bug#20751: wc -m doesn't count UTF-8 characters properly

[Glenn Morris]

You mailed address@hidden without specifying a Package:, so your bug
report ended up on the help-debbugs list. I have reassigned it to
coreutils. (Please note there is no "wc" package.)

(My mailer is messing up the UTF-8 characters in your report.
Interested parties can see the original at http://debbugs.gnu.org/20751#5 .)

Valdis V toli   wrote:

> Version: wc (GNU coreutils) 8.21
>
> When 'wc -m' is invoked, it should print character count, but it counts
> incorrectly UTF-8 encoded characters. Attached files have 3, 4 an 6
> bytes in them, but all have only two UTF-8 encoded characters, which you
> can see with any modern text editor. 
>
> wc -c chows correct number of bytes:
> wc -c *
>  3 3bytes.txt
>  4 4bytes.txt
>  6 6bytes.txt
> 13 total
>
> But wc -m shows incorrect number of characters:
> wc -m *
>  3 3bytes.txt
>  3 4bytes.txt
>  3 6bytes.txt
>  9 total
>
> But should be:
> wc -m *
>  2 3bytes.txt
>  2 4bytes.txt
>  2 6bytes.txt
>  6 total
>
>  I am using Lubuntu 14.04.2 LTS (lsb_release reports Ubuntu), x86_64  
> GNU/Linux 3.13.0-53-generic kernel
>
> P.S.
> If attachments will not pass through system, you can test it by creating
> files with following content:
>
> 3bytes.txt: aa
> 4bytes.txt: aā
> 6bytes.txt: að  

Attachments at http://debbugs.gnu.org/20751#5