[document intelligence] Fix copy operation polling

[catalinaperalta]

Clean up fix from #39957 to be more similar to generated layer and leverage available lro options config.

[Copilot]

Pull Request Overview

This PR addresses issues with the copy operation polling logic in the Document Intelligence SDK by cleaning up the implementation to be more aligned with the generated layer and using the available long-running operation (LRO) options configuration. Key changes include removing the custom polling algorithm (DocumentModelAdministrationPolling), updating response deserialization logic to use .get("result"), and refactoring polling configuration to use the lro_options parameter.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/aio/_operations/_patch.py	Removed the custom polling algorithm import and modified the AsyncLROBasePolling configuration; updated deserialization calls to use .get("result")
sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/_operations/_patch.py	Removed the custom polling class and updated LROBasePolling usage to use lro_options with "final-state-via": "operation-location"; updated deserialization accordingly

Comments suppressed due to low confidence (6)

sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/aio/_operations/_patch.py:269

• Switching from direct indexing to .get() could lead to a None value if 'result' is missing; if this behavior is intentional to handle absent keys, consider adding a clarifying comment.

deserialized = _deserialize(_models.DocumentModelDetails, response.json().get("result"))

sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/aio/_operations/_patch.py:284

• Ensure that replacing lro_algorithms with lro_options aligns with the intended behavior for retrieving the final state of the long-running operation.

lro_options={"final-state-via": "operation-location"},

sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/aio/_operations/_patch.py:348

• Verify that the updated polling configuration using lro_options provides the same final state behavior as the previous implementation with DocumentModelAdministrationPolling.

AsyncLROBasePolling(lro_delay, path_format_arguments=path_format_arguments, lro_options={"final-state-via": "operation-location"}, **kwargs),

sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/_operations/_patch.py:287

• Switching to .get() for the 'result' field might silently pass None to _deserialize; confirm that this is acceptable for handling missing keys.

deserialized = _deserialize(_models.DocumentModelDetails, response.json().get("result"))

sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/_operations/_patch.py:302

• Review the updated LROBasePolling configuration to ensure that the removal of the previous polling algorithm does not affect the correct final state handling of the operation.

lro_options={"final-state-via": "operation-location"},

sdk/documentintelligence/azure-ai-documentintelligence/azure/ai/documentintelligence/_operations/_patch.py:368

• Ensure that the refactored LROBasePolling call with lro_options maintains consistent behavior in retrieving the final operation state compared to the previous implementation.

LROBasePolling(lro_delay, path_format_arguments=path_format_arguments, lro_options={"final-state-via": "operation-location"}, **kwargs)

[azure-sdk]

API change check

API changes are not detected in this pull request.