Class: testable_question_attempt  - X-Ref

Makes some protected methods of question_attempt public to facilitate testing.

Class: testable_question_engine_unit_of_work  - X-Ref

Test subclass to allow access to some protected data so that the correct
behaviour can be verified.

Class: question_test_helper  - X-Ref

Base class for question type test helpers.

Class: test_question_maker  - X-Ref

This class creates questions of various types, which can then be used when
testing.

Class: testing_db_record_builder  - X-Ref

Helper for tests that need to simulate records loaded from the database.

Class: data_loading_method_test_base  - X-Ref

Helper base class for tests that need to simulate records loaded from the
database.

Class: question_testcase  - X-Ref

Class: question_contains_tag_with_contents  - X-Ref

Class: question_check_specified_fields_expectation  - X-Ref

Class: question_contains_select_expectation  - X-Ref

Class: question_does_not_contain_tag_with_attributes  - X-Ref

Class: question_contains_tag_with_attribute  - X-Ref

Class: question_contains_tag_with_attributes  - X-Ref

Class: question_pattern_expectation  - X-Ref

Class: question_no_pattern_expectation  - X-Ref

Class: qbehaviour_walkthrough_test_base  - X-Ref

Helper base class for question walk-through tests.

The purpose of tests that use this base class is to simulate the entire
interaction of a student making an attempt at a question. Therefore,
these are not really unit tests. They would more accurately be described
as integration tests. However, whether they are unit tests or not,
it works well to implement them in PHPUnit.

Historically, tests like this were made because Moodle did not have anything
like Behat for end-to-end testing. Even though we do now have Behat, it makes
sense to keep these walk-through tests. They run massively faster than Behat
tests, which gives you a much faster feedback loop while doing development.
They also make it quite easy to test things like regrading the attempt after
the question has been edited, which would be possible but very fiddly in Behat.

Ideally, the full set of tests for the question class of a question type would be:

1. A lot of unit tests for each qtype_myqtype_question class method
like grade_response, is_complete_response, is_same_response, ...

2. Several of these walk-through tests, to test the end-to-end interaction
of a student with a question, for example with different behaviours.

3. Just one Behat test, using question preview, to verify that everything
is plugged together correctly and works when used through the UI.

What one would expect to see in one of these walk-through tests is:

// 1. Set up a question: $q.

// 2. A call to $this->start_attempt_at_question($q, ...); with the relevant options.

// 3. Some number of calls to $this->process_submission passing an array of simulated
//    POST data that matches what would be sent back be submitting a form that contains
//    the form fields that are output by rendering the question. This is like clicking
//    the 'Check' button in a question, or navigating to the next page in a quiz.

// 4. A call to $this->finish(); which is the equivalent of clicking
//    'Submit all and finish' in the quiz.

// 5. After each of steps 2-4 above, one would expect to see a certain amount of
//    validation of the state of the question and how the question is rendered,
//    using methods like $this->check_current_state(), $this->check_current_output, etc.

The best way to work out how to write tests like this is probably to look at
some examples in other question types or question behaviours.

In writing these tests, it is worth noting the following points:

a) The easiest mistake to make is at step 3. You need to ensure that your
simulated post data actually matches what gets sent back when the
question is submitted in the browser. Try checking it against the
HTTP POST requests you see in your browser when the question is submitted.
Some question types have a $q->prepare_simulated_post_data() method that
can help with this.

b) In the past, tests like these used to contain even more repetitive code,
and so they were re-factored to add the helper methods like
start_attempt_at_question, process_submission, finish. That change had
good effects, like reducing duplicate code. However, there were down-sides.
The extra layers of indirection hide what is going on, which means these
tests are harder to understand until you know what the helpers are doing.
If you want an interesting exercise, take one of the walk-through tests,
and inline all the helpers. This might be a good way to understand more about
the question engine API. However, having made the everything-inlined code
and learned from the process, you should then just throw it away.

c) The way check_current_output works is weird. When these tests were first written
Moodle used SimpleTest for unit tests and check_current_output os written in a
style that made sense there. When we moved to PHPUnit, a quick and dirty
conversion was done. That was a pragmatic move at the time, and we just have
to live with the result. Sorry. (And: don't copy that style for new things.)

Class: question_test_recordset  - X-Ref

Simple class that implements the {@link moodle_recordset} API based on an
array of test data.

See the {@link question_attempt_step_db_test} class in
question/engine/tests/testquestionattemptstep.php for an example of how
this is used.