Re: Using one file to edit a second file.

On 2014-11-04 05:01, [email protected] [sed-users] wrote:
> Maybe I should have given a little more info or the more realistic
> issue. 

You know what they say, GIGO :-)  Give an incomplete example, you'll
get answers that don't actually solve your issues.

> file1 is actually  a list of know exceptions. The format is given
> below and is tab delimited but I can squash to a single space ... 
> 
> icmp snmp ssh   
> N       N       N       hostname              10.10.10.10

So if I understand correctly, you're only interested in the last
field of this file.

> I then have file2, a "seedfile", Im basically trying to remove the
> know exceptions from the seed file. Its format is 
> 
> 10.101.101.12 hostname Router (some other variable data)

Then you want to filter this file based on the first tab-delimited
column, excluding those items found in that first file.  I think the
following would do the trick:

  grep -vf <(sed '1d;s/.*\t\(.*\)/^\1\t/;s/\./\\./g' exceptions.txt)
  seedfile.txt 

This uses a bashism of <([command]) to create a FIFO.  If you use
another shell that doesn't support it, you can just pipe those
results to a file, use it, then delete it:

  sed '1d;s/.*\t\(.*\)/^\1\t/;s/\./\\./g' exceptions.txt > bad_ips.txt
  grep -vf bad_ips.txt seedfile.txt
  rm bad_ips.txt

This does assume that *both* files are tab-delimited.

-tim