Have colors in "ls", depending on file status

[holzschu]

Builiding on release #71 by @IMcD23 , this PR activates "color mode" for ls: directories are in blue, executable files in red, etc. Colors can be configured by setting the environment variable LSCOLORS. "color mode" can be deactivated by unsetting the variable CLICOLOR.

Color mode is automatically deactivated when the output of ls is sent to another command (e.g. ls | grep)

[ian-mcdowell]

Why are there so many commits in this PR?

[ian-mcdowell]

The colors are a bit hard to read. How does ‘ls’ determine which colors to show?

[holzschu]

From what I gathered, foreground color is set from LSCOLORS, by pairs of letters.
First letter = foreground color, color number = letter minus 'a'. Bold if capital letter.
2nd letter = background color, "x" if not changed, color number = letter minus 'a'
In that example, LSCOLORS = "ExFxBxDxCxegedabagacad".
First color = directory, 2nd color = symbolic link, 4th = executable (see enum Colors in ls/print.c).

So that would be directories = color #4 + bold, links = color #5 + bold, executables = color #3 + bold (so the color of executables is set by \033[33m and \033[1m). Inside Blink, with the same LSCOLORS (and the same binaries), I get different colors:

[ian-mcdowell]

I’ll take a look at the color handling in ANSITextState.swift

Perhaps those colors need to be adjusted...

[holzschu]

I think I should also check where the "bold" escape sequence is placed.

[ian-mcdowell]

I want to start building some test cases for these kind of things. Could you send me a sample color output string, with escape characters?

[holzschu]

Here is the output for the same directory, before parsing:

cacert.pem    ctd.cpp    echoTest    input    known_hosts
^[[1m^[[34mlua^[[39;49m^[[0m
path
^[[1m^[[32mtest^[[39;49m^[[0m
test.tar.gz
^[[1m^[[35mtest2^[[39;49m^[[0m

Where ^[[ is the escape sequence (esc + [). According to https://en.wikipedia.org/wiki/ANSI_escape_code , that should be lua in bold blue, test in bold green and test2 in bold magenta.

Raw version:
ls_result.txt

[louisdh]

This PR only changes code in ios_system, and was opened before ios_system was included as a git submodule. I suppose this PR can be closed now? Assuming these changes are available on the master branch of the ios_system repo.

[holzschu]

Yes, it can be closed.