CornerTools

CSV to JSON Converter

Convert CSV data to JSON with support for formatting, type inference, and clean output.

1
1
Rows: 0Characters: 0
JSON format

Converts CSV — exported from Excel, Google Sheets or a database — into a JSON array of objects, with numbers and booleans typed correctly. Conversion runs in your browser; the data is not uploaded.

Options

  • CSV delimiter — left empty, it is detected from the data and named under the field. To set it, type a single character such as ; or |, or \t for a tab.
  • Header included — on by default: the first row names the keys. Off, keys are column, column_2 and so on; repeated header names get the same numbering.
  • Trim values — removes spaces around each value; on by default.
  • Auto type inference — on by default (see below). Off, every value stays a string.
  • Use null for empty values — empty cells become null instead of "".
  • Nest dotted headers — off by default. On, a header such as address.city puts its value inside an address object, the reverse of flattening in the JSON to CSV converter. A header stays as written when it has an empty part (a..b) or when another column already holds a plain value under the same key (a beside a.b). Parts that are numbers are keys, not array positions.
  • JSON format — pretty with 1 to 8 spaces of indent, or minified.

Download saves the output as data.json.

How values are read

Detection tries a comma, a semicolon, a tab and a pipe on the first 20 rows and takes the one that splits every row into the same number of values, more than one; delimiters inside quotes do not count. When two qualify, the one giving more columns wins, then the earlier in that list. For a file with uneven rows it takes the one that splits the first row furthest, and a comma when nothing splits it. A first line of sep=;, which Excel writes to name the delimiter, sets it and is not read as a row.

Values in double quotes may contain the delimiter and line breaks, and a doubled quote inside stands for one quote. A quote opens a quoted value only at the start of a value, after any spaces; anywhere else, as in 5" disk, it is an ordinary character. A quoted value that is never closed is reported with the line it opened on. Blank lines between rows are skipped, and a byte order mark at the start of the file, as Excel writes, is ignored. Extra values beyond the header's columns are dropped; missing ones become empty.

With type inference, true, false and null become JSON values, plain numbers become numbers written digit for digit — long IDs are not rounded — and values that are valid JSON arrays or objects are parsed. Numbers with a leading zero, like ZIP codes and phone numbers, stay strings so the zeros are kept.

Examples

Quotes, types and empty cells

With the default options:

CSV
id,name,zip,active,score,tags
1,"Sharma, Anna",0012,true,4.5,"[""a"",""b""]"
2,Ben,,false,,
JSON
[
  {
    "id": 1,
    "name": "Sharma, Anna",
    "zip": "0012",
    "active": true,
    "score": 4.5,
    "tags": [
      "a",
      "b"
    ]
  },
  {
    "id": 2,
    "name": "Ben",
    "zip": "",
    "active": false,
    "score": "",
    "tags": ""
  }
]

A value across two lines

CSV
id,address
1,"12 High Street
London"
JSON
[
  {
    "id": 1,
    "address": "12 High Street\nLondon"
  }
]

Three lines of CSV, two rows: the line break inside the quotes belongs to the value, written \n in JSON.

Semicolons and dotted headers

With Nest dotted headers on and the delimiter left empty:

CSV
name;price;address.city;address.pin
Anna;1,5;Pune;411001
JSON
[
  {
    "name": "Anna",
    "price": "1,5",
    "address": {
      "city": "Pune",
      "pin": 411001
    }
  }
]

The semicolon is detected: a comma would split only the second row. 1,5 stays text, since a decimal comma is not a JSON number.

See also