GraphQL

Introduction

Specmatic supports service virtualization, contract testing and backward compatibility for GraphQL APIs based on GraphQL SDL (Schema Definition Language) files, similar to its support for REST (OpenAPI).

What Can You Achieve with Specmatic’s GraphQL Support?

With Specmatic GraphQL support you will be to leverage your GraphQL SDL files as executable contracts.

  1. Intelligent Service Virtualisation: Stub out GraphQL services for testing and development
  2. Contract Testing: Validate requests and responses against your GraphQL SDL files
  3. Backward Compatibility Checks: Compare two versions of your GraphQL SDL files to identify breaking changes without writing any code
  4. Central Contract Repo: Store your GraphQL SDL files in central Git repo which will serve as single source of truth for both providers and consumers
  5. API resiliency : Generate negative and edge cases to verify the resiliency of your API implementation based on your GraphQL SDL files.

Quick start with sample projects

Here are sample projects in different languages and frameworks that demonstrate below features in various languages and scenarios. Refer to the latest projects here:

README of each of these projects include

  • detailed animated architecture diagram
  • instructions on how to start stub server using graphql spec with custom examples, simulating delays, errors etc.
  • instructions on how to isolate the System Under Test when running contract tests
  • example CI workflow setups using GitHub Actions

Starting a stub server

To start a stub server using Specmatic, you can use the following steps:

Step 1: Create a graphql SDL file

Create a GraphQL SDL file named proudct-api.graphql as shown below.

enum ProductType {
  gadget
  book
  food
  other
}

type Query {
  findAvailableProducts(type: ProductType!, pageSize: Int): [Product]
}

type Product {
  id: ID!
  name: String!
  inventory: Int!
  type: ProductType!
}

Step 2: Start the stub server

Run the following command to start the GraphQL stub:

docker run -v "$PWD:/sandbox" -p 9000:9000 \
  specmatic/specmatic-graphql virtualize /sandbox/product-api.graphql

That’s it! You now have a GraphQL stub server running on localhost:9000/graphql serving the schema defined in product-api.graphql.

Step 3: Make a request to the stub server

Let’s try sending GraphQL queries to this server.

Use the following curl command to make a request:

curl -X POST http://localhost:9000/graphql \
 -H "Content-Type: application/json" \
 -d '{ "query": "query { findAvailableProducts(type: gadget, pageSize: 10) { id name inventory type } }" }'

You will receive a response with a random product ID, name, inventory, and type as shown below. The values are generated based on the GraphQL SDL file product-api.graphql you provided in the Step 2.

{
 "data": {
   "findAvailableProducts": [
     {
       "id": "YOTHQ",
       "name": "HAXVO",
       "inventory": 536,
       "type": "gadget"
     }
   ]
 }
}

Stubbing out specific values using example data

Above, we saw Specmatic generating random output based on the provided product-api.graphql spec. Instead, in order to receive specific example values, you can create an example YAML files containing specific data for findAvailableProducts query.

Let’s see how to do that.

Step 1. Create an examples file

Create product-api_examples/findAvailableProducts.yaml with following content

request:
  body: |
    query {
        findAvailableProducts(type: gadget, pageSize: 10) { id name inventory type }
    }

response: [
  {
    "id": "10",
    "name": "The Almanac",
    "inventory": 10,
    "type": "book"
  },
  {
    "id": "20",
    "name": "iPhone",
    "inventory": 15,
    "type": "gadget"
  }
]

Note: Keep the file product-api.graphql and directory product-api_examples at same level.

Step 2: Start the stub server

Run the following command to start the GraphQL stub:

docker run -v "$PWD:/spec" -p 9000:9000 \
  specmatic/specmatic-graphql virtualize /spec/product-api.graphql

Step 3: Make a request to the stub server

Let’s try sending the same request as earlier.

curl -X POST http://localhost:9000/graphql \
 -H "Content-Type: application/json" \
 -d '{ "query": "query { findAvailableProducts(type: gadget, pageSize: 10) { id name inventory type } }" }'

This time though, you will receive a response with the specified example defined in Step 1.

{
 "data": {
   "findAvailableProducts": [
     {
       "id": "10",
       "name": "The Almanac",
       "inventory": 10,
       "type": "book"
     },
     {
       "id": "20",
       "name": "iPhone",
       "inventory": 15,
       "type": "gadget"
     }
   ]
 }
}

Providing a custom examples directory

By convention, Specmatic automatically loads all examples defined for the GraphQL SDL file, say <graphql_sdl_filename>.graphql, from a colocated directory called <graphql_sdl_filename>_examples. However, in case your examples are in a different directory, you can pass it as an argument programmatically or through CLI args while running tests or service virtualization.

Example file file format

Let us now take deeper look at the external example format.

  • At the top level we have two YAML root nodes called request and response
  • request can hold the following keys:
    • body: this can contain either querys or mututations with exact values where necessary
    • headers: you can add your HTTP headers here
  • response holds responses with JSON syntax for readability, syntax highlighting and also as an aid to copy and paste of real responses from actual application logs etc.

HTTP Headers

Although GraphQL SDL files do not support HTTP request headers, you may directly add them to your Specmatic example files in httpHeaders under the request key, as seen in the example yaml below. The headers will be leveraged if present both by the contract tests as well as service virtualization.

request:
  httpHeaders:
    X-region: north-west
  body: |
    query {
        findAvailableProducts(type: gadget, pageSize: 10) { id name inventory type }
    }

response: [
  {
    "id": "10",
    "name": "The Almanac",
    "inventory": 10,
    "type": "book"
  },
  {
    "id": "20",
    "name": "iPhone",
    "inventory": 15,
    "type": "gadget"
  }
]

GraphQL Variables

Specmatic supports usage of GraphQL variables seamlessly. You only need to make sure that the externalised example is structured such that it uses the actual field values inline instead of variables in the query. Here is an example.

Say suppose, below request is that is being sent by your GraphQL Consumer to Specmatic GraphQL service virtualization server.

{
  "operationName": "FindAvailableProducts",
  "variables": {
    "type": "gadget",
    "pageSize": 10
  },
  "query": "query FindAvailableProducts($type: ProductType!, $pageSize: Int!) { findAvailableProducts(type: $type, pageSize: $pageSize) { id name inventory type } }"
}

As you can see, the above request from GraphQL consumer includes a variable called $pageSize. However in our example we will not be using it, instead we will be using the actual value (pageSize: 10) to match a request that comes with that value.

request:
  body: |
    findAvailableProducts(type: "gadget", pageSize: 10) {
      id
      name
      inventory
      type
    }
response: [
  {
    "id": "10",
    "name": "The Almanac",
    "inventory": 10,
    "type": "book"
  },
  {
    "id": "20",
    "name": "iPhone",
    "inventory": 15,
    "type": "gadget"
  }
]

Note: It is recommended to specify the type of the variable e.g. $pageSize: Int!. If the type is not specified explicitly you may face some issues since Specmatic will implicitly cast the variable to a certain type which may be invalid sometimes.

Delayed responses

To simulate delay in the response, you can use the delay-in-milliseconds key in your example YAML file. For example:

request:
  httpHeaders:
    X-region: north-west
  body: |
    query {
        findAvailableProducts(type: gadget, pageSize: 10) { id name inventory type }
    }

response: [
  {
    "id": "10",
    "name": "The Almanac",
    "inventory": 10,
    "type": "book"
  },
  {
    "id": "20",
    "name": "iPhone",
    "inventory": 15,
    "type": "gadget"
  }
]

delay-in-milliseconds: 5000

This will introduce a 5-second delay before the response is sent back to the client.

Dynamic Field Selection from Example Responses

When running a GraphQL stub using Specmatic, you can provide an example query that includes all possible fields and sub-selections. Specmatic uses this example to generate the stub response.

If a request is made using the exact query provided in the example, Specmatic will return the full response as defined. However, if a request is made with a subset of the fields specified in the original example, Specmatic will automatically tailor the response to include only those requested fields.

This means that a comprehensive example (an “uber example”) covering all fields can be reused by Specmatic to generate responses for queries that request only a subset of those fields.

Here are some simple steps to try this out:

1. Create a GraphQL SDL file named product-api.graphql with the following content:

   schema {
       query: Query
       mutation: Mutation
   }

   enum ProductType {
     gadget
     book
     food
     other
   }

   type Query {
     findAvailableProducts(type: ProductType!, pageSize: Int): [Product]
   }

   type Product {
     id: ID!
     name: String!
     inventory: Int!
     type: ProductType!
   }

2. Create a folder named examples.

3. Create an example file named find_available_gadgets.yaml in the examples folder with the following content:

   request:
     body: |
       query {
         findAvailableProducts(type: gadget, pageSize: 10) {
           id
           name
           inventory
           type
         }
       }
   response: [
       {
           "id": "10",
           "name": "The Almanac",
           "inventory": 10,
           "type": "book"
       },
       {
           "id": "20",
           "name": "iPhone",
           "inventory": 15,
           "type": "gadget"
       }
   ]

4. Create a specmatic.yaml file with the following content in the root folder where product-api.graphql file is present:

   contract_repositories:
     - type: filesystem
       consumes:
         - product-api.graphql

5. Run the following command to start the GraphQL stub on localhost:9000/graphql:

   docker run -v "$PWD/product-api.graphql:/usr/src/app/product-api.graphql" \
     -v "$PWD/examples:/usr/src/app/examples" \
     -v "$PWD/specmatic.yaml:/usr/src/app/specmatic.yaml" \
     -p 9000:9000 specmatic/specmatic-graphql virtualize --port=9000 --examples=examples

6. Make a request to query all the sub-selected fields specified in the example. You will get the exact response specified in the example. Use the following curl command to make this request:

   curl -X POST http://localhost:9000/graphql \
     -H "Content-Type: application/json" \
     -d '{
       "query": "query { findAvailableProducts(type: gadget, pageSize: 10) { id name inventory type } }"
     }'

Expected Response:

   {
     "data": {
       "findAvailableProducts": [
         {
           "id": "10",
           "name": "The Almanac",
           "inventory": 10,
           "type": "book"
         },
         {
           "id": "20",
           "name": "iPhone",
           "inventory": 15,
           "type": "gadget"
         }
       ]
     }
   }

7. Make a request to query only a subset of the sub-selected fields specified in the example. You will get a response with only those sub-selected fields with values picked up from the response. Use the following curl command to make this request:

   curl -X POST http://localhost:9000/graphql \
     -H "Content-Type: application/json" \
     -d '{
       "query": "query { findAvailableProducts(type: gadget, pageSize: 10) { id name } }"
     }'

Expected Response:

   {
     "data": {
       "findAvailableProducts": [
         {
           "id": "10",
           "name": "The Almanac"
         },
         {
           "id": "20",
           "name": "iPhone"
         }
       ]
     }
   }

This setup allows you to test how Specmatic reuses the example provided, adapting the response to the requested fields.

Multi-Query Requests

The Specmatic GraphQL stub server supports multi-query requests, allowing you to send a single request with multiple queries and receive a consolidated response. This feature is useful when you want to retrieve data from different queries in a single API call. Additionally, multi-query requests with variables are supported, making it flexible for dynamic requests.

To showcase this, let’s reuse the folder structure established in the previous section on dynamic field selection.

Steps to Try Out Multi-Query Requests

  1. Create the GraphQL SDL File: Create a file named product-api.graphql with the following schema:
    schema {
        query: Query
        mutation: Mutation
    }
    
    enum ProductType {
      gadget
      book
      food
      other
    }
    
    type Query {
      findAvailableProducts(type: ProductType!, pageSize: Int): [Product]
      productById(id: ID!): Product
    }
    
    type Product {
      id: ID!
      name: String!
      inventory: Int!
      type: ProductType!
    }
    
  2. Set Up Example Files: In the examples folder, create two example files:

    • find_available_gadgets.yaml:
      request:
        body: |
          query {
            findAvailableProducts(type: gadget, pageSize: 10) {
              id
              name
              inventory
              type
            }
          }
      response: [
          {
              "id": "10",
              "name": "The Almanac",
              "inventory": 10,
              "type": "book"
          },
          {
              "id": "20",
              "name": "iPhone",
              "inventory": 15,
              "type": "gadget"
          }
      ]
      
    • product_by_id.yaml:
      request:
        body: |
          query {
            productById(id: "10") {
              id
              name
              type
            }
          }
      response: {
          "id": "10",
          "name": "The Almanac",
          "type": "book"
      }
      
  3. Create specmatic.yaml File: Add the following specmatic.yaml file to the root folder:

    contract_repositories:
      - type: filesystem
        consumes:
          - product-api.graphql
    
  4. Run the GraphQL Stub:
    docker run -v "$PWD/product-api.graphql:/usr/src/app/product-api.graphql" \
      -v "$PWD/examples:/usr/src/app/examples" \
      -v "$PWD/specmatic.yaml:/usr/src/app/specmatic.yaml" \
      -p 9000:9000 specmatic/specmatic-graphql virtualize --port=9000 --examples=examples
    
  5. Send a Multi-Query Request: Use the following curl command to make a multi-query request:
    curl -X POST http://localhost:9000/graphql \
      -H "Content-Type: application/json" \
      -d '{
        "query": "query { findAvailableProducts(type: gadget, pageSize: 10) { id name inventory type } productById(id: \"10\") { id name type } }"
      }'
    

    Expected Response:

    {
      "data": {
        "findAvailableProducts": [
          {
            "id": "10",
            "name": "The Almanac",
            "inventory": 10,
            "type": "book"
          },
          {
            "id": "20",
            "name": "iPhone",
            "inventory": 15,
            "type": "gadget"
          }
        ],
        "productById": {
          "id": "10",
          "name": "The Almanac",
          "type": "book"
        }
      }
    }
    

    Here’s an example of a multi-query request using variables. This demonstrates how Specmatic can handle a single request with multiple queries, where each query may use variables.

  6. Send a Multi-Query Request with Variables: Use the following curl command:

    curl -X POST http://localhost:9000/graphql \
      -H "Content-Type: application/json" \
      -d '{
        "query": "query($type: ProductType!, $pageSize: Int, $id: ID!) { findAvailableProducts(type: $type, pageSize: $pageSize) { id name inventory type } productById(id: $id) { id name type } }",
        "variables": {
          "type": "gadget",
          "pageSize": 10,
          "id": "10"
        }
      }'
    

    Expected Response:

    {
      "data": {
        "findAvailableProducts": [
          {
            "id": "10",
            "name": "The Almanac",
            "inventory": 10,
            "type": "book"
          },
          {
            "id": "20",
            "name": "iPhone",
            "inventory": 15,
            "type": "gadget"
          }
        ],
        "productById": {
          "id": "10",
          "name": "The Almanac",
          "type": "book"
        }
      }
    }
    

In this example:

  • The variable $type is used to specify the ProductType for findAvailableProducts.
  • $pageSize controls the number of products returned.
  • $id is used to fetch a specific product by ID in the productById query.

This request showcases how Specmatic’s GraphQL stub server can process multi-query requests with variables, offering flexibility and efficiency in response generation.

This example demonstrates how Specmatic processes multiple queries in a single request and returns the expected responses for each query. You can adapt this setup for various use cases, leveraging the existing folder structure for organizing examples.

GraphQL Scalar Types

In GraphQL, you can define custom scalar types to handle specialized data, such as dates or monetary values, that require specific serialization and deserialization logic. For example:

schema {
  query: Query
  mutation: Mutation
}

scalar Date

type Offer {
  offerCode: String!
  validUntil: Date!
}

type Query {
  findOffersForDate(date: Date!): [Offer]!
}

In this schema, Date is a custom scalar. While GraphQL doesn’t provide details on how this scalar is serialized or deserialized, your application code defines the logic.

When testing queries like findOffersForDate using Specmatic, the tool doesn’t know how to correctly handle the Date scalar and might pass an incorrect value, such as a random string, leading to test failures.

To guide Specmatic, you can provide externalized examples that specify valid inputs for custom scalars like Date. For instance:

request:
  body: |
    query {
      findOffersForDate(date: "2024/12/31") { offerCode, validUntil }
    }

response: [
  {
    "offerCode": "WKND30",
    "validUntil": "2024/12/12"
  },
  {
    "offerCode": "SUNDAY20",
    "validUntil": "2024/12/25"
  }
]

Here, the Date scalar is provided with a valid value ("2024/12/31"). This ensures that during testing, Specmatic uses the correct format for the Date argument, allowing your tests to run smoothly without failures caused by incorrect data types.

Running contract tests

To run contract test:

docker run --network host -v "$(pwd)/specmatic.yml:/usr/src/app/specmatic.yml" -v "$(pwd)/build/reports/specmatic:/usr/src/app/build/reports/specmatic"  -e SPECMATIC_GENERATIVE_TESTS=true specmatic/specmatic-graphql test --port=8080

This command mounts your specmatic.yaml file and runs tests against a service running on port 8080 by generating GraphQL requests based on the GrapqhQL SDL files listed under provides section along with examples if any provided in the colocated directory named <GraphQL SDL file without extension>_examples. Also, it mounts the build artifacts from the docker container onto your local machine once the test run is completed.

Using your GraphQL files as your API Contracts from Central Contract Repository

  1. Store your GraphQL SDL files in a central repository for easy access and version control.
  2. Create a specmatic.yaml file in the root of your project to reference these contracts. Here’s an example:
version: 2
contracts:
  - git:
      url: https://github.com/specmatic/specmatic-order-contracts.git
    provides:
      - io/specmatic/examples/store/graphql/products_bff.graphqls
    consumes:
      - io/specmatic/examples/store/openapi/api_order_v3.yaml

Make sure to update the repository, provides and consumes sections to reflect your actual contract repository and .graphqls file locations.