Shell Programming and Scripting

hi i have a file content like

my_file:
1US8738297897918[space][space][space]0[space][space]0[space][space]0
[1US8738297897918][space][space][space]0[space][space]0[space][space]0
[1US8738297897918][space][space][space]0[space][space]0[space][space]0

i used
while IFS= read line
do
echo "$line" | grep '^\['
if [ "$?" -eq 0 ]
then
last=`echo "$line" | cut -c 3-20`
echo "$last"
fi
done < $my_file


but it is writing blank spaces into the o/p fie

please help

thanks
Satya

First off, i'd like to state for the record that i usually skip threads with "urgent" in their title out of habit. If you have an urgent demand for work you can't do on your own please consider hiring an expert.

Ok, having said this, it would also be a nice gesture to enclose code in "code"-tags. This is what they are for and they are here for a reason - perhaps for the same reason the telephone directory is printed in a tabular, formatted way instead of a stream of names and numbers: READABILITY.

Now back to the problem. I took the liberty to reformat your script a little:

The first thing i notice is a useless use of "grep". Your goal is to limit your operation to all lines starting with a "[" character, yes? (If this is not the case your script is just wrong.) This could also be done in a sed (awk, ...) script:

Code:

sed -n '/^\[/ {
               some command
               some other command
               ....
               p
             }' /some/file > /some/other/file

A second thing is the - quite antiquated - use of backticks. Do NOT use them any more! We are in the 21st century now, we do not use morse code (we have picture telephones now), horse-pulled carts (we have automobiles) and we do not use backticks for similar reasons - we have the "$(....)" construct, which can be nested and even enclosed in normal double quotes.

(As an aside: some hold that backticks are backwards compatible with older shells. True. I hope these are writing their scripts in cuneiform writing on clay tablets to be "backwards compatible" with sumeric texts. Antiquity has ended - get over it.)

In this case not even the "$(....)", which opens a subshell, is necessary. There are enough powerful shell textprocessing commands to cut character 3-20 out of a string:

Code:

a="123456789012345678901234567890"
b="${a##??}"                  # chop off the first two characters, leaving char 3-
c="${b%%??????????????????}"  # chop off everything but the first 18 chars, leaving char 3-20 from the original string

You might want to find out what a big impact on execution time this has by reading this thread.

But in your case even this is not necessary because it can be done inside the sed script we already use to filter for lines:

Code:

sed -n '/^\[/ {
               s/^..\(.\{18\}\).*$/\1/p
             }' /some/file > /some/other/file

which does the same as the shell construct: replace the complete line with characters 3-20 and put it to <stdout>. This is what is left of your script, it will do exactly the same as the shell script you presented here: it takes every line which starts with "[" and extracts

If doesn't do what you want it to do you might tell us what your goals are and we might find out together.

I hope this helps.

bakunin