CSV to JSON

CSV is what a spreadsheet gives you when you export it, and it is how most systems hand over a table of data. It is simple and it is everywhere. JSON is what almost every modern API, script and front end wants to read instead, because it carries the shape of the data along with the values. So turning one into the other is a small job that turns up all the time, and it is dull to do by hand.

The useful part is doing it without surprises. A good converter keeps every value exactly as it was, reads quoted fields properly, and tells you plainly when a file is not what it claims to be. That is what this page is built to do.

With the header option on, the first row is read as the names of the columns. Every row after it becomes one object, and each cell sits under the name of its column. A file of three data rows gives you a list of three objects, which is exactly the shape most code wants to loop over.

Turn the header option off when the file has no header of its own. Each row then becomes a plain list of values in the order they appear, so nothing is invented and no cell is mislabelled. Blank lines are skipped either way, so a stray empty row at the end of a file never becomes an empty entry.

Most CSV work is not one file, it is a stack of them: a month of exports, a report per region, a file per client. So you can drop many files onto the upload area, or pick them all in one go. Each file is converted on its own and named after the file it came from, with a .json ending in place of the old one.

When the batch is done you can save each result on its own, or download the whole set as a single zip. Files are never merged, so ten CSV files give you ten JSON files. If one file in the set cannot be read, it is flagged on its own and the rest still convert, so one bad file never sinks the whole job.

Because guessing here destroys real data. A postcode with a leading zero becomes a smaller number. A phone number loses its leading zero entirely. An order reference like 007 becomes 7. A version like 1.10 becomes 1.1, which is a different version. Every one of those is a real bug that has shipped somewhere because a converter was helpful without being asked.

So the default is to leave every value exactly as the CSV had it. When you know your values are truly numeric, tick the option. Numbers and true or false then become real JSON types, while anything that would change on the trip, like 0042 or 18.50, stays as text. That way the output is safe to trust.

Auto-detect looks at the first row and picks the separator that splits it most cleanly, which is right almost every time. Files from other countries often use a semicolon, and exports from some tools use a tab or a pipe, so you can set any of those by hand when you want no guessing at all.

Quoted fields are read the way the format intends. A value wrapped in quotes can hold the delimiter itself, a line break, or quote marks of its own, and it is still read as one field rather than split in two. That is what keeps an address with a comma in it, or a note that runs onto a second line, in one piece.

Yes, and it matters more here than for most tools. The CSV people need to convert is rarely a toy file: it is a real customer export, a payroll run, a list of orders, or a statement from a bank. Many free converters post whatever you upload to their own server, which means all of that ends up in someone else’s logs.

This page parses and converts in your browser, so the file never leaves your device. For related work that runs the same way, tidy your output with the JSON minifier, convert markup with XML to JSON, or see the rest of the developer tools.