Test cases

Every command on real files: the input, the command, what it printed and the output. Word, PowerPoint and Excel files are shown as macOS Quick Look renders them, and books as an e-reader engine lays them out. Pick a group, a command or a single run from the tree.

About this run

Run on commit 668647b, which is ahead of the latest release, so some commands here are not in a released version yet. Unit tests have no files to show and are not listed.

Credits

• CSpeed2.png: Benchmark chart from the zstd README. BSD-3-Clause (zstd), source.
• DCspeed5.png: Benchmark chart from the zstd README. BSD-3-Clause (zstd), source.
• DSCN0010.jpg: Camera photo with EXIF and GPS coordinates. MIT (exif-samples), source.
• DSpeed3.png: Benchmark chart from the zstd README. BSD-3-Clause (zstd), source.
• article.html: Wikipedia article Comma-separated values, saved whole: menus, sidebar, scripts, infobox, table. CC BY-SA 4.0 (Wikipedia), source.
• attachments.pdf: Acrobat XI AES-256 file that encrypts only its attachments (password attachment). Apache-2.0 (qpdf test suite), source.
• azure-interior.pdf: Odoo invoice: four products with units, a discount and tax. MIT (invoice2data), source.
• bat.toml: The Cargo.toml of bat. MIT (bat), source.
• books.md: Long Chinese Markdown list of programming books. CC BY 4.0, source.
• bunny.mp4: Big Buck Bunny, 10 seconds, 720p H.264, no audio track. CC BY 3.0 (Blender Foundation), source.
• bunny.webm: Big Buck Bunny, 10 seconds, VP9 WebM. CC BY 3.0 (Blender Foundation), source.
• catalog.xml: US House roll call vote 1 of 2023: 435 repeated votes with attributes, accented names. Public domain (US federal government work, 17 U.S.C. 105), source.
• clip.mov: iPhone 6 clip: 4 s H.264 stored 568x320 with a -90 rotation, two timed-metadata streams and GPS. Free to use (metadata-extractor-images README), source.
• clipping.pdf: New York Times column of 1880, a scan whose only text is a one-line archive stamp. Public domain (article of 1880); the footer names the NYT archive, source.
• cmyk.jpg: CMYK JPEG, as exported for print. HPND (Pillow test suite), source.
• commonmark.md: CommonMark spec README with code and links. CC BY-SA 4.0, source.
• commons-io.xml: The Maven pom.xml of Commons IO. Apache-2.0 (Commons IO), source.
• commons-lang.xml: The Maven pom.xml of Commons Lang. Apache-2.0 (Commons Lang), source.
• config.yaml: Helm values of the Artifact Hub chart. Apache-2.0 (Artifact Hub), source.
• coolblue.pdf: Coolblue invoice in Dutch: five products, euro amounts with commas. MIT (invoice2data), source.
• countries.csv: Country names and codes. PDDL-1.0, source.
• dice.png: PNG with real transparency. CC BY-SA 3.0, source.
• dict-cr.png: Benchmark chart from the zstd README. BSD-3-Clause (zstd), source.
• dict-cs.png: Benchmark chart from the zstd README. BSD-3-Clause (zstd), source.
• dict-ds.png: Benchmark chart from the zstd README. BSD-3-Clause (zstd), source.
• earth.gif: Animated GIF of the rotating Earth. CC BY-SA 3.0, source.
• earth.jpg: Apollo 17 Blue Marble, a large real photo (about 3000px). Public domain (NASA), source.
• example.ogg: Ogg Vorbis music clip. CC BY-SA 3.0, source.
• express.json: The package.json of Express. MIT (Express), source.
• f1040.pdf: IRS Form 1040. Public domain (US government work), source.
• frankenstein.epub: Frankenstein, EPUB 3 (Project Gutenberg). Public domain (Project Gutenberg), source.
• frontend-deployment.yaml: A Kubernetes deployment (guestbook example). Apache-2.0 (Kubernetes examples), source.
• fw9.pdf: IRS Form W-9, a real fillable form. Public domain (US government work), source.
• gallery.webp: WebP photo from Google's gallery. CC BY 3.0 (WebP gallery), source.
• gdp.csv: World Bank GDP by country and year, about 13,000 rows. PDDL-1.0, source.
• hal.pdf: HAL archive deposit whose cover page is tightly justified two-column text. CC BY 4.0 (HAL), source.
• hdr.mp4: Android phone HDR10 clip: 4 s HEVC Main 10, PQ, bt2020, rotated -90. Apache-2.0 (androidx/media), source.
• hopper.png: Small PNG portrait. HPND (Pillow test suite), source.
• huck-finn-scan.pdf: A real scan with no text layer: a page of Huckleberry Finn (OCRmyPDF test file). Public domain (Project Gutenberg scan), source.
• issues.ods: OpenDocument spreadsheet. MIT (calamine), source.
• issues.xls: Old binary Excel file. MIT (calamine), source.
• issues.xlsx: Workbook with several sheets (calamine test file). MIT (calamine), source.
• japanese.pdf: Japanese text with a non-embedded font. Apache-2.0 (pdf.js test suite), source.
• jellyfish.mp4: Jellyfish, 10 seconds, 720p H.264. CC BY 3.0, source.
• jfk.flac: President Kennedy speaking, with natural pauses (Whisper test file). MIT (Whisper); the speech is a US government work, source.
• koa.json: The package.json of Koa. MIT (Koa), source.
• landscape_6.jpg: Photo stored sideways with EXIF orientation 6. MIT (exif-samples), source.
• latex-form.pdf: pdflatex hyperref form whose checkbox 'on' appearance is an empty dictionary. See py-pdf/sample-files README (010), source.
• letter.docx: Two-page Word document (Apache POI test data). Apache-2.0 (Apache POI test data), source.
• letter.pdf: Handwritten letter of 1920 (Harry Burn's mother), seven scanned pages OCR cannot read. Public domain (Wikimedia Commons), source.
• locked.pdf: LibreOffice Writer PDF with the user password openpassword. CC BY-SA 4.0 (py-pdf/sample-files), source.
• manual.pdf: Washer-dryer manual whose only picture is a one-ink (DeviceN Black) image. Copyright Hotpoint, archived by the Internet Archive; used for testing only, source.
• markdown-article.html: A saved Wikipedia article, with its navigation and sidebars. CC BY-SA 4.0 (Wikipedia), source.
• memo.pdf: National Archives JFK release: 1953 FBI memo, 4 one-bit scanned pages, no text. Public domain (US federal record), source.
• movie.mp4: NASA STS-116 booster separation: 28 s H.264 800x450 with sound. Public domain (NASA), source.
• music.mp3: Kevin MacLeod, Greta Sting: 18 s MP3 320k with a 720x720 cover picture. CC BY 3.0 (Kevin MacLeod, incompetech.com), source.
• nda.pdf: University mutual NDA with its own footer at the bottom centre and right. Copyright Youngstown State University; used for testing only, source.
• news.webm: VOA news clip: 34 s VP8 852x480 with sound. Public domain (Voice of America), source.
• paper-v2.pdf: arXiv 2512.03139 version 2: pages 1 and 3 changed. CC BY 4.0 (arXiv 2512.03139), source.
• paper.pdf: 14-page academic paper with charts (TraceMonkey, PLDI 2009). Apache-2.0 (pdf.js test suite), source.
• paper.pdf: arXiv 2512.03139 version 1, 5 pages. CC BY 4.0 (arXiv 2512.03139), source.
• pauses.mp3: LibriVox reading of Lovecraft's Sunset: 55 s mono speech with pauses and a long silent end. Public domain (LibriVox), source.
• photo.heic: iPhone 11 Pro HEIC photo, 3024x4032 in 512x512 tiles. Free to use (metadata-extractor-images README), source.
• photos.pdf: NASA Moon lithograph: 2.6 MB, page 1 of large photographs. Public domain (NASA), source.
• phototest.tif: A scanned page of text (Tesseract test image). Apache-2.0 (Tesseract), source.
• portrait.heic: iPhone XR HEIC stored 4032x3024 with irot 270 and EXIF orientation 6: a portrait photo. Free to use (metadata-extractor-images README), source.
• portrait_8.jpg: Photo with EXIF orientation 8. MIT (exif-samples), source.
• prizes.json: Nobel Prize API answer for 2023: minified, nested, unsorted keys, Swedish text. CC0 (Nobel Prize API terms of use), source.
• prod.yaml: The production override used to deploy artifacthub.io. Apache-2.0 (Artifact Hub), source.
• protected.pdf: Encrypted with an empty user password. Apache-2.0 (pdf.js test suite), source.
• pub17.pdf: IRS Publication 17, a long real document. Public domain (US government work), source.
• qr-photo.png: A phone photo of a QR code on a screen (ZXing test images). Apache-2.0 (ZXing), source.
• quality-hosting.pdf: German hosting invoice: seven positions over two pages, page 1 text in reverse order. MIT (invoice2data), source.
• redis-leader-deployment.yaml: A second Kubernetes deployment (guestbook example). Apache-2.0 (Kubernetes examples), source.
• report.md: The zstd README: tables, code, badges and charts in doc/images placed inside tables. BSD-3-Clause (zstd), source.
• ride.gpx: A real recorded track with waypoints (gpxpy test files). Apache-2.0 (gpxpy), source.
• ripgrep.toml: The Cargo.toml of ripgrep. MIT (ripgrep), source.
• rotated.pdf: Pages with /Rotate set. Apache-2.0 (pdf.js test suite), source.
• rotation.pdf: Several pages with different rotations. Apache-2.0 (pdf.js test suite), source.
• rust-readme.md: Rust project README. MIT OR Apache-2.0, source.
• scan.pdf: Eisenhower Library scan, Project CORONA memo of 1959: 4 colour image pages, no text, 1.8 MB. Public domain (US federal record), source.
• shaky.ogv: Handheld clip of a calling marsh frog: 11 s Theora 1280x720, visibly shaky. CC BY-SA 3.0 (User:Calimo), source.
• show.pptx: Two slides with speaker notes (Apache POI test data). Apache-2.0 (Apache POI test data), source.
• table.pdf: A real report page with a table (Camelot test file). MIT (Camelot), source.
• tables.pdf: Census quarterly e-commerce report: two tables with notes among running text. Public domain (US Census Bureau), source.
• talk.srt: English subtitles for the VOA clip, 9 cues from 0 s. CC BY-SA (Wikimedia Commons TimedText), source.
• temperature.xlsx: Small real workbook. MIT (calamine), source.
• track.kml: A Google Earth track in gx:Track form (togeojson test data). BSD-2-Clause (togeojson), source.
• trailer.mkv: Sintel trailer as MKV: H.264 848x360, AAC and 8 subtitle tracks. CC BY 3.0 (Blender Foundation), source.
• turned.pdf: One scanned page four times, turned 0, 90, 180 and 270 degrees inside the image (OCRmyPDF test file). CC-BY-SA-4.0 (OCRmyPDF test resources), source.
• ventas.csv: INE retail sales index for large stores: semicolons, UTF-8 BOM, CRLF, decimal commas, empty cells and ... CC BY 4.0 (INE, Fuente: INE), source.
• zh.srt: Simplified Chinese subtitles for the VOA clip, 9 cues. CC BY-SA (Wikimedia Commons TimedText), source.
• 销售.csv: Daily shop sales with Chinese headers, UTF-8 with BOM, 400 rows. MIT (Incalos/Sales-Predict-With-LSTM), source.
• Tables from Visual Autoregressive Modeling. CC0 1.0.
• Tables from 1000 Layer Networks for Self-Supervised RL. CC BY 4.0.
• Tables from Artificial Hivemind. CC BY 4.0.

Other inputs were generated for these tests, some from the files above.