Upload handlers fail on names, not contents: a space, an accent, a 200-character stem, a leading dot that hides the file. Turn on awkward names and a share of your batch arrives shaped exactly like the ones that cause support tickets.
What you can control
- Six naming conventions from kebab-case to content hashes.
- Awkward mode produces spaces, accents, repeated dots, very long stems and leading dots.
- File sizes are skewed toward the middle range with occasional large outliers, as real uploads are.
- MIME types are mapped correctly per extension, for testing content-type validation.
What this is not
These are names and metadata only — no actual files are produced. Use them to test naming, storage keys and validation rather than upload throughput.
Questions
Why test awkward file names?
Because they are the ones that reach production. A name with a space breaks a shell script, an accent breaks a badly configured storage key, and a leading dot hides the file on Unix.
Are the MIME types accurate?
Yes, mapped from a standard extension-to-type table. Never trust a client-supplied MIME type in production — sniff the file contents instead.
What is the content-hash convention for?
Content-addressed storage. Naming a file by the hash of its contents means identical uploads deduplicate automatically.