Crate jtd_infer

Infers JSON Type Definition schemas from example inputs.

JSON Type Definition, aka RFC 8927, is an easy-to-learn, standardized way to define a schema for JSON data. You can use JSON Typedef to portably validate data across programming languages, create dummy data, generate code, and more.

This Rust crate can generate a JSON Typedef schema from example data. If you are looking to use this package as a CLI tool, see this crate’s README. The remainder of these docs are focused on this crate as a Rust library, and so focuses on the Rust API for using jtd_fuzz.

Quick start

Here’s how you can use this crate to infer a schema:

use serde_json::json;
use jtd_infer::{Inferrer, Hints, HintSet, NumType};

let mut inferrer = Inferrer::new(Hints::new(
    NumType::Uint8,
    HintSet::new(vec![]),
    HintSet::new(vec![]),
    HintSet::new(vec![]),
));

inferrer = inferrer.infer(json!({ "foo": true, "bar": "xxx" }));
inferrer = inferrer.infer(json!({ "foo": false, "bar": null, "baz": 5 }));

let inference = inferrer.into_schema();

assert_eq!(
    json!({
        "properties": {
            "foo": { "type": "boolean" },
            "bar": { "type": "string", "nullable": true },
        },
        "optionalProperties": {
            "baz": { "type": "uint8" },
        },
    }),
    serde_json::to_value(inference.into_serde_schema()).unwrap(),
)

Structs

HintSet

A set of paths to parts of the input that are subject to a hint in Hints.

Hints

Hints for Inferrer.

Inferrer

Keeps track of a sequence of example inputs, and can be converted into an inferred schema.

Enums

NumType

A type of number to infer by default.