IDLE - colorize `type` as a soft keyword when it begins a statememt

[Locked-chess-official]

Bug report

Bug description:

The type was colorized as a built-in variable instead of keyword. However, type here is a keyword.

CPython versions tested on:

CPython main branch

Operating systems tested on:

No response

[StanFromIreland]

Marking as easy, it's not quite trivial but I think a reasonable start. It will require writing a regex for the soft keyword, but it's not a terribly complicated one.

Maybe @robsdedude is interested in this?

[robsdedude]

Thanks for the ping. I'm travelling this week without a laptop. If the issue still open next week I'd love to work on it!

[omkar-334]

hi, can I work on this

[StanFromIreland]

Hi @omkar-334, I'm sorry but this issue already has someone assigned.

[terryjreedy]

I missed this syntax addition in 3.12. Soft keywords can sometimes be a bit nasty. All keywords and builtins, including new ones, are automatically recognized, whereas each soft keyword needs custom additions to both colorizer.py and test_colorizer.py*. A test that would currently fail in test_colorizer.ColorDelegatorTest is that the class has a 'test_soft...' test for each soft keyword other than '' in keyword.softkwlist. (EDIT: '' is included with 'case _')

• Each file needs additions in about 3 places. I should add a comment to each. 'lazy' needs to be added to 'source' in the test.

The statement syntax for 'type' is most similar to that of 'lazy'. The re for identifiers is currently buried in the idprog definition, but should be separately defined before make_pat.

@robsdedude If you want comments on an incomplete patch, post it and mark it as a draft.

[terryjreedy]

After reading #151390, I discovered that IDLE colorizes beginning-of-line 'match' and only uncolorizes it when followed by something identified as 'wrong', like a keyword or certain chars such as =. REPL does not colorize it until followed by something right. Either could be considered correct; I prefer the immediate coloring of initial 'match' as 'likely a keyword'. But IDLE does not always uncolorize when it should. For instance, match + remains colored because + could be valid, as in match +1. But IDLE stops there and also does not uncolorize on +=. case in handled similarly differently.

Like the REPL, IDLE does not colorize 'lazy' until confirmed as a keyword by 'import' or 'from'. type <identifier> should do the same, although the simple re for identifier is only approximate.