AWK script error

Solox · 05-28-2013, 12:21 AM

Hi all,

I am not involved in scripting as the scripts we use have already been written. However, we (my team of 4) are in the midst of migrating from AIX Server platform to a LINUX server platform in which we are responsible for migrating "our" portion.

Through our curent testing, we have discovered several AIX Unix AWK scripts that have compatibility issues with Linux. Most of them we have researched and fixed them. BUT...this one is driving me crazy.

Below is the script:

Code:

BEGIN {
	good_count = 0
	bad_count = 0
	line_count = 0
}

NR == 1 {
	rec_count = substr($0, 15, 8)
        date_str = substr($0, 32, 10) 
		
	print "Date in Header Record : " date_str
}

NR > 1 {
    line_count++
  
    if (length($0) == 710)
    {
        print $0 > GoodFile
        good_count++
    }
    else
    {
        print $0 > BadFile
        bad_count++
    }
}

END {
	if ( line_count == rec_count && bad_count==0 )
	then
    {
   	 print "********************************************"
         print "Count in header        :  " rec_count  
	 print "Count in file received :  " line_count 
	 print "  "
	 print "Number of Good Records :  " good_count
	 print "Number of Bad  Records :  " bad_count  	
    }
    else
    {
	  print "********************************************"
          print "                             PROBLEM WITH FILE SENT FROM SOURCE SYSTEM " 
          print "  "
          print "Count in header        :  "  rec_count  
	  print "Count in file received :  "  line_count 
	  print "  "
          print "Number of Good Records :  " good_count
	  print "Number of Bad  Records :  " bad_count

	  exit 99
    }
}

The error I get when I run it:

Code:

dnt_UIS_header_process.awk: line 1: BEGIN: command not found
dnt_UIS_header_process.awk: line 8: syntax error near unexpected token `('
dnt_UIS_header_process.awk: line 8: `   rec_count = substr ($0, 15, 8)'

I'm at a loss. If somebody can help me uncover WHY is spitting out these errors in Linux while the scripts works in AIX...I would be eternally grateful.

Thank you.

pan64 · 05-28-2013, 01:19 AM

probably there is a "strange" - non-printable char at the beginning of the script?
you can try od -xc to check it.

Ygrex · 05-28-2013, 01:25 AM

how do you run this? at first, «then» is not a part of «if» statement in awk, but it seems like the file is executed by shell, not awk

konsolebox · 05-28-2013, 03:51 AM

I think you simply missed the header.

Code:

#!/usr/bin/env gawk -f

murugesan · 05-28-2013, 10:28 AM

As informed by Ygrex:

Code:

Replace:
           if ( line_count == rec_count && bad_count==0 )
	   then
	   {
With:
 	   if ( line_count == rec_count && bad_count==0 )
	   {

As informed by konsolebox:

Code:

Replace:
           BEGIN {
With:
           /bin/awk 'BEGIN {

Also at the end

Code:

Replace:
	     exit 99
       }
   }
With:
	     exit 99
       }
   }' Any.File.Name.text

David the H. · 05-28-2013, 11:20 AM

Quote:

Originally Posted by murugesan

As informed by konsolebox:

Code:

Replace:
    /bin/awk 'BEGIN {

Code:

    }' Any.File.Name.text

This is incorrect. Your "corrected" version is the shell syntax for launching an awk command. For a standalone script you use #!/bin/awk -f at the top of the script, as konsolebox said, and then go straight into the code without extra quoting.

The same goes for the ending. You don't put the input file directly in the script itself (that's shell syntax again), but use it as an argument when executing the script, or feed it through stdin.

murugesan · 05-28-2013, 11:43 AM

Hi,

Kindly look at the posted comments, Replace With for "/bin/awk 'BEGIN {"
>You don't put the input file directly in the script itself
>You don't put the input file directly in the script itself
I feel not to agree with this step.
It depends upon the test also with the type of the document being submitted(Example SAP related document with multi-language document)
based on which the script is being executed.

Yes, agreed with

Code:

#!/bin/awk -f

but based on the writer of the tester to handle the exceptions

I always used to write scripts like on HP-UX OpenSource

Code:

#!/bin/ksh

David the H. · 05-28-2013, 12:24 PM

Whether you agree or not, it's still wrong. A stand-alone awk script contains only the code itself, and the shebang line with the interpreter at the top. The input comes from the outside (as an argument or stdin), just as with any other script or command.

You could also put just the code itself in the file, with no shebang, and run it in a shell with "awk -f codefile inputfile". It works the same way.

What you posted is the syntax for a shell wrapper script with an awk command inside it. That will work too, with the proper shell shebang, but it's not an awk script, and it adds an unnecessary second execution layer.

What will not work is using an awk shebang, and then also putting the awk command name in the file after it. That would be telling awk to try to launch itself. Not to mention other syntax errors occurring from improper quoting.

Edit: Here's the (rather brief) gnu awk documentation page on creating stand-alone scripts:

http://www.gnu.org/software/gawk/man...e-Scripts.html