vest/mr_vest_reduce.cc

   1 #include <sstream>
   2 #include <iostream>
   3 #include <fstream>
   4 #include <vector>
   5
   6 #include <boost/program_options.hpp>
   7 #include <boost/program_options/variables_map.hpp>
   8
   9 #include "sparse_vector.h"
  10 #include "error_surface.h"
  11 #include "line_optimizer.h"
  12 #include "hg_io.h"
  13
  14 using namespace std;
  15 namespace po = boost::program_options;
  16
  17 void InitCommandLine(int argc, char** argv, po::variables_map* conf) {
  18   po::options_description opts("Configuration options");
  19   opts.add_options()
  20         ("loss_function,l",po::value<string>(), "Loss function being optimized")
  21         ("help,h", "Help");
  22   po::options_description dcmdline_options;
  23   dcmdline_options.add(opts);
  24   po::store(parse_command_line(argc, argv, dcmdline_options), *conf);
  25   bool flag = conf->count("loss_function") == 0;
  26   if (flag || conf->count("help")) {
  27     cerr << dcmdline_options << endl;
  28     exit(1);
  29   }
  30 }
  31
  32 int main(int argc, char** argv) {
  33   po::variables_map conf;
  34   InitCommandLine(argc, argv, &conf);
  35   const string loss_function = conf["loss_function"].as<string>();
  36   ScoreType type = ScoreTypeFromString(loss_function);
  37   LineOptimizer::ScoreType opt_type = LineOptimizer::MAXIMIZE_SCORE;
  38   if (type == TER || type == AER) {
  39     opt_type = LineOptimizer::MINIMIZE_SCORE;
  40   }
  41   string last_key;
  42   vector<ErrorSurface> esv;
  43   while(cin) {
  44     string line;
  45     getline(cin, line);
  46     if (line.empty()) continue;
  47     size_t ks = line.find("\t");
  48     assert(string::npos != ks);
  49     assert(ks > 2);
  50     string key = line.substr(2, ks - 2);
  51     string val = line.substr(ks + 1);
  52     if (key != last_key) {
  53       if (!last_key.empty()) {
  54         float score;
  55         double x = LineOptimizer::LineOptimize(esv, opt_type, &score);
  56         cout << last_key << "|" << x << "|" << score << endl;
  57       }
  58       last_key = key;
  59       esv.clear();
  60     }
  61     if (val.size() % 4 != 0) {
  62       cerr << "B64 encoding error 1! Skipping.\n";
  63       continue;
  64     }
  65     string encoded(val.size() / 4 * 3, '\0');
  66     if (!B64::b64decode(reinterpret_cast<const unsigned char*>(&val[0]), val.size(), &encoded[0], encoded.size())) {
  67       cerr << "B64 encoding error 2! Skipping.\n";
  68       continue;
  69     }
  70     esv.push_back(ErrorSurface());
  71     esv.back().Deserialize(type, encoded);
  72   }
  73   if (!esv.empty()) {
  74     // cerr << "ESV=" << esv.size() << endl;
  75     // for (int i = 0; i < esv.size(); ++i) { cerr << esv[i].size() << endl; }
  76     float score;
  77     double x = LineOptimizer::LineOptimize(esv, opt_type, &score);
  78     cout << last_key << "|" << x << "|" << score << endl;
  79   }
  80   return 0;
  81 }