lib/gentranslit.c

   1 /* Copyright (C) 1999-2001 Free Software Foundation, Inc.
   2    This file is part of the GNU LIBICONV Library.
   3
   4    The GNU LIBICONV Library is free software; you can redistribute it
   5    and/or modify it under the terms of the GNU Library General Public
   6    License as published by the Free Software Foundation; either version 2
   7    of the License, or (at your option) any later version.
   8
   9    The GNU LIBICONV Library is distributed in the hope that it will be
  10    useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12    Library General Public License for more details.
  13
  14    You should have received a copy of the GNU Library General Public
  15    License along with the GNU LIBICONV Library; see the file COPYING.LIB.
  16    If not, write to the Free Software Foundation, Inc., 59 Temple Place -
  17    Suite 330, Boston, MA 02111-1307, USA.  */
  18
  19 /*
  20  * Generates a table of small strings, used for transliteration, from a table
  21  * containing lines of the form
  22  *   Unicode <tab> utf-8 replacement <tab> # comment
  23  */
  24
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27 #include <stdbool.h>
  28
  29 int main (int argc, char *argv[])
  30 {
  31   unsigned short data[0x100000];
  32   int uni2index[0x10000];
  33   int index;
  34
  35   if (argc != 1)
  36     exit(1);
  37
  38   printf("/*\n");
  39   printf(" * Copyright (C) 1999-2001 Free Software Foundation, Inc.\n");
  40   printf(" * This file is part of the GNU LIBICONV Library.\n");
  41   printf(" *\n");
  42   printf(" * The GNU LIBICONV Library is free software; you can redistribute it\n");
  43   printf(" * and/or modify it under the terms of the GNU Library General Public\n");
  44   printf(" * License as published by the Free Software Foundation; either version 2\n");
  45   printf(" * of the License, or (at your option) any later version.\n");
  46   printf(" *\n");
  47   printf(" * The GNU LIBICONV Library is distributed in the hope that it will be\n");
  48   printf(" * useful, but WITHOUT ANY WARRANTY; without even the implied warranty of\n");
  49   printf(" * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU\n");
  50   printf(" * Library General Public License for more details.\n");
  51   printf(" *\n");
  52   printf(" * You should have received a copy of the GNU Library General Public\n");
  53   printf(" * License along with the GNU LIBICONV Library; see the file COPYING.LIB.\n");
  54   printf(" * If not, write to the Free Software Foundation, Inc., 59 Temple Place -\n");
  55   printf(" * Suite 330, Boston, MA 02111-1307, USA.\n");
  56   printf(" */\n");
  57   printf("\n");
  58   printf("/*\n");
  59   printf(" * Transliteration table\n");
  60   printf(" */\n");
  61   printf("\n");
  62   {
  63     int c;
  64     int j;
  65     for (j = 0; j < 0x10000; j++)
  66       uni2index[j] = -1;
  67     index = 0;
  68     for (;;) {
  69       c = getc(stdin);
  70       if (c == EOF)
  71         break;
  72       if (c == '#') {
  73         do { c = getc(stdin); } while (!(c == EOF || c == '\n'));
  74         continue;
  75       }
  76       ungetc(c,stdin);
  77       if (scanf("%x",&j) != 1)
  78         exit(1);
  79       c = getc(stdin);
  80       if (c != '\t')
  81         exit(1);
  82       for (;;) {
  83         c = getc(stdin);
  84         if (c == EOF || c == '\n')
  85           exit(1);
  86         if (c == '\t')
  87           break;
  88         if (uni2index[j] < 0) {
  89           uni2index[j] = index;
  90           data[index++] = 0;
  91         }
  92         if (c >= 0x80) {
  93           /* Finish reading an UTF-8 character. */
  94           if (c < 0xc0)
  95             exit(1);
  96           else {
  97             unsigned int i = (c < 0xe0 ? 2 : c < 0xf0 ? 3 : c < 0xf8 ? 4 : c < 0xfc ? 5 : 6);
  98             c &= (1 << (8-i)) - 1;
  99             while (--i > 0) {
 100               int cc = getc(stdin);
 101               if (!(cc >= 0x80 && cc < 0xc0))
 102                 exit(1);
 103               c <<= 6; c |= (cc & 0x3f);
 104             }
 105           }
 106         }
 107         data[index++] = (unsigned short) c;
 108       }
 109       if (uni2index[j] >= 0)
 110         data[uni2index[j]] = index - uni2index[j] - 1;
 111       do { c = getc(stdin); } while (!(c == EOF || c == '\n'));
 112     }
 113   }
 114   printf("static const unsigned short translit_data[%d] = {",index);
 115   {
 116     int i;
 117     for (i = 0; i < index; i++) {
 118       if (data[i] < 32)
 119         printf("\n %3d,",data[i]);
 120       else if (data[i] == '\'')
 121         printf("'\\'',");
 122       else if (data[i] == '\\')
 123         printf("'\\\\',");
 124       else if (data[i] < 127)
 125         printf(" '%c',",data[i]);
 126       else if (data[i] < 256)
 127         printf("0x%02X,",data[i]);
 128       else
 129         printf("0x%04X,",data[i]);
 130     }
 131     printf("\n};\n");
 132   }
 133   printf("\n");
 134   {
 135     bool pages[0x100];
 136     int line[0x2000];
 137     int tableno;
 138     struct { int minline; int maxline; int usecount; const char* suffix; } tables[0x2000];
 139     int i, j, p, j1, j2, t;
 140
 141     for (p = 0; p < 0x100; p++)
 142       pages[p] = false;
 143     for (j = 0; j < 0x10000; j++)
 144       if (uni2index[j] >= 0)
 145         pages[j>>8] = true;
 146     for (j1 = 0; j1 < 0x2000; j1++) {
 147       bool all_invalid = true;
 148       for (j2 = 0; j2 < 8; j2++) {
 149         j = 8*j1+j2;
 150         if (uni2index[j] >= 0)
 151           all_invalid = false;
 152       }
 153       if (all_invalid)
 154         line[j1] = -1;
 155       else
 156         line[j1] = 0;
 157     }
 158     tableno = 0;
 159     for (j1 = 0; j1 < 0x2000; j1++) {
 160       if (line[j1] >= 0) {
 161         if (tableno > 0
 162             && ((j1 > 0 && line[j1-1] == tableno-1)
 163                 || ((tables[tableno-1].maxline >> 5) == (j1 >> 5)
 164                     && j1 - tables[tableno-1].maxline <= 8))) {
 165           line[j1] = tableno-1;
 166           tables[tableno-1].maxline = j1;
 167         } else {
 168           tableno++;
 169           line[j1] = tableno-1;
 170           tables[tableno-1].minline = tables[tableno-1].maxline = j1;
 171         }
 172       }
 173     }
 174     for (t = 0; t < tableno; t++) {
 175       tables[t].usecount = 0;
 176       j1 = 8*tables[t].minline;
 177       j2 = 8*(tables[t].maxline+1);
 178       for (j = j1; j < j2; j++)
 179         if (uni2index[j] >= 0)
 180           tables[t].usecount++;
 181     }
 182     for (t = 0, p = -1, i = 0; t < tableno; t++) {
 183       if (tables[t].usecount > 1) {
 184         char* s;
 185         if (p == tables[t].minline >> 5) {
 186           s = (char*) malloc(5+1);
 187           sprintf(s, "%02x_%d", p, ++i);
 188         } else {
 189           p = tables[t].minline >> 5;
 190           s = (char*) malloc(2+1);
 191           sprintf(s, "%02x", p);
 192         }
 193         tables[t].suffix = s;
 194       } else
 195         tables[t].suffix = NULL;
 196     }
 197     {
 198       p = -1;
 199       for (t = 0; t < tableno; t++)
 200         if (tables[t].usecount > 1) {
 201           p = 0;
 202           printf("static const short translit_page%s[%d] = {\n", tables[t].suffix, 8*(tables[t].maxline-tables[t].minline+1));
 203           for (j1 = tables[t].minline; j1 <= tables[t].maxline; j1++) {
 204             if ((j1 % 0x20) == 0 && j1 > tables[t].minline)
 205               printf("  /* 0x%04x */\n", 8*j1);
 206             printf(" ");
 207             for (j2 = 0; j2 < 8; j2++) {
 208               j = 8*j1+j2;
 209               printf(" %4d,", uni2index[j]);
 210             }
 211             printf(" /* 0x%02x-0x%02x */\n", 8*(j1 % 0x20), 8*(j1 % 0x20)+7);
 212           }
 213           printf("};\n");
 214         }
 215       if (p >= 0)
 216         printf("\n");
 217     }
 218     printf("#define translit_index(wc) \\\n  (");
 219     for (j1 = 0; j1 < 0x2000;) {
 220       t = line[j1];
 221       for (j2 = j1; j2 < 0x2000 && line[j2] == t; j2++);
 222       if (t >= 0) {
 223         if (j1 != tables[t].minline) abort();
 224         if (j2 > tables[t].maxline+1) abort();
 225         j2 = tables[t].maxline+1;
 226       }
 227       if (t == -1) {
 228       } else {
 229         if (t >= 0 && tables[t].usecount == 0) abort();
 230         if (t >= 0 && tables[t].usecount == 1) {
 231           if (j2 != j1+1) abort();
 232           for (j = 8*j1; j < 8*j2; j++)
 233             if (uni2index[j] >= 0) {
 234               printf("wc == 0x%04x ? %d", j, uni2index[j]);
 235               break;
 236             }
 237         } else {
 238           if (j1 == 0) {
 239             printf("wc < 0x%04x", 8*j2);
 240           } else {
 241             printf("wc >= 0x%04x && wc < 0x%04x", 8*j1, 8*j2);
 242           }
 243           printf(" ? translit_page%s[wc", tables[t].suffix);
 244           if (tables[t].minline > 0)
 245             printf("-0x%04x", 8*j1);
 246           printf("]");
 247         }
 248         printf(" : \\\n   ");
 249       }
 250       j1 = j2;
 251     }
 252     printf("-1)\n");
 253   }
 254
 255   fflush(stdout);
 256   if (ferror(stdout))
 257     exit(1);
 258   exit(0);
 259 }