Problem in Adding char* about cuckoofilter

[dnbaker]

Here's the patch I've made. It compiles -- I'm not sure if it fixes the problem, as I haven't built any test cases. Want to give it a whirl, srbehera?

[chialunwu]

The reason you cannot find the existing items is because you have to clear the char array to make sure the chars behind '\0' are the same.
ex:
char[31] str;
bzero(str, 31);

[srbehera]

It is still not working !
This is my code

CuckooFilter<string, 12> filter(total_items);
string line;
while(getline(inp, line))
{
    if(filter.Contain(line) == cuckoofilter::Ok)
        cout<<"Found:"<<line<<endl;
    else
        filter.Add(line);
}

[chialunwu]

I think no char* and string now. It needs primative and non-pointer type because there is sizeof(item) in the code. So sizeof(string) will go wrong.

[apc999]

@bingo4508 you are right. currently the way to calculate hash index is here .
As a result, the string is converted into char*, getting into problems.

@srbehera for now string is not supported but only primary types (where sizeof(item) makes sense). I planed to support string but still not able to find time doing that :(

[srbehera]

@bingo4508 @apc999 Got it !! Thanks !!

[dnbaker]

Would a templated function that defaulted to sizeof() for other structs/types with a specialization for std::string (using std::string.c_str() rather than (const char *) and std::string.size() rather than sizeof) be a quick fix?

[apc999]

yes, I think so.
Deaniel, feel free to submit a patch.
otherwise, I will probably add one later .

best,

On Wed, Apr 27, 2016 at 5:31 PM, Daniel Baker [email protected]
wrote:

Would a templated function that defaulted to sizeof() for other
structs/types with a specialization for std::string (using
std::string.c_str() rather than (const char *) and std::string.size()
rather than sizeof) be a quick fix?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#2 (comment)

[srbehera11]

Daniel, it is not working for string.
Here is the simple code I was testing

int main(int argc, char** argv) {
    CuckooFilter<string, 12> filter2(total_items);
    srand (time(NULL));
    ifstream inp;
    inp.open("test.txt");
    string line;
    int c1 = 0;
    int i = 0;
    while(getline(inp, line))
    {
      cout<<line<<endl;
      if(filter2.Contain(line) == cuckoofilter::Ok){
        cout<<"Found:"<<line<<endl;
        c++;
      }
      else{
        filter2.Add(line);
      }
      i++;
    }
    cout<< "c = "<< c <<endl;
    cout<< "i = "<< i <<endl;
    inp.close();

Where test.txt has following entries

AAA
BBB
CCC
AAA
CCC
DDD
XYZ
CCC
XYZ
AAA

[dnbaker]

Is there any output? What does it do?

[srbehera11]

The program was putting the strings in the filter. If it is already there, it was printing "Found". The following output shows that it only puts the first string and then says "Found" all other strings.

AAA
BBB
Found:BBB
CCC
Found:CCC
AAA
Found:AAA
CCC
Found:CCC
DDD
Found:DDD
XYZ
Found:XYZ
CCC
Found:CCC
XYZ
Found:XYZ
AAA
Found:AAA
c1=9
i=10

[siara-cc]

I faced the same issue and it looks like according to the implementation the first parameter can only be uint64_t.
So I used following code:

std::string mystring = "Hello World";
cuckoofilter::CuckooFilter<size_t, 12> test(1000);
test.Add(CityHash64(mystring.c_str(), mystring.length()));

CitiHash64 can be found in src/City.h in https://github.com/aappleby/smhasher