Tweak terminal colour handling.

[hjoliver]

Follow-up to #4562

• strip colorama markup from command help if not needed
• strip hardwired ansimarkup tags from command help if not needed (cylc scan --help)

Requirements check-list

• I have read CONTRIBUTING.md and added my name as a Code Contributor.
• Contains logically grouped changes (else tidy your branch by rebase).
• Does not contain off-topic changes (use other PRs for other changes).
• Applied any dependency changes to both setup.cfg and conda-environment.yml.
• Appropriate tests are included (unit and/or functional).
• Appropriate change log entry included.
• No documentation update required.

[hjoliver]

Note, it's kind of annoying that we have to treat command help separately for color switching (because it is printed by the option parser during option parsing) but I don't think it matters much. (There are more important things to think about right now).

[hjoliver]

(Assigned reviewers from the precursor PR)

[hjoliver]

@oliver-sanders - I've gone with a hybrid approach: a help formatter that uses my crude sys.argv parsing to get at the color options, because as I noted above help formatters do not have access to parsed option values. (I suppose that makes sense: help is often printed in response to errors on the command line).

This is not really any different than what I had before, except that the help formatter is formally nicer than overriding parser.print_help() 😁 However, the internal logic is robust enough (not fragile) it seems to me.

[oliver-sanders]

help formatters do not have access to parsed option values

That didn't make a lick of sense to me since the CLI args have to be parsed in order for the parser to activate the help formatter.

I think I've found the problem, if the --help argument was used any other args were stripped from the invoked command causing the strange dual state where the args passed to the parser did not match the args in sys.argv.

I think this change should fix the problem making args available to the help formatter.

diff --git a/cylc/flow/scripts/cylc.py b/cylc/flow/scripts/cylc.py
index 036cf4c05..6fa753315 100644
--- a/cylc/flow/scripts/cylc.py
+++ b/cylc/flow/scripts/cylc.py
@@ -532,7 +532,7 @@ def main():
                 # check if this is a command abbreviation or exit
                 command = match_command(command)
             if opts.help_:
-                execute_cmd(command, "--help")
+                execute_cmd(command, *cmd_args, "--help")
             else:
                 if opts.version:
                     cmd_args.append("--version")

MB!

[hjoliver]

I think I've found the problem, if the --help argument was used any other args were stripped from the invoked command causing the strange dual state where the args passed to the parser did not match the args in sys.argv.

OK, brilliant, well done 👍

[hjoliver]

Over-confident declaration of rightness retracted on Element chat!

[hjoliver]

I came back to this late in the day (out of time) I'm not sure why the new unit tests are failing. Monkeypatching of sys.argv appears to have worked (e.g. put print(sys.argv) in there) but is not affecting command parsing in the tests. I must have something wrong... [Fixed]

[oliver-sanders]

Tests timed out, poking them to see if it is reproducible.

[oliver-sanders]

... verily verily, not sure why.

[oliver-sanders]

Lets see if a merge sorts things out...

[hjoliver]

Minor tweak, should be good to go now.

[oliver-sanders]

🌈

[oliver-sanders]

(maybe squash)