Define package via template

How to read this guide

Each section of this guide provides 2 tabs, which are linked throughout (once you swap in one section, all other sections will automatically reflect that same level of detail):

• Simple — when you are just starting out, follow these tabs to understand the basic structure of the toolkit and the fundamental elements that you must use.
• Detailed — as you start to wonder about additional complexity, consider changing to these tabs, which cover additional (optional) possibilities.

Start by creating a Pkl file that amends our published package config toolkit model:

amends "package://developer.atlan.com/toolkits/custom-package/config@5.0.3#/Framework.pkl"

If this is the first time you're creating a package, hover over that line and download the Pkl package.

Define overall metadata

Then you can start defining your package. All packages should have at least these main components:

amends "package://developer.atlan.com/toolkits/custom-package/config@5.0.3#/Framework.pkl"

import "pkl:semver" // (1)
import "package://developer.atlan.com/toolkits/custom-package/config@5.0.3#/Connectors.pkl" // (2)

packageId = "@csa/openapi-spec-loader" // (3)
packageName = "OpenAPI Spec Loader" // (4)
version = semver.Version("1.0.0") // (5)
description = "Loads API specs and paths from an OpenAPI (v3) definition." // (6)
iconUrl = "http://assets.atlan.com/assets/apispec.png" // (7)
docsUrl = "https://developer.atlan.com/samples/loaders/openapi/" // (8)
implementationLanguage = "Kotlin" // (9)
containerImage = "ghcr.io/atlanhq/\(name):\(version)" // (10)
containerCommand { // (11)
  "/dumb-init"
  "--"
  "java"
  "OpenAPISpecLoaderKt"
}
outputs { // (12)
  files {
    ["debug-logs"] = "/tmp/debug.log"
  }
}
keywords { // (13)
  "kotlin"
  "crawler"
  "openapi"
}
category = "custom" // (14)
preview = true // (15)
certified = false // (16)
connectorType = Connectors.API // (17)

uiConfig { // (18)
  ...
}

credentialConfig { // (19)
  ...
}

1. Import the standard Pkl library for semantic versioning, to define your package's version.
2. (Optional) If your package could create a connection, import the pre-defined set of connections from the package toolkit.
3. A packageId, which gives a unique namespaced identity for your package. It must be of the form @namespace/kebab-case-name.
4. A packageName, which gives the human-readable name of your package, as it should show in the UI.
5. A version for your package, which must follow semantic versioning.
6. A description for your package, as it should show in the UI.
7. An iconUrl giving an online location where the icon to use in the UI for this package can be accessed.
8. A docsUrl giving an online location where additional documentation about the package can be found. This will be linked in the UI.
9. The implementationLanguage defining the coding language you will use to implement the logic of your package. This is used by the toolkit to generate a strongly-typed data class that captures the inputs a user provides during setup of the package.

10. The containerImage that will encapsulate your package's code.

    Use variables where possible

    Note that Pkl is itself a language, so you can use variables here as well (for example, if your container images follow a naming convention that can be generated from the packageId or to automatically set the version).

11. The full containerCommand used to run your custom logic within the container. This should be specified as a list, rather than a single spaced-out string.

12. (Optional) Any outputs your package's logic will produce, that you want a user to be able to download. In this example, our package will write debug logs out to /tmp/debug.log and we will make these available for download.
13. (Optional) Any keywords you want your package to have associated with it.
14. (Optional) The category controls the name of the pill in the marketplace in the UI where the package will be listed. It will default to custom if not specified.
15. (Optional) Setting preview to true will show a small orange badge on the package indicating it is experimental (by default it will not be experimental).
16. (Optional) Setting certified to false will remove the green tick badge from the package indicating it is certified (by default it will be certified).
17. (Optional) When your package could create a connection, use the connectorType to define what kind of connection it should create.
18. A uiConfig section where you will define the inputs you want your package to receive from the user during setup.
19. (Optional) A credentialConfig section where you can define any sensitive inputs you want the user to provide, typically to connect to an external system. Any details provided by a user within these inputs will be encrypted and stored in a secure vault in Atlan.

Define configurable inputs

Use the uiConfig section to define the configurable inputs you want the user to provide when setting up your package.

These inputs fit into a two-level hierarchy:

• Tasks are the top-level steps you see along the left side when setting up a new package.
• Inputs are the individual widgets a user can enter information into, select from, etc within each task (step).

In addition to these inputs, you can (optionally) define one or more rules to control which inputs appear on the screen, based on what a user has selected in some other input.

Simple Detailed

uiConfig {
  tasks { // (1)
    ["Configuration"] { // (2)
      description = "OpenAPI spec configuration" // (3)
      inputs { // (4)
        ...
      }
    }
    ["Connection"] { // (5)
      description = "Connection details"
      inputs {
        ...
      }
    }
  }
  rules { // (6)
    ...
  }
}

1. You must define one (and only one) list of tasks.
2. Use the name as you want it to show in the UI for each top-level step, surrounded in square brackets. You can then enclose the definition of that top-level step in curly braces.
3. Include a description, which will show in the UI as part of the top-level step (very space-limited).
4. Also include the list of inputs that should be configured as part of that top-level step.
5. You can define as many of these top-level steps as you like. They will be displayed in the order (top-down) they are listed here in the model.
6. (Optional) You can then specify any rules to control which inputs appear on the screen (or not), based on what a user has selected in some other input.

uiConfig {
  tasks { // (1)
    ["Configuration"] { // (2)
      description = "OpenAPI spec configuration" // (3)
      inputs { // (4)
        ["spec_url"] = new TextInput { // (5)
          title = "Specification URL" // (6)
          required = true // (7)
          helpText = "Full URL to the JSON form of the OpenAPI specification." // (8)
          placeholderText = "https://petstore3.swagger.io/api/v3/openapi.json" // (9)
        }
      }
    }
    ["Connection"] { // (10)
      description = "Connection details"
      inputs {
        ["connection_usage"] = new Radio {
          title = "Connection"
          required = true
          possibleValues {
            ["CREATE"] = "Create"
            ["REUSE"] = "Reuse"
          }
          default = "REUSE"
          fallback = default // (11)
          helpText = "Whether to create a new connection to hold these API assets, or reuse an existing connection."
        }
        ["connection"] = new ConnectionCreator {
          title = "Connection"
          required = true
          helpText = "Enter details for a new connection to be created."
        }
        ["connection_qualified_name"] = new ConnectionSelector {
          title = "Connection"
          required = true
          helpText = "Select an existing connection to load assets into."
        }
      }
    }
  }
  rules { // (12)
    new UIRule { // (13)
      whenInputs { ["connection_usage"] = "REUSE" } // (14)
      required { "connection_qualified_name" } // (15)
    }
    new UIRule { // (16)
      whenInputs {
        ["connection_usage"] = "CREATE"
      }
      required {
        "connection"
      }
    }
  }
}

1. You must define one (and only one) list of tasks.
2. Use the name as you want it to show in the UI for each top-level step, surrounded in square brackets. You can then enclose the definition of that top-level step in curly braces.
3. Include a description, which will show in the UI as part of the top-level step (very space-limited).
4. Also include the list of inputs that should be configured as part of that top-level step.
5. For each input, specify a unique variable name (in lower_snake_case) surrounded by square brackets. You can choose from a variety of widgets to use for receiving that input from a user.
6. All inputs will at least need a title, which defines the text to show in the UI for that widget.
7. All inputs can also be configured as mandatory using required = true (by default they'll be optional).
8. All inputs can also be given a helpText, which when provided shows an information icon next to the title that can be hovered over for more information on how the input should be used.
9. Other configurable options will vary by the kind of input, but could include for example some light grey text to show in the box by default to give the user some idea of the value you're expecting from them.
10. You can define as many of these top-level steps as you like, and within each as many inputs as you like. They will be displayed in the order (top-down) they are listed here in the model.
11. (Optional) You can specify a fallback value to use for any input, in case the user does not make any selection or enter any value in the UI.
12. (Optional) You can then specify any rules to control which inputs appear on the screen (or not), based on what a user has selected in some other input.
13. Each new rule needs to be defined as new UIRule, with its definition enclosed in curly braces.
14. You must specify a whenInputs that defines an input variable name and value.
15. When the input variable has the value indicated above, all input variables defined in this required will be shown in the UI (otherwise they will be hidden).
16. When you have multiple conditions or inputs to show or hide, you can use this more spaced out form. This example means that when a user has selected CREATE for the connection_usage input, the UI will show the connection input.

Define sensitive inputs

Use the credentialConfig section to define sensitive inputs you want the user to provide when setting up your package.

Encrypted and secured in a vault

Any details a user provides within this section will be encrypted and secured in a vault within Atlan, for added protection.

These inputs fit into two types:

• Common inputs are inputs that are common irrespective of the type of credential the user is configuring.
• Options define the different types of credentials a user can configure (for example, basic username and password vs shared secrets).

Simple Detailed

credentialConfig {
  name = "csa-connector-custom" // (1)
  source = Connectors.API // (2)
  icon = "https://assets.atlan.com/assets/apispec.svg" // (3)
  helpdesk = "https://ask.atlan.com/hc/en-us/articles/1234567890" // (4)
  logo = "https://assets.atlan.com/assets/apispec.svg" // (5)
  commonInputs { // (6)
    ...
  }
  options { // (7)
    ...
  }
}

1. You must give the credentials configuration a unique name. (This allows you to reuse credential configurations without needing to redefine them in different packages.)
2. Set the source to the type of connector through which these credentials can be used.
3. Define the icon to use when representing this particular source and credential in the UI.
4. Provide a link to any documentation about how to set up these credentials, for example any pre-requites on the permissions that are required on the external system.
5. Define the logo to use when representing this particular source and credential in the UI.
6. (Optional) You can define any common inputs that are shared across different connectivity options.
7. You must define at least one option for connecting to the external system, and could define many. For example, you may want to support both a username and password as one option, and another option that uses some shared secrets or API tokens.

credentialConfig {
  name = "csa-connector-custom" // (1)
  source = Connectors.API // (2)
  icon = "https://assets.atlan.com/assets/apispec.svg" // (3)
  helpdesk = "https://ask.atlan.com/hc/en-us/articles/1234567890" // (4)
  logo = "https://assets.atlan.com/assets/apispec.svg" // (5)
  commonInputs { // (6)
    ["host"] = new TextInput { // (7)
      title = "Hostname" // (8)
      placeholderText = "petstore3.swagger.io" // (9)
      prepend = "https://"
      width = 6
    }
    ["port"] = new NumericInput {
      title = "Port"
      default = 443
      width = 2
    }
  }
  options { // (10)
    ["basic"] { // (11)
      title = "Basic" // (12)
      inputs { // (13)
        ["username"] = new TextInput { // (14)
          title = "Username"
          required = true
          defaultValue = "jane.smith"
          width = 4
        }
        ["password"] = new PasswordInput {
          title = "Password"
          required = true
          width = 4
        }
        ["extra"] = new NestedInput { // (15)
          title = "Role" // (16)
          inputs { // (17)
            ["role"] = new DropDown {
              title = "Role"
              possibleValues {
                ["USER"] = "user"
                ["ADMIN"] = "admin"
              }
              width = 4
            }
          }
        }
      }
    }
    ["keypair"] {
      title = "Keypair"
      inputs {
        ["username"] = new TextInput {
          title = "Username"
          placeholderText = "jsmith"
          width = 4
        }
        ["password"] = new TextBoxInput {
          title = "Encrypted Private Key"
          placeholderText = "-----BEGIN ENCRYPTED PRIVATE KEY-----MIIE6TAbBgkqhkiG9w0BBQMwDgQILYPyCppzOwECAggABIIEyLiGSpeeGSe3xHP1wHLjfCYycUPennlX2bd8    yX8xOxGSGfvB+99+PmSlex0FmY9ov1J8H1H9Y3lMWXbL...-----END ENCRYPTED PRIVATE KEY-----"
          width = 4
        }
        ["extra"] = new NestedInput {
          title = "Private Key Password"
          inputs {
            ["private_key_password"] = new PasswordInput {
              title = "Private Key Password"
              width = 5
            }
          }
        }
      }
    }
  }
}

1. You must give the credentials configuration a unique name. (This allows you to reuse credential configurations without needing to redefine them in different packages.)
2. Set the source to the type of connector through which these credentials can be used.
3. Define the icon to use when representing this particular source and credential in the UI.
4. Provide a link to any documentation about how to set up these credentials, for example any pre-requites on the permissions that are required on the external system.
5. Define the logo to use when representing this particular source and credential in the UI.
6. (Optional) You can define any common inputs that are shared across different connectivity options.
7. For each input, specify a unique variable name (in lower_snake_case) surrounded by square brackets. You can choose from a variety of widgets to use for receiving that input from a user.
8. All inputs will need at least a title, which defines the text to show in the UI for that widget.
9. Other configurable options will vary by the kind of input, but could include for example some light grey text to show in the box by default to give the user some idea of the value you're expecting from them.
10. You must define at least one option for connecting to the external system, and could define many. For example, you may want to support both a username and password as one option, and another option that uses some shared secrets or API tokens.

Delegate publishing

You can now use the toolkit to delegate publishing of assets through packages that use CSV inputs. Your own logic simply needs to produce a CSV output file that matches the required format of one of these publishing package's CSV inputs.

Handles many common scenarios automatically

These publishing packages are designed to handle common scenarios for you directly, such as:

• Calculating differences (deltas) between a previous load and the current load, to automatically:
  • Identify which assets should be deleted (or archived), and do so
  • Optimize the publishing to only save any assets that have changed
• Support only updating assets (avoiding any accidental asset creation), if desired
• Allow case-insensitive matching of qualifiedNames of assets
• Resolve any ambiguity between tables or views for you, automatically

To add a delegated publish step to your package, you need only add a publishConfig to your template:

Detailed

publishConfig = new AssetImport {  // (1)
  versionTag = "5.0.1"  // (2)
  assetsFile = transferFile(  // (3)
    outputs,
    "transformed_file",
    "current.csv"
  )
  assetsUpsertSemantic = transferConfigInput(  // (4)
    uiConfig,
    "assets_upsert_semantic"
  )
  assetsPreviousFilePrefix = "csa-relational-assets-builder" // (5)
  assetsDeltaSemantic = transferConfigInput(uiConfig, "delta_semantic")
  assetsDeltaRemovalType = transferConfigInput(uiConfig, "delta_removal_type")
  assetsDeltaReloadCalculation = transferConfigInput(uiConfig, "delta_reload_calculation")
  assetsFailOnErrors = transferConfigInput(uiConfig, "assets_fail_on_errors")
  assetsFieldSeparator = transferConfigInput(uiConfig, "assets_field_separator")
  assetsBatchSize = transferConfigInput(uiConfig, "assets_batch_size")
  assetsCmHandling = transferConfigInput(uiConfig, "assets_cm_handling")
  assetsTagHandling = transferConfigInput(uiConfig, "assets_tag_handling")
  assetsLinkIdempotency = transferConfigInput(uiConfig, "assets_link_idempotency")
  trackBatches = transferConfigInput(uiConfig, "track_batches")
}

1. When configuring which publish package to delegate, you can choose from either AssetImport, RelationalAssetsBuilder or LineageBuilder. Each can be configured with different options, which are described further in the sections below.
2. You need to provide the specific version of the package you want to use for publishing. This version will match the version published at atlanhq/atlan-java.

3. For any file-based inputs, use the toolkit's transferFile() helper to indicate how to transfer the file from the outputs your own package produces:

   • outputs is the name of the variable in your package defining its outputs (will always be this)
   • "transformed_file" is the key of the entry in that outputs variable that points to the file your package produces that you want to delegate for publishing
   • "current.csv" is the name of the file you want to use when you pass it along to the publishing package

4. For any configuration-based inputs, use the toolkit's transferConfigInput() helper to indicate which UI-based input variable's value to transfer to the publishing package.

   • uiConfig is the name of the variable in your package defining its UI-based inputs (will always be this)
   • "assets_upsert_semantic" is the name of the UI input (widget variable) from which to take the value to be transferred

5. You can also use a literal primitive value for any of the options, if you just want to hard-code it rather than transfer anything across from the UI-based configuration or your own package's outputs.

AssetImport

The AssetImport publish config accepts the following configurable inputs. You can use any combination of these, and any you leave out will use sensible defaults.

Glossaries, categories, terms

Variable Name	Data Type	Description
glossariesFile	FrameworkRenderer.NamePathPair	File containing glossaries, categories and terms to import; typically passed through using transferFile().
glossariesUpsertSemantic	ImportSemantic	Whether to allow the creation of new glossaries, categories and terms from the input CSV, or ensure these are only updated if they already exist in Atlan.
glossariesConfig	ConfigType	Options to optimize how glossaries, categories and terms are imported.
glossariesAttrToOverwrite	Listing<String>	List of attributes you want to clear (remove) from glossaries, categories and terms if their value is blank in the provided file.
glossariesFailOnErrors	Boolean	Whether an invalid value in a field should cause the import to fail (true) or log a warning, skip that value, and proceed (false).
glossariesFieldSeparator	String	Single character used to separate fields in the input file (for example, , or ;).
glossariesCmHandling	CustomMetadataSemantic	How custom metadata in the input should be handled: ignore it, merge it with any existing asset custom metadata, or overwrite the existing asset custom metadata.
glossariesTagHandling	AtlanTagSemantic	How Atlan tags on assets in the input should be handled: ignore them, append them to any existing asset tags, replace the existing asset tags, or remove them from the assets.
glossariesLinkIdempotency	LinkIdempotencyInvariant	How linked resources on glossaries in the input should be updated: based on their unique URL or their unique name.
glossariesBatchSize	Int	Maximum number of rows to process at a time (per API request).

Data domains and products

Variable Name	Data Type	Description
dataProductsFile	FrameworkRenderer.NamePathPair	File containing data domains and data products to import; typically passed through using [transferFile()].
dataProductsUpsertSemantic	ImportSemantic	Whether to allow the creation of new domains and data products from the input CSV, or ensure these are only updated if they already exist in Atlan.
dataProductsConfig	ConfigType	Options to optimize how domains and data products are imported.
dataProductsAttrToOverwrite	Listing<String>	Select attributes you want to clear (remove) from domains and data products if their value is blank in the provided file.
dataProductsFailOnErrors	Boolean	Whether an invalid value in a field should cause the import to fail (true) or log a warning, skip that value, and proceed (false).
dataProductsFieldSeparator	String	Single character used to separate fields in the input file (for example, , or ;).
dataProductsCmHandling	CustomMetadataSemantic	How custom metadata in the input should be handled: ignore it, merge it with any existing asset custom metadata, or overwrite the existing asset custom metadata.
dataProductsTagHandling	AtlanTagSemantic	How Atlan tags on assets in the input should be handled: ignore them, append them to any existing asset tags, replace the existing asset tags, or remove them from the assets.
dataProductsLinkIdempotency	LinkIdempotencyInvariant	How linked resources on assets in the input should be updated: based on their unique URL or their unique name.
dataProductsBatchSize	Int	Maximum number of rows to process at a time (per API request).

Tag (structural) definitions

Variable Name	Data Type	Description
tagsFile	FrameworkRenderer.NamePathPair	File containing tag definitions to manage; typically passed through using transferFile().
tagsConfig	ConfigType	Options to optimize how tag definitions are imported.
tagsFailOnErrors	Boolean	Whether an invalid value in a field should cause the import to fail (true) or log a warning, skip that value, and proceed (false).
tagsFieldSeparator	String	Single character used to separate fields in the input file (for example, , or ;).
tagsBatchSize	Int	Maximum number of rows to process at a time (per API request).

All other assets

Variable Name	Data Type	Description
assetsFile	FrameworkRenderer.NamePathPair	File containing assets to import, typically passed through using transferFile().
assetsUpsertSemantic	ImportSemantic	Whether to allow the creation of new assets from the input CSV (full or partial assets), or ensure assets are only updated if they already exist in Atlan.
assetsDeltaSemantic	DeltaSemantic	Whether to treat the input file as an initial load, full replacement (deleting any existing assets not in the file) or only incremental (no deletion of existing assets).
assetsDeltaRemovalType	RemovalType	How to delete any assets not found in the latest file.
assetsDeltaReloadCalculation	ReloadCalculation	Which assets to reload from the latest input CSV file. Changed assets only will calculate which assets have changed between the files and only attempt to reload those changes.
assetsPreviousFileDirect	FrameworkRenderer.NamePathPair	Path to a direct file (locally) to use for delta processing. Note: providing a value for this will ignore any other previously-processed file in the object store, so please be sure this is the option you want to use (should be rare).
assetsPreviousFilePrefix	String	Object store prefix in which previous files exist for delta processing.
assetsConfig	ConfigType	Options to optimize how assets are imported.
assetsAttrToOverwrite	Listing<String>	List of attributes you want to clear (remove) from assets if their value is blank in the provided file.
assetsFailOnErrors	Boolean	Whether an invalid value in a field should cause the import to fail (true) or log a warning, skip that value, and proceed (false).
assetsCaseSensitive	Boolean	Whether to use case-sensitive matching when running in update-only mode (true) or try case-insensitive matching (false).
assetsTableViewAgnostic	Boolean	Whether to treat tables, views and materialized views as interchangeable (true) or strictly adhere to specified types in the input (false).
assetsFieldSeparator	String	Single character used to separate fields in the input file (for example, , or ;).
assetsCmHandling	CustomMetadataSemantic	How custom metadata in the input should be handled: ignore it, merge it with any existing asset custom metadata, or overwrite the existing asset custom metadata.
assetsTagHandling	AtlanTagSemantic	How Atlan tags on assets in the input should be handled: ignore them, append them to any existing asset tags, replace the existing asset tags, or remove them from the assets.
assetsLinkIdempotency	LinkIdempotencyInvariant	How linked resources on assets in the input should be updated: based on their unique URL or their unique name.
assetsBatchSize	Int	Maximum number of rows to process at a time (per API request).
trackBatches	Boolean	Whether to track details about every asset across batches (true) or only counts (false).

RelationalAssetsBuilder

The RelationalAssetsBuilder publish config accepts the following configurable inputs. You can use any combination of these, and any you leave out will use sensible defaults.

Relational assets

Variable Name	Data Type	Description
assetsFile	FrameworkRenderer.NamePathPair	File containing assets to import, typically passed through using transferFile().
assetsUpsertSemantic	ImportSemantic	Whether to allow the creation of new (full or partial) assets from the input CSV, or ensure assets are only updated if they already exist in Atlan.
deltaSemantic	DeltaSemantic	Whether to treat the input file as an initial load, full replacement (deleting any existing assets not in the file) or only incremental (no deletion of existing assets).
deltaRemovalType	RemovalType	How to delete any assets not found in the latest file.
deltaReloadCalculation	ReloadCalculation	Which assets to reload from the latest input CSV file. Changed assets only will calculate which assets have changed between the files and only attempt to reload those changes.
previousFileDirect	FrameworkRenderer.NamePathPair	Path to a direct file (locally) to use for delta processing. Note: providing a value for this will ignore any other previously-processed file in the object store, so please be sure this is the option you want to use (should be rare).
assetsAttrToOverwrite	Listing<String>	List of attributes you want to clear (remove) from assets if their value is blank in the provided file.
assetsFailOnErrors	Boolean	Whether an invalid value in a field should cause the import to fail (true) or log a warning, skip that value, and proceed (false).
assetsFieldSeparator	String	Single character used to separate fields in the input file (for example, , or ;).
assetsBatchSize	Int	Maximum number of rows to process at a time (per API request).
assetsCmHandling	CustomMetadataSemantic	How custom metadata in the input should be handled: ignore it, merge it with any existing asset custom metadata, or overwrite the existing asset custom metadata.
assetsTagHandling	AtlanTagSemantic	How Atlan tags on assets in the input should be handled: ignore them, append them to any existing asset tags, replace the existing asset tags, or remove them from the assets.
trackBatches	Boolean	Whether to track details about every asset across batches (true) or only counts (false).

LineageBuilder

The LineageBuilder publish config accepts the following configurable inputs. You can use any combination of these, and any you leave out will use sensible defaults.

Lineage assets

Variable Name	Data Type	Description
lineageFile	FrameworkRenderer.NamePathPair	File containing lineage to import, typically passed through using transferFile().
lineageUpsertSemantic	ImportSemantic	Whether to allow the creation of new (full or partial) assets from the input CSV, or ensure assets are only updated if they already exist in Atlan.
lineageFailOnErrors	Boolean	Whether an invalid value in a field should cause the import to fail (true) or log a warning, skip that value, and proceed (false).
fieldSeparator	String	Single character used to separate fields in the input file (for example, , or ;).
batchSize	Int	Maximum number of rows to process at a time (per API request).
cmHandling	CustomMetadataSemantic	How custom metadata in the input should be handled: ignore it, merge it with any existing asset custom metadata, or overwrite the existing asset custom metadata.
tagHandling	AtlanTagSemantic	How Atlan tags on assets in the input should be handled: ignore them, append them to any existing asset tags, replace the existing asset tags, or remove them from the assets.

---