tools/rdf2binary.c

   1 /***************************************************************************
   2  *             __________               __   ___.
   3  *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
   4  *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
   5  *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
   6  *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
   7  *                     \/            \/     \/    \/            \/
   8  * $Id$
   9  *
  10  * Copyright (C) 2005 Miika Pekkarinen
  11  *
  12  * All files in this archive are subject to the GNU General Public License.
  13  * See the file COPYING in the source tree root for full license agreement.
  14  *
  15  * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
  16  * KIND, either express or implied.
  17  *
  18  ****************************************************************************/
  19
  20 /*
  21 This tool converts the rdf file to the binary data used in the dict plugin.
  22 */
  23
  24 #include <sys/types.h>
  25 #include <sys/stat.h>
  26 #include <fcntl.h>
  27 #include <string.h>
  28 #include <stdio.h>
  29
  30 /* maximum word lenght, has to be the same in dict.c */
  31 #define WORDLEN 32
  32
  33 /* struckt packing */
  34 #ifdef __GNUC__
  35 #define STRUCT_PACKED __attribute__((packed))
  36 #else
  37 #define STRUCT_PACKED
  38 #pragma pack (push, 2)
  39 #endif
  40
  41
  42 struct word
  43 {
  44     char word[WORDLEN];
  45     long offset;
  46 } STRUCT_PACKED;
  47
  48 /* convert offsets here, not on device. */
  49 long reverse (long N) {
  50     unsigned char B[4];
  51     B[0] = (N & 0x000000FF) >> 0;
  52     B[1] = (N & 0x0000FF00) >> 8;
  53     B[2] = (N & 0x00FF0000) >> 16;
  54     B[3] = (N & 0xFF000000) >> 24;
  55     return ((B[0] << 24) | (B[1] << 16) | (B[2] << 8) | (B[3] << 0));
  56 }
  57
  58
  59 int main()
  60 {
  61     FILE *in, *idx_out, *desc_out;
  62     struct word w;
  63     char buf[10000];
  64     long cur_offset = 0;
  65
  66     in = fopen("dict.preparsed", "r");
  67     idx_out = fopen("dict.index", "wb");
  68     desc_out = fopen("dict.desc", "wb");
  69
  70     if (in == NULL || idx_out == NULL || desc_out == NULL)
  71     {
  72         fprintf(stderr, "Error: Some files couldn't be opened\n");
  73         return 1;
  74     }
  75
  76     while (fgets(buf, sizeof buf, in) != NULL)
  77     {
  78         /* It is safe to use strtok here */
  79         const char *word = strtok(buf, "\t");
  80         const char *desc = strtok(NULL, "\t");
  81
  82         if (word == NULL || desc == NULL)
  83         {
  84             fprintf(stderr, "Parse error!\n");
  85             fprintf(stderr, "word: %s\ndesc: %s\n", word, desc);
  86
  87             return 2;
  88         }
  89
  90         /* We will null-terminate the words */
  91         strncpy(w.word, word, WORDLEN - 1);
  92         w.offset = reverse(cur_offset);
  93         fwrite(&w, sizeof(struct word), 1, idx_out);
  94
  95         while (1)
  96         {
  97             int len = strlen(desc);
  98             cur_offset += len;
  99             fwrite(desc, len, 1, desc_out);
 100
 101             desc = strtok(NULL, "\t");
 102             if (desc == NULL)
 103                 break ;
 104
 105             cur_offset++;
 106             fwrite("\n", 1, 1, desc_out);
 107
 108         }
 109     }
 110
 111     return 0;
 112 }
 113