JSON formatter and validator

JSON syntax rules in brief

JSON (JavaScript Object Notation) is defined by RFC 8259, a short standard from the Internet Engineering Task Force. Almost every error comes from breaking one of these rules:

• A value is an object { }, a list (array) [ ], a string, a number, or one of true, false and null, always in lowercase.
• Strings and property names go in straight double quotes. Single quotes and curly quotes aren’t allowed.
• Inside a string, a double quote, a backslash and control characters such as line breaks and tabs must be escaped: \", \\, \n, \t, or \u with four hex digits.
• Commas go between items and between name-value pairs, never after the last one. A colon separates each name from its value.
• Numbers are plain decimals: no leading zeros (except 0 itself), no plus sign, no hexadecimal, and no NaN or Infinity.
• Spaces, tabs and line breaks between values are ignored. Comments are not part of JSON.
• A document holds exactly one value, usually an object or a list. Names should be unique within an object; when one repeats, programs disagree about which value wins.
• JSON shared between systems must be encoded as UTF-8, and shouldn’t start with a byte order mark.

Common errors and how to fix them

The messages below are the ones the checker above gives for each mistake, with the position it points to. Each browser words its own JSON errors differently and none suggests a fix, which is why this tool uses its own parser.

Lenient mode fixes the first five, plus curly quotes and unusual spaces copied from web pages, and lists what it changed. The rest need a decision from you, such as whether 02134 is a number or a code.

Formatting or minifying?

Formatted JSON is for people: one value per line, nested levels indented by 2 spaces, 4 spaces or a tab. Minified JSON is for machines: every space outside the strings removed. A small order record takes 266 bytes formatted with 2 spaces and 163 bytes minified, 39% less, and the saving grows with deeper nesting. Web servers often compress responses anyway, so minifying matters most for files you store or embed. The data is identical either way, and you can switch back and forth without losing anything.

The tree view is the quickest way to find your way around a long API response: open and close each level, see the type and size of every object and list, and copy the path to a value, such as items[1].sku, for use in code or a spreadsheet formula.

Big numbers and precision

JSON itself puts no limit on numbers, but RFC 8259 warns that whole numbers are only reliably exchanged between -9,007,199,254,740,991 and 9,007,199,254,740,991 (2⁵³ − 1), the range a 64-bit floating-point number holds exactly. Most formatters read your JSON into JavaScript numbers and write it back, so a long ID such as 12345678901234567890 comes out as 12345678901234567000. This one formats from its own parse of your text and writes {"id":12345678901234567890}, digit for digit, and it warns you when a number would be rounded by other programs. If you control the data, send long IDs as strings.

To see what changed between two versions of a file, paste both into text compare, after sorting the keys here so that the order doesn’t show up as a difference. The binary translator turns text into hex or binary bytes, and the case converter switches names between camelCase and snake_case.