[SOLVED] Identify line in hierarchical file bash script

Issue

I need to identify the head of a list of configuration entries in a file. There is not predictable, it could be any string, but it always be the line that starts closer to the left than the others (excluding "exit"):

Here is an example:

    vpls 2662 customer 1 v-vpls vlan 2662 create
        description "RES_2662"
        mac-move
            allow-res-res
            allow-reg-res
        exit
        stp
            shutdown
        exit
        ingress
            qos 2
        exit
        sap lt:1/1/1:2662 create
            description "RES_2662"
            enable-stats
            no shutdown
        exit
        sap lag-1:2662 create
            no shutdown
        exit
        no shutdown
    exit
    vpls 2663 customer 1 v-vpls vlan 2663 create
        description "RES_2663"
        mac-move
            allow-res-res
            allow-reg-res
        exit
        stp
            shutdown
        exit
        ingress
            qos 2
        exit
        sap lt:1/1/1:2663 create
            description "RES_2663"
            enable-stats
            no shutdown
        exit
        sap lag-1:2663 create
            no shutdown

In this case, I need to be able to identify the two lines that start with: vpls 266X customer 1 v-vpls vlan 266X create The script should know that these are the lines that I'm looking for.

The output not always will show spaces at the left, like in this example:

port vlan-port:1/1/1/3/7/4/4:824
  admin-up
  severity no-value
exit
port vlan-port:1/1/1/3/7/4/4:1224
  admin-up
  severity no-value
exit

In this case, the desired lines are: port vlan-port:x/x/x/x/x/x/x/x

I don't know if it can be done using grep/sed/awk.

Thanks for your assistance.

Solution

The following will work using any awk in any shell on every Unix box and will retain the order of the input lines for the output in case that matters:

$ cat tst.awk
$1 != "exit" {
    match($0,/^ */)
    if ( (min == "") || (RLENGTH <= min) ) {
        min = RLENGTH
        lines[min,++cnt[min]] = $0
    }
}
END {
    for (i=1; i<=cnt[min]; i++) {
        print lines[min,i]
    }
}

$ awk -f tst.awk file
    vpls 2662 customer 1 v-vpls vlan 2662 create
    vpls 2663 customer 1 v-vpls vlan 2663 create

Answered By - Ed Morton

Answer Checked By - Gilberto Lyons (WPSolving Admin)