Shell Programming and Scripting

Forgive me if this is a dumb question...I'm a Windows sys admin with little programming knowledge.

I have files containing anywhere from 3 to 200 lines. Using SED, I want to extract only lines containing a specific instance of "ISS." It is possible that "ISS" will occur several times in a line or not, but the only instance of "ISS" I want considered will always be located at, let's say, the 87th character position if you're counting from left to right on the line.

So if "ISS" is located at position 87, remove that instance of "ISS" from the line, then append the line to a new file.

If anyone could help with this I'd be eternally grateful!

-Tony

I am not sure exactly what you want, but maybe this is it...

sed -n 's/^\(.\{86\}\)ISS.*$/\1/p' < inputfile >> anotherfile

No wait, maybe you want this...

sed -n 's/^\(.\{86\}\)ISS\(.*/)$/\1\2/p' < inputfile >> anotherfile

Hi Perderabo,

Thanks very much for your response. I've tried both of your suggestions, and they both output a blank file, though its entirely possible I could be doing something wrong. Was it intentional that the command counts to the 86th character position instead of the 87th?

Thanks,
Tony

No way I'm typing in an example with 90+ characters. Here is one with ISS at position 11 in a couple of lines....

Code:

$ cat data
12345678901234567890
ISS 123   ISS
sjjsISSjjjISS1234
$ sed -n 's/^\(.\{10\}\)ISS.*$/\1/p' < data
ISS 123
sjjsISSjjj
$

Only lines 2 and 3 were output. Line 1 was ignored, no "ISS" at position 11. The early ISS in line 3 is retained. The ISS at position 11 and all characters following that ISS were dropped. The second version would drop only the ISS, so in line 3, that trailing "1234" would move over to position 11.

So it is working exactly like I expected. And as I said, I am only guessing as to what you wanted. I guess that I guessed wrong. *shrug* You can't win them all.

Hey, I'm thankful you took the time to guess at all!

I figured out one reason why it didn't work for me...I had to use double quotes before s and after p instead of the single quotes.
Once I switched those out, your first example worked for me.

Your second example is actually closer to what I wanted, but I couldn't get it working. Let me offer some clarification (which I'm sure you'd have appreciated beforehand...sorry! hard for me to think in programmer terms)

After finding "ISS" in position x, that instance should be replaced with 3 empty spaces, and everything following should be retained and written to the new file. All other instances of ISS in that line are retained as well.

I should also mention I'm using sed utilities that have been ported to Windows. The two I'm playing with were found below:

http://users.cybercity.dk/~bse26236/.../SED.HTM#11.80

When I try your second script using this one, I get this error: sed: garbled command s/^\(.\{86\}\)ISS\(.*/)$/\1\2/p


http://sed.sourceforge.net/grabbag/ssed/

With this one (super sed) I get this: ssed: -e expression #1, char 25: unknown option to `s'

Maybe these ported version aren't as good as the real thing? If you have any ideas, I'd greatly appreciate it. And thanks again for your help.

Code:

ruby -ne "print if sub!(/^(.{87})ISS/,'\1   ')" old >new