Huge CSV / Generator
File System Access API

CSV test-file generator

Generate a synthetic, multilingual CSV file of any size up to 50 GB directly on your machine and stream it to disk. Use it to benchmark the CSV viewer, your own parser, or any CSV tooling. Nothing is uploaded — the bytes go straight to a file you pick.

Generate

streamed to disk, never buffered in memory
base schema is 13; pads with extra_N
% of rows with an embedded comma / quote / newline
runs in a background worker; typically disk-bound — a few hundred MB/s on a fast SSD.

        

        

Output format

An RFC 4180 CSV with a header row, then one record per line. The default schema:

id,uuid,first_name,last_name,email,country,city,signup_date,age,balance,active,bio,tags

Name, city, and free-text cells are drawn from curated word lists across the selected scripts (incl. Arabic, Hebrew, Japanese, Ukrainian), so the file exercises the full UTF-8 range. A configurable share of bio/tags cells embed the delimiter, doubled quotes, or a literal newline — quoted per RFC 4180 — to stress a parser's quoting across block boundaries.

How it works

The page uses the File System Access API (showSaveFilePicker) to open a writable stream, then hands the handle to a module Web Worker that streams ~1 MB chunks until the target size is reached. The next chunk is generated while the previous one writes, and there is no in-memory buffer of the whole file — the same path works for 1 KB and 50 GB. Curious how the matching parser reads it back so fast? See how the WASM parser works.

Browser support: Chrome, Edge, Opera, and other Chromium browsers. Safari and Firefox don't yet implement showSaveFilePicker — the generator shows an error there.