Shell Programming and Scripting

I want to pass an array of variables to be inserted by awk in the 2nd column of a file.

Now, I want to pass an array having three different ages which need to replace the ages in the above table. Can I pass them to awk for it to replace the second column with new ages from an array?

Usually you'd pass big tables of data into awk as a file. You can use the NR==FNR trick to process only the first file, and let the rest of the code handle everything else.

Code:

awk 'NR==FNR { A[++L]=$0 ; next } { rest of stuff }' tablefile otherfiles

Thanks Corona. But it looks like I wasn't quite clear about what I wanted.

I have an array that I want to pass to awk as an argument, maybe to go in the -v variable. All I know is the variable used with -v is a scalar variable. If I wished to pass an array instead, would that be possible?

I could be using the wrong approach to my problem, wherein I want to replace the 2nd column of a file with new values contained in my array. But I guess that is unsafe as I'd rather do a key mapping as opposed to try and randomly fit in data from elsewhere. Thanks anyway!

If you really need entire huge blocks of data stored in arrays, perhaps the entire thing could be rewritten in awk. It's good at that sort of thing, and doesn't have limits except the main memory of the computer.

To answer your question: No. There's no generic way to shove an array from shell into awk, or anything else for that matter. You have to convert it back into a block of text, then split that block back apart on the other side manually. That's trivial to do with files, which awk is designed to process.

Forgive me, though, but I've seen a lot of new shell programmers shell shove absolutely everything into arrays because that's the only thing they know. If they don't have arrays, they simply shove everything into single variables instead. Then they post about what they managed to accomplish despite the difficulty, and people think that's how it's supposed to be done, and the habit spreads. What they're missing out on is easy redirection of files and streams -- which are the whole advantage of using a shell as opposed to, say, perl.

But doing so causes a whole new set of problems. How is that text stored, exactly? Line one as element one, line two as element two, and so forth? Or token 1 as the first element, token 2 as the second, and so forth? Do you need to worry about the size of a shell variable? -- that's quite small, on some systems! Will you need to feed awk its dimensions? Do you actually know its dimensions? Will you need to do all kinds of splitting specially instead of the stuff awk can do for you all by itself if used correctly? Might it have been simpler to not use an array at all?

Quite often, it really is simpler, to use things as they were intended.

Yeah you're right. I'm actually trying to complicate something by using an array which can otherwise be done using awk alone (if I could, in the correct way). But I intend to abandon that method as I also believe using the array couldn't land me with accurate data and I'd be beating around the bush to even get those array values before I "shove" them into my awk variable

So I'll explore the other way of using awk in the right way to get exactly what I want. Again, many thanks.

Maybe:

Code:

set -A arr aaa bbb ccc ddd
 
awk -v arr="${arr[*]}" 'NR>1 {$2=arr} 1' infile

Wow! Looks so workable. Now I'm tempted into using it (after almost pledging to Corona that i'll not go that way!)

Thanks RDRTX1!