UNIX for Dummies Questions & Answers

Hi,
I need to replace the record count which is in the last line of the text file to a new record count.

The file is a fixed length file, and whole file is list of numbers and the last line looks like this.

9#EOF# 00000000000000000016 00000 0000000000

The character between 12 to 30 represents the record count excluding the header and footer. I need to change the count in the last line.

ex 00000000000000000016 in the above needs to be replaced by 00000000000000000036
The characters "9#EOF# " is fixed.

Please suggest. Perl or sed or awk is fine.
Thanks.

If I understood correct:

Code:

$> cat infile
header
eins
zwei
drei
9#EOF# 00000000000000000016 00000 0000000000
$> awk '{l[x++]=$0} END{for(z=0; z<x-1; z++){print l[z]}; $2=sprintf("%020d", NR-2); print}' infile
header
eins
zwei
drei
9#EOF# 00000000000000000003 00000 0000000000

Thanks for the above reply, would you pls explain me this logic. I missed mentioning that the input file contains only numbers and the file would be having 10 to 15 millions of records. So traversing like this will it be fine.

Sure:

{l[x++]=$0} _________________ Store every line ($0) into the array "l" (small L) where x is it's element's index and increment it every line by 1 (++).
END ________________________ When all lines are processed, we are here, so do the following:
for(z=0; z<x-1; z++) __________ A for-loop counting up from zero to the number of lines which is equal to x, but subract 1, so we pass the last element, since this will be treated now. Automatically in the END{} part, $0 is equal to the last line of those been processed.
{print l[z]} __________________ First we print out all elements but the last of the array, so you'll see all lines but the last.
$2=sprintf("%020d", NR-2) _____ Currently $2 of $0 ie. field 2 of the current (last) line has the value we want to change. So we alter field 2 by printing out the number of lines -2 (lines in awk are called Records and N stands for number, so NR means Number of Records) because we want to subtract the header and the footer line.
print _______________________ After modifying field 2 ($2) we just print out the whole (new) line.


If the input lines contain only numbers and sowith is completely differently structured than the example you gave, then this code will most probably not work.
You should post a realistic example then of course.

Hi Zaxxon,
Thanks for the explanation.
the input file looks like this.



I am little confused with the explanation:
"$2=sprintf("%020d", NR-2) _____ Currently $2 of $0 ie. field 2 of the current (last) line has the value we want to change. So we alter field 2 by printing out the number of lines -2 (lines in awk are called Records and N stands for number, so NR means Number of Records) because we want to subtract the header and the footer line"

I want to modify the last line itself.. so is it going to do the last line or last but one ?

thanks in advance.

As in the new excerpt $2 has 19 characters instead of 20, try this:

Code:

awk '{l[x++]=$0} END{for(z=0; z<x-1; z++){print l[z]}; $2=sprintf("%019d", NR-2); print}' infile

Since $2 is still the field which has or should have the number of lines minus header and footer in it, it will work with just this change. The now missing $3 and $4 of the former example is no problem.

It only modifies the last line, ie. field 2 ($2) of the last line and then prints out the modified line.

Should memory become an issue at some point, here's a variation of zaxxon's most recent solution that instead of storing all records in memory only stores one:

Code:

awk 'NR>1 {print prev} {prev=$0} END {$2=sprintf("%019d", NR-2); print}' infile

Regards,
Alister