@adbros/typegen
v3.0.0
Published
A simple generator of TypeScript types from OpenAPI docs
Downloads
709
Readme
Typegen
A simple generator of TypeScript types from OpenAPI docs.
Installation
- Install the NPM package:
npm install --save-dev @adbros/typegen
. - Create a
typegen.config.mjs
file and export a valid config object from it. See Configuration (below) for details.
Usage
Each time the OpenAPI doc changes, run npx typegen
in the directory containing typegen.config.mjs
.
Configuration
The typegen.config.mjs
file should be present in the current working directory when running the typegen
utility. (Usually, it makes most sense for it to reside in the project's root directory -- next to package.json
, tsconfig.json
, etc.).
The config object exported from typegen.config.mjs
(using ESM default export) should have the following properties:
inputDoc
(required) -- A string containing the path to the input OpenAPI 3.0.x doc.outputDir
(required) -- A string containing the path to the output directory.hooks
-- An optional child config object containing convenience funtions to be called before/after the type generation itself. Both functions are optional. Both are called with zero arguments. Both may return a promise; the program's execution will be blocked until the promise's resolution. Both may throw or return an eventually rejected promise, in which case the program will be prematurely terminated with a nonzero exit code.before
-- Called before the input doc is loaded. Useful e.g. for fetching the doc from an external source, generating the doc from other data, performing custom validation on the doc (and throwing/rejecting on failure to prevent the type generation itself), etc.after
-- Called after the type generation is finished and the files written. Useful e.g. for custom post-processing of the generated files.
endpoints
-- An optional child config object containing options related to the rendering of endpoints as arbitrary TypeScript constructs. If not given,endpoints.ts
is not generated at all. If given, it must contain:renderEach
-- A function, called for each endpoint (i.e. each path-method pair), that maps available info about the given endpoint to a string containing arbitrary TypeScript code. If a nullish value is returned from the function instead, the given endpoint is excluded from the output. The function is always called with a single argument -- an object containing the following properties:path
(string
) -- the path as specified in the input doc;method
(string
) -- the name of the HTTP method in lower case, e.g.post
;pathType
(string
) -- a string literal type or template literal type describing the path including path parameters, e.g.`/customer/${number}/invoices`
;requestType
(string
) -- the type of the request body, e.g.Invoice
;responseType
(string
) -- the type of the response body for status code200
, e.g.{ invoices: Invoice[]; totalCount: number }
;paramsType
(string
) -- the type describing an object, where each property corresponds to a single query parameter accepted by the endpoint, e.g.GetInvoicesParams
(defined inparams.ts
as e.g.{ limit: number; offset: number }
);paramsExpected
(boolean
) -- whether there is at least one query parameter accepted by the endpoint;paramsRequired
(boolean
) -- whether there is at least one query parameter required by the endpoint;bodyExpected
(boolean
) -- whether the endpoint expects a request body;bodyReturned
(boolean
) -- whether the endpoint returns a body within each response with status code200
.
renderModule
-- A function which maps the concatenated output ofrenderEach
calls to a string containing the final TypeScript module. It takes an object with the sole propertycontent
.
Example
The following config defines the generation of a valid subtype of an Axios instance, which only permits calls to existing endpoints using overloaded method signatures. This is useful because:
- calls to nonexistent endpoints are disallowed;
- correct request bodies/parameters are enforced;
- the return type -- corresponding to the matching endpoint's response type -- is automatically inferred for each valid call.
export default {
inputDoc: 'path/to/openapi.yaml',
outputDir: 'path/to/output/directory'
endpoints: {
renderModule: ({ content }) => `export type Api = { ${content} };`,
renderEach: ({
method,
pathType,
requestType,
responseType,
paramsType,
paramsExpected,
paramsRequired,
}) => {
switch (method) {
case 'get':
return paramsExpected
? `${method}(
path: ${pathType},
config${paramsRequired ? '' : '?'}:
{ params: ${paramsType} },
): Promise<{ data: ${responseType} }>;`
: `${method}(
path: ${pathType},
): Promise<{ data: ${responseType} }>;`;
case 'post':
case 'put':
return `${method}(
path: ${pathType},
body: ${requestType},
): Promise<{ data: ${responseType} }>;`;
default:
return `${method}(
path: ${pathType},
): Promise<{ data: ${responseType} }>;`;
}
},
},
};