Extract width and height from JPEG format

jpeg_markers.rb

module JpegMarkers
  MARKERS = Hash[*%W(
    SOF0  \xff\xc0
    SOF1  \xff\xc1
    SOF2  \xff\xc2
    SOF3  \xff\xc3
    DHT   \xff\xc4
    SOF5  \xff\xc5
    SOF6  \xff\xc6
    SOF7  \xff\xc7
    JPG   \xff\xc8
    SOF9  \xff\xc9
    SOF10 \xff\xca
    SOF11 \xff\xcb
    DAC   \xff\xcc
    SOF13 \xff\xcd
    SOF14 \xff\xce
    SOF15 \xff\xcf
    RST0  \xff\xd0
    RST1  \xff\xd1
    RST2  \xff\xd2
    RST3  \xff\xd3
    RST4  \xff\xd4
    RST5  \xff\xd5
    RST6  \xff\xd6
    RST7  \xff\xd7
    SOI   \xff\xd8
    EOI   \xff\xd9
    SOS   \xff\xda
    DQT   \xff\xdb
    DNL   \xff\xdc
    DRI   \xff\xdd
    DHP   \xff\xde
    EXP   \xff\xdf
    APP0  \xff\xe0
    APP1  \xff\xe1
    APP2  \xff\xe2
    APP3  \xff\xe3
    APP4  \xff\xe4
    APP5  \xff\xe5
    APP6  \xff\xe6
    APP7  \xff\xe7
    APP8  \xff\xe8
    APP9  \xff\xe9
    APP10 \xff\xea
    APP11 \xff\xeb
    APP12 \xff\xec
    APP13 \xff\xed
    APP14 \xff\xee
    APP15 \xff\xef
    JPG0  \xff\xf0
    JPG1  \xff\xf1
    JPG2  \xff\xf2
    JPG3  \xff\xf3
    JPG4  \xff\xf4
    JPG5  \xff\xf5
    JPG6  \xff\xf6
    JPG7  \xff\xf7
    JPG8  \xff\xf8
    JPG9  \xff\xf9
    JPG10 \xff\xfa
    JPG11 \xff\xfb
    JPG12 \xff\xfc
    JPG13 \xff\xfd
    COM   \xff\xfe
    TEM   \xff\x01
    RES1  \xff\x02
    RES2  \xff\xbf
  )]

  MARKERS.each do |k,v|
    const_set k, v
  end

  def marker?(bytes)
    MARKERS.values.include? bytes
  end

  def sof?(bytes)
    (0..15).any? {|i| MARKERS["SOF#{i}"] == bytes }
  end

  def inspect_marker(chr)
    key = MARKERS.key(chr)
    if key
      "#{key}(#{chr.unpack('n').first.to_s(16)})"
    else
      chr.inspect
    end
  end
end

jpeg_parser.rb

require_relative './jpeg_markers.rb'

class JpegParser
  include JpegMarkers
  attr_reader :data
  class ParseError < Exception; end

  def initialize(io)
    @io = io
    @context_stack = []
    @data = {}
  end

  def parse
    read 2
    assert SOI
    parse_frame
    @data
  end

  def parse_frame
    read 2
    parse_extra
    parse_frame_header
  end

  def parse_extra
    while marker?(current) && !sof?(current)
      context "#{inspect_marker(current)}" do
        length = read_int(2) - 2
        log "length: #{length}"
        skip length
      end
      read 2
    end
  end

  def parse_frame_header
    assert true, sof?(current)
    context "frame_header" do
      length = read_int(2) - 2
      log "length: #{length}"
      skip 1
      @data[:height] = read_int 2
      @data[:width]  = read_int 2
      skip length - 5
    end
  end

  def current
    @current
  end

  def log(msg)
    header = @context_stack.map {|c| "[#{c}]" }.join
    puts [header, msg].join(' ')
  end

  def context(name)
    @context_stack << name
    yield
    @context_stack.pop
  end

  def read_int(bytes)
    read(bytes).unpack('n').first
  end

  def read(bytes)
    @current = @io.read bytes
    log "read(#{bytes}): #{inspect_marker(@current)}"
    @current
  end

  def skip(bytes)
    @io.read bytes
    log "skip(#{bytes})"
  end

  def assert(expected, actual=nil)
    actual ||= current
    if expected != actual
      msg = <<-END
Parse error at #{@io.pos} bytes:
  got:      #{inspect_marker(actual)}
  expected: #{inspect_marker(expected)}
      END
      raise ParseError, msg
    end
  end
end

reader = JpegParser.new ARGF
data = reader.parse
p data

output.txt

$ ruby jpeg_parser.rb input.jpg
 read(2): SOI(ffd8)
 read(2): APP0(ffe0)
[APP0(ffe0)] read(2): "\x00\x10"
[APP0(ffe0)] length: 14
[APP0(ffe0)] skip(14)
 read(2): APP2(ffe2)
[APP2(ffe2)] read(2): "\x02@"
[APP2(ffe2)] length: 574
[APP2(ffe2)] skip(574)
 read(2): APP1(ffe1)
[APP1(ffe1)] read(2): "\x00t"
[APP1(ffe1)] length: 114
[APP1(ffe1)] skip(114)
 read(2): DQT(ffdb)
[DQT(ffdb)] read(2): "\x00C"
[DQT(ffdb)] length: 65
[DQT(ffdb)] skip(65)
 read(2): DQT(ffdb)
[DQT(ffdb)] read(2): "\x00C"
[DQT(ffdb)] length: 65
[DQT(ffdb)] skip(65)
 read(2): SOF0(ffc0)
[frame_header] read(2): "\x00\x11"
[frame_header] length: 15
[frame_header] skip(1)
[frame_header] read(2): "\x05\xAD"
[frame_header] read(2): "\x04>"
[frame_header] skip(10)
{:height=>1453, :width=>1086}