|
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 |
std::size_t max_num_results = 10;
|
|
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 |
" -n, --max-num-results=[N] limit the number of outputs to N"
|
|
Packit |
de3218 |
" (default: 10)\n"
|
|
Packit |
de3218 |
" 0: no limit\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 predictive_search(const char * const *args, std::size_t num_args) {
|
|
Packit |
de3218 |
if (num_args == 0) {
|
|
Packit |
de3218 |
std::cerr << "error: dictionary is not specified" << std::endl;
|
|
Packit |
de3218 |
return 10;
|
|
Packit |
de3218 |
} else if (num_args > 1) {
|
|
Packit |
de3218 |
std::cerr << "error: more than one dictionaries are specified"
|
|
Packit |
de3218 |
<< std::endl;
|
|
Packit |
de3218 |
return 11;
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
|
|
Packit |
de3218 |
marisa::Trie trie;
|
|
Packit |
de3218 |
if (mmap_flag) {
|
|
Packit |
de3218 |
try {
|
|
Packit |
de3218 |
trie.mmap(args[0]);
|
|
Packit |
de3218 |
} catch (const marisa::Exception &ex) {
|
|
Packit |
de3218 |
std::cerr << ex.what() << ": failed to mmap a dictionary file: "
|
|
Packit |
de3218 |
<< args[0] << std::endl;
|
|
Packit |
de3218 |
return 20;
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
} else {
|
|
Packit |
de3218 |
try {
|
|
Packit |
de3218 |
trie.load(args[0]);
|
|
Packit |
de3218 |
} catch (const marisa::Exception &ex) {
|
|
Packit |
de3218 |
std::cerr << ex.what() << ": failed to load a dictionary file: "
|
|
Packit |
de3218 |
<< args[0] << std::endl;
|
|
Packit |
de3218 |
return 21;
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
|
|
Packit |
de3218 |
marisa::Agent agent;
|
|
Packit |
de3218 |
marisa::Keyset keyset;
|
|
Packit |
de3218 |
std::string str;
|
|
Packit |
de3218 |
while (std::getline(std::cin, str)) {
|
|
Packit |
de3218 |
try {
|
|
Packit |
de3218 |
agent.set_query(str.c_str(), str.length());
|
|
Packit |
de3218 |
while (trie.predictive_search(agent)) {
|
|
Packit |
de3218 |
keyset.push_back(agent.key());
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
if (keyset.empty()) {
|
|
Packit |
de3218 |
std::cout << "not found" << std::endl;
|
|
Packit |
de3218 |
} else {
|
|
Packit |
de3218 |
std::cout << keyset.size() << " found" << std::endl;
|
|
Packit |
de3218 |
const std::size_t end = std::min(max_num_results, keyset.size());
|
|
Packit |
de3218 |
for (std::size_t i = 0; i < end; ++i) {
|
|
Packit |
de3218 |
std::cout << keyset[i].id() << '\t';
|
|
Packit |
de3218 |
std::cout.write(keyset[i].ptr(), keyset[i].length()) << '\t';
|
|
Packit |
de3218 |
std::cout << str << '\n';
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
keyset.reset();
|
|
Packit |
de3218 |
} catch (const marisa::Exception &ex) {
|
|
Packit |
de3218 |
std::cerr << ex.what() << ": predictive_search() failed: "
|
|
Packit |
de3218 |
<< str << std::endl;
|
|
Packit |
de3218 |
return 30;
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
|
|
Packit |
de3218 |
if (!std::cout) {
|
|
Packit |
de3218 |
std::cerr << "error: failed to write results to standard output"
|
|
Packit |
de3218 |
<< std::endl;
|
|
Packit |
de3218 |
return 31;
|
|
Packit |
de3218 |
}
|
|
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 |
{ "max-num-results", 1, NULL, 'n' },
|
|
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, "n:mrh", 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 |
std::cerr << "error: option `-n' with an invalid argument: "
|
|
Packit |
de3218 |
<< cmdopt.optarg << std::endl;
|
|
Packit |
de3218 |
}
|
|
Packit |
de3218 |
if ((value == 0) || ((unsigned long long)value > MARISA_SIZE_MAX)) {
|
|
Packit |
de3218 |
max_num_results = MARISA_SIZE_MAX;
|
|
Packit |
de3218 |
} else {
|
|
Packit |
de3218 |
max_num_results = (std::size_t)value;
|
|
Packit |
de3218 |
}
|
|
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 predictive_search(cmdopt.argv + cmdopt.optind,
|
|
Packit |
de3218 |
cmdopt.argc - cmdopt.optind);
|
|
Packit |
de3218 |
}
|