i18npool/source/breakiterator/gendict.cxx

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2008 by Sun Microsystems, Inc.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * $RCSfile: gendict.cxx,v $
  10  * $Revision: 1.12 $
  11  *
  12  * This file is part of OpenOffice.org.
  13  *
  14  * OpenOffice.org is free software: you can redistribute it and/or modify
  15  * it under the terms of the GNU Lesser General Public License version 3
  16  * only, as published by the Free Software Foundation.
  17  *
  18  * OpenOffice.org is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21  * GNU Lesser General Public License version 3 for more details
  22  * (a copy is included in the LICENSE file that accompanied this code).
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * version 3 along with OpenOffice.org.  If not, see
  26  * <http://www.openoffice.org/license.html>
  27  * for a copy of the LGPLv3 License.
  28  *
  29  ************************************************************************/
  30
  31 // MARKER(update_precomp.py): autogen include statement, do not remove
  32 #include "precompiled_i18npool.hxx"
  33
  34 #include <stdio.h>
  35 #include <string.h>
  36 #include <stdlib.h>
  37 #include <sal/main.h>
  38 #include <sal/types.h>
  39 #include <rtl/strbuf.hxx>
  40 #include <rtl/ustring.hxx>
  41
  42 using namespace ::rtl;
  43
  44 /* Main Procedure */
  45
  46 SAL_IMPLEMENT_MAIN_WITH_ARGS(argc, argv)
  47 {
  48     FILE *sfp, *cfp;
  49
  50     if (argc < 3) exit(-1);
  51
  52     sfp = fopen(argv[1], "rb"); // open the source file for read;
  53     if (sfp == NULL)
  54     {
  55         printf("Open the dictionary source file failed.");
  56         return -1;
  57     }
  58
  59     // create the C source file to write
  60     cfp = fopen(argv[2], "wb");
  61     if (cfp == NULL) {
  62         fclose(sfp);
  63         printf("Can't create the C source file.");
  64         return -1;
  65     }
  66
  67     fprintf(cfp, "/*\n");
  68     fprintf(cfp, " * Copyright(c) 1999 - 2000, Sun Microsystems, Inc.\n");
  69     fprintf(cfp, " * All Rights Reserved.\n");
  70     fprintf(cfp, " */\n\n");
  71     fprintf(cfp, "/* !!!The file is generated automatically. DONOT edit the file manually!!! */\n\n");
  72     fprintf(cfp, "#include <sal/types.h>\n\n");
  73     fprintf(cfp, "extern \"C\" {\n");
  74
  75     sal_Int32 count, i, j;
  76     sal_Int32 lenArrayCurr = 0, lenArrayCount = 0, lenArrayLen = 0, *lenArray = NULL, charArray[0x10000];
  77     sal_Bool exist[0x10000];
  78     for (i = 0; i < 0x10000; i++) {
  79         exist[i] = sal_False;
  80         charArray[i] = 0;
  81     }
  82
  83     // generate main dict. data array
  84     fprintf(cfp, "static const sal_Unicode dataArea[] = {");
  85     sal_Char str[1024];
  86     sal_Unicode current = 0;
  87     count = 0;
  88     while (fgets(str, 1024, sfp)) {
  89         // input file is in UTF-8 encoding
  90         // don't convert last new line character to Ostr.
  91         OUString Ostr((const sal_Char *)str, strlen(str) - 1, RTL_TEXTENCODING_UTF8);
  92         const sal_Unicode *u = Ostr.getStr();
  93
  94         sal_Int32 len = Ostr.getLength();
  95
  96         i=0;
  97         Ostr.iterateCodePoints(&i, 1);
  98         if (len == i) continue; // skip one character word
  99
 100         if (*u != current) {
 101         if (*u < current)
 102         printf("u %x, current %x, count %d, lenArrayCount %d\n", *u, current,
 103                     sal::static_int_cast<int>(count), sal::static_int_cast<int>(lenArrayCount));
 104         current = *u;
 105         charArray[current] = lenArrayCount;
 106         }
 107
 108         if (lenArrayLen <= lenArrayCount+1)
 109         lenArray = (sal_Int32*) realloc(lenArray, (lenArrayLen += 1000) * sizeof(sal_Int32));
 110         lenArray[lenArrayCount++] = lenArrayCurr;
 111
 112         exist[u[0]] = sal_True;
 113         for (i = 1; i < len; i++) {             // start from second character,
 114         exist[u[i]] = sal_True;         // since the first character is captured in charArray.
 115         lenArrayCurr++;
 116         if ((count++) % 0x10 == 0)
 117             fprintf(cfp, "\n\t");
 118         fprintf(cfp, "0x%04x, ", u[i]);
 119         }
 120     }
 121     lenArray[lenArrayCount++] = lenArrayCurr; // store last ending pointer
 122     charArray[current+1] = lenArrayCount;
 123     fprintf(cfp, "\n};\n");
 124
 125     // generate lenArray
 126     fprintf(cfp, "static const sal_Int32 lenArray[] = {\n\t");
 127     count = 1;
 128     fprintf(cfp, "0x%x, ", 0); // insert one slat for skipping 0 in index2 array.
 129     for (i = 0; i < lenArrayCount; i++) {
 130         fprintf(cfp, "0x%lx, ", static_cast<long unsigned int>(lenArray[i]));
 131         if (count == 0xf) {
 132         count = 0;
 133         fprintf(cfp, "\n\t");
 134         } else count++;
 135     }
 136     fprintf(cfp, "\n};\n");
 137
 138     free(lenArray);
 139
 140     // generate index1 array
 141     fprintf (cfp, "static const sal_Int16 index1[] = {\n\t");
 142     sal_Int16 set[0x100];
 143     count = 0;
 144     for (i = 0; i < 0x100; i++) {
 145         for (j = 0; j < 0x100; j++)
 146         if (charArray[(i*0x100) + j] != 0)
 147             break;
 148
 149         fprintf(cfp, "0x%02x, ", set[i] = (j < 0x100 ? sal::static_int_cast<sal_Int16>(count++) : 0xff));
 150         if ((i+1) % 0x10 == 0)
 151         fprintf (cfp, "\n\t");
 152     }
 153     fprintf (cfp, "};\n");
 154
 155     // generate index2 array
 156     fprintf (cfp, "static const sal_Int32 index2[] = {\n\t");
 157     sal_Int32 prev = 0;
 158     for (i = 0; i < 0x100; i++) {
 159         if (set[i] != 0xff) {
 160         for (j = 0; j < 0x100; j++) {
 161             sal_Int32 k = (i*0x100) + j;
 162             if (prev != 0 && charArray[k] == 0) {
 163             for (k++; k < 0x10000; k++)
 164                 if (charArray[k] != 0)
 165                 break;
 166             }
 167             prev = charArray[(i*0x100) + j];
 168             fprintf(
 169                 cfp, "0x%lx, ",
 170                 sal::static_int_cast< unsigned long >(
 171                     k < 0x10000 ? charArray[k] + 1 : 0));
 172             if ((j+1) % 0x10 == 0)
 173             fprintf (cfp, "\n\t");
 174         }
 175         fprintf (cfp, "\n\t");
 176         }
 177     }
 178     fprintf (cfp, "\n};\n");
 179
 180     // generate existMark array
 181     count = 0;
 182     fprintf (cfp, "static const sal_uInt8 existMark[] = {\n\t");
 183     for (i = 0; i < 0x1FFF; i++) {
 184         sal_uInt8 bit = 0;
 185         for (j = 0; j < 8; j++)
 186         if (exist[i * 8 + j])
 187             bit |= 1 << j;
 188         fprintf(cfp, "0x%02x, ", bit);
 189         if (count == 0xf) {
 190         count = 0;
 191         fprintf(cfp, "\n\t");
 192         } else count++;
 193     }
 194     fprintf (cfp, "\n};\n");
 195
 196     // create function to return arrays
 197     fprintf (cfp, "\tconst sal_uInt8* getExistMark() { return existMark; }\n");
 198     fprintf (cfp, "\tconst sal_Int16* getIndex1() { return index1; }\n");
 199     fprintf (cfp, "\tconst sal_Int32* getIndex2() { return index2; }\n");
 200     fprintf (cfp, "\tconst sal_Int32* getLenArray() { return lenArray; }\n");
 201     fprintf (cfp, "\tconst sal_Unicode* getDataArea() { return dataArea; }\n");
 202     fprintf (cfp, "}\n");
 203
 204     fclose(sfp);
 205     fclose(cfp);
 206
 207     return 0;
 208 }       // End of main