#schema-validation #schema #reflection #validation #json-schema #macro-derive


A schema DSL which can be interpreted to implement multiple purposes

34 releases (7 breaking)

0.8.0 Nov 3, 2023
0.6.3 Nov 2, 2023

#1007 in Rust patterns

Used in avocado-schema-derive

MIT license

2.5K SLoC

Avocado Schema

Inspired by the JSON schema, the main purpose of Avocado Schema is to avoid defining a static schema or validation in macro, so with a flexible, separate schema defined with, for example, json string, the schema can be changed/inspected and/or saved into/load from, for example, database. Meanwhile, it can also be interpreted to implement multiple purposes (like perform validation of the data, or generate GUIs dynamically).

Crates.io MIT licensed codecov

How to use

The src/core/value.rs defines an FieldValue enum to implement the reflection of struct value, so any struct that requires to be validated against the schema needs to implement the Reflect trait. Several useful implementation has been already included in the file. The schema derive crate provides a derive macro for deriving the FieldValue for struct.

Please refer to the sources/tests code for both how to write a visitor and how to validate data by the schema. Here is a quick example:

struct Client {
    first_name: String,
    last_name: String,
    age: u64,

let schema_json = r#"
    "name": "client",
    "properties": {
        "first_name": {
            "type": "string",
            "name": "first_name",
            "max_length": 32,
            "min_length": 8
        "last_name": {
            "type": "string",
            "name": "last_name",
            "max_length": 32,
            "min_length": 8
        "age": {
            "type": "uinteger",
            "name": "age",
            "maximum": 200,
            "minimum": 0
let schema: ObjectField = serde_json::from_str(schema_json).unwrap();
let validator = Validator::new(schema);

let valid_client = Client {
first_name: "Robert".to_string(),
last_name: "Li".to_string(),
age: 32,

let invalid_client = Client {
first_name: "Robert".to_string(),
last_name: "Li".to_string(),
age: 201,
let result = validator.validate(&invalid_client);
    .contains("value 201 is larger then 200 (Maximum)"));

If any error occurs, the error will be returned in the format of BTreeMap<String, Vec<ValidationError>>. The key is path to the field where has validation error, and the ValidationError just contains the message of the error.

Besides creating the schema based on json, the builder pattern is also implemented to build the schema by code.


~171K SLoC