Question

I would like to sort alphanumeric strings the way a human being would sort them. I.e., “A2” comes before “A10”, and “a” certainly comes before “Z”! Is there any way to do with without writing a mini-parser? Ideally it would also put “A1B1” before “A1B10”. I see the question “Natural (human alpha-numeric) sort in Microsoft SQL 2005” with a possible answer, but it uses various library functions, as does “Sorting Strings for Humans with IComparer”.

Below is a test case that currently fails:

#include <set>
#include <iterator>
#include <iostream>
#include <vector>
#include <cassert>

template <typename T>
struct LexicographicSort {
  inline bool operator() (const T& lhs, const T& rhs) const{
    std::ostringstream s1,s2;
    s1 << toLower(lhs); s2 << toLower(rhs);
    bool less = s1.str() < s2.str();
    //Answer: bool less = doj::alphanum_less<std::string>()(s1.str(), s2.str());
    std::cout<<s1.str()<<" "<<s2.str()<<" "<<less<<"\n";
    return less;
  }

  inline std::string toLower(const std::string& str) const {
    std::string newString("");
    for (std::string::const_iterator charIt = str.begin();
         charIt!=str.end();++charIt) {
          newString.push_back(std::tolower(*charIt));
        }
        return newString;
      }
};


int main(void) {
  const std::string reference[5] = {"ab","B","c1","c2","c10"};
  std::vector<std::string> referenceStrings(&(reference[0]), &(reference[5]));

  //Insert in reverse order so we know they get sorted
  std::set<std::string,LexicographicSort<std::string> > strings(referenceStrings.rbegin(), referenceStrings.rend());

  std::cout<<"Items:\n";
  std::copy(strings.begin(), strings.end(), std::ostream_iterator<std::string>(std::cout, "\n"));
  std::vector<std::string> sortedStrings(strings.begin(), strings.end());
  assert(sortedStrings == referenceStrings);
}

Answer 1

Is there any way to do with without writing a mini-parser?

Let someone else do that?

I’m using this implementation: http://www.davekoelle.com/alphanum.html, I’ve modified it to support wchar_t, too.