Don't attempt to format a tuple with "%s"

[agashlin]

This fixes #10736, errors in mousemove events over some MSHTML elements
(such as buttons).

Link to issue number:

fixes #10736

Summary of the issue:

mouse move over buttons doesn't read the button text in Internet Explorer 8 MSHTML, fails with a logged TypeError exception.

Description of how this pull request fixes the issue:

In order to keep the % operator from trying to format each element of the Point namedtuple, wrap the position in a one element tuple. This way the right exception gets thrown and caught. I went into more detail in #10736 (comment)

Testing performed:

I confirmed that this allowed the button to be read in the embedded MSHTML instance in the Firefox stub installer. Build 1 here can be used to demonstrate the issue, though it has other problems: the Re-install button does not read out when moused over unless this fix is made.

Known issues with pull request:

There are possibly other similar issues, but I didn't go looking for them.

Change log entry:

Bug fixes
Fix mouse tracking for some MSHTML elements

[josephsl]

Hi, can you verify that this works in IE11, as IE8 is too old at this time? Thanks.

[AppVeyorBot]

• PR introduces Flake8 errors
• Build (for testing PR)

See test results for failed build of commit c35370856f

[agashlin]

Hi, can you verify that this works in IE11, as IE8 is too old at this time? Thanks.

Yes, I confirmed that this issue is also present in IE11, and this fixes it.

source/NVDAObjects/IAccessible/MSHTML.py:333:44: E228 missing whitespace around modulo operator

I'll fix this and try running a local flake8.

[AppVeyorBot]

• System test failure
• Build (for testing PR)

See test results for failed build of commit 9b3f4a82d7

[agashlin]

I suspect this system test failure is unrelated to my change, as that test uses Chrome and this only touches MSHTML. Is it possible to re-run the system test?

[josephsl]

Hi, I think a better solution is reverting system tests commit, which will require everyone to rebase their PR’s afterwards. Thanks.

[feerrenrut]

I suspect this system test failure is unrelated to my change

@agashlin Yes we are getting intermittent failures that I haven't identified. I have restarted the build for you.

I think a better solution is reverting system tests commit

Rather than revert, I think we should just disable until we have time to look at what is causing the failure. I'll look at this today.

[feerrenrut]

Thanks for the investigation and fix @agashlin

Another approach might be to use an f-string eg: f"position: {position}"

[feerrenrut]

Since this is a small, low risk change, I think it should go into the next release which is currently stalling anyway. I'll change the target to beta

[feerrenrut]

But the PR will need to be rebased onto beta. I'm happy to do that.