#!/usr/bin/perl -w
#
# jpginfo.pl version 0.22 - describe JPEG infomation (just experimental)
# copyright (c) hanawa <y@hnw.jp>
#
#
# The Adobe Photoshop format which begins with '8BPS' is not supported
# (I don't know the details of the format...)

# verbose level
#  0: print incorrect JPEG only
#  1: print information with 1 line
#  2: print information including all marker or other info

# sample usage:
#
# 1. You can move all regular JPEG files to ../jpeg. Such as:
#   $ mkdir ../jpeg
#   $ jpginfo * |perl -ne 'if (/^(.*): *\d+ x \d+$/){rename($1,"../jpeg/$1")}'



my ($vernum) = ("0.24");
my ($my_name);
my ($arcfile, $action, @files, $buf, $field_len);
my (%symbol);

my ($verbose, $debug);

$verbose = 1;
$debug = 0;

$/ = "\xff";

$my_name = $0;
$my_name =~ s|^.*/||;

$action = "info";

%symbol =
    (
     0xc0, "SOF0",		# Baseline DCT
     0xc1, "SOF1",		# Extended sequential DCT, Huffman coding
     0xc2, "SOF2",		# Progressive DCT, Huffman coding
     0xc3, "SOF3",		# Lossless (sequential), Huffman coding
     0xc5, "SOF5",
     0xc6, "SOF6",
     0xc7, "SOF7",
     0xc8, "JPG",
     0xc9, "SOF9",		# Extended sequential DCT, arithmetic coding
     0xca, "SOF10",		# Progressive DCT, arithmetic coding
     0xcb, "SOF11",		# Lossless (sequential), arithmetic coding
     0xcd, "SOF13",
     0xce, "SOF14",
     0xcf, "SOF15",
     0xc4, "DHT",
     0xcc, "DAC",
     0xd8, "SOI*",
     0xd9, "EOI*",
     0xda, "SOS",
     0xdb, "DQT",
     0xdc, "DNL",
     0xdd, "DRI",
     0xde, "DHP",
     0xdf, "EXP",

     0xe0, "APP0",
     0xe1, "APP1",
     0xe2, "APP2",
     0xe3, "APP3",
     0xe4, "APP4",
     0xe5, "APP5",
     0xe6, "APP6",
     0xe7, "APP7",
     0xe8, "APP8",
     0xe9, "APP9",
     0xea, "APP10",
     0xeb, "APP11",
     0xec, "APP12",
     0xed, "APP13",
     0xee, "APP14",
     0xef, "APP15",

     0xf0, "JPG0",
     0xf1, "JPG1",
     0xf2, "JPG2",
     0xf3, "JPG3",
     0xf4, "JPG4",
     0xf5, "JPG5",
     0xf6, "JPG6",
     0xf7, "JPG7",
     0xf8, "JPG8",
     0xf9, "JPG9",
     0xfa, "JPG10",
     0xfb, "JPG11",
     0xfc, "JPG12",
     0xfd, "JPG13",

     0xfe, "COM",
     );


while (@ARGV) {
    $_ = shift(@ARGV);
    if (/^--$/) {
	push(@files, @ARGV);
	last;
    } elsif ($_ eq "-l") {
        $action = "list";
    } elsif ($_ eq "-x") {
        $action = "extract";
    } elsif ($_ eq "-e") {
        $action = "extract";
    } elsif ($_ eq "-p") {
        $action = "print";
    } elsif ($_ eq "-c" || $_ eq "--check") {
        $action = "check";
        $verbose = 0;
    } elsif ($_ eq "-h" || $_ eq "--help") {
        $action = "help";
    } elsif ($_ eq "-q" || $_ eq "--quiet") {
        $verbose = 1;
    } elsif ($_ eq "-v" || $_ eq "--verbose") {
        $verbose = 2;
    } elsif ($_ eq "--version") {
        $action = "version";
    } elsif ($_ eq "-d" || $_ eq "--debug") {
        $debug = 1;
    } elsif (/^-/) {

    } else {
	push(@files, $_, @ARGV);
	last;
    }
}

if (@files) {
} elsif ($action ne "version") {
    $action = "help";
}

if ($action eq "version") {
    print STDERR "jpginfo.pl version $vernum\n";
    exit;
}

if ($action eq "help") {
    print STDERR <<"EOM";
jpginfo.pl version $vernum - describe JPEG infomation
copyright (c) hanawa <y\@hnw.jp>
usage: $0 [JPEGfiles ...]
EOM

	exit;
}

my($max_length) = (0);
foreach (@files) {
    $max_length = length($_) if ($max_length < length($_));
}

if (@files >= 2 || $verbose <= 1) {
    $print_filename = 1;
}

my (@printbuf, $is_no_problem);

FILE: while (@files) {
  $filename = shift(@files);
  @printbuf = ();

  if (-d $filename) {
    push(@printbuf, "directory");
    push(@printbuf, "\n") if ($verbose >= 2);
    next;
  }

  open(FH, $filename) || die "file not found: $filename\n";
  binmode(FH);

    $is_no_problem = 0;
    $error_occured = 0;
    $error_msg = "";
    $is_jpg = 0;
    $jpg_ended = 0;

    read(FH, $_, 32);
    if (/^\xff\xd8\xff/) {
      $usojpeg = 0;
      if (/^\xff\xd8\xff\xe0\x00\x10JFIF.{10}2\.reVGEPJOSU/) {
        $usojpeg = 1;
        seek(FH, -32, 2); # end of the file -32
        read(FH, $_, 32);
        $_ = reverse($_);
      } elsif (/^\xff\xd8\xff\xe0\x00\x10JFIF.{10}USOJPEGVer\.2/) {
        push(@printbuf, "(not reversed)"); ### will be commented out
        $usojpeg = 1;
        seek(FH, 32, 0); # beginning of the file +32
        read(FH, $_, 32);
      }
      if ($usojpeg) {
        push(@printbuf, "usojpeg");
        if (/^Rar!/) {
          push(@printbuf, " ", "rar?");
        } elsif (/^PK/) {
          push(@printbuf, " ", "zip?");
        } elsif (/^.RMF/) {
          push(@printbuf, " ", "rm?");
        }
	push(@printbuf, "\n") if ($verbose >= 2);
        next;
      } 
      # SOI
      push(@printbuf, "SOI\n") if ($verbose >= 2);
    } elsif (/^MZ/) {
	push(@printbuf, "EXE?");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^GIF8[79]a/) {
	push(@printbuf, "GIF?");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^\x89PNG\x0d\x0a\x1a\x0a/) {
	push(@printbuf, "PNG?");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^BM/) {
	push(@printbuf, "BMP?");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^..-lh[d0-9]-/) {
	push(@printbuf, "lzh");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^PK/) {
	push(@printbuf, "zip");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^Rar!/) {
	push(@printbuf, "rar");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^MSCF/) {
	push(@printbuf, "cab");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^Pokonya/) {
	push(@printbuf, "Pokonyan?");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^%PDF/) {
	push(@printbuf, "PDF?");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } elsif (/^.RMF/) {
	push(@printbuf, "rm?");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    } else {
	push(@printbuf, "not JPEG");
	push(@printbuf, "\n") if ($verbose >= 2);
	next;
    }

    seek(FH, 3, 0); # beginning of the file +3
    $is_jpg = 1; # hmm...

    while (<FH>) {
	if (/^\x00/) {
	    next;
	} elsif (/^\xda/) {
	    # SOS
	    push(@printbuf, "SOS\n") if ($verbose >= 2);
	    next;
#	    last;
	} elsif (/^([\xd0-\xd7])/) {
	    # RST (d0-d7???)
	    push(@printbuf, "RST", ord($1)-0xd0, "\n") if ($verbose >= 2);
	    next;
	} elsif (/^\xd9/) {
	    # EOI
	    push(@printbuf, "EOI\n") if ($verbose >= 2);
	    $length = length($_);
	    $jpg_ended = 1;
	    if ($length == 1) {
		# this file is correct JPEG.
		$is_no_problem = 1;
	    } elsif ($length > 1) {
		seek(FH, 1-length($_), 1);
		$length = read(FH, $buf, 18) || die;

		$fake_jpeg = "";
		if ($buf =~ /^....ImplantArchive/s) {
		    $fake_jpeg = "implant";
		} elsif ($buf =~ /^Pokonya/) {
		    $fake_jpeg = "pokonyan";
#			print " ", unpack("H2", substr($buf, 7, 1)), " ";
#			print " ", unpack("H2", substr($buf, 8, 1)), " ";
#			print " ", unpack("H2", substr($buf, 9, 1)), " ";
#			print " ", unpack("H2", substr($buf, 10, 1)), " ";
		}
		while ($read_len = read(FH, $buf, 1024)) {
		    $length += $read_len;
		}
		if ($action eq "info" || $action eq "check") {
		    push(@printbuf, " / ") if ($verbose == 1);
		    push(@printbuf, "  ", $length);
		    if ($length >= 2) {
			push(@printbuf, "bytes");
		    } else {
			push(@printbuf, "byte");
		    }

		    if (!$fake_jpeg) {
			push(@printbuf, " ignored after EOI.");
		    } elsif ($fake_jpeg eq "implant") {
			push(@printbuf, " implant data found.");
		    } elsif ($fake_jpeg eq "pokonyan") {
			push(@printbuf, " pokonyan data found.");
		    }
		    push(@printbuf, "\n") if ($verbose >= 2);
		}
	    }
	    last;
	}


	# exclude EOI & RSTn,
	# field length is placed right after marker.
	# (which occupies 2bytes)
	# After SOS field, there is scan data with no marker.
	# '0xff' is escaped to '0xff 0x00' in the scan data.

	if (!(s/^(.)(..)//s)) {
	    $error_occured = 1;
	    $error_msg = "Premature end.";
	    next FILE;
	}
	$symbol = $symbol{ord($1)};
	$length = unpack("n", $2);
	if (!(defined $symbol)) {
	    # unknown symbol
	    $error_occured = 1;
	    $error_msg = "Unsupported marker type 0x";
	    $error_msg .= unpack("H2", $1);
	    $error_msg .= " with length ";
	    $error_msg .= unpack("H4", $2);
	    next FILE;
	}

	push(@printbuf, $symbol, "\t: ", $length, "\n") if ($verbose >= 2);

	# exclude field length part and line separater 0xff (-2+1)
	$length -= 1;
	while (length($_) < $length) {
	    if ($tmp = <FH>) {
		$_ .= $tmp;
	    } else {
		last;
	    }
	}

	if (length($_) != $length) {
	    $error_occured = 1;
	    if (length($_) > $length) {
		$error_msg = "Illegal block size";
	    } else {
		$error_msg = "Premature end."
	    }

	    # Dealing with bug for "Arles Image Web Page Creator",
	    # which creates JPEG files with illegal COM block.
	    if ($symbol eq "COM" &&
		length($_) == 70 &&
		$length == 68) {
		
		$length = 70;
		# continues analysis.
	    } else {
		next FILE;
	    }
	}
	$length += 1;

	# show additional infomation
	if ($symbol eq "APP0") {
	    if (s/^JFIF\x00(..)(.)(..)(..)(.)(.)//s) {
		for $i (0,16,32) {
		    $mask = "kIUCHIuCHINO"x4;
		    if (length($_) >= $i+16) {
			$cp = substr($_, $i, 16) ^ substr($mask, $i, 16);
			$cp =~ s/[\x00-\x1f\x80-\xff].*$//s;
			push(@printbuf, "  ", $cp) if ($verbose >= 1);
			push(@printbuf, "\n") if ($verbose >= 2);
		    }
		}
#		print "APP0 ok\n";
	    }
	} elsif ($symbol =~ /^SOF/) {
	    if (length($_) >= 5) {
#		$p = unpack("C", substr($_, 0, 1));
		$x = unpack("n", substr($_, 1, 2));
		$y = unpack("n", substr($_, 3, 2));

		if ($verbose >= 2) {
                    push(@printbuf, " ", $y, " x ", $x, "\n");
		} elsif ($verbose >= 1) {
                    unshift(@printbuf, " ", $y, " x ", $x);
		}
	    }
	} elsif ($symbol =~ /^COM/) {
	    # remove 0xff
	    chop;
	    # ?
	    s/\x00$//;
	    push(@printbuf, $_, "\n") if ($verbose >= 2);
	}
    }
} continue {
    close(FH);

    if ($is_jpg && !$jpg_ended && !$error_occured) {
	$error_occured = 1;
	$error_msg = "Premature end."
    }
    if ($error_occured) {
	push(@printbuf, " / ") if ($verbose <= 1 && @printbuf >= 1);
	push(@printbuf, $error_msg);
	push(@printbuf, "\n") if ($verbose >= 2);
    }

    if ($action ne "check" ||
	$action eq "check" && !$is_no_problem) {

	if ($print_filename) {
	    if ($verbose >= 2) {
		print "\n==> ", $filename, " <==\n";
	    } else {
                print $filename, ":", " " x ($max_length-length($filename)+1);
	    }
	}
	print @printbuf, "\n";
    }

#	print "\n" if ($action eq "info");
}