example.cpp

#include <iostream>
#include <future>
#include <thread>
#include <algorithm>
#include <cstdlib>
#include <sstream>
#include <mutex>
#include <map>
#include <string>

#define BOOST_FILESYSTEM_NO_DEPRECATED
#define BOOST_FILESYSTEM_VERSION 3
#include <boost/filesystem/operations.hpp>
#include <boost/filesystem/fstream.hpp>

namespace fs = boost::filesystem;

std::mutex m;
typedef std::pair<uint32_t, fs::path> hash_pair ;
//typedef std::pair<uint32_t, std::string> hash_pair ;

void get_hashes(
    fs::path &root,
    std::vector< std::future<hash_pair> > &results
) {
  const fs::recursive_directory_iterator end;

  std::for_each(
    fs::recursive_directory_iterator(root),
    end,
    [&results](const fs::directory_entry &entry) {
      results.push_back(async(std::launch::async, [&entry]() -> hash_pair {
        hash_pair hp;

        m.lock();
        std::cout << "Starting new thread: " << std::this_thread::get_id() << std::endl;
        m.unlock();
        sleep(5);
        // the second value for the pair ends up being trash or empty strings.
        hp = std::make_pair(rand(), entry.path());
        // below works fine
        //hp = std::make_pair(rand(), std::string{"a string value"});
        return hp;
      }));
  });
}

int main()
{
  std::mutex m;
  std::vector<std::future<hash_pair>> futures;
  std::map<uint32_t, fs::path> hashes;
  fs::path root = "./data";
  get_hashes(root, futures);

  for (auto &result : futures) {
    auto pair = result.get();
    hashes[pair.first] = pair.second;
  }

  for (auto &result : hashes) {
    std::cout << result.first << " : " << result.second << std::endl;
  }

  std::cout << "number of hashes: " << hashes.size() << std::endl;
  return 0;
}