bpo-44026: Idle get Python's 'did you mean'

[E-Paine]

Based on the test_exception code included on bpo by @sweeneyde. If the exception is an AttributeError or NameError, we remove the last line from the normal traceback and replace it with one captured from sys.__excepthook__. This implementation utilises test.support.captured_stderr, like the original.

https://bugs.python.org/issue44026

[E-Paine]

Pipelines failure is an unrelated test_concurrent_futures issue

[shreyanavigyan]

AFAICT this PR needs a news entry

[E-Paine]

The pipelines test is a different test_concurrent_futures failure. I cannot see how these failures could be linked to this PR so am inclined to ignore them.

[sweeneyde]

Perhaps it would be better to use contextlib.redirect_stderr(to), since, from the docs:

Note test.support is not a public module. It is documented here to help Python developers write tests. The API of this module is subject to change without backwards compatibility concerns between releases. 

[E-Paine]

Perhaps it would be better to use contextlib.redirect_stderr(to)

I did not know that existed; thanks for pointing me to it!

[E-Paine]

OK, that's the third consecutive pipelines Windows failure of test_concurrent_futures. Is this an issue with other PRs, or could this (somehow) be to do with my changes? I'll try a merge and see if that fixes it.

[shreyanavigyan]

The test you're talking about has failed in lot of PRs recently.

[terryjreedy]

Paine, I am currently reviewing and revising, so please no more changes.

[terryjreedy]

Functions should only be nested if they need to be. They cannot be tested properly if they are. I expanded the scope slightly to make it more 'complete'. I then reduced the scope of the new tests to testing the new unit, and used subTest to combine and add a 'normal' case.

print_exc should be checked to see if it should be nested. get_message_lines should become get_message. Currently, format_exception_only only returns multiple lines. Combining them could be done within the function. I left that for another issue.

I think this is ready to merge, but am leaving for others and myself to look at again (and, I hope, for someone else to fix the test failures if not done yet).

Current coverage is 54% on my machine. Should update if any other edits.

[E-Paine]

Sorry Terry, I know you asked for no more changes, but I need to patch it to correctly deal with multiple errors in a traceback (I will also add a test for this).

>>> try: ... abc ... except NameError: ... bcd ... Traceback (most recent call last): File "<pyshell#8>", line 2, in <module> abc NameError: name 'bcd' is not defined During handling of the above exception, another exception occurred: Traceback (most recent call last): File "<pyshell#8>", line 4, in <module> bcd NameError: name 'bcd' is not defined 

[E-Paine]

MacOS failure is an unrelated issue with test_ssl.

[terryjreedy]

Sorry, I meant no more changes until I posted my edits, to avoid any merge conflicts. Once I did, I did not merge immediately to allow for further review and possible change. Good thing.

In your bad example above, should not the first bad name have been 'abc'? Was 'bcd' a result of fetching sys.exec info again, in get_message_lines, during processing? In any case, with the last patch:

try: abc except NameError: ing Traceback (most recent call last): File "<pyshell#2>", line 1, in <module> try: abc NameError: name 'abc' is not defined. Did you mean: 'abs'? During handling of the above exception, another exception occurred: Traceback (most recent call last): File "<pyshell#2>", line 2, in <module> except NameError: ing NameError: name 'ing' is not defined. Did you mean: 'int'? 

Both bad names are listed and corrected. Great.

I am impressed by and like the new test, but want to try to improve it and the previous one further. So no changes while I do ;-). (I am assuming that you are not editing right now.)

[terryjreedy]

I did edits in comprehensible chunks, testing after each. I will again wait a bit for you to check before merging.

[E-Paine]

[from the news] Include parser's typo fix suggestions

Just checking: is it the parser?

[miss-islington]

Thanks @E-Paine for the PR, and @terryjreedy for merging it 🌮🎉.. I'm working now to backport this PR to: 3.10, 3.9.
🐍🍒⛏🤖

[miss-islington]

Sorry, @E-Paine and @terryjreedy, I could not cleanly backport this to 3.9 due to a conflict.
Please backport using cherry_picker on command line.
cherry_picker 092f9ddb5e85665552c8207972cd393d492f764e 3.9

[bedevere-bot]

GH-25973 is a backport of this pull request to the 3.10 branch.

[bedevere-bot]

GH-25975 is a backport of this pull request to the 3.9 branch.

[terryjreedy]

@E-Paine The conflict was in the idlelib NEWS file. I fixed it and then remembered that this was not be be backported to 3.9 because it does not have the hints.

[terryjreedy]

Buildbot failure due to test asyncio deleted.