Convert Excel to JSON

JSON with types, not everything in quotes

The usual conversion — copy the table and wrap every cell in quotes — produces JSON where 1500 is the string "1500" and true is the string "true". Whoever consumes it has to convert it back, and a formatting mistake in one forgotten row only shows up much later.

Here every cell is emitted with the type it had in the sheet:

• A number comes out as a JSON number, unquoted and without thousands separators.
• A boolean comes out as true or false.
• A date comes out as a YYYY-MM-DD string, which any ISO 8601 parser understands.
• An empty cell comes out as null.
• Everything else comes out as a string.

Objects or arrays

The shape option decides the structure of the result:

• Objects produces an array of objects, one per row, using the first row's titles as property names. This is the natural choice for feeding an API or a map() in JavaScript.
• Arrays produces an array of arrays: each row is a list of values in column order. It is considerably smaller and works when the consumer already knows what each position means.

Titles are used verbatim, with no renaming or normalising. If two columns share a name, the JSON will contain two properties with the same name and most readers will keep only the last one, so it is worth checking the header row first.

Indentation, empty cells and accents

Pretty print is on by default and indents the JSON so you can read and review it. Turn it off and everything comes out on one line, which is what you want when the file goes straight to a service and size matters.

Skip empty values omits blank cells entirely instead of writing them as null. It only applies to the "objects" shape: in the "arrays" shape positions cannot be skipped without shifting the columns.

Accented characters are written directly rather than as ó escapes, and the file is downloaded as UTF-8 with a BOM. Both forms are valid JSON, but this one can be read without decoding it.