remove part of a text file based on a string pattern

ddaas · 01-30-2017, 03:21 AM

Hi there,
I want a bash script that removes part of a text file based on a specific pattern.
For example let's take the dhcpd.conf that contains many host containers like this:

host dellxx {
hardware ethernet 01:23:47AC:99:14;
fixed-address 10.0.0.1;

}

I want my script to take the MAC as argument and remove that host.
Like: ./remome_host 01:23:47AC:99:14
and it removes:
host dellxx {
hardware ethernet 01:23:47AC:99:14;
fixed-address 10.0.0.1;

}

How cad I do that in bash? I thaught about awk.

Thank you

Turbocapitalist · 01-30-2017, 03:24 AM

That's doable in sed, awk, or perl. Which one are you trying? Can you show us how far you have gotten and where you are stuck?

ddaas · 01-30-2017, 03:36 AM

I tried in awk but I am beginner. I'll put everything together and show you here.
Thanks

Turbocapitalist · 01-30-2017, 06:06 AM

Great. In sed and awk, you don't get much flexibility so don't try to make a generic dhcpd.conf parser. Just make one for your use case. If you need something bigger, fancier, or more complex then escalate to perl and try Net::ISC::DHCPd

grail · 01-30-2017, 12:54 PM

I would add that this is doable in bash, but awk and others are easier and probably preferable based on the task

Like many others, please remember to include a proper example, ie. a file that has both what needs to be removed and what needs to be kept.
Also include the output of the operation so it is unambiguous as to your need.

Lastly, please use [code][/code] tags around both code and example data.

MadeInGermany · 01-31-2017, 01:10 PM

Here is an awk solution

Code:

awk '
{
  if (b==0 && NR>1) {
    if (!found) { print buf } else { found=0 }
    buf=sep=""
  }
  if (index($0,search)) found=1
  b+=gsub(/\{/,"&")-gsub(/\}/,"&")
  buf=buf sep $0; sep=ORS
}
END {
  if (NR>1) if (!found) print buf
}
' search="01:23:47AC:99:14" file

Short explanation: the lines are stored into a string buffer. If leaving the { } block the buffer is printed if no search was met.

szboardstretcher · 01-31-2017, 01:20 PM

I've gotten part of it. This will return the searched for block:

Code:

awk -v RS='[^\n]*{|}' 'RT ~ /{/{p=RT} /01:23:47AC:99:14/{ print p $0 RT }' inputfile

Turbocapitalist · 01-31-2017, 01:31 PM

Quote:

Originally Posted by szboardstretcher

I've gotten part of it. This will return the searched for block:

Code:

awk -v RS='[^\n]*{|}' 'RT ~ /{/{p=RT} /01:23:47AC:99:14/{ print p $0 RT }' inputfile

Sweet. I hadn't seen RT before.

RT is only in gawk though. Debian-derivative systems seem to use mawk but gawk can be added.

dhcpd.conf files can be much more complex and ddaas might (or might not) have published just an excerpt.

pan64 · 01-31-2017, 01:39 PM

in awk it looks quite simple:
the "usual" way is to set the RS to a keyword, like \nhost<space> (or something similar), and now you can skip "lines" containing the given filter.

Code:

awk -v RS='\nhost ' '/01:23:47AC:99:14/{next}1' inputfile

MadeInGermany · 01-31-2017, 02:01 PM

pan64, the RS is missing in the output. You need to extra print it

Code:

awk 'BEGIN{RS="\nhost "} /01:23:47AC:99:14/{next} {print (NR>1?RS $0:$0)}' file

pan64 · 02-01-2017, 12:37 AM

there was a 1, probably you missed:

Code:

awk -v RS='\nhost ' '/01:23:47AC:99:14/{next}1' inputfile

Turbocapitalist · 02-01-2017, 12:46 AM

Quote:

Originally Posted by pan64

there was a 1, probably you missed:

Code:

awk -v RS='\nhost ' '/01:23:47AC:99:14/{next}1' inputfile

Even with the 1 it will show "host foo {" for the first record and then just "bar {" for the subsequent records.
The ORS needs to be set, too:

Code:

awk -v RS='\nhost ' -v ORS='\nhost ' '/01:23:47AC:99:14/{next}1' inputfile

pan64 · 02-01-2017, 01:20 AM

yes, you are right. thanks

grail · 02-01-2017, 04:19 AM

Quote:

Originally Posted by Turbocapitalist

Even with the 1 it will show "host foo {" for the first record and then just "bar {" for the subsequent records.
The ORS needs to be set, too:

Code:

awk -v RS='\nhost ' -v ORS='\nhost ' '/01:23:47AC:99:14/{next}1' inputfile

The above seems to leave a dangling 'host'

ddaas · 02-01-2017, 05:49 AM

The solution from MadeInGermany works great. I dind't understand everything, but I am on my way..

Also the first solution from szboardstretcher is ok. It returns the searched string, I still have to remove that from the initial file.

The dhcpd.conf file looks like that:

Code:

 authoritative;
option domain-name "mydomain.com";
option domain-name-servers 8.8.4.4,8.8.8.8;
default-lease-time 6000;
max-lease-time 72000;
log-facility local7;


subnet 192.168.0.0 netmask 255.255.255.0 {    
        range                      192.168.0.10 192.168.0.100;
        allow                       unknown-clients;
        default-lease-time          3600;    
        max-lease-time              64000;    
        option routers                     192.168.0.1;
        option subnet-mask                 255.255.255.0;
        option domain-name-servers         8.8.8.8;
        option ntp-servers    	           192.168.0.1;
        option domain-name                  "mydomain.com";
        
        
}
 
host hp1 {
           hardware ethernet         01:AF:55:F6:B0:22;
           fixed-address                192.168.0.130;
           option routers               192.168.0.254;
}
   


host hp2 {
           hardware ethernet        01:AF:55:F6:B0:23;
           fixed-address                192.168.0.131;
}
   

host hp3 {
           hardware ethernet        01:AF:55:F6:B0:24;
           fixed-address                192.168.0.13;
}
  
#and so on