Large JSON to CSV with predictable memory use.

The difficult example: huge identifiers and exponent notation

Large exports often carry database IDs that exceed JavaScript's exact integer range, plus exponent notation from metrics systems. jsonnorm keeps these numeric values as written instead of rounding or reformatting them.

[
  {"id":9007199254740993,"metrics":{"cpu":5e-324},"message":"batch-1"},
  {"id":18446744073709551615,"metrics":{"cpu":1e-300,"disk":"99%"},"message":"batch-2"}
]

id,metrics.cpu,message,metrics.disk
9007199254740993,5e-324,batch-1,
18446744073709551615,1e-300,batch-2,99%

Measured boundary: The engine has been tested with multi-gigabyte JSON files. The web converter accepts files up to 100 MiB.

What “large-file support” means here

• Large inputs: JSON is processed as it is read, so the entire document does not need to be loaded at once.
• Complete headers: fields that appear later are included in the CSV header.
• Web limit: each upload to the web converter can be up to 100 MiB.