JPEG (Joint Photographic Experts Group) File Interchange Format: GraphViz block diagram (.dot) source

JPEG File Interchange Format, or JFIF, or, more colloquially known as just "JPEG" or "JPG", is a popular 2D bitmap image file format, offering lossy compression which works reasonably well with photographic images.

Format is organized as a container format, serving multiple "segments", each starting with a magic and a marker. JFIF standard dictates order and mandatory apperance of segments:

  • SOI
  • APP0 (with JFIF magic)
  • APP0 (with JFXX magic, optional)
  • everything else
  • SOS
  • JPEG-compressed stream
  • EOI

File extension

["jpg", "jpeg", "jpe", "jif", "jfif", "jfi"]

KS implementation details

License: CC0-1.0

References

This page hosts a formal specification of JPEG (Joint Photographic Experts Group) File Interchange Format using Kaitai Struct. This specification can be automatically translated into a variety of programming languages to get a parsing library.

GraphViz block diagram source

jpeg.dot

Download 

digraph {
	rankdir=LR;
	node [shape=plaintext];
	subgraph cluster__jpeg {
		label="Jpeg";
		graph[style=dotted];

		jpeg__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
			<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
			<TR><TD PORT="segments_pos">0</TD><TD PORT="segments_size">...</TD><TD>Segment</TD><TD PORT="segments_type">segments</TD></TR>
			<TR><TD COLSPAN="4" PORT="segments__repeat">repeat to end of stream</TD></TR>
		</TABLE>>];
		subgraph cluster__segment {
			label="Jpeg::Segment";
			graph[style=dotted];

			segment__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
				<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
				<TR><TD PORT="magic_pos">0</TD><TD PORT="magic_size">1</TD><TD></TD><TD PORT="magic_type">magic</TD></TR>
				<TR><TD PORT="marker_pos">1</TD><TD PORT="marker_size">1</TD><TD>u1→MarkerEnum</TD><TD PORT="marker_type">marker</TD></TR>
				<TR><TD PORT="length_pos">2</TD><TD PORT="length_size">2</TD><TD>u2be</TD><TD PORT="length_type">length</TD></TR>
				<TR><TD PORT="data_pos">4</TD><TD PORT="data_size">...</TD><TD>switch (marker)</TD><TD PORT="data_type">data</TD></TR>
				<TR><TD PORT="image_data_pos">...</TD><TD PORT="image_data_size">⇲</TD><TD></TD><TD PORT="image_data_type">image_data</TD></TR>
			</TABLE>>];
segment__seq_data_switch [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
	<TR><TD BGCOLOR="#F0F2E4">case</TD><TD BGCOLOR="#F0F2E4">type</TD></TR>
	<TR><TD>:marker_enum_app1</TD><TD PORT="case0">SegmentApp1</TD></TR>
	<TR><TD>:marker_enum_app0</TD><TD PORT="case1">SegmentApp0</TD></TR>
	<TR><TD>:marker_enum_sof0</TD><TD PORT="case2">SegmentSof0</TD></TR>
	<TR><TD>:marker_enum_sos</TD><TD PORT="case3">SegmentSos</TD></TR>
</TABLE>>];
		}
		subgraph cluster__segment_sos {
			label="Jpeg::SegmentSos";
			graph[style=dotted];

			segment_sos__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
				<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
				<TR><TD PORT="num_components_pos">0</TD><TD PORT="num_components_size">1</TD><TD>u1</TD><TD PORT="num_components_type">num_components</TD></TR>
				<TR><TD PORT="components_pos">1</TD><TD PORT="components_size">2</TD><TD>Component</TD><TD PORT="components_type">components</TD></TR>
				<TR><TD COLSPAN="4" PORT="components__repeat">repeat num_components times</TD></TR>
				<TR><TD PORT="start_spectral_selection_pos">...</TD><TD PORT="start_spectral_selection_size">1</TD><TD>u1</TD><TD PORT="start_spectral_selection_type">start_spectral_selection</TD></TR>
				<TR><TD PORT="end_spectral_pos">...</TD><TD PORT="end_spectral_size">1</TD><TD>u1</TD><TD PORT="end_spectral_type">end_spectral</TD></TR>
				<TR><TD PORT="appr_bit_pos_pos">...</TD><TD PORT="appr_bit_pos_size">1</TD><TD>u1</TD><TD PORT="appr_bit_pos_type">appr_bit_pos</TD></TR>
			</TABLE>>];
			subgraph cluster__component {
				label="Jpeg::SegmentSos::Component";
				graph[style=dotted];

				component__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
					<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
					<TR><TD PORT="id_pos">0</TD><TD PORT="id_size">1</TD><TD>u1→ComponentId</TD><TD PORT="id_type">id</TD></TR>
					<TR><TD PORT="huffman_table_pos">1</TD><TD PORT="huffman_table_size">1</TD><TD>u1</TD><TD PORT="huffman_table_type">huffman_table</TD></TR>
				</TABLE>>];
			}
		}
		subgraph cluster__segment_app1 {
			label="Jpeg::SegmentApp1";
			graph[style=dotted];

			segment_app1__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
				<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
				<TR><TD PORT="magic_pos">0</TD><TD PORT="magic_size">...</TD><TD>str(ASCII)</TD><TD PORT="magic_type">magic</TD></TR>
				<TR><TD PORT="body_pos">...</TD><TD PORT="body_size">...</TD><TD>switch (magic)</TD><TD PORT="body_type">body</TD></TR>
			</TABLE>>];
segment_app1__seq_body_switch [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
	<TR><TD BGCOLOR="#F0F2E4">case</TD><TD BGCOLOR="#F0F2E4">type</TD></TR>
	<TR><TD>&quot;Exif&quot;</TD><TD PORT="case0">ExifInJpeg</TD></TR>
</TABLE>>];
		}
		subgraph cluster__segment_sof0 {
			label="Jpeg::SegmentSof0";
			graph[style=dotted];

			segment_sof0__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
				<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
				<TR><TD PORT="bits_per_sample_pos">0</TD><TD PORT="bits_per_sample_size">1</TD><TD>u1</TD><TD PORT="bits_per_sample_type">bits_per_sample</TD></TR>
				<TR><TD PORT="image_height_pos">1</TD><TD PORT="image_height_size">2</TD><TD>u2be</TD><TD PORT="image_height_type">image_height</TD></TR>
				<TR><TD PORT="image_width_pos">3</TD><TD PORT="image_width_size">2</TD><TD>u2be</TD><TD PORT="image_width_type">image_width</TD></TR>
				<TR><TD PORT="num_components_pos">5</TD><TD PORT="num_components_size">1</TD><TD>u1</TD><TD PORT="num_components_type">num_components</TD></TR>
				<TR><TD PORT="components_pos">6</TD><TD PORT="components_size">3</TD><TD>Component</TD><TD PORT="components_type">components</TD></TR>
				<TR><TD COLSPAN="4" PORT="components__repeat">repeat num_components times</TD></TR>
			</TABLE>>];
			subgraph cluster__component {
				label="Jpeg::SegmentSof0::Component";
				graph[style=dotted];

				component__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
					<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
					<TR><TD PORT="id_pos">0</TD><TD PORT="id_size">1</TD><TD>u1→ComponentId</TD><TD PORT="id_type">id</TD></TR>
					<TR><TD PORT="sampling_factors_pos">1</TD><TD PORT="sampling_factors_size">1</TD><TD>u1</TD><TD PORT="sampling_factors_type">sampling_factors</TD></TR>
					<TR><TD PORT="quantization_table_id_pos">2</TD><TD PORT="quantization_table_id_size">1</TD><TD>u1</TD><TD PORT="quantization_table_id_type">quantization_table_id</TD></TR>
				</TABLE>>];
				component__inst__sampling_x [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
					<TR><TD BGCOLOR="#E0FFE0">id</TD><TD BGCOLOR="#E0FFE0">value</TD></TR>
					<TR><TD>sampling_x</TD><TD>((sampling_factors &amp; 240) &gt;&gt; 4)</TD></TR>
				</TABLE>>];
				component__inst__sampling_y [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
					<TR><TD BGCOLOR="#E0FFE0">id</TD><TD BGCOLOR="#E0FFE0">value</TD></TR>
					<TR><TD>sampling_y</TD><TD>(sampling_factors &amp; 15)</TD></TR>
				</TABLE>>];
			}
		}
		subgraph cluster__exif_in_jpeg {
			label="Jpeg::ExifInJpeg";
			graph[style=dotted];

			exif_in_jpeg__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
				<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
				<TR><TD PORT="extra_zero_pos">0</TD><TD PORT="extra_zero_size">1</TD><TD></TD><TD PORT="extra_zero_type">extra_zero</TD></TR>
				<TR><TD PORT="data_pos">1</TD><TD PORT="data_size">⇲</TD><TD>Exif</TD><TD PORT="data_type">data</TD></TR>
			</TABLE>>];
		}
		subgraph cluster__segment_app0 {
			label="Jpeg::SegmentApp0";
			graph[style=dotted];

			segment_app0__seq [label=<<TABLE BORDER="0" CELLBORDER="1" CELLSPACING="0">
				<TR><TD BGCOLOR="#E0FFE0">pos</TD><TD BGCOLOR="#E0FFE0">size</TD><TD BGCOLOR="#E0FFE0">type</TD><TD BGCOLOR="#E0FFE0">id</TD></TR>
				<TR><TD PORT="magic_pos">0</TD><TD PORT="magic_size">5</TD><TD>str(ASCII)</TD><TD PORT="magic_type">magic</TD></TR>
				<TR><TD PORT="version_major_pos">5</TD><TD PORT="version_major_size">1</TD><TD>u1</TD><TD PORT="version_major_type">version_major</TD></TR>
				<TR><TD PORT="version_minor_pos">6</TD><TD PORT="version_minor_size">1</TD><TD>u1</TD><TD PORT="version_minor_type">version_minor</TD></TR>
				<TR><TD PORT="density_units_pos">7</TD><TD PORT="density_units_size">1</TD><TD>u1→DensityUnit</TD><TD PORT="density_units_type">density_units</TD></TR>
				<TR><TD PORT="density_x_pos">8</TD><TD PORT="density_x_size">2</TD><TD>u2be</TD><TD PORT="density_x_type">density_x</TD></TR>
				<TR><TD PORT="density_y_pos">10</TD><TD PORT="density_y_size">2</TD><TD>u2be</TD><TD PORT="density_y_type">density_y</TD></TR>
				<TR><TD PORT="thumbnail_x_pos">12</TD><TD PORT="thumbnail_x_size">1</TD><TD>u1</TD><TD PORT="thumbnail_x_type">thumbnail_x</TD></TR>
				<TR><TD PORT="thumbnail_y_pos">13</TD><TD PORT="thumbnail_y_size">1</TD><TD>u1</TD><TD PORT="thumbnail_y_type">thumbnail_y</TD></TR>
				<TR><TD PORT="thumbnail_pos">14</TD><TD PORT="thumbnail_size">((thumbnail_x * thumbnail_y) * 3)</TD><TD></TD><TD PORT="thumbnail_type">thumbnail</TD></TR>
			</TABLE>>];
		}
	}
	jpeg__seq:segments_type -> segment__seq [style=bold];
	segment__seq:data_type -> segment__seq_data_switch [style=bold];
	segment__seq_data_switch:case0 -> segment_app1__seq [style=bold];
	segment__seq_data_switch:case1 -> segment_app0__seq [style=bold];
	segment__seq_data_switch:case2 -> segment_sof0__seq [style=bold];
	segment__seq_data_switch:case3 -> segment_sos__seq [style=bold];
	segment__seq:marker_type -> segment__seq:data_type [color="#404040"];
	segment_sos__seq:components_type -> component__seq [style=bold];
	segment_sos__seq:num_components_type -> segment_sos__seq:components__repeat [color="#404040"];
	segment_app1__seq:body_type -> segment_app1__seq_body_switch [style=bold];
	segment_app1__seq_body_switch:case0 -> exif_in_jpeg__seq [style=bold];
	segment_app1__seq:magic_type -> segment_app1__seq:body_type [color="#404040"];
	segment_sof0__seq:components_type -> component__seq [style=bold];
	segment_sof0__seq:num_components_type -> segment_sof0__seq:components__repeat [color="#404040"];
	component__seq:sampling_factors_type -> component__inst__sampling_x [color="#404040"];
	component__seq:sampling_factors_type -> component__inst__sampling_y [color="#404040"];
	exif_in_jpeg__seq:data_type -> exif__seq [style=bold];
	segment_app0__seq:thumbnail_x_type -> segment_app0__seq:thumbnail_size [color="#404040"];
	segment_app0__seq:thumbnail_y_type -> segment_app0__seq:thumbnail_size [color="#404040"];
}