--- tamtam/tamtam2socialtext.pl	2007/12/12 18:14:19	15
+++ tamtam/tamtam2socialtext.pl	2007/12/13 11:38:36	20
@@ -10,8 +10,11 @@
 use Encode;
 use HTTP::Date;
 use POSIX qw/strftime/;
+use File::Slurp;
+use File::MMagic::XS;
 use Data::Dump qw/dump/;
 
+my $debug = 0;
 my $max = 999;
 
 my $page;
@@ -19,13 +22,23 @@
 
 my @page_names;
 
+print "Collecting pages...\n";
+
 find({
 	wanted => sub {
 		my $path = $File::Find::name;
 		return unless -f $path;
 
-		warn "# $path\n";
-		my $ref = XMLin( $path ) || die "can't open $path: $!";
+		warn "+ $path\n";
+		my $ref = XMLin( $path,
+			KeyAttr => {
+				'attachment' => '+name',
+				'meta' => 'name',
+			},
+			ForceArray => [ 'attachment', 'widget' ],
+		) || die "can't open $path: $!";
+
+		warn "## $path = ",dump( $ref ) if $debug;
 
 		my $name = $ref->{name} || die "no name in $path";
 
@@ -36,22 +49,43 @@
 			warn "SKIP: no LastModified in $path $name";
 			return;
 		}
-		my $data =
-			$ref->{widgets}->{widget}->{data} ||
-			$ref->{widgets}->{widget}->{Body}->{data} ||
-			die "no data in $path ",dump( $ref );
+
+		my $data;
+
+		foreach my $w ( @{ $ref->{widgets}->{widget} } ) {
+
+			warn "## w = ",dump( $w ) if $debug;
+
+			$data .= "\n----\n" if $data;
+			$data .= $w->{data} || die "no data?";
+		}
+
+		my $attachments;
+
+		if ( my $a = $ref->{attachment} ) {
+			foreach my $name ( keys %$a ) {
+				my $full_path = $path;
+				$full_path =~ s,pages/,attachments/,;
+				$full_path .= '.' . $name;
+				die "$full_path doesn't exist" unless -e $full_path;
+				push @$attachments, {
+					full_path => $full_path,
+					name => ( $name || $a->{$name}->{desc} || 'noname' ),
+				};
+			}
+		}
 
 		$page->{ $name } = {
 			content => convert_markup( $data ),
 			date => convert_date( $date ),
+			attachments => $attachments,
 		};
 
-		# strip path from page name
 		$name =~ s,^.+/([^/]+)$,$1,;
 		push @page_names, $name;
 
-#		warn dump( $ref );
 	},
+	no_chdir=>1,
 }, shift @ARGV || '.');
 
 my @pages = ( keys %$page );
@@ -64,8 +98,8 @@
 	username => 'tamtam',
 	password => 'import',
 	server   => 'http://saturn.ffzg.hr/',
+	workspace => 'razmjenavjestina',
 );
-$Rester->workspace('razmjenavjestina');
 $Rester->put_workspacetag('TamTam');
 
 sub convert_date {
@@ -119,15 +153,24 @@
 	$body =~ s,(\S+)----,$1\n----,gs;
 	$body =~ s,----(\S+),----\n$1,gs;
 
+	# attachments
+	$body =~ s,\[attachment:([^\]]+)(gif|png|jpg|jpeg)\],{image: $1$2},gis;
+	$body =~ s,\[attachment:([^\]]+)\],{file: $1},gs;
+
 	return $body;
 }
 
 my $count = 0;
 
+my $m = File::MMagic::XS->new;
+
 foreach my $name ( keys %$page ) {
 	last if $count++ == $max;
 
 	my $p = $page->{$name};
+
+	warn "## $name = ",dump( $p ) if $debug;
+
 	my $body = $p->{content} || die "no content?";
 	my $date = $p->{date} || die "no date?";
 
@@ -142,24 +185,28 @@
 	}
 
 	# link named pages
-	$body =~ s,$page_link_re,[$1],gs;
+	$body =~ s,\b$page_link_re\b,[$1],gs;
+	$body =~ s,``,,gs;
 
 	$body .= qq{
 ----
 
-Original sa http://www.razmjenavjestina.org/$full_name zadnja promjena {date: $date}
+"original"<http://www.razmjenavjestina.org/$full_name> {date: $date}
 };
 
 	Encode::_utf8_off( $body );
 
-	$Rester->put_page( $name, {
-		content => $body,
-		date => $date,
-	});
-	print "+ $name $date\n";
+	$Rester->put_page( $name, { content => $body, date => $date });
+	print "$name $date\n";
 	foreach ( @tags ) {
 		$Rester->put_pagetag( $name, $_ );
-		print "+ $name [$_]\n";
+		print "+ tag $_\n";
+	}
+	foreach my $a ( @{ $p->{attachments} } ) {
+		my $type = $m->get_mime( $a->{full_path} );
+		my $content = read_file( $a->{full_path} );
+		print "+ attachment ", $a->{name}," $type ", length($content), " bytes\n";
+		$Rester->post_attachment($name, $a->{name}, $content, $type );
 	}
 
 }