Subscription Faker

A GraphQL server implementing "faked" GraphQL queries, mutations and subscriptions. You supply the (annotated) schema, this server generates your test data.

The faked data is configured using the special GraphQL schema directives @fake and @fake_list (see Usage below). The server implements a hot schema reload mechanism whereby schema changes take effect without the server needing to be restarted.

Subscription-faker differs from other GraphQL fakers in that it supports subscriptions as well as queries and mutations. When a subscription is initiated, data is faked as though it were generated with a query - updates are published by successively "morphing" the original data according to probabilities specified in the @fake() and @fake_list() directives in the schema. This emulates datasets being modified over time in a multi-user system. In turn, this can be handy for testing how your client responds to updates.

Installation

npm install subscription-faker

Usage

If installed with the optional -g flag the server can be started using the subscription-faker command. If installed locally (without the -g flag) it will need to be started using npx subscription-faker.

Command Line Options

Usage: main [options]

Options:
  -V, --version          output the version number
  -s, --schema <file>    Annotated GraphQL schema file (default: "schema.graphql")
  -h, --host <hostname>  Hostname or IP address to run server under (default: "localhost")
  -p, --port <number>    port number to run the server under (default: 4000)
  -u, --pathname <path>  The server pathname (default: "/graphql")
  -c, --period <period>  The update cycle period (in seconds) for subscriptions (default: 1)
  -v, --verbose          Echo generated data to stdout (default: false)
  -h, --help             output usage information

The @fake() schema directive

directive @fake(
    fake: String,                   # a JavaScript expression
    probability: Float = 0,         # 0.0 => never morphs, 1.0 => always morphs
) on FIELD_DEFINITION

The @fake() directive can be meaningfully applied to any scalar GraphQL field. The fake property of the directive should be a valid JavaScript expression; this will be evaluated and the resultant value assigned to the given datum. Within the expression, this references an instance of faker, see https://www.npmjs.com/package/faker for details of the faker API. See also the examples below. If no @fake() is associated with a scalar, the default value from the Apollo GraphQL server will be used (see https://www.apollographql.com/docs/apollo-server/features/mocking/ for more details).

The optional @fake() probability controls how the data changes if that field is returned by a GraphQL subscription. It should range from 0 (the data never changes - the default) to 1 (the data changes every subscription cycle). If the probability is set to 0.5, then on each subscription cycle there is a 50% probability that the data will be updated. The new, fake, data is regenerated using the original fake expression.

The @fake_list() schema directive

directive @fake_list(
    min: Int
    max: Int
    p_add: Float
    p_delete: Float
) on FIELD_DEFINITION

The fake_list() directive can be meaningfully attached to any GraphQL list. The min and max properties specify the minimum and maximum list lengths that should be generated; a random length will be chosen at run time.

The p_add and p_delete properties control how the list changes if it is returned as part of a GraphQL subscription. These probabilities (again ranging from 0 to 1) control how likely an element will be randomly added to the list or removed. The length of the list will not fall below the specified min or exceed the specified max - if it were to do so the update will be ignored.

Example

type Course {
    title: String           @fake(fake: "this.lorem.words()")
    author: String          @fake(fake: "this.name.findName()")
    description: String     
    url: String             @fake(fake: "this.internet.url()", probability: 0.1)
    sections: [Section]     @fake_list(min: 3, max: 7, p_add: 0.2, p_delete: 0.3)
    size: Int               @fake(fake: "this.random.number(100)", probability: 0.5)
}

Defines a GraphQL type called Course. When queried, subscribed to or returned from a mutation:

• The title property will be generated using the lorem/words method in the faker library.
• The author property will be generated with the faker name/findName method.
• The description property will fallback to the default mock value specified by the Apollo server.
• The url property will be generated with the faker internet/url method.
• The sections property will be generated with a randomly selected length of between 3 and 7 elements inclusive. Furthermore, on each subscription cycle there will be a 20% probability that a new element will be added (at a random location along the list) and a 30% probability that a randomly selected element will be removed. However, the list will always have at least 3 elements and never more than 7.
• Finally, the size property will be generated as a random number between 0 and 100 using random/number - illustrating that parameters can be passed into faker functions. On each subscription cycle there is a 50% probability that the number will be re-generated.

Support

Please raise issues on Github.

Contributing

Github pull requests are very welcome.

License

MIT

Project Status

Always interested in fault reports and enhancement suggestions.