Re: [Boost-users] Using Boost string library

From: Pavol Droba (droba_at_[hidden])
Date: 2007-02-18 11:33:14

Hi,

Meryl Silverburgh wrote:
> Thanks for your idea.
> 
> I have cases like this:
> 
> /1/1
> /1/2/3/2
> 
> Since the first string /1/1 is smaller than the /1/2 (1 is < 2) of the
> /1/2/3/2, I should able to stop comparison by looking at the 4th
> character (/1/2) of '/1/2/3/2'.
> 
> So for cases like that, I would like to stop as soon as i can tell one
> string is shorter than the other, instead of splitting the whole
> string.
> 
> 
> 

This is exactly what would happen you use the second approach I gave you.
Constructing the iterator_range from the split iterator does not perform
any actual searching, and lexicographical_compare stops on the first
possible index (i.e on the first position where inputs are different).

>>
>> You can even throw away the vector, and directly create an
>> iterator_range from the split_iterator. This way you may spare some
>> unnecessary tokenization.
>>
>> typedef split_iterator<string::const_iterator> string_split;
>> iterator_range<string_split> r1(
>>         make_split_iterator(s1, is_any_of("/")), string_split())
>>
>> iterator_range<string_split> r2(
>>         make_split_iterator(s2, is_any_of("/")), string_split())
>>
>> if(lexicographical_compare(r1, r2, integer_compare))
>> {
>>         // s1 is less
>> }

Regards,
Pavol