#!/opt/bin/perl
#
###########################################################################
#                                                                         #
#  Tool for sorting the records of two input files into two output files  #
#  each (!) depending on wether the record has a unique key with respect  #
#  to the two input files or a key shared by both of the two input files. #
#  Each line in the input files is considered to be a record, which is    #
#  supposed to be divided into fields by ':'.                             #
#                                                                         #
###########################################################################
#                                                                         #
#    Version 1.0     -   Written 06.11.95 by Steffen Beyer                #
#                                                                         #
###########################################################################
#                                                                         #
#    Copyright (C) 1995 by software design & management GmbH & Co. KG     #
#                                                                         #
###########################################################################
#
# Some important default settings...
#
$version = 'version 1.0';
#
$self = $0;
$self = s!^.*/!!;
#
# Display usage if tool was called without parameters:
#
if (@ARGV == 0) { $help = 1; }
#
# Get and check command line options:
#
while (@ARGV)
{
    $_ = shift;
    if    (/^-\?$/)   { $help = 1; }
    elsif (/^-h$/)    { $help = 1; }
    elsif (/^-.*/)    { $error = 1; unless (defined $option) { $option = $_; } }
    else              { push(@filename, $_); }
}
#
# Was help requested?
#
if ($help)
{
    print <<"@@";

'$self' $version

Usage:  $self  [ <options> ]*  <filename_1>  <filename_2>

where <options> is one of the following:

  -h                      produces this help screen

  -?                      produces this help screen

@@
    exit;
}
#
# Unknown option encountered?
#
if ($error)
{
    die
  "Error: Unknown option '$option' encountered!\nEnter '$self -h' for help.\n";
}
#
# Are there two filenames specified?
#
if (@filename != 2)
{
    die "Error: You must specify two filenames!\nEnter '$self -h' for help.\n";
}
#
# Extract the two filenames:
#
$file_a = $filename[$[];
$file_b = $filename[$[+1];
#
# Do the specified files exist?
#
unless (($file_a ne "") && (-f $file_a))
{
    die "Error: Can't find file '$file_a'!\n";
}
unless (($file_b ne "") && (-f $file_b))
{
    die "Error: Can't find file '$file_b'!\n";
}
#
# Scan first file, Pass 1:
#
open(FILE_A, "<$file_a") || die "Can't open '$file_a': $!\n";
#
while (<FILE_A>)
{
    chop if /\n$/;
    ($user, $name) = (split(/:/,$_,6))[0,4];
    $name =~ s/[\t\r\n\f]/ /g;
    $name =~ s/  / /g;
    $name =~ s/^\s*//;
    $name =~ s/\s*$//;
    $name =~ s/\s*\(.*\)\s*$//;
    $name =~ s/^\s*Dr\.\s*//;
    $name =~ tr/A-Z/a-z/;
    $name =~ m/\s*(\S+)\s*$/;
    $lastname = $1;
    $firstname = $`;
    $firstname =~ s/^\s*//;
    $firstname =~ s/\s*$//;
    if ($lastname eq "")
    {
        $lastname = $firstname;
        $firstname = "";
    }
    $users{$user} = 1;
    $names{$name} = 1;
}
#
close(FILE_A);
#
# Scan second file, Pass 1:
#
$empty_intersection = 1;
#
open(FILE_B, "<$file_b") || die "Can't open '$file_b': $!\n";
#
while (<FILE_B>)
{
    chop if /\n$/;
    ($user, $name) = (split(/:/,$_,6))[0,4];
    $name =~ s/[\t\r\n\f]/ /g;
    $name =~ s/  / /g;
    $name =~ s/^\s*//;
    $name =~ s/\s*$//;
    $name =~ s/\s*\(.*\)\s*$//;
    $name =~ s/^\s*Dr\.\s*//;
    $name =~ tr/A-Z/a-z/;
    $name =~ m/\s*(\S+)\s*$/;
    $lastname = $1;
    $firstname = $`;
    $firstname =~ s/^\s*//;
    $firstname =~ s/\s*$//;
    if ($lastname eq "")
    {
        $lastname = $firstname;
        $firstname = "";
    }
    if ($users{$user} == 1)
    {
        $users{$user} = 3;
        $empty_intersection = 0;
    }
    else
    {
        $users{$user} = 2;
    }
    if ($names{$name} == 1)
    {
        $names{$name} = 3;
        $empty_intersection = 0;
    }
    else
    {
        $names{$name} = 2;
    }
}
#
close(FILE_B);
#
# No common keys in the two files?
#
if ($empty_intersection)
{
    printf("The intersection of the two sets of keys of\n");
    printf("file '%s'\n", $file_a);
    printf("and\n");
    printf("file '%s'\n", $file_b);
    printf("is empty.\n\n");
    exit;
}
#
# Prepare output file names:
#
$file_a_0 = $file_a . '.0';
$file_a_1 = $file_a . '.1';
#
$file_b_0 = $file_b . '.0';
$file_b_1 = $file_b . '.1';
#
# Scan first file, Pass 2:
#
open(FILE_A, "<$file_a")     || die "Can't open '$file_a': $!\n";
open(FILE_A_0, ">$file_a_0") || die "Can't write '$file_a_0': $!\n";
open(FILE_A_1, ">$file_a_1") || die "Can't write '$file_a_1': $!\n";
#
while (<FILE_A>)
{
    chop if /\n$/;
    ($user, $name) = (split(/:/,$_,6))[0,4];
    $name =~ s/[\t\r\n\f]/ /g;
    $name =~ s/  / /g;
    $name =~ s/^\s*//;
    $name =~ s/\s*$//;
    $name =~ s/\s*\(.*\)\s*$//;
    $name =~ s/^\s*Dr\.\s*//;
    $name =~ tr/A-Z/a-z/;
    $name =~ m/\s*(\S+)\s*$/;
    $lastname = $1;
    $firstname = $`;
    $firstname =~ s/^\s*//;
    $firstname =~ s/\s*$//;
    if ($lastname eq "")
    {
        $lastname = $firstname;
        $firstname = "";
    }
    if (($users{$user} == 3) || ($names{$name} == 3))
    {
        print FILE_A_0 $_, "\n";
    }
    else
    {
        print FILE_A_1 $_, "\n";
    }
}
#
close(FILE_A);
close(FILE_A_0);
close(FILE_A_1);
#
# Scan second file, Pass 2:
#
open(FILE_B, "<$file_b")     || die "Can't open '$file_b': $!\n";
open(FILE_B_0, ">$file_b_0") || die "Can't write '$file_b_0': $!\n";
open(FILE_B_1, ">$file_b_1") || die "Can't write '$file_b_1': $!\n";
#
while (<FILE_B>)
{
    chop if /\n$/;
    ($user, $name) = (split(/:/,$_,6))[0,4];
    $name =~ s/[\t\r\n\f]/ /g;
    $name =~ s/  / /g;
    $name =~ s/^\s*//;
    $name =~ s/\s*$//;
    $name =~ s/\s*\(.*\)\s*$//;
    $name =~ s/^\s*Dr\.\s*//;
    $name =~ tr/A-Z/a-z/;
    $name =~ m/\s*(\S+)\s*$/;
    $lastname = $1;
    $firstname = $`;
    $firstname =~ s/^\s*//;
    $firstname =~ s/\s*$//;
    if ($lastname eq "")
    {
        $lastname = $firstname;
        $firstname = "";
    }
    if (($users{$user} == 3) || ($names{$name} == 3))
    {
        print FILE_B_0 $_, "\n";
    }
    else
    {
        print FILE_B_1 $_, "\n";
    }
}
#
close(FILE_B);
close(FILE_B_0);
close(FILE_B_1);
#
# Display results:
#
printf("Wrote records of file '%s'\n", $file_a);
printf("with shared keys into '%s'\n", $file_a_0);
printf("with unique keys into '%s'\n\n", $file_a_1);
printf("Wrote records of file '%s'\n", $file_b);
printf("with shared keys into '%s'\n", $file_b_0);
printf("with unique keys into '%s'\n\n", $file_b_1);
#
# Done.
#