files.placeholder.com

Dummy files

Real files to test uploads, downloads, previews and parsers: documents, data, archives, images, audio and video, files of exact sizes, broken and mislabelled files, and every text encoding that breaks importers. All generated, all synthetic, the same bytes forever.

https://files.placeholder.com/<category>/<file>
https://files.placeholder.com/index.json        every file with size, type and sha256

https://files.placeholder.com/documents/sample.pdfPDF 1.4, 3 pages: text, a table and an embedded JPEG
https://files.placeholder.com/documents/sample.docxWord: headings, list, table, image, page break
https://files.placeholder.com/data/sample-10000-rows.csvCSV, 10,000 rows
https://files.placeholder.com/sizes/10mb.binexactly 10,485,760 bytes
https://files.placeholder.com/edge/truncated.pdfa PDF cut off at 60%
https://files.placeholder.com/index.jsonthe machine-readable catalog
https://files.placeholder.com/browse them all

What's there

107 files in nine folders. This is a summary; index.json is the full list.

Folder	Files
`documents/`	`sample.pdf`, `.docx`, `.xlsx` (2 sheets, formulas with cached values), `.pptx` (3 slides, speaker notes), `.odt`, `.ods`, `.rtf`, `.txt`, `.md`, `.html`
`data/`	`sample.csv` (quoted fields with commas, quotes and newlines), `sample-semicolon.csv`, `sample-10000-rows.csv`, `.tsv`, `.json`, `.ndjson`, `.xml` (namespace, CDATA), `.yaml`
`archives/`	`sample.zip`, `.tar`, `.tar.gz`, `.tar.bz2`, `.tar.xz` (the same five files in nested folders), `sample.txt.gz`
`images/`	PNG (also transparent, 100 × 100, 1920 × 1080), JPEG (baseline, progressive, 1920 × 1080), animated GIF, WebP (lossy, lossless), AVIF, SVG, BMP, TIFF, ICO
`audio/`	2-second tone as WAV, MP3, Ogg Vorbis, Opus, FLAC and M4A; `silence-1s.wav`
`video/`	`sample.mp4` (H.264 + AAC) and `sample.webm` (VP9 + Opus), 320 × 240, 2 s
`sizes/`	Exact byte counts: `.bin` from 1 KB to 100 MB; `.zip` 1 to 25 MB; `.pdf`, `.png`, `.jpg` at 1, 5 and 10 MB. 1 MB = 1,048,576 bytes.
`edge/`	Empty files, odd names, wrong extensions, truncated and corrupted files, invalid JSON/XML/CSV, password-protected ZIP and PDF
`encodings/`	The same text and the same CSV as UTF-8, UTF-8 with BOM, UTF-16 LE and BE, Latin-1, CRLF and CR line endings; mixed line endings; no final newline

Edge cases

https://files.placeholder.com/edge/empty.txt0 bytes (also empty.csv, and empty.zip: a valid ZIP with no entries)
https://files.placeholder.com/edge/file%20with%20spaces.txta space in the name
https://files.placeholder.com/edge/%C3%BCn%C3%AFc%C3%B6d%C3%A9-%E3%83%95%E3%82%A1%E3%82%A4%E3%83%AB-%E6%96%87%E4%BB%B6-%F0%9F%98%80.txtünïcödé-ファイル-文件-😀.txt; there is also a 240-byte file name
https://files.placeholder.com/edge/png-named.jpgPNG bytes with a .jpg extension (and jpeg-named.png, pdf-without-extension)
https://files.placeholder.com/edge/text-named.pdfplain text pretending to be a PDF
https://files.placeholder.com/edge/truncated.zipno central directory (also truncated PDF, PNG and JPEG)
https://files.placeholder.com/edge/corrupted-crc.zipdata doesn't match the CRC-32 (also corrupted.png)
https://files.placeholder.com/edge/invalid.jsontrailing commas (also invalid.xml, malformed.csv)
https://files.placeholder.com/edge/password-protected.zipZipCrypto, password placeholder
https://files.placeholder.com/edge/password-protected.pdfRC4-128, user password placeholder, owner password placeholder-owner

In index.json, deliberately broken files have "broken": true, files whose bytes don't match their extension have actual_type, and the protected files list their password.

index.json

{"name": "files.placeholder.com", "count": 107, "total_size": 296038390,
 "units": "Size labels use binary units: 1 KB = 1,024 bytes, 1 MB = 1,048,576 bytes.",
 "files": [
   {"path": "documents/sample.pdf", "url": "https://files.placeholder.com/documents/sample.pdf",
    "size": 28462, "sha256": "c488dcf5…", "media_type": "application/pdf",
    "category": "documents", "description": "PDF 1.4, 3 pages: text, table and an embedded JPEG."},
   {"path": "edge/png-named.jpg", …, "media_type": "image/jpeg", "actual_type": "image/png"},
   {"path": "edge/truncated.pdf", …, "broken": true},
   …]}

# Download a file and check it is the one you expect
curl -fsSLO https://files.placeholder.com/sizes/5mb.zip
sha256sum 5mb.zip   # compare with index.json

Behaviour

Content-Type follows the file extension, which is the point of the mislabelled files: png-named.jpg is served as image/jpeg.
Range requests work (206 Partial Content), so resumable downloads can be tested.
Files are cached for a year (immutable, no-transform, so proxies don't recompress them); index.json for five minutes.
CORS is open (Access-Control-Allow-Origin: *) and every response has X-Placeholder: files.placeholder.com.
The text inside the files is synthetic. These are files for testing, not content to ship. placeholder-check flags links to files.placeholder.com in a build.