Add dataset tags to SDK for identification (DE-7033) by vinay553 · Pull Request #456 · scaleapi/nucleus-python-client

vinay553 · 2026-04-06T20:38:33Z

Summary

Add tags field to DatasetInfo model so dataset.info() returns dataset tags
Add get_tags(), add_tags(), remove_tags() methods to Dataset class for programmatic tag management
Enables customers (e.g. Redwoods) to identify via API whether a dataset was labeled by Scale or another vendor

Test plan

Verify dataset.info() returns tags for a dataset with tags set in the UI
Verify dataset.add_tags(["Labeled by: Scale"]) adds tags
Verify dataset.get_tags() returns the current tag list
Verify dataset.remove_tags(["Labeled by: Scale"]) removes tags
Verify backward compatibility — dataset.info() works against a server that doesn't yet return tags (defaults to [])

🤖 Generated with Claude Code

Greptile Summary

Adds tags field to DatasetInfo (with a null-coercing validator for backward compatibility) and three new Dataset methods — get_tags(), add_tags(), and remove_tags() — that delegate to a new /dataset/{id}/tags endpoint.
remove_tags uses HTTP DELETE with a JSON body, which works with the current make_request plumbing but may be silently stripped by some intermediary infrastructure.
The coerce_null_tags Pydantic validator is missing the @classmethod decorator, relying on a pylint: disable comment instead of following the recommended v1 style.

Confidence Score: 5/5

Safe to merge — no P0 or P1 issues; only P2 style suggestions.

The change is well-scoped, backward-compatible (null-coercing validator + field default), and consistent with existing codebase patterns. Both findings are P2: a missing @classmethod decorator on the validator and a note about DELETE-with-body infrastructure risk. Neither blocks correctness.

No files require special attention; nucleus/dataset.py and dataset_info.py are the only changed files worth reviewing closely.

Important Files Changed

Filename	Overview
nucleus/data_transfer_object/dataset_info.py	Adds `tags: List[str] = []` field with a null-coercing validator for backward compatibility; validator is missing `@classmethod` decorator.
nucleus/dataset.py	Adds `get_tags()`, `add_tags()`, and `remove_tags()` methods; consistent with existing `make_request` patterns; `remove_tags` uses HTTP DELETE with a JSON body.
tests/test_dataset.py	Adds `test_dataset_tags` covering happy path, idempotency, and type-safety guard; comprehensive coverage of the new API surface.
CHANGELOG.md	Adds 0.18.2 changelog entry describing the new dataset tags feature.
pyproject.toml	Bumps version from 0.18.1 to 0.18.2.

Sequence Diagram

sequenceDiagram
    participant User
    participant Dataset
    participant NucleusClient
    participant API

    User->>Dataset: dataset.info()
    Dataset->>NucleusClient: "make_request({}, dataset/{id}/info, GET)"
    NucleusClient->>API: "GET /dataset/{id}/info"
    API-->>NucleusClient: "{tags: [...], ...}"
    NucleusClient-->>Dataset: response dict
    Dataset-->>User: "DatasetInfo(tags=[...])"

    User->>Dataset: dataset.get_tags()
    Dataset->>NucleusClient: "make_request({}, dataset/{id}/tags, GET)"
    NucleusClient->>API: "GET /dataset/{id}/tags"
    API-->>NucleusClient: "{tags: [...]}"
    NucleusClient-->>Dataset: response[tags]
    Dataset-->>User: List[str]

    User->>Dataset: dataset.add_tags([Labeled by Scale])
    Dataset->>NucleusClient: "make_request({tags:[...]}, dataset/{id}/tags, POST)"
    NucleusClient->>API: "POST /dataset/{id}/tags"
    API-->>NucleusClient: "{tags: [...]}"
    NucleusClient-->>Dataset: response[tags]
    Dataset-->>User: List[str]

    User->>Dataset: dataset.remove_tags([Labeled by Scale])
    Dataset->>NucleusClient: "make_request({tags:[...]}, dataset/{id}/tags, DELETE)"
    NucleusClient->>API: "DELETE /dataset/{id}/tags body:{tags:[...]}"
    API-->>NucleusClient: "{tags: [...]}"
    NucleusClient-->>Dataset: response[tags]
    Dataset-->>User: List[str]

Prompt To Fix All With AI

Fix the following 2 code review issues. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 2
nucleus/data_transfer_object/dataset_info.py:38-39
Pydantic v1's `@validator` is a classmethod under the hood, but the recommended practice (and what mypy expects) is to also apply `@classmethod` explicitly. Without it, type checkers may infer `cls` as the first argument type incorrectly, and the `# pylint: disable=no-self-argument` suppression is papering over the real fix.

```suggestion
    @validator("tags", pre=True, always=True)  # pylint: disable=used-before-assignment
    @classmethod
    def coerce_null_tags(cls, v):
```

### Issue 2 of 2
nucleus/dataset.py:474-477
`remove_tags` passes a request body via `requests.delete`. The `requests` library supports this, and `make_request` correctly forwards it as `json=payload`. However, some HTTP intermediaries (proxies, load balancers, AWS API Gateway) silently strip bodies from DELETE requests, which would cause the server to receive an empty `tags` list and could delete all tags instead of just the specified ones. If the backend supports a query-parameter approach or a dedicated `DELETE /dataset/{id}/tags/{tag}` sub-resource route, that would be more reliable across infrastructure.

_{Reviews (9): Last reviewed commit: "Add dataset tags to SDK for identificati..." | Re-trigger Greptile}

edwinpav

Can we add some unit/integration tests that call this on the datasets used by the tests?

Expose dataset tags through the Python SDK so customers can identify datasets labeled by Scale vs other vendors via the API. - Add `tags` field to DatasetInfo model (returned by dataset.info()) - Add get_tags(), add_tags(), remove_tags() methods to Dataset class - Use POST /tags/remove instead of DELETE to avoid proxy body-stripping - Use pydantic v1/v2 compat shim for null-coercion validator - Guard against passing a bare string instead of a list Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

vinay553 force-pushed the vinayparakala/de-7033-add-dataset-tags-to-sdk branch 2 times, most recently from 3eff859 to 8f06f7a Compare April 6, 2026 21:52

vinay553 requested a review from a team April 6, 2026 21:56

vinay553 force-pushed the vinayparakala/de-7033-add-dataset-tags-to-sdk branch from 8f06f7a to bf092a2 Compare April 6, 2026 21:57

edwinpav approved these changes Apr 7, 2026

View reviewed changes

vinay553 force-pushed the vinayparakala/de-7033-add-dataset-tags-to-sdk branch from bf092a2 to ccb48bc Compare April 8, 2026 21:42

vinay553 force-pushed the vinayparakala/de-7033-add-dataset-tags-to-sdk branch from ccb48bc to 9561851 Compare May 8, 2026 15:48

vinay553 force-pushed the vinayparakala/de-7033-add-dataset-tags-to-sdk branch from 9561851 to ce37cc4 Compare May 8, 2026 16:00

vinay553 merged commit 5c4b847 into master May 8, 2026
7 of 8 checks passed

vinay553 deleted the vinayparakala/de-7033-add-dataset-tags-to-sdk branch May 8, 2026 16:02

Navigation Menu

Uh oh!

Add dataset tags to SDK for identification (DE-7033) - #456

Add dataset tags to SDK for identification (DE-7033)#456
vinay553 merged 1 commit into
masterfrom
vinayparakala/de-7033-add-dataset-tags-to-sdk

vinay553 commented Apr 6, 2026 •

edited by greptile-apps Bot

Loading

Uh oh!

edwinpav left a comment •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

vinay553 commented Apr 6, 2026 • edited by greptile-apps Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

Greptile Summary

Confidence Score: 5/5

Important Files Changed

Sequence Diagram

Uh oh!

edwinpav left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

vinay553 commented Apr 6, 2026 •

edited by greptile-apps Bot

Loading

edwinpav left a comment •

edited

Loading