Shell Programming and Scripting

Hi all,

I have to generate a tab-delimited flat file. Since few days, I have been getting * in random accounts.

For example, an account supposed to have the value 123456789123,123

Now this is supposed to be in a 12,3 format.

Please note that this being a German field, the comma (,) here means a decimal (.) point.

For past few days, I have been getting 123456789123,12*
This occurs ONLY in one field.

I need a script that will help me remove this asterix from the affected accounts, ONLY in that field. If there are asterix in other fields, I dont want to remove them.

Can anyone help me?

Hi,

You can use awk to replace the "*" only in the corresponding field (the 3rd one in the example bellow).

Example:

(I am assuming your input/output files will be TAB separated)

input:
a 123 123456789123,123 HA***
b 124 123456123456,12* HB123

Code:

awk 'BEGIN {FS="\t"; OFS="\t"} {gsub("\*","",$3); print $0}' input_file > output_file

output:
a 123 123456789123,123 HA***
b 124 123456123456,12 HB123

So, only the "*" characters in the 3rd field are removed. In the 4th field "*" are not removed, as desired.

Let me know if it worked.

Try

Code:

sed "s/\*//g" < input_file > output_file

Better yet, do NOT try that, because it will remove *any* asterisks instead of only the ones you want to remove.

First, you have to determine where in line the field is, where you want to remove the asterisk. As an example I will suppose the field is on column 20-30, if you have a delimited format instead of a fixed length format you will have to change accordingly.

To phrase the substitute-expression to only remove asterisks in the columns 5-15 you will have to:

cut out columns 5-15 from the line
run the substitution on it
put the changed part of the line back in

This could be done by sed alone, but would be rather complicated, involving the use of the hold space and in the end the whole script would be pretty counterintuitive (I'll provide the solution upon reqquest if anybody is interested, but it is nothing for the ones with poor nerves ;-) ). An easy way to do it is in a shell script though:

Code:

#!/bin/ksh

typeset line=""
typeset begin=""
typeset end=""
typeset middle=""

exec 3> file.changed

cat file | while read line ; do
     # chop off the first 4 chars from $line, storing them in $begin
     begin="${line%%${line##????}}" ; line="${line##${begin}}"
     # chop off the next 10 chars from $line, storing them in $middle
     # and put the remainder in $end
     middle="${line%%${line##??????????}}" ; end="${line##${middle}}"

     # here we change the cut out field
     middle="$( print - $middle | sed 's/\*//g')"

     # put everything together

     print -u3 "${begin}${middle}${end}"
done

exec 3>&-
exit 0

bakunin