Shell Programming and Scripting

I would just use a slight simplification of RudiC's suggestions:

Code:

file="/path/to/desired/file"
awk -F'|' '{$4=$4}1' OFS='|' "$file" > "$file.$$" && cp "$file.$$" "$file"
rm -f "$file.$$"

If you insist on using sed we can show you how to do it, but you'll need to explain why you need to use sed when awk provides a much simpler solution. And, if you need to use an editor instead of awk, ed or ex would provide a more portable way to overwrite an existing file than trying to use the (sometimes dangerous) -i option that is available on some versions of sed (especially since you haven't told us what operating system you're using).

even more simplification:

Code:

awk -F"|" NF=4 OFS="|" infile

Yes, it is simple; it just isn't portable. With a BSD based awk and the submitter's sample input file, the output produced is a copy of the input:

Code:

awk -F'|' NF=4 OFS='|' file
040|14300|40.0|563000
042|13200000|40.0
041|100|40.0
043|10000|40.0
045|102|40.0

And, even if you expand that to actually specify a program to be run (as required by the standards):

Code:

awk -F'|' 1 NF=4 OFS='|' file
040|14300|40.0|563000
042|13200000|40.0
041|100|40.0
043|10000|40.0
045|102|40.0

There is nothing in your script that tells awk to make any change to the input when it is copied to the output. Some implementations may do what you suggest, but the standards do not specify that behavior.