WWW::Search::Profusion - class for searching Profusion.com!

    require WWW::Search;
    $search = new WWW::Search('Profusion');

# Profusion.pm
# by Jim Smyser
# Copyright (c) 1999 by Jim Smyser & USC/ISI
# $Id: Profusion.pm,v 2.22 2008/02/01 02:50:27 Daddy Exp $

package WWW::Search::Profusion;

use strict;
use warnings;

use base 'WWW::Search';

our
$VERSION = do { my @r = ( q$Revision: 2.22 $ =~ /\d+/g ); sprintf "%d." . "%03d" x $#r, @r };

my $MAINTAINER = 'Jim Smyser <jsmyser@bigfoot.com>';
my $TEST_CASES = <<"ENDTESTCASES";
&test('Profusion', '$MAINTAINER', 'zero', \$bogus_query, \$TEST_EXACTLY);
&test('Profusion', '$MAINTAINER', 'one', 'astronomy', \$TEST_GREATER_THAN, 10);
ENDTESTCASES

use Carp ();
use WWW::Search(generic_option);
require WWW::SearchResult;

# private
sub native_setup_search
{
    my($self, $native_query, $native_options_ref) = @_;
    $self->{agent_e_mail} = 'jsmyser@bigfoot.com';
    $self->user_agent('user');
    $self->{_next_to_retrieve} = 0;
    if (!defined($self->{_options})) {
    $self->{_options} = {
     'search_url' => 'http://www.profusion.com/cgi-bin/nph-ProFusion.pl',
     'option' => 'default', # see USAGE OPTIONS
     'display' => 'all&summary=yes&totalverify=0&auto=all&engine1=AltaVista&engine4=InfoSeek&engine5=LookSmart&engine2=Excite&engine8=Magellan&engine6=WebCrawler&engine9=GoTo&engine3=Google&engine7=Yahoo&search=web&log=yes&current=0&customEngines=1%2C2%2C3%2C4%2C5%2C6%2C7%2C8%2C9&pid=profusion',

      };
        };
    my($options_ref) = $self->{_options};
    if (defined($native_options_ref)) {
    # Copy in new options.
    foreach (keys %$native_options_ref) {
        $options_ref->{$_} = $native_options_ref->{$_};
    };
    };
    # Process the options.
    my($options) = '';
    foreach (keys %$options_ref) {
    # printf STDERR "option: $_ is " . $options_ref->{$_} . "\n";
    next if (generic_option($_));
    $options .= $_ . '=' . $options_ref->{$_} . '&';
    };
    $self->{_debug} = $options_ref->{'search_debug'};
    $self->{_debug} = 2 if ($options_ref->{'search_parse_debug'});
    $self->{_debug} = 0 if (!defined($self->{_debug}));
    # Finally figure out the url.
    $self->{_base_url} = 
    $self->{_next_url} =
    $self->{_options}{'search_url'} .
    "?" . $options .
    "queryterm=" . $native_query;
    print $self->{_base_url} . "\n" if ($self->{_debug});
}

# private
sub native_retrieve_some
{
    my ($self) = @_;

    # fast exit if already done
    return undef if (!defined($self->{_next_url}));

    # get some
    print STDERR "**Fetching some....\n" if 2 <= $self->{_debug};
    my($response) = $self->http_request('GET', $self->{_next_url});
    $self->{response} = $response;
    if (!$response->is_success) {
    return undef;
    };
    # parse the output
    my($HEADER, $HITS, $DESC) = (1..10);
    my($hits_found) = 0;
    my($state) = ($HEADER);
    my($hit, $raw, $title, $url, $desc) = ();
   foreach ($self->split_lines($response->content())) {
        next if m@^$@; # short circuit for blank lines
    if ($state == $HEADER && m@<td valign="top"><br>@i) { 
        print STDERR "PARSE(HEADER->HITS-1): $_\n" if ($self->{_debug} >= 2);
        $state = $HITS;                                                                                      

 } elsif ($state == $HITS && m@.*?<a href="/cgi-bin/process_result.pl.?url==([^"]+)\&engine.*?>(.*)</a>&nbsp;.*?(<i>(.*))</font><br>@i) { 
        print STDERR "**Parsing URL, Title & Desc...\n" if 2 <= $self->{_debug};
        my ($url, $title, $description) = ($1,$2,$3);
        my($hit) = new WWW::SearchResult;
        $title =~ s/<b>//g;
        $hit->add_url($url);
        $hit->title($title);
        $hit->description($description);
        $hit->raw($_);
        $hits_found++;
        push(@{$self->{cache}}, $hit);
        $state = $HITS;
#  Ignoring 'next' page url's since I am not defining them. 
    print STDERR "**All done!**\n" if 2 <= $self->{_debug};
    };
    if (defined($hit)) {
        push(@{$self->{cache}}, $hit);
    };
    $self->{_next_url} = undef;
    };
    return $hits_found;
}

1;

__END__