Re: Problem with uniq command

[Philip Rowlands]

On Wed, 21 Feb 2007, Miroslaw Kozielski wrote:

> I think I found a bug in 'uniq' command. I have a text file: bla.txt
>
> / address@hidden:~/tmp$ cat bla2
> ba111.
> ba112.
> bs112.
> bm123.
> ba123.
> ba111.
>
> /After use option - c (  recede each output line with a count of the number of 
> times the line occurred in the input) I receive wrong output:
>
> /address@hidden:~/tmp$ uniq -c bla2
>  * 1 ba111.*
>   1 ba112.
>   1 bs112.
>   1 bm123.
>   1 ba123.
>   *1 ba111.*
>
> /I think that output should be:
>
> /    *2 ba111*.
>   1 ba112.
>   1 bs112.
>   1 bm123.
>   1 ba123.
> /
> How can I resolve this problem?

Thanks for the "expected output" part - it's rare to see a complete bug 
report, but in this case the problem is the assumption that uniq will 
sort its input.

Looking at the manpage, we see "Discard all but one of successive 
identical lines"; only adjacent lines will be grouped together for 
counting or dupe-suppression. In practise, this means that input to uniq 
should be sorted (in the same locale).


Cheers,
Phil