#!/usr/bin/perl -w
#
# unpack a japanese zipfile, converting sjis filenames to unicode on the way
#
use lib $ENV{'HOME'} . "/src/perl";
use Archive::Zip qw( :ERROR_CODES :CONSTANTS );
use Unicode::Japanese;
use File::Path;
use Data::Dumper;


do "jcode.pl";
jcode::init();


while ( @ARGV ) {
  my $zipfile = new Archive::Zip;
  my $err = $zipfile->read( $ARGV[0] );
  if ( $err != AZ_OK ) {
	die "Failed: $err $!\n";
  } else {
	my @members = $zipfile->members();
#	print scalar( @members ) . " items in zipfile\n";
	for my $member ( @members ) {
	  my $filename = $member->fileName;
	  my @coding = jcode::getcode( \$filename );
	  my $coding = ( $coding[1] || "" );

#	  if ( @coding ) {
#		print "Coding is " . ( $coding[1] || "unknown") . "\n";
#	  }

	  print $filename . "\n";

	  # Now convert filename to sjis
	  if ( $coding and $coding ne "sjis" ) {
		$filename = jcode::sjis( $filename, $coding );
		print "sjis filename = $filename\n";
		$coding = "sjis";
	  }

	  # and convert sjis to unicode
	  if ( $coding ) {
		print "SJIS: " . get_in_hex_bytes( $filename ) . "\n";
		$filename = Unicode::Japanese->new( $filename, 'sjis' )->get;
		print "UTF8: " . get_in_hex_bytes( $filename ) . "\n";
	  }

	  # next, make directories
#	  my ( $dir ) = $filename =~ m@(.*)/.*?@; #@?
#	  mkpath( $dir, 0, 0755 ) if $dir;
#	  open( FILE, ">$filename" );
#	  binmode( FILE );
#	  
#	  close( FILE );
	  $zipfile->extractMember( $member, $filename );
	}
  }
  shift;
}


sub get_in_hex_bytes {
    my(@bytes) = split(//, shift);
    map { $_ = sprintf("%02x", ord($_)) } @bytes;
    return join(" ", @bytes);
}

__DATA__
SJIS:
83 51 = 0x8351	0x2532	0x30B2	# KATAKANA LETTER GE
83 43 = 0x8343	0x2524	0x30A4	# KATAKANA LETTER I
83 69 = 0x8369	0x254A	0x30CA	# KATAKANA LETTER NA
81 5b = 0x815B	0x213C	0x30FC	# KATAKANA-HIRAGANA PROLONGED SOUND MARK
2f 54 68 75 6d 62 73 2e 64 62       # /Thumbs.db
UTF8:
e3 82 b2 e3 82 a4 e3 83 8a e3 83 bc 2f 54 68 75 6d 62 73 2e 64 62