Turn CSV into clean JSON in your browser. Works offline once loaded; your data stays on your device.
FeedbackTurn a spreadsheet export into a clean JSON array. The parser follows RFC 4180, so quoted fields containing commas, quotes or line breaks survive intact, and numbers, booleans and null are restored as real JSON types. Everything runs in your browser, so the file never leaves your device.
Paste your CSV, or copy it straight out of Excel, Google Sheets or Numbers.
Keep the first row as your header row: it becomes the keys of each JSON object.
Leave 'Detect numbers and booleans' on to get real JSON types, or turn it off to keep every value as text.
Press Convert, then copy the JSON or download it as a .json file.
CSV is what spreadsheets and analytics tools export, but APIs, config files and JavaScript all expect JSON. Converting between the two is the small chore that sits between a data export and actually using it in code.
A field like "Doe, John" keeps its comma instead of splitting into two columns.
42 becomes a number and true becomes a boolean, so you do not have to cast every value afterwards.
The conversion happens in your browser, so customer data or exports never reach a server.
Reading CSV is harder than it looks, and every difficulty in a CSV to JSON conversion comes from the CSV side. JSON has one specification and one meaning. CSV has a loose convention and a great deal of local variation, so the first job is working out what the file in front of you actually is.
Delimiters are the first question. Comma is the name of the format and semicolon is extremely common in European exports, because locales that use a comma as the decimal separator cannot also use it to separate fields. Tab-separated files are widespread in data work. A converter that guesses will usually guess right and is worth overriding when it does not, which shows up immediately as every row arriving as a single field.
Quoting is where correctness lives. A field containing the delimiter must be quoted, and a quoted field can contain the delimiter, newlines, and escaped quotes. This is why splitting a CSV on commas is wrong for any file containing an address or a free-text comment: the naive split breaks mid-field and every column after it shifts. A real parser tracks quote state, which is the difference between a conversion that works on your sample and one that works on the file.
The header row determines your JSON keys, so its quality is the output's quality. Duplicate column names, empty header cells, and names with leading or trailing spaces all produce keys you will then have to work around in code. It is worth fixing the header before converting rather than after.
Types are the decision that changes the result most. Everything in a CSV is text, so the converter must decide whether "123" becomes the number 123 or stays a string, and whether "true" becomes a boolean. Inferring types is convenient and it is how identifiers lose their leading zeros, how postcodes become integers, and how a column that is mostly numbers but occasionally contains a dash arrives with mixed types. Keeping everything as strings is lossless and puts the decision back on you, which for identifiers is usually the right place for it.
Encoding is the last trap. A CSV exported from a spreadsheet may be UTF-8, UTF-8 with a byte order mark, or a legacy regional encoding, and the symptom of getting it wrong is accented characters arriving as pairs of unrelated symbols. If names look mangled, the encoding was the problem rather than the conversion.
All of this happens in your browser. The file is read locally, parsed locally, and the JSON is produced locally, which matters because a CSV is very often a customer list, an order export or a payroll extract.
Convert JSON to CSV in your browser and download the file. Your data never reaches a server.
Format, validate and minify JSON in your browser. Paste, prettify, copy: nothing is uploaded.
Validate JSON and jump to the exact line and column of each error, locally in your browser.
The same tool answers on HTTP. Free, 1,000 calls a day, no card, and one token works across every Astral API.
POST https://astraljson.com/api/v1/csv-to-json