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 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();
}

// path: ./src/api/[api-name]/controllers/my-controller.ts

export default {
  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.ts

export default 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 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.

Example: Throwing an error in a service**#

This example shows wrapping a core service and doing a custom validation on the create method:

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 errors.ApplicationError('Something went wrong', { foo: 'bar' });
    }

    const result = await super.create(params);

    return result;
  }
});

import { errors } from '@strapi/utils';
import { factories } from '@strapi/strapi';

const { ApplicationError } = errors;

export default factories.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 errors.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.

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 errors.ApplicationError('Something went wrong', { foo: 'bar' });
    }
  },
};

import { errors } from '@strapi/utils';
const { ApplicationError } = errors;

export default {
  beforeCreate(event) {
    let okay = false;

    // Throwing an error will prevent the entity from being created
    if (!okay) {
      throw new errors.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 .

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.

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

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

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

import { errors } from '@strapi/utils';
const { PolicyError } = errors;

export default (policyContext, config, { strapi }) => {
  let isAllowed = false;

  if (isAllowed) {
    return true;
  } else {
    throw new errors.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:

throw new errors.ApplicationError('Something went wrong', { foo: 'bar' });

The PaginationError class is a specific error class that is typically used when parsing the pagination information from REST, GraphQL, or the Document Service. It accepts the following parameters:

throw new errors.PaginationError('Exceeded maximum pageSize limit');

The NotFoundError class is a generic error class for throwing 404 status code errors. It accepts the following parameters:

throw new errors.NotFoundError('These are not the droids you are looking for');

The ForbiddenError class is a specific error class used when a user either doesn't provide any or the correct authentication credentials. It accepts the following parameters:

throw new errors.ForbiddenError('Ah ah ah, you didn\'t say the magic word');

The UnauthorizedError class is a specific error class used when a user doesn't have the proper role or permissions to perform a specific action, but has properly authenticated. It accepts the following parameters:

throw new errors.UnauthorizedError('You shall not pass!');

The NotImplementedError class is a specific error class used when the incoming request is attempting to use a feature that is not currently implemented or configured. It accepts the following parameters:

throw new errors.NotImplementedError('This isn\'t implemented', { feature: 'test', implemented: false });

The PayloadTooLargeError class is a specific error class used when the incoming request body or attached files exceed the limits of the server. It accepts the following parameters:

throw new errors.PayloadTooLargeError('Uh oh, the file too big!');

The PolicyError class is a specific error designed to be used with route policies. The best practice recommendation is to ensure the name of the policy is passed in the details parameter. It accepts the following parameters:

throw new errors.PolicyError('Something went wrong', { policy: 'my-policy' });