跳到主要内容

Error handling

Strapi is natively handling errors with a standard format.

There are 2 use cases for error handling:

  • As a developer querying content through the REST or GraphQL APIs, you might receive errors in response to the requests.
  • As a developer customizing the backend of your Strapi application, you could use controllers and services to throw errors.

Receiving errors​

Errors are included in the response object with the error key and include information such as the HTTP status code, the name of the error, and additional information.

REST errors​

Errors thrown by the REST API are included in the response that has the following format:

{
  "data": null,
  "error": {
    "status": "", // HTTP status
    "name": "", // Strapi error name ('ApplicationError' or 'ValidationError')
    "message": "", // A human readable error message
    "details": {
      // error info specific to the error type
    }
  }
}

GraphQL errors​

Errors thrown by the GraphQL API are included in the response that has the following format:

{ "errors": [
    {
      "message": "", // A human reable error message
      "extensions": {
        "error": {
          "name": "", // Strapi error name ('ApplicationError' or 'ValidationError'),
          "message": "", // A human reable error message (same one as above);
          "details": {}, // Error info specific to the error type
        },
        "code": "" // GraphQL error code (ex: BAD_USER_INPUT)
      }
    }
  ],
  "data": {
    "graphQLQueryName": null
  }
}

Throwing errors​

Controllers and middlewares​

The recommended way to throw errors when developing any custom logic with Strapi is to have the controller or middleware respond with the correct status and body.

This can be done by calling an error function on the context (i.e. ctx). Available error functions are listed in the http-errors documentation but their name should be lower camel-cased to be used by Strapi (e.g. badRequest).

Error functions accept 2 parameters that correspond to the error.message and error.details attributes received by a developer querying the API:

  • the first parameter of the function is the error message
  • and the second one is the object that will be set as details in the response received
// path: ./src/api/[api-name]/controllers/my-controller.js

module.exports = {
  renameDog: async (ctx, next) => {
    const newName = ctx.request.body.name;
    if (!newName) {
      return ctx.badRequest('name is missing', { foo: 'bar' })
    }
    ctx.body = strapi.service('api::dog.dog').rename(newName);
  }
}

// path: ./src/api/[api-name]/middlewares/my-middleware.js

module.exports = async (ctx, next) => {
  const newName = ctx.request.body.name;
  if (!newName) {
    return ctx.badRequest('name is missing', { foo: 'bar' })
  }
  await next();
}

Services and models lifecycles​

Once you are working at a deeper layer than the controllers or middlewares there are dedicated error classes that can be used to throw errors. These classes are extensions of Node Error class and are specifically targeted for certain use-cases.

These error classes are imported through the @strapi/utils package and can be called from several different layers. The following examples use the service layer but error classes are not just limited to services and model lifecycles. When throwing errors in the model lifecycle layer, it's recommended to use the ApplicationError class so that proper error messages are shown in the admin panel.

Note

See the default error classes section for more information on the error classes provided by Strapi.

Details

Example: Throwing an error in a service This example shows wrapping a core service and doing a custom validation on the create method:

path: ./src/api/restaurant/services/restaurant.js

const { errors } = require('@strapi/utils');
const { ApplicationError } = errors;
const { createCoreService } = require('@strapi/strapi').factories;

module.exports = createCoreService('api::restaurant.restaurant', ({ strapi }) =>  ({
  async create(params) {
    let okay = false;

    // Throwing an error will prevent the restaurant from being created
    if (!okay) {
      throw new ApplicationError('Something went wrong', { foo: 'bar' });
    }
  
    const result = await super.create(params);

    return result;
  }
});
Example: Throwing an error in a model lifecycle

This example shows building a custom model lifecycle and being able to throw an error that stops the request and will return proper error messages to the admin panel. Generally you should only throw an error in beforeX lifecycles, not afterX lifecycles.

path: ./src/api/[api-name]/content-types/[api-name]/lifecycles.js

const { errors } = require('@strapi/utils');
const { ApplicationError } = errors;

module.exports = {
  beforeCreate(event) {
    let okay = false;

    // Throwing an error will prevent the entity from being created
    if (!okay) {
      throw new ApplicationError('Something went wrong', { foo: 'bar' });
    }
  },
};

Policies​

Policies are a special type of middleware that are executed before a controller. They are used to check if the user is allowed to perform the action or not. If the user is not allowed to perform the action and a return false is used then a generic error will be thrown. As an alternative, you can throw a custom error message using a nested class extensions from the Strapi ForbiddenError class, ApplicationError class (see Default error classes for both classes), and finally the Node Error class.

The PolicyError class is available from @strapi/utils package and accepts 2 parameters:

  • the first parameter of the function is the error message
  • (optional) the second parameter is the object that will be set as details in the response received; a best practice is to set a policy key with the name of the policy that threw the error.
Example: Throwing a PolicyError in a custom policy

This example shows building a custom policy that will throw a custom error message and stop the request.

path: ./src/api/[api-name]/policies/my-policy.js

const { errors } = require('@strapi/utils');
const { PolicyError } = errors;

module.exports = (policyContext, config, { strapi }) => {
  let isAllowed = false;

  if (isAllowed) {
    return true;
  } else {
    throw new PolicyError('You are not allowed to perform this action', {
      policy: 'my-policy',
      myCustomKey: 'myCustomValue',
    });
  }
}

Default error classes​

The default error classes are available from the @strapi/utils package and can be imported and used in your code. Any of the default error classes can be extended to create a custom error class. The custom error class can then be used in your code to throw errors.

The ApplicationError class is a generic error class for application errors and is generally recommended as the default error class. This class is specifically designed to throw proper error messages that the admin panel can read and show to the user. It accepts the following parameters:

ParameterTypeDescriptionDefault
messagestringThe error messageAn application error occured
detailsobjectObject to define additional details{}
throw new ApplicationError('Something went wrong', { foo: 'bar' });