filehandle in warning not what I expected...

[spstansbury]

Greetings...

I have a script that parses a csv file, looks up some data in another file, and generates a report. Pertinent code is here:

my $report_files_dir = "../Working/New_data";

my $filename_pattern = "Remedy.csv";

$interface_file = "nodes.csv";

open ( my $interface_data, "<", "$report_files_dir/$interface_file" ) 
+or die $!;

opendir ( my $report_files, $report_files_dir) || die "can't opendir $
+report_files_dir: $!";

for ( sort readdir $report_files ) {
    next unless $_ =~ /$filename_pattern/;
    
    open ( my $remedy_file, "<", "$report_files_dir/$_" ) or die $!;

    while (<$remedy_file>) {
        chomp;
        next if ($. == 1);

        # Parse info from the current line
        # Use a returned value ($first_node) to look up data in anothe
+r file:
        
        LOOKUP: while (<$interface_data>) {
            chomp;
            my @list = split/\|/;
            my $ref_id = &clean( $list[2] );
                        
            if ( $ref_id =~ /$first_node/ ) {
            
                # The CMTS name is in Column 0
                $cmts_name = $list[0];
                
                # The CMTS Interface is in Column 1
                $cmts_interface = $list[1];
                
                last LOOKUP;
            }
        }
        seek ( $interface_data, 0, 0 );

        # Finish doing the report generation...

    }
    close ( $remedy_file );
}

closedir ( $report_files );
close ( $benchmark_data );
close ( $interface_data );
close ( $progress_file );

But I get the following warnings:

Use of uninitialized value in subtraction (-) at ./load_Remedy.pl line
+ 503, <$interface_data> line 1894801.
Use of uninitialized value in subtraction (-) at ./load_Remedy.pl line
+ 503, <$interface_data> line 5825736.
Use of uninitialized value in subtraction (-) at ./load_Remedy.pl line
+ 503, <$interface_data> line 6546698.

The issue is not the uninitialized value ( I'm comparing two dates in line 503, and I may indeed be missing a date value in the original data, but the fact the the data file is $remedy_file, not $interface_data, and the line values are nonsensical, none of the files has more that 100K lines. Am I not using seek properly?

[Corion]

I think that you're throwing off Perl's line number counters by using seek.

But if your files are not "that big", why do you search through the $interface_data every time on disk instead of reading it once into a hash?

[Anonymous Monk]

Not related to your question, but when you use $filename_pattern = "Remedy.csv" and then /$filename_pattern/, that period character is used as the regex special character (like using [^\n]), which may match more than you intend (eg, it matches "Remedy_csv"). If you want to match strings in which that string occurs literally, quote it (like /\Q$filename_pattern\E/ or with quotemeta). However, I suspect you really want equality: $_ eq $filename_pattern;, or maybe case-insensitive equality (eg, for Windows): lc $_ eq lc $filename_pattern.

[JavaFan]

Perl only resets the line counter when opening the file. For each line in $remedy_file, you read many lines of $interface_data. It seems very inefficient to me, but that's a different issue. If, instead of seeking, you would just (re)open the file, you'll see more sensible numbers.

[spstansbury]

Thanks for the input - you're right, it's inefficient, I will follow Corian's advice and read the file into a hash.

But what really confused me is that the warning referenced the interface file and not the actual data file that the undef field was parsed from...

[spstansbury]

Or it's simply showing the last filehandle that I read from...

[ysth]

Perl is just reporting the last file you read from. If you prefer to see the $remedy_file line number in the warning, you can localize perl's recollection of the last file read by putting:

local $.;

just before LOOKUP:. This is a special feature of the $. variable.