Re: printf doesn't work properly for non-ASCII
On Wed, Mar 10, 2004 at 03:35:07PM +0200, Anton Zinoviev wrote:
> I decided to try how partman works with localised strings and observed
> that some of the strings almost overlap on the screen. Investigation
> showed that this was caused by printf.
> In several places in the code of partman are used commands like
> printf "(%5s)\n" $var
> The problem is that printf considers the length of the strings in
> bytes as length in symbols which is not true for UTF-8. For example
> if var='пя' (two Cyrillic letters) then the quoted printf command will
> output "( пя)" instead of "( пя)". For Asian languages the
> situation can be even worse because there 3 or 4 bytes represent only
> one character.
> All versions of printf I tested are broken -- the internal printf
> command in bash, /usr/bin/printf and busybox. Does someone know how
> this can be fixed?
Using of wide character functions may help but this will require
rewrite of all code that works with localized strings.