#!/usr/bin/perl

# A script to calculate the hit rate of papers before and after a 
# journal reference was added to them.
# Author: Ian Hickman

use strict;

use Time::Local;

my %months=( "Jan", "00", "Feb", "01", "Mar", "02", "Apr", "03", "May", "04", "Jun", "05", "Jul", "06", "Aug", "07", "Sep",
"08", "Oct", "09", "Nov", "10", "Dec", "11" );

my ( $startoflogs, $endoflogs );
my ( %firstpublished, %lengthbeforejr, %lengthafterjr );
my ( %hitsbeforejr, %hitsafterjr, %jrdate, %normaldate);
my ($total, $papercount, $hitcount)=(0,0,0);

openchanges();
print STDERR "Total: $total\n";

while( <STDIN> ){
  my ($paper, $time, @field);
  my ($sec, $min, $hour, $day, $month, $monthtext, $year);
  
  @field=split;

  # use diffinjournal < alldownloads

  # get the date (and time) of the hit
  $_=$field[3];
  ($day, $monthtext, $year, $hour, $min, $sec)=/.(\d\d).(\w+).(\d{4}).(\d\d).(\d\d).(\d\d)/;
  $month=$months{$monthtext};
  $time=timelocal( $sec, $min, $hour, $day, $month, $year );

  $papercount++;

  # if the paper that was downloaded is in the hash table
  $_=$field[6];
  ($paper)=/.*\/(.*\/\d{7}$)/;

  if( $firstpublished{$paper} ){

    # is the paper valid?
    if( $time>$firstpublished{$paper} ){

      $hitcount++;
      print STDERR "papers: $papercount hits: $hitcount\n";
      # this is our paper, but is the hit before of after the journal ref?
      if( $time<$jrdate{$paper} ){
        $hitsbeforejr{$paper}++;
      } else {
        $hitsafterjr{$paper}++;
      } # end if $time<$jrdate

    } # end if this is our paper
  } # end if in table
} # end while

my $paper;
foreach $paper ( keys ( %firstpublished ) ){
  if( $hitsbeforejr{$paper} ne "0" || $hitsafterjr{$paper} ne "0" ){
    $hitsbeforejr{$paper};
    $hitsafterjr{$paper};
    print"$lengthbeforejr{$paper} ";
    print"$hitsbeforejr{$paper} ";
    print"$lengthafterjr{$paper} ";
    print"$hitsafterjr{$paper}\n";
  } 
}

sub openchanges{
  open( PAPERLIST, "papswithjourns" ) || die "Cant open list of papers: $!";

  my ($datejournalref);
  my ($day, $month, $year, $datepublished);

  # calculate the start and end times we have logs for
  # All months -1.
  $startoflogs=timelocal( 0, 0, 0, 24, 7, 1999 );
  $endoflogs=timelocal( 0, 0, 0, 9, 4, 2000 );

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

    my @field;

    # paperlist line format is:
    # $field[0]="paper reference" 
    # $field[1]="submission date in last 6 months"
    # $field[2]="date journal ref was added"

    @field=split;

    # get the date that a journal ref was added
    $_=$field[2];
    ($year, $month, $day)=/(....)(..)(..)/;
    $datejournalref=timelocal( 0, 0, 0, $day, $month-1, $year );

    # check the the date the journal ref was added is within the boundarys
    if( $datejournalref<$endoflogs && $datejournalref>$startoflogs ){

      $total++;
      $jrdate{$field[0]}=$datejournalref;
      $normaldate{$field[0]}=$field[2];

      # calculate the date first published
      $_=$field[1];
      ($year, $month, $day)=/(....)(..)(..)/;
      $datepublished=timelocal( 0, 0, 0, $day, $month-1, $year );
      if( $datepublished<$startoflogs ){
        $datepublished=$startoflogs;
      } #endif

      # set some global hash table values
      $firstpublished{$field[0]}=$datepublished;
      $lengthbeforejr{$field[0]}=$datejournalref-$datepublished;
      $lengthafterjr{$field[0]}=$endoflogs-$datejournalref;
      $hitsbeforejr{$field[0]}=0;
      $hitsafterjr{$field[0]}=0;

    } # endif
  } # endwhile
  close( PAPERLIST ) || die "Cant close paper list: $!";
}