TOML Formatter + Converter — TOML 1.0, ↔ JSON, ↔ YAML, line-numbered errors

Real-world use cases

• Bring a noisy `Cargo.toml` into shape before committing

  You added three new dependencies during a debugging session and your `Cargo.toml` now mixes 2-space and 4-space blocks, with one `features = ["serde", "tokio", "reqwest", "rustls"]` line stretching to column 95. Paste it here, click "Format TOML", toggle "Sort keys A→Z" so `[dependencies]` lands alphabetized for your team's CI lint. Copy back. The diff against `main` now only shows the actual three additions instead of 40 lines of whitespace churn.

• Pull a config out of `pyproject.toml` into JSON for a Python script

  You're writing a script that reads project metadata (`[project] name`, `version`, `dependencies`) and the rest of your tooling speaks JSON. Paste `pyproject.toml`, switch to "TOML → JSON", copy. Now your `meta = json.loads(open(...).read())` gets the same shape as `[project]` without you having to add `tomllib` or remember Python 3.11's stdlib oddities.

• Convert a YAML CI config to TOML for a Rust tool

  You have a `.github/workflows/build.yml` you want to port the job matrix from into a Cargo-side `xtask/config.toml`. Paste the YAML block, switch to "YAML → TOML", copy. The output uses `[[matrix]]` array-of-tables for each job, and your `xtask` binary picks it up with `toml::from_str` — same job matrix, same source of truth, two configs.

• Debug a `netlify.toml` that fails the build with no good error

  Netlify's build error is "TOML parse error at line ?? col ??" with no context. Paste your `netlify.toml`, you get the exact line, column, AND three lines of context with a `^` under the offending character. Usually it's an unquoted value with a colon (`status = 301 force` instead of `force = true`) or a missing `]` in a `[[redirects]]` header. Fix in 5 seconds, push.

• Multi-environment DB config sync across staging and prod

  You keep `[default]`, `[development]`, `[staging]`, `[production]` blocks in one `db.toml`. You want to make sure `staging` and `production` differ ONLY in `host` and `pool_size`. Format both files with "Sort keys A→Z", diff with `git diff --no-index`, you see exactly the intentional differences — nothing snuck in because someone hand-edited the file.

• Generate a Discord/Telegram bot config block from a JSON template

  You have a JSON-shaped bot config you're prototyping, but the target runtime (Rust serenity, Python aiogram with a TOML extension) wants TOML. Paste the JSON, switch to "JSON → TOML", copy into `config.toml`. The `[[commands]]` array of objects becomes `[[commands]]` array-of-tables — exactly the form the framework's `serde` derives expect.

Common pitfalls

• Pasting a `.env` file as TOML and watching it fail. `.env` allows `KEY=value` without quoting; TOML requires quotes around any string with spaces or special chars. Wrap each value in quotes (`KEY="value"`) and it'll parse.

• Expecting `inf` / `nan` to survive the TOML → JSON conversion. JSON forbids both, so we flatten them to `null` with a note in the FAQ. If your value is genuinely `nan`, the right fix is upstream — that's almost always a math bug, not a serialization gap.

• Trying to keep comments after a re-format. The parser intentionally drops comments because preserving them across a re-emit is a 3x-larger parser for a feature most workflows don't use (Prettier's TOML mode does the same). Use git diff to see the actual content change.

• Treating dotted keys as nesting in JSON output. TOML's `a.b.c = 1` is identical to `[a.b] c = 1` — the JSON output is `{a: {b: {c: 1}}}` either way. There is no separate flat-with-dots JSON form.

• Pasting JSON with a root array and expecting TOML. TOML requires a table (object) at the root. Wrap your array in `{ "items": [...] }` first; the converter will then emit `[[items]]` array-of-tables.

• Expecting sort-keys to reorder array elements. Sort-keys only sorts table KEYS — `[[products]]` array elements and `[1, 3, 2]` array values stay in source order because TOML semantics treat array order as data.

Privacy