Shell Programming and Scripting

That's incorrect. In a POSIX ERE, ^ is only an anchor when it is the first character in an expression.

\|^ should be portable, but, sadly, some implementations botch ^ handling. Some (perhaps all?) mawk implementations definitely do not handle this correctly.

Regards,
Alister

No. What you say is true for a BRE; not for an ERE. Quoting from IEEE Std. 1003.1-2008, P190, L6174-6183:

I stand corrected. Thank you.

Also, apologies to mawk devs

Regards,
Alister

Hi,

Thanks for your response..

Now my file is having around 300 columns and awk is giving an error message "awk cannot have more than 199 fields" while summing up the columns 80 and 138.

Please suggest

Thanks, Jram

What system are you using?

I see nothing in the POSIX standard that would allow this limitation.
Input files for awk are text files. Text files can't contain null bytes. If there are any characters in a text file, the last character in the file must be a <newline> character and no line in a text file (including its trailing <newline> character can be more than {LINE_MAX} bytes long. The minimum value for {LINE_MAX} on conforming implementations is 2,048, so in theory, I believe the standard requires awk implementations to support at least 2,049 fields (more if the implementation's value of {LINE_MAX} is larger than the minimum).

Assuming your input line are not longer than {LINE_MAX} bytes long, the following should work around your awk's limit on the number of fields:

Code:

awk -F"\n" '# Call getfield(N) to extract field N from $0 with "|^"
# as the field separator.  f1toN is a local temp variable.
function getfield(n,f1toN) {
        if(match($0,"(\([^^]*\)[.|.][.^.]){"n"}") > 0) {
                f1toN=substr($0,RSTART,RLENGTH-2)
                match(f1toN,"(\([^^]*\)[.|.][.^.]){1,"n-1"}")
                return substr(f1toN,RLENGTH+1)
        }
        return ""
}
        {sum+=(getfield(80)+getfield(138))}
END     {print "Sum of 2 columns 80 & 138 which is",sum
}' in

Setting FS to <newline> means each input line will be treated as a single field. The getfield(n) function will extract field n from that line (assuming the field delimiter is still "|^") by storing the 1st n fields in f1toN and then extracting just the nth field from the end of that. (This could be done with a single call to substr() by calculating the start and end points of the desired field using the RSTART and RLENGTH from the two calls to match(). But I'll leave that as an exercise for the reader.) Note that the "n" and "n-1", respectively, in the two calls to match() are not inside a "..." string!