Unless otherwise noted, the *.gzfiles were generated by the gzip command line tool and the *.deflate and *.zlib versions were then generated by script/extract-deflate-offsets.go. Similarly, the *.giflzw files were generated by script/extract-giflzw.go and the *.palette and *.indexes files were generated by script/extract-palette-indexes.go. The *.tokens files were generated by script/print-json-token-debug-format.c.

The *.jpeg files are usually the canonical versions of the test/data images, and other versions (*.bmp, *.gif, *.png etc) were usually generated by ImageMagick's convert command line tool. The *.wbmp versions were generated by the script/convert-png-to-wbmp.go command line tool. The *.webp versions were generated by the cwebp command line tool. The *.no-ancillary.png files were generated by the script/strip-png-ancillary-chunks.go command line tool. The *.qoi files were generated by the qoiconv command line tool. The *.pkm files were generated by the ETCPACK command line tool, except the *.etc1s.pkm files were generated by a custom etc1tool.

The *.apng files were generated by gif2apng. Those with finite animation loop counts were manually patched to correct for a gif2apng bug. https://bugs.chromium.org/p/chromium/issues/detail?id=592735 argues that GIF and APNG/WEBP have different semantics but gif2apng does not cater for that.

The artificial-* directories hold artificially generated test data, often to explicitly test corner cases of various file formats. The files there usually come in families whose names have a common prefix, such as three files foo.bar, foo.bar.qux, foo.bar.commentary.txt all prefixed by foo.bar. The file whose name is that prefix is usually the canonical (typically hand-crafted) file, and the other files are then derived from that. Outside of the artificial-* directories, the other files in this directory are typically real world examples of various file formats, or deriviations of them.

The non-ascii directory holds trivial UTF-8 (but not ASCII) text files.

36.png and 49.png are simple, artificially generated images. The generation script is gen-36-49.go from https://github.com/nigeltao/etc2 and the 49.*.nie variants were generated from a customized version of that script.

DCI-P3-D65.icc comes from color.org. DCI-P3-D65.icc.zlib is a zlib-compresion of that, created by Go's standard library.

archive.* archives a subset of other files in this directory.

animated-red-blue.gif is an original animation by Nigel Tao nigeltao@golang.org. animated-red-blue.nia and animated-red-blue.*.nie were then generated by example/convert-to-nia and script/extract-nia-frames.go.

australian-abc-local-stations.json was crawled from data.gov.au.

bricks-* are various encodings of an original photo by Nigel Tao nigeltao@golang.org.

cbor-rfc-7049-examples.cbor is the concatenated examples from RFC 7049. The cbor-rfc-7049-examples.*.json files were then generated by example/cbor-to-json.

checkerboard.*.png was generated by script/make-red-blue-gradient.go.

crude-flag.* is an original animation by Nigel Tao nigeltao@golang.org. See the lib/nie documentation.

enwik5 is the first 1e5 bytes of the enwik8 or enwik9 data, derived from the English Wikipedia as of 2006.

file-sizes.json was created by running script/print-file-sizes-json.go in this repository's root directory.

gifplayer-muybridge.gif is an original animation by Nigel Tao nigeltao@golang.org.

harvesters.* are various encodings of a photo of “The Harvesters” by Pieter Bruegel the Elder, held by the Metropolitan Museum of Art. www.metmuseum.org lists that image as in the public domain.

github-tags.json was crawled from api.github.com.

hat.* are various encodings of a photo of “Self-Portrait with a Straw Hat (obverse: The Potato Peeler)” by Vincent van Gogh, held by the Metropolitan Museum of art. www.metmuseum.org lists that image as in the public domain.

hello.sh is a trivial shell script.

hibiscus.regular.* are various encodings of a photo of “Hibiscus and Parrots” by Louis Comfort Tiffany, held by the Metropolitan Museum of Art. www.metmuseum.org lists that image as in the public domain.

hibiscus.primitive.* is the result of running the hibiscus.regular.png image through github.com/fogleman/primitive and other programs:

• primitive -i hibiscus.regular.png -o hibiscus.primitive.svg -s 442 -n 32
• inkscape -z -e hibiscus.primitive.png hibiscus.primitive.svg
• Various invocations of convert and cwebp

hippopotamus.* are various encodings of a cropping of a photo of “Hippopotamus (William)”, held by the Metropolitan Museum of Art. www.metmuseum.org lists that image as in the public domain.

www.metmuseum.org says that “You are welcome to use images of artworks in The Met collection that the Museum believes to be in the public domain, or those to which the Museum waives any copyright it might have, for any purpose, including commercial and noncommercial use, free of charge and without requiring permission from the Museum.”

json-things.* are original JSON objects by Nigel Tao nigeltao@golang.org.

midsummer.txt is an excerpt of Shakespeare‘s "A Midsummer Night’s Dream", copied from shakespeare.mit.edu.

mona-lisa.* is derived from a Wikimedia Commons photo of the Mona Lisa.

muybridge.gif is derived from en.wikipedia.org which is in the public domain.

nobel-prizes.json was crawled from api.nobelprize.org.

peacock.* is derived from this peacock (pavo cristatus) image, “available under the Creative Commons CC0 1.0 Universal Public Domain Dedication”:

• wget --quiet https://upload.wikimedia.org/wikipedia/commons/thumb/6/6f/Pavo_Real_Venezolano.jpg/1024px-Pavo_Real_Venezolano.jpg
• convert 1024px-Pavo_Real_Venezolano.jpg -resize 100x75 peacock.ppm
• convert peacock.ppm peacock.png
• rm 1024px-Pavo_Real_Venezolano.jpg
• cjpeg peacock.ppm > peacock.default.jpeg
• cjpeg -optimize peacock.ppm > peacock.optimize.jpeg
• cjpeg -grayscale peacock.ppm > peacock.grayscale.jpeg
• cjpeg -progressive peacock.ppm > peacock.progressive.jpeg
• cjpeg -quality 30 peacock.ppm > peacock.q30.jpeg
• cjpeg -quality 99 peacock.ppm > peacock.q99.jpeg
• cjpeg -sample 2x2 peacock.ppm > peacock.s2x2-420.jpeg
• cjpeg -sample 2x1 peacock.ppm > peacock.s2x1-422.jpeg
• cjpeg -sample 1x1 peacock.ppm > peacock.s1x1-444.jpeg
• cjpeg -sample 1x3 peacock.ppm > peacock.s1x3.jpeg
• cjpeg -sample 2x2,2x1,1x1 peacock.ppm > peacock.s-weird.jpeg
• cjpeg -sample 4x1,2x1,2x2 peacock.ppm > peacock.s-very-weird.jpeg
• go run ../../script/truncate-progressive-jpeg.go peacock.progressive.jpeg

pi.txt contains the digits of pi.

pjw-thumbnail.* are various encodings of an image derived from an iconic, original photo of Peter J. Weinberger by Rob Pike r@golang.org.

red-blue-gradient.*.png was generated by script/make-red-blue-gradient.go.

rfc-6901-json-pointer.json is the example JSON document given in the RFC 6901 “JavaScript Object Notation (JSON) Pointer” specification.

rgb24png.bmp comes from BMP Suite, which states that its generated images “are in the public domain”.

ridiculously-fast.png is a screenshot of a Richard Geldreich tweet from January 2022. It was lightly edited to darken the non-text areas.

romeo.txt is an excerpt of Shakespeare's “Romeo and Juliet”, copied from shakespeare.mit.edu.

romeo.txt.fixed-huff.deflate was derived from romeo.txt by a custom program to use fixed (not dynamic) Huffman tables for the deflate encoding.

sheep-more.rac is a RAC-compression of original text by Nigel Tao nigeltao@golang.org.