#!/usr/bin/perl

# E-mail filter tool for Aptech

# version 0.2.0

# Copyright 2011, David Veliev (gotoindvdum@gmail.com).

# This program may be used under Apache License 2.0.

use strict;

# Глобальные переменные

my $fileName;

my $isRemoveDuplicates = 0;

my $isSplitByDomens = 0;

my $excludeDomen;

my $outputFile;

parseCommandLine();

processFile();

exit 0;

# Парсер параметров командной строки

sub parseCommandLine {

	my $arg = shift(@ARGV);

	my $isHasArgs = 0;

	until ($arg eq '') {

		$isHasArgs = 1;

		if($arg =~ /^-{1,2}.+$/){

			if ($arg =~ /^-{1,2}version$/){

				about();

				exit 0;

			} elsif ($arg =~ /^-{1,2}help$/){

				help();

				exit 0;

			} elsif ($arg eq '--remove-duplicates'){

				$isRemoveDuplicates = 1;

			} elsif ($arg eq '--split-by-domens'){

				$isSplitByDomens = 1;

			} elsif ($arg =~ /--exclude-domen[=]{0,1}([\.\w]*)/){

				$excludeDomen = $1;

				illegalUse() if $excludeDomen ne '' && $arg !~ /--exclude-domen=[\.\w]*/;

				$excludeDomen = shift(@ARGV) if $excludeDomen eq '';

				illegalUse() if $excludeDomen eq '';

			} elsif ($arg =~ /--output-file[=]{0,1}([\.\w]*)/){

				$outputFile = $1;

				illegalUse() if $outputFile ne '' && $arg !~ /--output-file=[\.\w]*/;

				$outputFile = shift(@ARGV) if $outputFile eq '';

				illegalUse() if $outputFile eq '';

			} else {

				illegalUse();

			}

		} else {

			illegalUse() if $fileName ne '';

			$fileName = $arg;

		}

		$arg = shift(@ARGV);

	};

	if (!$isHasArgs) {

		about();

		exit 0;

	}

}

# Вывод информации о программе

sub about {

	my $about = q {

E-mail filter tool for Aptech

version 0.2.0

Copyright 2011, David Veliev (gotoindvdum@gmail.com).

This program may be used under Apache License 2.0.

};

	print $about;

}

# Вывод доступных параметров командной строки

sub help {

	my $help = q {usage: }.$0. q { [KEYS] FILENAME [KEYS]

Parse for e-mails file FILENAME with arguments KEYS and print result to standart output stream.

Arguments:

    --help                          print this help

    --version                       print version and information about this script

    --remove-duplicates             remove e-mail duplicates

    --split-by-domens               split e-mails by domens

    --exclude-domen[=]DOMEN         exclude e-mails with domen DOMEN

    --output-file[=]OUTFILENAME     redirect standart output to this file

};

	print $help;

}

# Неправильный формат параметров командной строки

sub illegalUse {

	print "Illegal use!\n\n";

	help();

	exit 1;	

}

# Обработка файла

sub processFile {

	open FILE, $fileName or die "Failed to open $fileName: $!\n";

	my @emails;

	while(<FILE>){

		push @emails, $1 while s/([\.\w]+@[\.\w]+)//;

	}

	close FILE;

	if($isRemoveDuplicates){

		my %tmp;

		@emails = grep {! $tmp{$_}++} @emails;

	}

	if($excludeDomen){

		@emails = grep {

			$1 ne $excludeDomen if $_ =~ /[\.\w]+@([\.\w]+)/;

		} @emails;

	}

	if($isSplitByDomens){

		@emails = sort {

			my $a_domen = $1 if $a =~ /[\.\w]+@([\.\w]+)/;

			my $b_domen = $1 if $b =~ /[\.\w]+@([\.\w]+)/;

			$a_domen cmp $b_domen;

		} @emails;

		my @tmp;

		my %tmp;

		while($#emails >= 0){

			my $email = shift @emails;

			my $domen = $1 if $email =~ /[\.\w]+@([\.\w]+)/;

			push @tmp, '' if !$tmp{$domen}++;

			push @tmp, $email;

		}

		@emails = @tmp;

	}

	if($outputFile){

		open STDOUT, ">$outputFile" or die "Failed to open $outputFile: $!\n";

	}

	print "$_\n" for (@emails);

}