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

use WWW::Mechanize;
use HTML::TreeBuilder;

my ($propid, $address, $price, $header, $summary);
my ($status, $description, $agent);
my ($officephone, $officefax, $proptype, $bed, $bath);
my ($land, $carnumport, $carnumgar, $municipality);
my ($title);

#my $mech = WWW::Mechanize->new( autocheck => 1 );
#my $url = "http://www.realestate.com.au/cgi-bin/rsearch?a=o&id=106023887";
#$mech->get($url);

# Use local copy of page
my $file_copy = $ENV{HOME}."/rsearch.html";

#my $root = HTML::TreeBuilder->new_from_content($mech->{content});
my $root = HTML::TreeBuilder->new_from_file($file_copy);

my $titleTag = $root->look_down('_tag', 'title');
$title = $titleTag ? $titleTag->as_trimmed_text() : "";



my $propertyDetail = $root->look_down('_tag', 'div', 'id', 'propertyDetail');
die("Cannot find propertyDetail") if !defined $propertyDetail;

my $detailsHeader = $propertyDetail->look_down('_tag', 'div', 'class', qr{\bdetailsHeader\b});
die("Cannot find detailsHeader") if !defined $detailsHeader;

my $propertyID = $detailsHeader->look_down('_tag', 'p', 'class', qr{\bpropertyID\b});
die("Cannot find propertyID") if !defined $propertyID;
$propid = $propertyID->as_trimmed_text();

my $addrH1 = $detailsHeader->look_down('_tag', 'h1');
die("Cannot find addrH1") if !defined $addrH1;
$address = $addrH1->as_trimmed_text();

my $priceStrong = $detailsHeader->look_down('_tag', 'strong', 'class', qr{\bprice\b});
die("Cannot find priceStrong") if !defined $priceStrong;
$price = $priceStrong->as_trimmed_text();



my $textualDiv = $propertyDetail->look_down('_tag', 'div', 'class', qr{\btextual\b});
die("Cannot find textualDiv") if !defined $textualDiv;

my $sizeH2 = $textualDiv->look_down('_tag', 'h2');
die("Cannot find sizeH2") if !defined $sizeH2;
$header = $sizeH2->as_trimmed_text();

my $propSumH2 = $textualDiv->look_down('_tag', 'h2', 'class', qr{\bpropertySummary\b});
die("Cannot find propSumH2") if !defined $propSumH2;
$summary = $propSumH2->as_trimmed_text();

# contract/offer might be missing
my $offerH3 = $textualDiv->look_down('_tag', 'h3', 'class', qr{\bhighlighted\b});
$status = $offerH3 ? $offerH3->as_trimmed_text() : "";

my $descriptionDiv = $textualDiv->look_down('_tag', 'div', 'class', qr{\bdescription\b});
die("Cannot find descriptionDiv") if !defined $descriptionDiv;
$description = $descriptionDiv->as_trimmed_text();



my $propertySummary = $propertyDetail->look_down('_tag', 'div', 'id', 'propertySummary');
die("Cannot find propertySummary") if !defined $propertySummary;

# Property summary in a definition list, no id
my $propdl = $propertySummary->look_down('_tag', 'dl');
die("Cannot find propdl") if !defined $propdl;

# Collect all dt/dd in definition list
my @propdtdd = $propdl->look_down('_tag', qr{\b(?:dt|dd)\b});
die("Empty property details list") if scalar(@propdtdd) == 0;
my %prop;
my $lastdt = "";
foreach (@propdtdd)
{
    next if !ref;
    my $thingy = $_;
    if ($thingy->attr('_tag') eq "dt")
    {
        $lastdt = $thingy->as_trimmed_text();
        $lastdt =~ s/\s*:$//;  # strip trailing colon
    }
    elsif ($thingy->attr('_tag') eq "dd" && $lastdt)
    {
        $prop{lc($lastdt)} = $thingy->as_trimmed_text();
        $lastdt = "";
    }
}

$proptype     = $prop{category}     || "";
$bed          = $prop{bedrooms}     || "";
$bath         = $prop{bathrooms}    || "";
$land         = $prop{land}         || "";
$carnumport   = $prop{carport}      || "";
$carnumgar    = $prop{garage}       || "";
$municipality = $prop{municipality} || "";




my $contactAgentDetails = $root->look_down('_tag', 'div', 'id', 'contactAgentDetails');
die("Cannot find contactAgentDetails") if !defined $contactAgentDetails;

# First paragraph has name
my $firstp = $contactAgentDetails->look_down('_tag', 'p');
die("Cannot find firstp") if !defined $firstp;
$agent = $firstp->as_trimmed_text();

my $officePhone = $contactAgentDetails->look_down('_tag', 'p', 'class', qr{\bofficePhone\b});
die("Cannot find officePhone") if !defined $officePhone;
$officephone = $officePhone->as_trimmed_text();

my $officeFax = $contactAgentDetails->look_down('_tag', 'p', 'class', qr{\bofficeFax\b});
die("Cannot find officeFax") if !defined $officeFax;
$officefax = $officeFax->as_trimmed_text();


$root->delete();

print_all();

sub print_all
{
print "PropertyID: $propid \n";
print "Status: $status \n";
print "Title: $title \n";
print "Address: $address \n";
print "Price: $price \n";
print "Type: $proptype \n";
print "Bedrooms: $bed \n";
print "Bathrooms: $bath \n";
print "Carport: $carnumport \n";
print "Garage: $carnumgar \n";
print "Header: $header \n";
print "Summary: $summary \n";
print "Description: $description \n";
print "Agent Details: \n";
print "Agent: $agent \n";
print "OfficePhone: $officephone \n";
print "OfficeFax: $officefax \n";
}