filename(CJK utf8) problem on OSX.

filename(CJK utf8) problem on OSX.

[Park SungMin]

follow code….. I just test openfile(CJK filename).
"open-file" and "call-with-input-file" are raise error.  so I test use low C(Posix) API, then run ok!
I don’t know is it bug? or option problem?
 

(use-modules  (rnrs io ports)
                	(rnrs bytevectors)
	                (system foreign)
	                (ice-9 rdelim))

(fluid-set! %default-port-encoding "utf-8")
(set-port-encoding! (current-output-port) "utf-8")
(set-port-encoding! (current-input-port) "utf-8")

(define filename "/Users/byul/Desktop/사진.gif")

(define my-open-file
  (lambda (filename)
    (let* ((fd ((pointer->procedure
		 int
		 (dynamic-func "open" (dynamic-link))
		 (list '* int))
		(bytevector->pointer
		 (string->utf8 filename))
		O_RDONLY)))
      (when (= fd -1)
	(throw 'can't-find-file))
      (fdes->inport fd))))

(define port (my-open-file filename))

(get-bytevector-n port 10)  ;; OK. I get bytevector.

;;but….
(file-exists? filename) ;; #f
(open-file filename "r") ; ERROR: In procedure open-file: No such file or directory: "/Users/byul/Desktop/사진.gif"

Re: filename(CJK utf8) problem on OSX.

[Alex Vong]

Park SungMin <byulparan_eng@icloud.com> writes:

> follow code….. I just test openfile(CJK filename).
> "open-file" and "call-with-input-file" are raise error.  so I test use low C(Posix) API, then run ok!
> I don’t know is it bug? or option problem?
>  
>
> (use-modules  (rnrs io ports)
>                 	(rnrs bytevectors)
> 	                (system foreign)
> 	                (ice-9 rdelim))
>
> (fluid-set! %default-port-encoding "utf-8")
> (set-port-encoding! (current-output-port) "utf-8")
> (set-port-encoding! (current-input-port) "utf-8")
>
> (define filename "/Users/byul/Desktop/사진.gif")
>
> (define my-open-file
>   (lambda (filename)
>     (let* ((fd ((pointer->procedure
> 		 int
> 		 (dynamic-func "open" (dynamic-link))
> 		 (list '* int))
> 		(bytevector->pointer
> 		 (string->utf8 filename))
> 		O_RDONLY)))
>       (when (= fd -1)
> 	(throw 'can't-find-file))
>       (fdes->inport fd))))
>
> (define port (my-open-file filename))
>
> (get-bytevector-n port 10)  ;; OK. I get bytevector.
>
> ;;but….
> (file-exists? filename) ;; #f
> (open-file filename "r") ; ERROR: In procedure open-file: No such file or directory: "/Users/byul/Desktop/사진.gif"

I cannot reproduce it on GNU/Linux. Perhaps others know what's going
wrong. For now, you could try setting port encoding to "UTF-8" instead
of "utf-8". Also, I remember OS X ships an ancient version of libiconv,
maybe you can try installing the latest version.

(use-modules  (rnrs io ports)
                	(rnrs bytevectors)
	                (system foreign)
	                (ice-9 rdelim))

(fluid-set! %default-port-encoding "utf-8")
(set-port-encoding! (current-output-port) "utf-8")
(set-port-encoding! (current-input-port) "utf-8")

(define filename "/home/alexvong1995/사진.gif")
(system* "touch" filename) ;; create the file if not exist
;;; $1 = 0

(define my-open-file
  (lambda (filename)
    (let* ((fd ((pointer->procedure
		 int
		 (dynamic-func "open" (dynamic-link))
		 (list '* int))
		(bytevector->pointer
		 (string->utf8 filename))
		O_RDONLY)))
      (when (= fd -1)
	(throw 'can't-find-file))
      (fdes->inport fd))))

(define port (my-open-file filename))

(get-bytevector-n port 10)
;;; $2 = #<eof>

(file-exists? filename)
;;; $3 = #t

(open-file filename "r")
;;; $4 = #<input: /home/alexvong1995/사진.gif 10>

Re: filename(CJK utf8) problem on OSX.

[tomas]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Mon, Mar 28, 2016 at 10:21:46AM +0900, Park SungMin wrote:
> follow code….. I just test openfile(CJK filename).
> "open-file" and "call-with-input-file" are raise error.  so I test use low C(Posix) API, then run ok!
> I don’t know is it bug? or option problem? 

[...]

> ;;but….
> (file-exists? filename) ;; #f
> (open-file filename "r") ; ERROR: In procedure open-file: No such file or directory: "/Users/byul/Desktop/사진.gif"

I think the port encoding has nothing to do with the encoding
of the file name. It is effective only once you've opened the
file (i.e. for the file content). For the file name, I'd say
your current locale is the relevant thing; start here:

  https://www.gnu.org/software/guile/manual/html_node/Locales.html

regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)

iEYEARECAAYFAlb4wrUACgkQBcgs9XrR2kZQLACbBVv12GD1IbReUY/cks1gavv5
lAMAni15UHKElYjKhSkimM48fMum27QN
=bfF7
-----END PGP SIGNATURE-----

Re: filename(CJK utf8) problem on OSX.

[Eli Zaretskii]

> From: Park SungMin <byulparan_eng@icloud.com>
> Date: Mon, 28 Mar 2016 10:21:46 +0900
> 
> (define filename "/Users/byul/Desktop/사진.gif")
> 
> (define my-open-file
>   (lambda (filename)
>     (let* ((fd ((pointer->procedure
> 		 int
> 		 (dynamic-func "open" (dynamic-link))
> 		 (list '* int))
> 		(bytevector->pointer
> 		 (string->utf8 filename))
> 		O_RDONLY)))
>       (when (= fd -1)
> 	(throw 'can't-find-file))
>       (fdes->inport fd))))
> 
> (define port (my-open-file filename))
> 
> (get-bytevector-n port 10)  ;; OK. I get bytevector.
> 
> ;;but….
> (file-exists? filename) ;; #f
> (open-file filename "r") ; ERROR: In procedure open-file: No such file or directory: "/Users/byul/Desktop/사진.gif"

Isn't this because OS X filesystems store file names in decomposed
form?