#!/usr/bin/perl

# Calculates the number of downloads of papers with and without journal
# references.
# Author: Ian Hickman

use strict;

my (%downloaded, @field);
my ($paperfield, $papernumber, $paper);
my ($jr, $nojr, $total)=(0, 0, 0);
openpapers();

# loop through all downloads
while( <STDIN> ){
  @field=split;
  if( $field[6]=~/papers/ ){
    $_=$field[6];
    ($paperfield, $papernumber)=/\/ftp\/(.*)\/papers.*(\d{7})/; 
  } else {
    $_=$field[6];
    ($paperfield, $papernumber)=/\/.*\/(.*)\/\??(\d{7})/;
  }
  $paper=$paperfield."/".$papernumber;
  if( exists( $downloaded{$paper} ) ){
    $jr++;
  } else {
    $nojr++;
  }
}

print"jr\t$jr\tno\t$nojr\n";
print" total with jr is $total\n";
# load all papers with journal references 
sub openpapers {
  open( IN, "../q1/d_paperswithjr" ) || die"Cant open: $!";
  while( <IN> ){
    @field=split;
    $downloaded{$field[2]}=0;
    $total++;
  }
  close( IN ) || die"Cant close: $!";
}