tests/storage/test_chewing_table.cpp

   1 /*
   2  *  libpinyin
   3  *  Library to deal with pinyin.
   4  *
   5  *  Copyright (C) 2011 Peng Wu <alexepico@gmail.com>
   6  *
   7  *  This program is free software; you can redistribute it and/or modify
   8  *  it under the terms of the GNU General Public License as published by
   9  *  the Free Software Foundation; either version 2 of the License, or
  10  *  (at your option) any later version.
  11  *
  12  *  This program is distributed in the hope that it will be useful,
  13  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15  *  GNU General Public License for more details.
  16  *
  17  *  You should have received a copy of the GNU General Public License
  18  *  along with this program; if not, write to the Free Software
  19  *  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
  20  */
  21
  22 #include "timer.h"
  23 #include <string.h>
  24 #include "pinyin_internal.h"
  25 #include "tests_helper.h"
  26
  27 size_t bench_times = 1000;
  28
  29 int main(int argc, char * argv[]) {
  30     pinyin_option_t options = USE_TONE | PINYIN_INCOMPLETE;
  31     ChewingLargeTable largetable(options);
  32     FacadePhraseIndex phrase_index;
  33
  34     if (!load_phrase_table(&largetable, NULL, &phrase_index))
  35         exit(ENOENT);
  36
  37     MemoryChunk * new_chunk = new MemoryChunk;
  38     largetable.store(new_chunk);
  39     largetable.load(new_chunk);
  40
  41     char* linebuf = NULL; size_t size = 0;
  42     while( getline(&linebuf, &size, stdin) ){
  43         linebuf[strlen(linebuf)-1] = '\0';
  44         if ( strcmp ( linebuf, "quit" ) == 0)
  45             break;
  46
  47         FullPinyinParser2 parser;
  48         ChewingKeyVector keys = g_array_new(FALSE, FALSE, sizeof(ChewingKey));
  49         ChewingKeyRestVector key_rests =
  50             g_array_new(FALSE, FALSE, sizeof(ChewingKeyRest));
  51
  52         parser.parse(options, keys, key_rests, linebuf, strlen(linebuf));
  53         if (0 == keys->len) {
  54             fprintf(stderr, "Invalid input.\n");
  55             continue;
  56         }
  57
  58         guint32 start = record_time();
  59         PhraseIndexRanges ranges;
  60         memset(ranges, 0, sizeof(PhraseIndexRanges));
  61
  62         phrase_index.prepare_ranges(ranges);
  63
  64         for (size_t i = 0; i < bench_times; ++i) {
  65             largetable.search(keys->len, (ChewingKey *)keys->data, ranges);
  66         }
  67
  68         phrase_index.clear_ranges(ranges);
  69
  70         print_time(start, bench_times);
  71
  72         largetable.search(keys->len, (ChewingKey *)keys->data, ranges);
  73
  74         for (size_t i = 0; i < PHRASE_INDEX_LIBRARY_COUNT; ++i) {
  75             GArray * & range = ranges[i];
  76             if (!range)
  77                 continue;
  78
  79             if (range->len)
  80                 printf("range items number:%d\n", range->len);
  81
  82             for (size_t k = 0; k < range->len; ++k) {
  83                 PhraseIndexRange * onerange =
  84                     &g_array_index(range, PhraseIndexRange, k);
  85                 printf("start:%d\tend:%d\n", onerange->m_range_begin,
  86                        onerange->m_range_end);
  87
  88                 PhraseItem item;
  89                 for ( phrase_token_t token = onerange->m_range_begin;
  90                       token != onerange->m_range_end; ++token){
  91
  92                     phrase_index.get_phrase_item( token, item);
  93
  94                     /* get phrase string */
  95                     ucs4_t buffer[MAX_PHRASE_LENGTH + 1];
  96                     item.get_phrase_string(buffer);
  97                     char * string = g_ucs4_to_utf8
  98                         ( buffer, item.get_phrase_length(),
  99                           NULL, NULL, NULL);
 100                     printf("%s\t", string);
 101                     g_free(string);
 102
 103                     ChewingKey chewing_buffer[MAX_PHRASE_LENGTH];
 104                     size_t npron = item.get_n_pronunciation();
 105                     guint32 freq;
 106                     for (size_t m = 0; m < npron; ++m){
 107                         item.get_nth_pronunciation(m, chewing_buffer, freq);
 108                         for (size_t n = 0; n < item.get_phrase_length();
 109                              ++n){
 110                             printf("%s'",
 111                                    chewing_buffer[n].get_pinyin_string());
 112                         }
 113                         printf("\b\t%d\t", freq);
 114                     }
 115                 }
 116                 printf("\n");
 117             }
 118             g_array_set_size(range, 0);
 119         }
 120
 121         phrase_index.destroy_ranges(ranges);
 122         g_array_free(keys, TRUE);
 123         g_array_free(key_rests, TRUE);
 124     }
 125
 126     if (linebuf)
 127         free(linebuf);
 128     return 0;
 129 }