Add a QTI assessment-item type and renderable preset-bit ordering

[rtibbles]

Overview

Add the QTI assessment-item type and a stable preset→bit ordering for the renderable bitmask — both consumed by Studio and Kolibri.

Context

• AssessmentItem.type values come from le_utils.constants.exercises; Studio Adding raw deployments app to fle-utils #1 needs a QTI value.
• File.included_presets (studio Separated constants #8, kolibri Trying init file #10) is a bitmask whose bit positions derive from a stable preset ordering — like Kolibri's metadata-label bitmasks, which take each label's bit from its index in the le-utils ordered list.

The Change

• Add a QTI assessment-item type to constants.exercises.
• Define the preset→bit ordering for the renderable bitmask as append-only: existing presets keep their bit; new presets get the next.

Acceptance Criteria

• constants.exercises exposes a QTI assessment-item type
• A stable, append-only preset→bit mapping backs the renderable bitmask
• Studio and Kolibri can both import the type and the bit mapping

References

• studio Adding raw deployments app to fle-utils #1, studio Separated constants #8, kolibri Trying init file #10

AI usage

Architecture decided with the maintainer across an iterative session: blanket QTI type with the item XML in raw_data; XSD-authoritative validation across all sources; a legacy→QTI global migration with an API-layer dual-read; ricecooker upload delegating to the AssessmentItem serializer; and a Perseus custom-interaction contract confirmed against the QTI 3.0 specification. Claude mapped the existing publish/validation/ricecooker code, proposed the breakdown, and drafted each issue; the maintainer steered every decision and reviewed throughout.

[rtibblesbot]

Scoping notes before implementation

Ran this through a spec pass since it touches two ambiguous-looking spots in the codebase. Posting the resolution here rather than asking, since the codebase itself settles both — flag if either read is wrong.

QTI assessment-item type — constants.exercises gets QTI = "QTI", appended to question_choices. Uppercase is deliberate (breaks the file's existing lowercase-snake convention) because Studio's AssessmentItem.type choices are built directly from question_choices (contentcuration/models.py), and studio#5999 already fixes the literal as type='QTI'.

Preset→bit ordering — format_presets.py currently has two orderings of the same preset set that have already drifted apart: the hand-written choices display tuple and PRESETLIST (sourced from resources/presetlookup.json, dict/JSON order). E.g. video_dependency/video_thumbnail/video_subtitle and audio_thumbnail/audio_dependency appear in different relative order between the two, and choices has a precedent for silently dropping an entry (VIDEO_VECTOR, commented out). tests/test_presets.py::test_format_presets_are_synced only checks the two share the same set of keys, never the same order — confirming choices' order was never meant to be significant.

So the bit order is taken from PRESETLIST's existing order (i.e. presetlookup.json's), exposed as a small dedicated append-only id list (name TBD at implementation, e.g. RENDERABLE_PRESETS_ORDER) rather than a hand-maintained third ordering. Consumers compute 2 ** index themselves — le-utils only owns the order, mirroring the existing LEARNINGACTIVITIESLIST / Kolibri metadata-label-bitmask pattern this issue already points at (le-utils exposes the ordered list; kolibri/core/content/utils/search.py does the 2**j arithmetic). Not placed under constants/labels/ since that package is explicitly metadata-only and generated from specs — this is a functional preset concern, so it lives in format_presets.py next to PRESETLIST.

Full writeup: .specs/2026-07-02-qti-preset-bitmask-design.md (local scratch, not committed).

No open questions — ready for implementation.

[rtibblesbot]

Plan: Add a QTI assessment-item type to constants.exercises and a stable, append-only preset→bit ordering list to constants.format_presets, so Studio and Kolibri can both import them for issue #224 (studio #1, studio #8, kolibri #10).

• Add QTI assessment-item type
• Add RENDERABLE_PRESETS_ORDER preset→bit ordering list

---

@rtibblesbot's comments are generated by an LLM, and should be evaluated accordingly

How was this generated?

• Implemented code following the pre-approved plan using test-driven development (red-green-refactor)
• Ran the project's full test suite and linter before committing
• For UI changes: captured live screenshots to verify rendering
• For backend changes: verified with example API/CLI output