Given a phrase like this: WORD1 WORD2: WORD3 WORD4 WORD5 I just want WORD3 WORD4 WORD5. I can tail the file it comes from to get the last line, but how do I get only those last few words? It needs to be done autonomously (i.e. I can't do it through vi or something like that- it's part of a script). I thought I could use sed to replace the spaces with newlines and then just get the lines I want, and swap 'em back, but then I realized that sed can't substitute newlines in, and getting 'em back out is no easy solution either. Any suggestions? -- N. McSpadden

Two ways: echo "WORD1 WORD2: WORD3 WORD4 WORD5"|cut -d " " -f3- echo "WORD1 WORD2: WORD3 WORD4 WORD5"|awk ' { e=NF-2 for(i=e; i<=NF; i++) if(i < NF) printf("%s ", $i) else printf("%s\n ", $i) }

Thanks! Using cut worked great. One more question: If I do a command that has results that look like this: 49 Mount Point: /A/Stuff 65 Mount Point: / How do I specifically grep for the line that has the mount point of /? If I do a grep for "Mount Point: ", I get the results above. However, I need to differentiate the root volume from a secondary one, but I can't find a regular expression that will allow me to do so. I thought this might work: results | grep "/[\n]" Using the regular expression for newline after the / so that the line /A/stuff would be ignored, but it gives me an empty result. (Assume that "results" is a command that gives the output shown above). I've tried a number of different regular expressions, but none of them seem to work. How do I distinguish the / that is followed by nothing from the / that is followed by by other stuff? Thanks!

How about searching for the string with a / at the end of the line: grep 'Mount Point: /$' data.file This works fine on my Solaris 7 box. If grep doesn't work for you, try egrep.

Wow, that was way easier. /sigh, time to go look over my Regular Expressions guides again. One last question, since you've already been so helpful: So let's say I have the line: 65 Mount Point: / Now I can use cut to get the line count at the beginning of this input (the line is a result of a command | cat -n | grep "Mount Point: /$"). Using cut, I can get the characters that are the numbers: 65 However, if I store this in a variable, it stores it literally, like a string. If I want to use this for mathematical operations (i.e. subtracting this from the total line count given via wc -l), how can I do that? If I try: LineDifference = $LineCount (65) - 5 results in echo $LineDifference = 65 -5 which doesn't do what I want at all. Any last suggestions? -- N. McSpadden

With the modern shells such as ksh and bash, this is how you do arithmetic: #!/bin/ksh LineCount=65 LineDifference=$((LineCount-5)) echo $LineDifferenc

with python , it's even easier. Assuming a string "WORD1 WORD2: WORD3 WORD4 WORD5" and ":" separates them theline = "WORD1 WORD2: WORD3 WORD4 WORD5" last3words = theline.split(":")[1].strip() word1,word2,word3 = last3words.split() print word1,word2,word3 Its more readable compared to shell commands.