topic Re: EOF within awk? in Operating System - Linux

EOF within awk?

Tim Howell — Tue, 22 Aug 2006 11:36:16 GMT

If the following was invoked...
awk -f script.awk `ls -t /tmp/logdir/log.*`

Is there a way within script.awk to know when you have reached the end of an individual file? (or know when all files have been processed?) script.awk matches patterns and processes an indeterminant number of records in each file. I want to totalize the value of these records.
TIA

Re: EOF within awk?

Steven E. Protter — Tue, 22 Aug 2006 11:41:45 GMT

Shalom,

how about inbedding awk in a read loop

while not EOF
do
nice awk stuff
done < filename

SEP

Re: EOF within awk?

A. Clay Stephenson — Tue, 22 Aug 2006 11:45:25 GMT

Yes awk has BEGIN and END blocks. When the normal block is finished (or if an explicit exit is done) the END block is executed.
This example should total the sizes of the files using an ls -l:

my.awk
-------------------------
BEGIN {
tot = 0
}
{
tot += ($5 + 0)
}
END {
print "Total: ",tot
}
--------------------------------------

Invoke like this:
ls -l | awk -f my.awk

Note the ($5 + 0). That is intential as it forces the 5th field (the file length in ls -l) into a numeric context. The standard idiom to force a field to string context is to append a null string.

Re: EOF within awk?

Tim Howell — Tue, 22 Aug 2006 12:14:58 GMT

Thanks,
the BEGIN & END were what I needed - the proper format and syntax was what I was lacking...your example answered my question

Re: EOF within awk?

Tim Howell — Tue, 22 Aug 2006 12:16:44 GMT

The END block is executed when the last record in the last file is processed apparently, and that's what I needed.

Re: EOF within awk?

A. Clay Stephenson — Tue, 22 Aug 2006 12:18:45 GMT

Note that the BEGIN block is not strictly needed as the variables are initialized as null strings -- which would evaluate in numeric context as zero but I don't never trust nobody for nothing so I always initialize values like "tot" in this example.

Re: EOF within awk?

Mike Stroyan — Tue, 22 Aug 2006 12:23:09 GMT

You can track individual files using FILENAME. It will be updated line by line. This example accumulates values in an array indexed by FILENAME.

awk '{c[FILENAME]+=$2}END {for (f in c) {print f,c[f]}}' *

Re: EOF within awk?

James R. Ferguson — Tue, 22 Aug 2006 12:37:38 GMT

Hi Tim:

How about(?):

# cat .nbrlines
#!/usr/bin/perl
while (<>) {
if (eof) {
printf "%8d %s\n", $., $ARGV;
close ARGV if eof;
}
}

...now:

# ./nbrlines /etc/hosts /etc/services
21 /etc/hosts
187 /etc/services

Regards!

...JRF...