Question

Using any combination of Linux tools (without going into any full featured programming language) how can I sort this list

A,C 1
C,B 2
B,A 3

into

A,B 3
A,C 1
B,C 2

Answer 1

In case somebody is interested. I was not realy satisfied with any suggestions. Probably because I hoped for view lines solution and such doesn’t exist as far as I know.
Anyway I did wrote an utility, called ljoin (for left join like in databases) which does exactly what I was asking for (of course :D)

#!/usr/bin/perl
=head1 NAME

ljoin.pl - Utility to left join files by specified key column(s)

=head1 SYNOPSIS

ljoin.pl [OPTIONS] <INFILE1>..<INFILEN> <OUTFILE>

To successfully join rows one must suply at least one input file and exactly one output file. Input files can be real file names or a patern, like [ABC].txt or *.in etc.


=head1 DESCRIPTION

This utility merges multiple file into one using specified column as a key

=head2 OPTIONS

=item --field-separator=<separator>, -fs <separator>

Specifies what string should be used to separate columns in plain file. Default value for this option is tab symbol.

=item --no-sort-fields, -no-sf

Do not sort columns when creating a key for merging files

=item --complex-key-separator=<separator>, -ks <separator>

Specifies what string should be used to separate multiple values in multikey column. For example "A B" in one file can be presented as "B A" meaning that this application should somehow understand that this is the same key. Default value for this option is space symbol.

=item --no-sort-complex-keys, -no-sk

Do not sort complex column values when creating a key for merging files

=item --include-primary-field, -i

Specifies whether key which is used to find matching lines in multiple files should be included in the output file. First column in output file will be the key in any case, but in case of complex column the value of first column will be sorted. Default value for this option is false.

=item --primary-field-index=<index>, -f <index>

Specifies index of the column which should be used for matching lines.  You can use multiple instances of this option to specify a multi-column key made of more than one column like this "-f 0 -f 1"

=item --help, -?

Get help and documentation

=cut


use strict;
use warnings;
use Getopt::Long;
use Pod::Usage;

my $fieldSeparator = "\t";
my $complexKeySeparator = " ";
my $includePrimaryField = 0;
my $containsTitles = 0;
my $sortFields = 1;
my $sortComplexKeys = 1;
my @primaryFieldIndexes;

GetOptions(
    "field-separator|fs=s" => \$fieldSeparator,
    "sort-fields|sf!" => \$sortFields,
    "complex-key-separator|ks=s" => \$complexKeySeparator,
    "sort-complex-keys|sk!" => \$sortComplexKeys,
    "contains-titles|t!" => \$containsTitles,
    "include-primary-field|i!" => \$includePrimaryField,
    "primary-field-index|f=i@" => \@primaryFieldIndexes,
    "help|?!" => sub { pod2usage(0) }
) or pod2usage(2);

pod2usage(0) if $#ARGV < 1;

push @primaryFieldIndexes, 0 if $#primaryFieldIndexes < 0;

my %primaryFieldIndexesHash;
for(my $i = 0; $i <= $#primaryFieldIndexes; $i++)
{
    $primaryFieldIndexesHash{$i} = 1;
}

print "fieldSeparator = $fieldSeparator\n";
print "complexKeySeparator = $complexKeySeparator \n";
print "includePrimaryField = $includePrimaryField\n";
print "containsTitles = $containsTitles\n";
print "primaryFieldIndexes = @primaryFieldIndexes\n";
print "sortFields = $sortFields\n";
print "sortComplexKeys = $sortComplexKeys\n";

my $fieldsCount = 0;
my %keys_hash = ();
my %files = ();
my %titles = ();


# Read columns into a memory
foreach my $argnum (0 .. ($#ARGV - 1)) 
{
    # Find files with specified pattern
    my $filePattern = $ARGV[$argnum];
    my @matchedFiles = < $filePattern >;
    foreach my $inputPath (@matchedFiles) 
    {
        open INPUT_FILE, $inputPath or die $!;

        my %lines;
        my $lineNumber = -1;
        while (my $line = <INPUT_FILE>) 
        {
            next if $containsTitles && $lineNumber == 0;

            # Don't use chomp line. It doesn't handle unix input files on windows and vice versa
            $line =~ s/[\r\n]+$//g;

            # Skip lines that don't have columns
            next if $line !~ m/($fieldSeparator)/;

            # Split fields and count them (store maximum number of columns in files for later use)
            my @fields = split($fieldSeparator, $line);
            $fieldsCount = $#fields+1 if $#fields+1 > $fieldsCount;

            # Sort complex key
            my @multipleKey;
            for(my $i = 0; $i <= $#primaryFieldIndexes; $i++)
            {
                my @complexKey = split ($complexKeySeparator, $fields[$primaryFieldIndexes[$i]]);
                @complexKey = sort(@complexKey) if $sortFields;
                push @multipleKey, join($complexKeySeparator, @complexKey)
            }

            # sort multiple keys and create key string
            @multipleKey = sort(@multipleKey) if $sortFields;
            my $fullKey = join $fieldSeparator, @multipleKey;

            $lines{$fullKey} = \@fields;
            $keys_hash{$fullKey} = 1;
        }
        close INPUT_FILE;

        $files{$inputPath} = \%lines;
    }
}

# Open output file
my $outputPath = $ARGV[$#ARGV];
open OUTPUT_FILE, ">" . $outputPath or die $!;
my @keys = sort keys(%keys_hash); 

# Leave blank places for key columns
for(my $pf = 0; $pf <= $#primaryFieldIndexes; $pf++)
{
    print OUTPUT_FILE $fieldSeparator;
}

# Print column headers
foreach my $argnum (0 .. ($#ARGV - 1)) 
{
    my $filePattern = $ARGV[$argnum];
    my @matchedFiles = < $filePattern >;
    foreach my $inputPath (@matchedFiles) 
    {
        print OUTPUT_FILE $inputPath;

        for(my $f = 0; $f < $fieldsCount - $#primaryFieldIndexes - 1; $f++)
        {
            print OUTPUT_FILE $fieldSeparator;
        }
    }
}

# Print merged columns
print OUTPUT_FILE "\n";
foreach my $key ( @keys )
{
    print OUTPUT_FILE $key;

    foreach my $argnum (0 .. ($#ARGV - 1)) 
    {
        my $filePattern = $ARGV[$argnum];
        my @matchedFiles = < $filePattern >;
        foreach my $inputPath (@matchedFiles) 
        {
            my $lines = $files{$inputPath};

            for(my $i = 0; $i < $fieldsCount; $i++)
            {
                next if exists $primaryFieldIndexesHash{$i} && !$includePrimaryField;
                print OUTPUT_FILE $fieldSeparator;
                print OUTPUT_FILE $lines->{$key}->[$i] if exists $lines->{$key}->[$i];
            }
        }
    }

    print OUTPUT_FILE "\n";
}
close OUTPUT_FILE;