Configure mapping

The mapping of an integration's data source defines the ingested data and its destination. It allows you to specify:

Which data you wish to ingest from the integrated tool.
Which properties in the integration's blueprints will be filled with the ingested data.

How does mapping work?

Integration mapping is configured in the data sources page of your portal, under Exporters.
Each integration has its own mapping, written in YAML.

To understand how mapping works, let's take a look at an example. After you complete the onboarding and connect your Git provider to Port, you will see an exporter entry in your data sources page:

Clicking on this entry will open the mapping configuration. In the bottom left panel, you will see the YAML configuration of the mapping.
Note that Port provides default mapping, providing values to the properties defined in the relevant blueprint:

Configuration structure

This section will explain each part of the configuration YAML.
Some of the keys use JQ queries to filter the data ingested from the tool's API.

The resources key is the root of the YAML configuration:
```
resources:
  - kind: repository
    ...
```

The kind key is a specifier for the object you wish to map from the tool's API (in this example, a Github repository).
To see which kinds are available for mapping, refer to the integration's documentation. In this example, the available kinds are listed in the Github integration page.
```
  resources:
    - kind: repository
      selector:
      ...
```

The selector and the query keys let you filter exactly which objects of the specified kind will be ingested into Port:
```
resources:
  - kind: repository
    selector:
      query: "true" # JQ boolean query. If evaluated to false - skip syncing the object.
    port:
```
Using a JQ query, you can define your desired conditions. For example, to ingest only repositories that have a name starting with "service", use the query key like this:
```
query: .name | startswith("service")
```

The port.entity.mappings key contains the section used to map the object fields to Port entities.
Here you can specify the blueprint in Port to which the data should be mapped, and which API object will be ingested to each of its properties.

resources:
  - kind: repository
    selector:
      query: "true"
    port:
      entity:
        mappings: # Mappings between one GitHub API object to a Port entity. Each value is a JQ query.
          identifier: ".name"
          title: ".name"
          blueprint: '"service"'
          properties:
            description: ".description"
            url: ".html_url"
            defaultBranch: ".default_branch"

To create multiple mappings of the same kind, you can add another item to the resources array:

resources:
  - kind: repository
    selector:
      query: "true"
    port:
      entity:
        mappings: # Mappings between one GitHub API object to a Port entity. Each value is a JQ query.
          identifier: ".name"
          title: ".name"
          blueprint: '"service"'
          properties:
            description: ".description"
            url: ".html_url"
            defaultBranch: ".default_branch"
  - kind: repository # In this instance repository is mapped again with a different filter
    selector:
      query: '.name == "MyRepositoryName"'
    port:
      entity:
        mappings: ...

Additional options

Several more advanced options are available in the mapping configuration:

createMissingRelatedEntities - used to enable the creation of missing related entities in Port. This is useful when you want to create an entity and its related entities in one call, or if you want to create an entity whose related entity does not exist yet.
deleteDependentEntities - used to enable deletion of dependent Port entities. This is useful when you have two blueprints with a required relation, and the target entity in the relation should be deleted. In this scenario, the delete operation will fail if this parameter is set to false. If set to true, the source entity will be deleted as well.

To use these options, add them to the root of the mapping configuration:

createMissingRelatedEntities: true
deleteDependentEntities: true
resources:
  - kind: repository
    ...

Test your mapping - JQ playground

The mapping configuration window contains a JQ playground that allows you to test your JQ queries against example responses from the API of the integrated tool. This is useful for validating your queries and ensuring they return the expected results.

For integrations based on the Ocean framework, examples will be automatically generated for each resource kind in your mapping, based on real data ingested from the tool. You can disable this behavior by setting the sendRawDataExamples flag to false in the integration's configuration.

To test your mapping against the example data, click on the Test mapping button in the bottom-right panel.

Manually add test examples

For each resource kind in your mapping (in the bottom-left panel), you can add an example in the Test examples section.
Click on the Add kind button to add an example:

After adding your example, click on the Test mapping button in the bottom-right panel to test your mapping against the example data.

Port's JQ playground

In addition to the aforementioned JQ playground, Port provides a general JQ playground where you can test any JSON snippet against JQ expressions with real-time filters and AI-powered assistance.

Edit an integration's mapping

Once you have configured an integration's mapping to your liking, click the Resync button in the bottom-right to save your changes.

To edit an integration's mapping using Port's API, you can use the Patch integration route.

Resync via the API

To perform a simple resync of an integration via the API without changing its mapping, use the same Patch integration route with the integration identifier and an empty body.

Mapping relations

You can use the mapping YAML file to set the value of a relation between entities. This is very useful when you want to automatically assign an entity to the relation of another entity using a convention of your choice.

For example, say we have a service blueprint and a PagerDuty Service blueprint with a relation between them:

After ingesting all of our services and PagerDuty services, we want to connect each service to its corresponding PagerDuty service. To achieve this, we have two options:

Option 1 - use the integration's mapping YAML. In our example, we can add an entry to the mapping of the PagerDuty integration:
- Go to your data sources page and click on the PagerDuty exporter:
- Add the following entry to the mapping YAML:
```
- kind: services
  selector:
    query: "true"
  port:
    entity:
      mappings:
        identifier: .name
        blueprint: '"service"'
        properties: {}
        relations:
          pagerduty_service: .id
```
  Now, if a service's identifier is equal to a PagerDuty service's name, that service will automatically have its on-call property filled with the relevant PagerDuty service.
  This is just the convention we chose for this example, but you can use a different one if you'd like.
Option 2 - manually assign a PagerDuty service to each service using the UI:
- Go to the Services page of your software catalog.
- Choose a service you want to assign a PagerDuty service to. Hover over it, click on the ... button on the right, and select Edit.
- In the PagerDuty service field, select the relevant PagerDuty service from the dropdown list, then click Update:

Mapping relations using search queries

In the example above we map a relation using a direct reference to the related entity's identifier.

Port also allows you to use a search query rule to map relations based on a property of the related entity.
This is useful in cases where you don't have the identifier of the related entity, but you do have one of its properties.

For example, consider the following scenario:
Say we have a service blueprint that has a relation (named service_owner) to a user blueprint. The user blueprint has a property named github_username.

Now, we want to map the service_owner relation based on the github_username property of the user entity.
To achieve this, we can use the following mapping configuration:

- kind: repository
  selector:
    query: 'true'
  port:
    entity:
      mappings:
        identifier: .name
        title: .name
        blueprint: '"service"'
        relations:
          service_owner:
            combinator: '"and"'
            rules:
              - property: '"github_username"'
                operator: '"="'
                value: .owner.login

Instead of directly referencing the user entity's identifier, we use a search query rule to find the user entity whose github_username property matches the .owner.login value returned from GitHub's API.

When using a search query rule to map a relation, Port will query all entities of the related blueprint (in this case - user) and return the one/s that match the rule.

Limitations

One or more entities can be returned by the search query rule. Note the relation's type when using this method:
- A "single type" relation expects a single entity to be returned.
- A "many type" relation expects an array of entities to be returned.
The maximum number of entities returned by the search query rule is 100.
Mirror and calculation properties are currently not supported.
Only the = and in operators are supported for the search query rule.

Map by property

In some cases, we may not know the identifier of the entity we want to map to. If that entity has a property that we do know, we can use it to map the data.
This is especially useful when patching entities whose identifiers are not known in advance. Take the following example:

Say we installed Port's PagerDuty integration, and we want to connect each service (Git repository) to the relevant PagerDuty service.
We can create a property in our service blueprint named pagerduty_service_id, containing the identifier of the relevant PagerDuty service.
Then, in the PagerDuty integration mapping, we can use this property to map each PagerDuty service to the relevant service.
This way, we would not need to have a separate blueprint for PagerDuty services, since the integration maps directly to the service blueprint.

Mapping by property is done using a search query rule in the following format:

resources:
  - kind: services
    selector:
      query: 'true'
    port:
      entity:
        mappings:
          identifier:
            combinator: '"and"'
            rules:
              - operator: '"="'
                property: '"pagerduty_service_id"'
                value: .id
          blueprint: '"service"'
          properties:
            oncall: .__oncall_user | sort_by(.escalation_level) | .[0].user.email
            ... # Any other properties you want to map

In the example above, we search for a service entity whose pagerduty_service_id property is equal to the id of the PagerDuty service, and map data from the PagerDuty service to it.

API usage

Searching by property can also be used when using Port's API to create an entity.

Limitations

The search query must return exactly one entity (else the entire request will fail).
If the search query returns no entities, a new entity will not be created.
The query will be executed on the same blueprint from the request’s url.
Only the = and in operators is supported for the search query rule.
Calculation and mirror properties are not supported.

Create multiple entities from an array API object

In some cases, an application's API returns an array of objects that you want to map to multiple entities in Port.
To achieve this, Port provides you with the itemsToParse key, its value should be a JQ query that returns an array.
In order to reference an array item attribute, use .item in your JQ expression.

Here is an example mapping configuration of a Jira issue, where we want to map each of the issue's comments to a separate comment entity:

- kind: issue
  selector:
    query: .item.name != 'test-item' and .issueType == 'Bug' 
  port:
    itemsToParse: .fields.comments
    entity:
      mappings:
        identifier: .item.id
        blueprint: '"comment"'
        properties:
          text: .item.text
        relations:
            issue: .key

The object returned from Jira for which we would apply this mapping might look like this (note the comments array):

Example Jira API response (click to expand)

{
  "url": "https://example.com/issue/1",
  "status": "Open",
  "issueType": "Bug",
  "comments": [
    {
      "id": "123",
      "text": "This issue is not reproducing"
    },
    {
      "id": "456",
      "text": "Great issue!"
    }
  ],
  "assignee": "user1",
  "reporter": "user2",
  "creator": "user3",
  "priority": "High",
  "created": "2024-03-18T10:00:00Z",
  "updated": "2024-03-18T12:30:00Z",
  "key": "ISSUE-1"
}

Common use-cases

Splitting a `kind` block

Sometimes the CreateMissingRelatedEntities flag is passed as false to prevent generation of additional entities for relations. This can lead to cases where entity ingestion will not happen because the target entity for a relation does not exist in your catalog.
To handle such cases, you can split a single kind to multiple mappings like this:

createMissingRelatedEntities: false
  - kind: services
    selector:
      query: "true"
    port:
      entity:
        mappings:
          identifier: .name
          blueprint: '"service"'
          properties: 
            #Properties mapping
  - kind: services
    selector:
      query: "true"
    port:
      entity:
        mappings:
          identifier: .name
          blueprint: '"service"'
          properties: {}
          relations:
            pagerduty_service: .id

Looking at this mapping configuration we see the following:

The first kind block is used to create the entity along with all of its properties.
The second kind block is used to update the same entity (notice the mapping for the identifier is the same in both configurations) with relations. If the target entity of the relation does not exist (i.e. you have no matching PagerDuty service), the update itself will fail, but the data ingestion will still complete successfully.

This case can also be expanded for handling multiple relation, for each relation that might not be established, you can split it into another kind mapping.

How does mapping work?​

Configuration structure​

Additional options​

Test your mapping - JQ playground​

Manually add test examples​

Edit an integration's mapping​

Mapping relations​

Mapping relations using search queries​

Limitations​

Map by property​

Limitations​

Create multiple entities from an array API object​

Common use-cases​

Splitting a kind block​

How does mapping work?

Configuration structure

Additional options

Test your mapping - JQ playground

Manually add test examples

Edit an integration's mapping

Mapping relations

Mapping relations using search queries

Limitations

Map by property

Limitations

Create multiple entities from an array API object

Common use-cases

Splitting a `kind` block