Using the converter
Paste JSON Lines, one JSON value per line, and they come out as a single JSON array, formatted for reading. Blank lines are skipped. If a line isn't valid JSON, the error says which line, so a broken record in a long log is quick to find.
Swap to go the other way: paste a JSON array and each item comes out on a line of its own, compact, ready to append to a .jsonl file.
What JSON Lines is
A JSON Lines file holds one complete JSON value on each line, usually an object. It's also called NDJSON, newline-delimited JSON. Because every line stands alone, a program can read the file a line at a time without loading the whole thing, and can add a record just by appending a line.
That makes it common for logs, data exports and machine learning datasets, where files grow long and get written a record at a time. A normal JSON array can't do that: it isn't valid until its closing bracket, so the whole file has to be read before any of it can be used.
A worked example
| Line | JSON Lines | Item in the JSON array |
|---|---|---|
| 1 | {"id":1,"tags":["new"]} | [{"id":1,"tags":["new"]}, |
| 2 | {"id":2,"note":"line one\nline two"} | {"id":2,"note":"line one\nline two"}, |
| 3 | null | null] |
Reading the example
The array is shown compact in the table; the converter formats it for reading. Each line is a whole JSON value on its own, so there are no commas between records and no brackets around them. The line break inside the second note is written as \n, because a real line break would end the record. Line 3 is just null: any JSON value is allowed on a line, not only objects, though objects are by far the most common.
Put a stray comma after the last field of a record, as in {"id": 3, "name": "Katherine",}, and the converter stops and names line 3. In a log with thousands of lines, that's the line to fix.
JSON Lines and NDJSON side by side
| JSON Lines (jsonlines.org) | NDJSON (ndjson-spec) | |
|---|---|---|
| Each line | One valid JSON value | One JSON text, per RFC 8259 |
| Line ending | \n, and \r\n works too | \n, which may follow a \r |
| Blank lines | Not valid | A parser may skip them, if it says so |
| File extension | .jsonl (.jsonl.gz when compressed) | .ndjson |
| Media type | application/jsonl, not yet standardized | application/x-ndjson |
| Text encoding | UTF-8 | UTF-8, through RFC 8259, which requires it between systems |
Questions
What's the difference between JSONL and NDJSON?
In practice, nothing: both mean one JSON value per line, separated by newlines. The converter reads either.
Can a JSON Lines record span several lines?
No. Each record has to fit on one line, so any line breaks inside a string are written as \n.
Why won't my JSON convert to JSONL?
JSON Lines is a list, so the JSON has to be an array, starting with [. To turn a single object into one line, wrap it in [ and ] first.
Why does the converter skip blank lines when JSON Lines says they aren't valid?
A blank line usually means a trailing newline at the end of the file or a gap left by hand editing, not a lost record, so stopping on it would only get in the way. The NDJSON spec allows a parser to skip them as long as it says so, and this one does. The JSON to JSONL direction never writes a blank line.
Sources
Added . Last updated and reviewed . What's new






