tg-resources

Abstractions on-top of superagent and fetch (or other Ajax libraries) for communication with REST. Targeted mostly against Django Rest Framework (DRF) running on Django so some logic might not be applicable for other frameworks.

Installing

Using NPM

1npm i tg-resources
2
3# And add fetch backend
4npm i @tg-resources/fetch
5
6# Or use superagent backend
7npm i @tg-resources/superagent

Or using Yarn

1yarn add tg-resources
2
3# And fetch backend
4yarn add @tg-resources/fetch
5
6# Or use superagent backend
7yarn add @tg-resources/superagent

Polyfills for fetch

When you are targeting browsers without native support for fetch or running this on node versions before 17 then you need to provide polyfills for the fetch globals. The easiest way to do it is to add @tg-resources/fetch-runtime.

Alternatively you can also just use your own polyfill if you want to. In that case the methods should be available in the root scope (e.g. window/self for browsers or global for node).

Does it work on react native?

YES

Using with hermes engine

With the version 4.0.0 hermes is fully supported as we have replaced lodash.template with our own url token replacement helper. :tada:

Using signal with react-native

Use abortcontroller-polyfill until https://github.com/facebook/react-native/issues/18115 is resolved in react-native core. The polyfill does not actually close the connection, but instead ensures the fetch rejects the promise with AbortError. To use the polyfill add the following to the top of your app entrypoint:

import 'abortcontroller-polyfill/dist/polyfill-patch-fetch'

Basic Usage

1import { Router } from "tg-resources"
2import { FetchResource: Resource } from "@tg-resources/fetch"
3
4const onLoad = result => console.log(result);
5const onError = result => console.error(result);
6
7
8const api = new Router({
9    cats: new Resource('/cats'),
10    cat: new Resource('/cats/${pk}')
11}, {
12    apiRoot: '/api/v1' // Set api root
13});
14
15const apiRouter = createRouter({
16    cats: '/cats',
17    cat: '/cats/${pk}',
18}, {
19    apiRoot: '/api/v1', // Set api root
20}, Resource);
21
22// Do a get request to /api/v1/cats?gender=M
23api.cats.fetch(null, {gender: 'M'}).then(onLoad, onError);
24apiRouter.cats.fetch(null, {gender: 'M'}).then(onLoad, onError);
25
26// Do a head request to /api/v1/cats?gender=M
27api.cats.head(null, {gender: 'M'}).then(onLoad, onError);
28apiRouter.cats.head(null, {gender: 'M'}).then(onLoad, onError);
29
30// Do a post request to /api/v1/cats with data: {name: 'Twinky', gender: 'M'}
31api.cats.post(null, {name: 'Twinky', gender: 'M'}).then(onLoad, onError);
32apiRouter.cats.post(null, {name: 'Twinky', gender: 'M'}).then(onLoad, onError);
33
34// Do a patch request to /api/v1/cats/1 with data: {name: 'Tinkelberg'}
35api.cat.patch({pk: 1}, {name: 'Tinkelberg'}).then(onLoad, onError);
36apiRouter.cat.patch({pk: 1}, {name: 'Tinkelberg'}).then(onLoad, onError);
37
38// Do a put request to /api/v1/cats with data: {pk: 1, name: 'Twinky'}
39api.cats.put(null, {pk: 1, name: 'Twinky', gender: 'M'}).then(onLoad, onError);
40apiRouter.cats.put(null, {pk: 1, name: 'Twinky', gender: 'M'}).then(onLoad, onError);
41
42// Do a delete request to /api/v1/cats/1 with data: {'free':'yes'}
43api.cat.del({pk: 1}, {free: 'yes'}).then(onLoad, onError);
44apiRouter.cat.del({pk: 1}, {free: 'yes'}).then(onLoad, onError);

Please note that the router is useful for providing default configuration and grouping endpoints. It's still possible to use Resources without a router(see Resource api)

Configuration

• apiRoot (String): Base for all resource paths
• headers (Object|Function: Object): Optional Function or Object which can be used to add any additional headers to requests.
• cookies (Object|Function): Optional Function or Object which can be used to add any additional cookies to requests. Please note that in modern browsers this is disabled due to security concerns.
• mutateResponse (Function): Optional function with signature (responseData, rawResponse: ResponseWrapper, resource: Resource, requestConfig: Object) => responseData which can be used to mutate response data before resolving it. E.g. This can be used to provide access to raw response codes and headers to your success handler.
• mutateError (Function): Optional function with signature (error: ResourceErrorInterface, rawResponse: ResponseWrapper, resource: Resource, requestConfig: Object) => error which can be used to mutate errors before rejecting them. E.g. This can be used to provide access to raw response codes and headers to your error handler.
• statusSuccess (Array[int]|number): Array (or a single value) of status codes to treat as a success. Default: [200, 201, 204]
• statusValidationError (Array[int]|number): Array (or a single value) of status codes to treat as ValidationError. Default: [400]
• defaultAcceptHeader (String): Default accept header that is automatically added to requests (only if headers.Accept=undefined). Default: 'application/json'
• parseErrors (Function): Function with signature (errorText, parentConfig) => [nonFieldErrors, errors] which is used to parse response errors into a ValidationError object. The default handler is built for Django/DRF errors.
• prepareError (Function): Function with signature (err, parentConfig) => mixed which is used to normalize a single error. The default handler is built for Django/DRF errors.
• mutateRawResponse (Function): Advanced usage: Optional function with signature (rawResponse: ResponseWrapper, requestConfig: Object) => rawResponse which can be used to mutate the response before it is resolved to responseData or a ResourceErrorInterface subclass. Use the source of ResponseWrapper, SuperagentResponse and Resource::ensureStatusAndJson for guidance.
• withCredentials (bool): Allow request backend to send cookies/authentication headers, useful when using same API for server-side rendering.
• allowAttachments (bool): Allow POST like methods to send attachments.
• signal: (AbortSignal): Pass in an AbortSignal object to abort the request when desired. Only supported via request config. Default: [null]. For react-native a polyfill is needed.

Error handling

With tg-resources, all errors are Rejected. The logic is best described with an example:

1const resource = new Resource('user/login');
2
3const errorHandler = (error) => {
4    // Network error occurred
5    if (error.isNetworkError) {
6        console.error({
7            type: 'NETWORK_FAILED',
8            error,
9        });
10    } else if (error.isAbortError) {
11        // Request was aborted
12        console.error({
13            type: 'ABORTED',
14            error,
15        });
16    } else if (error.isValidationError) {
17        // Validation error occurred (e.g.: wrong credentials)
18        console.error({
19            type: 'VALIDATION_ERROR',
20            error,
21        });
22    } else {
23        // As a last resort, also handle invalid response codes
24        console.error({
25            type: 'SERVER_ERROR',
26            error,
27        });
28    }
29};
30
31const payload = {
32    user: 'foo',
33    passwrod: 'bar',
34};
35
36resource.post(null, payload).then(
37    (user) =>
38        console.log({
39            type: 'LOGGED_IN',
40            data: {
41                user,
42            },
43        }),
44    errorHandler
45);

API

createRouter

Creates type-safe Router instance.

Arguments

1. routes (Object): Object matching pattern { [key]: string | { [key]: string } }. String values are used as endpoints to create resource. For more info see Resource API This can be nested, meaning new router is created for object types found.
2. config (Object): Object containing config for top level router. See Configuration
3. resourceKlass Resource: Resource class that implements backend. This allows any of the backends to be used when creating Router.

Resource

Construct a new resource for loading data from a single (or dynamic) endpoint

Arguments

1. apiEndpoint (string): Endpoint used for this resource. Supports ES6 token syntax, e.g: "/foo/bar/${pk}"
2. config (Object): Object containing config for this resource. See Configuration

Tokenized endpoints

The Resource module also supports dynamic urls by supporting ES6 token syntax. Request methods can then provide values as an object using the first argument kwargs.

So for example:

1new Resource('/foo/bar/${pk}').get({ pk: 1 }).then((x) => x);

Would result in a GET request to /foo/bar/1

Returns

(Resource): Returns instance of Resource.

Resource.fetch

Do a get request to the resource endpoint with optional kwargs and query parameters.

Arguments

1. kwargs=null (Object): Object containing the replacement values if the resource uses tokenized urls
2. query=null (Object|string): Query parameters to use when doing the request.
3. requestConfig=null (Object): Configuration overrides, useful when using same API for server-side rendering.

Resource.options

Alias for Resource.fetch(kwargs, query, requestConfig) with options method.

Resource.head

Alias for Resource.fetch(kwargs, query, requestConfig) with head method.

Returns

(Promise): Returns a Promise that resolves to the remote result or throws if errors occur.

Resource.post

Do a method request to the resource endpoint with optional kwargs and query parameters.

Arguments

1. kwargs=null (Object): Object containing the replacement values if the resource uses tokenized urls
2. data=null (Object|string): Query parameters to use when doing the request.
3. query=null (Object|string): Query parameters to use when doing the request.
4. attachments=null (Array): Attachments, creates multipart request
5. requestConfig=null (Object): Configuration overrides, useful when using same API for server-side rendering.

Returns

(Promise): Returns a Promise that resolves to the remote result or throws if errors occur.

Resource.patch

Alias for Resource.post(kwargs, data, query, attachments, requestConfig) with patch method.

Resource.put

Alias for Resource.post(kwargs, data, query, attachments, requestConfig) with put method.

Resource.del

Alias for Resource.post(kwargs, data, query, attachments, requestConfig) with del method.

ResourceErrorInterface

Generic base class for all errors that can happen during requests

Attributes

• isNetworkError (bool): Always false
• isInvalidResponseCode (bool): Always false
• isValidationError (bool): Always false

NetworkError

Error class used for all network related errors

Extends ResourceErrorInterface and overwrites:

• isNetworkError (bool): Always true

Attributes

• error (Error): Original Error object that occured during network transport

AbortError

Error class used when a request is aborted

Extends ResourceErrorInterface and overwrites:

• isAbortError (bool): Always true

Attributes

• error (Error): Original Error object that was raised by the request engine

InvalidResponseCode

Error class used when unexpected response code occurs

Extends ResourceErrorInterface and overwrites:

• isInvalidResponseCode (bool): Always true

Attributes

• statusCode (string): Response status code
• responseText (int): Response body text

RequestValidationError

Error class used when backend response code is in config.statusValidationError.

Extends InvalidResponseCode and overwrites:

• isInvalidResponseCode (bool): Always false
• isValidationError (bool): Always true

Attributes

• errors: (ValidationErrorInterface|any): The result from requestConfig.parseError

ValidationErrorInterface

Error types returned by the default error parser.

Supports iteration (map/forEach/for .. of/etc)

Attributes

• errors: (any): Errors and error messages.

Types

Since DRF errors can be arbitrarily nested and one field can have multiple errors, some specific types of interest:

• SingleValidationError: Errors for a single field the .errors attribute is a list of strings.
• ValidationError: Errors for an object, .errors is an object with field names as keys.
• ListValidationError: Errors related to list of objects. .errors is a list of ValidationErrorInterface.

Methods

(*) Not applicable to SingleValidationError

hasError

Returns

(bool): True if there are any errors.

getError*

Get field specific error

Arguments

1. fieldName (Array|string): Field name or path to child error, e.g ['parent', 'child'] or array indexes for ListValidationError
2. [allowNonField=false] (bool): If true, also check for nonFieldErrors if the specified field does not have an error

Returns

(any): Returns a normalized error for fieldName or null

firstError*

Get first error normalized to a string for this ValidationError

Arguments

1. [allowNonField=false] (bool): If true, also check for nonFieldErrors

Returns

(any): First error as a string or null

License

MIT © Thorgate