All: I have a file, following this format: [START FILE] CN=user01 telephoneNumber=123-4567 whenChanged=20081118132545.0Z employeeID=00001885 mail=user.01@home.com CN=user02 telephoneNumber=234-5678 whenChanged=20081116194043.0Z employeeID=00001984 mail=user.02@home.com CN=user03 telephoneNumber=345-6789 whenChanged=20081118124212.0Z employeeID=00001988 [END FILE] I want to reformat like this (BUT ONLY FOR THOSE RECORDS WHOSE whenChanged VALUE IS 20081118*): employeeID:telephoneNumber:mail 00001885:123-4567:user.01@home.com 00001984:234-5678:user.02@home.com 00001988:345-6789:: Any ideas? Thanks in advance. Preferable shell is KSH. ./Brett

Parse by the = sign, then f1 is your main field name, employeeid, whenchanged,etc. and f2 is the value. Print it out when the new header flag CN appears. Need to skip the first header CN and print out the last record: #!/bin/ksh tnum="" wchg="" eid="" mm="" ff=0 while IFS="=" read f1 f2 f3 do if [[ $ff -eq 0 && "$f1" = "CN" ]] then # skip the first CN ff=1 continue fi if [[ "$f1" = "telephoneNumber" ]] then tnum="$f2" fi if [[ "$f1" = "whenChanged" ]] then wchg="$f2" fi if [[ "$f1" = "employeeID" ]] then eid="$f2" fi if [[ "$f1" = "mail" ]] then mm="$f2" fi # print if first field is CN if [[ "$f1" = "CN" ]] then case "$wchg" in 20081118*) echo "$eid:$tnum:$mm" tnum="" wchg="" eid="" mm="" ;; esac fi done < data4.txt # print the last record case "$wchg" in 20081118*) echo "$eid:$tnum:$mm" ;; esac

Nails: Well done, thank you very much for the example. //Brett