Convert CSV records to NDJSON for import

Create one JSON object per CSV record for a line-oriented import. Preserve identifiers as strings and validate headers and record widths before export.

Check the destination contract

Use this when an importer expects one JSON object per line rather than a JSON array. This output does not include Elasticsearch action metadata or database commands. Confirm the required field names and types with the destination before importing.

Strings stay strings

CSV id,active with record 00123,true becomes {"id":"00123","active":"true"}. Empty cells become empty strings, not null. Dates, booleans and numbers are never guessed; apply any required schema conversion separately.

Quoted newlines remain valid JSON

The existing CSV parser handles quoted delimiters, quotes and embedded line breaks. JSON escaping keeps each generated object on one physical line. Record numbers count CSV records, which can differ from physical source line numbers.

Reject ambiguous records

Duplicate or blank headers and records with missing or extra cells stop the conversion. Choose comma, semicolon or tab explicitly. At most 2 million input characters, 10,000 data records and 200 columns are accepted.

Review before importing

Download the .ndjson file and parse sample lines in your destination’s test workflow. Header-only input produces zero objects and is reported as such. This tool neither connects to an importer nor uploads or executes records.

Related existing tool →