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

package bnParser;
use HTML::Parser;
use vars qw(@ISA $VERSION @EXPORT @EXPORT_OK %EXPORT_TAGS);

@ISA = ('HTML::Parser');
sub new {
    my $type = shift;
    my $class = ref($type) || $type;
    my $self = bless(HTML::Parser->new, $class);

    push @EXPORT, '&getbookinfo';
    
    my @authors;
    $self->{'authors'} = \@authors;
    $self->{'state'} = '';
    $self->{'tagtext'} = '';
    return $self;
}

sub start($tag,$attr, $attrseq, $origtext)
{
    my ($self,$tag,$attr, $attrseq, $origtext) = @_;
    if ($tag eq 'a')
    {
	$self->{'tagtext'} = '';
    }
    if ($tag eq 'br')
    {
	if ($self->{'state'} eq 'getauthor')
	{
	    my $authors = $self->{'authors'};
	    my $author = $self->{'tagtext'};
	    $author =~ s/^\s+//g;
	    $author =~ s/\s+$//g;
	    push @$authors, $author;
	}
	if ($self->{'state'} eq 'gettitle')
	{
	    $self->{'title'} = $self->{'tagtext'};
	}
	$self->{'tagtext'} = '';
	$self->{'state'} = '';
    }
}
sub end($tag, $origtext)
{
    my ($self,$tag, $origtext) = @_;

    if ($tag eq 'a')
    {
	if ($self->{'state'} eq 'getauthor')
	{
	    my $authors = $self->{'authors'};
	    my $author = $self->{'tagtext'};
	    $author =~ s/^\s+//g;
	    $author =~ s/\s+$//g;
	    push @$authors, $author;
	}
	$self->{'tagtext'} = '';
	$self->{'state'} = '';
    }
}
sub text($text)
{
    my ($self,$text) = @_;
    $self->{'tagtext'} .= $text;
}

sub comment($comment)
{
    my ($self,$comment) = @_;

    if ($comment =~ /^\s*title\s*$/)
    {
	$self->{'state'} = 'gettitle';
    }
    if ($comment =~ /^\s*author\s*$/)
    {
	$self->{'state'} = 'getauthor';
    }
    $self->{'tagtext'} = '';
}
sub getbookinfo()
{
    my ($self) = @_;
    my $authors;
    $authors = $self->{'authors'};
    $self->{'title'} =~ s/\&nbsp\;/ /g;
    $self->{'title'} =~ s/^\s+//g;
    $self->{'title'} =~ s/\s+$//g;
    $self->{'title'} =~ s/\s+/ /g;
    return ($self->{'title'}, @$authors);
}



package amazonParser;
use HTML::Parser;
use vars qw(@ISA $VERSION @EXPORT @EXPORT_OK %EXPORT_TAGS);

@ISA = ('HTML::Parser');
sub new {
    my $type = shift;
    my $class = ref($type) || $type;
    my $self = bless(HTML::Parser->new, $class);

    push @EXPORT, '&getbookinfo';
    
    my @authors;
    $self->{'authors'} = \@authors;
    $self->{'state'} = '';
    $self->{'tagtext'} = '';
    return $self;
}

sub start($tag,$attr, $attrseq, $origtext)
{
    my ($self,$tag,$attr, $attrseq, $origtext) = @_;

    if ($tag eq 'form')
    {
	if ($attr->{'action'} =~ 
	    /^\/exec\/obidos\/handle\-buy\-box\=[0-9]+\/book-glance/)
	{
	    $self->{'state'} = 'gettitle';
	    $self->{'tagtext'} = '';
	}
    }
    if (($tag eq 'a') && defined($attr->{'href'}))
    {
	if ($attr->{'href'} =~
	    /^\/exec\/obidos\/ts\/book-customer-review-form\//)
	{
	    $self->{'pastreviewsection'} = 1;
	}

	if (($attr->{'href'} =~
	    /^\/exec\/obidos\/Author\=/)
	    && !defined($self->{'pastreviewsection'}))
	{
	    $self->{'state'} = 'getauthor';
	    $self->{'tagtext'} = '';
	}
    }
    if ($tag eq 'b')
    {
	$self->{'tagtext'} = '';
    }
}
sub end($tag, $origtext)
{
    my ($self,$tag, $origtext) = @_;

    if (($tag eq 'b') || ($tag eq 'a'))
    {
	if ($self->{'state'} eq 'gettitle')
	{
	    $self->{'title'} = $self->{'tagtext'};
	    $self->{'state'} = '';
	}
	if (($self->{'state'} eq 'getauthor') && defined($self->{'title'}))
	{
	    my $authors = $self->{'authors'};
	    push @$authors, $self->{'tagtext'};
	    $self->{'state'} = '';
	}
    }
}
sub text($text)
{
    my ($self,$text) = @_;
    $self->{'tagtext'} .= $text;
}

sub comment($comment)
{
    my ($self,$comment) = @_;
}
sub getbookinfo()
{
    my ($self) = @_;
    my $authors;
    $authors = $self->{'authors'};
    return ($self->{'title'}, @$authors);
}



package bordersParser;
use HTML::Parser;
use vars qw(@ISA $VERSION @EXPORT @EXPORT_OK %EXPORT_TAGS);

@ISA = ('HTML::Parser');
sub new {
    my $type = shift;
    my $class = ref($type) || $type;
    my $self = bless(HTML::Parser->new, $class);

    push @EXPORT, '&getbookinfo';
    
    my @authors;
    $self->{'authors'} = \@authors;
    $self->{'state'} = '';
    $self->{'tagtext'} = '';
    return $self;
}

sub start($tag,$attr, $attrseq, $origtext)
{
    my ($self,$tag,$attr, $attrseq, $origtext) = @_;

    if ($tag eq 'big')
    {
	$self->{'tagtext'} = '';
    }
    if ($tag eq 'b')
    {
	$self->{'tagtext'} = '';
    }
    if ($tag eq 'br')
    {
	if ($self->{'state'} eq 'geteditor')
	{
	    my $authors = $self->{'authors'};
	    my ($author, @authlist);
	    if (!(@authlist = split /\//, $self->{'tagtext'}))
	    {
		push(@authlist, $self->{'tagtext'});
	    }
	    foreach $author (@authlist)
	    {
		$author =~ s/\&nbsp\;/ /g;
		$author =~ s/^\s+//;
		$author =~ s/\s+$//;
		$author =~ s/\s+/ /;
		push @$authors, $author;
	    }
	}
	if ($self->{'state'} eq 'getauthor')
	{
	    my $authors = $self->{'authors'};
	    my ($author, @authlist);
	    if (!(@authlist = split /\//, $self->{'tagtext'}))
	    {
		push(@authlist, $self->{'tagtext'});
	    }
	    foreach $author (@authlist)
	    {
		$author =~ s/\&nbsp\;/ /g;
		$author =~ s/^\s+//;
		$author =~ s/\s+$//;
		$author =~ s/\s+/ /;
		push @$authors, $author;
	    }
	}
	$self->{'tagtext'} = '';
	$self->{'state'} = '';
    }
}
sub end($tag, $origtext)
{
    my ($self,$tag, $origtext) = @_;

    if ($tag eq 'big')
    {
	$self->{'title'} = $self->{'tagtext'};
    }
    if ($tag eq 'b')
    {
	if ($self->{'tagtext'} =~ /^Author.*\:/)
	{
	    $self->{'state'} = 'getauthor';
	    $self->{'tagtext'} = '';
	}
	if ($self->{'tagtext'} =~ /^Editor.*\:/)
	{
	    $self->{'state'} = 'geteditor';
	    $self->{'tagtext'} = '';
	}
    }
}
sub text($text)
{
    my ($self,$text) = @_;
    $self->{'tagtext'} .= $text;
}

sub comment($comment)
{
    my ($self,$comment) = @_;
}
sub getbookinfo()
{
    my ($self) = @_;
    my $authors;
    $authors = $self->{'authors'};
    return ($self->{'title'}, @$authors);
}


package main;

use LWP::UserAgent;

my %bookstores = (
	'borders',
	'http://search.borders.com/fcgi-bin/db2www/search/search.d2w/Details?code=%s&mediaType=Book&searchType=ISBNUPC&prodID=',

	'amazon', 
	'http://www.amazon.com/exec/obidos/ISBN=%s',
	'bn', 
	'http://shop.barnesandnoble.com/bookSearch/isbnInquiry.asp?isbn=%s',
);

my $ua;
$ua = new LWP::UserAgent;
$ua->agent("Flutterby.ComLibraryInfo/0.1 ".$ua->agent);

open I, "isbn.3";

while (<I>)
{
    chop;
    my $isbn = $_;
    my $bookstore;

    foreach $bookstore (keys %bookstores)
    {
	my ($p, $t);
	$p = eval $bookstore.'Parser->new()';
	print "Trying $bookstore for $isbn\n";
	my $url;
	$url = sprintf $bookstores{$bookstore}, $isbn;
	my $req = new HTTP::Request GET => $url;
	my $res = $ua->request($req);


	$t = $res->as_string;
	$p->parse($t);

	my ($title, @authors, $author);
	($title, @authors) = $p->getbookinfo();
	print "Title: $title\n";
	print " authors @authors\n";
    }
}