You might want to check out and extend the piece of code I've included.
It's purpose is to setup a grammar to break a string and replace
color codes by the sequences (tune the get_colors to return sequences,
switch is fastest way to do this inside it). parse_colors() returns an
array where the string is split by '*' or *color-code*, the color codes
are replaced in get_colors function. By imploding the returned array to a
string with "" delimiter you get a colored string. 

example:
--
compile color.c
code return $1->parse_string("zyx**yellow* \n  *   *green*");
$1 = ({ "zyx","*","COLOR-*yellow*"," \n  ","*","   ","COLOR-*green*" })
--

To optimize your color code further, you might want to use some less often
separator for marking colors, maybe @^ or something like that.

If you wish, it's also easy to split the lines by \n in the same
grammar.

Notice that the grammar is compiled on first time the code is called, only
one grammar can be simultanously compiled on one object. So if you start
using parse_string, you should have one parse_string grammar per object to
keep it fast.

--- cuticuti, code here ---
 
string *get_color(string *s)
{
    switch(s[0]) {
    case "*yellow*":
    case "*red*":
    case "*black*":
    case "*green*":
    case "*blue*":
    case "*magenta*":
    case "*cyan*":
    case "*gray*":
    case "*orange*":
	return ({ "COLOR-"+s[0] });
	break;
    default:
	error("This code should not be reached");
	break;
    }
}

string *parse_colors(string input)
{
    return parse_string("\
anystr = /[^*]+/ \
"+"\
prod: sub_color               \
prod: anystr                  \
prod: '*'                     \
prod: prod prod               \
"+"\
color: '*yellow*'   \
color: '*red*'      \
color: '*black*'    \
color: '*green*'    \
color: '*blue*'     \
color: '*magenta*'  \
color: '*cyan*'     \
color: '*gray*'     \
color: '*orange*'   \
"+"\
sub_color: color ? get_color \
",input,0);
}

--- cuticuti ---