Shell Programming and Scripting

Hi,

I'm looking for a way for me to load a set of variables (threshold values) from a file using AWK so that when my AWK script processes a file, I can use the threshold values (loaded into an AWK variable) to compare against the ones in the file being processed. If the threshold exceeds or something like that, print it into a file etc... The reason why I want to have have AWK load the threshold values from a file is so that these values can be changed on the fly. I know I can hard code the values but it's not practical.

Here's an example of my config file:

Code:

var1=100
var2=1000
var3=10000
var4=10

I haven't written the script yet because I need to figure out this first... So, the script should do something like this :

Code:

awk '
BEGIN{
 <load the config file here>
} <do the comparison here, etc>' file

Is this possible at all using AWK? If not, I will have to use shell scripting which would probably be slower consider the file I am going to process is going to be big. Any pointers would be very much appreciated.

Thanks.

if you only have these four variables, then source your file

Code:

 
. variables.txt
 
awk -v var1="$var1" -v var2="$var2" -v var3="$var3" -v var4="$var4" '{do your process and access the vairable as var1, var2, var3, var4}'

Well, I actually have more than 4. Right now, I have 27 of them and the list may just increase later on. The idea is to make it as dynamic as possible so that no changes to the script would be required in the future.

Awk doesn't have an eval so you can't just set arbitrary variable names in BEGIN.

What's your system? If you have GNU awk, you can just add 'BEGIN {' and '}' to your configfile, then do this:

Code:

awk -f configfile.awk -f mainscript.awk ...

It's perfectly valid to have more than one BEGIN { } section, by the way, so this won't interfere with mainscript. The advantage of this is you get full awk syntax in your config file, including quotes and escapes behaving as expected, comment lines with #, and such.

Or you could cram them all onto the commandline programmatically:

Code:

set --

# Rarely-appreciated property of xargs:  it can understand quotes.
# We use it to convert var="asdf" into var=asdf
xargs -n 1 < configfile > /tmp/$$

while read LINE
do
        set -- "$@" -v "$LINE"
done < /tmp/$$

rm -f /tmp/$$

awk "$@" -f script.awk ...

Like this may be:

Code:

awk 'BEGIN{FS="=";while(getline < "configfile") a[$1]=$2} <use variables as a["var1"],a["var2"],etc.>'

Thanks Corona688. I didn't think of using the -f option I'll try this out.

I would also try out the suggestion posted by elixir and see which works out better for me.

I don't see anything GNU AWK specific there. Perhaps some obsolete implementations do not support multiple -f options, but it's common functionality with modern implementations, e.g., gawk, nawk, mawk, busybox. POSIX has required it for at least 15 years now.

Regards,
Alister