#!/usr/bin/perl

# Author: Ian Hickman

use strict;

use Time::Local;

my %months=( "Jan", 0, "Feb", 1, "Mar", 2, "Apr", 3, "May", 4, "Jun", 5, "Jul", 6, "Aug", 7, "Sep", 8,
"Oct", 9, "Nov", 10, "Dec", 11 );

my (%table, %subtimes, %subdates);
my (%rank, %top, %middle, %bottom, %unknown);
my ($topt, $middlet, $bottomt, $unknownt)=(0, 0, 0, 0);
my ($day, $month, $monthtext, $year);
my ($paper, $paperid, $paperfield, @field);
my ($hittime, $subtime, $diff);

my $line=0;

print STDERR "opening papers\n";
openpapers();
print STDERR "opening ranks\n";
openranks();
print STDERR "doing the rest\n";

while( <STDIN> ){
  
  @field=split;

  # get the date of the hit (nearest day)
  $_=$field[3];
  ($day, $monthtext, $year)=/^.(\d\d).(\w+).(\d{4}).*/;
  $month=$months{$monthtext};
  $hittime=timelocal( 0, 0, 0, $day, $month, $year );
    
  # get the unique id of the paper
  $_=$field[6];
  ($paperid)=/(\d{7})/;
  if( $field[6]=~/papers/ ){
    ($paperfield)=/\/ftp\/([\w-]+)\.?.*\/papers/;
  } else {
    ($paperfield)=/\/.*\/([\w-]+)\.?.*\/\??\d{7}/;
  }
  $paper="$paperfield"."/"."$paperid";

  # get the date the paper was first submitted (nearest day)
  if( exists($subtimes{$paper}) ){
    $subtime=$subtimes{$paper};
    $diff=($hittime-$subtime)/(3600*24);
    if( $diff>=0 ){
      if( roundoff($rank{$paper}) eq "3"){
        $top{roundoff($diff)}++;
        $topt++;
      } elsif( roundoff($rank{$paper}) eq "2"){
        $middle{roundoff($diff)}++;
        $middlet++;
      } elsif( roundoff($rank{$paper}) eq "1"){
        $bottom{roundoff($diff)}++;
        $bottomt++;
      } else {
        $unknown{roundoff($diff)}++;
        $unknownt++;
      }
    } else {
      print STDERR "$line $diff $field[3] $subdates{$paper} $paper\n";
    }
  } else {
    print STDERR "$line $paper $field[6] not found\n";
  }
  $line++;
} # end while

my $time;

# save top hit distribution
open( FILE, "> topdistribution" ) || die"Cant open: $!";
foreach $time ( keys ( %top ) ){
  print FILE ("$time\t".($top{$time}/$topt)."\n");
}
close( FILE ) || die"Cant close: $!";

# save middle hit distribution
open( FILE, "> middledistribution" ) || die"Cant open: $!";
foreach $time ( keys ( %middle ) ){
  print FILE ("$time\t".($middle{$time}/$middlet)."\n");
}
close( FILE ) || die"Cant close: $!";

# save bottom hit distribution
open( FILE, "> bottomdistribution" ) || die"Cant open: $!";
foreach $time ( keys ( %bottom ) ){
  print FILE ("$time\t".($bottom{$time}/$bottomt)."\n");
}
close( FILE ) || die"Cant close: $!";

# save unknown hit distribution
open( FILE, "> unknowndistribution" ) || die"Cant open: $!";
foreach $time ( keys ( %unknown ) ){
  print FILE ("$time\t".($unknown{$time}/$unknownt)."\n");
}
close( FILE ) || die"Cant close: $!";


sub roundoff{
  my ($num)=@_;
  if( int($num) eq $num ){
    return $num;
  } elsif( ($num-int($num))<0.5 ){
    return int($num);
  } else {
    return (int($num)+1);
  }
}

sub openpapers{

  open( PAPERLIST, "../../d_firstsubmit" ) || die"Cant open paperlist: $!";

  # loop through all the papers
  while( defined( $_ = <PAPERLIST> )){

    # loop through list of papers getting date and paperid
    my @field;
    @field=split;
    $_=$field[0];
    ($year, $month, $day)=/(....)(..)(..)/;
    if( $month>0 ){
      $time=timelocal( 0, 0, 0, $day, $month-1, $year );
      if( !(exists ($subtimes{$field[2]})) ){
        $subtimes{$field[2]}=$time;
        $subdates{$field[2]}=$field[0];
      } else {
        print"clash\n";
      }
    }
  }
  close( PAPERLIST ) || die"Cant close paper list: $!";
}

sub openranks {
  open( FILE, "myrankedpapers" ) || die"Cant open: $!";
  while( <FILE> ){
    @field=split;
    $rank{$field[0]}=$field[1];
  }
  close( FILE );
}