Hi , Within Unix from a large file containing text (not always seperated with spaces) I would like to be able to look for a word , and then extract the word with the next n amount of characters or the next amount of characters untill it hits a certain character (such as a / or a ], but not extract the whole line. Any thoughts/help would be great! Jama

Just to add to it, The specific word that I want to find may not always be in the same place on each line.

I think a script is required. This is my take. Let me know if you have any questions: #!/bin/ksh word=myword while read line do if echo "$line"|grep "$word" > /dev/null then # if word exists in line str="" len=${#line} # length of line mypos=$(echo "$line"|awk ' { print index($0, "'"$word"'")} ') while [[ $mypos -le $len ]] do # stop at the end of the line # grab the character charpos=$(echo "$line"| cut -c$mypos) # get the next char position mypos=$((mypos+1)) if [[ $charpos = "/" || $charpos == "]" ]] then # stop at / or ] break fi # build the string str="$str"${charpos} done echo "$str" fi done < myfile If efficiency is an issue, you might consider greping your original file first, and place the results in a temp file and run the script against the temp file: #!/bin/ksh word=myword grep "$word" myfile > tmpmyfile while read line do str="" len=${#line} # length of line mypos=$(echo "$line"|awk ' { print index($0, "'"$word"'") } ') while [[ $mypos -le $len ]] do # stop at the end of the line # grab the character charpos=$(echo "$line"| cut -c$mypos) # get the next char position mypos=$((mypos+1)) if [[ $charpos = "/" || $charpos == "]" ]] then # stop at / or ] break fi # build the string str="$str"${charpos} done echo "$str" done < tmpmyfile

If I am understanding this right, you can probably do this one liner. grep beginword infile | sed -e "s/.*beginword/beginword/g" -e "s/endword.*/endword/g" This will look for any line with your beginning word, cut everything before it, then everything after your ending word. With the following file, using one as my begin and three as my end: nine one two three asdf asdf asdf one four seven three asdf asdf asdf four six nine asdf asdf asdf seventeen one five thirteen six three asdf asdf asdf asdf eleven twelve asdf asdf asdf asdf asdf a one six six six six five three xxsd asdfa asdf asdf asdf it produces: one two three one four seven three one five thirteen six three one six six six six five three

Thank you both for your replies . lankrypt0 have not had chance to try your one liner out , but it looks good!

grep -E There's actually a link to this option as 'egrep', because it's so common: egrep -o '(match|this|or|this)' file.txt