An API for robber language

I just added a small site and API at rovarspraket.tammergard.se for translating to and from robber language (Swedish: rövarspråket). The background — the rules, the edge cases, and eleven progressive JavaScript implementations — is in the older post Robber language generator.

All endpoints live under /api/v1/, return JSON, and have CORS enabled for all origins. Responses are CDN-cached for a week. Full docs at rovarspraket.tammergard.se/api.

To robber

await fetch("https://rovarspraket.tammergard.se/api/v1/to/hej")
// {
//   text: "hej",
//   robber: "hohejoj"
// }

Any string works — consonants get the + "o" + consonant treatment, vowels and other characters pass through unchanged, and x is expanded to ks first (since x is pronounced "ks" in Swedish):

await fetch("https://rovarspraket.tammergard.se/api/v1/to/yxa")
// {
//   text: "yxa",
//   robber: "ykoksosa"
// }

From robber

await fetch("https://rovarspraket.tammergard.se/api/v1/from/hohejoj")
// {
//   robber: "hohejoj",
//   text: "hej"
// }

The reverse is lossy in one specific way: since x and ks both encode to the same robber output, decoding always gives back ks:

await fetch("https://rovarspraket.tammergard.se/api/v1/from/ykoksosa")
// {
//   robber: "ykoksosa",
//   text: "yksa"
// }

This isn't a bug — it's a fundamental property of robber language. The encoding step throws away the distinction between x and ks.

Validate

await fetch("https://rovarspraket.tammergard.se/api/v1/validate/hohejoj")
// {
//   input: "hohejoj",
//   valid: true
// }

A string is valid robber language if every consonant is immediately followed by "o" + the same consonant (case-insensitive on the trailing one), and no standalone x or X appears.

Errors

Errors follow RFC 9110 and distinguish malformed input from semantically invalid input. Missing parameters return 400:

await fetch("https://rovarspraket.tammergard.se/api/v1/to/")
// {
//   error: {
//     status: 400,
//     message: "Saknar text att översätta."
//   }
// }

Decoding a string that isn't valid robber language returns 422:

await fetch("https://rovarspraket.tammergard.se/api/v1/from/bod")
// {
//   error: {
//     status: 422,
//     message: '"bod" är inte en giltig rövarspråkssträng.'
//   }
// }

That's the whole API. It's built on the @tammergard/robber package on npm, which exposes the same toRobber, fromRobber and pattern you can use directly in any JavaScript project.

rovarspraket.tammergard.se

Behind the API sits a small playful site at rovarspraket.tammergard.se where you can paste in any text and see it translated either way. It picks the direction automatically — if the input looks like robber language, it decodes; otherwise it encodes.

The deeper write-up of how the encoding actually works lives in the original post: Robber language generator. It walks through two strategies and eleven progressive examples before landing on a two-line regex.