Performance enhancement calling expand_macros()

Hello,
I have a (very simple) proposal to optimize a portion of the code that is used more than 40 times in the project:

msc_string *str = (msc_string *)apr_pcalloc(msr->mp, sizeof(msc_string));
str->value = (char *)input_string;
str->value_len = strlen(str->value);
expand_macros(msr, str, rule, msr->mp);
output_string = str->value;

We could optimize that code by first checking that input_string contains a '%' (or even "%{"). If not, no need to allocate the memory, call the function, etc. We add a (very) little overhead for the check, but avoid a much bigger overhead in most of the cases where the value doesn't contain a macro.
That could be done in a wrapper function (the name could be adapted), that would simplify the code as well:

char* expand_macros_wrapper(modsec_rec *msr, char *var, msre_rule *rule, apr_pool_t *mptmp) {
 if (!strstr(var, "%{")) return var;
 msc_string *str = (msc_string *)apr_pcalloc(msr->mp, sizeof(msc_string));
 str->value = var;
 str->value_len = strlen(str->value);
 int changed = expand_macros(msr, str, rule, msr->mp);
if (!changed) return var; // Does this line add anything? Should we always return str->value?
 return str->value;
}

The calling code would simply be:
output_string = expand_macros_wrapper(msr, input_string, rule, msr->mp);

Any remark?

Start by locating expand_macros() and its roughly 40 call sites; inspect how each prepares msc_string and handles the returned value. Compare the proposed wrapper's behavior with current macro expansion, then verify that unchanged inputs avoid allocation and expansion without altering outputs.