recover jpg version of image out of corrupt/weird raw/dng tiff

tiff.rb

#!/usr/bin/env ruby

require 'rubygems'
require 'ruby-debug'

def out s = ''
  puts "#{$indent}#{s}"
end

def read_ifd f, order = 'n', indent = ''
  old_indent, $indent = $indent, indent
  n = f.read(2).unpack(order).first
  out
  out "\e[42m#{n} IFD entries.\e[00m"
  entries = f.read 12 * n
  next_ifd = f.read(4).unpack(order.upcase).first
  sub_ifd = nil
  jpg_offset = jpg_length = nil

  (0 ... n).each do |i|
    entry = entries[12 * i, 12]
    tag, type, count, value = entry.unpack "#{order}2#{order.upcase}2"
    next if type == 2
    out "ifd#{i}"
    out sprintf "  0x%04x %d %2d 0x%08x (%d)", tag, type, count, value, value
    short_value = value >> 16
    case tag
    when 0x106
      out "  PhotometricInterpretation"
      case short_value
      when 0
        out "  white is zero"
      when 1
        out "  black is zero"
      when 2
        out "  RGB"
      when 32803
        out "  CFA (Color Filter Array)"
      else
        out "  unknown value"
      end
    when 0x112
      out "  Orientation"
    when 0x11c
      out "  PlanarConfiguration"
      case short_value
      when 1
        out "  chunky"
      else
        out "  unknown value"
      end
    when 0x115
      out "  SamplesPerPixel"
    when 0x102
      out "  BitsPerSample"
    when 0x103
      out "  Compression"
      case short_value
      when 1
        out "  no compression"
      when 2
        out "  ccitt group 3"
      when 6
        out "  JPEG (old)"
      when 32773
        out "  packbits"
      else
        out "  unknown value"
      end
    when 0x101
      out "  ImageLength"
    when 0x100
      out "  ImageWidth"
    when 0x128
      out "  ResolutionUnit"
      case short_value
      when 1
        out "  no absolute unit of measurement"
      when 2
        out "  inch"
      when 3
        out "  centimeter"
      else
        out "  unknown value"
      end
    when 0x11a
      out "  XResolution"
    when 0x11b
      out "  YResolution"
    when 0x116
      out "  RowsPerStrip"
    when 0x111
      out "  StripOffsets"
    when 0x117
      out "  StripByteCounts"
    when 0xfe
      out "  NewSubfileType"
    when 0x214
      out "  ReferenceBlackWhite"
    when 0x10f
      out "  Make"
    when 0x110
      out "  Model"
    when 0x131
      out "  Software"
    when 0x132
      out "  DateTime"
    when 0x201
      jpg_offset = value
    when 0x202
      jpg_length = value
    when 0x14a
      out "  \e[44mSubIFDs\e[00m"
      f.seek value
      offsets = f.read(4 * count).unpack "#{order.upcase}#{count}"
      count.times do |i|
        out "  subifd#{i}: #{offsets[i]}"
        f.seek offsets[i]
        read_ifd f, order, indent + '    '
      end
    else
      out "  \e[31mUNKNOWN\e[00m"
    end

    case type
    when 2
      if count > 4
        f.seek value
        out "  \"#{f.read(count)}\""
      else
        out [value].pack "c#{count-1}"
      end
    when 3
      case count
      when 1
        out sprintf "  0: 0x%04x (%d)", short_value, short_value
      when 2
        out sprintf "  0: 0x%04x (%d)", short_value, short_value
        out sprintf "  1: 0x%04x (%d)", value & 0xffff, value & 0xffff
      else
        f.seek value
        count.times do |i|
          v = f.read(2).unpack(order).first
          out sprintf "  %d: 0x%04x (%d)", i, v, v
        end
      end
    when 5
      f.seek value
      count.times do |i|
        n, d = f.read(8).unpack("#{order}2")
        out "  #{i}: #{n}/#{d}"
      end
    end

    out
  end

  if jpg_offset && jpg_length
    out "  outputting jpg to jpg/#{ARGV.first}.jpg"
    open "jpg/#{ARGV.first}.jpg", 'w' do |w|
      f.seek jpg_offset
      w.write f.read(jpg_length)
    end
  end

  $indent = old_indent
  next_ifd
end

open ARGV.first do |f|
  # decode header
  header = f.read 8

  order =
    case header[0, 2]
    when 'II'
      puts "little-endian."
      'v'
    when 'MM'
      puts "big-endian."
      'n'
    else
      warn "not a tiff file\n"
      return
    end

  tiff = header[2, 2].unpack(order).first
  puts "tiff file." if tiff == 42
  offset = header[4, 4].unpack(order.upcase).first
  puts "offset: #{offset}."
  f.seek offset

  # read image file directory (IFD)
  next_ifd = read_ifd f, order

  puts "next ifd: #{next_ifd}"
end