Guide: Setting up JSON validation

Track

Test bed setup

This guide walks you through the points to consider when setting up JSON validation and the steps to bring your validation service online.

What you will achieve

At the end of this guide you will have understood what you need to consider when starting to implement validation services for your JSON-based specification. You will also have gone through the steps to bring it online and make it available to your users.

An JSON validation service can be created using multiple approaches depending on your needs. You can have an on-premise (or local to your workstation) service through Docker or use the test bed’s resources and, with minimal configuration, bring online a public service that is automatically kept up-to-date.

For the purpose of this guide you will be presented the options to consider and start with a Docker-based instance that could be replaced (or complemented) by a setup through the test bed. Interestingly, the configuration relevant to the validator is the same regardless of the approach you choose to follow.

What you will need

  • About 30 minutes.

  • A text editor.

  • A web browser.

  • Access to the Internet.

  • Docker installed on your machine (only if you want to run the validator as a Docker container).

  • A basic understanding of JSON and JSON schema. A good source for more information here is the Understanding JSON schema tutorial site.

How to complete this guide

The steps described in this guide are for the most part hands-on, resulting in you creating a fully operational validation service. For these practical steps there are no prerequisites and the content for all files to be created are provided in each step. In addition, if you choose to try your setup as a Docker container you will also be issuing commands on a command line interface (all commands are provided and explained as you proceed).

Steps

You can complete this guide by following the steps described in this section. Not all steps are required, with certain ones being optional or complementary depending on your needs. The following diagram presents an overview of all steps highlighting the ones that apply in all cases (marked as mandatory):

../_images/step_overview.png

When and why you should skip or consider certain steps depends on your testing needs. Each step’s description covers the options you should consider and the next step(s) to follow depending on your choice.

Step 1: Determine your testing needs

Before proceeding to setup your validator you need to clearly determine your testing needs. A first outline of the approach to follow would be provided by answering the following questions:

  • Will the validator be available to your users as a tool to be used on an ad-hoc basis?

  • Do you plan on measuring the conformance of your community’s members to the JSON-based specification?

  • Is the validator expected to be used in a larger conformance testing context (e.g. during testing of a message exchange protocol)?

  • Should the validator be publicly accessible?

  • Should test data and validation reports be treated as confidential?

The first choice to make is on the type of solution that will be used to power your validation service:

  • Standalone validator: A service allowing validation of JSON content based on a predefined configuration of JSON schemas The service supports fine-grained customisation and configuration of different validation types (e.g. specification versions) and supported communication channels. Importantly, use of the validator is anonymous and it is fully stateless in that none of the test data or validation reports are maintained once validation completes.

  • Complete test bed: The test bed is used to realise a full conformance testing campaign. It supports the definition of test scenarios as test cases, organised in test suites that are linked to specifications. Access is account-based allowing users to claim conformance to specifications and execute in a self-service manner their defined test cases. All results are recorded to allow detailed reporting, monitoring and eventually certification. Test cases can address JSON validation but are not limited to that, allowing validation of any complex exchange of information.

It is important to note that these two approaches are by no means exclusive. It is often the case that a standalone validator is defined as a first step that is subsequently used from within test cases in the test bed. The former solution offers a community tool to facilitate work towards compliance supporting ad-hoc data validation, whereas the latter allows for rigorous conformance testing to take place where proof of conformance is required. This could apply in cases where conformance is a qualification criterion before receiving funding or before being accepted as a partner in a distributed system. Finally, it is interesting to consider that non-trivial JSON validation may involve multiple validation artefacts (e.g. different schemas for different message types). In such a case, even if ad-hoc data validation is not needed, defining a separate validator simplifies management of the validation artefacts by consolidating them in a single location, as opposed to bundling them within test suites.

Regardless of the choice of solution, the next point to consider will be the type of access. If public access is important then the obvious choice is to allow access over the Internet. An alternative would be an installation that allows access only through a restricted network, be it an organisation’s internal network or a virtual private network accessible only by your community’s members. Finally, an extreme case would be access limited to individual workstations where each community member would be expected to run the service locally (albeit of course without the expectation to test message exchanges with remote parties).

If access to your validation services over the Internet is preferred or at least acceptable, the simplest case is to opt for using the shared ISA² test bed resources, both regarding the standalone validator and the test bed itself. If such access is not acceptable or is technically not possible (e.g. access to private resources is needed), the proposed approach would be to go for a Docker-based on-premise installation of all components.

Summarising the options laid out in this section, you will first want to choose:

  • Whether you will be needing a standalone validator, a complete test bed or both.

  • Whether the validator and/or test bed will be accessible over the Internet or not.

Your choices here can help you better navigate the remaining steps of this guide. Specifically:

Step 2: Prepare validation artefacts

As an example case for JSON validation we will consider a variation of the EU purchase order case first seen in Guide: Creating a test suite. In short, for the purposes of this guide you are considered to be leading an EU cross-border initiative to define a new common specification for the exchange of purchase orders between retailers.

To specify the content of purchase orders your experts have created the following JSON schema:

{
    "$id": "http://itb.ec.europa.eu/sample/PurchaseOrder.schema.json",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "description": "A JSON representation of EU Purchase Orders",
    "type": "object",
    "required": [ "shipTo", "billTo", "orderDate", "items" ],
    "properties": {
      "orderDate": { "type": "string" },
      "shipTo":    { "$ref": "#/definitions/address" },
      "billTo":    { "$ref": "#/definitions/address" },
      "comment":   { "type": "string" },
      "items":     { 
          "type": "array",
          "items": { "$ref": "#/definitions/item" },
          "minItems": 1,
          "additionalItems": false
      }
    },
    "definitions": {
      "address": {
            "type": "object",
          "properties": {
                "name":   { "type": "string" },
                "street": { "type": "string" },
                "city":   { "type": "string" },
                "zip":    { "type": "number" }
            },
            "required": ["name", "street", "city", "zip"]
      },
      "item": {
            "type": "object",
          "properties": {
              "partNum":	   { "type": "string" },
                "productName": { "type": "string" },
                "quantity":    { "type": "number", "minimum": 0 },
                "priceEUR":    { "type": "number", "minimum": 0 },
                "comment":     { "type": "string" }
            },
            "required": ["partNum", "productName", "quantity", "priceEUR"]
      }    
    }
  }

Based on this, a sample purchase order would be as follows:

{
	"shipTo": {
		"name": "John Doe",
		"street": "Europa Avenue 123",
		"city": "Brussels",
		"zip": 1000
	},
	"billTo": {
		"name": "Jane Doe",
		"street": "Europa Avenue 210",
		"city": "Brussels",
		"zip": 1000
	},
	"orderDate": "2020-01-22",
	"comment": "Send in one package please",
	"items": [
		{
			"partNum": "XYZ-123876",
			"productName": "Mouse",
			"quantity": 20,
			"priceEUR": 15.99,
			"comment": "Confirm this is wireless"
		},
		{
			"partNum": "ABC-32478",
			"productName": "Keyboard",
			"quantity": 15,
			"priceEUR": 25.50
		}
	]
}

A first obvious validation for purchase orders would be against the defined JSON schema. However, your business requirements also define the concept of a large purchase order which is one that includes more than 10 of each ordered item. This restriction is not reflected in the JSON schema which is considered as a base for all purchase orders but rather in a separate JSON schema file that checks this only for orders that are supposed to be “large”. Such a rule file would be as follows:

{
    "$id": "http://itb.ec.europa.eu/sample/PurchaseOrder-large.schema.json",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "description": "Business rules for large EU Purchase Orders expressed in JSON",
    "type": "object",
    "required": [ "items" ],
    "properties": {
      "items": { 
          "type": "array",
          "items": {
              "type": "object"
          },
          "minItems": 10
      }
    }
  }

As you see in the content of the two schemas, the first one defines the structure of the expected JSON objects and their properties, whereas the second one does not replicate structural checks, focusing only on the number of items. In this case a valid large purchase order would be expected to validate against both schemas.

Given these requirements and validation artefacts we want to support two types of validation (or profiles):

  • basic: For all purchase orders acting as a common base. This is realised by validating against PurchaseOrder.schema.json.

  • large: For large purchase orders. This is realised by validating against PurchaseOrder.schema.json and PurchaseOrder-large.schema.json.

As the first configuration step for the validator we will prepare a folder with the required resources. For this purpose create a root folder named validator with the following subfolders and files:

validator
 |
 +-- resources
      |
      +-- order
           |
           +-- schemas
                |
                +-- PurchaseOrder.schema.json
                +-- PurchaseOrder-large.schema.json

Regarding the PurchaseOrder.schema.json and PurchaseOrder-large.schema.json files you can create them from the above content or download them (here: PurchaseOrder.schema.json and PurchaseOrder-large.schema.json). Finally, note that you are free to use any names for the files and folders; the ones used here will however be the ones considered in this guide’s subsequent steps.

Step 3: Prepare validator configuration

After having defined your testing needs and the validation artefacts for your specific case, the next step will be to configure the validator. The validator is defined by a core engine maintained by the test bed team and a layer of configuration, provided by you, that defines its use for a specific scenario. In terms of features the validator supports the following:

  • Validation channels including a SOAP web service API and a web user interface.

  • Configuration of JSON schemas to drive the validation that can be local or remote.

  • Definition of different validation types as logically-related sets of validation artefacts.

  • Support per validation type allowing user-provided schemas.

  • Definition of separate validator configurations that are logically split but run as part of a single validator instance. Such configurations are termed “validation domains”.

  • Customisation of all texts presented to users.

Configuration is provided by means of key-value pairs in a property file. This file can be named as you want but needs to end with the .properties extension. In our case we will name this config.properties and place it within the order folder. Recall that the purpose of this folder is to store all resources relevant to purchase order validation. These are the validation artefacts themselves (PurchaseOrder.schema.json and PurchaseOrder-large.schema.json) and the configuration file (config.properties).

Define the content of the config.properties file as follows:

# The different types of validation to support. These values are reflected in other properties.
validator.type = basic, large
# Labels to describe the defined types.
validator.typeLabel.basic = Basic purchase order
validator.typeLabel.large = Large purchase order
# Validation artefacts (JSON schema) to consider for the "basic" type.
validator.schemaFile.basic = schemas/PurchaseOrder.schema.json
# Validation artefacts (JSON schema) to consider for the "large" type.
validator.schemaFile.large = schemas/PurchaseOrder.schema.json, schemas/PurchaseOrder-large.schema.json
# The title to display for the validator's user interface.
validator.uploadTitle = Purchase Order Validator

All validator properties share a validator. prefix. The validator.type property is key as it defines one or more types of validation that will be supported (multiple are provided as a comma-separated list of values). The values provided here are important not only because they define the available validation types but also because they drive most other configuration properties. Regarding the validation artefacts themselves, these are provided by means of the validator.schemaFile properties:

  • validator.schemaFile.TYPE defines one or more (comma-separated) file paths (relative to the configuration file) to lookup schema files.

Using these properties you define the validator’s validation artefacts as local files, where in both cases each provided path can be for a file or a folder. If a folder is referenced it will load all contained top-level files (i.e. ignoring subfolders).

Note

Further validation artefact configuration: You may also define validation artefacts as remote resource references and/or as being user-provided.

The purpose of the remaining properties is to customise the text descriptions presented to users:

  • validator.typeLabel defines a label to present to users on the validator’s user interface for the type in question.

  • validator.uploadTitle defines the title label to present to users on the validator’s user interface.

Once you have created the config.properties file, the validator folder should be as follows:

validator
 |
 +-- resources
      |
      +-- order
           |
           +-- config.properties
           +-- schemas
                |
                +-- PurchaseOrder.schema.json
                +-- PurchaseOrder-large.schema.json

When you are defining multiple schema files for a given validation type you may also want to specify how these are combined. This is done by means of the validator.schemaFile.TYPE.combinationApproach property (for a given TYPE) that accepts of of three values:

  • allOf: Content must validate against all defined schemas. This is the default if not specified.

  • anyOf: Content must validate against any of the defined defined schemas.

  • oneOf: Content must validate against one, and only one, of the defined schemas.

In our configuration for large purchase orders we define two schemas which by default are applied with allOf semantics. If the schemas were rather two possible alternatives a anyOf value would be more appropriate. This would be configured as follows:

validator.type = basic, large
...
validator.schemaFile.large = schemas/PurchaseOrder.schema.json, schemas/PurchaseOrder-large.schema.json
validator.schemaFile.large.combinationApproach = anyOf

The limited configuration file we have prepared assumes numerous default configuration properties. An important example is that by default, the validator will expose a web user interface and a SOAP web service API. This configuration is driven through the validator.channels property that by default is set to form, soap_api (for a user form and SOAP web service respectively). All configuration properties provided in config.properties relate to the specific domain in question, notably purchase orders, reflected in the validator’s resources as the order folder. Although rarely needed, you may define additional validation domains each with its own set of validation artefacts and configuration file (see Configuring additional validation domains for details on this). Finally, if you are planning to host your own validator instance you can also define configuration at the level of the complete validator (see Additional configuration options regarding application-level configuration options).

For the complete reference of all available configuration properties and their default values refer to section Validator configuration properties.

Remote validation artefacts

Defining the validator’s artefacts as local files is not the only option. If these are available online you can also reference them remotely by means of property validator.schemaFile.TYPE.remote.N.url. The N element in the properties’ names is a zero-based positive integer allowing you to define more than one entries to match the number of remote files.

The example that follows illustrates the loading of two remote schemas for a validation type named v2.2.1 from a remote location:

validator.type = v2.2.1
...
validator.schemaFile.v2.2.1.remote.0.url = https://my.server.com/my_schema_1.json
validator.schemaFile.v2.2.1.remote.1.url = https://my.server.com/my_schema_2.json

You may also combine local and remote Schematron files by defining a validator.schemaFile.TYPE property and one or more validator.schemaFile.TYPE.remote.N.url properties. In all cases, the schemas from all sources will be aggregated into a single model for the validation. Such combinations are not possible for XML Schemas where only one schema source is considered.

Note

Remote schema caching: Caching is used to avoid constant lookups of remote schema files. Once loaded, remote schemas will be automatically refreshed every hour.

User-provided validation artefacts

Another available option on schema file configuration is to allow a given validation type to support user-provided schemas. Such schemas would be considered in additional to any pre-configured local and remote schemas. Enabling user-provided schemas is achieved through the validator.externalSchemas property:

...
validator.externalSchemas.TYPE = required

These properties allow three possible values:

  • required: The relevant schema(s) must be provided by the user.

  • optional: Providing the relevant schema(s) is allowed but not mandatory.

  • none (the default value): No such schema(s) are requested or considered.

Specifying that for a given validation type you allow users to provide schemas will result in any such schemas being combined with your pre-defined ones. This could be useful in scenarios where you want to define a common validation base but allow also ad-hoc extensions for e.g. restrictions defined at user-level (e.g. National validation rules to consider in addition to a common set of EU rules). Similarly to pre-defined schemas, you can also define the validator.externalSchemaCombinationApproach.TYPE with values allOf (the default), anyOf and oneOf to specify how they are combined. Note that when you have pre-configured schemas and user-provided ones, these are validated separately based on the defined combination semantics (properties validator.schemaFile.TYPE.combinationApproach and validator.externalSchemaCombinationApproach.TYPE) but for an overall success both sets of schemas need to succeed.

Note

Generic validator: It is possible to not predefine any schemas resulting in a validator that is truly generic, expecting all schemas to be provided by users. Such a generic instance actually exists at https://www.itb.ec.europa.eu/json/any/upload.

Step 4: Setup validator as Docker container

Note

When to setup a Docker container: The purpose of setting up your validator as a Docker container is to host it yourself or run it locally on workstations. If you prefer or don’t mind the validator being accessible over the Internet it is simpler to delegate hosting to the test bed team by reusing the test bed’s infrastructure. If this is the case skip this section and go directly to Step 5: Setup validator on test bed. Note however that even if you opt for a validator managed by the test bed, it may still be interesting to create a Docker image for development purposes (e.g. to test new validation artefact versions) or to make available to your users as a complementary service (i.e. use online or download and run locally).

Once the validator’s configuration is ready (configuration file and validation artefacts) you can proceed to create a Docker image.

The configuration for your image is driven by means of a Dockerfile. Create this file in the validator folder with the following contents:

FROM isaitb/json-validator:latest
COPY resources /validator/resources/
ENV validator.resourceRoot /validator/resources/

This Dockerfile represents the most simple Docker configuration you can provide for the validator. Let’s analyse each line:

FROM isaitb/json-validator:latest

This tells Docker that your image will be built over the latest version of the test bed’s isaitb/json-validator image. This represents the validator’s core that expects configuration to drive the validation. It is available on the public Docker Hub and as such can be directly used through any Docker installation with Internet access.

COPY resources /validator/resources/

This copies your resources folder to the image under path /validator/resources/.

ENV validator.resourceRoot /validator/resources/

This instructs the validator that it should consider as the root of all its configuration resources the /validator/resources/ folder (which was just copied into it).

The contents of the validator folder should now be as follows:

validator
 |
 +-- Dockerfile
 +-- resources
      |
      +-- order
           |
           +-- config.properties
           +-- shapes
                |
                +-- PurchaseOrder.schema.json
                +-- PurchaseOrder-large.schema.json

That’s it. To build the Docker image open a command prompt to the validator folder and issue:

docker build -t po-validator .

This command will create a new local Docker image named po-validator based on the Dockerfile it finds in the current directory. It will proceed to download missing images (e.g. the isaitb/json-validator:latest image) and eventually print the following output:

Sending build context to Docker daemon  32.77kB
Step 1/3 : FROM isaitb/json-validator:latest
---> 39ccf8d64a50
Step 2/3 : COPY resources /validator/resources/
---> 66b718872b8e
Step 3/3 : ENV validator.resourceRoot /validator/resources/
---> Running in d80d38531e11
Removing intermediate container d80d38531e11
---> 175eebf4f59c
Successfully built 175eebf4f59c
Successfully tagged po-validator:latest

The new po-validator:latest image can now be pushed to a local Docker registry or to the Docker Hub. In our case we will proceed directly to run this as follows:

docker run -d --name po-validator -p 8080:8080 po-validator:latest

This command will create a new container named po-validator based on the po-validator:latest image you just built. It is set to run in the background (-d) and expose its internal listen port through the Docker machine (-p 8080:8080). Note that by default the listen port of the container (which you can map to any available host port) is 8080.

Your validator is now online and ready to validate JSON content. If you want to try it out immediately skip to Step 6: Use the validator. Otherwise, read on to see additional configuration options for the image.

Configuring additional validation domains

Up to this point you have configured validation for purchase orders which defines one or more validation types (basic and large). This configuration can be extended by providing additional types to reflect:

  • Additional profiles with different business rules (e.g. minimal).

  • Specification versions (e.g. basic_v1.0, large_v1.0, basic_v1.1_beta).

  • Other types of content that are linked to purchase orders (e.g. purchase_order_basic_v1.0 and order_receipt_v1.0).

All such extensions would involve defining potentially additional validation artefacts and updating the config.properties file accordingly.

Apart from extending the validation possibilities linked to purchase orders you may want to configure a completely separate validator to address an unrelated specification that would most likely not be aimed to the same user community. To do so you have two options:

  • Repeat the previous steps to define a separate configuration and a separate Docker image. In this case you would be running two separate containers that are fully independent.

  • Reuse your existing validator instance to define a new validation domain. The result will be two validation services that are logically separate but are running as part of a single validator instance.

The rationale behind the second option is simply one of required resources. If you are part of an organisation that needs to support validation for dozens of different types of JSON content that are unrelated, it would probably be preferable to have a single application to host rather than one per specification.

In your current single domain setup, the purchase order configuration is reflected through folder order. The name of this folder is also by default assumed to match the name of the domain. A new domain could be named invoice that is linked to JSON invoices. This is represented by an invoice folder next to order that contains similarly its validation artefacts and domain-level configuration property file. Considering this new domain, the contents of the validator folder would be as follows:

validator
 |
 +-- Dockerfile
 +-- resources
      |
      +-- invoice
        (Further contents skipped)
      +-- order
        (Further contents skipped)

If you were now to rebuild the validator’s Docker image this would setup two logically-separate validation domains (invoice and order).

Note

Validation domains vs types: In almost all scenarios you should be able to address your validation needs by having a single validation domain with multiple validation types. Validation types under the same domain will all be presented as options for users. Splitting in domains would make sense if you don’t want the users of one domain to see the supported validation types of other domains.

Important: Support for such configuration is only possible if you are defining your own validator as a Docker image. If you plan to use the test bed’s shared validator instance (see Step 5: Setup validator on test bed), your configuration needs to be limited to a single domain. Note of course that if you need additional domains you can in this case simply repeat the configuration process multiple times.

Additional configuration options

We have seen up to now that configuring how validation takes place is achieved through domain-level configuration properties provided in the domain configuration file (file config.properties in our example). When setting up the validator as a Docker image you may also make use of application-level configuration properties to adapt the overall validator’s operation. Such configuration properties are provided as environment variables through ENV directives in the Dockerfile.

We already saw this when defining the validator.resourceRoot property that is the only mandatory property for which no default exists. Other such properties that you may choose to override are:

  • validator.domain: A comma-separated list of names that are to be loaded as the validator’s domains. By default the validator scans the provided validator.resourceRoot folder and selects as domains all subfolders that contain a configuration property file (folder order in our case). You may want to configure the list of folder names to consider if you want to ensure that other folders get ignored.

  • validator.domainName.DOMAIN: A mapping for a domain (replacing the DOMAIN placeholder) that defines the name that should be presented to users. This would be useful if the folder name itself (order in our example) is not appropriate (e.g. if the folder was named files).

The following example Dockerfile illustrates use of these properties. The values set correspond to the applied defaults so the resulting Docker images from this Dockerfile and the original one (see Step 4: Setup validator as Docker container) are in fact identical:

FROM isaitb/json-validator:latest
COPY resources /validator/resources/
ENV validator.resourceRoot /validator/resources/
ENV validator.domain order
ENV validator.domainName.order order

See Application-level configuration for the full list of supported application-level properties.

Finally, it may be the case that you need to adapt further configuration properties that relate to how the validator’s application is ran. The validator is built as a Spring Boot application which means that you can override all configuration properties by means of environment variables. This is rarely needed as you can achieve most important configuration through the way you run the Docker container (e.g. defining port mappings). Nonetheless the following adapted Dockerfile shows how you could ensure the validator’s application starts up on another port (9090) and uses a specific context path (/ctx).

FROM isaitb/json-validator:latest
COPY resources /validator/resources/
ENV validator.resourceRoot /validator/resources/
ENV server.servlet.context-path /ctx
ENV server.port 9090

Note

Custom port: Even if you define the server.port property to a different value other than the default 8080 this remains internal to the Docker container. The port through which you access the validator will be the one you map on your host through the -p flag of the docker run command.

The full list of such application configuration properties, as well as their default values, are listed in the Spring Boot configuration property documentation.

Step 5: Setup validator on test bed

Note

When to setup on test bed resources: Setting up your validator on the test bed’s resources removes hosting concerns and allows you to benefit from automatic service reloads for configuration changes. In doing so however you need to keep in mind that the validator will be exposed over the Internet. If this approach is not suitable for you (e.g. you want to expose the validator within a restricted network) you should consider setting up the validator as a Docker container (see Step 4: Setup validator as Docker container) that you can then host as you see fit.

To configure a validator using the test bed’s resources all you need to do is get in touch with the test bed team and provide the validator’s configuration. Specifically:

  1. Send an email to DIGIT-ITB@ec.europa.eu describing your case: This step is needed for two reasons. Firstly you may want to have a further discussion and potentially a demo to better understand the available options. Secondly the test bed’s team would need to ensure that you qualify to use its resources (to e.g. avoid that you are a private company planning to offer commercial validation services).

  2. Share the configuration for the validator: Once contact has been established you need to provide the initial configuration for the validator.

Regarding the second step, the validator’s configuration to be shared is the contents of the validator folder as described in Step 3: Prepare validator configuration. The eventual goal here will be to have the configuration available through an accessible Git repository. This can be done in a number of ways listed below in decreasing order of preference:

  • Create a new Git repository: You can push all resources (the validator folder) to a new Git repository (e.g. on GitHub or the European Commission’s CITNet Bitbucket server). You can of course add any other resources to this repository as you see fit (e.g. a README file). Once done provide the repository’s URL to the test bed team.

  • Provide the resources to the test bed team: You can send the configuration files themselves to the test bed’s team (e.g. make an archive of the validator folder). Ideally you should define the configuration file but if in doubt you can simply describe the resources and the test bed team will prepare the initial configuration for you. When following this approach a new Git repository will be created for you on the European Commission’s CITNet Bitbucket server or on GitHub for which you will be assigned write access (assuming you have a relevant user account).

  • Update an existing Git repository: If you already have a Git repository to maintain the validation artefacts you can reuse this by adding to it the required configuration file (config.properties in our case). When ready you will need to provide the test bed team with the URL to the repository and the location of the configuration file.

Following the initial configuration, the resulting Git repository will be monitored to detect any changes to the validation artefacts or the configuration file. If such a change is detected, the validation service will be automatically updated within a few minutes.

Note

Using a dedicated Git repository for the validator: Whether you define a new Git repository yourself or the test bed team creates one for you, the result is a repository that is dedicated to the validator. This approach is preferable to reusing an existing Git repository to avoid unwanted changes to the validator. whether or not this is done through GitHub, CITNet’s Bitbucket or another service depends on what best suits your needs.

As part of the initial setup for the validator the test bed team will also configure how it is accessed. The name used will match the name of the folder that contains your configuration file (order in the considered example), but this can differ according to your preferences. If this is the case make sure to inform the test bed team of your preferred naming.

Considering our example, for a name of order, the resulting root URL through which the validator will be accessed is https://www.itb.ec.europa.eu/json/order. The specific paths will depend on the supported validation channels as described in Step 6: Use the validator.

Step 6: Use the validator

Well done! At this step your validator has been successfully configured and is ready to use. Depending on which approach was followed, this may have been done either:

The validation channels that are supported depend on the configuration you have supplied. This is done through the validator.channels property of your configuration file (config.properties) that defaults to form, soap_api. The supported channels are as follows:

  • form: A web user interface allowing a user to provide the JSON content to validate.

  • soap_api: A SOAP API allowing contract-based machine-to-machine integration using SOAP web service calls.

The following sub-sections describe how each channel can be used considering the example EU purchase order specification.

Validation via user interface

The validator’s user interface is available at the /json/DOMAIN/upload path. The exact path depends on how this is deployed:

The first page that you see is a simple form to provide the JSON content to validate.

../_images/validator_upload.png

This form expects the following input:

  • Content to validate: The JSON content that will be submitted for validation. The preceding dropdown selection determines how this will be provided, specifically as a file input (pre-selected), as a URI to be loaded remotely or as content to be provided using an editor.

  • Validate as: The type of validation to apply.

Note that all displayed labels can be adapted through the config.properties configuration file (see Properties related to UI labels). The available validation types match the ones defined in the validator.type property, displayed using the validator.typeLabel.TYPE labels. Moreover, the text title could be replaced by a configurable HTML banner, and further complemented with a HTML footer (see Domain-level configuration).

../_images/validator_upload_selected.png

It is worth noting also that if your configuration defined only a single validation type, the user interface would be simplified by presenting only the content input controls (i.e. considering the single validation type as pre-selected).

../_images/validator_upload_simple.png

In addition, if your configuration for the selected validation type allows for user-provided schemas, the form also includes the controls to manage the files you provide. Files can be defined via file upload or remote URI and they can be mandatory or optional depending on your configuration. In case multiple user-provided schemas are defined the interface is also extended with the option on how these should be combined.

../_images/validator_upload_external.png

Once you have provided your input click the Validate button to trigger the validation. Upon completion you will be presented with the validation results:

../_images/validator_result.png

This screen includes an overview of the result listing:

  • The validation timestamp (in UTC), the name of the validated file and the applied validation type (if more than one are configured).

  • The overall result (SUCCESS or FAILURE).

  • The number of errors, warnings and information messages.

This section is followed by the Details panel, where the details of each report item are listed:

  • It’s type (whether this is an error, warning or information message).

  • It’s description.

Clicking on each item’s details will open a popup that shows within the provided content the specific point that triggered the issue:

../_images/validator_result_errordetail.png

In terms or reporting, apart from the on-screen display, buttons are available allowing you to download the validation report:

  • Download XML report: Download as XML (in GITB TRL syntax - sample here).

  • Download PDF report: Download as PDF (sample here).

Note that these download buttons are initially disabled but are enabled as soon as the respective reports become available.

Finally, to trigger a new validation you may either use the form from the top of the result screen or click on the form’s title that will take you back to the previous page.

Validation via minimal user interface

If you are exposing a web user interface (see Validation via user interface) for your validator you also have the option of enabling an alternative minimal interface that could be used as an embedded component in another web page (e.g. via an iframe). This is enabled through the validator.supportMinimalUserInterface property in your domain configuration (file config.properties).

...
validator.supportMinimalUserInterface = true

The result of this is to expose a /uploadm path. The path depends on how this is deployed:

The minimal interface offers largely the same functionality as the complete one but with a more condensed layout and minimal styling. The initial input page you see for the validator is as follows:

../_images/validator_upload_minimal.png

The most significant difference is the result page which provides only an overview and the relevant download controls:

../_images/validator_result_minimal.png

The meaning of the provided controls and displayed information for both input and result pages is identical to the complete user interface (see Validation via user interface).

Validation via SOAP web service API

The validator’s SOAP API is available under the /json/soap path. The exact path depends on how this is deployed (path to WSDL provided):

The SOAP API used is the GITB validation service API, meaning that the validator is a GITB-compliant validation service. The importance of this is that apart from using it directly, this SOAP API allows integration of the validator in more complex conformance testing scenarios as a validation step in GITB TDL test cases. This potential is covered further in Step 7: Use the validator in GITB TDL test cases.

The operations supported are as follows:

  • getModuleDefinition: Called to return information on how to call the service (i.e. what inputs are expected).

  • validate: Called to trigger validation for provided content.

You can download this SOAP UI project that includes sample calls of these operations (make sure to change the service URL to match your setup).

Regarding the getModuleDefinition operation, a request of:

<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:v1="http://www.gitb.com/vs/v1/">
   <soapenv:Header/>
   <soapenv:Body>
      <v1:GetModuleDefinitionRequest/>
   </soapenv:Body>
</soapenv:Envelope>

Will produce a response as follows:

<soap:Envelope xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
   <soap:Body>
      <ns4:GetModuleDefinitionResponse xmlns:ns2="http://www.gitb.com/core/v1/" xmlns:ns3="http://www.gitb.com/tr/v1/" xmlns:ns4="http://www.gitb.com/vs/v1/">
         <module operation="V" id="ValidatorService">
            <ns2:metadata>
               <ns2:name>ValidatorService</ns2:name>
               <ns2:version>1.0.0</ns2:version>
            </ns2:metadata>
            <ns2:inputs>
               <ns2:param type="binary" name="contentToValidate" use="R" kind="SIMPLE" desc="The content to validate, provided as a string, BASE64 or a URI."/>
               <ns2:param type="string" name="embeddingMethod" use="O" kind="SIMPLE" desc="The embedding method to consider for the 'contentToValidate' input ('BASE64', 'URL' or 'STRING')."/>
               <ns2:param type="string" name="validationType" use="O" kind="SIMPLE" desc="The type of validation to perform (if multiple types are supported)."/>
               <ns2:param type="list[map]" name="externalSchemas" use="O" kind="SIMPLE" desc="A list of maps that defines external schemas to consider in addition to any preconfigured ones. Each map item corresponds to a schema file and defines the following keys: 'content' (the schema content to consider, see 'contentToValidate' for its semantics), 'embeddingMethod' (the way to consider the 'content' value)."/>
               <ns2:param type="boolean" name="externalSchemaCombinationApproach" use="O" kind="SIMPLE"/>
               <ns2:param type="boolean" name="locationAsPointer" use="O" kind="SIMPLE" desc="Whether or not the location reported for returned errors will be a JSON pointer (default false). False will return the line number in the input."/>
            </ns2:inputs>
         </module>
      </ns4:GetModuleDefinitionResponse>
   </soap:Body>
</soap:Envelope>

This response can be customised through configuration properties in config.properties to provide descriptions specific to your setup. For example, extending config.properties with the following:

...
validator.webServiceId = PurchaseOrderValidator
validator.webServiceDescription.contentToValidate = The purchase order content to validate
validator.webServiceDescription.validationType = The type of validation to perform ('basic' or 'large')

Will produce a response as follows:

<soap:Envelope xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
   <soap:Body>
      <ns4:GetModuleDefinitionResponse xmlns:ns2="http://www.gitb.com/core/v1/" xmlns:ns3="http://www.gitb.com/tr/v1/" xmlns:ns4="http://www.gitb.com/vs/v1/">
         <module operation="V" id="ValidatorService">
            <ns2:metadata>
               <ns2:name>PurchaseOrderValidator</ns2:name>
               <ns2:version>1.0.0</ns2:version>
            </ns2:metadata>
            <ns2:inputs>
               <ns2:param type="binary" name="contentToValidate" use="R" kind="SIMPLE" desc="The purchase order content to validate"/>
               <ns2:param type="string" name="embeddingMethod" use="O" kind="SIMPLE" desc="The embedding method to consider for the 'contentToValidate' input ('BASE64', 'URL' or 'STRING')."/>
               <ns2:param type="string" name="validationType" use="O" kind="SIMPLE" desc="The type of validation to perform ('basic' or 'large')"/>
               <ns2:param type="list[map]" name="externalSchemas" use="O" kind="SIMPLE" desc="A list of maps that defines external schemas to consider in addition to any preconfigured ones. Each map item corresponds to a schema file and defines the following keys: 'content' (the schema content to consider, see 'contentToValidate' for its semantics), 'embeddingMethod' (the way to consider the 'content' value)."/>
               <ns2:param type="boolean" name="externalSchemaCombinationApproach" use="O" kind="SIMPLE"/>
               <ns2:param type="boolean" name="locationAsPointer" use="O" kind="SIMPLE" desc="Whether or not the location reported for returned errors will be a JSON pointer (default false). False will return the line number in the input."/>
            </ns2:inputs>
         </module>
      </ns4:GetModuleDefinitionResponse>
   </soap:Body>
</soap:Envelope>

Running the validation itself is done through the validate operation. This expects the following inputs:

  • contentToValidate: The JSON content to validate.

  • validationType: The type of validation to perform (optional if a single type is defined).

  • embeddingMethod: The way to consider the content provided for the contentToValidate input (STRING the default, BASE64 or URI).

  • externalSchema: A list of user-provided JSON schemas to be considered with any predefined ones. These are accepted only if explicitly allowed in the configuration for the validation type in question.

  • externalSchemaCombinationApproach: The way to combine externally provided schemas in case multiple are defined (allOf, anyOf, oneOf). Default is allOf.

  • locationAsPointer: Whether or not the location reported for returned errors will be a JSON pointer (default false). False will return the line number in the input.

The content to validate can be provided by any of three means that are determined by the input element’s embeddingMethod attribute. Specifically:

  • STRING: The content is provided as an embedded text within the request.

  • BASE64: The content is provided as a BASE64 encoded string.

  • URI: The content is to be loaded remotely from the provided URI.

An alternative to using this attribute is to use the optional embeddingMethod input. This is provided to address a known issue in the GITB software where not all embedding methods can be leveraged within test cases (see Step 7: Use the validator in GITB TDL test cases).

Regarding the externalSchema input, each item is treated as a map with one property:

  • schema: The schema content to consider.

The sample SOAP UI project includes sample requests per case. As an example, validating via URI would be done using the following envelope:

<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/" xmlns:v1="http://www.gitb.com/vs/v1/" xmlns:v11="http://www.gitb.com/core/v1/">
   <soapenv:Header/>
   <soapenv:Body>
      <v1:ValidateRequest>
         <sessionId>?</sessionId>
         <input name="contentToValidate" embeddingMethod="URI">
            <v11:value>https://www.itb.ec.europa.eu/files/json/sample.json</v11:value>
         </input>
         <input name="validationType" embeddingMethod="STRING">
            <v11:value>large</v11:value>
         </input>
      </v1:ValidateRequest>
   </soapenv:Body>
</soapenv:Envelope>

With the resulting report provided as follows:

<soap:Envelope xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
   <soap:Body>
      <ns4:ValidationResponse xmlns:ns2="http://www.gitb.com/core/v1/" xmlns:ns3="http://www.gitb.com/tr/v1/" xmlns:ns4="http://www.gitb.com/vs/v1/">
         <report>
            <ns3:date>2020-05-19T15:31:11.409+02:00</ns3:date>
            <ns3:result>FAILURE</ns3:result>
            <ns3:counters>
               <ns3:nrOfAssertions>0</ns3:nrOfAssertions>
               <ns3:nrOfErrors>1</ns3:nrOfErrors>
               <ns3:nrOfWarnings>0</ns3:nrOfWarnings>
            </ns3:counters>
            <ns3:context type="map">
               <ns2:item name="contentToValidate" embeddingMethod="STRING" type="string">
                  <ns2:value>{
  "shipTo": {
    "name": "John Doe",
    "street": "Europa Avenue 123",
    "city": "Brussels",
    "zip": 1000
  },
  "billTo": {
    "name": "Jane Doe",
    "street": "Europa Avenue 210",
    "city": "Brussels",
    "zip": 1000
  },
  "orderDate": "2020-01-22",
  "comment": "Send in one package please",
  "items": [
    {
      "partNum": "XYZ-123876",
      "productName": "Mouse",
      "quantity": 20,
      "priceEUR": 15.99,
      "comment": "Confirm this is wireless"
    },
    {
      "partNum": "ABC-32478",
      "productName": "Keyboard",
      "quantity": 15,
      "priceEUR": 25.50
    }
  ]
}</ns2:value>
               </ns2:item>
            </ns3:context>
            <ns3:reports>
               <ns3:error xsi:type="ns3:BAR" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
                  <ns3:description>The array must have at least 10 element(s), but actual number is 2.</ns3:description>
                  <ns3:location>contentToValidate:30:3</ns3:location>
               </ns3:error>
            </ns3:reports>
         </report>
      </ns4:ValidationResponse>
   </soap:Body>
</soap:Envelope>

The returned report uses the GITB TRL syntax and is the same as the XML report you can download from the user interface (see Validation via user interface). It includes:

  • The validation timestamp (in UTC).

  • The overall result (SUCCESS or FAILURE).

  • The count of errors, warnings and information messages.

  • The context for the validation (i.e. the JSON content that was validated).

  • The list of report items displaying per item its description and location in the validated content.

Step 7: Use the validator in GITB TDL test cases

As a next step over the standalone JSON validator you may consider using it from within GITB TDL test cases running in the test bed. You would typically do this for the following reasons:

  • You want to control access to the validation service based on user accounts.

  • You prefer to record all data linked to validations (for e.g. subsequent inspection).

  • You want to build complete conformance testing scenarios that are either focused on the validator or that use it as part of validation steps.

As described in Validation via SOAP web service API, the standalone JSON validator offers by default a SOAP API for machine-to-machine integration that realises the GITB validation service specification. In short this means that the service can be easily included in any GITB TDL test case as the handler of a verify step. This is done by supplying as the handler value the full URL to the service’s WSDL, as illustrated in the following example that requests the user to upload the file to validate:

<?xml version="1.0" encoding="UTF-8"?>
<testcase id="testCase1_upload" xmlns="http://www.gitb.com/tdl/v1/" xmlns:gitb="http://www.gitb.com/core/v1/">
    <metadata>
        <gitb:name>testCase1_upload</gitb:name>
        <gitb:version>1.0</gitb:version>
        <gitb:description>Test case that allows the developer of an EU retailer system to upload a purchase order for validation.</gitb:description>
    </metadata>
    <variables>
        <var name="purchaseOrderToValidate" type="binary"/>
    </variables>
    <actors>
        <gitb:actor id="Retailer" name="Retailer" role="SUT"/>
    </actors>
    <steps>
        <interact desc="Upload content">
            <request desc="Purchase order to validate:">$purchaseOrderToValidate</request>
        </interact>
        <verify handler="https://www.itb.ec.europa.eu/json/soap/order/validation?wsdl" desc="Validate purchase order">
            <input name="contentToValidate">$purchaseOrderToValidate</input>
            <input name="validationType">"basic"</input>
        </verify>
    </steps>
</testcase>

Notice in this example how the contentToValidate and validationType inputs are provided as input elements to the verify step. We included the validationType input because we define two validation types (basic and large) but this could be omitted if only a single validation type is supported. In addition, note that although you can define the service’s WSDL URL directly, a better approach to improve portability is to define this in the test bed’s domain configuration as a domain parameter. Defining a validationService parameter in the domain you could thus redefine the verify step as:

...
<verify handler="$DOMAIN{validationService}" desc="Validate purchase order">
    ...
</verify>
...

Summary

Congratulations! You have just setup a validation service for your JSON specification. In doing so you considered your needs and defined your service through configuration on the ISA² test bed or as a Docker container. In addition, you used this service via its different APIs and considered how this could be used as part of complete conformance testing scenarios.

See also

In Step 7: Use the validator in GITB TDL test cases we briefly touched upon using the test bed for complete conformance testing scenarios. If this interests you, several additional guides are available that can provide you with further information:

For the full information on GITB TDL test cases check out the GITB TDL documentation, the reference for all test step constructs as well as a source of numerous complete examples.

In case you need to consider validation of further content types, be aware that the test bed provides similar support for:

Finally, for more information on Docker and the commands used in this guide, check out the Docker online documentation.

References

This section contains additional references linked to this guide.

Validator configuration properties

The following sections list the configuration properties you can use to customise the operation of your validation service.

Domain-level configuration

The properties in this section are to be provided in the configuration property file (one per configured validation domain) you define as part of your validator configuration.

Property

Description

Type

Default value

validator.channels

Comma-separated list of validation channels to have enabled. Possible values are (form, soap_api).

Comma-separated Strings

form, soap_api

validator.type

Comma-separated list of supported validation types. Values need to be reflected in properties validator.typeLabel, validator.schemaFile.

Comma-separated Strings

validator.typeLabel.XYZ

Label to display in the web form for a given validation type (added as a postfix of validator.typeLabel). Only displayed if there are multiple types.

String

validator.webServiceId

The ID of the web service.

String

ValidatorService

validator.webServiceDescription.contentToValidate

The description of the SOAP web service for element “contentToValidate”.

String

The content to validate, provided as a string, BASE64 or a URI.

validator.webServiceDescription.validationType

The description of the SOAP web service for element “validationType”. Only displayed if there are multiple types.

String

The type of validation to perform (if multiple types are supported).

validator.webServiceDescription.embeddingMethod

The description of the SOAP web service for element “embeddingMethod”.

String

The embedding method to consider for the 'contentToValidate' input ('BASE64', 'URL' or 'STRING').

validator.webServiceDescription.externalSchemas

The description of the SOAP web service for element “externalSchemas”.

String

A list of maps that defines external schemas to consider in addition to any preconfigured ones. Each map item corresponds to a schema file and defines the following keys: 'content' (the schema content to consider, see 'contentToValidate' for its semantics), 'embeddingMethod' (the way to consider the 'content' value).

validator.webServiceDescription.externalSchemaCombinationApproach

The description of the SOAP web service for element “externalSchemaCombinationApproach”.

String

The way to combine externally provided schemas in case multiple are defined ('allOf', 'anyOf', 'oneOf'). Default is 'allOf'.

validator.webServiceDescription.locationAsPointer

The description of the SOAP web service for element “locationAsPointer”.

String

Whether or not the location reported for returned errors will be a JSON pointer (default false). False will return the line number in the input.

validator.schemaFile.XYZ

Comma-separated list of schema files loaded for a given validation type (added as a postfix). These can be a files or folders.

Comma-separated Strings

validator.schemaFile.XYZ.remote.N.url

Reference for a remotely loaded schema file for a given validation type (added as the XYZ placeholder). One or more such entries can be defined by incrementing the zero-based N counter.

String

validator.schemaFile.XYZ.combinationApproach

The approach to follow when multiple schemas are defined for the validation type. Possible values are (allOf, anyOf, oneOf).

String

allOf

validator.externalSchemas.XYZ

Whether or not user-provided schemas are allowed for the given validation type (added as a postfix). Possible values are (required, optional, none).

String

required

validator.externalSchemaCombinationApproach.XYZ

The approach to follow when multiple external schemas are provided. Possible values are (allOf, anyOf, oneOf).

String

allOf

validator.showAbout

Whether or not to show the about panel on the web UI.

Boolean

true

validator.supportMinimalUserInterface

Enable a minimal user interface useful for embedding in other UIs or portals (applies only if the form validation channel is enabled).

Boolean

false

validator.bannerHtml

Configurable HTML banner replacing the text title.

String

validator.footerHtml

Configurable HTML banner for the footer.

String

Application-level configuration

These properties govern the validator’s application instance itself. They apply only when you are defining your own validator as a Docker image in which case they are supplied as environment variables (ENV directives in a Dockerfile). Note that apart from these properties any Spring Boot configuration property can also be supplied.

Note

Mandatory property: The only mandatory property that needs to be defined is validator.resourceRoot.

Property

Description

Type

Default value

validator.resourceRoot

The root folder under which domain subfolders will be loaded from.

String

validator.domain

The names of the domain subfolders to consider. By default all folders under validator.resourceRoot will be considered.

Comma-separated Strings

validator.domainName.XYZ

The name to display for a given domain folder. This value will also be used in request paths.

String

The folder name is used.

logging.path

Path to a folder that will hold the validator’s log output.

String

/validator/logs

validator.tmpFolder

Path to a folder that contains temporary data and reports.

String

/validator/tmp

validator.acceptedSchemaExtensions

Accepted local schema file extensions. All other files found in validator.schemaFile.XYZ (when folders are defined) are ignored.

Comma-separated Strings

json

validator.cleanupRate

The rate at which the external file cache is refreshed (in milliseconds).

Long

3600000

validator.cleanupWebRate

The rate at which temporary files linked to the web form are cleaned (in milliseconds).

Long

600000

validator.minimumCachedInputFileAge

The minimum time for which input provided through the web interface is cached (in milliseconds).

Long

600000

validator.minimumCachedReportFileAge

The minimum time for which reports generated through use of the web interface are cached (in milliseconds).

Long

600000

validator.acceptedMimeTypes

The accepted mime types for user-provided content through the user interface.

Comma-separated Strings

application/json, text/plain