#!/usr/bin/perl -w
use strict;


##################
#
#  bib.pl - parseBib function
#
#  convert HIP v2/v3 XML output into a Perl data structure
#
#  Knocked together by Dave Pattern <d.c.pattern@hud.ac.uk>
#  If you spot any bugs, please let me know!
#
#############
#
#  version 0.01 - last updated 22/Apr/2006
#
#########
#
# (cc) 2005
#
# http://creativecommons.org/licenses/by-nc-sa/2.5/
#
#####




    use LWP::UserAgent;
    use XML::Simple;  
    use Data::Dumper;  
    use Encode;

#    use HTML::Entities qw(:DEFAULT encode_entities_numeric);
#    use HTML::Entities::Numbered;


    $Data::Dumper::Indent = 2;
    $Data::Dumper::Sortkeys = 1;


    my $proxy = '';
#    $proxy = 'http://leed-cache-2.server.ntli.net:8080';


    # BIB URL TO PARSE...

    my $url = 'http://webcat.hud.ac.uk/ipac20/ipac.jsp?session=TS45720408K82.6491&profile=cls&source=~!horizon&view=items&uri=full=3100001~!26231~!54&ri=9&aspect=subtab33&menu=search&ipp=20&spp=20&staffonly=&term=Fran%C3%A7ois&index=.GW&uindex=&aspect=subtab33&menu=search&ri=9';

    $url =~ s/\#focus//g;
    if( $url !~ /&GetXML=true/ ) { $url .= '&GetXML=true' }


    my $ua  = LWP::UserAgent->new;
    $ua->timeout(20);
    if( $proxy ) { $ua->proxy( ['http'], $proxy ) }
    my $response = $ua->get( $url );


    unless( $response->is_success ) 
    {
        die $response->status_line
    }
    my $content = $response->content;

    my $info    = parseBib( \$content );


    # DUMP THE DATA STRUCTURE FOR REFERENCE...

    open( OUT, ">./dump_output.txt" );
    print OUT Dumper($info);
    close( OUT );

    print "TITLE = ".$info->{title}."\n";










sub parseBib
{
    my $content = shift;
    if( ref($content) eq 'SCALAR' ) { $content = $$content }

    my %ret = ( );


    # USE EVAL JUST IN CASE WE GET AN XML PARSING ERROR...

    my $xml = eval { XMLin( $content, ForceArray => 1 ) };
    if( $@ )
    { 
        $ret{error} = 'XML parse failed';
        $ret{errorText} = $@;
        $ret{errorType} = 'fatal';
        return( \%ret );
    }


    # DUMP THE XML STRUCTURE FOR REFERENCE...

    open( OUT, ">./dump_input.txt" );
    print OUT Dumper($xml);
    close( OUT );


    # ITEM TITLE(S)
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{TITLE}[0]->{data} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        if( $count )
        {
            $ret{title} = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{TITLE}[0]->{data}[0]->{text}[0] || '';
        }

        foreach ( 1 .. $count )
        {
            my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{TITLE}[0]->{data}[($_-1)]->{text}[0] || '';
            if( $value && ref( $value ) ne 'HASH' ) 
            { 
                push @array, $value;
            }
        }
	if( @array )
        {
            $ret{titleCount} = scalar(@array);
            $ret{titles}     = \@array;
        }
    }

    # AUTHOR(S)
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{AUTHOR}[0]->{data} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        foreach ( 1 .. $count )
        {
            my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{AUTHOR}[0]->{data}[($_-1)]->{text}[0] || '';
            if( $value && ref( $value ) ne 'HASH' ) 
            { 
                push @array, $value 
            }
        }
        if( @array ) 
        {
            $ret{authorCount} = scalar(@array);
            $ret{authors}     = \@array;
        }
        else { $ret{authorCount} = 0 }
    }

    # SUBJECT(S)
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{SUBJECT}[0]->{data} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        foreach ( 1 .. $count )
        {
            my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{SUBJECT}[0]->{data}[($_-1)]->{text}[0] || '';
            if( $value && ref( $value ) ne 'HASH' ) 
            { 
                push @array, $value;
            }
        }
        if( @array ) 
        {
            $ret{subjectCount} = scalar(@array);
            $ret{subjects}     = \@array;
        }
        else { $ret{subjectCount} = 0 }
    }

    # PUBLISHER(S)
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{PUBLISHER}[0]->{data} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        foreach ( 1 .. $count )
        {
            my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{PUBLISHER}[0]->{data}[($_-1)]->{text}[0] || '';
            if( $value && ref( $value ) ne 'HASH' ) 
            { 
                push @array, convert( $value );
            }
        }
        if( @array ) 
        {
            $ret{publisherCount} = scalar(@array);
            $ret{publishers}     = \@array;
        }
        else { $ret{publisherCount} = 0 }
    }

    # PUBLISH DATE(S)
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{PUBDATE}[0]->{data} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        foreach ( 1 .. $count )
        {
            my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{PUBDATE}[0]->{data}[($_-1)]->{text}[0] || '';
            if( $value && ref( $value ) ne 'HASH' ) 
            { 
                push @array, $value;
            }
        }
        if( @array ) 
        {
            $ret{publishDateCount} = scalar(@array);
            $ret{publishDates}     = \@array;
        }
        else { $ret{publishDateCount} = 0 }
    }

    # CLASS/CALL NUMBER(S)
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{CALL}[0]->{data} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        foreach ( 1 .. $count )
        {
            my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{CALL}[0]->{data}[($_-1)]->{text}[0] || '';
            if( $value && ref( $value ) eq 'HASH' ) { $value = '' }
            push @array, $value;
            $ret{call}{$_} = $value;
        }
        if( @array ) 
        {
            $ret{callCount} = $count;
            $ret{calls}     = \@array;
        }
        else { $ret{callCount} = 0 }
    }

    # 856 URL(S)
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{URL856}[0]->{data} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        foreach ( 1 .. $count )
        {
            my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{URL856}[0]->{data}[($_-1)]->{text}[0] || '';
            if( $value && ref( $value ) ne 'HASH' ) 
            { 
                push @array, $value;
                $ret{url}{$_} = $value;
            }
        }
        if( @array ) 
        {
            $ret{urlCount} = $count;
            $ret{urls}     = \@array;
        }
        else{ $ret{urlCount} = 0 }
    }

    # ISBN
    {
        $ret{isbn} = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{isbn}[0] || '';
    }

    # BIB NUMBER
    {
        $ret{bib} = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{key}[0] || '';
    }

    # BIB HEADERS & CONTENT
    {
        my $check = $xml->{fullnonmarc}[0]->{searchresults}[0]->{header}[0]->{col} || [ ];
        my $count = scalar(@$check) || 0;
        my @array = ( );

        my %dupes = ( );

        foreach my $x ( 1 .. $count )
        {
            my $header = $xml->{fullnonmarc}[0]->{searchresults}[0]->{header}[0]->{col}[($x-1)]->{label}[0] || '';

            $dupes{$header}++;

            if( $dupes{$header} > 1 ) { $header .= " {$dupes{$header}}" }

            push @array, $header;

            my $check2 = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{cell}[($x-1)]->{data} || [ ];
            my $count2 = scalar(@$check2) || 0;
            my @array2 = ( );        

            foreach my $y ( 1 .. $count2 )
            {
                my $value = $xml->{fullnonmarc}[0]->{searchresults}[0]->{results}[0]->{row}[0]->{cell}[($x-1)]->{data}[($y-1)]->{text}[0] || '';
                if( ref( $value ) eq 'HASH' )
                {
                    $value = '';
                }
                if( $value ) 
                {  
                    push @array2, $value;
                }
            }

            $ret{bibContents}{$header} = \@array2;
        }

        $ret{bibHeaderCount} = $count;
        $ret{bibHeaders}     = \@array;
    }


    $ret{itemCount} = 0;


    # ITEM INFO
    {
        my $check1 = $xml->{items}[0]->{searchresults}[0]->{header}[0]->{col} || [ ];
        my $headCount = scalar(@$check1) || 0;
        my @headArray = ( );
        my @allItemDetails = ( );
        my @allItemFixedDetails = ( );

        foreach my $x ( 1 .. $headCount )
        {
            my $value = $xml->{items}[0]->{searchresults}[0]->{header}[0]->{col}[($x-1)]->{label}[0] || '';
            if( $value && ref( $value ) ne 'HASH' ) 
            { 
                push @headArray, $value;
            }
        }

        if( @headArray ) 
        {        
            $ret{itemHeaderCount} = $headCount;
            $ret{itemHeaders} = \@headArray;
        }
        else { $ret{itemHeaderCount} = 0 }

        my $check2 = $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row} || [ ];
        my $itemCount = scalar(@$check2) || 0;

        $ret{itemCount} = $itemCount;

        my @itemNumbers = ( );

        foreach my $x ( 1 .. $itemCount )
        {
            push @itemNumbers, $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row}[($x-1)]->{key}[0] || '';

            my %i = ( );

            $i{RESTRICTIONS}      = $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row}[($x-1)]->{RESTRICTIONS}[0]->{data}[0]->{text}[0]      || '';
            $i{MIDSPINE}          = $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row}[($x-1)]->{MIDSPINE}[0]->{data}[0]->{text}[0]          || '';
            $i{AVAILABLETHRU}     = $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row}[($x-1)]->{AVAILABLETHRU}[0]->{data}[0]->{text}[0]     || '';
            $i{TEMPORARYLOCATION} = $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row}[($x-1)]->{TEMPORARYLOCATION}[0]->{data}[0]->{text}[0] || '';
            $i{AVAILABILITYDATE}  = $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row}[($x-1)]->{AVAILABILITYDATE}[0]->{data}[0]->{text}[0]  || '';

            push @allItemFixedDetails, \%i;

            my @itemArray = ( );
            foreach my $y ( 1 .. $headCount )
            {
                my $value = $xml->{items}[0]->{searchresults}[0]->{results}[0]->{row}[($x-1)]->{cell}[($y-1)]->{data}[0]->{text}[0] || '';
                if( ref( $value ) eq 'HASH' )
                { 
                    $value = '';
                }
                push @itemArray, $value;
            }
            push @allItemDetails, \@itemArray;        
        }

        if( @allItemFixedDetails ) { $ret{itemFixedDetails} = \@allItemFixedDetails }
        if( @allItemDetails      ) { $ret{itemDetails}      = \@allItemDetails      }
        if( @itemNumbers         ) { $ret{itemNumbers}      = \@itemNumbers         }
    }


    # SERIALS INFO
    {
        my $check1 = $xml->{subscriptionsummary}[0]->{header}[0]->{col} || [ ];
        my $headCount1 = scalar(@$check1) || 0;
        my @headArray1 = ( );
        my @serialCopyDetails = ( );

        foreach my $x ( 1 .. $headCount1 )
        {
            push @headArray1, $xml->{subscriptionsummary}[0]->{header}[0]->{col}[($x-1)]->{label}[0] || '';
        }
        if( @headArray1 ) { $ret{serialHeaders} = \@headArray1 }

        my $check3 = $xml->{subscriptionsummary}[0]->{serial} || [ ];
        my $itemCount = scalar(@$check3) || 0;
        $ret{serialCount} = $itemCount;
        if( $itemCount ) { $ret{serialCopyDetailCount} = $itemCount }

        my @serialCopyKeys = ( );
        my @serialCopyLocs = ( );
        $ret{serialCopyNotes} = 0;

        my @allNotesTop = ( );
        my @allTextsTop = ( );
        my @allLabelTop = ( );

        foreach my $x ( 1 .. $itemCount )
        {
            push @serialCopyKeys, $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{copykey}[0]  || '';
            push @serialCopyLocs, $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{location}[0] || '';

            my @copyArray = ( );
            foreach my $y ( 1 .. $headCount1 )
            {
                my $value = $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{copy}[0]->{cell}[($y-1)]->{data}[0]->{text}[0] || '';
                if( ref( $value ) eq 'HASH' ) { $value = '' }
                push @copyArray, $value;
            }

                  my $check4 = $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{runlist}[0]->{run} || [ ];
            my $itemCount4 = scalar(@$check4) || 0;

            my @allNotes   = ( );
            my @allTexts   = ( );
            my @allLabel   = ( );

            foreach my $y ( 1 .. $itemCount4 )
            {
                my @runDetails = ( );
                my @runNotes   = ( );

                my $check5 = $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{runlist}[0]->{run}[($y-1)]->{data}[0]->{rundata} || [ ];
                my $itemCount5 = scalar(@$check5) || 0;

                push @allLabel, $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{runlist}[0]->{run}[($y-1)]->{runlabel}[0] || '';

                my @notes = ( );
                my @texts = ( );

                foreach my $z ( 1 .. $itemCount5 )
                {
                    my $value = $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{runlist}[0]->{run}[($y-1)]->{data}[0]->{rundata}[($z-1)]->{text}[0] || '';
                    my $notes = $xml->{subscriptionsummary}[0]->{serial}[($x-1)]->{runlist}[0]->{run}[($y-1)]->{data}[0]->{rundata}[($z-1)]->{note}[0] || '';
                    if( ref( $value ) eq 'HASH' ) { $value = '' }
                    if( ref( $notes ) eq 'HASH' ) { $notes = '' }
                    push @runDetails, $value;
                    push @runNotes, $notes;
                    push @notes, $notes;
                    push @texts, $value;
                }


                push @allNotes, \@notes;
                push @allTexts, \@texts;

                if( @runNotes ) { $ret{serialCopyNotes} = 1 }
            }

            push @allNotesTop, \@allNotes;
            push @allTextsTop, \@allTexts;
            push @allLabelTop, \@allLabel;

            push @serialCopyDetails, \@copyArray;
        }

        if( @allNotesTop       ) { $ret{serialCopyRunNotes}   = \@allNotesTop       }
        if( @allTextsTop       ) { $ret{serialCopyRunDetails} = \@allTextsTop       }
        if( @allLabelTop       ) { $ret{serialCopyRunLabels}  = \@allLabelTop       }

        if( @serialCopyKeys    ) { $ret{serialCopyKeys}       = \@serialCopyKeys    }
        if( @serialCopyLocs    ) { $ret{serialCopyLocations}  = \@serialCopyLocs    }
        if( @serialCopyDetails ) { $ret{serialDetails}        = \@serialCopyDetails }
    }

    # CHECK WE'VE FOUND A BIB!

    unless( $ret{bib} )
    {
        $ret{error} = 'no bib number found';
        $ret{errorType} = 'warning';
    }

    return( \%ret );

}




sub convert
{
    my $str = shift;
    return( Encode::encode_utf8( $str ) );
}