regex - How do you ignore but preserve ANSI escape codes with sed? -

if print string "hello, \e[38;5;200mworld\e[0m", might see "hello, world" word "world" highlighted strong magenta color (ff00df) assuming terminal supports it.

however, if pipe output of program tree column, you'll see alignment broken, since escape codes incorrectly counted in line length calculations.

as more direct example, tried split string characters in sed using following:

sed 's/\(.\)/\1 /g'

...which yields...

h e l l o , 3 8 ; 5 ; 2 0 0 m w o r l d 0 m

... when piped string shown earlier.

how go matching escape codes purposes of counting, replacing, preserving, etc. while editing stream?

that not exactly, looking for, can develop idea further:

you can remove control characters using col:

$ col -bx 

or using sed:

$ sed -r "s/\x1b\[([0-9]{1,2}(;[0-9]{1,2})?)?[m|k]//g" 

you can use regular expression extract sequences should not counted or processed other way.

another interesting project, can useful ansifilter.