Home | History | Annotate | Download | only in tools 
      1 #include <cstdlib>
      2 #include <iostream>
      3 #include <string>
      4 
      5 #include <marisa.h>
      6 
      7 #include "./cmdopt.h"
      8 
      9 namespace {
     10 
     11 bool mmap_flag = true;
     12 
     13 void print_help(const char *cmd) {
     14   std::cerr << "Usage: " << cmd << " [OPTION]... DIC\n\n"
     15       "Options:\n"
     16       "  -m, --mmap-dictionary  use memory-mapped I/O to load a dictionary"
     17       " (default)\n"
     18       "  -r, --read-dictionary  read an entire dictionary into memory\n"
     19       "  -h, --help             print this help\n"
     20       << std::endl;
     21 }
     22 
     23 int lookup(const char * const *args, std::size_t num_args) {
     24   if (num_args == 0) {
     25     std::cerr << "error: a dictionary is not specified" << std::endl;
     26     return 10;
     27   } else if (num_args > 1) {
     28     std::cerr << "error: more than one dictionaries are specified"
     29         << std::endl;
     30     return 11;
     31   }
     32 
     33   marisa::Trie trie;
     34   marisa::Mapper mapper;
     35   if (mmap_flag) {
     36     try {
     37       trie.mmap(&mapper, args[0]);
     38     } catch (const marisa::Exception &ex) {
     39       std::cerr << ex.filename() << ':' << ex.line() << ": " << ex.what()
     40           << ": failed to mmap a dictionary file: " << args[0] << std::endl;
     41       return 20;
     42     }
     43   } else {
     44     try {
     45       trie.load(args[0]);
     46     } catch (const marisa::Exception &ex) {
     47       std::cerr << ex.filename() << ':' << ex.line() << ": " << ex.what()
     48           << ": failed to load a dictionary file: " << args[0] << std::endl;
     49       return 21;
     50     }
     51   }
     52 
     53   std::string str;
     54   while (std::getline(std::cin, str)) {
     55     const marisa::UInt32 key_id = trie.lookup(str);
     56     if (key_id != trie.notfound()) {
     57       std::cout << key_id << '\t' << str << '\n';
     58     } else {
     59       std::cout << "-1\t" << str << '\n';
     60     }
     61     if (!std::cout) {
     62       std::cerr << "error: failed to write results to standard output"
     63           << std::endl;
     64       return 30;
     65     }
     66   }
     67 
     68   return 0;
     69 }
     70 
     71 }  // namespace
     72 
     73 int main(int argc, char *argv[]) {
     74   std::ios::sync_with_stdio(false);
     75 
     76   ::cmdopt_option long_options[] = {
     77     { "mmap-dictionary", 0, NULL, 'm' },
     78     { "read-dictionary", 0, NULL, 'r' },
     79     { "help", 0, NULL, 'h' },
     80     { NULL, 0, NULL, 0 }
     81   };
     82   ::cmdopt_t cmdopt;
     83   ::cmdopt_init(&cmdopt, argc, argv, "mrh", long_options);
     84   int label;
     85   while ((label = ::cmdopt_get(&cmdopt)) != -1) {
     86     switch (label) {
     87       case 'm': {
     88         mmap_flag = true;
     89         break;
     90       }
     91       case 'r': {
     92         mmap_flag = false;
     93         break;
     94       }
     95       case 'h': {
     96         print_help(argv[0]);
     97         return 0;
     98       }
     99       default: {
    100         return 1;
    101       }
    102     }
    103   }
    104   return lookup(cmdopt.argv + cmdopt.optind, cmdopt.argc - cmdopt.optind);
    105 }
    106