[OPIK-2031] Add tags support to datasets

[BorisTkachenko]

Details

Add tags support to datasets

Testing

Integration tests

[github-actions]

Backend Tests Results

  184 files    184 suites   20m 46s ⏱️
4 282 tests 4 279 ✅ 3 💤 0 ❌
4 271 runs  4 268 ✅ 3 💤 0 ❌

Results for commit ea1158a.

[Copilot]

Pull Request Overview

This PR adds support for tagging datasets across the backend API, persistence layer, filtering, sorting, and tests.

• Adds a new tags column via Liquibase and updates the domain model and DAO to persist and query tags.
• Extends the filtering and sorting infrastructure to handle tags on datasets.
• Updates API resources, criteria, and integration tests to accept and validate dataset tags.

Reviewed Changes

Copilot reviewed 13 out of 14 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
DatasetsResourceTest.java	Updated tests to cover sorting and filtering by tags
000022_add_tags_to_datasets.sql	Liquibase migration adding the tags column
FilterStrategy.java	Added DATASET strategy for dataset-level filters
FilterQueryBuilder.java	Mapped DatasetField.TAGS to the DB column
DatasetService.java	Wired filter query builder into service find methods
DatasetDAO.java	Extended SQL queries and mapper for tags and filters
DatasetCriteria.java	Added filters list to criteria
SortingFactoryDatasets.java	Included TAGS in sortable fields
DatasetsResource.java	Accepts filters query param and populates criteria
DatasetFilter.java	New class representing dataset filter payloads
DatasetField.java	Enum for dataset filterable fields
DatasetUpdate.java	Added tags field to update payload
Dataset.java	Added tags field to the dataset API model

Comments suppressed due to low confidence (4)

apps/opik-backend/src/test/java/com/comet/opik/api/resources/v1/priv/DatasetsResourceTest.java:1923

• List in Java does not have a reversed() method. Consider using Collections.reverse on a mutable list or ListUtils.reversed() to produce a reversed view.

            List<Dataset> expectedDatasets = getExpectedDatasets.apply(datasets).reversed();

apps/opik-backend/src/test/java/com/comet/opik/api/resources/v1/priv/DatasetsResourceTest.java:1936

• List in Java does not provide a getFirst() method. You should use datasets.get(0) to access the first element.

                                    .value(datasets.getFirst().tags().iterator().next())

apps/opik-backend/src/test/java/com/comet/opik/api/resources/v1/priv/DatasetsResourceTest.java:1938

• List in Java does not provide a getFirst() method; use datasets.get(0) when creating the singleton list.

                            (Function<List<Dataset>, List<Dataset>>) datasets -> List.of(datasets.getFirst())),

apps/opik-backend/src/main/java/com/comet/opik/api/filter/DatasetField.java:9

• The symbol TAGS_QUERY_PARAM is not defined in this enum. Define a constant for the request query parameter name (e.g., "tags") or reference the correct constant.

    TAGS(TAGS_QUERY_PARAM, FieldType.STRING_STATE_DB),

[YarivHashaiComet]

do we need the "," ? or we can just ; in the same line?

[andrescrz]

This is better for extending to additional values. Nicer to review in those cases as well.

[YarivHashaiComet]

plz make it the same line as 13 (dont break line)

[andrescrz]

It should be like it is, and these are handled and formatted automatically by spotless.

[andrescrz]

LGTM, just a couple of minor recommendations.

[andrescrz]

This is better for extending to additional values. Nicer to review in those cases as well.

[andrescrz]

It should be like it is, and these are handled and formatted automatically by spotless.

[andrescrz]

Minor: id, name, created_at and last_updated_at already exists. I'd reuse them and make the name neutral of the DB, as we follow the same naming convention in both. If you prefer not to, I wouldn't tie this naming to the dataset. I'd just rename them for the state DB.

[andrescrz]

Minor: add new line to all files. Maybe we can automate this with spotless.