On 02/13/2011 11:18 PM, Graham Leggett wrote:
> On 13 Feb 2011, at 5:22 PM, Ruediger Pluem wrote:
>
>>> + /* skip characters in sep (will terminate at '\0') */
>>> + while (*str && strchr(sep, *str)) {
>>> + ++str;
>>> + }
>>> +
>>> + if (!*str) { /* no more tokens */
>>> + return NULL;
>>> + }
>>> +
>>> + token = str;
>>> +
>>> + /* skip valid token characters to terminate token and
>>> + * prepare for the next call (will terminate at '\0)
>>> + * on the way, ignore all quoted strings, and within
>>> + * quoted strings, escaped characters.
>>> + */
>>> + *last = token + 1;
>>
>> What happens if str is supplied as "a, b"?
>> I mean why token + 1 and not token?
>
> I guess it's because we know *token isn't a separator, so there is no
> point checking if it is one a second time.
*token might not be a separator, but it might be ".
>
> The same pattern exists in apr_strtok.c:
>
> https://svn.apache.org/repos/asf/apr/apr/trunk/strings/apr_strtok.c
Maybe, but that doesn't mean that it is correct :-).
IMHO "a, b" would result in the two tokens
a
b"
which would be wrong. The result should be one token:
"a, b"
Regards
Rüdiger
