Shell Programming and Scripting

Spelling things wrongly usually is fatal in IT unless specially taken care of in the code. So instead of softening the conditions for your replacement, you might want to install error checking upfront.
Howsoever, for your problem, try

Code:

sed -r "\$s/(([^,],){2}).*/\1$TUNA/g" file
Line 1
Line 2
Line 1
ABC,2,TUNA,PIE

or, if your sed version doesn't allow for extended regexes,

Code:

sed  "\$s/\(\([^,],\)\{2\}\).*/\1$TUNA/g" file

Unfortunately, when using double quotes, you need to escape the $ sign designating the last line of the input stream.

Hi RudiC,

I used you first sample code and it seems to only work with files with fewer lines, i have a file which consists of 40+ lines and it doesn't seem to work. Is there a way to resolve this?

Also would it matter if i have a file that consists of multiple Commasas a delimiter like below? Would it potentially affect the way the script works?

Code:

"FOOD",CAT,20160606,20160606,001
"FOOD1",TUNA,"PASTA"
"FOOD2",CHICKEN,"PASTA"
"FOOD3",HAM,"PASTA"
"FOOD4",EGG,"PASTA"
"FOOD5",NOODLE,"PASTA"
"FOOD6",BACON,"PASTA"
"FOOD7",FISH,"PASTA"
"FOOD8",BIG,"PASTA"
"TOTAL",8,FOOD,ABC1234567

Also apologies in advance if i keep changing the way the code in the file looks it's that i'm trying to keep it as generic as possible.

Generic is one thing. But, without knowing how it doesn't work and without knowing what the last line really looks like in a file where it doesn't work, we are all left guessing with no solid idea of what might be wrong.

If I were to make a bunch of wild guesses, might I be correct in saying that the sed command RudiC suggested in post #8:

Code:

sed  "\$s/\(\([^,],\)\{2\}\).*/\1$TUNA/g" file

works just fine as long as there is exactly one digit in the second field in the last line in file (as found in all of your sample input files), but it doesn't do anything if there is more than one digit in that field? If my wild guess is correct, would the following slight modification of RudiC's code work correctly for you?

Code:

sed "\$s/^\(\([^,]*,\)\{2\}\).*/\1$TUNA/" file

We don't necessarily need the exact contents of the file(s) you're working on to help you get working code, but we do need representative contents that accurately show us the format of the data you're going to be processing with a clear English description of the variations that would be expected in your real data.

As the sed script is going for the last line in the input stream, it doesn't matter if it's 40 lines or thousands of lines of data.
You may want to anchor the regex (after applying Don Cragun's "patch") at the begin-of-line to be even safer to retain the FIRST two fields.

What do you mean by "multiple Commas as a delimiter like below"? I can't recognize any in your sample...