If you want more consistent results using explode() and implode():
(exploded strings consisting of only separators, or having them in the
beginning and at the end doesn't give a consistent "backwards" compatible
result with implode()..)

NOTE: THIS CODE IS AS-IS, as it is found in the MudOSa4DGD library.

/* big_explode(str, separator)
   This function works like explode(), except that it also counts
   separators at the start and end of the string. Basically that means
   that the array returned is more practical to parse. And big_implode()
   will return the original string if needed be.
*/
static string *big_explode(string str, string sep) {
  int stlen, seplen;
  string *expl;
  expl = ::explode(str, sep);
  if ((stlen=strlen(str)) > (seplen=strlen(sep)) && seplen > 0) {
    if (str[..seplen-1] == sep)
      /* matching separator found at beginning */
      expl = ({ "" }) + expl;
    if (str[stlen-seplen..] == sep && ::sizeof(expl - ({ "" })))
      /* matching separator found at end */
      expl += ({ "" });
  }
  /* got correct array now, return */
  return expl;
}

/* big_implode(str, separator)
   fully implodes an array of strings, regaining the string from
   big_explode(), unlike the relation between explode() and implode()
   of certain strings.
*/
static string big_implode(string *arr, string sep) {
  if (!::sizeof(arr - ({ "" }))) {
    /* special array consisting only of separators, add one extra */
    arr += ({ "" });
  }
  return ::implode(arr, sep);
}