I am a subclass of ImageReadWriter that understands JFIF file streams, and can decode JPEG images.
This code is based upon the Independent Joint Photographic Experts Group (IJPEG) software, originally written in C by Tom Lane, Philip Gladstone, Luis Ortiz, Jim Boucher, Lee Crocker, Julian Minguillon, George Phillips, Davide Rossi, Ge' Weijers, and other members of the Independent JPEG Group.

	colorspace conversion
	colorConvertFloatYCbCrMCU colorConvertGrayscaleMCU colorConvertIntYCbCrMCU colorConvertMCU primColorConvertGrayscaleMCU primColorConvertGrayscaleMCU:bits:residuals:ditherMask: primColorConvertIntYCbCrMCU primColorConvertYCbCrMCU:bits:residuals:ditherMask: sampleFloatRangeLimit: sampleRangeLimit:

	dct
	dctFloatRangeLimit: idctBlockFloat:component: idctBlockInt:component: idctBlockInt:qt: idctMCU primIdctBlockInt:component: primIdctInt:qt: scaleQuantizationTable:

	error handling
	notSupported:

	huffman encoding
	decodeBlockInto:component:dcTable:acTable: decodeMCU getBits: primDecodeBlockInto:component:dcTable:acTable:stream: processRestart scaleAndSignExtend:inFieldWidth:

	marker parsing
	initialSOSSetup okToIgnoreMarker: parseAPPn parseDecoderRestartInterval parseFirstMarker parseHuffmanTable parseNOP parseNextMarker parseQuantizationTable parseStartOfFile parseStartOfInput parseStartOfScan perScanSetup skipMarker

	preferences
	useFloatingPoint

	private
	on:

	public access
	decompressionTest nextImage nextImageDitheredToDepth: setStream:

	testing
	understandsImageFormat

	accessing	top
	hACTable
	hDCTable
	qTable

	colorspace conversion	top
	colorConvertFloatYCbCrMCU
	colorConvertGrayscaleMCU
	colorConvertIntYCbCrMCU
	colorConvertMCU
	primColorConvertGrayscaleMCU
	primColorConvertGrayscaleMCU:bits:residuals:ditherMask:
	primColorConvertIntYCbCrMCU
	primColorConvertYCbCrMCU:bits:residuals:ditherMask:
	sampleFloatRangeLimit:
	sampleRangeLimit:

	dct	top
	dctFloatRangeLimit:
	idctBlockFloat:component:
	idctBlockInt:component:
	idctBlockInt:qt:
	idctMCU
	primIdctBlockInt:component:
	primIdctInt:qt:
	scaleQuantizationTable:

	error handling	top
	notSupported:

	huffman encoding	top
	decodeBlockInto:component:dcTable:acTable:
	decodeMCU
	getBits:
	primDecodeBlockInto:component:dcTable:acTable:stream:
	processRestart
	scaleAndSignExtend:inFieldWidth:

	marker parsing	top
	initialSOSSetup
	okToIgnoreMarker:
	parseAPPn
	parseDecoderRestartInterval
	parseFirstMarker
	parseHuffmanTable
	parseNOP don't need to do anything, here
	parseNextMarker Parse the next marker of the stream
	parseQuantizationTable
	parseStartOfFile
	parseStartOfInput
	parseStartOfScan
	perScanSetup
	skipMarker

	preferences	top
	useFloatingPoint

	private	top
	on:

	public access	top
	decompressionTest Test decompression; don't generate actual image
	nextImage Dencoding an image on stream and answer the image.
	nextImageDitheredToDepth:
	setStream: Feed it in from an existing source

	testing	top
	understandsImageFormat Answer true if the image stream format is understood by this decoder.

	initialization	top
	initialize JPEGReadWriter initialize