xref: /aosp_15_r20/external/marisa-trie/tools/marisa-build.cc (revision ab8db090fce404b23716c4c9194221ee27efe31c)
1*ab8db090SAndroid Build Coastguard Worker #ifdef _WIN32
2*ab8db090SAndroid Build Coastguard Worker  #include <fcntl.h>
3*ab8db090SAndroid Build Coastguard Worker  #include <io.h>
4*ab8db090SAndroid Build Coastguard Worker  #include <stdio.h>
5*ab8db090SAndroid Build Coastguard Worker #endif  // _WIN32
6*ab8db090SAndroid Build Coastguard Worker 
7*ab8db090SAndroid Build Coastguard Worker #include <cstdlib>
8*ab8db090SAndroid Build Coastguard Worker #include <fstream>
9*ab8db090SAndroid Build Coastguard Worker #include <iostream>
10*ab8db090SAndroid Build Coastguard Worker #include <string>
11*ab8db090SAndroid Build Coastguard Worker 
12*ab8db090SAndroid Build Coastguard Worker #include <marisa.h>
13*ab8db090SAndroid Build Coastguard Worker 
14*ab8db090SAndroid Build Coastguard Worker #include "cmdopt.h"
15*ab8db090SAndroid Build Coastguard Worker 
16*ab8db090SAndroid Build Coastguard Worker namespace {
17*ab8db090SAndroid Build Coastguard Worker 
18*ab8db090SAndroid Build Coastguard Worker int param_num_tries = MARISA_DEFAULT_NUM_TRIES;
19*ab8db090SAndroid Build Coastguard Worker marisa::TailMode param_tail_mode = MARISA_DEFAULT_TAIL;
20*ab8db090SAndroid Build Coastguard Worker marisa::NodeOrder param_node_order = MARISA_DEFAULT_ORDER;
21*ab8db090SAndroid Build Coastguard Worker marisa::CacheLevel param_cache_level = MARISA_DEFAULT_CACHE;
22*ab8db090SAndroid Build Coastguard Worker const char *output_filename = NULL;
23*ab8db090SAndroid Build Coastguard Worker 
print_help(const char * cmd)24*ab8db090SAndroid Build Coastguard Worker void print_help(const char *cmd) {
25*ab8db090SAndroid Build Coastguard Worker   std::cerr << "Usage: " << cmd << " [OPTION]... [FILE]...\n\n"
26*ab8db090SAndroid Build Coastguard Worker       "Options:\n"
27*ab8db090SAndroid Build Coastguard Worker       "  -n, --num-tries=[N]  limit the number of tries"
28*ab8db090SAndroid Build Coastguard Worker       " [" << MARISA_MIN_NUM_TRIES << ", " << MARISA_MAX_NUM_TRIES
29*ab8db090SAndroid Build Coastguard Worker       << "] (default: 3)\n"
30*ab8db090SAndroid Build Coastguard Worker       "  -t, --text-tail      build a dictionary with text TAIL (default)\n"
31*ab8db090SAndroid Build Coastguard Worker       "  -b, --binary-tail    build a dictionary with binary TAIL\n"
32*ab8db090SAndroid Build Coastguard Worker       "  -w, --weight-order   arrange siblings in weight order (default)\n"
33*ab8db090SAndroid Build Coastguard Worker       "  -l, --label-order    arrange siblings in label order\n"
34*ab8db090SAndroid Build Coastguard Worker       "  -c, --cache-level=[N]    specify the cache size"
35*ab8db090SAndroid Build Coastguard Worker       " [1, 5] (default: 3)\n"
36*ab8db090SAndroid Build Coastguard Worker       "  -o, --output=[FILE]  write tries to FILE (default: stdout)\n"
37*ab8db090SAndroid Build Coastguard Worker       "  -h, --help           print this help\n"
38*ab8db090SAndroid Build Coastguard Worker       << std::endl;
39*ab8db090SAndroid Build Coastguard Worker }
40*ab8db090SAndroid Build Coastguard Worker 
read_keys(std::istream & input,marisa::Keyset * keyset)41*ab8db090SAndroid Build Coastguard Worker void read_keys(std::istream &input, marisa::Keyset *keyset) {
42*ab8db090SAndroid Build Coastguard Worker   std::string line;
43*ab8db090SAndroid Build Coastguard Worker   while (std::getline(input, line)) {
44*ab8db090SAndroid Build Coastguard Worker     const std::string::size_type delim_pos = line.find_last_of('\t');
45*ab8db090SAndroid Build Coastguard Worker     float weight = 1.0F;
46*ab8db090SAndroid Build Coastguard Worker     if (delim_pos != line.npos) {
47*ab8db090SAndroid Build Coastguard Worker       char *end_of_value;
48*ab8db090SAndroid Build Coastguard Worker       weight = (float)std::strtod(&line[delim_pos + 1], &end_of_value);
49*ab8db090SAndroid Build Coastguard Worker       if (*end_of_value == '\0') {
50*ab8db090SAndroid Build Coastguard Worker         line.resize(delim_pos);
51*ab8db090SAndroid Build Coastguard Worker       }
52*ab8db090SAndroid Build Coastguard Worker     }
53*ab8db090SAndroid Build Coastguard Worker     keyset->push_back(line.c_str(), line.length(), weight);
54*ab8db090SAndroid Build Coastguard Worker   }
55*ab8db090SAndroid Build Coastguard Worker }
56*ab8db090SAndroid Build Coastguard Worker 
build(const char * const * args,std::size_t num_args)57*ab8db090SAndroid Build Coastguard Worker int build(const char * const *args, std::size_t num_args) {
58*ab8db090SAndroid Build Coastguard Worker   marisa::Keyset keyset;
59*ab8db090SAndroid Build Coastguard Worker   if (num_args == 0) try {
60*ab8db090SAndroid Build Coastguard Worker     read_keys(std::cin, &keyset);
61*ab8db090SAndroid Build Coastguard Worker   } catch (const marisa::Exception &ex) {
62*ab8db090SAndroid Build Coastguard Worker     std::cerr << ex.what() << ": failed to read keys" << std::endl;
63*ab8db090SAndroid Build Coastguard Worker     return 10;
64*ab8db090SAndroid Build Coastguard Worker   }
65*ab8db090SAndroid Build Coastguard Worker 
66*ab8db090SAndroid Build Coastguard Worker   for (std::size_t i = 0; i < num_args; ++i) try {
67*ab8db090SAndroid Build Coastguard Worker     std::ifstream input_file(args[i], std::ios::binary);
68*ab8db090SAndroid Build Coastguard Worker     if (!input_file) {
69*ab8db090SAndroid Build Coastguard Worker       std::cerr << "error: failed to open: " << args[i] << std::endl;
70*ab8db090SAndroid Build Coastguard Worker       return 11;
71*ab8db090SAndroid Build Coastguard Worker     }
72*ab8db090SAndroid Build Coastguard Worker     read_keys(input_file, &keyset);
73*ab8db090SAndroid Build Coastguard Worker   } catch (const marisa::Exception &ex) {
74*ab8db090SAndroid Build Coastguard Worker     std::cerr << ex.what() << ": failed to read keys" << std::endl;
75*ab8db090SAndroid Build Coastguard Worker     return 12;
76*ab8db090SAndroid Build Coastguard Worker   }
77*ab8db090SAndroid Build Coastguard Worker 
78*ab8db090SAndroid Build Coastguard Worker   marisa::Trie trie;
79*ab8db090SAndroid Build Coastguard Worker   try {
80*ab8db090SAndroid Build Coastguard Worker     trie.build(keyset, param_num_tries | param_tail_mode | param_node_order |
81*ab8db090SAndroid Build Coastguard Worker         param_cache_level);
82*ab8db090SAndroid Build Coastguard Worker   } catch (const marisa::Exception &ex) {
83*ab8db090SAndroid Build Coastguard Worker     std::cerr << ex.what() << ": failed to build a dictionary" << std::endl;
84*ab8db090SAndroid Build Coastguard Worker     return 20;
85*ab8db090SAndroid Build Coastguard Worker   }
86*ab8db090SAndroid Build Coastguard Worker 
87*ab8db090SAndroid Build Coastguard Worker   std::cerr << "#keys: " << trie.num_keys() << std::endl;
88*ab8db090SAndroid Build Coastguard Worker   std::cerr << "#nodes: " << trie.num_nodes() << std::endl;
89*ab8db090SAndroid Build Coastguard Worker   std::cerr << "size: " << trie.io_size() << std::endl;
90*ab8db090SAndroid Build Coastguard Worker 
91*ab8db090SAndroid Build Coastguard Worker   if (output_filename != NULL) {
92*ab8db090SAndroid Build Coastguard Worker     try {
93*ab8db090SAndroid Build Coastguard Worker       trie.save(output_filename);
94*ab8db090SAndroid Build Coastguard Worker     } catch (const marisa::Exception &ex) {
95*ab8db090SAndroid Build Coastguard Worker       std::cerr << ex.what() << ": failed to write a dictionary to file: "
96*ab8db090SAndroid Build Coastguard Worker           << output_filename << std::endl;
97*ab8db090SAndroid Build Coastguard Worker       return 30;
98*ab8db090SAndroid Build Coastguard Worker     }
99*ab8db090SAndroid Build Coastguard Worker   } else {
100*ab8db090SAndroid Build Coastguard Worker #ifdef _WIN32
101*ab8db090SAndroid Build Coastguard Worker     const int stdout_fileno = ::_fileno(stdout);
102*ab8db090SAndroid Build Coastguard Worker     if (stdout_fileno < 0) {
103*ab8db090SAndroid Build Coastguard Worker       std::cerr << "error: failed to get the file descriptor of "
104*ab8db090SAndroid Build Coastguard Worker           "standard output" << std::endl;
105*ab8db090SAndroid Build Coastguard Worker       return 31;
106*ab8db090SAndroid Build Coastguard Worker     }
107*ab8db090SAndroid Build Coastguard Worker     if (::_setmode(stdout_fileno, _O_BINARY) == -1) {
108*ab8db090SAndroid Build Coastguard Worker       std::cerr << "error: failed to set binary mode" << std::endl;
109*ab8db090SAndroid Build Coastguard Worker       return 32;
110*ab8db090SAndroid Build Coastguard Worker     }
111*ab8db090SAndroid Build Coastguard Worker #endif  // _WIN32
112*ab8db090SAndroid Build Coastguard Worker     try {
113*ab8db090SAndroid Build Coastguard Worker       std::cout << trie;
114*ab8db090SAndroid Build Coastguard Worker     } catch (const marisa::Exception &ex) {
115*ab8db090SAndroid Build Coastguard Worker       std::cerr << ex.what()
116*ab8db090SAndroid Build Coastguard Worker           << ": failed to write a dictionary to standard output" << std::endl;
117*ab8db090SAndroid Build Coastguard Worker       return 33;
118*ab8db090SAndroid Build Coastguard Worker     }
119*ab8db090SAndroid Build Coastguard Worker   }
120*ab8db090SAndroid Build Coastguard Worker   return 0;
121*ab8db090SAndroid Build Coastguard Worker }
122*ab8db090SAndroid Build Coastguard Worker 
123*ab8db090SAndroid Build Coastguard Worker }  // namespace
124*ab8db090SAndroid Build Coastguard Worker 
main(int argc,char * argv[])125*ab8db090SAndroid Build Coastguard Worker int main(int argc, char *argv[]) {
126*ab8db090SAndroid Build Coastguard Worker   std::ios::sync_with_stdio(false);
127*ab8db090SAndroid Build Coastguard Worker 
128*ab8db090SAndroid Build Coastguard Worker   ::cmdopt_option long_options[] = {
129*ab8db090SAndroid Build Coastguard Worker     { "max-num-tries", 1, NULL, 'n' },
130*ab8db090SAndroid Build Coastguard Worker     { "text-tail", 0, NULL, 't' },
131*ab8db090SAndroid Build Coastguard Worker     { "binary-tail", 0, NULL, 'b' },
132*ab8db090SAndroid Build Coastguard Worker     { "weight-order", 0, NULL, 'w' },
133*ab8db090SAndroid Build Coastguard Worker     { "label-order", 0, NULL, 'l' },
134*ab8db090SAndroid Build Coastguard Worker     { "cache-level", 1, NULL, 'c' },
135*ab8db090SAndroid Build Coastguard Worker     { "output", 1, NULL, 'o' },
136*ab8db090SAndroid Build Coastguard Worker     { "help", 0, NULL, 'h' },
137*ab8db090SAndroid Build Coastguard Worker     { NULL, 0, NULL, 0 }
138*ab8db090SAndroid Build Coastguard Worker   };
139*ab8db090SAndroid Build Coastguard Worker   ::cmdopt_t cmdopt;
140*ab8db090SAndroid Build Coastguard Worker   ::cmdopt_init(&cmdopt, argc, argv, "n:tbwlc:o:h", long_options);
141*ab8db090SAndroid Build Coastguard Worker   int label;
142*ab8db090SAndroid Build Coastguard Worker   while ((label = ::cmdopt_get(&cmdopt)) != -1) {
143*ab8db090SAndroid Build Coastguard Worker     switch (label) {
144*ab8db090SAndroid Build Coastguard Worker       case 'n': {
145*ab8db090SAndroid Build Coastguard Worker         char *end_of_value;
146*ab8db090SAndroid Build Coastguard Worker         const long value = std::strtol(cmdopt.optarg, &end_of_value, 10);
147*ab8db090SAndroid Build Coastguard Worker         if ((*end_of_value != '\0') || (value <= 0) ||
148*ab8db090SAndroid Build Coastguard Worker             (value > MARISA_MAX_NUM_TRIES)) {
149*ab8db090SAndroid Build Coastguard Worker           std::cerr << "error: option `-n' with an invalid argument: "
150*ab8db090SAndroid Build Coastguard Worker               << cmdopt.optarg << std::endl;
151*ab8db090SAndroid Build Coastguard Worker           return 1;
152*ab8db090SAndroid Build Coastguard Worker         }
153*ab8db090SAndroid Build Coastguard Worker         param_num_tries = (int)value;
154*ab8db090SAndroid Build Coastguard Worker         break;
155*ab8db090SAndroid Build Coastguard Worker       }
156*ab8db090SAndroid Build Coastguard Worker       case 't': {
157*ab8db090SAndroid Build Coastguard Worker         param_tail_mode = MARISA_TEXT_TAIL;
158*ab8db090SAndroid Build Coastguard Worker         break;
159*ab8db090SAndroid Build Coastguard Worker       }
160*ab8db090SAndroid Build Coastguard Worker       case 'b': {
161*ab8db090SAndroid Build Coastguard Worker         param_tail_mode = MARISA_BINARY_TAIL;
162*ab8db090SAndroid Build Coastguard Worker         break;
163*ab8db090SAndroid Build Coastguard Worker       }
164*ab8db090SAndroid Build Coastguard Worker       case 'w': {
165*ab8db090SAndroid Build Coastguard Worker         param_node_order = MARISA_WEIGHT_ORDER;
166*ab8db090SAndroid Build Coastguard Worker         break;
167*ab8db090SAndroid Build Coastguard Worker       }
168*ab8db090SAndroid Build Coastguard Worker       case 'l': {
169*ab8db090SAndroid Build Coastguard Worker         param_node_order = MARISA_LABEL_ORDER;
170*ab8db090SAndroid Build Coastguard Worker         break;
171*ab8db090SAndroid Build Coastguard Worker       }
172*ab8db090SAndroid Build Coastguard Worker       case 'c': {
173*ab8db090SAndroid Build Coastguard Worker         char *end_of_value;
174*ab8db090SAndroid Build Coastguard Worker         const long value = std::strtol(cmdopt.optarg, &end_of_value, 10);
175*ab8db090SAndroid Build Coastguard Worker         if ((*end_of_value != '\0') || (value < 1) || (value > 5)) {
176*ab8db090SAndroid Build Coastguard Worker           std::cerr << "error: option `-c' with an invalid argument: "
177*ab8db090SAndroid Build Coastguard Worker               << cmdopt.optarg << std::endl;
178*ab8db090SAndroid Build Coastguard Worker           return 2;
179*ab8db090SAndroid Build Coastguard Worker         } else if (value == 1) {
180*ab8db090SAndroid Build Coastguard Worker           param_cache_level = MARISA_TINY_CACHE;
181*ab8db090SAndroid Build Coastguard Worker         } else if (value == 2) {
182*ab8db090SAndroid Build Coastguard Worker           param_cache_level = MARISA_SMALL_CACHE;
183*ab8db090SAndroid Build Coastguard Worker         } else if (value == 3) {
184*ab8db090SAndroid Build Coastguard Worker           param_cache_level = MARISA_NORMAL_CACHE;
185*ab8db090SAndroid Build Coastguard Worker         } else if (value == 4) {
186*ab8db090SAndroid Build Coastguard Worker           param_cache_level = MARISA_LARGE_CACHE;
187*ab8db090SAndroid Build Coastguard Worker         } else if (value == 5) {
188*ab8db090SAndroid Build Coastguard Worker           param_cache_level = MARISA_HUGE_CACHE;
189*ab8db090SAndroid Build Coastguard Worker         }
190*ab8db090SAndroid Build Coastguard Worker         break;
191*ab8db090SAndroid Build Coastguard Worker       }
192*ab8db090SAndroid Build Coastguard Worker       case 'o': {
193*ab8db090SAndroid Build Coastguard Worker         output_filename = cmdopt.optarg;
194*ab8db090SAndroid Build Coastguard Worker         break;
195*ab8db090SAndroid Build Coastguard Worker       }
196*ab8db090SAndroid Build Coastguard Worker       case 'h': {
197*ab8db090SAndroid Build Coastguard Worker         print_help(argv[0]);
198*ab8db090SAndroid Build Coastguard Worker         return 0;
199*ab8db090SAndroid Build Coastguard Worker       }
200*ab8db090SAndroid Build Coastguard Worker       default: {
201*ab8db090SAndroid Build Coastguard Worker         return 1;
202*ab8db090SAndroid Build Coastguard Worker       }
203*ab8db090SAndroid Build Coastguard Worker     }
204*ab8db090SAndroid Build Coastguard Worker   }
205*ab8db090SAndroid Build Coastguard Worker   return build(cmdopt.argv + cmdopt.optind,
206*ab8db090SAndroid Build Coastguard Worker       static_cast<std::size_t>(cmdopt.argc - cmdopt.optind));
207*ab8db090SAndroid Build Coastguard Worker }
208