Skip to content

Creating an asset

All objects in the SDK that you can create within Atlan implement the builder pattern. This allows you to progressively build-up the object you want to create. In addition, each object provides a method that takes the minimal set of required fields to create that asset.

Each type of asset has a different containment hierarchy

Every asset in Atlan can have slightly different parent objects in which they exist. For example, a GlossaryTerm cannot exist outside a Glossary. A Column cannot exist outside a Table, View or MaterializedView; these cannot exist outside a Schema; which cannot exist outside a Database; which cannot exist outside a Connection.

The minimal required fields for each asset type will therefore be slightly different.

Creation order is important

As a result of this containment, creation order is important. Parent objects must be created (exist) before child objects can be created.

Build minimal object needed

2.0.0 1.0.0

For example, to create a glossary term you need to provide the name of the term and either the GUID or qualifiedName of the glossary in which to create the term:

Build minimal asset necessary for creation
1
2
3
GlossaryTermBuilder<?,?> termCreator = GlossaryTerm
        .creator("Example Term", // (1)
                  "b4113341-251b-4adc-81fb-2420501c30e6"); // (2)
  1. A name for the new term.
  2. The GUID or qualifiedName of the glossary in which to create the term.
Build minimal asset necessary for creation
1
2
3
4
5
6
7
8
from pyatlan.client.atlan import AtlanClient
from pyatlan.model.assets import AtlasGlossaryTerm

client = AtlanClient()
term = AtlasGlossaryTerm.creator(
    name="Example Term", # (1)
    glossary_guid="b4113341-251b-4adc-81fb-2420501c30e6" # (2)
)
  1. A name for the new term.
  2. The GUID of the glossary in which to create the term.
Build minimal asset necessary for creation
1
2
3
4
5
val termCreator = GlossaryTerm
    .creator(
        "Example Term",  // (1)
        "b4113341-251b-4adc-81fb-2420501c30e6",  // (2)
    )
  1. A name for the new term.
  2. The GUID or qualifiedName of the glossary in which to create the term.

Implicit in the API calls below

There is nothing specific to do for this step when using the raw APIs — constructing the object is simply what you place in the payload of the API calls in the steps below.

Create the asset from the object

1.4.0 4.0.0

This term object will have the minimal required information for Atlan to create it. You must then actually persist the object in Atlan1:

Create the asset
 5
 6
 7
 8
 9
10
11
12
13
14
15
GlossaryTerm term = termCreator.build(); // (1)
AssetMutationResponse response = term.save(client); // (2)
Asset created = response.getCreatedAssets().get(0); // (3)
if (created instanceof GlossaryTerm) {
    term = (GlossaryTerm) created; // (4)
}
Asset updated = response.getUpdatedAssets().get(0); // (5)
Glossary glossary;
if (updated instanceof Glossary) {
    glossary = (Glossary) updated; // (6)
}
  1. Before you can take actions on the builder object you've been interacting with, you need to build() it into a full object.
  2. Then you can do operations, like save(), which will either:

    • create a new asset, if Atlan does not have a term with the same name in the same glossary
    • update an existing asset, if Atlan already has a term with the same name in the same glossary

    Because this operation will persist the asset in Atlan, you must provide it an AtlanClient through which to connect to the tenant.

  3. You can distinguish what was created or updated:

    • getCreatedAssets() lists assets that were created
    • getUpdatedAssets() lists assets that were updated

    Note that the save() method always returns objects of type Asset, though.

  4. The Asset class is a superclass of all assets. So you need to cast to more specific types (like GlossaryTerm) after verifying the object that was actually returned.

  5. In this example, creating the GlossaryTerm actually also updates the parent Glossary. This is why the response contains generic Asset objects rather than specific types — any operation could side-effect a number of different assets.

  6. Like with the GlossaryTerm, you can check and cast the generic Asset returned by the response into its more specific type (Glossary).

Create the asset
 9
10
11
12
13
14
15
response = client.asset.save(term) # (1)
created = response.assets_created(asset_type=AtlasGlossaryTerm) # (2)
if created: # (3)
    term = created[0] # (4)
updated = response.assets_updated(asset_type=AtlasGlossaryTerm) # (5) 
if updated: # (6)
    term = updated[0] # (7)
  1. Call the save method which will create or update the asset in atlan.

  2. You can distinguish what was created or updated:

    • assets_created(asset_type=AtlasGlossaryTerm) returns a lists assets of the specified type that were created.
    • assets_updated(asset_type=AtlasGlossaryTerm) returns a lists assets of the specified type that were updated.
  3. Check if the list is empty to determine if an AtlasGlossaryTerm was created.

  4. Get the new AtlasGlossaryTerm that was created.

  5. In this example, creating the AtlasGlossaryTerm actually also updates the parent AtlasGlossary. This is why the response contains an AtlasGlossary.

  6. Check if the list is empty to determine if an AtlasGlossary was updated.

  7. Get the AtlasGlossary that was updated.

Create the asset
 6
 7
 8
 9
10
11
12
13
var term = termCreator.build() // (1)
val response = term.save(client) // (2)
val created = response.createdAssets[0] // (3)
if (created is GlossaryTerm) {
    term = created // (4)
}
val updated = response.updatedAssets[0] // (5)
val glossary = if (updated is Glossary) updated else null // (6)
  1. Before you can take actions on the builder object you've been interacting with, you need to build() it into a full object.
  2. Then you can do operations, like save(), which will either:

    • create a new asset, if Atlan does not have a term with the same name in the same glossary
    • update an existing asset, if Atlan already has a term with the same name in the same glossary

    Because this operation will persist the asset in Atlan, you must provide it an AtlanClient through which to connect to the tenant.

  3. You can distinguish what was created or updated:

    • getCreatedAssets() lists assets that were created
    • getUpdatedAssets() lists assets that were updated

    Note that the save() method always returns objects of type Asset, though.

  4. The Asset class is a superclass of all assets. So you need to cast to more specific types (like GlossaryTerm) after verifying the object that was actually returned.

  5. In this example, creating the GlossaryTerm actually also updates the parent Glossary. This is why the response contains generic Asset objects rather than specific types — any operation could side-effect a number of different assets.

  6. Like with the GlossaryTerm, you can check and cast the generic Asset returned by the response into its more specific type (Glossary).

POST /api/meta/entity/bulk
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
{
  "entities": [ // (1)
    {
      "typeName": "AtlasGlossaryTerm", // (2)
      "attributes": {
        "name": "Example Term", // (3)
        "qualifiedName": "Example Term", // (4)
        "anchor": { // (5)
          "typeName": "AtlasGlossary",
          "guid": "b4113341-251b-4adc-81fb-2420501c30e6"
        }
      }
    }
  ]
}
  1. All assets must be wrapped in an entities array.
  2. You must provide the exact type name for the asset (case-sensitive). For a term, this is AtlasGlossaryTerm.
  3. You must provide the exact name of the asset (case-sensitive).
  4. You must provide a qualifiedName of the asset (case-sensitive). In the case of glossary objects (like terms), this will actually be replaced in the back-end with a generated qualifiedName, but you must provide some value when creating the object.
  5. You must also specify the parent object in which this object is contained (if any). In the case of a term, it can only exist within a glossary. So here we specify the details of the parent glossary through the anchor relationship (specific to glossary assets).

(Optional) Enrich before creating

2.0.0 4.0.0

If you want to further enrich the asset before creating it, you can do this using the builder pattern:

Alternatively, further enrich the asset before creating it
 5
 6
 7
 8
 9
10
11
GlossaryTerm term = termCreator // (1)
        .certificateStatus(CertificateStatus.VERIFIED) // (2)
        .announcementType(AtlanAnnouncementType.INFORMATION)
        .announcementTitle("Imported")
        .announcementMessage("This term was imported from ...")
        .build(); // (3)
AssetMutationResponse response = term.save(client); // (4)
  1. We'll create an object you can take actions on from this creator.
  2. In this example, you're adding a certificate and announcement to the object.
  3. To persist the enrichment back to the object, you must build() the builder.
  4. You can call the save() operation against this enriched object, the same as shown earlier. Because this operation will persist the asset in Atlan, you must provide it an AtlanClient through which to connect to the tenant.

Assign the result back

Remember to assign the result of the build() operation back to a variable! (In the example above this happens on line 5 with GlossaryTerm term =.)

Alternatively, further enrich the asset before creating it
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
from pyatlan.client.atlan import AtlanClient
from pyatlan.model.assets import AtlasGlossary, AtlasGlossaryTerm
from pyatlan.model.enums import AnnouncementType, CertificateStatus

client = AtlanClient()
term = AtlasGlossaryTerm.creator(
    name="Example Term",
    glossary_guid="b4113341-251b-4adc-81fb-2420501c30e6"
)
term.certificate_status = CertificateStatus.VERIFIED
announcement = Announcement(
    announcement_type=AnnouncementType.INFORMATION,
    announcement_title="Imported",
    announcement_message="This term was imported from ..",
)
term.set_announcement(announcement)
response = client.asset.save(term) # (1)
  1. You can call the save() operation against this enriched object, the same as shown earlier.
Alternatively, further enrich the asset before creating it
 6
 7
 8
 9
10
11
12
val term = termCreator // (1)
    .certificateStatus(CertificateStatus.VERIFIED) // (2)
    .announcementType(AtlanAnnouncementType.INFORMATION)
    .announcementTitle("Imported")
    .announcementMessage("This term was imported from ...")
    .build() // (3)
val response = term.save(client) // (4)
  1. We'll create an object you can take actions on from this creator.
  2. In this example, you're adding a certificate and announcement to the object.
  3. To persist the enrichment back to the object, you must build() the builder.
  4. You can call the save() operation against this enriched object, the same as shown earlier. Because this operation will persist the asset in Atlan, you must provide it an AtlanClient through which to connect to the tenant.

Assign the result back

Remember to assign the result of the build() operation back to a variable! (In the example above this happens on line 6 with val term =.)

POST /api/meta/entity/bulk
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
{
  "entities": [ // (1)
    {
      "typeName": "AtlasGlossaryTerm",
      "attributes": {
        "name": "Example Term",
        "qualifiedName": "Example Term",
        "anchor": {
          "typeName": "AtlasGlossary",
          "guid": "b4113341-251b-4adc-81fb-2420501c30e6"
        },
        "certificateStatus": "VERIFIED", // (2)
        "announcementType": "information",
        "announcementTitle": "Imported",
        "announcementMessage": "This term was imported from..."
      }
    }
  ]
}
  1. You would still create the asset by wrapping it within the entities array.
  2. But you can also extend the information you store on the asset. In this example, you're adding a certificate and announcement to the object when it is created.

  1. Why no distinction between creation and update? This has to do with how Atlan detects changes — see the Importance of identifiers for a more detailed explanation.