I would point out you should never use $a or $b as ordinary variables. At best they are confusing both due to ambiguity in name and conflict in namespace and at worst you'll clobber a sort.
Given that what you want to do is collect results by your first field, I would suggest using a hash as your base structure. Since you want to keep more than one piece of data per key, a hash of hashes would be helpful (usually called a HoH, see perlreftut, perldsc and/or perllol). In that way, the original order is irrelevant. I would then sort the keys using a caseless alphabetical sort, as you did in Re^2: Sorting and Arranging data from input file, before outputting. Swapping your filehandles to indirect ones, using 3-argument open and testing your opens, your code might look something like:
#!/usr/bin/perl use strict; use warnings; use Data::Dumper; my $inFile = "input file"; my $outFile = "output file"; # By putting an "or die" clause after an open, the code will exit grac +efully # with an emitted error message if a file cannot be opened for the int +ended # operation open my $fin, "<", $inFile or die "Open fail on $inFile: $!"; open my $fout, ">", $outFile or die "Open fail on $outFile: $!"; my %results; print $fout "|CSED Form|OrderNumber|Date|Total Documents|Total Pages|\ +n"; # Process the input file 1 line at a time, storing the results in a ha +sh of # hashes in %results while (my $line = <$fin>) { # Implicit tests if $line is defined my @fields = split /\|/, $line; # Split $line on | my $key = lc($fields[1]); # Lowercase here so results get associat +ed # $results{$key} is a hash reference, containing information on al +l records # keyed on $key (m737, q569, m729...). After processing, the hash + reference # in $results{$key} will have data keyed on 'lines', 'sum1' and 's +um2' push @{$results{$key}{lines}}, $line; # $results{$key}{lines} is an array reference. As each line is en +countered, # it is added to the end of the array. This code uses autovivific +ation - # this means that since $results{$key}{lines} is undefined when pe +rl first # encounters it AND it is treated as an array reference when perl +first # encounters it, it will be set to an empty array reference. It i +s # roughly equivalent to being preceded by the line # $results{$key}{lines} = [] if not defined $results{$key}{lin +es}; $results{$key}{sum1} += $fields[-3]; # Third to last column # Accumulate the number in $fields[-3] in $results{$key}{sum1} $results{$key}{sum2} += $fields[-2]; # Second to last column # Accumulate the number in $fields[-2] in $results{$key}{sum2} } # Use a second loop to sort and output the accumulated results. It tr +ansits all # keys in alphabetical order since there is no sorting code block for my $key (sort keys %results) { # prints the unchanged lines to the file (print in list context) a +t $fout # In particular, note all lines still contain newline characters a +t the ends print $fout @{$results{$key}{lines}}; my $count = @{$results{$key}{lines}}; # List in scalar context yi +elds length # Print accumulated results, as stored in the variables $count, $k +ey, # $results{$key}{sum1}, $results{$key}{sum2} print $fout "There were $count ${key}s and total of 2nd to last ro +w was $results{$key}{sum1} and total of last row was $results{$key}{s +um2}\n"; } # Handles close automatically when they go out of scope # Output to terminal a dump of %results so operator can see internal s +tructure print Dumper \%results;
Terminal output:
$VAR1 = { 'q569' => { 'sum1' => 1989, 'sum2' => 18248, 'lines' => [ '|Q569|5555|Jun 05,2010|584|4562| ', '|Q569|5555|Jun 05,2010|345|4562| ', '|Q569|5555|Jun 05,2010|215|4562| ', '|Q569|5555|Jun 05,2010|845|4562| + ' ] }, 'm737' => { 'sum1' => 876, 'sum2' => 9124, 'lines' => [ '|M737|5555|Jun 05,2010|753|4562| ', '|M737|5555|Jun 05,2010|123|4562| ' ] }, 'm729' => { 'sum1' => 2780, 'sum2' => 22810, 'lines' => [ '|M729|5652|Jun 11,2010|198|4562| ', '|M729|5876|Jun 15,2010|298|4562| ', '|M729|5726|Jun 18,2010|428|4562| ', '|M729|5147|Jun 20,2010|918|4562| ', '|M729|5632|Jun 01,2010|938|4562| ' ] } };
input file:
|M737|5555|Jun 05,2010|753|4562| |M737|5555|Jun 05,2010|123|4562| |Q569|5555|Jun 05,2010|584|4562| |Q569|5555|Jun 05,2010|345|4562| |Q569|5555|Jun 05,2010|215|4562| |Q569|5555|Jun 05,2010|845|4562| |M729|5652|Jun 11,2010|198|4562| |M729|5876|Jun 15,2010|298|4562| |M729|5726|Jun 18,2010|428|4562| |M729|5147|Jun 20,2010|918|4562| |M729|5632|Jun 01,2010|938|4562|
output file:
|CSED Form|OrderNumber|Date|Total Documents|Total Pages| |M729|5652|Jun 11,2010|198|4562| |M729|5876|Jun 15,2010|298|4562| |M729|5726|Jun 18,2010|428|4562| |M729|5147|Jun 20,2010|918|4562| |M729|5632|Jun 01,2010|938|4562| There were 5 m729s and total of 2nd to last row was 2780 and total of +last row was 22810 |M737|5555|Jun 05,2010|753|4562| |M737|5555|Jun 05,2010|123|4562| There were 2 m737s and total of 2nd to last row was 876 and total of l +ast row was 9124 |Q569|5555|Jun 05,2010|584|4562| |Q569|5555|Jun 05,2010|345|4562| |Q569|5555|Jun 05,2010|215|4562| |Q569|5555|Jun 05,2010|845|4562| There were 4 q569s and total of 2nd to last row was 1989 and total of +last row was 18248
Update: At the OP's request via msg, I added significant in-line documentation for the above code as well as the trailing Dumper code.
In reply to Re^3: Sorting by value??
by kennethk
in thread Sorting by value??
by Homerhrdz
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |