Is there a sed or perl expression to increment a specific number in a file?-CodePudding

I have a file with two lines, The numbers in the file when written to will vary ranging from 01 to over 100. For example:

ln -s /dev/vg_ifmx_test_core4/rdsk_web59 /dev/itest17/dbs_web59
/usr/sbin/lvcreate -L 2048 -n rdsk_web59 /dev/vg_ifmx_test_core4

I need to increment the ending number (in this case it is 59) by one.

Resulting in:

ln -s /dev/vg_ifmx_test_core4/rdsk_web60 /dev/itest17/dbs_web60
/usr/sbin/lvcreate -L 2048 -n rdsk_web60 /dev/vg_ifmx_test_core4

Notice how /dev/vg_ifmx_test_core4 does not get incremented,

So is it possible to take in account the scenario of:

ln -s /dev/vg_ifmx_test_core4/rdsk_web04 /dev/itest17/dbs_web04
/usr/sbin/lvcreate -L 2048 -n rdsk_web04 /dev/vg_ifmx_test_core4

Resulting in

ln -s /dev/vg_ifmx_test_core4/rdsk_web05 /dev/itest17/dbs_web05
/usr/sbin/lvcreate -L 2048 -n rdsk_web05 /dev/vg_ifmx_test_core4

The lines in the file can vary (the path that is) such as:

/usr/sbin/lvcreate -L 2048 -n dbs_saleitem_f105 /dev/vg_ifmx_test_std3
ln -s/dev/vg_ifmx_test_std3/rdbs_saleitem_f105/dev/itest7/saleitem_f105

Would result in:

/usr/sbin/lvcreate -L 2048 -n dbs_saleitem_f106 /dev/vg_ifmx_test_std3
ln -s/dev/vg_ifmx_test_std3/rdbs_saleitem_f106/dev/itest7/saleitem_f106

So using something as perl -i -pe '/ln -s /dev/vg_ifmx_test_core4/rdsk_web/ would not be possible since the path is not always the same in the file.

Is there a way to use sed or awk/perl to create an expression to do this? I am new to shell scripting and using these commands. Thanks in advance!

CodePudding user response：

perl -pe 'if ($. == 1) {
              ($n) = ($m) = /([0-9] )$/;
                $m;
              s/$n/$m/g;
          } else {
              s/$n(?= )/$m/;
          }' file

CodePudding user response：

With GNU awk for the third arg to match():

awk '
    match($0,/(.*[^0-9])([0-9] )( .*[^0-9])([0-9] )/,a) {
        a[2]  
        a[4]  = (NR==1 ? 1 : 0)
        $0 = a[1] a[2] a[3] a[4]
    }
    { print }
' file
ln -s /dev/vg_ifmx_test_core4/rdsk_web60 /dev/itest17/dbs_web60
/usr/sbin/lvcreate -L 2048 -n rdsk_web60 /dev/vg_ifmx_test_core4

To write the output back to the original file just change awk to awk -i inplace.

CodePudding user response：

$ cat file
ln -s /dev/vg_ifmx_test_core4/rdsk_web04 /dev/itest17/dbs_web04
/usr/sbin/lvcreate -L 2048 -n rdsk_web04 /dev/vg_ifmx_test_core4

awk '{
      gsub(/web[0-9][0-9]/,
           "web"sprintf("%0.2d",gensub(/.*web([0-9][0-9]).*/, "\\1",1) 1))
}1' file

ln -s /dev/vg_ifmx_test_core4/rdsk_web05 /dev/itest17/dbs_web05
/usr/sbin/lvcreate -L 2048 -n rdsk_web05 /dev/vg_ifmx_test_core4

CodePudding user response：

I would prefer to either set a variable that can be trivially changed, or use a templating language to create the file. Either seems simpler to maintain and less brittle than awk/perl/etc.

If "file" will be executed by a shell:

N=60
ln -s /dev/vg_ifmx_test_core4/rdsk_web$N /dev/itest17/dbs_web$N
/usr/sbin/lvcreate -L 2048 -n rdsk_web$N /dev/vg_ifmx_test_core4

Now it is trivial to change the first line.

Or, if you want to keep the original format:

# generate the file from template
mkfile(){
    cat <<EOD
ln -s /dev/vg_ifmx_test_core4/rdsk_web$1 /dev/itest17/dbs_web$1
/usr/sbin/lvcreate -L 2048 -n rdsk_web$1 /dev/vg_ifmx_test_core4
EOD
}

# ...
mkfile 60 >/path/to/file
# ...

for n in 59 60 04 05; do
    # ...
    mkfile $n >/path/to/file
    # ...
fi