Unsupported Features / Remaining Issues

Single Response Per Route
Limited Parameter Section Parsing
Heterogeneous Arrays of Objects
Enums
Multiple Parameter Ref tags
Swagger Inheritance and Composition

Single Response per Route

Due to Mountebanks weird style of “round-robining” through multiple responses set on a single path (uri + verb), SwaggerBank will only extract a single response that is defined with a status code of 200, 201, or default. Extracting each response means posting multiple responses to Mountebank for a single path which means round-robin behavior. I wanted to avoid this because round-robining will result in non-idempotent behavior when testing (hitting the same mocked route could return different responses depending on number of times route was hit prior). If you have any suggestions on a more elegant solution please open an issue and let me know.

Limited Parameter Section Parsing

My focus when I started SwaggerBank was to be able to intelligently and efficiency parse out response definitions and object schemas and generate random data to adhere to those schemas and definitions. I have not placed a lot of focus on the parameter sections of the spec. This means that for the mocked out routes, they place very little enforcement on the incoming request. ie: If the request matches the path + verb specified in the spec and includes the correct types for url parameters, then the request is considered a valid request. Any necessary body parameters, header values, query parameters, and/or auth headers that are required as per the spec are ignored. Again my initial focus was on the responses, not the requests. SwaggerBank 2.0 will include a completely revamped parsing of the parameter sections of the spec and a respective rehaul to MountebankHelper. With SwaggerBank 2.0, any required components for the request to a certain route (query parameters, body parameters, headers, etc) will also be required when hitting the mocked API. That way only a valid http request will result in the return of a mocked out response

Heterogeneous Arrays of Objects

Yeah, probably not going to support this. This is weird

  items:
    - properties:
        text:
          type: string
        ts:
          type: string
        type:
          type: string
        user:
          type: string
      type: object
    - properties:
        is_starred:
          type: boolean
        text:
          type: string
        ts:
          type: string
        type:
          type: string
        user:
          type: string
      type: object
    - properties:
        ts:
          type: string
        type:
          type: string
        wibblr:
          type: boolean
      type: object
  type: array

Enums

This is a big issues. WE DO NOT SUPPORT THIS STYLE OF DECLARING ENUMS
We have a type array, where each item of that array can only be one of the enum values

events:
  items:
    enum:
      - push
      - issues
      - issue_comment
      - commit_comment
      - pull_request
      - pull_request_review_comment
      - gollum
      - watch
      - download
      - fork
      - fork_apply
      - member
      - public
      - team_add
      - status
  type: array

We have a property, in this case encoding, that DOES NOT have a type (string, object, array, etc), but instead has an enum value. We will set the type to string (by default) and the format as enum

blob:
  properties:
    content:
      type: string
    encoding:
      enum:
        - utf-8
        - base64
    sha:
      type: string
    size:
      type: integer
  type: object

We have a property, in this mode, that HAS A TYPE (string) and also has a specified enum. This means the property must be a string, and must only be one of the enum values

  
  properties:
    mode:
      description: 'One of 100644 for file (blob), 100755 for executable (blob), 040000 for subdirectory (tree), 160000 for submodule (commit) or 120000 for a blob that specifies the path of a symlink.'
      enum:
        - '100644'
        - '100755'
        - '040000'
        - '160000'
        - '120000'
      type: string

MULTIPLE PARAMETER REF tags

Perhaps the biggest remaining issue left with SwaggerBank, this will cause the initial api.validate() method to fail (essentially SwaggerBank can’t even run). See explanation below

paths:
  '/calendars':
    parameters:
      - $ref: '#/parameters/alt'
      - $ref: '#/parameters/fields'
      - $ref: '#/parameters/key'
      - $ref: '#/parameters/oauth_token'
      - $ref: '#/parameters/prettyPrint'
      - $ref: '#/parameters/quotaUser'
      - $ref: '#/parameters/userIp'

Having multiple references to parameter objects within a single parameters object. SwaggerParser (the tool we use for parsing) DOES NOT SUPPORT THIS directly. The issue is that we DON’T RESOLVE INTERNAL references (because we generate our global template hash from these unresolved references) and therefore the parameters section remains as a LIST OF $REFS. SwaggerParser then assumes that the parameters section is a list of RESOLVED PARAMETERS containing “name”, “in”, “description”, etc which they are not Ex. It assumes that the first field for each “resolved parameter object” is the parameters name and the value is the parameter object. Because these “resolved parameter objects” are actually just a series of $ref: ‘#/parameters/blah1’ , $ref: ‘#/parameters/blah2’ statements, SwaggerParser raises an error on “duplicate parameter names” because it sees several consecutive ‘$ref’s

POSSIBLE SOLUTIONS:

Indeed resolve internal references and then reimplement logic for generating template hash (new logic wouldn’t be bad, would have to do performance tests on each, see if we can keep that large of JSON object in memory)
Continue to not resolve internal references and find another way to PREPROCESS this parameters section to manually resolve it.
Considering that the SwaggerParser.validate() method can take in a path to a file OR a COMPLETE SWAGGER OBJECT, we can actually give it a Swagger object that already has this parameters section (for each path) pre-resolved
Do not support list of parameter references

Swagger Inheritance and Composition

This could be supported, would require a lot of work

"Pet": {
    "discriminator": "petType",
    "properties": {
      "name": {
        "type": "string"
      },
      "petType": {
        "type": "string"
      }
    },
    "required": [
      "name",
      "petType"
    ]
  },
  "Cat": {
    "description": "A representation of a cat",
    "allOf": [
      {
        "$ref": "#/definitions/Pet"
      },
      {
        "properties": {
          "huntingSkill": {
            "type": "string",
            "description": "The measured skill for hunting",
            "default": "lazy",
            "enum": [
              "clueless",
              "lazy",
              "adventurous",
              "aggressive"
            ]
          }
        },
        "required": [
          "huntingSkill"
        ]
      }
    ]
  },
  "Dog": {
    "description": "A representation of a dog",
    "allOf": [
      {
        "$ref": "#/definitions/Pet"
      },
      {
        "properties": {
          "packSize": {
            "type": "integer",
            "format": "int32",
            "description": "the size of the pack the dog is from",
            "default": 0,
            "minimum": 0
          }
        },
        "required": [
          "packSize"
        ]
      }
    ]
  }

Potentially Problematic Things

A definition for an object that does not have the type: object field but is inferred to be an object based on the properties field. If this brings up problems, simply add the type: object field

thumbnail:
  properties:
    height:
      description: Thumnail height
      type: integer
    source:
      description: Thumbnail image URI
      type: string
    width:
      description: Thumbnail width
      type: integer

swagger-bank

A library to bridge Swagger API specifications and the Mountebank Test-Double tool