How to parse columns from a log with empty string using scripts

Suppose a log result includes several columns delimited with wild spaces. For some columns like PREEMPTIBLE, EXTERNAL_IP, sometimes they have values, sometime they do not have. See below as an example. In this case, how can I get other columns like INTERNAL_IP?

run: gcloud compute instances list  --project=XXX, get:
NAME  ZONE  MACHINE_TYPE  PREEMPTIBLE INTERNAL_IP EXTERNAL_IP STATUS                                       
Name1  zone1  type1       true        5.6.7.8     1.2.3.4     RUNNING
Name2  zone1  type1       true        6.5.8.7                 RUNNING
Name3  zone2  type2                   6.5.8.7                 RUNNING

In fact, if all the columns are non-empty, then I can get INTERNAL_IP using:

gcloud compute instances list  --project=XXX |tr -s ' ' |awk -F' ' '{if (NR>1)print $5}'

because INTERNAL_IP is the 5th column, or

gcloud compute instances list  --project=XXX |tr -s ' ' |awk -F' ' '{if (NR>1)print $(NF-2)}'

because INTERNAL_IP is the 3rd of the last column

But if the columns like PREEMPTIBLE and EXTERNAL_IP have empty string, or sometimes they are completely empty, then it is hard to parse columns like INTERNAL_IP because for some rows INTERNAL_IP is the 5th column, for some rows INTERNAL_IP is the 4th column, etc.

So, ow to get INTERNAL_IP in this case? Thanks.

Solution

You can get the index of INTERNAL_IP on the first line and use it to reach internal ip on the following lines:

gcloud compute instances list  --project=XXX |\
awk 'NR==1{ii=index($0,"INTERNAL_IP")}NR>1{$0=substr($0,ii);sub(" +.*","");print}'