[SOLVED] Replace sequential numbers in a file with a different sequence using sed

thefiend · 04-11-2010, 04:16 PM

Hello,

I am trying to find a way to replace a set of sequential numbers in a file with a different sequence using sed. This might be done easier using awk or some sort of bash script, but it seems to me there must be a way to do this easily with sed. Basically, what I am editing is a Cisco switch config. I want to change the sequence of ports to a different numbered sequence. Here is an example of what I am trying to do.

I want to change for example, the file:

Code:

cat testfile
interface FastEthernet0/1
 speed 10
 switchport access vlan 10
 spanning-tree portfast
!
interface FastEthernet0/2
 speed 10
 switchport access vlan 10
 spanning-tree portfast
!
interface FastEthernet0/3
 speed 100
 switchport access vlan 11
 spanning-tree portfast
!
interface FastEthernet0/4
 speed 10
 switchport access vlan 11
 spanning-tree portfast

To:

Code:

interface FastEthernet0/5
 speed 10
 switchport access vlan 10
 spanning-tree portfast
!
interface FastEthernet0/6
 speed 10
 switchport access vlan 10
 spanning-tree portfast
!
interface FastEthernet0/7
 speed 100
 switchport access vlan 11
 spanning-tree portfast
!
interface FastEthernet0/8
 speed 10
 switchport access vlan 11
 spanning-tree portfast

All that is being changed is the line "interface FastEthernet0/X" where X is the sequential value I want to change from 1-4 to 5-8.

I am fairly new to sed and linux in general, but I expected the replacement field in sed to also be able to handle a regular expression. However this command did not work:

cat testfile | sed 's/\/[1-4]/\/[5-8]/g'

This just replaces the FastEthernet0/X lines all with FastEthernet0/[5-8]. I have also tried the y/ switch in sed, which works for this example, but not for the full file I am trying to edit:

cat testfile | sed '/0\/[1-4]$/ y/[1234]/[5678]/'

This works, but I really want to replace the range 1-22 with 26-48, so this would be a really long sed statement:

cat testfile | sed '/0\/[1-22]$/ y/[123456789,10,11...]/[26,27,28,...]/'

Plus I have yet to figure out how to make y/ work with double digit entries.... so the above is probably a horrible example of y/.

Any help would really be appreciated, even if it is suggestions for awk or some sort of script.

Regards,

thefiend

Tinkster · 04-11-2010, 05:34 PM

Hi, welcome to LQ!

Given that the ranges are crossing the ones/tens border I'd suggest awk
which can do some maths for you.
Something like this (quick and dirty):

Code:

/FastEthernet0/{
  new="0"gensub(/interface FastEthernet0\/(.*)/, "\\1","1",$0)
  print "interface FastEthernet0/"(new + 22)
  next
}
{
  print
}

grail · 04-11-2010, 09:02 PM

You could go a little simpler assuming the whole file looks as you have posted:

Code:

awk 'BEGIN{FS="/"}$0~FS{$2=$2+22}1' testfile

Tinkster · 04-12-2010, 12:28 AM

Nice ... I was thinking way too complicated :}

Kenhelm · 04-12-2010, 11:12 AM

This needs GNU sed and bash.
It uses the 's' command to replace the line in the pattern space with bash code made with back references, \1 and \2, to the line.
The GNU sed 'e' command executes the shell code found in the pattern space, replacing the code with the code's output.

Code:

sed -r 's|(interface FastEthernet0/)(.+)|echo \1$((\2+25))|e'

# Pattern space contents at each step
interface FastEthernet0/1               # Original line
echo interface FastEthernet0/$((1+25))  # bash code to be executed
interface FastEthernet0/26              # New line

thefiend · 04-12-2010, 02:37 PM

Hello Everyone,

Thank you for the quick answers, I really appreciate it. This was my first post here. I always find answers here when searching google, so I figured I would sign up and contribute as best I can. Also, this particular question had me puzzled for a few days!

@Tinkster, thanks for the quick and dirty version

@grail, your solution was great too. However, I did have to adjust it a bit to get the result I wanted. Your example replaced the slash with a space:

Code:

awk 'BEGIN{FS="/"}$0FS{$2=$2+22}1' testfile
interface FastEthernet0 32
...

So I had to modify it (quick and dirty) with sed to replace the slash:

Code:

awk 'BEGIN{FS="/"}$0~FS{$2=$2+22}1' testfile | sed 's/t0 /t0\//g'
interface FastEthernet0/32
...

@Kenhelm, thank you so much for that one, I really felt like there must be a way to do this with sed. I was just starting to read about the back references and such, but was far from figuring it out. Your solution worked perfectly and uses sed, so I will mark this post as solved

Thank you!!

thefiend

grail · 04-12-2010, 10:29 PM

Nah ... it's much easier than that:

Code:

awk 'BEGIN{OFS=FS="/"}$0~FS{$2=$2+22}1' testfile