schema.js

Universal, tiny schema for JavaScript data types.

Design

For this project, we will define the schema based on JavaScript data types.

JavaScript can be classified into two data types called primitives and objects.

If we further classify them according to whether they are Unit type or Collective types, they can be classified into 8 data types.

Unit type is a type that has only a single value. Collection type is set of Unit type.

	Unit type	Collective type
Primitive	`undefined`	`string`, `number`, `bigint`, `boolean`, `symbol`
Object	`null`	`object`

A subtype of object, Function is a special type that JavaScript treats as first class.

So for this project, we define Function as a supertype in addition to the 8 types above.

Core schema

Create JavaScript primitive data schema.

import {
  BigintSchema,
  BooleanSchema,
  NumberSchema,
  StringSchema,
  SymbolSchema,
  UndefinedSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const stringSchema = new StringSchema();
const numberSchema = new NumberSchema();
const bigintSchema = new BigintSchema();
const booleanSchema = new BooleanSchema();
const undefinedSchema = new UndefinedSchema();
const symbolSchema = new SymbolSchema();

Create JavaScript objective data schema.

import {
  FunctionSchema,
  NullSchema,
  ObjectSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const objectSchema = new ObjectSchema();
const functionSchema = new FunctionSchema();
const nullSchema = new NullSchema();

Assert with schema

Assert whether the value satisfies the schema.

import {
  assertSchema,
  BooleanSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const value: unknown = true;
assertSchema(new BooleanSchema(), value);
// value is `boolean`

Validate with schema

If you do not want to throw an error, you can use the validateSchema function.

import {
  ObjectSchema,
  StringSchema,
  validateSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const schema = new ObjectSchema({
  name: new StringSchema(),
  type: "dog",
});

const result = validateSchema(schema, {});
if (result.pass) {
  result.data; // { name: string, type: "dog" }
} else {
  result.errors; // SchemaError[]
}

Additional subtype assertion (narrowing)

For the Collective type, you can add assertions of subtypes.

The Collective type has the and method. It accept subtype schema and returns a new Collective type. The new Collective type will be type narrowed by the subtype schema.

Example of creating a tuple ([0, "hi"]) schema from an object schema:

import {
  ArraySchema,
  assertSchema,
  NumberSchema,
  ObjectSchema,
  SchemaError,
  StringSchema,
  TupleSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const value: unknown = undefined;

const tupleSchema = new ObjectSchema().and(new ArraySchema()).and(
  new TupleSchema(
    new NumberSchema(),
    new StringSchema(),
  ),
);
assertSchema(tupleSchema, value);
// value is `[number, string]`

Logical operation schema

Provides the logical operations of the schema. Several schemas can be multiplied together to create a new schema.

Logical OR

The logical OR schema (logical disjunction) for a set of schemas is true if and only if one or more of its schemas is true.

Type inference works correctly.

import {
  assertSchema,
  NullSchema,
  NumberSchema,
  OrSchema,
  StringSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const schema = new OrSchema(
  new StringSchema(),
  new NumberSchema(),
  new NullSchema(),
);
const value: unknown = undefined;
assertSchema(schema, value);
// value is `string` | `number` | null

Logical AND

The logical AND schema (logical conjunction) for a set of schemas will be true if and only if all the schemas are true. Otherwise it will be false.

Type inference works correctly.

import {
  AndSchema,
  assertSchema,
  ObjectSchema,
  PartialSchema,
  StringSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const schema = new AndSchema(
  new ObjectSchema({
    type: new StringSchema(),
  }),
  new PartialSchema({
    payload: new ObjectSchema(),
  }),
);
assertSchema(schema, {});

Logical NOT

The logical NOT schema (logical complement, negation) takes valid schema to invalid and vice versa.

Type inference works correctly.

import {
  assertSchema,
  BooleanSchema,
  NotSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const value: unknown = undefined;
assertSchema(new NotSchema(new BooleanSchema()), value);
// value is `string` | `number` | ...
assertSchema(new NotSchema(true), value);
// value is `false` | `string` | `number` | ...

String subtype schema

Provides schema for string subtypes.

import {
  assertSchema,
  LengthSchema,
  MaxLengthSchema,
  MinLengthSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const maxLengthSchema = new MaxLengthSchema(255);
const minLengthSchema = new MinLengthSchema(10);
const lengthSchema = new LengthSchema(20);

const value: string = "This is string subtype type.";
assertSchema(maxLengthSchema, value);
assertSchema(minLengthSchema, value);
assertSchema(lengthSchema, value); // throw SchemaError

Email schema

∈ string

Schema of string subtype of email format.

import {
  assertSchema,
  EmailFormatSchema,
  MaxLengthSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const emailFormatAndLessThan20 = new EmailFormatSchema().and(
  new MaxLengthSchema(20),
);
assertSchema(emailFormatAndLessThan20, "contact@test.test");

UUID format schema

∈ string

Schema of UUID format. This is string subtype.

import {
  assertSchema,
  UuidFormatSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new UuidFormatSchema();
assertSchema(schema, "00000000-0000-0000-0000-000000000000");
assertThrows(() => assertSchema(schema, "invalid UUID"));

Hostname format schema

∈ string

Schema of hostname format. This is string subtype.

Compliant with RFC 1123, 2.1 Host Names and Numbers

import {
  assertSchema,
  HostnameFormatSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new HostnameFormatSchema();
assertSchema(schema, "a");
assertThrows(() => assertSchema(schema, "a".repeat(64)));
assertThrows(() => assertSchema(schema, "invalid hostname"));

URL format schema

∈ string

Schema of URL format. This is string subtype.

import {
  assertSchema,
  UrlFormatSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new UrlFormatSchema();
assertSchema(schema, "http://localhost");
assertThrows(() => assertSchema(schema, "invalid URL"));

Date format schema

∈ string

Schema of date format. This is string subtype.

Compliant with RFC 3339, section-5.6, full-date

import {
  assertSchema,
  DateFormatSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new DateFormatSchema();
assertSchema(schema, "1000-01-01");
assertThrows(() => assertSchema(schema, "0000-00-00"));
assertThrows(() => assertSchema(schema, "invalid Date"));

Time format schema

∈ string

Schema of time format. This is string subtype.

Compliant with RFC 3339, section-5.6, full-time

import {
  assertSchema,
  TimeFormatSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new TimeFormatSchema();
assertSchema(schema, "00:00:00Z");
assertSchema(schema, "23:59:59+19:59");
assertThrows(() => assertSchema(schema, "00:00:00"));
assertThrows(() => assertSchema(schema, "invalid Time"));

Date time format schema

∈ string

Schema of date time format. This is string subtype.

Compliant with RFC 3339, section-5.6, date-time

import {
  assertSchema,
  DateTimeFormatSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new DateTimeFormatSchema();
assertSchema(schema, "1000-01-01T00:00:00Z");
assertSchema(schema, "9999-12-31T23:59:59+19:59");
assertThrows(() => assertSchema(schema, "0000-00-00:00:00:00Z"));
assertThrows(() => assertSchema(schema, "invalid date time"));

Partial schema

Schema of optional properties.

import {
  assertSchema,
  FunctionSchema,
  PartialSchema,
  StringSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const abilitySchema = new PartialSchema({
  fly: new FunctionSchema(),
});
const model = { type: "bird" } as const;
assertSchema(abilitySchema, model);
// { type: "bird", fly?: Function }

Record schema

Schema of Record object.

import {
  NumberSchema,
  RecordSchema,
  StringSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const schema = new RecordSchema(new StringSchema(), new NumberSchema());
// Record<string, number>

TypeScript accepts PropertyKey ( string | number |symbol ) for Record type, but this is restricted to string and symbol only.

This is because JavaScript always casts the object key from a number to a string.

Union subtype schema

The union subtype schema is a schema that can be used for multiple types.

Max schema

type ∈ number ⋃ bigint

Schema of max value for number or bigint subtype.

import {
  assertSchema,
  MaxSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

assertSchema(new MaxSchema(10), 5);
assertThrows(() => assertSchema(new MaxSchema(10), 11));

Min schema

type ∈ number ⋃ bigint

Schema of min value for number or bigint subtype.

import {
  assertSchema,
  MinSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import { assertThrows } from "https://deno.land/std@$VERSION/testing/asserts.ts";

assertSchema(new MinSchema(5), 10);
assertThrows(() => assertSchema(new MinSchema(5), 0));

Count schema

type ∈ Iterable<unknown>

Schema of number of elements for Iterable data types.

import {
  assertSchema,
  CountSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import {
  assertEquals,
  assertThrows,
} from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new CountSchema(10);
assertSchema(schema, "abcdefghij");
assertThrows(() => assertSchema(schema, []));

Min count schema

type ∈ Iterable<unknown>

Schema of min number of elements for Iterable data types.

import {
  assertSchema,
  MinCountSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import {
  assertEquals,
  assertThrows,
} from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new MinCountSchema(8);
assertSchema(schema, "password");
assertThrows(() => assertSchema(schema, new Array(4)));

Max count schema

type ∈ Iterable<unknown>

Schema of max number of elements for Iterable data types.

import {
  assertSchema,
  MaxCountSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import {
  assertEquals,
  assertThrows,
} from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new MaxCountSchema(255);
assertSchema(schema, "https://test.com");
assertThrows(() => assertSchema(schema, new Array(1000)));

Difference of Length schema

Length retrieves a value from the length property.
Count counts the actual number of elements.
They differ in the way they count strings.

In strings, Length schema counts the number of code units. In contrast, Count schema counts the number of characters.

import { assertEquals } from "https://deno.land/std@$VERSION/testing/asserts.ts";
assertEquals("A\uD87E\uDC04Z".length, 4);
assertEquals([..."A\uD87E\uDC04Z"].length, 3);

You should use the count schema in most cases.

Built-in Objects schema

This project provide built-in object schemas.

All list

Array schema

import {
  ArraySchema,
  assertSchema,
  StringSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const value: unknown = undefined;
assertSchema(new ArraySchema(), value);
// value is `{}[]`
assertSchema(new ArraySchema([new StringSchema()]), value);
// value is `string[]`

Tuple schema

It is a sub-type of the Array object and represents an array of a finite number of elements.

import {
  ArraySchema,
  assertSchema,
  NumberSchema,
  StringSchema,
  TupleSchema,
  UndefinedSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const value: any[] = [];
const tupleSchema = new TupleSchema(
  new NumberSchema(),
  new StringSchema(),
  new UndefinedSchema(),
);
assertSchema(tupleSchema, value);
// value is [number, string, undefined]

Date schema

Schema of Date object.

import {
  assertSchema,
  DateSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";
import {
  assertThrows,
} from "https://deno.land/std@$VERSION/testing/asserts.ts";

const schema = new DateSchema();
assertSchema(schema, new Date());
assertThrows(() => assertSchema(schema, {}));

Type inference

You can derive the correct type inference by assertSchema, but you can do the same from TypeScript types.

import {
  ArraySchema,
  InferSchema,
  NumberSchema,
  ObjectSchema,
  StringSchema,
  TupleSchema,
} from "https://deno.land/x/schema_js@$VERSION/mod.ts";

const schema = new ObjectSchema({
  a: new StringSchema(),
  b: new ArraySchema().and(
    new TupleSchema(new StringSchema(), new NumberSchema()),
  ),
  c: new ObjectSchema({
    d: 0,
  }),
});

type Schema = InferSchema<typeof schema>;
type EqualTo = {
  a: string;
  b: [string, number];
  c: { d: 0 };
};

API

All APIs can be found in the deno doc.

Performance

Benchmark script with comparison to several popular schema library is available.

deno bench --unstable

You can check the benchmark.

License

Released under the MIT license