package VROOM::Search::Google;

use strict;

use VROOM::Search qw(escape_query unescape_sequence);
use Time::HiRes   qw(gettimeofday);

@VROOM::Search::Google::ISA = qw(VROOM::Search);

sub prepare_request
{
    my $self   = shift;
    my $query  = escape_query(shift);
    my $params = shift;
    my $uri    = 'http://www.google.com';

    $params->{baseurl} = $uri unless defined $params->{baseurl};
    $params->{hl}      = 'en' unless defined $params->{hl};

    $self->{baseurl} = $uri = $params->{baseurl};

    $uri .= '/search?q='.$query;

    while (my ($name, $value) = each %$params) {
    next if $name =~ /baseurl/;
    $uri .= '&'.$name.'='.$value;
    }

    $self->{initime} = $self->{endtime} = [gettimeofday];

    $self->{request} = new HTTP::Request(GET => $uri);
}

sub store_results
{
    my $self = shift;
    my $res  = shift;

    $self->{endtime} = [gettimeofday];

    if ($res->code != 200) {
    $self->{request} = undef;
    return undef;
    }

    #
    # Google doesn't return Content-Length, 
    # so ($res->headers)->content_length will be zero. We're forced to
    # use Perl function - length.
    #
    $self->{fetch}++;
    $self->{pgsize} += length($res->content);

    #
    # If we reach here, HTTP response is OK. Proceed to parse the html
    # document for search results
    #
    my ($HIT, $ENTRY, $NEXT) = (0, 1, 2);
    my $rank   = $self->count;
    my $hits   = 0;
    my $wish   = $HIT;
    my $result = undef;
    
    foreach (split(/(<p>|\n|<\/div>)/i, $res->content)) {
        next if /^$/;            # short circuit for blank lines
    last if $wish == $NEXT;
    
    if ($self->count == $self->maximum) {
        $self->{request} = undef;
        return $hits;
    }

    #print "#################################################\n";
    #print $_, "\n";

    #
    # Ah,found some results. Get approximate results and wish to 
        # see the title/url of the first result.
    # 
    if ($wish == $HIT && /Results.*?of.*?([0-9,]+).*?\./i) {
        my $count = $1;
        $self->approximate($count);
        $wish = $ENTRY;
    }
    #
    # Extract the url/title and wish to have abstract text
    #
    elsif ($wish == $ENTRY && 
           /^<a href=(.*?)>(.*?)<\/a><br><font.*?>(.*?)$/i) {
        
        my $url      = $1;
        my $title    = $2;
        my $abstract = $3;

        $url    =~ s/%([0-9A-Fa-f]{2})/chr(hex($1))/eg;
        $url    =~ s/(^http:\/\/|\/(index.htm|index.html)*$)//g;
        $title  =~ s/<.*?>//g;
 
        $result = new VROOM::Search::Result;
        $result->url($url);
        $result->title(unescape_sequence($title));
        $result->text(unescape_sequence($abstract));
        $result->rank(++$rank);
        $result->engine('Google');

        $self->add_result($result);
        $self->{pool}->insert($result) if $self->{pool};
        $hits++;
        }
    #
    # Extract the url for the next page
    #
    elsif ($wish == $ENTRY && 
           /<td nowrap><a href=(.*?)>.*?<span.*?>Next<\/span><\/a>/i) 
+{

        $self->{request}->uri($self->{baseurl}.$1);

        $wish = $NEXT;
    }
    }

    #
    # This is important. It signals the search agent not to fetch more
+ pages.
    #
    $self->{request} = undef if $wish != $NEXT;


    return $hits;
    
}

1;

__END__
[download]

#! Perl\bin\perl -w

use VROOM::Search::Google;

open FILE1, "> sample1.txt" or die "$!";

my $oSearch = new VROOM::Search::Google( );

my $sQuery = VROOM::Search::Google::escape_query('"telefonos" "mundial
+"');

$oSearch->native_query($sQuery);
  
while ( my $oResult = $oSearch->next_result() ) { 
    
    print "Adding: ", $oResult->url, "\n";
    print FILE1 $oResult->url, "\n";
   
    }
print ref($oSearch);
[download]