fix: enforce uniqueness on computed attributes (closes #7924)

[polmichel]

Why

Loading objects whose human-friendly id (or any unique: true field) is a computed Jinja2 attribute could create duplicates.

Schema used

---
version: "1.0"

nodes:
  - name: Stuff
    namespace: Random
    display_label: name__value
    human_friendly_id:
      - name__value
    attributes:
      - name: name
        kind: Text
        computed_attribute:
          kind: Jinja2
          jinja2_template: "{{ description__value | upper }}-STUFF"
        read_only: true
        unique: true
        optional: false
      - name: description
        kind: Text

Submitting four RandomStuff objects with the same description produced four nodes sharing the same computed name / HFID, because the uniqueness check silently skipped the computed field.

Root cause: the constraint widened its check filter only for optional unique attributes. A required computed attribute is never in the user-supplied data, and the macro populates its value only after the filter was captured, so the check was bypassed.

Closes #7924

What changed

• A required computed unique attribute is now enforced on create. Duplicates are rejected before reaching the database.
• The violation message names the input attributes that drive the computed value, e.g. Violates uniqueness constraint 'name' (computed from: model) (or owner.name for relationship-derived inputs), so the user knows which field to change rather than seeing an opaque reference to a read-only field.

What stayed the same

• NodeAttributeUniquenessConstraint deliberately left alone: it is registered as a dependency but not wired into the production NodeConstraintRunner.

How to test

uv run pytest backend/tests/unit/core/node/constraints/test_uniqueness_violation_message.py -v
uv run pytest backend/tests/component/core/node/test_create_node_computed_uniqueness.py -v
uv run pytest backend/tests/component/core/constraint_validators/ -v

Reproduction from the issue: load the RandomStuff schema (computed unique name), then load four objects with identical description. Before: four duplicates created. After: rejected with Violates uniqueness constraint 'name' (computed from: description).

Follow-up: complementary SDK-side guard

This PR fixes enforcement at the server boundary, which is the correct authoritative place. A complementary improvement could live in the Python SDK / infrahubctl object load to catch the common case before anything is sent to the server.

Checklist

• Tests added/updated
• Changelog entry added
• External docs updated (n/a — bug fix, no user-facing doc change)
• Internal .md docs updated (n/a)
• I have reviewed AI generated content

[cubic-dev-ai]

No issues found across 12 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

<sub>Shadow auto-approve: would auto-approve. The change fixes a specific bug where computed unique attributes were not enforced for uniqueness, and the implementation is well-contained with targeted logic, good tests, and a dedicated message builder that does not affect other parts of the system.

Re-trigger cubic</sub>

[codspeed-hq]

Merging this PR will not alter performance

✅ 12 untouched benchmarks

---

_{Comparing ai-bug-pipeline-7924-computed-attr-uniqueness (43964cb) with stable (edaeba7)¹}

Footnotes

1. No successful run was found on stable (969810f) during the generation of this report, so edaeba7 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[polmichel]

I was wondering if a Noop class could be useful to avoid instantiating this class every time we instantiate NodeGroupedUniquenessConstraint or if this is over-engineered. Not sure of the answer, so I let the code in this state for this iteration.

[cubic-dev-ai]

0 issues found across 2 files (changes from recent commits).

<sub>Shadow auto-approve: would auto-approve. This is a targeted bug fix that extends the uniqueness constraint check to computed attributes, preventing duplicate HFIDs; the logic change is minimal and well-encapsulated, the new message builder is isolated and testable, and comprehensive tests confirm correctness without altering critical...

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 1 file (changes from recent commits).

<sub>Shadow auto-approve: would auto-approve. This PR fixes a well-scoped bug where computed unique attributes were not being enforced for uniqueness, with a small logic change and thorough test coverage, making it low-risk for production.

Re-trigger cubic</sub>