CSV to JSON Converter

About this tool

What CSV is, and where files differ

RFC 4180 describes CSV as records separated by line breaks and fields separated by commas. A field that contains a comma, a quotation mark or a line break is enclosed in double quotes, and a quotation mark inside it is written twice. The format has no types, so every value is text, and a file cannot say which delimiter or which character encoding it uses.

Real files vary. A spreadsheet program set to a region that writes decimals with a comma usually exports with semicolons, database tools often export with tabs, and line endings can be CRLF, LF or CR. This tool reads all of these, removes a byte order mark, skips blank lines and keeps a stray quotation mark inside an unquoted field as text.

How CSV becomes JSON

With Auto-detect, the tool splits the first 20 rows with a comma, a semicolon, a tab and a pipe, and scores each one by the number of columns it gives and by how many rows have that same number. A delimiter that leaves quotation marks in the middle of a field scores lower. The preview shows the result, so a wrong guess is easy to see and to correct by hand.

With Infer types, an empty field becomes null, true and false become booleans, and a plain decimal number becomes a JSON number written exactly as it was, so 49.90 stays 49.90. Anything else stays a string, including values that only look like numbers: 007, +39 and 1e5 are more often codes than quantities.

A row with fewer or more fields than the header is still converted, and its line number is listed. In the array of objects, the missing keys are left out and the extra fields are named column_N after their position. A header cell that is empty gets the same kind of name, and a repeated name gets a suffix such as _2, so that no value overwrites another.

How JSON becomes CSV

An array of objects becomes one row for each object. The header is the union of all the keys, in the order in which they first appear, and an object that lacks a key gets an empty field. A nested array or object is written into its cell as JSON text, and null becomes an empty field. An array of arrays is written row by row as it is.

Fields are quoted only when they contain the delimiter, a quotation mark or a line break. The downloaded file separates the records with CRLF, as RFC 4180 specifies.