Blame tools/marisa-dump.cc

Packit de3218
#ifdef _WIN32
Packit de3218
 #include <fcntl.h>
Packit de3218
 #include <io.h>
Packit de3218
 #include <stdio.h>
Packit de3218
#endif  // _WIN32
Packit de3218
Packit de3218
#include <cstdlib>
Packit de3218
#include <iostream>
Packit de3218
#include <string>
Packit de3218
Packit de3218
#include <marisa.h>
Packit de3218
Packit de3218
#include "cmdopt.h"
Packit de3218
Packit de3218
namespace {
Packit de3218
Packit de3218
const char *delimiter = "\n";
Packit de3218
bool mmap_flag = true;
Packit de3218
Packit de3218
void print_help(const char *cmd) {
Packit de3218
  std::cerr << "Usage: " << cmd << " [OPTION]... DIC...\n\n"
Packit de3218
      "Options:\n"
Packit de3218
      "  -d, --delimiter=[S]    specify the delimier (default: \"\\n\")\n"
Packit de3218
      "  -m, --mmap-dictionary  use memory-mapped I/O to load a dictionary"
Packit de3218
      " (default)\n"
Packit de3218
      "  -r, --read-dictionary  read an entire dictionary into memory\n"
Packit de3218
      "  -h, --help             print this help\n"
Packit de3218
      << std::endl;
Packit de3218
}
Packit de3218
Packit de3218
int dump(const marisa::Trie &trie) {
Packit de3218
  std::size_t num_keys = 0;
Packit de3218
  marisa::Agent agent;
Packit de3218
  agent.set_query("");
Packit de3218
  try {
Packit de3218
    while (trie.predictive_search(agent)) {
Packit de3218
      std::cout.write(agent.key().ptr(), agent.key().length()) << delimiter;
Packit de3218
      if (!std::cout) {
Packit de3218
        std::cerr << "error: failed to write results to standard output"
Packit de3218
            << std::endl;
Packit de3218
        return 20;
Packit de3218
      }
Packit de3218
      ++num_keys;
Packit de3218
    }
Packit de3218
  } catch (const marisa::Exception &ex) {
Packit de3218
    std::cerr << ex.what() << ": predictive_search() failed" << std::endl;
Packit de3218
    return 21;
Packit de3218
  }
Packit de3218
  std::cerr << "#keys: " << num_keys << std::endl;
Packit de3218
  return 0;
Packit de3218
}
Packit de3218
Packit de3218
int dump(const char *filename) {
Packit de3218
  marisa::Trie trie;
Packit de3218
  if (filename != NULL) {
Packit de3218
    std::cerr << "input: " << filename << std::endl;
Packit de3218
    if (mmap_flag) {
Packit de3218
      try {
Packit de3218
        trie.mmap(filename);
Packit de3218
      } catch (const marisa::Exception &ex) {
Packit de3218
        std::cerr << ex.what() << ": failed to mmap a dictionary file: "
Packit de3218
            << filename << std::endl;
Packit de3218
        return 10;
Packit de3218
      }
Packit de3218
    } else {
Packit de3218
      try {
Packit de3218
        trie.load(filename);
Packit de3218
      } catch (const marisa::Exception &ex) {
Packit de3218
        std::cerr << ex.what() << ": failed to load a dictionary file: "
Packit de3218
            << filename << std::endl;
Packit de3218
        return 11;
Packit de3218
      }
Packit de3218
    }
Packit de3218
  } else {
Packit de3218
    std::cerr << "input: <stdin>" << std::endl;
Packit de3218
#ifdef _WIN32
Packit de3218
    const int stdin_fileno = ::_fileno(stdin);
Packit de3218
    if (stdin_fileno < 0) {
Packit de3218
      std::cerr << "error: failed to get the file descriptor of "
Packit de3218
          "standard input" << std::endl;
Packit de3218
      return 20;
Packit de3218
    }
Packit de3218
    if (::_setmode(stdin_fileno, _O_BINARY) == -1) {
Packit de3218
      std::cerr << "error: failed to set binary mode" << std::endl;
Packit de3218
      return 21;
Packit de3218
    }
Packit de3218
#endif  // _WIN32
Packit de3218
    try {
Packit de3218
      std::cin >> trie;
Packit de3218
    } catch (const marisa::Exception &ex) {
Packit de3218
      std::cerr << ex.what()
Packit de3218
          << ": failed to read a dictionary from standard input" << std::endl;
Packit de3218
      return 22;
Packit de3218
    }
Packit de3218
  }
Packit de3218
  return dump(trie);
Packit de3218
}
Packit de3218
Packit de3218
int dump(const char * const *args, std::size_t num_args) {
Packit de3218
  if (num_args == 0) {
Packit de3218
    return dump(NULL);
Packit de3218
  }
Packit de3218
  for (std::size_t i = 0; i < num_args; ++i) {
Packit de3218
    const int result = dump(args[i]);
Packit de3218
    if (result != 0) {
Packit de3218
      return result;
Packit de3218
    }
Packit de3218
  }
Packit de3218
  return 0;
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
    { "delimiter", 1, NULL, 'd' },
Packit de3218
    { "mmap-dictionary", 0, NULL, 'm' },
Packit de3218
    { "read-dictionary", 0, NULL, 'r' },
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, "d:mrh", long_options);
Packit de3218
  int label;
Packit de3218
  while ((label = ::cmdopt_get(&cmdopt)) != -1) {
Packit de3218
    switch (label) {
Packit de3218
      case 'd': {
Packit de3218
        delimiter = cmdopt.optarg;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'm': {
Packit de3218
        mmap_flag = true;
Packit de3218
        break;
Packit de3218
      }
Packit de3218
      case 'r': {
Packit de3218
        mmap_flag = 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
  return dump(cmdopt.argv + cmdopt.optind, cmdopt.argc - cmdopt.optind);
Packit de3218
}