Refactor scoring system to use Entity and Question API

I wonder if this is related to #2438421: Correct score is shown on quiz but incorrect score is shown in Quiz results table

Well, it didn't break anything :) Thanks!

Since the bug wasn't covered in the tests, let's get one written that fails without this patch.

I am running into a problem related to this as well.

Since some users have been testing and I manually reviewed all results, it seems I am able to locate the problem in my case:

The calculated score was only inconsistent for one test user who didn't finish a quiz in time (timed quiz).

• In the quiz report (node/%/quiz/results/%/view) the user has 17 of 30 possible points, which is calculated correctly to a score of 57%
• In a view where the calculated scores are generated (field => 'Quiz result: Score') for the logged in users(contextual filter => 'Quiz result: Uid'), for this user a score of '60% correct' is generated.

Because the user still has to 'submit' the current current question, after the timer reached it's end, it seems very plausible the last question which this user submitted was evaluated as correct for the result generated in the second point from the list above, but the last submitted question didn't count as correct (as desired) for the first point form the list above.

Since I am not a coder (yet;) I am not sure if this situation is already covered by the currently listed issues:

• Using a quiz with weighted scores, or optional questions with no points awarded even when correct - will NOT save a correct "percentage" of points earned.
• When finishing the quiz, in quiz_calculate_score, the function "quiz_question_quiz_question_score" is called. This function will return:
• When no quiz and no result ID are available, the maximum score for the question nid from the database
  When a result ID is available the maximum score as reported by the implementing QuizQuestionResponce class (which seems to in turn be calling the getMaximumScore() method of the question class)
• When the quiz is available, but not the result ID, the maximum score as reported by the implementing question class

Hope this might help.

I confirmed this is broken as it came up in a use case.

#5 fixes it, I'll try to write a test.

I noticed it doesn't work in weighted points but maybe that isn't a fix here but elsewhere.

For example

One quiz with

MCQ 1 (1 point)
Answer 1 correct 3pts
Answer 2 correct 3pts
Answer 3 correct 3pts

Multiple answers
Simple score off

Results in a quiz result score of 300%

STR

Create a quiz with three True/False questions

Set question 1 to be worth 1 point
Set question 2 to be worth 5 points
Set question 3 to be worth 20 points

Take the quiz and answer the last question correct.

On the results page you should see (correct)

You got 20 of 26 possible points.
Your score: 77%

On the manage results page you should see 33% (incorrect)

The quiz_node_results_answers is correct (20/26 points)

quiz_node_results table is wrong (score 33)

There was a lot of legacy code from Quiz 3 lying around so I removed it and replaced it with the Entity approach.

The score weighting wasn't being checked in every case.

There were also a bunch of different places where the weight was being calculated, so I'm trying to find all of those places and replace them with API calls.

Added back support for weighted random questions. Replaced old blocks of code.

This is fixed, with tests.

Thanks!

Rewriting the patch to be compatible with version 5.5

Rewriting the patch to be compatible with version 5.5

This Patch that is compatible with version 5.0-alpha10