JSON to CSV Converter

How the conversion works

The input must be a JSON array, so it starts with [ and ends with ]. Each item in the array becomes one row. For an object, each key becomes a column. The first row of the CSV holds the column names.

This is the example shown in the input box:

[
  {"name": "Alice", "age": 30, "address": {"city": "New York", "zip": "10001"}},
  {"name": "Bob", "age": 25, "address": {"city": "Chicago", "zip": "60601"}}
]

Convert to CSV gives this, and the status line says 2 rows, 4 columns:

name,age,address.city,address.zip
Alice,30,New York,10001
Bob,25,Chicago,60601

Rows do not need the same keys. The tool collects every key from every row, in the order it first sees them. A row without a key gets an empty cell. For example, [{"a":1,"b":2},{"b":3,"c":4}] becomes:

a,b,c
1,2,
,3,4

Nested objects and arrays

Nested objects are flattened at any depth. The keys are joined with a dot, as in address.city above. Arrays are a different story. The tool does not split them into rows or columns. It writes the whole array into one cell as compact JSON, so no data is lost and you can read it back with any JSON parser. This table shows what you get for each kind of value. The middle column is what a spreadsheet shows in the cell:

If you want one row per item of an inner array, such as one row per order line, reshape the data before you convert. Make one object per line item and copy the order ID into each one.

Items in the top-level array do not have to be objects. A number, a piece of text, true, null or an array goes into a single column called value. So [1, 2, "x"] gives:

value
1
2
x

Commas, quotes and line breaks in values

CSV has one main rule for special characters, described in RFC 4180. A value that has a comma, a double quote or a line break is wrapped in double quotes. Each double quote inside it is written twice. The tool follows this rule. This input:

[{"note": "He said \"hi\", then left", "multi": "line1\nline2", "plain": "ok"}]

gives this output. The second value really spans two lines:

note,multi,plain
"He said ""hi"", then left","line1
line2",ok

Two small differences from RFC 4180: rows end with a line feed only, not a carriage return plus line feed. Spreadsheet apps accept both. The separator is always a comma. There is no option for semicolons or tabs.

Protection against spreadsheet formulas

Excel, Google Sheets and LibreOffice treat a cell that starts with =, +, - or @ as a formula. This is called CSV injection or formula injection. If your JSON came from other people, for example a form or an API, a value such as =HYPERLINK("http://example.com","Click") could become a working link, and some formulas can fetch data or run commands after a warning.

The option Protect against spreadsheet formulas is on by default. It puts a single quote ' in front of any text cell that starts with =, +, -, @, a tab or a carriage return. Then the spreadsheet shows the value as plain text. Column names are checked too. Real JSON numbers are not changed, so -5 stays -5. But the text "-5" or a phone number such as "+44 20" gets the quote.

The quote is written into the file. Some apps hide it and some show it as part of the text. If your data is your own and you need the exact values, for example to load the CSV into a database, turn the option off. The output updates at once when you change it.

Opening the file in Excel

Download CSV saves a file called data.csv in UTF-8. The option UTF-8 BOM for Excel is on by default. It adds a byte order mark (BOM) to the start of the downloaded file. A BOM is a small hidden marker that says which encoding the file uses. Copy CSV never includes it. Excel is where most problems appear:

• Accented letters look wrong. Without a BOM, Excel on Windows may read a double-clicked UTF-8 file with an older encoding. München can then show as München. With the BOM option on, Excel reads it as UTF-8. Google Sheets, LibreOffice and Numbers read the file fine either way. Turn the option off if another program sees the BOM as part of the first column name.
• Everything lands in column A. In regions that use a comma as the decimal sign, Excel often expects semicolons. Import with From Text/CSV and pick comma as the delimiter.
• Leading zeros disappear. The tool writes 0012 as it is, but Excel turns it into 12. Long ID numbers can turn into scientific notation. Set those columns to Text when you import.
• Values that start with =, +, - or @. Excel treats them as formulas. With the protection option on, the tool writes =1+1 as '=1+1, so it stays text. See the section above.

Other things to know

• Large numbers are rounded. The tool reads JSON with the browser's JSON.parse, which stores numbers exactly only up to 9,007,199,254,740,991. 12345678901234567890 comes out as 12345678901234567000. Keep long IDs as strings in quotes.
• Number-like keys come first. {"b":1,"2":2,"a":3,"1":4} gives the header 1,2,b,a. Other keys keep their order.
• A key with a dot can clash. In {"a.b": 1, "a": {"b": 2}}, both would give the column a.b. The tool keeps both. The first is a.b and the second is a.b_2, and an orange note under the status line tells you about it.
• A single object is rejected. {"a": 1} shows Input must be a JSON array. Put square brackets around it.

If the input is not valid JSON, the status line shows the browser's error message. Our JSON Formatter page explains the common errors and how to fix them.