Added strict ASCII codec.

[lisp-utils.git] / charcode.lisp

diff --git a/charcode.lisp b/charcode.lisp
index 2a051ae..1d52dae 100644 (file)
--- a/charcode.lisp
+++ b/charcode.lisp
@@ -3,9 +3,10 @@
 
 (defpackage :charcode
   (:use :cl #+sbcl :sb-gray #-sbcl :gray)
-  (:export "MAKE-ENCODER" "MAKE-DECODER" "ENCODE-STRING" "DECODE-STRING"
+  (:export "MAKE-ENCODER" "MAKE-DECODER" "ENCODE-STRING" "DECODE-STRING" "SYSTEM-CHARSET"
           "CODING-ERROR"
-          "LATIN-1" "LATIN1" "UTF-8" "UTF8"))
+          "MAKE-CODEC-CHARACTER-STREAM"
+          "ASCII" "LATIN-1" "LATIN1" "UTF-8" "UTF8"))
 (in-package :charcode)
 
 ;;; General stuff
@@ -50,7 +51,11 @@
 (defun make-decoder (name)
   (the decoder-fun (values (funcall (get name 'make-decoder)))))
 
-(defun encode-string (string coding)
+(defun system-charset ()
+  ;; XXX: Replace me with something perhaps more sensible.
+  'utf-8)
+
+(defun encode-string (string &optional (coding (system-charset)))
   (declare (type string string))
   (let ((encoder (make-encoder coding))
        (buf (make-array (list (length string)) :element-type '(unsigned-byte 8) :adjustable t :fill-pointer 0)))
@@ -58,7 +63,7 @@
       (coding-error string (length string) buf "Encoding of string in ~A ended prematurely." coding))
     buf))
 
-(defun decode-string (buffer coding)
+(defun decode-string (buffer &optional (coding (system-charset)))
   (declare (type (array (unsigned-byte 8)) buffer))
   (let ((decoder (make-decoder coding))
        (buf (make-array (list (length buffer)) :element-type 'character :adjustable t :fill-pointer 0)))
@@ -75,6 +80,10 @@
    (read-pos :initform 0)
    (buffer :initform (make-array '(64) :element-type 'character :adjustable t :fill-pointer 0))))
 
+(defun make-codec-character-stream (real-stream &optional (charset (system-charset)))
+  (declare (type stream real-stream))
+  (make-instance 'codec-character-stream :decoder (make-decoder charset) :encoder (make-encoder charset) :back real-stream))
+
 (defmethod close ((stream codec-character-stream) &key abort)
   (with-slots (back) stream
     (close back :abort abort))
@@ -186,6 +195,53 @@
   (declare (type character char))
   (char-code char))
 
+;;; ASCII
+
+(defun decode-ascii (byteseq charseq &key (start 0) (end (length byteseq)))
+  (declare (type (array (unsigned-byte 8)) byteseq)
+          (type (array character) charseq)
+          (type fixnum start end))
+  (loop
+     (restart-case
+        (loop
+           (unless (< start end) (return-from decode-ascii t))
+           (let ((byte (aref byteseq (prog1 start (incf start)))))
+             (unless (< byte 128)
+               (coding-error byteseq start charseq "Invalid byte ~D in ASCII stream." byte))
+             (vector-push-extend (unicode->char byte) charseq)))
+       (:replace-char (&optional (replacement (unicode->char #xfffd)))
+        :report "Replace the invalid byte with a character."
+        (vector-push-extend replacement charseq))
+       (:skip-char ()
+        :report "Ignore the invalid byte."
+        nil))))
+
+(defun encode-ascii (charseq byteseq &key (start 0) (end (length charseq)))
+  (declare (type (array (unsigned-byte 8)) byteseq)
+          (type (array character) charseq)
+          (type fixnum start end))
+  (loop
+     (restart-case
+        (loop
+           (unless (< start end) (return-from encode-ascii t))
+           (vector-push-extend (let ((cp (char->unicode (aref charseq (prog1 start (incf start))))))
+                                 (unless (< cp 128)
+                                   (coding-error charseq start byteseq "ASCII cannot encode code-points higher than 128."))
+                                 cp)
+                               byteseq))
+       (:replace-char (&optional (replacement #\?))
+        :report "Replace this character with another."
+        (vector-push-extend (char->unicode replacement) byteseq))
+       (:skip-char ()
+        :report "Ignore this character."
+        nil))))
+
+(define-decoder (ascii)
+  #'decode-ascii)
+
+(define-encoder (ascii)
+  #'encode-ascii)
+
 ;;; Latin-1
 
 (defun decode-latin-1 (byteseq charseq &key (start 0) (end (length byteseq)))