Runtime validation for Rust, the Zod way

vld describes the shape of your data once and checks untrusted input against it: JSON bodies, query strings, config files, database rows. Errors list every problem with its path, and the same schema produces OpenAPI.

Get started GitHub

What you get

Schemas as values
Chain checks on strings, numbers, dates, collections and objects, then parse JSON or serde values into typed Rust.
Typed structs
schema! and #[derive(Validate)] generate parse(), field-level errors and lenient parsing for your own types.
Every error at once
Typed issue codes with paths, flatten/treeify/prettify helpers, custom messages and bundled translations.
OpenAPI from the same code
JSON Schema and OpenAPI 3.1 documents, nested components and breaking-change diffs.
Web extractors
Axum, Actix, ntex, Poem, Rama, Rocket, Salvo, Tower, Warp and Tonic with one error format and body limits.
Beyond HTTP
Databases, message queues, config files, CLI arguments, UI frameworks, TypeScript codegen and fake data.

Two minutes to the first parse

Add the crate, describe a struct and call parse(). Invalid input returns a VldError with every issue, not just the first.

Quick start Migrate from 0.4

Cargo.toml

[dependencies]
vld = { version = "0.5", features = ["derive", "openapi"] }

main.rs

use vld::prelude::*;

vld::schema! {
    #[derive(Debug)]
    pub struct User {
        pub name: String => vld::string().min(2).max(50),
        pub email: String => vld::string().email(),
        pub age: Option<i64> => vld::number().int().gte(18).optional(),
    }
}

let user = User::parse(r#"{"name": "Alex", "email": "alex@example.com"}"#)?;
assert_eq!(user.age, None);