@barusu/typescript-json-schema
typescript-json-schema
Generate json-schemas from your Typescript sources.
I sincerely suggest you use the original repository, as this repository/package is forked for demanding of myself (study and customization).
Peer Commit Id: https://github.com/YousefED/typescript-json-schema/commit/20a03a2d2fe81bea56a895cee7975f87fbf480f8
Features
- Compiles your Typescript program to get complete type information.
- Translates required properties, extends, annotation keywords, property initializers as defaults. You can find examples for these features in the test examples.
Usage
Programmatic use
import { resolve } from 'path'
import * as TJS from '@barusu/typescript-json-schema'
// optionally pass argument to schema generator
const settings: TJS.PartialArgs = {
ref: false,
required: true,
}
// optionally pass ts compiler options
const compilerOptions: TJS.CompilerOptions = {
strictNullChecks: true
}
// optionally pass a base path
const basePath = './my-dir'
const program = TJS.getProgramFromFiles([resolve('my-file.ts')], compilerOptions, basePath)
// We can either get the schema for one file and one type...
const schema = TJS.generateSchema(program, 'MyType', settings)
// ... or a generator that lets us incrementally get more schemas
const generator = TJS.buildGenerator(program, settings)
// generator can be also reused to speed up generating the schema if usecase allows:
const schemaWithReusedGenerator = TJS.generateSchema(program, 'MyType', settings, [], generator)
// all symbols
const symbols = generator.getUserSymbols()
// Get symbols for different types from generator.
generator.getSchemaForSymbol('MyType')
generator.getSchemaForSymbol('AnotherType')
// In larger projects type names may not be unique,
// while unique names may be enabled.
const settings: TJS.PartialArgs = {
uniqueNames: true
}
const generator = TJS.buildGenerator(program, settings)
// A list of all types of a given name can then be retrieved.
const symbolList = generator.getSymbols('MyType')
// Choose the appropriate type, and continue with the symbol's unique name.
generator.getSchemaForSymbol(symbolList[1].name)
// Also it is possible to get a list of all symbols.
const fullSymbolList = generator.getSymbols()
getSymbols('<SymbolName>')
and getSymbols()
return an array of SymbolRef
, which is of the following format:
type SymbolRef = {
name: string
typeName: string
fullyQualifiedName: string
symbol: ts.Symbol
}
getUserSymbols
and getMainFileSymbols
return an array of string
.
Annotations
The schema generator converts annotations to JSON schema properties.
For example
export interface Shape {
/**
* The size of the shape.
*
* @minimum 0
* @TJS-type integer
*/
size: number
}
will be translated to
{
"$ref": "#/definitions/Shape",
"$schema": "http://json-schema.org/draft-07/schema#",
"definitions": {
"Shape": {
"properties": {
"size": {
"description": "The size of the shape.",
"minimum": 0,
"type": "integer"
}
},
"type": "object"
}
}
}
Note that we needed to use @TJS-type
instead of just @type
because of an issue with the typescript compiler.
You can also override the type of array items, either listing each field in its own annotation or one annotation with the full JSON of the spec (for special cases). This replaces the item types that would have been inferred from the TypeScript type of the array elements.
Example:
export interface ShapesData {
/**
* Specify individual fields in items.
*
* @items.type integer
* @items.minimum 0
*/
sizes: number[];
/**
* Or specify a JSON spec:
*
* @items {"type":"string","format":"email"}
*/
emails: string[];
}
Translation:
{
"$ref": "#/definitions/ShapesData",
"$schema": "http://json-schema.org/draft-07/schema#",
"definitions": {
"Shape": {
"properties": {
"sizes": {
"description": "Specify individual fields in items.",
"items": {
"minimum": 0,
"type": "integer"
},
"type": "array"
},
"emails": {
"description": "Or specify a JSON spec:",
"items": {
"format": "email",
"type": "string"
},
"type": "array"
}
},
"type": "object"
}
}
}
This same syntax can be used for contains
and additionalProperties
.
integer
type alias
If you create a type alias integer
for number
it will be mapped to the integer
type in the generated JSON schema.
Example:
type integer = number;
interface MyObject {
n: integer;
}
Note: this feature doesn't work for generic types & array types, it mainly works in very simple cases.
require
a variable from a file
(for requiring typescript files is needed to set argument tsNodeRegister
to true)
When you want to import for example an object or an array into your property defined in annotation, you can use require
.
Example:
export interface InnerData {
age: number;
name: string;
free: boolean;
}
export interface UserData {
/**
* Specify required object
*
* @examples require("./example.ts").example
*/
data: InnerData;
}
file example.ts
export const example: InnerData[] = [{
age: 30,
name: "Ben",
free: false
}]
Translation:
{
"$schema": "http://json-schema.org/draft-07/schema#",
"properties": {
"data": {
"description": "Specify required object",
"examples": [
{
"age": 30,
"name": "Ben",
"free": false
}
],
"type": "object",
"properties": {
"age": { "type": "number" },
"name": { "type": "string" },
"free": { "type": "boolean" }
},
"required": ["age", "free", "name"]
}
},
"required": ["data"],
"type": "object"
}
Also you can use require(".").example
, which will try to find exported variable with name 'example' in current file. Or you can use require("./someFile.ts")
, which will try to use default exported variable from 'someFile.ts'.
Note: For examples
a required variable must be an array.