shell 脚本中的复杂排序

Question

I have written a script that fulfills the extraction requirement but I cannot sort the results according to the formatting requirements. This is my script:

awk '{print $9"  "$1}' ab.log | sort | uniq -c | awk '{print $2 "\t" $3}' | sort -nr

It shows the status code and IP address in descending order;

404   89.86.144.219
404   81.192.148.245
.
.
403   172.6.0.3
403   129.16.26.39
402   145.8.0.9
402   256.23.4.57
.
.
401   126.158.20.9

But the requirements are: status code groups have to be sorted by which status code group appears more often and the IP addresses have to be sorted by occurrence within each of the groups.

So the output should be:

404    127.0.0.1
404    xxx.xxx.xxx.xxx
.
.
200    xxx.xxx.xxx.xxx
200    xxx.xxx.xxx.xxx
.
.
403    xxx.xxx.xxx.xxx
403    xxx.xxx.xxx.xxx

How can I sort the result according to the above requirements?

Answer 1

There is probably an amazing single command you can use for this. However here is a pipeline using awk, sort and cut:

$ awk '{a[$1]++;b[$0]=$1}END{for(i in b) printf "%-9d%s\n",a[b[i]],i}' | sort -k1,1nr -k2,2n -k3,3 | cut -c9-

With GNU awk, you could even make it into a single command, but it just will start to look messy