What Is NDJSON?
TOC
Definition
Also called JSON Lines (JSONL) or Newline Delimited JSON (NDJSON), this format is a text format ( not a transport protocol ) for storing structured data.
Each line consists of a valid JSON value, separated by a newline character.
The format is designed to store and process structured data conveniently, one record at a time, which makes it ideal for streaming, logging, and append-only data scenarios.
That may not land yet, so here is what a live log screen built from mock data would look like.
Demo
mock-data.ndjson
{"id":1,"type":"click","target":"button#login","time":"10:01:02"}
{"id":2,"type":"navigate","target":"/dashboard","time":"10:01:05"}
{"id":3,"type":"input","target":"input#search","time":"10:01:08"}
{"id":4,"type":"click","target":"button#logout","time":"10:01:12"}
{"id":5,"type":"navigate","target":"/login","time":"10:01:15"}
mock-data.json
[
{ "id": 1, "type": "click", "target": "button#login", "time": "10:01:02" },
{ "id": 2, "type": "navigate", "target": "/dashboard", "time": "10:01:05" },
{ "id": 3, "type": "input", "target": "input#search", "time": "10:01:08" },
{ "id": 4, "type": "click", "target": "button#logout", "time": "10:01:12" },
{ "id": 5, "type": "navigate", "target": "/login", "time": "10:01:15" }
]
Characteristics
- You cannot wrap the whole thing in brackets ([ and ]) and parse it as a JSON array.
- There are no commas between lines.
- Each line stands on its own.
- Unlike CSV, every row is a complete JSON object, so nested objects, arrays, and every JSON data type are fully supported.
Rules
- Each line must be a valid JSON value. The most common type is a JSON object, but arrays, strings, numbers, booleans, and null are valid too.
- Each line must be separated by the newline character \n.
- The file must use UTF-8 encoding.
- Do not use commas between lines.
- Each JSON value must fit on one line. (Spreading it across several lines is not allowed.)
- It's best not to use blank lines.
When would you use it?
- Live log screens
- Streaming AI responses ( ChatGPT )
- Large volumes of data