Unless I have made an error,
Code:
char *this_field = aList, *next_field, *cp;
while (*this_field) // For each field in aList.
{
// To avoid the need to constantly check for buffer overflow (i.e. to keep it simple),
// just copy up to the limit of the buffer:
strlcpy(buf, this_field, sizeof(buf));
// Find the end of the field inside buf. In keeping with the tradition set by the Input command,
// this always uses comma rather than g_delimiter.
for (cp = buf, next_field = this_field; *cp; ++cp, ++next_field)
{
if (*cp == ',')
{
if (cp[1] == ',') // Make this pair into a single literal comma.
{
memmove(cp, cp + 1, strlen(cp + 1) + 1); // +1 to include the zero terminator.
++next_field; // An extra increment since the source string still has both commas of the pair.
}
else // this comma marks the end of the field.
{
*cp = '\0'; // Terminate the buffer to isolate just the current field.
break;
}
}
}
if (*next_field) // The end of the field occurred prior to the end of aList.
++next_field; // Point it to the character after the delimiter (otherwise, leave it where it is).
may be replaced with
Code:
char *this_field = aList, *next_field, *cp, *end_buf = buf + sizeof(buf) - 1;
while (*this_field) // For each field in aList.
{
for (cp = buf, next_field = this_field; cp < end_buf && *next_field; ++cp, ++next_field)
{
if (*next_field == ',') // Check if this is a delimiter or literal comma.
{
++next_field;
if (*next_field != ',') // Was it a delimiter?
break;
// Otherwise next_field now points at the second comma, so continue copying.
}
*cp = *next_field;
}
// Terminate the string in the buffer.
*cp = '\0';
to the same effect, but considerably faster in most cases. Actually, with fields of approximately 16300 characters, both if-in methods were roughly the same, and both were faster than Loop,Parse. I think short fields are probably more common.
Login
Register
FAQ
Search
