Script needed for deceptively difficult find and replace

mjl3434 · 01-23-2018, 07:09 PM

Guys,

I'm trying to automatically do a find and replace in a bind config file, and I'm having a heck of a time coming up with a simple bash script that will do what I want. The bind config file has several sections like this:

Code:

view "name-of-view1" {
    match-clients {
        x.y.z.0/24;
    };
    forward only;
    include "/etc/bind/forwarder-file";

    ....
};

I want to find only a subset of the overall views, and then replace the include line with something else. For example suppose there are 10 views, "name-of-view1", "name-of-view2", ... all 10 with the exact same include line. Suppose I want to only change the include lines for the 3rd, 7th and 10th views to a different forwarder file.

Ideally (but not required) I would like to do this efficiently in only one pass. I have no doubt I can come up with a way to accomplish it in multiple passes, but I'm stuck on it trying to do it efficiently.

keefaz · 01-23-2018, 07:58 PM

With perl, a small script:

Code:

#!/usr/bin/perl

my %configs = (
	3 => 'config1.conf', 
	7 => 'another_config.conf', 
	10 => 'other.conf'
);

my $conf;

while(<>) {
	/^view.*?(\d+)/ and $conf = $configs{$1};
	s:(\s+include ")[^"].*(".*):$1/etc/bind/$conf$2: and $conf=undef if $conf;
	print;
}

Fill %configs hash with the specific view numbers and configuration filenames (don't include path)
name the script and chmod +x it
execute it like

Code:

./script.pl < /etc/named_config > /etc/named_config.new

edit: didn't fully read the question and the request about changing specific views so used a hash with numeric keys instead of array

rigor · 01-23-2018, 08:10 PM

Hi mjl3434,

I have to say, I tend to think that bash might not be a great choice for such a task. Here's a trivial GAWK script that replaces the include file for each view. It's also rather trivial to modify it to contain a list of index of the views you would actually like to change, and then and in the further condition in the "if" statement. Alternatively, if you must use BASH, I believe a rather similar approach should work.

Code:

BEGIN   {
            view_include = "include \"/etc/bind/forwarder-file\";" ;
            rep_view_include = "include \"/etc/bind/new-forwarder-file\";" ;
        }

{
    if (  $0  !~  view_include  )
        print $0 ;
    else
        print "    "  rep_view_include ;
}

As a simple example:

Code:

#!/bin/bash

view_include='include "/etc/bind/forwarder-file";' ;

view_indexes=( 3  7  10 ) ;
include_files=( 'include "/etc/bind/forwarder-3-file";'   'include "/etc/bind/forwarder-7-file";'   'include "/etc/bind/forwarder-10-file";'  )  ;

read a_line
result=$? ;
include_count=1 ;
ra_index=0 ;

while  [[  $result  -eq  0  ]]
    do

        if [[  $a_line  =~  $view_include  ]]
            then

                if [[  $include_count  -eq  ${view_indexes[$ra_index]}  ]]
                    then
                        echo "    "  ${include_files[$ra_index]}
                        ra_index=$(( ra_index + 1 ))
                    else
                        /bin/echo $a_line
                fi

                include_count=$(( include_count + 1 ))

            else
                /bin/echo $a_line
        fi

        read a_line
        result=$?

    done

when fed a data file created from your example data file, but extended to contain 10 views, the script outputs a new config. file with the 3rd, 7th, and 10th includes replaced. It would just need a little tweaking to reproduce the whitespace as desired.

HTH.