OK. Unfortunately, you can't count how many lines you have written into a file before you write those lines into the file. So using cntRec like you tried can only show you how many lines were written into previous files.

But, since we know how many lines we've read and how many lines are in the input file, we can calculate how many lines we are going to write into this file before we write the header record. So, remove the new action you added:

Code:

{       # count no. of lines
        ++cntRec
}

and just change the printf() statement you changed to something like:

Code:

	printf("HD-%d-%03d\n", (NR - 1 + lpf) <= lc ? lpf : lc % lpf, nf) > fn

If the current line number (which is the 1st line in an output file) - 1 + the maximum number of lines that we will write to a file is less than or equal to the the number of lines in the input file, print the maximum number of lines to write to a file; otherwise, print the number of lines left over (which will only happen on the last file and only then if there are less than lpf lines left to go into that file).

Jim,
The reason for the script is that csplit doesn't add the desired headers and trailers in the split files. And, yes, that could be done with after csplit did the big part of the job; but why read and write the data again to add a header when awk can do it in one pass.