[QTI] Perseus archive of an exercise with raw Perseus questions keeps legacy input questions whose answers are not numbers

[rtibblesbot]

❌ This issue is not open for contribution. Visit Contributing guidelines to learn about the contributing process and how to find suitable issues.

Overview

An exercise holding a legacy perseus_question and no native QTI item publishes only a Perseus archive. A legacy input question in it whose accepted answers Perseus can't read as a finite number (Sphere, +3, 1E3, 1e400) becomes a numeric-input with answers: [], which nothing a learner enters can satisfy. #6210 leaves the blank-answer ones out and excludes these; #6273 sends mixed native QTI exercises holding them to QTI only, where they publish as text entries.

Complexity: Low
Target branch: unstable

Context

• An exercise with a perseus_question and no native QTI item gets PerseusExerciseGenerator only:

  studio/contentcuration/contentcuration/utils/publish.py

  Lines 408 to 409 in f65cb93

  	elif any_perseus_question:
  	generator_classes = [PerseusExerciseGenerator]

• _process_input_answers drops every answer extract_value does not read as a number:

  studio/contentcuration/contentcuration/utils/assessment/perseus.py

  Lines 80 to 88 in f65cb93

  	def _process_input_answers(self, processed_data):
  	"""Extract input answer processing logic"""
  	non_empty_answers = []
  	for answer in processed_data["answers"]:
  	answer["answer"] = extract_value(answer["answer"])
  	if answer["answer"] or answer["answer"] == 0:
  	non_empty_answers.append(answer)
  	return {**processed_data, "answers": non_empty_answers}

• An exercise with a perseus_question and one legacy input question, published on unstable and with Publish mixed exercises QTI only when Perseus can't read a legacy input answer #6273:

  input answer   unstable                     with #6273
  "Sphere"       numeric-input, answers []    numeric-input, answers []
  "+3"           numeric-input, answers []    numeric-input, answers []
  "1E3"          numeric-input, answers []    numeric-input, answers []
  "1e400"        item JSON does not parse     numeric-input, answers []
  

The Change

• Replace the parallel number readers with one, shared by the Perseus export, QTI conversion and Perseus derivation: extract_value (utils/parser.py), the _JS_NUMBER fallback (utils/assessment/qti/convert.py) and Normalise xsd:double text entry answers for Perseus derivation #6259's normalisation (utils/assessment/qti/perseus_derive.py).
• Route exercises whose legacy input answers that reader rejects as Publish mixed exercises QTI only when Perseus can't read a legacy input answer #6273 routes exercises mixing native QTI and legacy items.

How to Get There

• Publish an exercise holding one legacy perseus_question and one legacy input question answered Sphere, then read the input question's item JSON in the .perseus file.

Acceptance Criteria

• In an exercise holding a legacy perseus_question, a legacy input question answered ["+3"] or ["1E3"] publishes in the Perseus archive with numeric-input answers [3] or [1000].
• An exercise holding a legacy perseus_question and a legacy input question answered ["Sphere"] or ["1e400"] publishes a QTI package holding both questions and no Perseus archive.
• No published file serves a legacy input question as a numeric-input with answers: [].
• The published node's AssessmentMetaData.assessment_item_ids lists only items its published files contain.
• test_exportchannel.py covers those exercises.

AI usage

Claude Code found this during an automated self-review of #6273, reproduced it with a scratch publish test on unstable and on that branch, and drafted this issue.