Re: sanitized mm-string-to-multibyte

[Dave Love <fx@gnu.org>]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> Your version of `mm-string-to-multibyte' doesn't seem to convert
> a unibyte string to a multibyte string.  In Emacs 21.1~21.4 I got:
>
> (let* ((s1 (string-as-unibyte "a"))
>        (s2 (mm-with-preserved-unibyte (string-make-multibyte s1))))
>   (list (multibyte-string-p s1) (multibyte-string-p s2)))
>  => (nil nil)
>
> (let* ((s1 (string-as-multibyte "a"))
>        (s2 (mm-with-preserved-unibyte (string-make-multibyte s1))))
>   (list (multibyte-string-p s1) (multibyte-string-p s2)))
>  => (t t)
>
> Did I miss something?

I think it doesn't cons a new string in the trivial case like that, when
it won't matter.  Use this version if you want always to cons a
multibyte string.

Index: mm-util.el
===================================================================
RCS file: /usr/local/cvsroot/gnus/lisp/mm-util.el,v
retrieving revision 7.91
diff -u -r7.91 mm-util.el
--- mm-util.el	14 Jan 2009 00:52:01 -0000	7.91
+++ mm-util.el	9 Feb 2009 23:26:20 -0000
@@ -202,6 +202,22 @@
     (defalias 'mm-decode-coding-region 'decode-coding-region)
     (defalias 'mm-encode-coding-region 'encode-coding-region)))
 
+(defconst mm-identity-nat (let (l)
+			    (dotimes (i 256)
+			      (push (cons i i) l))
+			    (make-translation-table l))
+  "Non-ASCII translation table that applies the identity translation.")
+
+(defmacro mm-with-preserved-unibyte (&rest body)
+  "Execute BODY forms while preserving unibyte characters.
+Such characters are not converted automatically to multibyte ones
+when, for instance, inserted into a multibyte buffer within the
+BODY forms."
+  `(let ((nonascii-translation-table mm-identity-nat))
+     ,@body))
+(put 'mm-with-preserved-unibyte 'lisp-indent-function 0)
+(put 'mm-with-preserved-unibyte 'edebug-form-spec '(body))
+
 ;; `string-to-multibyte' is available only in Emacs 22.1 or greater.
 (defalias 'mm-string-to-multibyte
   (cond
@@ -210,11 +226,9 @@
    ((fboundp 'string-to-multibyte)
     'string-to-multibyte)
    (t
-    (lambda (string)
-      "Return a multibyte string with the same individual chars as STRING."
-      (mapconcat
-       (lambda (ch) (mm-string-as-multibyte (char-to-string ch)))
-       string "")))))
+    (lambda (s)
+      (mm-with-preserved-unibyte
+	(concat s (eval-when-compile (string-as-multibyte ""))))))))
 
 ;; `char-or-char-int-p' is an XEmacs function, not available in Emacs.
 (eval-and-compile