UNIX for Dummies Questions & Answers

Hi all,

1. I have a file that is getting continously refreshed (appended)
I want to grep all the strings containing substring of the type

Code:

abcdf123@aaa.xxx.yyy.zzz:portnumber:

where,

before @, any letters or numbers combination,
after @, IP address
then symbol :
then port number (any number)
then :

2. All these strings to be written in a filename in a format "username" "ip"

3. I already use tail -f to get the updates to this file (say x.txt)

Any ideas?
Thank you in advance.

Moderator's Comments:

Please use [code] and [/code] tags when posting code, data or logs etc. to preserve formatting and enhance readability, thanks.

What is the delimeter in front of the username? Or maybe post a complete example line using code tags, thanks.

before the username is a space.
after the :
there is a space too.

I have the + escaped with \. You will have to try out if this is necessary with your sed too; I am on cygwin at the moment so I am not sure.

Code:

$ cat infile
aaa bbb abcdf123@aaa.xxx.yyy.zzz:portnumber: ccc ddd

zaxxon@saburac ~
$ sed 's/.* \([^@]\+\)@\([^:]\+\):.*/\1 \2/' infile > newfile

zaxxon@saburac ~
$ cat newfile
abcdf123 aaa.xxx.yyy.zzz

In your case you would use it like:

Code:

tail -f somelog | sed 's/.* \([^@]\+\)@\([^:]\+\):.*/\1 \2/' >> user_ip.out

Hi Zaxxon.

Thank you very much for your assitstance.

The sed part worked fine, it cathced the requested type strings exactly, from a given file.
The

Code:

 tail -f <reportFile> | sed ......      >>  <outputFile>

didn' produce any output though
Also is it possible to provide some logic in constructing the regular expression in the sed part? It seems quite magical for me!

Thank you again.

---------- Post updated at 03:46 PM ---------- Previous update was at 10:45 AM ----------

I found the problem about no writing to output file. The file had to be preexisting...