Functional test of colour output

[MetRonnie]

There is a problem with this test

cylc-flow/tests/functional/cli/05-colour.t

Lines 34 to 35 in ca454d0

	script -q -c "cylc scan -t rich --color=never" log > /dev/null 2>&1
	grep_fail "$ANSI" log -P # no color

A color reset char is included on the last line of the output:

Script started on 2024-11-07 18:11:50+00:00 [TERM="xterm-256color" TTY="/dev/pts/1" COLUMNS="202" LINES="13"]
�[0m
Script done on 2024-11-07 18:11:53+00:00 [COMMAND_EXIT_CODE="0"]

This is despite autoreset=False and use_color=False (strip=True) here:

cylc-flow/cylc/flow/terminal.py

Line 292 in ca454d0

color_init(autoreset=False, strip=not use_color)

Whether this is just some strange interaction between colorama and the script command used in the test, or a bug in colorama, I'm not sure.

---

N.B. I had a look at how colorama works. It bills itself as a package meant only for making colour work on Windows, but I guess we are using it for its init(strip=not use_color) function that can automatically strip ANSI color chars from print() calls (it actually patches sys.stdout and sys.stderr). However there are >100 open issues and it is in a call-for-maintainers state: tartley/colorama#300.

Originally posted by @MetRonnie in #6463 (comment)