Normalize legacy multi-answer text-entry QTI items on publish

[rtibblesbot]

Summary

• Pre-QTI editor emits cardinality="multiple" and match_correct for multi-answer Numeric and Text entry questions #6187 multi-answer Numeric/Text entry items use cardinality="multiple" + match_correct, so Kolibri scores every response incorrect.
• Publish rewrites them to cardinality="single", first answer in <qti-correct-response>, one <qti-map-entry> per answer, map_response.xml template.
• Pure raw_data -> raw_data function (qti/normalize.py) called from _create_native_qti_item; stored raw_data is untouched.
• Chosen over a data migration, which would miss later writes from stale clients or ricecooker.
• Only declarations bound to qti-text-entry-interaction change.
• Existing <qti-mapping> is kept, preserving per-answer case-sensitive.
• Items with explicit response rules, several declarations, a DOCTYPE, or element content in qti-value are skipped with a warning.

References

Fixes #6223. Siblings under #6103: #6187, #6220.

Reviewer guidance

• pytest contentcuration/contentcuration/tests/utils/qti contentcuration/contentcuration/tests/utils/test_exercise_creation.py
• Perseus derivation still reads stored raw_data deliberately: the old shape lists every answer in <qti-correct-response>, which _derive_text needs.

AI usage

I used Claude Code to plan and implement this from the issue, with tests written alongside the code. I reviewed the diff and trimmed it for simplicity.

Deviations from the issue spec

Criterion	Why not
A saved multi-answer text-entry item is answerable without re-saving	Items with explicit response rules, several declarations, a DOCTYPE, or element content in qti-value are left as written (logged) and stay unscored; rewriting them safely needs per-item rule handling.

---

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

How was this generated?

• Ran pre-flight CI checks (lint, format, tests) and verified all pass
• Rebased onto the target branch and resolved any conflicts
• Reorganized commit history into clean, logical commits
• Audited the diff to ensure only issue-relevant files are changed
• Built PR body from the repository's PR template with evidence blocks

🟡 Waiting for feedback Last updated: 2026-09-30 22:16 UTC

[rtibblesbot]

Outside this PR's scope — tracked under #6103:

• Published QTI item with a single-quoted xml:lang is written as malformed XML #6230 Published QTI item with a single-quoted xml:lang is written as malformed XML

---

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

How was this generated?

• Ran pre-flight CI checks (lint, format, tests) and verified all pass
• Rebased onto the target branch and resolved any conflicts
• Reorganized commit history into clean, logical commits
• Audited the diff to ensure only issue-relevant files are changed
• Built PR body from the repository's PR template with evidence blocks

[AlexVelezLl]

The erroneous code has not yet reached production. So far, there are no QTI assessment items in the production DB. The only items potentially at risk are the QA items; in that case, we will ask our QA team to edit them again only after we have applied the fix for #6187.