JSON Formatter & Validator

Common JSON syntax errors and how to fix them

RFC 8259 allows only objects, arrays, double-quoted strings, numbers, true, false and null, so text that works in a .js file often fails here. Each item shows a small input and the exact message you'll get.

• Trailing comma: [1, 2,] gives Trailing comma before "]" (line 1, column 7). Delete the comma after the last item.
• Single quotes: {'a': 1} gives Expected a property name in double quotes, found character "'" (JSON strings need double quotes) (line 1, column 2). Use double quotes for keys and strings.
• Unquoted keys: {a: 1} gives Expected a property name in double quotes, found character "a" (line 1, column 2). Quote every key.
• Comments: {"a": 1 // note} gives Expected "," or "}" after property value, found character "/" (line 1, column 9). JSON has no comments, so remove them or move the note into a field.
• Values from other languages: NaN, Infinity, undefined, and Python's True and None all fail. {"a": NaN} gives Unknown value "NaN". Text must be in double quotes (only true, false and null have no quotes) (line 1, column 7). Use null, a string, or lowercase true and false. For a Python dict, use json.dumps instead of print.
• Leading zeros and hex: 007 gives Invalid number, and 0x1F is rejected too. JSON numbers are decimal only.
• Two values in one paste: copying two lines from a JSON Lines (NDJSON) log gives Unexpected character "{" after the end of the JSON value (line 2, column 1). Format one line at a time, or wrap them in [ ] with a comma between them.
• Byte order mark: a file saved as "UTF-8 with BOM" starts with an invisible character, and you get Unexpected invisible character U+FEFF (line 1, column 1). RFC 8259 says JSON writers must not add a BOM, so save the file as plain UTF-8.
• Windows paths: {"path": "C:\Users"} gives Bad escape "\U" in string (line 1, column 13). A backslash starts an escape, so write C:\\Users or use forward slashes.

Why many formatters change big numbers

A typical online formatter runs JSON.parse and then JSON.stringify. JSON.parse stores every number as an IEEE 754 double, so the largest integer it holds exactly is 9007199254740991 (2^53 - 1). Larger ones are rounded: 9007199254740993 becomes 9007199254740992. The same round trip turns 59.90 into 59.9 and silently drops the first of two duplicate keys.

That's a real problem for 64-bit database IDs, social media post IDs and payment amounts. RFC 8259 calls only integers from -(2^53 - 1) to 2^53 - 1 safe to exchange, so if you design the API, send bigger IDs as strings.

Other languages differ. Python's json module keeps big integers exact but prints 59.90 as 59.9. PHP's json_decode turns them into floats unless you pass JSON_BIGINT_AS_STRING. jq 1.7 and later keep number literals as written, but arithmetic on them still uses doubles.

Here, Format and Minify work on your original text, so nothing is rounded. Sort keys is the one option that parses the JSON into JavaScript values, and the note under it tells you when that rounds a number.

JSON, JSON5 and JSONC

A file with comments or trailing commas on purpose is usually JSON5 or JSONC; VS Code's settings.json and TypeScript's tsconfig.json are JSONC. Strict JSON parsers, this one included, reject both. For a config format that allows comments, convert the data with JSON to YAML.

Format and validate JSON in code

Each snippet parses the JSON, which throws or returns an error when it's invalid, and prints it with indentation. python3 -m json.tool and PHP's JSON_PRETTY_PRINT use 4 spaces by default. Go's json.Indent works on the raw bytes, so like this page it keeps numbers as written. AI assistants can call the same formatter as format_json and validate_json through the toolo MCP server.

const pretty = JSON.stringify(JSON.parse(text), null, 2);
// JSON.parse throws a SyntaxError for invalid JSON

import json
pretty = json.dumps(json.loads(text), indent=2)

python3 -m json.tool --indent 2 data.json

jq . data.json      # pretty-print with 2 spaces
jq -c . data.json   # minify
jq -S . data.json   # sort keys

var out bytes.Buffer
if err := json.Indent(&out, data, "", "  "); err != nil {
    return err // not valid JSON
}

// From a Go value instead of raw bytes:
b, err := json.MarshalIndent(v, "", "  ")

$pretty = json_encode(
    json_decode($text, flags: JSON_THROW_ON_ERROR),
    JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES
);