Convert CSV rows to predictable JSON objects

Treat the header, delimiter and quoting rules as part of the data format.

Choose the actual delimiter

Comma, semicolon and tab are supported. Select the character that separates your source columns. The tool does not guess based on your computer’s regional settings.

The first row becomes the object keys. Headers must be non-empty and unique. Every following row must contain the same number of cells; a missing final cell is different from a row ending with a delimiter.

name,city,note
Ada,London,"one, two"
Jane,İstanbul,"said ""hello"""

Quoting protects field content

A delimiter or line break inside a cell needs a quoted field. An inner quotation mark is represented by two quotation marks. Quoted multiline fields remain one cell and are preserved in the output string.

The parser accepts LF and CRLF row endings and removes an initial UTF-8 byte-order mark. It rejects an unfinished quoted field and text after a closing quote that is not a delimiter or row ending.

Cells remain strings

A CSV cell containing 0012 remains the string "0012". The converter does not guess numbers, dates, booleans or null. This avoids silently removing leading zeroes or changing date formats.

If an application needs typed values, perform an explicit conversion using its schema after inspecting the JSON. Empty CSV cells become empty strings, not null or absent fields.

Check a complete round trip

Download the JSON and verify that the number of objects equals the number of data rows. Compare a cell with a delimiter, a Unicode value and an empty cell. Do not count embedded newlines as extra records.

Convert the JSON back to CSV if needed. With formula protection enabled, risky text values may gain an apostrophe. That safety transformation means byte-for-byte round-trip equality is not promised.