MAME db scraper

mame_db.pl

It is a nice scraper to create a MAME game info database compatible with rcb XBMC add-on

$dir_main should exists and be RW (place where database will be created)

./mame_db.pl ROM_NAME

IE: ./mame_db.pl punisher

#!/usr/bin/perl -w

# Directory where the local database will be created:

$dir_main = "/games";

# Check arument:

my $game = {};

$game{filename} = $ARGV[0] or die("rom filename without extension is needed");

# Require CPAN modules:

require HTML::TreeBuilder;

require LWP::UserAgent;

# CHECK WRITE PERMISIONS ON DIRECTORIES TO SAVE

$game{info_dir} = "info";

( -w "$dir_main/$game{info_dir}" or mkdir("$dir_main/$game{info_dir}") ) or die("Directory: $dir_main/$game{info_dir} -> $!");

my $game_img = {};

$game_img{title}{dir} = "titles";

$game_img{snap}{dir} = "snap"; # TIP: Configure xmamerc to get your own snaps.

$game_img{gameover}{dir} = "gameover";

$game_img{score}{dir} = "scores";

$game_img{select}{dir} = "select";

$game_img{cabinet}{dir} = "cabinet";

$game_img{marquee}{dir} = "marquee";

$game_img{cpanel}{dir} = "cpanel";

$game_img{flyer}{dir} = "flyer";

$game_img{pcb}{dir} = "pcb";

foreach ( keys %game_img ) {

( -w "$dir_main/$game_img{$_}{dir}" or mkdir("$dir_main/$game_img{$_}{dir}") ) or die("Directory: $dir_main/$game_img{$_}{dir} -> $!");

}

# LWP browser rulez!:

my $ua = LWP::UserAgent->new;

$ua->timeout(10);

$ua->env_proxy;

$ua->agent('Mozilla/5.0');

sub get_minimaws {

# Database information:

my $info_db = "http://maws.mameworld.info/minimaws/set/$game{filename}";

$game{url} = $info_db;

my $info_str_ok = '\w\ \d\.\,\:\-\\\/';

my $info_raw = $ua->get($info_db);

$info_tree = HTML::TreeBuilder->new_from_content($info_raw->content);

foreach ( $info_tree->look_down('class', 'section') ) {

if ( $_->is_inside('h1') ) {

$_->objectify_text();

my @buffer = $_->look_down('_tag', '~text');

($game{title}, $game{region}) = split(/[\(\)\x{00A9}]/,$buffer[0]->attr_get_i('text'));

$game{year} = $buffer[1]->attr_get_i('text');

$game{developer} = $buffer[3]->attr_get_i('text');

foreach ($_->right()->look_down('_tag', 'li')) {

if ( $_->as_text =~ /genre\:\ ([$info_str_ok]*)/s) {

$game{genre} = $1;

}

elsif ( $_->as_text =~ /category\:\ ([$info_str_ok]*)/s) {

$game{category} = $1;

}

elsif ( $_->as_text =~ /gallery\:/) {

foreach ($_->look_down('_tag', 'a')) {

foreach my $image (keys %game_img) {

if ( $_->attr_get_i('href') =~ /$image/ ) {

$game_img{$image}{url} = $_->attr_get_i('href');

}

else {

}

}

}

}

elsif ( $_->as_text =~ /number\ of\ players\:\ ([$info_str_ok]*)/s) {

$game{nop} = $1;

}

elsif ( $_->as_text =~ /rating\:\ ([$info_str_ok]{2})/s) {

$game{rating} = $1;

}

else {

}

}

}

elsif ( $_->is_inside('h2') ) {

if ( $_->as_text =~ /emulation/ ) {

}

elsif ( $_->as_text =~ /display/ ) {

}

elsif ( $_->as_text =~ /input/ ) {

foreach ($_->right()->look_down('_tag', 'li')) {

if ( $_->as_text =~ /controls\:\ ([$info_str_ok]*)/) {

$game{controls} = $1;

}

if ( $_->as_text =~ /players\:\ ([$info_str_ok]*)/) {

$game{players} = $1;

}

else {

}

}

}

elsif ( $_->as_text =~ /hardware/ ) {

}

elsif ( $_->as_text =~ /data/ ) {

foreach ($_->right()->look_down('_tag', 'ol')) {

foreach ($_->look_down('_tag', 'li')) {

my ( $rom_name, $rom_size, $rom_CRC, $rom_set, $rom_status ) = split ( '/', $_->as_text );

$game{crc} .= $rom_CRC;

}

}

}

elsif ( $_->as_text =~ /links/ ) {

}

elsif ( $_->as_text =~ /MAMEInfo.dat/ ) {

$_->right()->objectify_text();

foreach ($_->right()->look_down('_tag', '~text')) {

$game{MAMEInfo} .= $_->attr_get_i('text')."\n";

}

}

elsif ( $_->as_text =~ /arcade-history.com/ ) {

$_->right()->objectify_text();

foreach ($_->right()->look_down('_tag', '~text')) {

$game{arcadehistory} .= $_->attr_get_i('text')."\n";

}

}

else {

}

}

else {

}

}

}

get_minimaws();

my $info_file = $dir_main."/".$game{info_dir}."/".$game{filename}.".txt";

open (INFO_FILE, ">$info_file");

# http://code.google.com/p/romcollectionbrowser/wiki/Main#parserConfig.xml

print INFO_FILE "Game: $game{title}

Platform:

Region: $game{region}

Media:

Controller: $game{controls}

Genre: $game{genre}

Release Year: $game{year}

Developer: $game{developer}

Publisher: $game{developer}

Players: $game{players}

URL: $game{url}

Description:

$game{arcadehistory}

********************************************************************\n";

close (INFO_FILE);

foreach ( keys %game_img ) {

if ( $game_img{$_}{url} ) {

print "GETTING: http://maws.mameworld.info".$game_img{$_}{url}."\n";

my $img_src = "http://maws.mameworld.info".$game_img{$_}{url};

my $img_dst = $dir_main."/".$game_img{$_}{dir}."/".$game{filename}.".png";

$ua->mirror($img_src, $img_dst);

}

}