Hi, I am having a file which has 9 million records. Few records in between are having no value for a field resulting in two consecutive pipes(||). Because of this, the data load is getting rejected. I want to replace these consecutive pipes(||) with a space between them(| |). Please provide a solution for this. Ex : abc|123|aa||bcd|23^M I want the above record as : abc|123|aa| |bcd|23^M Thanks, Anil

echo "abc|123|aa||bcd|23^M" | sed -e "s/||/| |/g" sed -e "s/||/| |/g" INFILE > OUTFILE

Perhaps something like: sed 's/||/| |/g' input.file > output.file

Jerry and David's sed solution works fine, except where you have two sequential null columns: abc|123|aa|||bcd|23 My solution is this: sed -e 's/|/|\&/g' -e 's/|\&|/| | /g' -e 's/\&//g' d2.file 1) replace each pipe symbol with a pipe symbol and a character not used in the data set. I chose & and escaped it cause sed uses it. 2) replace |\&| with a space 3) then delete all the unnneeded & haven't thoroughly tested, but seems to work.

That's a good catch, nails. But your code will insert an extra blank sometimes, such as before the bcd in this example: in:   abc|123|aa||||bcd|23 out: abc|123|aa| | | | bcd|23 Following code will keep reprocessing a line as long as changes are being made: sed -e :a -e 's/||/| |/g' -e ta

Jim: That's a good catch of my catch. I really feel dumb not thinking of using a loop. Nails

It's hard to get one up on the nails!