bug#72617: sort -n loses lines.

[Simon B]

On Wed, 14 Aug 2024 at 11:48, Pádraig Brady <P@draigbrady.com> wrote:
>
> tag 72617 notabug
> close 72617
> stop
>
> On 14/08/2024 09:43, Simon B wrote:
> > Hallo,
> >
> > The output of my grep command is:
> >
> > # grep -i "sshd" /root/access.report | egrep -o
> >   
> > '(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-
> > 9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)'
> > 64.227.127.122
> > 172.169.5.249
> > 172.169.6.164
>
> Adding the --debug option shows the issue.
> I.e. the '.' being considered as part of a number:
>
> $ sort --debug -rbn -s ips
> sort: text ordering performed using ‘en_IE.UTF-8’ sorting rules
> sort: note numbers use ‘.’ as a decimal point in this locale
> 178.128.44.128
> _______
> 172.169.6.164
> _______
> 172.169.5.249
> _______
>
>
> Taking the example for sorting IPv4 addresses from the manual,
> shows the desired comparisons being performed:
>
> $ sort --debug -t '.' -k 1,1rn -k 2,2rn -k 3,3rn -k 4,4rn -u ips
> sort: text ordering performed using ‘en_IE.UTF-8’ sorting rules
> sort: numbers use ‘.’ as a decimal point in this locale
> 178.128.44.128
> ___
>      ___
>          __
>             ___
> 172.169.6.164
> ___
>      ___
>          _
>            ___
> 172.169.5.249
> ___
>      ___
>          _
>            ___
>

Hi Pádraig

I am largely satisfied by your great explanation,

I am still confused why lines go "missing" though.

Unsorted output
# grep -i "sshd" /root/access.report | egrep -o
'(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)'
| wc -l
41
Sorted once:
# grep -i "sshd" /root/access.report |  egrep -o
'(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)'
| sort -urbn | wc -l
15
Sorted workaround:
# grep -i "sshd" /root/access.report | egrep -o
'(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)'
| sort -urb | sort -nr | wc -l
17

Even if the dot is being interpreted, it still should not lose the
line containing 172.169.6.164

Regards

Simon