JSON to CSV, YAML & XML Converter

This converter turns JSON into CSV, YAML or XML, and turns each of those back into JSON. It is for developers, analysts and anyone who has data in one format and a tool that only reads another.

Paste or open a file, pick a direction, and the result appears as you type. Nested objects become dot-path columns like address.city, so an API response lands in a spreadsheet with one column per field. Going the other way, those dot paths can be rebuilt into nested JSON. If your input has a syntax error, you get the line and column instead of a vague failure. Need to clean it up first? Format and validate the JSON and come back.

Everything runs in your browser. Exports from a CRM, a payment system or a user table often hold personal data, and none of it is sent anywhere.

CSV is flat: rows and columns, nothing inside anything. JSON is a tree. To make them meet, each path through the tree becomes one column name. A record like {"address": {"city": "London"}} becomes a column called address.city with the value London.

Arrays are the tricky part. With Columns, a tags array with two items fills tags.0 and tags.1, and a row with only one tag leaves tags.1 empty. With JSON text, the whole array sits in one cell as ["math","poetry"]. Columns suit filtering in a spreadsheet. JSON text keeps the column count down when arrays are long.

To go back, turn on Rebuild nesting from dot.paths. The converter splits each header on dots, and numeric parts become array positions. Empty cells under a numbered column are skipped, so a short array stays short instead of growing null items.

CSV has no types. RFC 4180 describes only fields, commas, quotes and line breaks, so every value is text until a program guesses otherwise. Spreadsheets guess aggressively. They drop the leading zero from a ZIP code like 02134 and show a 16-digit card or order number in scientific notation, rounded to 15 significant digits.

This tool writes exactly the text that was in your JSON, so the CSV file itself is correct. The guessing happens when the spreadsheet opens it. To keep codes intact, import the file with the column type set to Text instead of double-clicking it.

The same problem runs in reverse. When converting CSV to JSON with type detection on, this tool only turns a value into a number if printing that number gives back the exact same text. If a column of ZIP codes like 10001 should stay strings, turn type detection off.

Almost. Since YAML 1.2 (2009), nearly every JSON document is also valid YAML, which is why YAML to JSON and back usually round-trips. The differences show up in the other direction. YAML has comments, anchors and aliases (&base and *base), multiple documents in one file, and plain unquoted strings.

Unquoted strings cause the classic surprises. In the older YAML 1.1 rules, NO, yes, on and off are booleans, so a country code list could turn Norway into false. This converter reads YAML with the 1.2 core schema, where those stay text. When writing YAML, it quotes any string that would be misread, like 'true' or '10001'.

JSON has no attributes, so a naming convention marks them. By default, an attribute id="bk101" becomes the key @_id, and text next to child elements goes into #text. You can change the prefix, or drop attributes when you only care about element content.

Repeated elements become arrays. Two <book> elements turn into "book": [ {...}, {...} ], while a single <book> stays an object. Code that reads the JSON should handle both shapes. When building XML from JSON, keys that are not valid XML 1.0 names are fixed: spaces become underscores and names that start with a digit get an underscore in front.