#!/usr/bin/perl -w
# based on a script by cowboyd, found at <http://www.cogentdude.com/node/view/11&e=7370>
# crudely rehacked by Gustaf Erikson <gerikson@gmail.com>

use XML::SAX::Base;
use XML::SAX::ParserFactory;
use Time::Local;

my $debug = 0;
# this hash maps current MT categories to blosxom categories/directories
my %cat_dirs = ( 'books' => 'books/read',
		 'comp' => 'comp',
		 'life' => 'alt',
		 'music' => 'alt',
		 'meta' => 'comm/weblog',
		 'mobile' => 'comm/mobile',
		 'taco' => 'comm/mobile',
		 'politics' => 'alt/politics',
		 'se.pol' => 'alt/politics',
		 'huset' => 'huset',);

# where should the files be written?
my $dest_dir = '/home/gustaf/projects/migrate/testdata';

package ConversionFilter;

@ISA = qw(XML::SAX::Base);


sub characters {
	my ($self,$data) = @_;
	$self->{_characters} .= $data->{Data};
}

sub start_element {	
	my ($self, $element) = @_; 
	my $tagname = $element->{LocalName};
	my $handle = "start_$tagname";
	if ($self->can($handle)) {
		$self->$handle($element);
	} 
	$self->SUPER::start_element($element);
}

sub end_element {
	my ($self, $element) = @_;
	my $tagname = $element->{LocalName};
	my $handle = "end_$tagname";
	if ($self->can($handle)) {
		$self->$handle($element);
	} 
	$self->SUPER::end_element();
}

sub start_item {
	my $self = shift;
	$self->{_current_item} = new Node();
    
}

sub end_item {
	my $self = shift;
#	print $self->{_current_item}->print_item()
	$self->{_current_item}->write_to_file();
}

sub start_description {
	my $self = shift;
	$self->clear_characters();
}

sub end_description {
	my $self = shift;
	$self->{_current_item}->{description} = $self->get_characters();
}

sub start_title {
	my $self = shift;
	$self->clear_characters();
}

sub end_title {
	my $self = shift;
	$self->{_current_item}->{title} = $self->get_characters();
}

sub start_date {
	my $self = shift;
	$self->clear_characters();
}

sub end_date {
	my $self = shift;
	$self->{_current_item}->{created} = $self->get_characters();
}

sub start_link {
    my $self = shift;
    $self->clear_characters();
}

sub end_link {
    my $self = shift;
    $self->{_current_item}->{link} = $self->get_characters();
}

sub start_subject {
    my $self = shift;
    $self->clear_characters();
}

sub end_subject {
    my $self = shift;
    $self->{_current_item}->{category} = $self->get_characters();
}

sub clear_characters {
	my $self = shift;
	$self->{_characters} = "";
}

sub get_characters {
	my $self = shift;
	return $self->{_characters};
}

package Node;

sub new {
	my $class = shift;
	return bless {}, $class;
}


sub print_item { # used for debugging
    my $self = shift;
    my $title = $self->{title};
    my $timestamp = $self->{created};
    my $link = $self->{link};
    my $body = $self->{description};
    my $cat = $self->{category};
    return "==> $title <$timestamp> ($cat)\n" . $body . "\n". "-- $link\n\n";
}

sub write_to_file {
    my $self = shift;
    my $title = $self->{title};
    my $timestamp = $self->{created};
    my $link = $self->{link};
    my $body = $self->{description};
    my $cat = $self->{category};
    my $filename;
    my $epoch_s = str2date($timestamp);
    if ( $title =~ m!<a href=.*>(.*)</a>! ) {
	$filename = normalize($1) . ".txt";
    } else {
	$filename = normalize($title) . ".txt";
    }
    if ( $debug ) {
	print "<!-- $title ($timestamp / $epoch_s) => $filename -->\n";
	print "$body\n";
    } else {
	my $dest = '';
	if ( exists($cat_dirs{$cat})) {
	    $dest = "$dest_dir/$cat_dirs{$cat}/$filename";
	    print "==> writing to $dest\n";
	} else {
	    $dest = "$dest_dir/misc/$filename";
	    print "==> category $cat doesn't have a directory associated with it\n==> writing to $dest/misc/\n";
	}
	open OUT, ">$dest" or die "can' open file '$dest' for writing: $!";
	print OUT "$title\n";
	print OUT "$body\n";
	close OUT;
	utime $epoch_s, $epoch_s, $dest;
    }
}
	
sub normalize {
    my $string = shift;
    if ( $string =~ m!<.*>(.*)</em>! ) {
	$string = $1;
    }
    $string =~ s/[,\'\":!]//g;
    $string =~ s/[\ ]/-/g;
    $string =~ s/[هن]/a/g;
    $string =~ s/ِ/o/g;
  
    return $string;
}

sub str2date { 
    my $string = shift;
    if ( $string =~ m!(\d{4})-(\d{2})-(\d{2})T(\d{2}):(\d{2}):(\d{2})\+(\d{2}:\d{2})! ) {
	my ($yyyy, $mm, $dd, $hh, $mi, $ss, $tz) = ($1, $2, $3, $4, $5, $6, $7);
#	my $timestamp = "$yyyy-$mm-$dd $hh:$mi:$ss";
	my $epoch_s = Time::Local::timelocal( $ss, $mi, $hh, $dd, $mm - 1, $yyyy - 1900 );
	return ( $epoch_s );
    } else {
	return (  undef );
    }
}




sub mysql_escape {
	my $string = shift;
	$string =~ s/\n/\\n/mg;
	$string =~ s/(\'|\")/\\$1/g;
	return $string;
}

sub to_mysql_date {
	my $string = shift;
	$string =~ s/T/ /;
	$string =~ s/\+00:00$//;
	return $string;
}

package main;

my $filename = $ARGV[0] or die "usage: $0 filename\n";

# create dirs


foreach my $cat ( keys %cat_dirs ) {

    if ( ! -d "$dest_dir/$cat_dirs{$cat}" ) {
 	print "creating directory $dest_dir/$cat_dirs{$cat}\n";
	system( "mkdir -p $dest_dir/$cat_dirs{$cat}" );
	if ($? == -1) {
	    print "failed to execute: $!\n";
	}
	elsif ($? & 127) {
	    printf "child died with signal %d, %s coredump\n",
	    ($? & 127),  ($? & 128) ? 'with' : 'without';
	}
	else {
#	    printf "child exited with value %d\n", $? >> 8;
	}

    }
}

if ( ! -d "$dest_dir/misc" ) {
    print "creating directory $dest_dir/misc\n";
 	mkdir("$dest_dir/misc") or die "couldn't create directory $dest_dir/misc: $!\n";
}

exit 0 if $debug;

my $handler = new ConversionFilter();

my $parser = new XML::SAX::ParserFactory->parser(Handler => $handler);

$parser->parse_uri($filename);