lib/charsets.c

   1 /*
   2    Text conversion from one charset to another.
   3
   4    Copyright (C) 2001-2016
   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/charsets.h"
  40
  41 /*** global variables ****************************************************************************/
  42
  43 GPtrArray *codepages = NULL;
  44
  45 unsigned char conv_displ[256];
  46 unsigned char conv_input[256];
  47
  48 const char *cp_display = NULL;
  49 const char *cp_source = NULL;
  50
  51 /*** file scope macro definitions ****************************************************************/
  52
  53 #define UNKNCHAR '\001'
  54
  55 #define OTHER_8BIT "Other_8_bit"
  56
  57 /*** file scope type declarations ****************************************************************/
  58
  59 /*** file scope variables ************************************************************************/
  60
  61 /*** file scope functions ************************************************************************/
  62 /* --------------------------------------------------------------------------------------------- */
  63
  64 static codepage_desc *
  65 new_codepage_desc (const char *id, const char *name)
  66 {
  67     codepage_desc *desc;
  68
  69     desc = g_new (codepage_desc, 1);
  70     desc->id = g_strdup (id);
  71     desc->name = g_strdup (name);
  72
  73     return desc;
  74 }
  75
  76 /* --------------------------------------------------------------------------------------------- */
  77
  78 static void
  79 free_codepage_desc (gpointer data, gpointer user_data)
  80 {
  81     codepage_desc *desc = (codepage_desc *) data;
  82     (void) user_data;
  83
  84     g_free (desc->id);
  85     g_free (desc->name);
  86     g_free (desc);
  87 }
  88
  89 /* --------------------------------------------------------------------------------------------- */
  90 /* returns display codepage */
  91
  92 static void
  93 load_codepages_list_from_file (GPtrArray ** list, const char *fname)
  94 {
  95     FILE *f;
  96     char buf[BUF_MEDIUM];
  97     char *default_codepage = NULL;
  98
  99     f = fopen (fname, "r");
 100     if (f == NULL)
 101         return;
 102
 103     while (fgets (buf, sizeof buf, f) != NULL)
 104     {
 105         /* split string into id and cpname */
 106         char *p = buf;
 107         size_t buflen = strlen (buf);
 108
 109         if (*p == '\n' || *p == '\0' || *p == '#')
 110             continue;
 111
 112         if (buflen > 0 && buf[buflen - 1] == '\n')
 113             buf[buflen - 1] = '\0';
 114         while (*p != '\t' && *p != ' ' && *p != '\0')
 115             ++p;
 116         if (*p == '\0')
 117             goto fail;
 118
 119         *p++ = '\0';
 120         g_strstrip (p);
 121         if (*p == '\0')
 122             goto fail;
 123
 124         if (strcmp (buf, "default") == 0)
 125             default_codepage = g_strdup (p);
 126         else
 127         {
 128             const char *id = buf;
 129
 130             if (*list == NULL)
 131             {
 132                 *list = g_ptr_array_sized_new (16);
 133                 g_ptr_array_add (*list, new_codepage_desc (id, p));
 134             }
 135             else
 136             {
 137                 unsigned int i;
 138
 139                 /* whether id is already present in list */
 140                 /* if yes, overwrite description */
 141                 for (i = 0; i < (*list)->len; i++)
 142                 {
 143                     codepage_desc *desc;
 144
 145                     desc = (codepage_desc *) g_ptr_array_index (*list, i);
 146
 147                     if (strcmp (id, desc->id) == 0)
 148                     {
 149                         /* found */
 150                         g_free (desc->name);
 151                         desc->name = g_strdup (p);
 152                         break;
 153                     }
 154                 }
 155
 156                 /* not found */
 157                 if (i == (*list)->len)
 158                     g_ptr_array_add (*list, new_codepage_desc (id, p));
 159             }
 160         }
 161     }
 162
 163     if (default_codepage != NULL)
 164     {
 165         mc_global.display_codepage = get_codepage_index (default_codepage);
 166         g_free (default_codepage);
 167     }
 168
 169   fail:
 170     fclose (f);
 171 }
 172
 173 /* --------------------------------------------------------------------------------------------- */
 174
 175 static char
 176 translate_character (GIConv cd, char c)
 177 {
 178     gchar *tmp_buff = NULL;
 179     gsize bytes_read, bytes_written = 0;
 180     const char *ibuf = &c;
 181     char ch = UNKNCHAR;
 182
 183     int ibuflen = 1;
 184
 185     tmp_buff = g_convert_with_iconv (ibuf, ibuflen, cd, &bytes_read, &bytes_written, NULL);
 186     if (tmp_buff)
 187         ch = tmp_buff[0];
 188     g_free (tmp_buff);
 189     return ch;
 190 }
 191
 192 /* --------------------------------------------------------------------------------------------- */
 193 /*** public functions ****************************************************************************/
 194 /* --------------------------------------------------------------------------------------------- */
 195
 196 void
 197 load_codepages_list (void)
 198 {
 199     char *fname;
 200
 201     /* 1: try load /usr/share/mc/mc.charsets */
 202     fname = g_build_filename (mc_global.share_data_dir, CHARSETS_LIST, (char *) NULL);
 203     load_codepages_list_from_file (&codepages, fname);
 204     g_free (fname);
 205
 206     /* 2: try load /etc/mc/mc.charsets */
 207     fname = g_build_filename (mc_global.sysconfig_dir, CHARSETS_LIST, (char *) NULL);
 208     load_codepages_list_from_file (&codepages, fname);
 209     g_free (fname);
 210
 211     if (codepages == NULL)
 212     {
 213         /* files are not found, add defaullt codepage */
 214         fprintf (stderr, "%s\n", _("Warning: cannot load codepages list"));
 215
 216         codepages = g_ptr_array_new ();
 217         g_ptr_array_add (codepages, new_codepage_desc (DEFAULT_CHARSET, _("7-bit ASCII")));
 218     }
 219 }
 220
 221 /* --------------------------------------------------------------------------------------------- */
 222
 223 void
 224 free_codepages_list (void)
 225 {
 226     g_ptr_array_foreach (codepages, free_codepage_desc, NULL);
 227     g_ptr_array_free (codepages, TRUE);
 228 }
 229
 230 /* --------------------------------------------------------------------------------------------- */
 231
 232 const char *
 233 get_codepage_id (const int n)
 234 {
 235     return (n < 0) ? OTHER_8BIT : ((codepage_desc *) g_ptr_array_index (codepages, n))->id;
 236 }
 237
 238 /* --------------------------------------------------------------------------------------------- */
 239
 240 int
 241 get_codepage_index (const char *id)
 242 {
 243     size_t i;
 244     if (strcmp (id, OTHER_8BIT) == 0)
 245         return -1;
 246     if (codepages == NULL)
 247         return -1;
 248     for (i = 0; i < codepages->len; i++)
 249         if (strcmp (id, ((codepage_desc *) g_ptr_array_index (codepages, i))->id) == 0)
 250             return i;
 251     return -1;
 252 }
 253
 254 /* --------------------------------------------------------------------------------------------- */
 255 /** Check if specified encoding can be used in mc.
 256  * @param encoding name of encoding
 257  * @return TRUE if encoding is supported by mc, FALSE otherwise
 258  */
 259
 260 gboolean
 261 is_supported_encoding (const char *encoding)
 262 {
 263     gboolean result = FALSE;
 264     guint t;
 265
 266     for (t = 0; t < codepages->len; t++)
 267     {
 268         const char *id = ((codepage_desc *) g_ptr_array_index (codepages, t))->id;
 269         result |= (g_ascii_strncasecmp (encoding, id, strlen (id)) == 0);
 270     }
 271
 272     return result;
 273 }
 274
 275 /* --------------------------------------------------------------------------------------------- */
 276
 277 char *
 278 init_translation_table (int cpsource, int cpdisplay)
 279 {
 280     int i;
 281     GIConv cd;
 282
 283     /* Fill inpit <-> display tables */
 284
 285     if (cpsource < 0 || cpdisplay < 0 || cpsource == cpdisplay)
 286     {
 287         for (i = 0; i <= 255; ++i)
 288         {
 289             conv_displ[i] = i;
 290             conv_input[i] = i;
 291             cp_source = cp_display;
 292         }
 293         return NULL;
 294     }
 295
 296     for (i = 0; i <= 127; ++i)
 297     {
 298         conv_displ[i] = i;
 299         conv_input[i] = i;
 300     }
 301     cp_source = ((codepage_desc *) g_ptr_array_index (codepages, cpsource))->id;
 302     cp_display = ((codepage_desc *) g_ptr_array_index (codepages, cpdisplay))->id;
 303
 304     /* display <- inpit table */
 305
 306     cd = g_iconv_open (cp_display, cp_source);
 307     if (cd == INVALID_CONV)
 308         return g_strdup_printf (_("Cannot translate from %s to %s"), cp_source, cp_display);
 309
 310     for (i = 128; i <= 255; ++i)
 311         conv_displ[i] = translate_character (cd, i);
 312
 313     g_iconv_close (cd);
 314
 315     /* inpit <- display table */
 316
 317     cd = g_iconv_open (cp_source, cp_display);
 318     if (cd == INVALID_CONV)
 319         return g_strdup_printf (_("Cannot translate from %s to %s"), cp_display, cp_source);
 320
 321     for (i = 128; i <= 255; ++i)
 322     {
 323         unsigned char ch;
 324         ch = translate_character (cd, i);
 325         conv_input[i] = (ch == UNKNCHAR) ? i : ch;
 326     }
 327
 328     g_iconv_close (cd);
 329
 330     return NULL;
 331 }
 332
 333 /* --------------------------------------------------------------------------------------------- */
 334
 335 void
 336 convert_to_display (char *str)
 337 {
 338     if (!str)
 339         return;
 340
 341     while (*str)
 342     {
 343         *str = conv_displ[(unsigned char) *str];
 344         str++;
 345     }
 346 }
 347
 348 /* --------------------------------------------------------------------------------------------- */
 349
 350 GString *
 351 str_convert_to_display (const char *str)
 352 {
 353     return str_nconvert_to_display (str, -1);
 354
 355 }
 356
 357 /* --------------------------------------------------------------------------------------------- */
 358
 359 GString *
 360 str_nconvert_to_display (const char *str, int len)
 361 {
 362     GString *buff;
 363     GIConv conv;
 364
 365     if (!str)
 366         return g_string_new ("");
 367
 368     if (cp_display == cp_source)
 369         return g_string_new (str);
 370
 371     conv = str_crt_conv_from (cp_source);
 372
 373     buff = g_string_new ("");
 374     str_nconvert (conv, str, len, buff);
 375     str_close_conv (conv);
 376     return buff;
 377 }
 378
 379 /* --------------------------------------------------------------------------------------------- */
 380
 381 void
 382 convert_from_input (char *str)
 383 {
 384     if (!str)
 385         return;
 386
 387     while (*str)
 388     {
 389         *str = conv_input[(unsigned char) *str];
 390         str++;
 391     }
 392 }
 393
 394 /* --------------------------------------------------------------------------------------------- */
 395
 396 GString *
 397 str_convert_to_input (const char *str)
 398 {
 399     return str_nconvert_to_input (str, -1);
 400 }
 401
 402 /* --------------------------------------------------------------------------------------------- */
 403
 404 GString *
 405 str_nconvert_to_input (const char *str, int len)
 406 {
 407     GString *buff;
 408     GIConv conv;
 409
 410     if (!str)
 411         return g_string_new ("");
 412
 413     if (cp_display == cp_source)
 414         return g_string_new (str);
 415
 416     conv = str_crt_conv_to (cp_source);
 417
 418     buff = g_string_new ("");
 419     str_nconvert (conv, str, len, buff);
 420     str_close_conv (conv);
 421     return buff;
 422 }
 423
 424 /* --------------------------------------------------------------------------------------------- */
 425
 426 unsigned char
 427 convert_from_utf_to_current (const char *str)
 428 {
 429     unsigned char buf_ch[UTF8_CHAR_LEN + 1];
 430     unsigned char ch = '.';
 431     GIConv conv;
 432     const char *cp_to;
 433
 434     if (str == NULL)
 435         return '.';
 436
 437     cp_to = get_codepage_id (mc_global.source_codepage);
 438     conv = str_crt_conv_to (cp_to);
 439
 440     if (conv != INVALID_CONV)
 441     {
 442         switch (str_translate_char (conv, str, -1, (char *) buf_ch, sizeof (buf_ch)))
 443         {
 444         case ESTR_SUCCESS:
 445             ch = buf_ch[0];
 446             break;
 447         case ESTR_PROBLEM:
 448         case ESTR_FAILURE:
 449             ch = '.';
 450             break;
 451         default:
 452             break;
 453         }
 454         str_close_conv (conv);
 455     }
 456
 457     return ch;
 458 }
 459
 460 /* --------------------------------------------------------------------------------------------- */
 461
 462 unsigned char
 463 convert_from_utf_to_current_c (int input_char, GIConv conv)
 464 {
 465     unsigned char str[UTF8_CHAR_LEN + 1];
 466     unsigned char buf_ch[UTF8_CHAR_LEN + 1];
 467     unsigned char ch = '.';
 468     int res;
 469
 470     res = g_unichar_to_utf8 (input_char, (char *) str);
 471     if (res == 0)
 472         return ch;
 473
 474     str[res] = '\0';
 475
 476     switch (str_translate_char (conv, (char *) str, -1, (char *) buf_ch, sizeof (buf_ch)))
 477     {
 478     case ESTR_SUCCESS:
 479         ch = buf_ch[0];
 480         break;
 481     case ESTR_PROBLEM:
 482     case ESTR_FAILURE:
 483         ch = '.';
 484         break;
 485     default:
 486         break;
 487     }
 488
 489     return ch;
 490 }
 491
 492 /* --------------------------------------------------------------------------------------------- */
 493
 494 int
 495 convert_from_8bit_to_utf_c (char input_char, GIConv conv)
 496 {
 497     unsigned char str[2];
 498     unsigned char buf_ch[UTF8_CHAR_LEN + 1];
 499     int ch;
 500
 501     str[0] = (unsigned char) input_char;
 502     str[1] = '\0';
 503
 504     switch (str_translate_char (conv, (char *) str, -1, (char *) buf_ch, sizeof (buf_ch)))
 505     {
 506     case ESTR_SUCCESS:
 507         {
 508             int res;
 509
 510             res = g_utf8_get_char_validated ((char *) buf_ch, -1);
 511             ch = res >= 0 ? res : buf_ch[0];
 512             break;
 513         }
 514     case ESTR_PROBLEM:
 515     case ESTR_FAILURE:
 516     default:
 517         ch = '.';
 518         break;
 519     }
 520
 521     return ch;
 522 }
 523
 524 /* --------------------------------------------------------------------------------------------- */
 525
 526 int
 527 convert_from_8bit_to_utf_c2 (char input_char)
 528 {
 529     unsigned char str[2];
 530     int ch = '.';
 531     GIConv conv;
 532     const char *cp_from;
 533
 534     str[0] = (unsigned char) input_char;
 535     str[1] = '\0';
 536
 537     cp_from = get_codepage_id (mc_global.source_codepage);
 538     conv = str_crt_conv_to (cp_from);
 539
 540     if (conv != INVALID_CONV)
 541     {
 542         unsigned char buf_ch[UTF8_CHAR_LEN + 1];
 543
 544         switch (str_translate_char (conv, (char *) str, -1, (char *) buf_ch, sizeof (buf_ch)))
 545         {
 546         case ESTR_SUCCESS:
 547             {
 548                 int res;
 549
 550                 res = g_utf8_get_char_validated ((char *) buf_ch, -1);
 551                 ch = res >= 0 ? res : buf_ch[0];
 552                 break;
 553             }
 554         case ESTR_PROBLEM:
 555         case ESTR_FAILURE:
 556         default:
 557             ch = '.';
 558             break;
 559         }
 560         str_close_conv (conv);
 561     }
 562
 563     return ch;
 564 }
 565
 566 /* --------------------------------------------------------------------------------------------- */