How do I make sed delete a block of indented text below a non-indented line that also contains particular regexp, but not beyond?

Question

I've been fighting with GPT, the sed documentation, and regex for almost my entire day. I'm trying to parse through files that contain cisco prefix-list configurations and print out the lists that don't match a given regexp list name, and ignore the contents of that list or what is under the same list name and indented.

For example, my files look like this (networks obfuscated):

ip prefix-list ispA_in_pref-150
ip prefix-list ispB_in_pref-100
ip prefix-list ispC_out_deny
    seq 10 permit 23.1.24.3/24
ip prefix-list ispD_in_pref-50
    seq 10 permit 5.2.3.4/24
ip prefix-list ispD_in_pref-100
    seq 10 permit 4.18.27.58/24
ip prefix-list ispE_out_deny
    seq 10 permit 9.2.3.4/24
    seq 20 permit 4.4.4.1/24
ip prefix-list ispF_out_deny
    seq 10 permit 2.2.3.4/24
    seq 20 permit 28.2.3.1/24
ip prefix-list ispG_out_deny
    seq 10 permit 4.2.3.4/24
ip prefix-list ispH_in_pref-150
ip prefix-list ispI_out_deny
    seq 10 permit 3.2.3.4/24

What I want is:

ip prefix-list ispA_in_pref-150
ip prefix-list ispB_in_pref-100
ip prefix-list ispD_in_pref-50
    seq 10 permit 5.2.3.4/24
ip prefix-list ispD_in_pref-100
    seq 10 permit 4.18.27.58/24
ip prefix-list ispH_in_pref-150

What I keep getting is output like this (plus the sed command):

:~$ sed -e '/deny/{:a;N;/\n.*seq.*/!ba;d}' file
ip prefix-list ispA_in_pref-150
ip prefix-list ispB_in_pref-100
ip prefix-list ispD_in_pref-50
    seq 10 permit 5.2.3.4/24
ip prefix-list ispD_in_pref-100
    seq 10 permit 4.18.27.58/24
    seq 20 permit 4.4.4.1/24
    seq 20 permit 28.2.3.1/24
ip prefix-list ispH_in_pref-150

Notice how it's taking the other 2 "seq 20" lines from a different list entirely.

With GPT, it gave this as an example:

Line 1
Pattern Line
Line 2
This is a line below Line 2
Another line below Line 2
Pattern Line
Line 3

Gave this as a result:

:~$ sed -e '/Line 2/{:a;N;/\nPattern.*/!ba;d}' input_file
Line 1
Pattern Line
Line 2
Line 3

And I've been able to reproduce this, but not with the ACL files.

I could've sworn there was a way to do this with hold buffers that gave sed some semblance of context but I cannot relocate the post so I also don't know if other commands were ran with it. Is sed getting confused because of the "/" in the network notation? Is there a better way to do this?

Answer 1

Awk:

$ awk '/deny/ { deny=1; next }
       /^ip/  { deny=0; print; next }
       !deny' < data
ip prefix-list ispA_in_pref-150
ip prefix-list ispB_in_pref-100
ip prefix-list ispD_in_pref-50
    seq 10 permit 5.2.3.4/24
ip prefix-list ispD_in_pref-100
    seq 10 permit 4.18.27.58/24
ip prefix-list ispH_in_pref-150

Answer 2

Using any awk:

$ awk '/^ip/{f=/deny/} !f' file
ip prefix-list ispA_in_pref-150
ip prefix-list ispB_in_pref-100
ip prefix-list ispD_in_pref-50
    seq 10 permit 5.2.3.4/24
ip prefix-list ispD_in_pref-100
    seq 10 permit 4.18.27.58/24
ip prefix-list ispH_in_pref-150

Answer 3

Your challenge arises because the Cisco prefix-lists are hierarchical in nature, and using regular expressions with sed alone can become quite complex. It's better suited for single-line pattern matching and replacements. To achieve your desired result, you may need a more powerful text processing tool, such as awk or perl. Here's a solution using awk:

awk '
  /^ip prefix-list/ {  # If the line starts with "ip prefix-list"
    list_name = $3   # Set list_name to the third field
    match(list_name, /^.*\//)  # Check if list_name contains a slash
    if (RSTART) {  # If it does, it's a main list, print it
      print
      in_main_list = 1
    } else {  # Otherwise, it's a sub-list, ignore it
      in_main_list = 0
    }
  }
  /^ / && in_main_list {  # If the line starts with a space (sub-list) and in_main_list is true
    next  # Skip this line
  }
  in_main_list {  # If we are inside a main list, print the line
    print
  }
' file

Here's what this awk script does:

1. When it encounters a line starting with "ip prefix-list," it checks if the list name contains a slash ("/"). If it does, it's considered a main list, and the script prints it and sets a flag (in_main_list) to true.

2. If the line starts with a space and in_main_list is true, it skips the line (ignores sub-list entries).

3. If in_main_list is true, it prints the line.

This script should give you the desired output, filtering out sub-list entries.