+          total)))))
+
+(define get-token
   (lambda (file)
     (let ([c (lookahead-char file)])
       (cond
-        [(eof-object? c) '()]
+        [(eof-object? c) 'end-of-file]
         [(char-whitespace? c)
-          (get-char file)
+         (get-char file)
-          (get-words file)]
+         (get-token file)]
         [(char-alphabetic? c)
-          (cons (get-word file) (get-words file))]))))
+         (get-word file)]
+        [(char-numeric? c)
+         (get-number file)]
+        [(eq? c #\+) (get-char file) '+]
+        [(eq? c #\-) (get-char file) '-]
+        [(eq? c #\*) (get-char file) '*]
+        [(eq? c #\/) (get-char file) '/]
+        [(eq? c #\=) (get-char file) #\=]
+        [else (error 'get-token "Bad character" c)]))))
 
+(define get-tokens
+  (lambda (file)
+    (let ([token (get-token file)])
+      (if (eq? token 'end-of-file)
+          '()
+          (cons token (get-tokens file))))))
+    
 (define tokenize
   (lambda (filename)
     (let ([file (open-input-file filename)])
-      (get-words file))))
+      (get-tokens file))))
-  
+
-(display (tokenize "tokens.txt"))
+(define test-lexer
+  (lambda (input want)
+    (let ([port (open-string-input-port input)])
+      (let ([got (get-tokens port)])
+        (unless (eq? got want)
+          (text-lexer-fail input want got))))))
+
+(test-lexer "hello world" '("hello" "world"))
+(test-lexer "+ * / -", '(+ * / -))
+(test-lexer "identifier = 14", '("identifier" = 14))