#!/usr/bin/perl

#######################################################################
#
#  Filename:    removeDupLines.pl
#
#  Purpose:     This script will remove duplicate lines from the data directory
#               of the specified product. The lines it will remove are
#               defined at the top of the program.
#
#  Date:        Jul. 02, 2002
#               Saeid Asgari-Marani
#
#  Notes:
#
#######################################################################

if ($#ARGV < 0)
{
        print "\n\nUsage : $0 product <text to remove>\n";
        print "   which\n";
        print "        product        : product name like nms\n";
        print "        text           : text which duplicate lines containing that line will be removed\n\n";
        exit 1;
}

$product = $ARGV[0];

$CDNProducts =  'passport ' .
                'omsp ' .
                '6K ' .
                'dpn ' .
                'pnp ' .
                'expressmanager ' .
                'servicemonitor ' .
                'mdm ';


$dataroot = "";
$dataroot = "/d/cdn_metrics/clarifyWeekly/$product" if ($CDNProducts =~ $product);
$dataroot = "/d/mag_metrics/clarifyWeekly/$product" if ($product eq "6K");

#$dataroot = "/home/users/ironwood/TestData";
if (!$dataroot)
{
        print "Invalid product name $product\n";
        exit 1;
}

$numFiles = 0;

&FindIt($dataroot);

#print "No. of files which data exist : $numFiles\n";



# main body of script

sub FindIt {
        local ($directory) = @_;

        opendir(DATAROOT, $directory) || print "Error: opening $directory";
        local @datadirs = grep(!/^\.\.?$/, readdir(DATAROOT));
        closedir(DATAROOT);
#       print "Working on $directory\n";
        local $dir;

        foreach $dir (@datadirs) {

                if(-f "$directory/$dir") {
#                       print "\tChecking $directory/$dir\t";

                        open(FILECHECK, "$directory/$dir") || print "Error: reading $directory/$dir\n";
                        local @lines = <FILECHECK>;
                        close(FILECHECK);
# We have to check all lines to see if all starts with a year,
# so it will be a file to work on it.
                        $ItIsDataFile = 1;
                        foreach (@lines)
                        {
                                if (!(/^([0-9][0-9][0-9][0-9][0-9][0-9])\s+.*/))
                         #This file is not a data file
                                {
                                        $ItIsDataFile = 0;
                                        last;
                                }
                        }
                        if ($ItIsDataFile)
                        {
# First we sort them
                                @sortLines = sort(@lines);
                                $Changed = 0;
                                $dupWeek = 0;
                                $oldWeek = 0;
                                $oldYear = 0;
                                $oldBusWeek = 0;
                                $printText = "$directory/$dir : \n";
                                for ($i = 0; $i <= $#lines; $i++)
                                {
                                        $sortLines[$i] =~ /^([0-9][0-9][0-9][0-9])([0-9][0-9])\s+.*/;
                                        $curYear = $1;
                                        $curBusWeek = $2;
                                        $curWeek = "$curYear$curBusWeek";
                                        if ($oldWeek != "")
                                        {
                                                next if ($curWeek == $dupWeek);
                                                if ($curWeek - $oldWeek == 0 && $curWeek != $dupWeek)
                                                {
                                                        print "$printText\tDuplicates $curWeek\n";
                                                        $printText = "";
                                                        $dupWeek = $curWeek;
                                                        next;
                                                }
                                                if ($curWeek - $oldWeek != 1)
                                                {
                                                        if ($curYear - $oldYear != 1)
                                                        {
                                                                print "$printText\tMissing data between $oldWeek <-> $curWeek\n";
                                                                $printText = "";
                                                        } elsif (!($curBusWeek == "01" && ($oldBusWeek == "52" || $oldBusWeek == "53")))
                                                        {
                                                                print "$printText\tMissing data between $oldWeek <-> $curWeek\n";
                                                                $printText = "";
                                                        }
                                                }

                                        }
                                        $oldWeek = $curWeek;
                                        $oldYear = $curYear;
                                        $oldBusWeek = $curBusWeek;
                                        $Changed = 1 if ($sortLines[$i] ne $lines[$i]);
                                }
                                if ($Changed)
                                {
                                        print "\tSorting $directory/$dir\n";
                                        open OUT, ">$directory/$dir" || print "Error: writing  $directory/$dir\n";
                                        print OUT @sortLines;
                                        close OUT;
                                }
                        }
                        next;
                }
                &FindIt("$directory/$dir") if (-d "$directory/$dir");
        }
}