source: trunk/gsdl/perllib/plugins/ArcPlug.pm@ 4

# plugin which recurses through directories processing
# each file it finds

package ArcPlug;

use util;
use BasPlug;
use plugin;
use arcinfo;

BEGIN {
    @ISA = ('BasPlug');
}

sub new {
    my ($class) = @_;
    my $self = new BasPlug ();

    return bless $self, $class;
}

# return 1 if this class might recurse using $pluginfo
sub is_recursive {
    my $self = shift (@_);

    return 1;
}

# return 1 if processed, 0 if not processed
# Note that $base_dir might be "" and that $file might 
# include directories
sub read {
    my $self = shift (@_);
    ($pluginfo, $base_dir, $file, $metadata, $processor) = @_;

    # see if this has a archives information file within it
    $archive_info_filename = &util::filename_cat($base_dir,$file,"archives.inf");

    if (-e $archive_info_filename) {

    # read in the archives information file
    my $archive_info = new arcinfo ();
    $archive_info->load_info ($archive_info_filename);
    
    my $file_list = $archive_info->get_file_list();

    # process each file
    foreach $subfile (sort alphabetically @$file_list) {
        # note: metadata is not carried on to the next level
        my $tmp = &util::filename_cat ($file, $subfile->[0]);
        next if $tmp eq $file;
        &plugin::read ($pluginfo, $base_dir, $tmp, {}, $processor);
    }
    
    # all books have been processed so need to output classifications
    # to infodb - note that at present you have to import before building
    if ($processor->{'mode'} eq 'infodb') {
        print STDERR "ArcPlug: Adding classifications to infodb\n";
        $processor->process('classifications');
    }

    return 1;
    }

    # wasn't an archives directory, someone else will have to process it
    return 0;
}

# sort files so they're built (and displayed) in alphabetical order
sub alphabetically {
    open (A, &util::filename_cat($base_dir, $file, $a->[0])) || 
    die "couldn't open " . &util::filename_cat($base_dir, $file, $a->[0]) . "\n";
    open (B, &util::filename_cat($base_dir, $file, $b->[0])) || 
    die "couldn't open " . &util::filename_cat($base_dir, $file, $b->[0]) . "\n";
    my ($atitle, $btitle);
    my $line = "";
    while (defined ($line .= <A>)) {
    if ($line =~ /Title=\"([^\"]*)\"/) {$atitle = $1; last;}
    } 
    $line = "";
    while (defined ($line .= <B>)) {
    if ($line =~ /Title=\"([^\"]*)\"/) {$btitle = $1; last;}
    } 
    close A;
    close B;
    
    $atitle =~ tr/A-Z/a-z/;
    $atitle =~ s/[^a-z0-9 ]//g;
    $atitle =~ s/^\s+//;
    $atitle =~ s/^(the|a|an)\b//;
    $atitle =~ s/^\s+//;
    $atitle =~ s/^(.).*$/$1/;
    
    $btitle =~ tr/A-Z/a-z/;
    $btitle =~ s/[^a-z0-9 ]//g;
    $btitle =~ s/^\s+//;
    $btitle =~ s/^(the|a|an)\b//;
    $btitle =~ s/^\s+//;
    $btitle =~ s/^(.).*$/$1/;

    $atitle cmp $btitle;
}

1;