Shell Programming and Scripting

It does not work. I mean it does not change the character as requried

Code:

ISA*00* *00* *08*92 *08*123*030627*1304*U*00*00*0*P*>~ has nothing valid to change

It works fine when I try it. Please show us the output from the command:

Code:

head -n 1 hugefile|od -c

The sample you have shown us has the character to be changed as the 51st character on the 1st line of hugefile. Your response to my earlier questions said it was the 103rd character on the line. (Of course you also said that your 3GB file has a total length of 106 bytes, so we don't know what to believe.)

Something doesn't add up here, and your refusal to use CODE tags when displaying your sample input and output may be a contributing factor.

I have another script that might work, but it doesn't make any sense posting it if I can't get a sample to use to test it that bears some resemblance to the first few lines of your huge file.

Thanks Don...In my sample file it 51 character and in my original data it 103 character.
Bottom line, it has to be changed from P to T or from T to P. Also as I said file size in Giga bytes. could you please post your other script ? I will give a try.
I will post the current script in my next post

I don't write a lot of interactive awk scripts, but this seems to do what you want (as long as your version of awk supports the nextfile command. If awk on your system doesn't support nextfile and the character you want to change is in a fixed position, Corona688's suggestion (modified to loop across a list of files and to use whatever character position is in your real data (instead of 51 that was in your sample data)) would be a better approach.

The script below uses field 16 (which you refer to as ISA segment 15), no matter what character number that is within a line. If it finds a "P" or a "T", it tells you what it found and asks if you want to change it, otherwise it tells you what it found but does not change anything. In either case, it then moves on to the next file until all files you provided as operands are processed:

Code:

#!/bin/ksh
IAm=${0##*/}
if [ $# -lt 1 ]
then	printf 'Usage: %s pathname...\n' "$IAm"
	exit 1
fi
awk -F'*' '
BEGIN {	OFS = "*" }
{	old = $16
	new = (old == "P") ? "T" : (old == "T") ? "P" : old
	if(old != new) {
		printf("Change \"%s\" to \"%s\" in %s: (Y/N) ", old, new,
			FILENAME)
		getline resp < "/dev/tty"
		if(toupper(resp) ~ /^Y/) {
			$16 = new
			cmd = sprintf("dd of=\"%s\" conv=notrunc", FILENAME)
			printf("Will run commmand: %s\n", cmd)
			print | cmd
			close(cmd)
			# Clear resp here.  If user enters EOF on next round,
			# we do not want this response to be interpreted as a
			# blanket approval for future files...
			resp = ""
		} else	new = old
	}
	if(old == new) printf("\"%s\" in %s not changed.\n", old, FILENAME)
	nextfile
}' "$@"

It was written and tested using the Korn shell, but will work with any shell that handles POSIX-required variable expansions (such as bash).

If you are trying this on a Solaris system, I don't think it will work. I don't think /usr/xpg4/bin/awk, /usr/xpg6/bin/awk, or nawk support nextfile; and /bin/awk or /usr/bin/awk on Solaris systems can't handle this script.