1b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <cstdlib> 2b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <ctime> 3b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <fstream> 4b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <iostream> 5b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <limits> 6b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <string> 7b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <utility> 8b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <vector> 9b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 10b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include <marisa_alpha.h> 11b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 12b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak#include "./cmdopt.h" 13b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 14b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniaknamespace { 15b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 16b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniaktypedef std::pair<std::string, double> Key; 17b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 18b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint param_min_num_tries = 1; 19b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint param_max_num_tries = 10; 20b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint param_trie = MARISA_ALPHA_DEFAULT_TRIE; 21b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint param_tail = MARISA_ALPHA_DEFAULT_TAIL; 22b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint param_order = MARISA_ALPHA_DEFAULT_ORDER; 23b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakbool predict_strs_flag = false; 24b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakbool speed_flag = true; 25b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 26b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakclass Clock { 27b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak public: 28b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Clock() : cl_(std::clock()) {} 29b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 30b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak void reset() { 31b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak cl_ = std::clock(); 32b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 33b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 34b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak double elasped() const { 35b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::clock_t cur = std::clock(); 36b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return (cur == cl_) ? 0.01 : (1.0 * (cur - cl_) / CLOCKS_PER_SEC); 37b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 38b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 39b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak private: 40b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::clock_t cl_; 41b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak}; 42b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 43b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid print_help(const char *cmd) { 44b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "Usage: " << cmd << " [OPTION]... [FILE]...\n\n" 45b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "Options:\n" 46b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -N, --min-num-tries=[N] limits the number of tries to N" 47b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " (default: 1)\n" 48b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -n, --max-num-tries=[N] limits the number of tries to N" 49b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " (default: 10)\n" 50b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -P, --patricia-trie build patricia tries (default)\n" 51b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -p, --prefix-trie build prefix tries\n" 52b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -T, --text-tail build a dictionary with text TAIL (default)\n" 53b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -b, --binary-tail build a dictionary with binary TAIL\n" 54b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -t, --without-tail build a dictionary without TAIL\n" 55b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -w, --weight-order arrange siblings in weight order (default)\n" 56b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -l, --label-order arrange siblings in label order\n" 57b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -I, --predict-ids get key IDs in predictive searches (default)\n" 58b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -i, --predict-strs restore key strings in predictive searches\n" 59b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -S, --print-speed print speed [1000 keys/s] (default)\n" 60b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -s, --print-time print time [us/key]\n" 61b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak " -h, --help print this help\n" 62b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak << std::endl; 63b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 64b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 65b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid print_config() { 66b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "#tries: " << param_min_num_tries 67b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak << " - " << param_max_num_tries << std::endl; 68b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 69b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak switch (param_trie) { 70b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case MARISA_ALPHA_PATRICIA_TRIE: { 71b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "trie: patricia" << std::endl; 72b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 73b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 74b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case MARISA_ALPHA_PREFIX_TRIE: { 75b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "trie: prefix" << std::endl; 76b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 77b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 78b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 79b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 80b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak switch (param_tail) { 81b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case MARISA_ALPHA_WITHOUT_TAIL: { 82b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "tail: no" << std::endl; 83b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 84b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 85b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case MARISA_ALPHA_BINARY_TAIL: { 86b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "tail: binary" << std::endl; 87b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 88b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 89b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case MARISA_ALPHA_TEXT_TAIL: { 90b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "tail: text" << std::endl; 91b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 92b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 93b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 94b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 95b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak switch (param_order) { 96b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case MARISA_ALPHA_LABEL_ORDER: { 97b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "order: label" << std::endl; 98b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 99b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 100b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case MARISA_ALPHA_WEIGHT_ORDER: { 101b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "order: weight" << std::endl; 102b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 103b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 104b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 105b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 106b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (predict_strs_flag) { 107b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "predict: both IDs and strings" << std::endl; 108b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } else { 109b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "predict: only IDs" << std::endl; 110b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 111b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 112b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 113b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid print_time_info(std::size_t num_keys, double elasped) { 114b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (speed_flag) { 115b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (elasped == 0.0) { 116b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf(" %7s", "-"); 117b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } else { 118b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf(" %7.2f", num_keys / elasped / 1000.0); 119b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 120b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } else { 121b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (num_keys == 0) { 122b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf(" %7s", "-"); 123b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } else { 124b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf(" %7.3f", 1000000.0 * elasped / num_keys); 125b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 126b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 127b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 128b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 129b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid read_keys(std::istream *input, std::vector<Key> *keys) { 130b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Key key; 131b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::string line; 132b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak while (std::getline(*input, line)) { 133b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::string::size_type delim_pos = line.find_last_of('\t'); 134b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (delim_pos != line.npos) { 135b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak char *end_of_value; 136b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak key.second = std::strtod(&line[delim_pos + 1], &end_of_value); 137b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (*end_of_value == '\0') { 138b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak line.resize(delim_pos); 139b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } else { 140b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak key.second = 1.0; 141b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 142b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } else { 143b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak key.second = 1.0; 144b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 145b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak key.first = line; 146b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak keys->push_back(key); 147b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 148b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 149b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 150b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint read_keys(const char * const *args, std::size_t num_args, 151b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<Key> *keys) { 152b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (num_args == 0) { 153b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak read_keys(&std::cin, keys); 154b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 155b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (std::size_t i = 0; i < num_args; ++i) { 156b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::ifstream input_file(args[i], std::ios::binary); 157b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (!input_file) { 158b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: failed to open a keyset file: " 159b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak << args[i] << std::endl; 160b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return 10; 161b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 162b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak read_keys(&input_file, keys); 163b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 164b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "#keys: " << keys->size() << std::endl; 165b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::size_t total_length = 0; 166b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (std::size_t i = 0; i < keys->size(); ++i) { 167b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak total_length += (*keys)[i].first.length(); 168b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 169b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cout << "total length: " << total_length << std::endl; 170b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return 0; 171b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 172b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 173b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid benchmark_build(const std::vector<Key> &keys, int num_tries, 174b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak marisa_alpha::Trie *trie, std::vector<marisa_alpha::UInt32> *key_ids) { 175b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Clock cl; 176b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak trie->build(keys, key_ids, num_tries 177b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak | param_trie | param_tail | param_order); 178b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf(" %9lu", (unsigned long)trie->num_nodes()); 179b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf(" %9lu", (unsigned long)trie->total_size()); 180b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_time_info(keys.size(), cl.elasped()); 181b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 182b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 183b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid benchmark_restore(const marisa_alpha::Trie &trie, 184b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<Key> &keys, 185b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<marisa_alpha::UInt32> &key_ids) { 186b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Clock cl; 187b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::string key; 188b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (std::size_t i = 0; i < key_ids.size(); ++i) { 189b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak key.clear(); 190b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak trie.restore(key_ids[i], &key); 191b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (key != keys[i].first) { 192b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: restore() failed" << std::endl; 193b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return; 194b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 195b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 196b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_time_info(key_ids.size(), cl.elasped()); 197b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 198b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 199b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid benchmark_lookup(const marisa_alpha::Trie &trie, 200b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<Key> &keys, 201b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<marisa_alpha::UInt32> &key_ids) { 202b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Clock cl; 203b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (std::size_t i = 0; i < keys.size(); ++i) { 204b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const marisa_alpha::UInt32 key_id = trie.lookup(keys[i].first); 205b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (key_id != key_ids[i]) { 206b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: lookup() failed" << std::endl; 207b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return; 208b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 209b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 210b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_time_info(keys.size(), cl.elasped()); 211b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 212b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 213b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid benchmark_find(const marisa_alpha::Trie &trie, 214b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<Key> &keys, 215b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<marisa_alpha::UInt32> &key_ids) { 216b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Clock cl; 217b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<marisa_alpha::UInt32> found_key_ids; 218b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (std::size_t i = 0; i < keys.size(); ++i) { 219b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak found_key_ids.clear(); 220b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::size_t num_keys = trie.find(keys[i].first, &found_key_ids); 221b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if ((num_keys == 0) || (found_key_ids.back() != key_ids[i])) { 222b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: find() failed" << std::endl; 223b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return; 224b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 225b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 226b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_time_info(keys.size(), cl.elasped()); 227b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 228b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 229b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid benchmark_predict_breadth_first(const marisa_alpha::Trie &trie, 230b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<Key> &keys, 231b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<marisa_alpha::UInt32> &key_ids) { 232b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Clock cl; 233b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<marisa_alpha::UInt32> found_key_ids; 234b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<std::string> found_keys; 235b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<std::string> *found_keys_ref = 236b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak predict_strs_flag ? &found_keys : NULL; 237b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (std::size_t i = 0; i < keys.size(); ++i) { 238b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak found_key_ids.clear(); 239b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak found_keys.clear(); 240b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::size_t num_keys = trie.predict_breadth_first( 241b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak keys[i].first, &found_key_ids, found_keys_ref); 242b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if ((num_keys == 0) || (found_key_ids.front() != key_ids[i])) { 243b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: predict() failed" << std::endl; 244b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return; 245b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 246b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 247b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_time_info(keys.size(), cl.elasped()); 248b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 249b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 250b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid benchmark_predict_depth_first(const marisa_alpha::Trie &trie, 251b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<Key> &keys, 252b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::vector<marisa_alpha::UInt32> &key_ids) { 253b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak Clock cl; 254b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<marisa_alpha::UInt32> found_key_ids; 255b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<std::string> found_keys; 256b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<std::string> *found_keys_ref = 257b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak predict_strs_flag ? &found_keys : NULL; 258b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (std::size_t i = 0; i < keys.size(); ++i) { 259b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak found_key_ids.clear(); 260b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak found_keys.clear(); 261b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const std::size_t num_keys = trie.predict_depth_first( 262b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak keys[i].first, &found_key_ids, found_keys_ref); 263b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if ((num_keys == 0) || (found_key_ids.front() != key_ids[i])) { 264b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: predict() failed" << std::endl; 265b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return; 266b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 267b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 268b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_time_info(keys.size(), cl.elasped()); 269b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 270b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 271b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakvoid benchmark(const std::vector<Key> &keys, int num_tries) { 272b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("%6d", num_tries); 273b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak marisa_alpha::Trie trie; 274b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<marisa_alpha::UInt32> key_ids; 275b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak benchmark_build(keys, num_tries, &trie, &key_ids); 276b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (!trie.empty()) { 277b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak benchmark_restore(trie, keys, key_ids); 278b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak benchmark_lookup(trie, keys, key_ids); 279b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak benchmark_find(trie, keys, key_ids); 280b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak benchmark_predict_breadth_first(trie, keys, key_ids); 281b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak benchmark_predict_depth_first(trie, keys, key_ids); 282b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 283b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("\n"); 284b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 285b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 286b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint benchmark(const char * const *args, std::size_t num_args) try { 287b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::vector<Key> keys; 288b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const int ret = read_keys(args, num_args, &keys); 289b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (ret != 0) { 290b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return ret; 291b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 292b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("------+---------+---------+-------+" 293b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "-------+-------+-------+-------+-------\n"); 294b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("%6s %9s %9s %7s %7s %7s %7s %7s %7s\n", 295b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "#tries", "#nodes", "size", 296b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "build", "restore", "lookup", "find", "predict", "predict"); 297b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("%6s %9s %9s %7s %7s %7s %7s %7s %7s\n", 298b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "", "", "", "", "", "", "", "breadth", "depth"); 299b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if (speed_flag) { 300b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("%6s %9s %9s %7s %7s %7s %7s %7s %7s\n", 301b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "", "", "[bytes]", 302b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "[K/s]", "[K/s]", "[K/s]", "[K/s]", "[K/s]", "[K/s]"); 303b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } else { 304b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("%6s %9s %9s %7s %7s %7s %7s %7s %7s\n", 305b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "", "", "[bytes]", "[us]", "[us]", "[us]", "[us]", "[us]", "[us]"); 306b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 307b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("------+---------+---------+-------+" 308b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "-------+-------+-------+-------+-------\n"); 309b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak for (int i = param_min_num_tries; i <= param_max_num_tries; ++i) { 310b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak benchmark(keys, i); 311b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 312b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::printf("------+---------+---------+-------+" 313b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak "-------+-------+-------+-------+-------\n"); 314b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return 0; 315b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} catch (const marisa_alpha::Exception &ex) { 316b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << ex.filename() << ':' << ex.line() 317b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak << ": " << ex.what() << std::endl; 318b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return -1; 319b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 320b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 321b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} // namespace 322b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 323b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniakint main(int argc, char *argv[]) { 324b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::ios::sync_with_stdio(false); 325b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak 326b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak ::cmdopt_option long_options[] = { 327b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "min-num-tries", 1, NULL, 'N' }, 328b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "max-num-tries", 1, NULL, 'n' }, 329b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "patricia-trie", 0, NULL, 'P' }, 330b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "prefix-trie", 0, NULL, 'p' }, 331b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "text-tail", 0, NULL, 'T' }, 332b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "binary-tail", 0, NULL, 'b' }, 333b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "without-tail", 0, NULL, 't' }, 334b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "weight-order", 0, NULL, 'w' }, 335b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "label-order", 0, NULL, 'l' }, 336b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "predict-ids", 0, NULL, 'I' }, 337b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "predict-strs", 0, NULL, 'i' }, 338b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "print-speed", 0, NULL, 'S' }, 339b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "print-time", 0, NULL, 's' }, 340b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { "help", 0, NULL, 'h' }, 341b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak { NULL, 0, NULL, 0 } 342b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak }; 343b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak ::cmdopt_t cmdopt; 344b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak ::cmdopt_init(&cmdopt, argc, argv, "N:n:PpTbtwlIiSsh", long_options); 345b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak int label; 346b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak while ((label = ::cmdopt_get(&cmdopt)) != -1) { 347b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak switch (label) { 348b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'N': { 349b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak char *end_of_value; 350b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const long value = std::strtol(cmdopt.optarg, &end_of_value, 10); 351b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if ((*end_of_value != '\0') || (value <= 0) || 352b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak (value > MARISA_ALPHA_MAX_NUM_TRIES)) { 353b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: option `-n' with an invalid argument: " 354b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak << cmdopt.optarg << std::endl; 355b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 356b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_min_num_tries = (int)value; 357b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 358b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 359b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'n': { 360b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak char *end_of_value; 361b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak const long value = std::strtol(cmdopt.optarg, &end_of_value, 10); 362b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak if ((*end_of_value != '\0') || (value <= 0) || 363b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak (value > MARISA_ALPHA_MAX_NUM_TRIES)) { 364b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak std::cerr << "error: option `-n' with an invalid argument: " 365b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak << cmdopt.optarg << std::endl; 366b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 367b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_max_num_tries = (int)value; 368b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 369b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 370b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'P': { 371b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_trie = MARISA_ALPHA_PATRICIA_TRIE; 372b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 373b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 374b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'p': { 375b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_trie = MARISA_ALPHA_PREFIX_TRIE; 376b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 377b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 378b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'T': { 379b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_tail = MARISA_ALPHA_TEXT_TAIL; 380b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 381b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 382b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'b': { 383b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_tail = MARISA_ALPHA_BINARY_TAIL; 384b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 385b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 386b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 't': { 387b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_tail = MARISA_ALPHA_WITHOUT_TAIL; 388b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 389b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 390b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'w': { 391b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_order = MARISA_ALPHA_WEIGHT_ORDER; 392b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 393b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 394b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'l': { 395b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak param_order = MARISA_ALPHA_LABEL_ORDER; 396b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 397b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 398b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'I': { 399b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak predict_strs_flag = false; 400b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 401b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 402b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'i': { 403b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak predict_strs_flag = true; 404b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 405b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 406b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'S': { 407b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak speed_flag = true; 408b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 409b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 410b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 's': { 411b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak speed_flag = false; 412b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak break; 413b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 414b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak case 'h': { 415b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_help(argv[0]); 416b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return 0; 417b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 418b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak default: { 419b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return 1; 420b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 421b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 422b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak } 423b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak print_config(); 424b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak return benchmark(cmdopt.argv + cmdopt.optind, cmdopt.argc - cmdopt.optind); 425b4acd09676fe12e9d95b2a5f3f742e9a1a0483b1Przemyslaw Szczepaniak} 426