sed replace regex with regex

Question

This sounds like a really banal question, and it's bugging me dearly.

I have a text file, generated by python, that contains a list of host names and ip addresses in the following format:

host46.provider.com = 192.168.0.1

However, because the file was generated asynchronously, there's a few strange errors in the output. I'm getting lines that look like this:

us9831.provider.com = 262.99.124.183us9832.provider.com = 121.92.213.10

The problem is that it hasn't inserted a newline between the two instances. I've ascertained that (fortunately) the naming convention of the nodes doesn't allow for a number to come before a letter, unless it's before a dot (.)

So what I neeed to happen is:

• find an instance where a number comes before a dot
• add a newline between the two

I can't figure out how I'll get sed to maintain the regex match, and copy that over to the next regex.

I've tried this:

cat eg | sed '/[0-9][a-z]/s/[0-9][a-z]/[0-9]\n[a-z]/'

and I have tried this:

cat eg | sed "s/[0-9][a-z]/\n/"

Is sed the right tool? Should I be using awk?

Any suggestions would be warmly welcomed.

Answer 1

I'm not sure what the unless it's before a dot (.) part of your requirements means but if your requirement is really just that a number can't be immediately before a letter then if your sed supports \n to produce a newline in the replacement text:

$ sed 's/\([0-9]\)\([[:alpha:]]\)/\1\n\2/' file
us9831.provider.com = 262.99.124.183
us9832.provider.com = 121.92.213.10

Otherwise with any sed:

$ sed 's/\([0-9]\)\([[:alpha:]]\)/\1\
\2/' file
us9831.provider.com = 262.99.124.183
us9832.provider.com = 121.92.213.10

Answer 2

echo "us9831.provider.com = 262.99.124.183us9832.provider.com = 121.92.213.10"| sed "s/[0-9]*\.[0-9]*\.[0-9]*\.[0-9]*/&\n/g"

output

us9831.provider.com = 262.99.124.183
us9832.provider.com = 121.92.213.10