JSONL to CSV Converter

updated 25 April 2026

100% client-side. Your file never leaves your browser. No upload.

Before you start

You need one of the following:

• A .jsonl or .ndjson file (JSON Lines) on your computer, or
• JSONL text you can paste directly into the input pane.

Every line must be a valid JSON object. If a line is a string, a number, or a standalone array, my converter will flag it as an error and skip it. Because CSV is a table format, I need objects so I can map keys to column headers.

I use dot-notation to flatten your data. If you have a nested object like {"user": {"id": 1}}, it becomes a column named user.id. There is no hard file size limit, but since this runs entirely in your browser's memory, 100 MB is usually the "safe" ceiling before things get laggy.

How to use it

1. Paste your JSONL into the left pane, or drop a .jsonl file onto the drop zone.
2. Pick a Delimiter (comma is the default, but semicolon or tab is often better if your data contains lots of long-form text).
3. Click Convert. My script will scan every line to build a master list of all possible columns.
4. Review the output on the right. If any lines were malformed, I'll list the specific line numbers and errors at the bottom.
5. Click Copy or Download .csv to save your flattened table.

Options explained

Delimiter

This sets the character that separates your columns. While comma is the standard, Tab (TSV) is often safer for complex JSON data because it's much less likely to appear inside your actual values, making the resulting file easier for tools like Excel or pandas to parse without quoting headaches.

Example

Input (JSONL):

{"id": 1, "user": {"name": "Ada"}, "tags": ["admin", "dev"]}
{"id": 2, "user": {"name": "Linus"}, "notes": "Active"}

Output (CSV):

id,user.name,tags.0,tags.1,notes
1,Ada,admin,dev,
2,Linus,,,Active

Notice how the headers are a union of all keys found in the file, and nested arrays are expanded into indexed columns like tags.0.

Tips & common pitfalls

• Column Inference: I have to read the entire input before I can write the first line of CSV. This is because I need to know every possible key that exists in any row to build the header. If you have a 500 MB file, this will be slow.
• Dot-notation for nesting: Nested objects are always flattened. {"a": {"b": 1}} → a.b. If you have deeply recursive structures, your CSV might end up with hundreds of columns.
• Sparse Data: If only one record out of a thousand has a specific field, that column will still be created for every row, but it will be empty for the other 999 records.
• Arrays to Columns: Arrays are flattened by index (0, 1, 2). This works great for short, fixed-length lists but creates "column soup" if you have huge arrays.

Troubleshooting

Related tools

• JSONL ViewerInspect and search your records before flattening them.
• JSONL ValidatorFind and fix broken JSON lines that are failing the conversion.
• JSONL ↔ JSON ArrayConvert to a standard JSON array instead of a flat table.
• JSONL FormatterClean up your JSONL indentation or minify it for smaller file sizes.

See also: if you need to do something adjacent on this site, try JSONL to JSON to convert JSONL into a JSON array (or back), Formatter to pretty-print or minify each JSONL record, or OpenAI Fine Tune Validator to validate an OpenAI fine-tune file against the chat schema.

Frequently asked questions