[evakravi]

Problem

HubContent has no get_all. The resource generator skips a List API when the list summary lacks a member that the Describe input requires, and HubContentInfo omits HubName, the name the list call was made with. As a result, every caller pages list_hub_contents on the boto3 client by hand: jumpstart/search.py, jumpstart/hub/hub.py, recipe_utils.py, ai_registry/air_hub.py, training_session/_utils.py, and model_aliases.py in this repo, and three scripts in the SageMaker agent skills.

Two more defects follow from the same gap. HubContentInfo carries OriginalCreationTime, the date a model first entered the catalog, which DescribeHubContent does not return, and the HubContent class had no field for it, so ResourceIterator dropped it (the class forbids extra fields). And HubContent.get_all_versions() built each version from a summary with no hub_name, so the first refresh() failed:

botocore.exceptions.ParamValidationError: Missing required parameter in input: "HubName"

Solution

The generator learns two per-resource tables. GET_ALL_INPUT_ATTRIBUTES names list-input members to copy onto every listed resource (HubName to hub_name), and LIST_SUMMARY_EXTRA_ATTRIBUTES names summary-only members to keep as optional attributes (OriginalCreationTime). ResourceIterator takes the copied members as extra_init_data and applies them only where the summary has no value. resources.py is the regenerated output: HubContent.get_all(hub_name, hub_content_type, name_contains, max_schema_version, creation_time_before, creation_time_after, sort_by, sort_order, session, region), the original_creation_time attribute, and the hub_name binding on get_all_versions(). No other resource class changes.

Tests

New sagemaker-core/tests/unit/test_hub_content_get_all.py, 7 tests: paging with NextToken and hub_name bound on each resource, the list filters forwarded, original_creation_time kept from the summary, each per-item refresh sent with HubName, the session and region bound to each resource, get_all_versions() refreshing with the parent's hub_name, and extra_init_data not overriding a summary value.

cd sagemaker-core && python -m pytest tests/unit/test_hub_content_get_all.py -q
7 passed

On origin/master the same file reports 7 failed (AttributeError: type object 'HubContent' has no attribute 'get_all', and the HubName parameter error on get_all_versions).

Existing suites on this branch: tests/unit/tools, tests/unit/test_resource_client_context.py, tests/unit/generated/test_utils.py, tests/unit/generated/test_resources.py, tests/unit/jumpstart/hub, tests/unit/jumpstart/test_search_unit.py: 196 passed, 1 skipped. python src/sagemaker/core/tools/codegen.py regenerates resources.py to the committed content.

black --line-length 100, flake8, and pydocstyle pass on the changed hand-written files.

Merge Checklist

• I read the contribution guide.
• The change is backward compatible.
• The commits use the repository commit-message format.
• The change creates no S3 or STS client.
• The change has regression coverage and adds no dependency.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[evakravi]

Superseded by #6533, which combines the four changes into one PR and moves the hub-model metadata accessors onto HubContent itself (a HubContentMixin in sagemaker.core.hub_content) instead of a separate class.

[github-actions]

🤖 AI code review

This PR adds HubContent.get_all, keeps OriginalCreationTime on listed hub content, and binds hub_name onto resources produced by get_all/get_all_versions so their first refresh() succeeds. It does this by teaching the codegen two per-resource tables (GET_ALL_INPUT_ATTRIBUTES, LIST_SUMMARY_EXTRA_ATTRIBUTES) plus an extra_init_data path in ResourceIterator, and regenerating resources.py. The logic is sound: the Describe-required-but-summary-missing HubName is now excluded from the "skip get_all" check and copied onto each resource via setdefault (so summary values win), and the extra original_creation_time field is added to the class so it is no longer dropped. Tests are thorough and cover paging, filter forwarding, the retained attribute, per-item refresh, session/region binding, and the setdefault precedence. Overall it looks correct and well-tested; only one low-severity latent limitation noted.

_{Reviewed commit d5068d5. Automated review; verify before acting.}

[github-actions]

🟡 Low: The LIST_SUMMARY_EXTRA_ATTRIBUTES loop is added only to the get-method branch of _get_class_attributes. Since the following branch is elif "get_all" in class_methods: (mutually exclusive), a resource that has get_all but no get would be generated without the extra summary-only attribute, so ResourceIterator would again silently drop it. This is harmless today because the only entry (HubContent) also has a get method, but if the table is extended to a get_all-only resource the attribute would be lost without an obvious failure. Consider applying the same loop in the elif "get_all" branch (or factoring it out) to avoid the trap.