Add lexing for immediates, keywords, and associated tests

3 files changed, 53 insertions, 16 deletions

diff --git a/src/lex.lisp b/src/lex.lisp
index ad386ba..c6e9cf7 100644
--- a/src/lex.lisp
+++ b/src/lex.lisp
@@ -1,19 +1,34 @@
 (in-package #:lex)
 
+(define-condition invalid-immediate-or-keyword (error)
+  ((chr :initarg :chr
+        :initform nil
+        :reader chr)
+   (instance :initarg :instance
+             :initform nil
+             :reader instance))
+  (:report (lambda (condition stream)
+             (format stream
+                     "Lex failed--encountered ~a while reading ~a."
+                     (chr condition) (instance condition))))
+  (:documentation "Dedicated error for immediates/keywords which contain
+invalid characters."))
+
 (defun file->tokens (file)
   "Opens FILE and parses returns a list of tokens, or
 NIL if the file could not be opened."
 
-  (defun read-tokens (tokens-so-far)
-    "Collects tokens in FILE into TOKENS-SO-FAR."
+  (defun read-instr (lst tokens-so-far)
+    "Collects tokens in FILE into TOKENS-SO-FAR, splitting on a newline."
     (let ((token (read-token)))
-      (if token
-          (read-tokens (cons token tokens-so-far))
-          (reverse tokens-so-far))))
+      (cond ((null token) (reverse tokens-so-far))
+            ((eq token 'nl)
+             (cons (reverse tokens-so-far) (read-instr nil nil)))
+            (t (read-instr lst (cons token tokens-so-far))))))
 
   (and (probe-file file)
        (with-open-file (*standard-input* file :direction :input)
-         (read-tokens '()))))
+         (remove nil (read-instr '() '())))))
 
 (defun read-token ()
   "Reads *STANDARD-INPUT* and returns a token, or nil if the end
@@ -23,29 +38,49 @@ Comments start with a semi-colon ';' and all tokens after are ignored."
   (let ((chr (read-char *standard-input* nil)))
     (cond
       ((null chr) chr)
+
+      ((char= chr #\linefeed) 'nl)
+
       ((whitespace-char-p chr)
        (read-token))
 
       ((char= chr #\;)
        (progn (read-line *standard-input* nil)
-              (read-token)))
+              'nl))
 
       ((char= chr #\() 'left-paren)
       ((char= chr #\)) 'right-paren)
 
+      ((char= chr #\:) 'colon)
+      ((char= chr #\$) 'dollar)
+
       ((digit-char-p chr)
        (read-immediate chr))
 
       ((alpha-char-p chr)
-       (read-identifier chr))
+       (read-keyword chr))
 
       (t (error (format nil "~a is not a valid lexical symbol.~%" chr))))))
 
 (defun read-immediate (chr)
-  'immediate)
+  (defun read-immediate-helper (chrs-so-far)
+    (let ((chr (peek-char nil *standard-input* nil)))
+      (cond ((and (not (null chr)) (digit-char-p chr))
+             (read-immediate-helper (cons (read-char *standard-input* nil) chrs-so-far)))
+            ((and (not (null chr)) (alpha-char-p chr))
+             (error 'invalid-immediate-or-keyword :chr chr :instance "immediate"))
+            (t (reverse chrs-so-far)))))
+  (parse-integer (coerce (read-immediate-helper (list chr)) 'string)))
 
-(defun read-identifier (chr)
-  'id)
+(defun read-keyword (chr)
+  (defun read-keyword-helper (chrs-so-far)
+    (let ((chr (peek-char nil *standard-input* nil)))
+      (cond ((and (not (null chr)) (alpha-char-p chr))
+             (read-keyword-helper (cons (read-char *standard-input* nil) chrs-so-far)))
+            ((and (not (null chr)) (digit-char-p chr))
+             (error 'invalid-immediate-or-keyword :chr chr :instance "keyword"))
+            (t (reverse chrs-so-far)))))
+  (coerce (read-keyword-helper (list chr)) 'string))
 
 (defun whitespace-char-p (x)
   (or (char= #\space x)
diff --git a/src/main.lisp b/src/main.lisp
index 98176ec..f6e5754 100644
--- a/src/main.lisp
+++ b/src/main.lisp
@@ -41,13 +41,14 @@ _/_/                                  _/_/  "
          (emit? (not (clingon:getopt cmd :parse))))
     (cond
       ;; complain about num arguments
-      ((/= (length args) 1) (error "Wrong number of arguments."))
+      ((/= (length args) 1) (error "Wrong number of arguments.~%"))
       ((not (util:asm-extension? file))
-       (error "The file is not an asm source code file."))
+       (error "The file is not an asm source code file.~%"))
       (t (let ((tokens (lex:file->tokens file)))
            (if tokens
-               (format t "~a" tokens)
-               (error "The file does not exist, or it could not be opened."))
+               (progn (pprint tokens)
+                      (terpri))
+               (error "The file does not exist, or it could not be opened.~%"))
            (format t "Nitimur in Vetitum~%"))))))
 
 
diff --git a/src/package.lisp b/src/package.lisp
index 44399cb..670ed02 100644
--- a/src/package.lisp
+++ b/src/package.lisp
@@ -10,4 +10,5 @@
   (:use #:cl)
   (:export #:file->tokens
            ;; exported for testing only
-           #:read-token))
+           #:read-token
+           #:invalid-immediate-or-keyword))