Convert to NDJSON

Convert CSV, INI, JSON, TOML, TSV, XML, or YAML to newline-delimited JSON. Paste the source or open a local file, then choose the parser settings. The conversion runs in your browser and does not upload the input.

How conversion works

Link

A top-level array becomes one compact JSON value per line. A top-level object or primitive value produces one line. The converter does not wrap the output in square brackets or add commas between records.

CSV and TSV tables become one object per line when the first row contains headers. Dots in a header create nested properties. Columns with the same path are merged into one object. Without headers, each row becomes a JSON array. Type inference can turn recognised numbers, booleans, and null values into JSON types instead of strings. CSV also supports a custom delimiter.

JSON arrays are split across lines. JSON objects and primitive values stay on one line. TOML tables, INI sections, and YAML mappings become a single JSON object unless their parsed root is an array or sequence. XML elements become object properties, repeated elements become arrays, and attributes are kept under the configured attributes key.

NDJSON output

Link

NDJSON is also called JSON Lines or JSONL. Each line is a complete JSON value that a parser can read without looking at adjacent lines.

{"id":1,"name":"John","active":true}
{"id":2,"name":"Jane","active":false}

The output uses compact JSON without indentation. Strings and property names use double quotes. Reserved characters inside strings are escaped according to JSON rules. The converter inserts a line feed between values and does not add an extra line feed at the end.

Records and top-level values

Link

NDJSON commonly contains objects with a similar shape, but the format itself accepts any JSON value on each line. This converter can therefore write arrays, strings, numbers, booleans, and null as records too. Some import tools expect every line to be an object, so check the requirements of the software that will read the file.

An empty top-level array produces empty output. Empty CSV or TSV input also produces no records. A single object produces one line rather than one line per property.

Where NDJSON is used

Link

NDJSON works well for logs, event exports, command-line pipelines, bulk API requests, and datasets processed one record at a time. Software can read a line, process the JSON value, and continue without parsing one large outer array. Appending a record is also straightforward because existing records do not need a closing bracket or trailing comma change.

This browser tool still reads the complete source and output into memory. It does not stream files line by line, so use a streaming program for exports too large to fit comfortably in a browser tab.

What may be lost

Link

Conversion preserves values that JSON can represent, but it does not preserve comments, whitespace, or source formatting. TOML and YAML date-like values pass through the parsers’ supported data model. YAML anchors, aliases, tags, and scalar styles are discarded. XML comments, processing instructions, namespaces, mixed content, and the distinction between CDATA and ordinary text cannot be reconstructed.

CSV and TSV have no built-in type information. Without type inference, 25, true, and null remain JSON strings. With inference enabled, text that matches a supported JSON primitive becomes that type. Large integers pass through JavaScript, which cannot exactly represent every integer above 9,007,199,254,740,991.

Files, privacy, and saved work

Link

Select a local file to load it into the converter. A recognised extension selects the matching input format. The browser reads the file locally and does not upload it.

The selected format, each format’s input, and parser settings are stored in localStorage. Switching formats does not overwrite another format’s text. Clear the fields or site data after handling sensitive information on a shared device.

Link

Widget made with staark