typeshift

typeshift provides a Zod-like parse workflow with idiomatic Rust types.

The struct or enum is the single source of truth. With one attribute, you get:

Serialize
Deserialize
Validate
JsonSchema

Install

[dependencies]
typeshift = "0.5"

Quick start

use typeshift::typeshift;

#[typeshift]
struct User {
    #[validate(length(min = 3))]
    name: String,
}

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let user: User = typeshift::parse_str(r#"{"name":"Ada"}"#)?;
    let json = typeshift::to_json(&user)?;
    let schema = typeshift::schema_json::<User>();

    assert_eq!(json, r#"{"name":"Ada"}"#);
    assert_eq!(schema["type"], "object");
    Ok(())
}

API

parse_str<T: TypeShift>(&str) -> Result<T, TypeShiftError>
to_json<T: TypeShift>(&T) -> Result<String, serde_json::Error>
schema_json<T: TypeShift>() -> serde_json::Value

Notes

#[typeshift] supports structs and enums, including generics.
Enum variant fields with #[validate(...)] are validated during parse_str.
Unions are intentionally not supported.
#[derive(TypeShift)] is a legacy compatibility marker and does not generate impls.
Current MSRV is Rust 1.81.

Validation to schema mapping

typeshift uses schemars and forwards compatible #[validate(...)] attributes into generated JSON Schema. Common mappings include:

#[validate(email)] -> format: "email"
#[validate(url)] -> format: "uri"
#[validate(length(...))] -> minLength / maxLength (or minItems / maxItems)
#[validate(range(...))] -> minimum / maximum
#[validate(regex(path = ...))] -> pattern
#[validate(required)] on Option<T> -> schema treats field as required T

Some validators are runtime-only and not directly representable as JSON Schema (for example custom, must_match, and nested behavior).

More examples

cargo run -p typeshift --example basic
cargo run -p typeshift --example enum_tagged
cargo run -p typeshift --example generic_envelope
cargo run -p typeshift --example full_flow