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  * This program is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU General Public License
  14  * as published by the Free Software Foundation; either version 2
  15  * of the License, or (at your option) any later version.
  16  *
  17  * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
  18  * KIND, either express or implied.
  19  *
  20  ****************************************************************************/
  21
  22 /*
  23 This tool converts the rdf file to the binary data used in the dict plugin.
  24 */
  25
  26 #include <sys/types.h>
  27 #include <sys/stat.h>
  28 #include <fcntl.h>
  29 #include <string.h>
  30 #include <stdio.h>
  31
  32 /* maximum word lenght, has to be the same in dict.c */
  33 #define WORDLEN 32
  34
  35 /* struckt packing */
  36 #ifdef __GNUC__
  37 #define STRUCT_PACKED __attribute__((packed))
  38 #else
  39 #define STRUCT_PACKED
  40 #pragma pack (push, 2)
  41 #endif
  42
  43
  44 struct word
  45 {
  46     char word[WORDLEN];
  47     long offset;
  48 } STRUCT_PACKED;
  49
  50 /* convert offsets here, not on device. */
  51 long reverse (long N) {
  52     unsigned char B[4];
  53     B[0] = (N & 0x000000FF) >> 0;
  54     B[1] = (N & 0x0000FF00) >> 8;
  55     B[2] = (N & 0x00FF0000) >> 16;
  56     B[3] = (N & 0xFF000000) >> 24;
  57     return ((B[0] << 24) | (B[1] << 16) | (B[2] << 8) | (B[3] << 0));
  58 }
  59
  60
  61 int main()
  62 {
  63     FILE *in, *idx_out, *desc_out;
  64     struct word w;
  65     char buf[10000];
  66     long cur_offset = 0;
  67
  68     in = fopen("dict.preparsed", "r");
  69     idx_out = fopen("dict.index", "wb");
  70     desc_out = fopen("dict.desc", "wb");
  71
  72     if (in == NULL || idx_out == NULL || desc_out == NULL)
  73     {
  74         fprintf(stderr, "Error: Some files couldn't be opened\n");
  75         return 1;
  76     }
  77
  78     while (fgets(buf, sizeof buf, in) != NULL)
  79     {
  80         /* It is safe to use strtok here */
  81         const char *word = strtok(buf, "\t");
  82         const char *desc = strtok(NULL, "\t");
  83
  84         if (word == NULL || desc == NULL)
  85         {
  86             fprintf(stderr, "Parse error!\n");
  87             fprintf(stderr, "word: %s\ndesc: %s\n", word, desc);
  88
  89             return 2;
  90         }
  91
  92         /* We will null-terminate the words */
  93         strncpy(w.word, word, WORDLEN - 1);
  94         w.offset = reverse(cur_offset);
  95         fwrite(&w, sizeof(struct word), 1, idx_out);
  96
  97         while (1)
  98         {
  99             int len = strlen(desc);
 100             cur_offset += len;
 101             fwrite(desc, len, 1, desc_out);
 102
 103             desc = strtok(NULL, "\t");
 104             if (desc == NULL)
 105                 break ;
 106
 107             cur_offset++;
 108             fwrite("\n", 1, 1, desc_out);
 109
 110         }
 111     }
 112
 113     return 0;
 114 }
 115