Dictionary

When Specmatic generates requests while running tests or responses while running as a stub and no examples have been provided, Specmatic will generate values for fields based on the schema in the OpenAPI specifications.

While the generated values will conform to the schema, they may not be meaningful from the point of view of your business domain. Also at times you want to have more control on the exact values that are used for certain fields.

This is where dictionary capability helps you define domain-specific values which Specmatic can use in the absence of examples. Specmatic will lookup values for fields in the dictionary defined by you while generating requests or responses.

The dictionary can be supplied in either YAML or JSON format. When the dictionary file name follows the convention <spec-file-name>_dictionary.<format>, Specmatic will automatically pick it up in the context of the corresponding API specification file.

Structure

Fields are represented as nested properties that reflect the hierarchical structure of the schema.

  • Each entry in the dictionary maps a schema field to either a single value or a list of values.
  • If a single value is specified, it will be used directly.
  • If a list is specified, one of those values will be selected at random.

Basic Field Mapping

For example, given the Employee schema as follows:

components:
  schemas:
    Employee:
      type: object
      properties:
        name:
          type: string
        age:
          type: integer

Corresponding dictionary entries for the name and age properties would be:

  • Employee:
 name: John # Single-value
 age: # Multi-value
 - 20
 - 30
 - 40

  • {
  "Employee": {
    "name": "John",
    "age": [20, 30, 40]
  }
}
  • The name field is supplied as a single value and will be used directly.
  • The age field is supplied as a list of values, from which either 20, 30, or 40 will be pseudo-random selected.

Nested Properties

For nested structures, properties are represented as nested keys, where each level corresponds to a that level in the schema hierarchy. For example, given the Employee schema as follows:

components:
  schemas:
    Employee:
      type: object
      required:
        - name
      properties:
        name:
          type: object
          required:
            - first_name
          properties:
            first_name:
              type: string
            last_name:
              type: string

Corresponding dictionary entries for the first_name and last_name property would be:

  • Employee:
 name:
  first_name: John
  last_name: Smith

  • Employee:
 name:
  first_name: 
  - John
  - Jane
  last_name:
  - Smith
  - Doe

Handling Arrays

For properties that are arrays, the values must also be an array in the dictionary For example, given the Employee schema as follows:

components:
  schemas:
    Employee:
      type: object
      required:
        - aliases
      properties:
        aliases:
          type: array
          items:
            type: string

Corresponding dictionary entries for the aliases array would be:

  • Employee:
  aliases:
  - "John"
  - "Jane"

    The entire array will be used

  • Employee:
  aliases:
  - ["John", "Jane"] # JSON style array
  - - "May" # YAML style array
    - "Jones"

    One of the nested arrays will be pseudo-randomly selected and used

Nested properties in Arrays

For example, given the Employee schema as follows:

components:
  schemas:
    Employee:
      type: object
      required:
        - aliases
      properties:
        aliases:
          type: array
          items:
            type: object
            required:
              - first_name
            properties:
              first_name:

Corresponding dictionary entries for the first_name property within the aliases array would be:

  • Employee:
  aliases:
  - first_name: "John"

  • Employee:
  aliases:
  - first_name: "John"
  - first_name: "Jane"

This nesting behaves correctly because first_name is not defined as a top-level property in the schema. It is also not necessary to specify all properties of the object within the array, any missing properties will be populated by Specmatic.

Referencing Other Schemas

When a schema component utilizes $ref to reference another schema component, the dictionary for the referenced schema component with be defined with that schema component as the root entry (in other words directly start with the fields of the referenced schema)
For example, given the following Employee and Address schemas:

components:
  schemas:
    Employee:
      type: object
      required:
        - addresses
      properties:
        first_name:
          type: string
        addresses:
          $ref: '#/components/schemas/Address'
    Address:
      type: object
      required:
        - city
      properties:
        street:
          type: string
        city:
          type: string

Corresponding dictionary entries for any property defined in the Address schema would be:

  • Address:
  city: "New York"
  street: "Broadway"

  • Address:
  city:
  - "New York"
  - "London"
  street:
  - "Broadway"
  - "High Street"

Notice that the keys begin with Address instead than Employee, because the dictionary accesses the fields from the referenced schema. This makes for convenient dictionary definition without have to nest the values deeply.

Dictionary Generation

Commercial

Manually creating a dictionary can be quite an involved process, especially when the schema is complex. This is where, specmatic-openapi offers a convenient method to generate dictionaries from OpenAPI specifications and existing examples.

Automated dictionary generation is only available in the commercial version of Specmatic. For further details, please check the pricing page.

Let’s take a look at how we can generate a dictionary from an OpenAPI specification along with existing examples

Specification

Create an OpenApi Specification file named employees.yaml as follows:

openapi: 3.0.0
info:
  title: Employees
  version: '1.0'
servers: []
paths:
  /employees:
    post:
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/EmployeeDetails'
      responses:
        '200':
          description: Employee Created
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Employee'
        '400':
          description: Bad Request
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Error'
components:
  schemas:
    Employee:
      type: object
      required:
        - id
        - name
        - department
      properties:
        id:
          type: integer
        employeeCode:
          type: string
        name:
          type: string
        department:
          type: string
    EmployeeDetails:
      type: object
      required:
        - name
        - department
      properties:
        name:
          type: string
        department:
          type: string
    Error:
      type: object
      required:
        - message
      properties:
        message:
          type: string

Examples

Given that we have an OpenAPI specification, we should add a few examples for the /employees endpoint. Create a folder named employees_examples in the same directory as your employees.yaml file

  1. Let’s create an example for john named employee_john.json:

    {
 "http-request": {
     "path": "/employees",
     "method": "POST",
     "headers": {
         "Content-Type": "application/json"
     },
     "body": {
         "name": "John",
         "department": "IT"
     }
 },
 "http-response": {
     "status": 200,
     "body": {
         "id": 123,
         "employeeCode": "EMP123",
         "name": "John",
         "department": "IT"
     },
     "status-text": "OK",
     "headers": {
         "Content-Type": "application/json"
     }
 }
}
  2. Similarly, create an example for jane named employee_jane.json:

    {
 "http-request": {
     "path": "/employees",
     "method": "POST",
     "headers": {
         "Content-Type": "application/json"
     },
     "body": {
         "name": "Jane",
         "department": "HR"
     }
 },
 "http-response": {
     "status": 200,
     "body": {
         "id": 456,
         "employeeCode": "EMP456",
         "name": "Jane",
         "department": "HR"
     },
     "status-text": "OK",
     "headers": {
         "Content-Type": "application/json"
     }
 }
}
  3. Now let’s create a bad-request example named bad_request.json, where we’ve mutated the department field:

    {
 "http-request": {
     "path": "/employees",
     "method": "POST",
     "headers": {
         "Content-Type": "application/json"
     },
     "body": {
         "name": "James",
         "department": 123
     }
 },
 "http-response": {
     "status": 400,
     "body": {
         "message": "BAD REQUEST - Invalid field: department, value: 123 <number>, expected: <string>"
     },
     "status-text": "Bad Request",
     "headers": {
         "Content-Type": "application/json"
     }
 }
}

The _examples suffix is a naming convention that indicates to Specmatic to look for examples in that directory.

Generating the Dictionary

Create a file named employees_dictionary.yaml in the same directory as your employees.yaml file with an empty object i.e. {}, so we volume mount this file to the Docker container

After setting up, we can execute the examples dictionary command using the specmatic-openapi Docker image and provide the path to our employees.yaml file as shown below:

docker run --rm -v "$(pwd)/employees.yaml:/usr/src/app/employees.yaml" -v "$(pwd)/employees_examples:/usr/src/app/employees_examples" -v "$(pwd)/employees_dictionary.yaml:/usr/src/app/employees_dictionary.yaml" specmatic/specmatic-openapi examples dictionary --spec-file employees.yaml

Understanding the Dictionary

After executing the command, the employees_dictionary.yaml file will be updated with the following contents:

EmployeeDetails:
  name:
  - Jane
  - John
  - James
  department:
  - HR
  - IT
Employee:
  id:
  - 456
  - 123
  employeeCode:
  - EMP456
  - EMP123
  name:
  - Jane
  - John
  department:
  - HR
  - IT
Error:
  message:
  - 'BAD REQUEST - Invalid field: department, value: 123 <number>, expected: <string>'
  • Note how the hierarchy of the dictionary reflects the schema outlined in the OpenAPI specification.
  • Each key in the dictionary aligns with a property in the schema, while the values represent the possible values for that property.

We have not included the invalid value of department in the bad-request example. This filtering is not restricted to 4xx examples, the command will include only valid values in the dictionary, allowing it to be executed even with invalid examples.

Dictionary with Contract Testing

The Dictionary can be utilized in contract testing, allowing Specmatic to use the values defined in the dictionary when generating requests for tests. To illustrate this process, we will use the previous specification and dictionary as an example.

For the moment, we will remove the employees_examples directory to observe how contract testing operates without examples.

Run the tests

Now to execute contract tests on the specification using the dictionary a service is required, we will utilize service-virtualization for this purpose.

  • java -jar specmatic.jar stub employees.yaml

  • npx specmatic stub employees.yaml

  • docker run --rm --network host -v "$(pwd)/employees.yaml:/usr/src/app/employees.yaml" -v "$(pwd)/employees_dictionary.yaml:/usr/src/app/employees_dictionary.yaml" specmatic/specmatic stub "employees.yaml"

Next, execute the contract tests by running the following command:

  • java -jar specmatic.jar test employees.yaml

  • npx specmatic test employees.yaml

  • docker run --rm --network host -v "$(pwd)/employees.yaml:/usr/src/app/employees.yaml" -v "$(pwd)/employees_dictionary.yaml:/usr/src/app/employees_dictionary.yaml" specmatic/specmatic test "employees.yaml"

We can now examine the request sent to the service by reviewing the logs.

POST /employees
Accept-Charset: UTF-8
Accept: */*
Content-Type: application/json
{
    "name": "James",
    "department": "IT"
}

The values from the dictionary are used in the requests. Keep in mind that these values are selected in a pseudo-random manner from the list, so you may encounter different values each time.

Generative Tests

As it’s evident that only valid values can be included in the dictionary. hence, generative tests will ignore the values in the dictionary for the key being mutated. The other keys will still retrieve values from the dictionary if available; otherwise, random values will be generated.

For instance, if you execute the specification with generative tests enabled, one of the request will appear as follows:

POST /employees
Accept-Charset: UTF-8
Accept: */*
Content-Type: application/json
{
  "name": null,
  "department": "IT"
}

In this case, the key name is being mutated, which results in the value from the dictionary being disregarded. While the values for department is still being retrieved from the dictionary

Dictionary with Service Virtualization

The Dictionary can be utilized with service virtualization, where Specmatic will leverage the values in the dictionary to generate responses for incoming requests. To illustrate this functionality, let’s consider the previous specification and dictionary as an example. We will remove the employees_examples directory to observe how service virtualization functions without examples.

Run Service Virtualization

To start service virtualization, use the following command:

  • java -jar specmatic.jar stub employees.yaml

  • npx specmatic stub employees.yaml

  • docker run --rm -p 9000:9000 -v "$(pwd)/employees.yaml:/usr/src/app/employees.yaml" -v "$(pwd)/employees_dictionary.yaml:/usr/src/app/employees_dictionary.yaml" specmatic/specmatic stub "employees.yaml"

Making Requests

Execute the following curl command:

  • curl -X POST -H "Content-Type: application/json" -d "{\"name\":\"Jamie\",\"department\":\"IT\"}" http://localhost:9000/employees

  • Invoke-RestMethod -Uri "http://localhost:9000/employees" -Method Post -Headers @{"Content-Type"="application/json"} -Body '{"name":"Jamie","department":"IT"}'

You’ll get the following response:

{
  "id": 123,
  "employeeCode": "EMP123",
  "name": "Jane",
  "department": "IT"
}

The values from the dictionary are used in the responses. Keep in mind that these values are selected in a pseudo-random manner from the list, so you may encounter different values each time.

Dictionary with Examples

You can also utilize dictionary with the example commands and the examples interactive server, allowing values to be retrieved from the dictionary for both request and response generation of examples, refer to External Examples for more details.