I found a pretty low tech solution to avoid the "gotcha" without worrying about the array order of how things are replaced. I could not "order" the replacement array easily because it was being read from a database table.
Anyway if you add an identifiable token to each replaced word, then just filter this out at the very end, no nested search terms are found. I just dynamically add the %% after the first char of each word before pumping it into the str_ireplace function.
$find = array("as1", "as2", "as3", "flex");
$replace = array("<a href = \"#as1\">A%%uto S%%entry R%%ev. A%%</a>", "<a href = \"#as2\">A%%uto S%%entry E%%xp</a>", "<a href = \"#as3\">A%%uto S%%entry f%%lex</a>", "<a style = \"color: red;\" href = \"#flex\">f%%lex</a>");
$text = str_ireplace($find, $replace, $text);
echo str_ireplace("%%", "", $text);
In this case I am using %% as my token as this is an unlikely char combo for me.
