Reversing/inverting codecs

[franky47]

The docs show an example of bytesToUtf8:

const bytesToUtf8 = z.codec(z.instanceof(Uint8Array), z.string(), {
  decode: (bytes) => new TextDecoder().decode(bytes),
  encode: (str) => new TextEncoder().encode(str),
});
 
bytesToUtf8.decode(bytes);          // => "Hello, 世界!"
bytesToUtf8.encode("Hello, 世界!");  // => Uint8Array

Let's say one needs to do the reverse transform in a pipe (example), UTF-8 to bytes. It would be nice to have the ability to reverse/invert a codec, swapping the input & output schemas, and decode & encode function.

// Invert sounds more mathematically correct, but reverse is what came to mind initially
const utf8ToBytes = bytesToUtf8.invert()
const utf8ToBytes = bytesToUtf8.reverse()

If that sounds like something useful, I'd be down to try implementing it in a PR.

[inkeep-ai-assistant[bot]]

The current Zod (v4) API allows you to define codecs for bidirectional transformations, such as bytes-to-UTF8 and UTF8-to-bytes, by specifying both decode and encode functions when creating a codec(1). However, there is no documented built-in method to automatically "reverse" or invert a codec (i.e., to swap the input/output schemas and the encode/decode functions in-place). To achieve the reverse transformation, you would need to manually define a new codec with the input and output schemas swapped and the encode/decode functions inverted:

const utf8ToBytes = z.codec(
  z.string(), 
  z.instanceof(Uint8Array), 
  {
    decode: (str) => new TextEncoder().encode(str),
    encode: (bytes) => new TextDecoder().decode(bytes),
  }
);

(1)

If you think a utility for inverting codecs would be helpful, implementing it as a PR would be a valuable contribution. There is no mention of such a feature in the current documentation(1), (2).

📚 Sources:

1. Zod v4 LLMs-Full
2. Zod v4 LLMs-Full

---

Have another question? Just tag @inkeep.