JSON5 / JSONC ⇄ JSON Converter

About this tool

JSON5 / JSONC ⇄ JSON Converter takes the relaxed JSON people actually write in config files — tsconfig.json, VS Code settings.json, Babel and ESLint configs, .jsonc fixtures — and turns it into strict RFC 8259 JSON that any parser will accept. It also runs the other way, rewriting plain JSON as JSON5 with unquoted keys and single quotes.

The parser reads the full JSON5 grammar, which is a superset of both JSONC and strict JSON:

• Comments — // line and /* block */, anywhere whitespace is allowed.
• Trailing commas — after the last element of an array or the last member of an object.
• Unquoted keys — bare identifiers such as port: instead of "port":.
• Single-quoted strings — 'a' as well as "a".
• Line continuations — a backslash at end of line joins the next line into the same string.
• Extra escapes — \x41, \v, \0, and \a-style escapes that stand for the character itself.
• Loose numbers — hexadecimal (0xff), leading-dot (.5), trailing-dot (5.), leading-plus (+1) and leading-zero (007) literals.
• Non-finite literals — NaN, Infinity, -Infinity.

Everything JSON5-only is normalized on the way out: numbers become plain decimal literals, keys and strings get double quotes, comments and trailing commas disappear.

Worked example

JSONC input:

{
  // the dev server port
  port: 8080,
  /* only these hosts are allowed */
  hosts: ['a', 'b',],
}

Strict JSON output (Direction = JSON5 / JSONC → strict JSON, Indentation = 2 spaces):

{
  "port": 8080,
  "hosts": [
    "a",
    "b"
  ]
}

Going the other way, {"name": "ada", "tags": ["x", "y"]} with Direction = JSON → JSON5 and JSON5: add trailing commas ticked becomes:

{
  name: 'ada',
  tags: [
    'x',
    'y',
  ],
}

And loose numeric syntax is normalized — { mask: 0xff, ratio: .5, missing: NaN, cap: Infinity } with NaN / Infinity in strict JSON = Convert to a string gives:

{
  "mask": 255,
  "ratio": 0.5,
  "missing": "NaN",
  "cap": "Infinity"
}

Options

• Direction — JSON5 / JSONC → strict JSON (the default), JSON → JSON5, or Auto, which emits strict JSON when the input used any JSON5-only syntax and JSON5 when the input was already strict.
• Indentation — 2 spaces, 4 spaces, a tab, or Minify for a single compact line.
• Sort keys alphabetically — sorts every object's keys by Unicode code point, at every nesting level, so two configs diff cleanly.
• NaN / Infinity in strict JSON — strict JSON cannot spell them, so pick null (what JSON.stringify does), a string such as "Infinity", or refuse the conversion.
• JSON5 quote style — single quotes (JSON5 house style) or double quotes, for JSON5 output.
• JSON5: leave identifier keys unquoted — on by default; keys that are plain ASCII identifiers stay bare, anything else stays quoted.
• JSON5: add trailing commas — appends a comma after the last element of every non-empty array and object, so adding a line later touches one line in the diff.

Limits and edge cases

• Comments carry no data and are always dropped. JSON has no comment syntax, and this converter emits data only — it does not try to re-attach comments when writing JSON5. If your comments matter, keep the JSON5 file as the source of truth and generate the JSON.
• Key order is preserved. Objects keep the order they were written in, so a round trip does not reshuffle your config. Repeated keys collapse last-wins at the first occurrence's position — exactly what a JavaScript object literal does.
• Number precision is preserved. Numbers are normalized as text, not through a 64-bit float, so 12345678901234567890123 survives intact instead of being rounded.
• Input caps: 1 MB of text and 200 levels of nesting. Larger inputs are refused rather than risking an out-of-memory failure in the sandbox.
• Parse errors name the problem and the line and column where it was found, e.g. unterminated string starting at line 2, column 8.
• Only one top-level value is accepted. Newline-delimited JSON (one object per line) is a different format and is rejected as trailing content.

FAQ