Blame tools/marisa-benchmark.cc

Packit de3218
#include <cstdlib>
Packit de3218
#include <cstring>
Packit de3218
#include <ctime>
Packit de3218
#include <fstream>
Packit de3218
#include <iostream>
Packit de3218
#include <string>
Packit de3218
#include <vector>
Packit de3218
Packit de3218
#include <marisa.h>
Packit de3218
Packit de3218
#include "cmdopt.h"
Packit de3218
Packit de3218
namespace {
Packit de3218
Packit de3218
int param_min_num_tries = 1;
Packit de3218
int param_max_num_tries = 5;
Packit de3218
marisa::TailMode param_tail_mode = MARISA_DEFAULT_TAIL;
Packit de3218
marisa::NodeOrder param_node_order = MARISA_DEFAULT_ORDER;
Packit de3218
marisa::CacheLevel param_cache_level = MARISA_DEFAULT_CACHE;
Packit de3218
bool param_with_predict = true;
Packit de3218
bool param_print_speed = true;
Packit de3218
Packit de3218
class Clock {
Packit de3218
 public:
Packit de3218
  Clock() : cl_(std::clock()) {}
Packit de3218
Packit de3218
  void reset() {
Packit de3218
    cl_ = std::clock();
Packit de3218
  }
Packit de3218
Packit de3218
  double elasped() const {
Packit de3218
    std::clock_t cur = std::clock();
Packit de3218
    return 1.0 * (cur - cl_) / CLOCKS_PER_SEC;
Packit de3218
  }
Packit de3218
Packit de3218
 private:
Packit de3218
  std::clock_t cl_;
Packit de3218
};
Packit de3218
Packit de3218
void print_help(const char *cmd) {
Packit de3218
  std::cerr << "Usage: " << cmd << " [OPTION]... [FILE]...\n\n"
Packit de3218
      "Options:\n"
Packit de3218
      "  -N, --min-num-tries=[N]  limit the number of tries"
Packit de3218
      " [" << MARISA_MIN_NUM_TRIES << ", " << MARISA_MAX_NUM_TRIES
Packit de3218
      << "] (default: 1)\n"
Packit de3218
      "  -n, --max-num-tries=[N]  limit the number of tries"
Packit de3218
      " [" << MARISA_MIN_NUM_TRIES << ", " << MARISA_MAX_NUM_TRIES
Packit de3218
      << "] (default: 10)\n"
Packit de3218
      "  -t, --text-tail      build a dictionary with text TAIL (default)\n"
Packit de3218
      "  -b, --binary-tail    build a dictionary with binary TAIL\n"
Packit de3218
      "  -w, --weight-order   arrange siblings in weight order (default)\n"
Packit de3218
      "  -l, --label-order    arrange siblings in label order\n"
Packit de3218
      "  -c, --cache-level=[N]    specify the cache size"
Packit de3218
      " [1, 5] (default: 3)\n"
Packit de3218
      "  -P, --with-predict       include predictive search (default)\n"
Packit de3218
      "  -p, --without-predict    skip predictive search\n"
Packit de3218
      "  -S, --print-speed    print speed [1000 keys/s] (default)\n"
Packit de3218
      "  -s, --print-time     print time [ns/key]\n"
Packit de3218
      "  -h, --help           print this help\n"
Packit de3218
      << std::endl;
Packit de3218
}
Packit de3218
Packit de3218
void print_config() {
Packit de3218
  std::cout << "Number of tries: " << param_min_num_tries
Packit de3218
      << " - " << param_max_num_tries << std::endl;
Packit de3218
Packit de3218
  std::cout << "TAIL mode: ";
Packit de3218
  switch (param_tail_mode) {
Packit de3218
    case MARISA_TEXT_TAIL: {
Packit de3218
      std::cout << "Text mode" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
    case MARISA_BINARY_TAIL: {
Packit de3218
      std::cout << "Binary mode" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
  }
Packit de3218
Packit de3218
  std::cout << "Node order: ";
Packit de3218
  switch (param_node_order) {
Packit de3218
    case MARISA_LABEL_ORDER: {
Packit de3218
      std::cout << "Ascending label order" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
    case MARISA_WEIGHT_ORDER: {
Packit de3218
      std::cout << "Descending weight order" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
  }
Packit de3218
Packit de3218
  std::cout << "Cache level: ";
Packit de3218
  switch (param_cache_level) {
Packit de3218
    case MARISA_HUGE_CACHE: {
Packit de3218
      std::cout << "Huge cache" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
    case MARISA_LARGE_CACHE: {
Packit de3218
      std::cout << "Large cache" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
    case MARISA_NORMAL_CACHE: {
Packit de3218
      std::cout << "Normal cache" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
    case MARISA_SMALL_CACHE: {
Packit de3218
      std::cout << "Small cache" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
    case MARISA_TINY_CACHE: {
Packit de3218
      std::cout << "Tiny cache" << std::endl;
Packit de3218
      break;
Packit de3218
    }
Packit de3218
  }
Packit de3218
}
Packit de3218
Packit de3218
void print_time_info(std::size_t num_keys, double elasped) {
Packit de3218
  if (param_print_speed) {
Packit de3218
    if (elasped == 0.0) {
Packit de3218
      std::printf(" %8s", "-");
Packit de3218
    } else {
Packit de3218
      std::printf(" %8.2f", num_keys / elasped / 1000.0);
Packit de3218
    }
Packit de3218
  } else {
Packit de3218
    if ((elasped == 0.0) || (num_keys == 0)) {
Packit de3218
      std::printf(" %8s", "-");
Packit de3218
    } else {
Packit de3218
      std::printf(" %8.1f", 1000000000.0 * elasped / num_keys);
Packit de3218
    }
Packit de3218
  }
Packit de3218
}
Packit de3218
Packit de3218
void read_keys(std::istream &input, marisa::Keyset *keyset,
Packit de3218
    std::vector<float> *weights) {
Packit de3218
  std::string line;
Packit de3218
  while (std::getline(input, line)) {
Packit de3218
    const std::string::size_type delim_pos = line.find_last_of('\t');
Packit de3218
    float weight = 1.0F;
Packit de3218
    if (delim_pos != line.npos) {
Packit de3218
      char *end_of_value;
Packit de3218
      weight = (float)std::strtod(&line[delim_pos + 1], &end_of_value);
Packit de3218
      if (*end_of_value == '\0') {
Packit de3218
        line.resize(delim_pos);
Packit de3218
      }
Packit de3218
    }
Packit de3218
    keyset->push_back(line.c_str(), line.length());
Packit de3218
    weights->push_back(weight);
Packit de3218
  }
Packit de3218
}
Packit de3218
Packit de3218
int read_keys(const char * const *args, std::size_t num_args,
Packit de3218
    marisa::Keyset *keyset, std::vector<float> *weights) {
Packit de3218
  if (num_args == 0) {
Packit de3218
    read_keys(std::cin, keyset, weights);
Packit de3218
  }
Packit de3218
  for (std::size_t i = 0; i < num_args; ++i) {
Packit de3218
    std::ifstream input_file(args[i], std::ios::binary);
Packit de3218
    if (!input_file) {
Packit de3218
      std::cerr << "error: failed to open: " << args[i] << std::endl;
Packit de3218
      return 10;
Packit de3218
    }
Packit de3218
    read_keys(input_file, keyset, weights);
Packit de3218
  }
Packit de3218
  std::cout << "Number of keys: " << keyset->size() << std::endl;
Packit de3218
  std::cout << "Total length: " << keyset->total_length() << std::endl;
Packit de3218
  return 0;
Packit de3218
}
Packit de3218
Packit de3218
void benchmark_build(marisa::Keyset &keyset,
Packit de3218
    const std::vector<float> &weights, int num_tries, marisa::Trie *trie) {
Packit de3218
  for (std::size_t i = 0; i < keyset.size(); ++i) {
Packit de3218
    keyset[i].set_weight(weights[i]);
Packit de3218
  }
Packit de3218
  Clock cl;
Packit de3218
  trie->build(keyset, num_tries | param_tail_mode | param_node_order |
Packit de3218
      param_cache_level);
Packit de3218
  std::printf(" %10lu", (unsigned long)trie->io_size());
Packit de3218
  print_time_info(keyset.size(), cl.elasped());
Packit de3218
}
Packit de3218
Packit de3218
void benchmark_lookup(const marisa::Trie &trie,
Packit de3218
    const marisa::Keyset &keyset) {
Packit de3218
  Clock cl;
Packit de3218
  marisa::Agent agent;
Packit de3218
  for (std::size_t i = 0; i < keyset.size(); ++i) {
Packit de3218
    agent.set_query(keyset[i].ptr(), keyset[i].length());
Packit de3218
    if (!trie.lookup(agent) || (agent.key().id() != keyset[i].id())) {
Packit de3218
      std::cerr << "error: lookup() failed" << std::endl;
Packit de3218
      return;
Packit de3218
    }
Packit de3218
  }
Packit de3218
  print_time_info(keyset.size(), cl.elasped());
Packit de3218
}
Packit de3218
Packit de3218
void benchmark_reverse_lookup(const marisa::Trie &trie,
Packit de3218
    const marisa::Keyset &keyset) {
Packit de3218
  Clock cl;
Packit de3218
  marisa::Agent agent;
Packit de3218
  for (std::size_t i = 0; i < keyset.size(); ++i) {
Packit de3218
    agent.set_query(keyset[i].id());
Packit de3218
    trie.reverse_lookup(agent);
Packit de3218
    if ((agent.key().id() != keyset[i].id()) ||
Packit de3218
        (agent.key().length() != keyset[i].length()) ||
Packit de3218
        (std::memcmp(agent.key().ptr(), keyset[i].ptr(),
Packit de3218
            agent.key().length()) != 0)) {
Packit de3218
      std::cerr << "error: reverse_lookup() failed" << std::endl;
Packit de3218
      return;
Packit de3218
    }
Packit de3218
  }
Packit de3218
  print_time_info(keyset.size(), cl.elasped());
Packit de3218
}
Packit de3218
Packit de3218
void benchmark_common_prefix_search(const marisa::Trie &trie,
Packit de3218
    const marisa::Keyset &keyset) {
Packit de3218
  Clock cl;
Packit de3218
  marisa::Agent agent;
Packit de3218
  for (std::size_t i = 0; i < keyset.size(); ++i) {
Packit de3218
    agent.set_query(keyset[i].ptr(), keyset[i].length());
Packit de3218
    while (trie.common_prefix_search(agent)) {
Packit de3218
      if (agent.key().id() > keyset[i].id()) {
Packit de3218
        std::cerr << "error: common_prefix_search() failed" << std::endl;
Packit de3218
        return;
Packit de3218
      }
Packit de3218
    }
Packit de3218
    if (agent.key().id() != keyset[i].id()) {
Packit de3218
      std::cerr << "error: common_prefix_search() failed" << std::endl;
Packit de3218
      return;
Packit de3218
    }
Packit de3218
  }
Packit de3218
  print_time_info(keyset.size(), cl.elasped());
Packit de3218
}
Packit de3218
Packit de3218
void benchmark_predictive_search(const marisa::Trie &trie,
Packit de3218
    const marisa::Keyset &keyset) {
Packit de3218
  if (!param_with_predict) {
Packit de3218
    print_time_info(keyset.size(), 0.0);
Packit de3218
    return;
Packit de3218
  }
Packit de3218
Packit de3218
  Clock cl;
Packit de3218
  marisa::Agent agent;
Packit de3218
  for (std::size_t i = 0; i < keyset.size(); ++i) {
Packit de3218
    agent.set_query(keyset[i].ptr(), keyset[i].length());
Packit de3218
    if (!trie.predictive_search(agent) ||
Packit de3218
        (agent.key().id() != keyset[i].id())) {
Packit de3218
      std::cerr << "error: predictive_search() failed" << std::endl;
Packit de3218
      return;
Packit de3218
    }
Packit de3218
    while (trie.predictive_search(agent)) {
Packit de3218
      if (agent.key().id() <= keyset[i].id()) {
Packit de3218
        std::cerr << "error: predictive_search() failed" << std::endl;
Packit de3218
        return;
Packit de3218
      }
Packit de3218
    }
Packit de3218
  }
Packit de3218
  print_time_info(keyset.size(), cl.elasped());
Packit de3218
}
Packit de3218
Packit de3218
void benchmark(marisa::Keyset &keyset, const std::vector<float> &weights,
Packit de3218
    int num_tries) {
Packit de3218
  std::printf("%6d", num_tries);
Packit de3218
  marisa::Trie trie;
Packit de3218
  benchmark_build(keyset, weights, num_tries, &trie);
Packit de3218
  if (!trie.empty()) {
Packit de3218
    benchmark_lookup(trie, keyset);
Packit de3218
    benchmark_reverse_lookup(trie, keyset);
Packit de3218
    benchmark_common_prefix_search(trie, keyset);
Packit de3218
    benchmark_predictive_search(trie, keyset);
Packit de3218
  }
Packit de3218
  std::printf("\n");
Packit de3218
}
Packit de3218
Packit de3218
int benchmark(const char * const *args, std::size_t num_args) try {
Packit de3218
  marisa::Keyset keyset;
Packit de3218
  std::vector<float> weights;
Packit de3218
  const int ret = read_keys(args, num_args, &keyset, &weights);
Packit de3218
  if (ret != 0) {
Packit de3218
    return ret;
Packit de3218
  }
Packit de3218
  std::printf("------+----------+--------+--------+"
Packit de3218
      "--------+--------+--------\n");
Packit de3218
  std::printf("%6s %10s %8s %8s %8s %8s %8s\n",
Packit de3218
      "#tries", "size", "build", "lookup", "reverse", "prefix", "predict");
Packit de3218
  std::printf("%6s %10s %8s %8s %8s %8s %8s\n",
Packit de3218
      "", "", "", "", "lookup", "search", "search");
Packit de3218
  if (param_print_speed) {
Packit de3218
    std::printf("%6s %10s %8s %8s %8s %8s %8s\n",
Packit de3218
        "", "[bytes]",
Packit de3218
        "[K/s]", "[K/s]", "[K/s]", "[K/s]", "[K/s]");
Packit de3218
  } else {
Packit de3218
    std::printf("%6s %10s %8s %8s %8s %8s %8s\n",
Packit de3218
        "", "[bytes]", "[ns]", "[ns]", "[ns]", "[ns]", "[ns]");
Packit de3218
  }
Packit de3218
  std::printf("------+----------+--------+--------+"
Packit de3218
      "--------+--------+--------\n");
Packit de3218
  for (int i = param_min_num_tries; i <= param_max_num_tries; ++i) {
Packit de3218
    benchmark(keyset, weights, i);
Packit de3218
  }
Packit de3218
  std::printf("------+----------+--------+--------+"
Packit de3218
      "--------+--------+--------\n");
Packit de3218
  return 0;
Packit de3218
} catch (const marisa::Exception &ex) {
Packit de3218
  std::cerr << ex.what() << std::endl;
Packit de3218
  return -1;
Packit de3218
}
Packit de3218
Packit de3218
}  // namespace
Packit de3218
Packit de3218
int main(int argc, char *argv[]) {
Packit de3218
  std::ios::sync_with_stdio(false);
Packit de3218
Packit de3218
  ::cmdopt_option long_options[] = {
Packit de3218
    { "min-num-tries", 1, NULL, 'N' },
Packit de3218
    { "max-num-tries", 1, NULL, 'n' },
Packit de3218
    { "text-tail", 0, NULL, 't' },
Packit de3218
    { "binary-tail", 0, NULL, 'b' },
Packit de3218
    { "weight-order", 0, NULL, 'w' },
Packit de3218
    { "label-order", 0, NULL, 'l' },
Packit de3218
    { "cache-level", 1, NULL, 'c' },
Packit de3218
    { "predict-on", 0, NULL, 'P' },
Packit de3218
    { "predict-off", 0, NULL, 'p' },
Packit de3218
    { "print-speed", 0, NULL, 'S' },
Packit de3218
    { "print-time", 0, NULL, 's' },
Packit de3218
    { "help", 0, NULL, 'h' },
Packit de3218
    { NULL, 0, NULL, 0 }
Packit de3218
  };
Packit de3218
  ::cmdopt_t cmdopt;
Packit de3218
  ::cmdopt_init(&cmdopt, argc, argv, "N:n:tbwlc:PpSsh", long_options);
Packit de3218
  int label;
Packit de3218
  while ((label = ::cmdopt_get(&cmdopt)) != -1) {
Packit de3218
    switch (label) {
Packit de3218
      case 'N': {
Packit de3218
        char *end_of_value;
Packit de3218
        const long value = std::strtol(cmdopt.optarg, &end_of_value, 10);
Packit de3218
        if ((*end_of_value != '\0') || (value <= 0) ||
Packit de3218
            (value > MARISA_MAX_NUM_TRIES)) {
Packit de3218
          std::cerr << "error: option `-n' with an invalid argument: "
Packit de3218
              << cmdopt.optarg << std::endl;
Packit de3218
          return 1;
Packit de3218
        }
Packit de3218
        param_min_num_tries = (int)value;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'n': {
Packit de3218
        char *end_of_value;
Packit de3218
        const long value = std::strtol(cmdopt.optarg, &end_of_value, 10);
Packit de3218
        if ((*end_of_value != '\0') || (value <= 0) ||
Packit de3218
            (value > MARISA_MAX_NUM_TRIES)) {
Packit de3218
          std::cerr << "error: option `-n' with an invalid argument: "
Packit de3218
              << cmdopt.optarg << std::endl;
Packit de3218
          return 2;
Packit de3218
        }
Packit de3218
        param_max_num_tries = (int)value;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 't': {
Packit de3218
        param_tail_mode = MARISA_TEXT_TAIL;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'b': {
Packit de3218
        param_tail_mode = MARISA_BINARY_TAIL;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'w': {
Packit de3218
        param_node_order = MARISA_WEIGHT_ORDER;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'l': {
Packit de3218
        param_node_order = MARISA_LABEL_ORDER;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'c': {
Packit de3218
        char *end_of_value;
Packit de3218
        const long value = std::strtol(cmdopt.optarg, &end_of_value, 10);
Packit de3218
        if ((*end_of_value != '\0') || (value < 1) || (value > 5)) {
Packit de3218
          std::cerr << "error: option `-c' with an invalid argument: "
Packit de3218
              << cmdopt.optarg << std::endl;
Packit de3218
          return 3;
Packit de3218
        } else if (value == 1) {
Packit de3218
          param_cache_level = MARISA_TINY_CACHE;
Packit de3218
        } else if (value == 2) {
Packit de3218
          param_cache_level = MARISA_SMALL_CACHE;
Packit de3218
        } else if (value == 3) {
Packit de3218
          param_cache_level = MARISA_NORMAL_CACHE;
Packit de3218
        } else if (value == 4) {
Packit de3218
          param_cache_level = MARISA_LARGE_CACHE;
Packit de3218
        } else if (value == 5) {
Packit de3218
          param_cache_level = MARISA_HUGE_CACHE;
Packit de3218
        }
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'P': {
Packit de3218
        param_with_predict = true;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'p': {
Packit de3218
        param_with_predict = false;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'S': {
Packit de3218
        param_print_speed = true;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 's': {
Packit de3218
        param_print_speed = false;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'h': {
Packit de3218
        print_help(argv[0]);
Packit de3218
        return 0;
Packit de3218
      }
Packit de3218
      default: {
Packit de3218
        return 1;
Packit de3218
      }
Packit de3218
    }
Packit de3218
  }
Packit de3218
  print_config();
Packit de3218
  return benchmark(cmdopt.argv + cmdopt.optind, cmdopt.argc - cmdopt.optind);
Packit de3218
}