- #!/usr/bin/env perl
- use strict;
- use Time::HiRes qw( gettimeofday );
- use CGI;
- use DBI;
- use Geo::IP;
-
- my ( $start_sec, $start_micro ) = gettimeofday;
- my ($start_time) = ( $start_sec * 1000 ) + round( $start_micro / 1000 );
-
- my ($page);
-
- # Database connection.
- my ($dsn) = "DBI:mysql:database=eprintstats;host=localhost";
- my ($user_name) = "eprintstatspriv";
- my ($password) = "AuldGrizzel";
- my ( $connect, $query, %types, %unmapped, $stat, $row, $num_rows, $vtype );
- my ($where) = '';
-
- # Geolocation database.
- my ($gi);
- my ($gidb) = '/usr/local/share/GeoIP/GeoLiteCity.dat';
-
- # Miscellaneous variable.
- my ( %cities, %IPs );
- my ($num_entries) = -1; # include all entries from database
- my ($num_hits) = 0;
- my ( $ip, $count, $location );
- my ( $lat, $long, $city ) = ( 0, 0, '' );
- my ($show_only) = 'both'; # include both abstracts & downloads
- my ($eprint) = '';
-
- $page = new CGI;
- print $page->header( -type => "text/xml", -Pragma => 'no-cache', -Cache-Control => 'no-cache' );
- $num_entries = $page->param('top') if ( defined $page->param('top') );
- $show_only = $page->param('show') if ( defined $page->param('show') );
- $eprint = $page->param('eprint') if ( defined $page->param('eprint') );
-
- $gi = Geo::IP->open( $gidb, GEOIP_STANDARD )
- or die "Unable to open GeoIP database $gidb\n";
-
- $connect = DBI->connect( $dsn, $user_name, $password, { RaiseError => 1 } );
-
- $types{'download'} = $types{'abstract'} = 0;
- $unmapped{'download'} = $unmapped{'abstract'} = 0;
-
- # Set up query.
- if ( $show_only eq 'both')
- {
- $where = "view_type IN ('download', 'abstract')";
- }
- else
- {
- $where = "view_type = '$show_only'";
- }
-
- $where .= " AND archiveid IN ($eprint)" unless ( $eprint eq '' );
-
- $query = "SELECT ip, view_type, COUNT(*) AS count
- FROM view
- WHERE $where
- GROUP BY ip, view_type
- ORDER BY count DESC" . ( ( $num_entries > 0 ) ? " LIMIT $num_entries" : '' );
-
- $stat = $connect->prepare($query);
- $stat->execute();
- $num_rows = $stat->rows;
-
- if ( $num_rows > 0 )
- {
- $num_entries = $num_rows if ( $num_entries < 1 );
-
- while ( $row = $stat->fetchrow_hashref() )
- {
- $ip = $row->{'ip'};
- $count = $row->{'count'};
- $vtype = $row->{'view_type'};
-
- $IPs{$ip} = 1;
-
- $location = $gi->record_by_addr($ip);
-
- if ( defined($location) )
- {
- $lat = $location->latitude;
- $long = $location->longitude;
- $city = (
- ( $location->city eq '' )
- ? 'Unknown'
- : $location->city
- ) . " ($lat, $long)";
-
- if ( !defined( $cities{$city} ) )
- {
- $cities{$city}{'lat'} = $lat;
- $cities{$city}{'long'} = $long;
- $cities{$city}{'abstract'} = 0;
- $cities{$city}{'download'} = 0;
- }
- $cities{$city}{$vtype} += $count;
- $types{$vtype} += $count;
- }
- else
- {
- $unmapped{$vtype} += $count;
- }
- }
-
- # Need to wait until we have all the counts before writing the data out.
- print '<?xml version="1.0"?>' . "\n";
- print '<markers abs="'
- . $types{'abstract'}
- . '" dl="'
- . $types{'download'}
- . '" ips="'
- . scalar( keys %IPs )
- . '" ua="'
- . $unmapped{'abstract'}
- . '" ud="'
- . $unmapped{'download'} . '">' . "\n";
-
- # Generate markers for each city.
- foreach $city ( keys %cities )
- {
- print '<marker city="' . $city
- . '" lat="'
- . $cities{$city}{'lat'}
- . '" long="'
- . $cities{$city}{'long'}
- . '" abs="'
- . $cities{$city}{'abstract'}
- . '" dl="'
- . $cities{$city}{'download'} . '" />' . "\n";
- }
-
- print "</markers>\n";
- }
-
- $stat->finish();
- $connect->disconnect();
-
- my ( $finish_sec, $finish_micro ) = gettimeofday();
- my ($finish_time) = ( $finish_sec * 1000 ) + round( $finish_micro / 1000 );
-
- open RESULTS, ">>/tmp/google_results_$num_entries.txt" or die "Argh!\n";
- print RESULTS ( $finish_time - $start_time ) . "\n";
- close RESULTS;
-
- sub round
- {
- my ($n) = shift;
- return int( $n + 0.5 * ( $n <=> 0 ) );
- }