Please note, this is a large change, though most of the changes are unit tests. I have tried structure the commits in a logical order and keep changes separate. Please consider looking at the commits first to get an overview of the changes. The tests and extra debug logging is added first and the bugs are then fixed.

Note that this is blocking the 2020.2 release.

Link to issue number:

fixes #11132 - Say-all repeats some lines
fixes #11144 - Attempt to cancel speech for expired focus events: dialog title discarded when finding text in browse mode

Summary of the issue:

For #11132 - Say-all repeats some lines:

During say-all, after hitting an index, speechManager._handleIndex calls back to sayallHandler results in calling speechManager.speak and thus _pushNextSpeech, speechManager._handleIndex then goes on to call _pushNextSpeech itself.

For #11144 - Attempt to cancel speech for expired focus events: dialog title discarded when finding text in browse mode:

Titles were being dropped because they once had focus but when focus moves to the active control (automatically) they are considered previous focus and cancelled. In addition to this, removing cancelled speech was happening before all dependencies of the isValid callback were updated.

Wrapped indexes

During investigation I noticed that indexes were being compared naively, despite the fact that they are wrapped at a value of 10000. The indexes are not reset, it is feasible that this limit could be hit within a day and would cause some weird results.

Description of how this pull request fixes the issue:

This PR adds a lot of unit tests:

• Tests modeling the errors during say all
• Tests exercising the how cancellable speech commands should affect the speech manager.
• Tests based on the initial development of the SpeechManager (see New speech framework including callbacks, beeps, sounds, profile switches and prioritized queuing #7599), these were manual test cases.
• Tests for wrapped index comparisons

To support the unit tests, which could become hard to read and understand very quickly, a specific test harness was developed. There are a number of outputs to monitor, and it is useful to be able to assert the pre/post conditions for each step. To achieve this the test harness uses a context manager (see unit.test_speechManager.speechManagerTestHarness.SpeechManagerInteractions.expectation) which allows several post conditions to be added during the scope of a with block. These conditions are saved and used as the preconditions for the next with block.

Testing performed:

• Unit tests
• Manual testing of say-all double speaking case (with and without cancellable speech)
• Manual testing of changing focus rapidly in focus mode (moving through email subjects with up and down arrow) in Gmail in Chrome
• Manual testing of dialog title reporting with the run-dialog.

Known issues with pull request:

None.

Change log entry:

None