lib/charsets.c

   1 /*
   2    Text conversion from one charset to another.
   3
   4    Copyright (C) 2001-2024
   5    Free Software Foundation, Inc.
   6
   7    Written by:
   8    Walery Studennikov <despair@sama.ru>
   9
  10    This file is part of the Midnight Commander.
  11
  12    The Midnight Commander is free software: you can redistribute it
  13    and/or modify it under the terms of the GNU General Public License as
  14    published by the Free Software Foundation, either version 3 of the License,
  15    or (at your option) any later version.
  16
  17    The Midnight Commander is distributed in the hope that it will be useful,
  18    but WITHOUT ANY WARRANTY; without even the implied warranty of
  19    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20    GNU General Public License for more details.
  21
  22    You should have received a copy of the GNU General Public License
  23    along with this program.  If not, see <http://www.gnu.org/licenses/>.
  24  */
  25
  26 /** \file charsets.c
  27  *  \brief Source: Text conversion from one charset to another
  28  */
  29
  30 #include <config.h>
  31
  32 #include <stdio.h>
  33 #include <stdlib.h>
  34 #include <string.h>
  35
  36 #include "lib/global.h"
  37 #include "lib/strutil.h"        /* utf-8 functions */
  38 #include "lib/fileloc.h"
  39 #include "lib/util.h"           /* whitespace() */
  40
  41 #include "lib/charsets.h"
  42
  43 /*** global variables ****************************************************************************/
  44
  45 GPtrArray *codepages = NULL;
  46
  47 unsigned char conv_displ[256];
  48 unsigned char conv_input[256];
  49
  50 const char *cp_display = NULL;
  51 const char *cp_source = NULL;
  52
  53 /*** file scope macro definitions ****************************************************************/
  54
  55 #define UNKNCHAR '\001'
  56
  57 #define OTHER_8BIT "Other_8_bit"
  58
  59 /*** file scope type declarations ****************************************************************/
  60
  61 /*** forward declarations (file scope functions) *************************************************/
  62
  63 /*** file scope variables ************************************************************************/
  64
  65 /* --------------------------------------------------------------------------------------------- */
  66 /*** file scope functions ************************************************************************/
  67 /* --------------------------------------------------------------------------------------------- */
  68
  69 static codepage_desc *
  70 new_codepage_desc (const char *id, const char *name)
  71 {
  72     codepage_desc *desc;
  73
  74     desc = g_new (codepage_desc, 1);
  75     desc->id = g_strdup (id);
  76     desc->name = g_strdup (name);
  77
  78     return desc;
  79 }
  80
  81 /* --------------------------------------------------------------------------------------------- */
  82
  83 static void
  84 free_codepage_desc (gpointer data)
  85 {
  86     codepage_desc *desc = (codepage_desc *) data;
  87
  88     g_free (desc->id);
  89     g_free (desc->name);
  90     g_free (desc);
  91 }
  92
  93 /* --------------------------------------------------------------------------------------------- */
  94 /* returns display codepage */
  95
  96 static void
  97 load_codepages_list_from_file (GPtrArray ** list, const char *fname)
  98 {
  99     FILE *f;
 100     char buf[BUF_MEDIUM];
 101     char *default_codepage = NULL;
 102
 103     f = fopen (fname, "r");
 104     if (f == NULL)
 105         return;
 106
 107     while (fgets (buf, sizeof buf, f) != NULL)
 108     {
 109         /* split string into id and cpname */
 110         char *p = buf;
 111         size_t buflen;
 112
 113         if (*p == '\n' || *p == '\0' || *p == '#')
 114             continue;
 115
 116         buflen = strlen (buf);
 117
 118         if (buflen != 0 && buf[buflen - 1] == '\n')
 119             buf[buflen - 1] = '\0';
 120         while (*p != '\0' && !whitespace (*p))
 121             ++p;
 122         if (*p == '\0')
 123             goto fail;
 124
 125         *p++ = '\0';
 126         g_strstrip (p);
 127         if (*p == '\0')
 128             goto fail;
 129
 130         if (strcmp (buf, "default") == 0)
 131             default_codepage = g_strdup (p);
 132         else
 133         {
 134             const char *id = buf;
 135
 136             if (*list == NULL)
 137             {
 138                 *list = g_ptr_array_new_full (16, free_codepage_desc);
 139                 g_ptr_array_add (*list, new_codepage_desc (id, p));
 140             }
 141             else
 142             {
 143                 unsigned int i;
 144
 145                 /* whether id is already present in list */
 146                 /* if yes, overwrite description */
 147                 for (i = 0; i < (*list)->len; i++)
 148                 {
 149                     codepage_desc *desc;
 150
 151                     desc = (codepage_desc *) g_ptr_array_index (*list, i);
 152
 153                     if (strcmp (id, desc->id) == 0)
 154                     {
 155                         /* found */
 156                         g_free (desc->name);
 157                         desc->name = g_strdup (p);
 158                         break;
 159                     }
 160                 }
 161
 162                 /* not found */
 163                 if (i == (*list)->len)
 164                     g_ptr_array_add (*list, new_codepage_desc (id, p));
 165             }
 166         }
 167     }
 168
 169     if (default_codepage != NULL)
 170     {
 171         mc_global.display_codepage = get_codepage_index (default_codepage);
 172         g_free (default_codepage);
 173     }
 174
 175   fail:
 176     fclose (f);
 177 }
 178
 179 /* --------------------------------------------------------------------------------------------- */
 180
 181 static char
 182 translate_character (GIConv cd, char c)
 183 {
 184     gchar *tmp_buff = NULL;
 185     gsize bytes_read, bytes_written = 0;
 186     const char *ibuf = &c;
 187     char ch = UNKNCHAR;
 188     int ibuflen = 1;
 189
 190     tmp_buff = g_convert_with_iconv (ibuf, ibuflen, cd, &bytes_read, &bytes_written, NULL);
 191     if (tmp_buff != NULL)
 192         ch = tmp_buff[0];
 193     g_free (tmp_buff);
 194     return ch;
 195 }
 196
 197 /* --------------------------------------------------------------------------------------------- */
 198 /*** public functions ****************************************************************************/
 199 /* --------------------------------------------------------------------------------------------- */
 200
 201 void
 202 load_codepages_list (void)
 203 {
 204     char *fname;
 205
 206     /* 1: try load /usr/share/mc/mc.charsets */
 207     fname = g_build_filename (mc_global.share_data_dir, CHARSETS_LIST, (char *) NULL);
 208     load_codepages_list_from_file (&codepages, fname);
 209     g_free (fname);
 210
 211     /* 2: try load /etc/mc/mc.charsets */
 212     fname = g_build_filename (mc_global.sysconfig_dir, CHARSETS_LIST, (char *) NULL);
 213     load_codepages_list_from_file (&codepages, fname);
 214     g_free (fname);
 215
 216     if (codepages == NULL)
 217     {
 218         /* files are not found, add default codepage */
 219         fprintf (stderr, "%s\n", _("Warning: cannot load codepages list"));
 220
 221         codepages = g_ptr_array_new_with_free_func (free_codepage_desc);
 222         g_ptr_array_add (codepages, new_codepage_desc (DEFAULT_CHARSET, _("7-bit ASCII")));
 223     }
 224 }
 225
 226 /* --------------------------------------------------------------------------------------------- */
 227
 228 void
 229 free_codepages_list (void)
 230 {
 231     g_ptr_array_free (codepages, TRUE);
 232     /* NULL-ize pointer to make unit tests happy */
 233     codepages = NULL;
 234 }
 235
 236 /* --------------------------------------------------------------------------------------------- */
 237
 238 const char *
 239 get_codepage_id (const int n)
 240 {
 241     return (n < 0) ? OTHER_8BIT : ((codepage_desc *) g_ptr_array_index (codepages, n))->id;
 242 }
 243
 244 /* --------------------------------------------------------------------------------------------- */
 245
 246 int
 247 get_codepage_index (const char *id)
 248 {
 249     size_t i;
 250
 251     if (codepages == NULL)
 252         return -1;
 253     if (strcmp (id, OTHER_8BIT) == 0)
 254         return -1;
 255     for (i = 0; i < codepages->len; i++)
 256         if (strcmp (id, ((codepage_desc *) g_ptr_array_index (codepages, i))->id) == 0)
 257             return i;
 258     return -1;
 259 }
 260
 261 /* --------------------------------------------------------------------------------------------- */
 262 /** Check if specified encoding can be used in mc.
 263  * @param encoding name of encoding
 264  * @return TRUE if encoding is supported by mc, FALSE otherwise
 265  */
 266
 267 gboolean
 268 is_supported_encoding (const char *encoding)
 269 {
 270     gboolean result = FALSE;
 271     guint t;
 272
 273     for (t = 0; t < codepages->len; t++)
 274     {
 275         const char *id;
 276
 277         id = ((codepage_desc *) g_ptr_array_index (codepages, t))->id;
 278         result |= (g_ascii_strncasecmp (encoding, id, strlen (id)) == 0);
 279     }
 280
 281     return result;
 282 }
 283
 284 /* --------------------------------------------------------------------------------------------- */
 285
 286 char *
 287 init_translation_table (int cpsource, int cpdisplay)
 288 {
 289     int i;
 290     GIConv cd;
 291
 292     /* Fill inpit <-> display tables */
 293
 294     if (cpsource < 0 || cpdisplay < 0 || cpsource == cpdisplay)
 295     {
 296         for (i = 0; i <= 255; ++i)
 297         {
 298             conv_displ[i] = i;
 299             conv_input[i] = i;
 300         }
 301         cp_source = cp_display;
 302         return NULL;
 303     }
 304
 305     for (i = 0; i <= 127; ++i)
 306     {
 307         conv_displ[i] = i;
 308         conv_input[i] = i;
 309     }
 310     cp_source = ((codepage_desc *) g_ptr_array_index (codepages, cpsource))->id;
 311     cp_display = ((codepage_desc *) g_ptr_array_index (codepages, cpdisplay))->id;
 312
 313     /* display <- inpit table */
 314
 315     cd = g_iconv_open (cp_display, cp_source);
 316     if (cd == INVALID_CONV)
 317         return g_strdup_printf (_("Cannot translate from %s to %s"), cp_source, cp_display);
 318
 319     for (i = 128; i <= 255; ++i)
 320         conv_displ[i] = translate_character (cd, i);
 321
 322     g_iconv_close (cd);
 323
 324     /* inpit <- display table */
 325
 326     cd = g_iconv_open (cp_source, cp_display);
 327     if (cd == INVALID_CONV)
 328         return g_strdup_printf (_("Cannot translate from %s to %s"), cp_display, cp_source);
 329
 330     for (i = 128; i <= 255; ++i)
 331     {
 332         unsigned char ch;
 333         ch = translate_character (cd, i);
 334         conv_input[i] = (ch == UNKNCHAR) ? i : ch;
 335     }
 336
 337     g_iconv_close (cd);
 338
 339     return NULL;
 340 }
 341
 342 /* --------------------------------------------------------------------------------------------- */
 343
 344 void
 345 convert_to_display (char *str)
 346 {
 347     if (str != NULL)
 348         for (; *str != '\0'; str++)
 349             *str = conv_displ[(unsigned char) *str];
 350 }
 351
 352 /* --------------------------------------------------------------------------------------------- */
 353
 354 GString *
 355 str_nconvert_to_display (const char *str, int len)
 356 {
 357     GString *buff;
 358     GIConv conv;
 359
 360     if (str == NULL)
 361         return NULL;
 362
 363     if (cp_display == cp_source)
 364         return g_string_new (str);
 365
 366     conv = str_crt_conv_from (cp_source);
 367
 368     buff = g_string_new ("");
 369     str_nconvert (conv, str, len, buff);
 370     str_close_conv (conv);
 371     return buff;
 372 }
 373
 374 /* --------------------------------------------------------------------------------------------- */
 375
 376 void
 377 convert_from_input (char *str)
 378 {
 379     if (str != NULL)
 380         for (; *str != '\0'; str++)
 381             *str = conv_input[(unsigned char) *str];
 382 }
 383
 384 /* --------------------------------------------------------------------------------------------- */
 385
 386 GString *
 387 str_nconvert_to_input (const char *str, int len)
 388 {
 389     GString *buff;
 390     GIConv conv;
 391
 392     if (str == NULL)
 393         return NULL;
 394
 395     if (cp_display == cp_source)
 396         return g_string_new (str);
 397
 398     conv = str_crt_conv_to (cp_source);
 399
 400     buff = g_string_new ("");
 401     str_nconvert (conv, str, len, buff);
 402     str_close_conv (conv);
 403     return buff;
 404 }
 405
 406 /* --------------------------------------------------------------------------------------------- */
 407
 408 unsigned char
 409 convert_from_utf_to_current (const char *str)
 410 {
 411     unsigned char buf_ch[UTF8_CHAR_LEN + 1];
 412     unsigned char ch = '.';
 413     GIConv conv;
 414     const char *cp_to;
 415
 416     if (str == NULL)
 417         return '.';
 418
 419     cp_to = get_codepage_id (mc_global.source_codepage);
 420     conv = str_crt_conv_to (cp_to);
 421
 422     if (conv != INVALID_CONV)
 423     {
 424         switch (str_translate_char (conv, str, -1, (char *) buf_ch, sizeof (buf_ch)))
 425         {
 426         case ESTR_SUCCESS:
 427             ch = buf_ch[0];
 428             break;
 429         case ESTR_PROBLEM:
 430         case ESTR_FAILURE:
 431             ch = '.';
 432             break;
 433         default:
 434             break;
 435         }
 436         str_close_conv (conv);
 437     }
 438
 439     return ch;
 440 }
 441
 442 /* --------------------------------------------------------------------------------------------- */
 443
 444 unsigned char
 445 convert_from_utf_to_current_c (int input_char, GIConv conv)
 446 {
 447     unsigned char str[UTF8_CHAR_LEN + 1];
 448     unsigned char buf_ch[UTF8_CHAR_LEN + 1];
 449     unsigned char ch = '.';
 450     int res;
 451
 452     res = g_unichar_to_utf8 (input_char, (char *) str);
 453     if (res == 0)
 454         return ch;
 455
 456     str[res] = '\0';
 457
 458     switch (str_translate_char (conv, (char *) str, -1, (char *) buf_ch, sizeof (buf_ch)))
 459     {
 460     case ESTR_SUCCESS:
 461         ch = buf_ch[0];
 462         break;
 463     case ESTR_PROBLEM:
 464     case ESTR_FAILURE:
 465         ch = '.';
 466         break;
 467     default:
 468         break;
 469     }
 470
 471     return ch;
 472 }
 473
 474 /* --------------------------------------------------------------------------------------------- */
 475
 476 int
 477 convert_from_8bit_to_utf_c (char input_char, GIConv conv)
 478 {
 479     unsigned char str[2];
 480     unsigned char buf_ch[UTF8_CHAR_LEN + 1];
 481     int ch;
 482
 483     str[0] = (unsigned char) input_char;
 484     str[1] = '\0';
 485
 486     switch (str_translate_char (conv, (char *) str, -1, (char *) buf_ch, sizeof (buf_ch)))
 487     {
 488     case ESTR_SUCCESS:
 489         {
 490             int res;
 491
 492             res = g_utf8_get_char_validated ((char *) buf_ch, -1);
 493             ch = res >= 0 ? res : buf_ch[0];
 494             break;
 495         }
 496     case ESTR_PROBLEM:
 497     case ESTR_FAILURE:
 498     default:
 499         ch = '.';
 500         break;
 501     }
 502
 503     return ch;
 504 }
 505
 506 /* --------------------------------------------------------------------------------------------- */
 507
 508 int
 509 convert_from_8bit_to_utf_c2 (char input_char)
 510 {
 511     int ch = '.';
 512     GIConv conv;
 513     const char *cp_from;
 514
 515     cp_from = get_codepage_id (mc_global.source_codepage);
 516
 517     conv = str_crt_conv_to (cp_from);
 518     if (conv != INVALID_CONV)
 519     {
 520         ch = convert_from_8bit_to_utf_c (input_char, conv);
 521         str_close_conv (conv);
 522     }
 523
 524     return ch;
 525 }
 526
 527 /* --------------------------------------------------------------------------------------------- */