bug#40407: [PATCH] slow ENCODE_FILE and DECODE_FILE

["Mattias Engdegård" <mattiase@acm.org>]

[-- Attachment #1: Type: text/plain, Size: 743 bytes --]

6 apr. 2020 kl. 16.21 skrev Eli Zaretskii <eliz@gnu.org>:

> Kenichi, why was coding-type of UTF-7 systems set to 'utf-8'?
> Wouldn't it be better to set it to 'utf-16'?  Or is there some
> subtlety here that we should be aware of?  Do you have any comments on
> the patch below?

There is no reason why utf-7[-imap] should have utf-8 as coding-type, is there? utf-16 is definitely wrong (utf-7* are encoded in ASCII). What about the patch below instead?

By the way, there appears to be another, unrelated bug in utf-7-imap: According to RFC 2060, all C0 controls are base64-encoded, but in Emacs some of them are passed through unchanged (CR, LF and TAB). This is permitted by plain UTF-7 (RFC 1642) but not in the IMAP variant.


[-- Attachment #2: utf-7.diff --]
[-- Type: application/octet-stream, Size: 2474 bytes --]

diff --git a/lisp/international/mule-conf.el b/lisp/international/mule-conf.el
index e6e6135243..c5cfbaeb87 100644
--- a/lisp/international/mule-conf.el
+++ b/lisp/international/mule-conf.el
@@ -1511,20 +1511,25 @@ 'iso-safe
 
 (define-coding-system 'utf-7
   "UTF-7 encoding of Unicode (RFC 2152)."
-  :coding-type 'utf-8
+  :coding-type 'charset
+  :charset-list '(ascii)
   :mnemonic ?U
   :mime-charset 'utf-7
-  :charset-list '(unicode)
   :pre-write-conversion 'utf-7-pre-write-conversion
   :post-read-conversion 'utf-7-post-read-conversion)
+;; Having `ascii' in :charset-list automatically sets :ascii-compatible-p,
+;; but UTF-7 is not ASCII compatible; disable.
+(coding-system-put 'utf-7 :ascii-compatible-p nil)
 
 (define-coding-system 'utf-7-imap
   "UTF-7 encoding of Unicode, IMAP version (RFC 2060)"
-  :coding-type 'utf-8
+  :coding-type 'charset
+  :charset-list '(ascii)
   :mnemonic ?u
-  :charset-list '(unicode)
   :pre-write-conversion 'utf-7-imap-pre-write-conversion
   :post-read-conversion 'utf-7-imap-post-read-conversion)
+;; See comment for utf-7 above.
+(coding-system-put 'utf-7-imap :ascii-compatible-p nil)
 
 ;; Use us-ascii for terminal output if some other coding system is not
 ;; specified explicitly.
diff --git a/test/lisp/international/mule-tests.el b/test/lisp/international/mule-tests.el
index 91e3c2279f..b5fbb4ab8e 100644
--- a/test/lisp/international/mule-tests.el
+++ b/test/lisp/international/mule-tests.el
@@ -48,6 +48,19 @@ mule-cmds--test-universal-coding-system-argument
                         (append (kbd "C-x RET c u t f - 8 RET C-u C-u c a b RET") nil)))
                    (read-string "prompt:")))))
 
+(ert-deftest mule-utf-7 ()
+  ;; utf-7 and utf-7-imap are not ASCII-compatible.
+  (should-not (coding-system-get 'utf-7 :ascii-compatible-p))
+  (should-not (coding-system-get 'utf-7-imap :ascii-compatible-p))
+  ;; Invariant ASCII subset.
+  (let ((s (apply #'string (append (number-sequence #x20 #x25)
+                                   (number-sequence #x27 #x7e)))))
+    (should (equal (encode-coding-string s 'utf-7-imap) s))
+    (should (equal (decode-coding-string s 'utf-7-imap) s)))
+  ;; Escaped ampersand.
+  (should (equal (encode-coding-string "a&bcd" 'utf-7-imap) "a&-bcd"))
+  (should (equal (decode-coding-string "a&-bcd" 'utf-7-imap) "a&bcd")))
+
 ;; Stop "Local Variables" above causing confusion when visiting this file.
 \f
 

[-- Attachment #3: Type: text/plain, Size: 2 bytes --]