gconvert.c 55.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
/* GLIB - Library of useful routines for C programming
 *
 * gconvert.c: Convert between character sets using iconv
 * Copyright Red Hat Inc., 2000
 * Authors: Havoc Pennington <hp@redhat.com>, Owen Taylor <otaylor@redhat.com
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 02111-1307, USA.
 */

23
#include "config.h"
Owen Taylor's avatar
1.3.14  
Owen Taylor committed
24

25 26
#include <iconv.h>
#include <errno.h>
27
#include <stdio.h>
28
#include <string.h>
29 30
#include <stdlib.h>

Tor Lillqvist's avatar
Tor Lillqvist committed
31
#include "glib.h"
32
#include "gprintfint.h"
33
#include "gthreadinit.h"
Tor Lillqvist's avatar
Tor Lillqvist committed
34

35 36
#ifdef G_PLATFORM_WIN32
#define STRICT
37
#include <windows.h>
38
#undef STRICT
39
#endif
40

Owen Taylor's avatar
Owen Taylor committed
41
#include "glibintl.h"
42

43 44
#if defined(USE_LIBICONV_GNU) && !defined (_LIBICONV_H)
#error GNU libiconv in use but included iconv.h not from libiconv
45
#endif
46 47
#if !defined(USE_LIBICONV_GNU) && defined (_LIBICONV_H)
#error GNU libiconv not in use but included iconv.h is from libiconv
48 49
#endif

50 51
#include "galias.h"

52
GQuark 
53
g_convert_error_quark (void)
54
{
55
  return g_quark_from_static_string ("g_convert_error");
56 57
}

58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
static gboolean
try_conversion (const char *to_codeset,
		const char *from_codeset,
		iconv_t    *cd)
{
  *cd = iconv_open (to_codeset, from_codeset);

  if (*cd == (iconv_t)-1 && errno == EINVAL)
    return FALSE;
  else
    return TRUE;
}

static gboolean
try_to_aliases (const char **to_aliases,
		const char  *from_codeset,
		iconv_t     *cd)
{
  if (to_aliases)
    {
      const char **p = to_aliases;
      while (*p)
	{
	  if (try_conversion (*p, from_codeset, cd))
	    return TRUE;

	  p++;
	}
    }

  return FALSE;
}

91
extern const char **_g_charset_get_aliases (const char *canonical_name) G_GNUC_INTERNAL;
92

Havoc Pennington's avatar
docs  
Havoc Pennington committed
93 94 95 96 97
/**
 * g_iconv_open:
 * @to_codeset: destination codeset
 * @from_codeset: source codeset
 * 
98
 * Same as the standard UNIX routine iconv_open(), but
99
 * may be implemented via libiconv on UNIX flavors that lack
Havoc Pennington's avatar
docs  
Havoc Pennington committed
100 101
 * a native implementation.
 * 
Owen Taylor's avatar
Owen Taylor committed
102
 * GLib provides g_convert() and g_locale_to_utf8() which are likely
Havoc Pennington's avatar
docs  
Havoc Pennington committed
103 104
 * more convenient than the raw iconv wrappers.
 * 
105 106
 * Return value: a "conversion descriptor", or (GIConv)-1 if
 *  opening the converter failed.
Havoc Pennington's avatar
docs  
Havoc Pennington committed
107
 **/
108 109 110 111
GIConv
g_iconv_open (const gchar  *to_codeset,
	      const gchar  *from_codeset)
{
112
  iconv_t cd;
113
  
114 115 116
  if (!try_conversion (to_codeset, from_codeset, &cd))
    {
      const char **to_aliases = _g_charset_get_aliases (to_codeset);
117
      const char **from_aliases = _g_charset_get_aliases (from_codeset);
118 119 120 121 122 123 124

      if (from_aliases)
	{
	  const char **p = from_aliases;
	  while (*p)
	    {
	      if (try_conversion (to_codeset, *p, &cd))
125
		goto out;
126 127

	      if (try_to_aliases (to_aliases, *p, &cd))
128
		goto out;
129 130 131 132 133 134

	      p++;
	    }
	}

      if (try_to_aliases (to_aliases, from_codeset, &cd))
135
	goto out;
136 137
    }

138
 out:
139
  return (cd == (iconv_t)-1) ? (GIConv)-1 : (GIConv)cd;
140 141
}

Havoc Pennington's avatar
docs  
Havoc Pennington committed
142 143 144 145 146 147 148 149
/**
 * g_iconv:
 * @converter: conversion descriptor from g_iconv_open()
 * @inbuf: bytes to convert
 * @inbytes_left: inout parameter, bytes remaining to convert in @inbuf
 * @outbuf: converted output bytes
 * @outbytes_left: inout parameter, bytes available to fill in @outbuf
 * 
150
 * Same as the standard UNIX routine iconv(), but
151
 * may be implemented via libiconv on UNIX flavors that lack
Havoc Pennington's avatar
docs  
Havoc Pennington committed
152 153
 * a native implementation.
 *
Owen Taylor's avatar
Owen Taylor committed
154
 * GLib provides g_convert() and g_locale_to_utf8() which are likely
Havoc Pennington's avatar
docs  
Havoc Pennington committed
155 156 157 158
 * more convenient than the raw iconv wrappers.
 * 
 * Return value: count of non-reversible conversions, or -1 on error
 **/
159 160 161
size_t 
g_iconv (GIConv   converter,
	 gchar  **inbuf,
162
	 gsize   *inbytes_left,
163
	 gchar  **outbuf,
164
	 gsize   *outbytes_left)
165 166 167 168 169 170
{
  iconv_t cd = (iconv_t)converter;

  return iconv (cd, inbuf, inbytes_left, outbuf, outbytes_left);
}

Havoc Pennington's avatar
docs  
Havoc Pennington committed
171 172 173 174
/**
 * g_iconv_close:
 * @converter: a conversion descriptor from g_iconv_open()
 *
175
 * Same as the standard UNIX routine iconv_close(), but
176
 * may be implemented via libiconv on UNIX flavors that lack
Havoc Pennington's avatar
docs  
Havoc Pennington committed
177
 * a native implementation. Should be called to clean up
Matthias Clasen's avatar
Matthias Clasen committed
178
 * the conversion descriptor from g_iconv_open() when
Havoc Pennington's avatar
docs  
Havoc Pennington committed
179 180
 * you are done converting things.
 *
Owen Taylor's avatar
Owen Taylor committed
181
 * GLib provides g_convert() and g_locale_to_utf8() which are likely
Havoc Pennington's avatar
docs  
Havoc Pennington committed
182 183 184 185
 * more convenient than the raw iconv wrappers.
 * 
 * Return value: -1 on error, 0 on success
 **/
186 187 188 189 190 191 192 193
gint
g_iconv_close (GIConv converter)
{
  iconv_t cd = (iconv_t)converter;

  return iconv_close (cd);
}

194

Matthias Clasen's avatar
Matthias Clasen committed
195 196
#ifdef NEED_ICONV_CACHE

197 198 199 200 201 202
#define ICONV_CACHE_SIZE   (16)

struct _iconv_cache_bucket {
  gchar *key;
  guint32 refcount;
  gboolean used;
203
  GIConv cd;
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238
};

static GList *iconv_cache_list;
static GHashTable *iconv_cache;
static GHashTable *iconv_open_hash;
static guint iconv_cache_size = 0;
G_LOCK_DEFINE_STATIC (iconv_cache_lock);

/* caller *must* hold the iconv_cache_lock */
static void
iconv_cache_init (void)
{
  static gboolean initialized = FALSE;
  
  if (initialized)
    return;
  
  iconv_cache_list = NULL;
  iconv_cache = g_hash_table_new (g_str_hash, g_str_equal);
  iconv_open_hash = g_hash_table_new (g_direct_hash, g_direct_equal);
  
  initialized = TRUE;
}


/**
 * iconv_cache_bucket_new:
 * @key: cache key
 * @cd: iconv descriptor
 *
 * Creates a new cache bucket, inserts it into the cache and
 * increments the cache size.
 *
 * Returns a pointer to the newly allocated cache bucket.
 **/
239
static struct _iconv_cache_bucket *
240
iconv_cache_bucket_new (const gchar *key, GIConv cd)
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331
{
  struct _iconv_cache_bucket *bucket;
  
  bucket = g_new (struct _iconv_cache_bucket, 1);
  bucket->key = g_strdup (key);
  bucket->refcount = 1;
  bucket->used = TRUE;
  bucket->cd = cd;
  
  g_hash_table_insert (iconv_cache, bucket->key, bucket);
  
  /* FIXME: if we sorted the list so items with few refcounts were
     first, then we could expire them faster in iconv_cache_expire_unused () */
  iconv_cache_list = g_list_prepend (iconv_cache_list, bucket);
  
  iconv_cache_size++;
  
  return bucket;
}


/**
 * iconv_cache_bucket_expire:
 * @node: cache bucket's node
 * @bucket: cache bucket
 *
 * Expires a single cache bucket @bucket. This should only ever be
 * called on a bucket that currently has no used iconv descriptors
 * open.
 *
 * @node is not a required argument. If @node is not supplied, we
 * search for it ourselves.
 **/
static void
iconv_cache_bucket_expire (GList *node, struct _iconv_cache_bucket *bucket)
{
  g_hash_table_remove (iconv_cache, bucket->key);
  
  if (node == NULL)
    node = g_list_find (iconv_cache_list, bucket);
  
  g_assert (node != NULL);
  
  if (node->prev)
    {
      node->prev->next = node->next;
      if (node->next)
        node->next->prev = node->prev;
    }
  else
    {
      iconv_cache_list = node->next;
      if (node->next)
        node->next->prev = NULL;
    }
  
  g_list_free_1 (node);
  
  g_free (bucket->key);
  g_iconv_close (bucket->cd);
  g_free (bucket);
  
  iconv_cache_size--;
}


/**
 * iconv_cache_expire_unused:
 *
 * Expires as many unused cache buckets as it needs to in order to get
 * the total number of buckets < ICONV_CACHE_SIZE.
 **/
static void
iconv_cache_expire_unused (void)
{
  struct _iconv_cache_bucket *bucket;
  GList *node, *next;
  
  node = iconv_cache_list;
  while (node && iconv_cache_size >= ICONV_CACHE_SIZE)
    {
      next = node->next;
      
      bucket = node->data;
      if (bucket->refcount == 0)
        iconv_cache_bucket_expire (node, bucket);
      
      node = next;
    }
}

Havoc Pennington's avatar
Havoc Pennington committed
332
static GIConv
333
open_converter (const gchar *to_codeset,
334
		const gchar *from_codeset,
335 336
		GError     **error)
{
337 338 339 340 341 342
  struct _iconv_cache_bucket *bucket;
  gchar *key;
  GIConv cd;
  
  /* create our key */
  key = g_alloca (strlen (from_codeset) + strlen (to_codeset) + 2);
343
  _g_sprintf (key, "%s:%s", from_codeset, to_codeset);
344 345 346 347 348 349 350 351
  
  G_LOCK (iconv_cache_lock);
  
  /* make sure the cache has been initialized */
  iconv_cache_init ();
  
  bucket = g_hash_table_lookup (iconv_cache, key);
  if (bucket)
352
    {
353 354 355
      if (bucket->used)
        {
          cd = g_iconv_open (to_codeset, from_codeset);
356
          if (cd == (GIConv) -1)
357 358
            goto error;
        }
359
      else
360
        {
361 362 363 364
	  /* Apparently iconv on Solaris <= 7 segfaults if you pass in
	   * NULL for anything but inbuf; work around that. (NULL outbuf
	   * or NULL *outbuf is allowed by Unix98.)
	   */
365
	  gsize inbytes_left = 0;
366
	  gchar *outbuf = NULL;
367
	  gsize outbytes_left = 0;
368
		
369 370 371 372
          cd = bucket->cd;
          bucket->used = TRUE;
          
          /* reset the descriptor */
373
          g_iconv (cd, NULL, &inbytes_left, &outbuf, &outbytes_left);
374 375 376
        }
      
      bucket->refcount++;
377
    }
378 379 380
  else
    {
      cd = g_iconv_open (to_codeset, from_codeset);
381
      if (cd == (GIConv) -1)
382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
        goto error;
      
      iconv_cache_expire_unused ();
      
      bucket = iconv_cache_bucket_new (key, cd);
    }
  
  g_hash_table_insert (iconv_open_hash, cd, bucket->key);
  
  G_UNLOCK (iconv_cache_lock);
  
  return cd;
  
 error:
  
  G_UNLOCK (iconv_cache_lock);
  
  /* Something went wrong.  */
400 401 402 403 404 405 406 407 408 409 410
  if (error)
    {
      if (errno == EINVAL)
	g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NO_CONVERSION,
		     _("Conversion from character set '%s' to '%s' is not supported"),
		     from_codeset, to_codeset);
      else
	g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
		     _("Could not open converter from '%s' to '%s'"),
		     from_codeset, to_codeset);
    }
411
  
412
  return cd;
413
}
414

415 416 417 418 419
static int
close_converter (GIConv converter)
{
  struct _iconv_cache_bucket *bucket;
  const gchar *key;
420
  GIConv cd;
421
  
422
  cd = converter;
423
  
424
  if (cd == (GIConv) -1)
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
    return 0;
  
  G_LOCK (iconv_cache_lock);
  
  key = g_hash_table_lookup (iconv_open_hash, cd);
  if (key)
    {
      g_hash_table_remove (iconv_open_hash, cd);
      
      bucket = g_hash_table_lookup (iconv_cache, key);
      g_assert (bucket);
      
      bucket->refcount--;
      
      if (cd == bucket->cd)
        bucket->used = FALSE;
      else
        g_iconv_close (cd);
      
      if (!bucket->refcount && iconv_cache_size > ICONV_CACHE_SIZE)
        {
          /* expire this cache bucket */
          iconv_cache_bucket_expire (NULL, bucket);
        }
    }
  else
    {
      G_UNLOCK (iconv_cache_lock);
      
      g_warning ("This iconv context wasn't opened using open_converter");
      
      return g_iconv_close (converter);
    }
  
  G_UNLOCK (iconv_cache_lock);
  
  return 0;
462 463
}

Matthias Clasen's avatar
Matthias Clasen committed
464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504
#else  /* !NEED_ICONV_CACHE */

static GIConv
open_converter (const gchar *to_codeset,
		const gchar *from_codeset,
		GError     **error)
{
  GIConv cd;

  cd = g_iconv_open (to_codeset, from_codeset);

  if (cd == (GIConv) -1)
    {
      /* Something went wrong.  */
      if (error)
	{
	  if (errno == EINVAL)
	    g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NO_CONVERSION,
			 _("Conversion from character set '%s' to '%s' is not supported"),
			 from_codeset, to_codeset);
	  else
	    g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
			 _("Could not open converter from '%s' to '%s'"),
			 from_codeset, to_codeset);
	}
    }
  
  return cd;
}

static int
close_converter (GIConv cd)
{
  if (cd == (GIConv) -1)
    return 0;
  
  return g_iconv_close (cd);  
}

#endif /* NEED_ICONV_CACHE */

505 506 507
/**
 * g_convert_with_iconv:
 * @str:           the string to convert
508 509
 * @len:           the length of the string, or -1 if the string is 
 *                 nul-terminated<footnoteref linkend="nul-unsafe"/>. 
510 511 512
 * @converter:     conversion descriptor from g_iconv_open()
 * @bytes_read:    location to store the number of bytes in the
 *                 input string that were successfully converted, or %NULL.
Matthias Clasen's avatar
Matthias Clasen committed
513
 *                 Even if the conversion was successful, this may be 
514
 *                 less than @len if there were partial characters
515
 *                 at the end of the input. If the error
516 517
 *                 #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
 *                 stored will the byte offset after the last valid
518
 *                 input sequence.
519
 * @bytes_written: the number of bytes stored in the output buffer (not 
Matthias Clasen's avatar
Matthias Clasen committed
520
 *                 including the terminating nul).
521 522 523
 * @error:         location to store the error occuring, or %NULL to ignore
 *                 errors. Any of the errors in #GConvertError may occur.
 *
524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
 * Converts a string from one character set to another. 
 * 
 * Note that you should use g_iconv() for streaming 
 * conversions<footnote id="streaming-state">
 *  <para>
 * Despite the fact that @byes_read can return information about partial 
 * characters, the <literal>g_convert_...</literal> functions
 * are not generally suitable for streaming. If the underlying converter 
 * being used maintains internal state, then this won't be preserved 
 * across successive calls to g_convert(), g_convert_with_iconv() or 
 * g_convert_with_fallback(). (An example of this is the GNU C converter 
 * for CP1255 which does not emit a base character until it knows that 
 * the next character is not a mark that could combine with the base 
 * character.)
 *  </para>
 * </footnote>. 
540 541
 *
 * Return value: If the conversion was successful, a newly allocated
542 543
 *               nul-terminated string, which must be freed with
 *               g_free(). Otherwise %NULL and @error will be set.
544 545 546 547 548 549 550 551 552 553 554 555
 **/
gchar*
g_convert_with_iconv (const gchar *str,
		      gssize       len,
		      GIConv       converter,
		      gsize       *bytes_read, 
		      gsize       *bytes_written, 
		      GError     **error)
{
  gchar *dest;
  gchar *outp;
  const gchar *p;
556
  const gchar *shift_p = NULL;
557 558 559 560 561
  gsize inbytes_remaining;
  gsize outbytes_remaining;
  gsize err;
  gsize outbuf_size;
  gboolean have_error = FALSE;
562
  gboolean done = FALSE;
563 564 565
  
  g_return_val_if_fail (converter != (GIConv) -1, NULL);
     
566 567 568 569 570
  if (len < 0)
    len = strlen (str);

  p = str;
  inbytes_remaining = len;
571
  outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
572
  
573 574 575
  outbytes_remaining = outbuf_size - 1; /* -1 for nul */
  outp = dest = g_malloc (outbuf_size);

576
  while (!done && !have_error)
577
    {
578 579 580
      err = g_iconv (converter, (char **)&p, &inbytes_remaining, &outp, &outbytes_remaining);

      if (err == (size_t) -1)
581
	{
582 583 584 585
	  switch (errno)
	    {
	    case EINVAL:
	      /* Incomplete text, do not report an error */
586
	      done = TRUE;
587 588 589 590
	      break;
	    case E2BIG:
	      {
		size_t used = outp - dest;
591
		
592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624
		outbuf_size *= 2;
		dest = g_realloc (dest, outbuf_size);
		
		outp = dest + used;
		outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
	      }
	      break;
	    case EILSEQ:
	      if (error)
		g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
			     _("Invalid byte sequence in conversion input"));
	      have_error = TRUE;
	      break;
	    default:
	      if (error)
		g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
			     _("Error during conversion: %s"),
			     g_strerror (errno));
	      have_error = TRUE;
	      break;
	    }
	}
      else 
	{
	  if (!shift_p)
	    {
	      /* call g_iconv with NULL inbuf to cleanup shift state */
	      shift_p = p;
	      p = NULL;
	      inbytes_remaining = 0;
	    }
	  else
	    done = TRUE;
625 626 627
	}
    }

628 629 630
  if (shift_p)
    p = shift_p;

631 632 633 634
  *outp = '\0';
  
  if (bytes_read)
    *bytes_read = p - str;
635 636 637 638
  else
    {
      if ((p - str) != len) 
	{
Havoc Pennington's avatar
Havoc Pennington committed
639 640
          if (!have_error)
            {
641 642 643
	      if (error)
		g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_PARTIAL_INPUT,
			     _("Partial character sequence at end of input"));
Havoc Pennington's avatar
Havoc Pennington committed
644 645
              have_error = TRUE;
            }
646 647
	}
    }
648 649 650 651 652 653 654 655 656 657 658 659 660

  if (bytes_written)
    *bytes_written = outp - dest;	/* Doesn't include '\0' */

  if (have_error)
    {
      g_free (dest);
      return NULL;
    }
  else
    return dest;
}

661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733
/**
 * g_convert:
 * @str:           the string to convert
 * @len:           the length of the string, or -1 if the string is 
 *                 nul-terminated<footnote id="nul-unsafe">
                     <para>
                       Note that some encodings may allow nul bytes to 
                       occur inside strings. In that case, using -1 for 
                       the @len parameter is unsafe.
                     </para>
                   </footnote>. 
 * @to_codeset:    name of character set into which to convert @str
 * @from_codeset:  character set of @str.
 * @bytes_read:    location to store the number of bytes in the
 *                 input string that were successfully converted, or %NULL.
 *                 Even if the conversion was successful, this may be 
 *                 less than @len if there were partial characters
 *                 at the end of the input. If the error
 *                 #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
 *                 stored will the byte offset after the last valid
 *                 input sequence.
 * @bytes_written: the number of bytes stored in the output buffer (not 
 *                 including the terminating nul).
 * @error:         location to store the error occuring, or %NULL to ignore
 *                 errors. Any of the errors in #GConvertError may occur.
 *
 * Converts a string from one character set to another.
 *
 * Note that you should use g_iconv() for streaming 
 * conversions<footnoteref linkend="streaming-state"/>.
 *
 * Return value: If the conversion was successful, a newly allocated
 *               nul-terminated string, which must be freed with
 *               g_free(). Otherwise %NULL and @error will be set.
 **/
gchar*
g_convert (const gchar *str,
           gssize       len,  
           const gchar *to_codeset,
           const gchar *from_codeset,
           gsize       *bytes_read, 
	   gsize       *bytes_written, 
	   GError     **error)
{
  gchar *res;
  GIConv cd;

  g_return_val_if_fail (str != NULL, NULL);
  g_return_val_if_fail (to_codeset != NULL, NULL);
  g_return_val_if_fail (from_codeset != NULL, NULL);
  
  cd = open_converter (to_codeset, from_codeset, error);

  if (cd == (GIConv) -1)
    {
      if (bytes_read)
        *bytes_read = 0;
      
      if (bytes_written)
        *bytes_written = 0;
      
      return NULL;
    }

  res = g_convert_with_iconv (str, len, cd,
			      bytes_read, bytes_written,
			      error);

  close_converter (cd);

  return res;
}

734 735 736
/**
 * g_convert_with_fallback:
 * @str:          the string to convert
737 738
 * @len:          the length of the string, or -1 if the string is 
 *                nul-terminated<footnoteref linkend="nul-unsafe"/>. 
739 740 741
 * @to_codeset:   name of character set into which to convert @str
 * @from_codeset: character set of @str.
 * @fallback:     UTF-8 string to use in place of character not
742 743 744 745
 *                present in the target encoding. (The string must be
 *                representable in the target encoding). 
                  If %NULL, characters not in the target encoding will 
                  be represented as Unicode escapes \uxxxx or \Uxxxxyyyy.
746 747
 * @bytes_read:   location to store the number of bytes in the
 *                input string that were successfully converted, or %NULL.
Matthias Clasen's avatar
Matthias Clasen committed
748
 *                Even if the conversion was successful, this may be 
749
 *                less than @len if there were partial characters
750
 *                at the end of the input.
751
 * @bytes_written: the number of bytes stored in the output buffer (not 
Matthias Clasen's avatar
Matthias Clasen committed
752
 *                including the terminating nul).
753 754 755
 * @error:        location to store the error occuring, or %NULL to ignore
 *                errors. Any of the errors in #GConvertError may occur.
 *
Matthias Clasen's avatar
Matthias Clasen committed
756
 * Converts a string from one character set to another, possibly
757 758 759 760
 * including fallback sequences for characters not representable
 * in the output. Note that it is not guaranteed that the specification
 * for the fallback sequences in @fallback will be honored. Some
 * systems may do a approximate conversion from @from_codeset
761
 * to @to_codeset in their iconv() functions, 
Owen Taylor's avatar
Owen Taylor committed
762
 * in which case GLib will simply return that approximate conversion.
763
 *
764 765 766
 * Note that you should use g_iconv() for streaming 
 * conversions<footnoteref linkend="streaming-state"/>.
 *
767
 * Return value: If the conversion was successful, a newly allocated
768 769
 *               nul-terminated string, which must be freed with
 *               g_free(). Otherwise %NULL and @error will be set.
770 771 772
 **/
gchar*
g_convert_with_fallback (const gchar *str,
773
			 gssize       len,    
774 775 776
			 const gchar *to_codeset,
			 const gchar *from_codeset,
			 gchar       *fallback,
777 778
			 gsize       *bytes_read,
			 gsize       *bytes_written,
779 780 781 782 783 784 785
			 GError     **error)
{
  gchar *utf8;
  gchar *dest;
  gchar *outp;
  const gchar *insert_str = NULL;
  const gchar *p;
786
  gsize inbytes_remaining;   
787
  const gchar *save_p = NULL;
788 789 790
  gsize save_inbytes = 0;
  gsize outbytes_remaining; 
  gsize err;
791
  GIConv cd;
792
  gsize outbuf_size;
793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
  gboolean have_error = FALSE;
  gboolean done = FALSE;

  GError *local_error = NULL;
  
  g_return_val_if_fail (str != NULL, NULL);
  g_return_val_if_fail (to_codeset != NULL, NULL);
  g_return_val_if_fail (from_codeset != NULL, NULL);
     
  if (len < 0)
    len = strlen (str);
  
  /* Try an exact conversion; we only proceed if this fails
   * due to an illegal sequence in the input string.
   */
  dest = g_convert (str, len, to_codeset, from_codeset, 
		    bytes_read, bytes_written, &local_error);
  if (!local_error)
    return dest;

  if (!g_error_matches (local_error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE))
    {
      g_propagate_error (error, local_error);
      return NULL;
    }
  else
    g_error_free (local_error);

Havoc Pennington's avatar
Havoc Pennington committed
821 822
  local_error = NULL;
  
823 824 825 826
  /* No go; to proceed, we need a converter from "UTF-8" to
   * to_codeset, and the string as UTF-8.
   */
  cd = open_converter (to_codeset, "UTF-8", error);
827
  if (cd == (GIConv) -1)
828 829 830 831 832 833 834 835 836 837 838 839 840
    {
      if (bytes_read)
        *bytes_read = 0;
      
      if (bytes_written)
        *bytes_written = 0;
      
      return NULL;
    }

  utf8 = g_convert (str, len, "UTF-8", from_codeset, 
		    bytes_read, &inbytes_remaining, error);
  if (!utf8)
841
    {
842
      close_converter (cd);
843 844 845 846
      if (bytes_written)
        *bytes_written = 0;
      return NULL;
    }
847 848 849 850 851 852 853 854 855 856

  /* Now the heart of the code. We loop through the UTF-8 string, and
   * whenever we hit an offending character, we form fallback, convert
   * the fallback to the target codeset, and then go back to
   * converting the original string after finishing with the fallback.
   *
   * The variables save_p and save_inbytes store the input state
   * for the original string while we are converting the fallback
   */
  p = utf8;
857 858

  outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
859 860 861 862 863
  outbytes_remaining = outbuf_size - 1; /* -1 for nul */
  outp = dest = g_malloc (outbuf_size);

  while (!done && !have_error)
    {
864
      size_t inbytes_tmp = inbytes_remaining;
865
      err = g_iconv (cd, (char **)&p, &inbytes_tmp, &outp, &outbytes_remaining);
866
      inbytes_remaining = inbytes_tmp;
867 868 869 870 871 872 873 874 875 876 877

      if (err == (size_t) -1)
	{
	  switch (errno)
	    {
	    case EINVAL:
	      g_assert_not_reached();
	      break;
	    case E2BIG:
	      {
		size_t used = outp - dest;
878

879 880 881 882 883
		outbuf_size *= 2;
		dest = g_realloc (dest, outbuf_size);
		
		outp = dest + used;
		outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
884 885 886 887 888 889 890 891 892
		
		break;
	      }
	    case EILSEQ:
	      if (save_p)
		{
		  /* Error converting fallback string - fatal
		   */
		  g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
893
			       _("Cannot convert fallback '%s' to codeset '%s'"),
894 895 896 897
			       insert_str, to_codeset);
		  have_error = TRUE;
		  break;
		}
898
	      else if (p)
899 900 901 902
		{
		  if (!fallback)
		    { 
		      gunichar ch = g_utf8_get_char (p);
903
		      insert_str = g_strdup_printf (ch < 0x10000 ? "\\u%04x" : "\\U%08x",
904 905 906 907 908 909 910 911 912
						    ch);
		    }
		  else
		    insert_str = fallback;
		  
		  save_p = g_utf8_next_char (p);
		  save_inbytes = inbytes_remaining - (save_p - p);
		  p = insert_str;
		  inbytes_remaining = strlen (p);
913
		  break;
914
		}
915
	      /* fall thru if p is NULL */
916
	    default:
917
	      g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
918
			   _("Error during conversion: %s"),
Owen Taylor's avatar
Owen Taylor committed
919
			   g_strerror (errno));
920 921 922 923 924 925 926 927 928 929 930 931 932 933
	      have_error = TRUE;
	      break;
	    }
	}
      else
	{
	  if (save_p)
	    {
	      if (!fallback)
		g_free ((gchar *)insert_str);
	      p = save_p;
	      inbytes_remaining = save_inbytes;
	      save_p = NULL;
	    }
934 935 936 937 938 939
	  else if (p)
	    {
	      /* call g_iconv with NULL inbuf to cleanup shift state */
	      p = NULL;
	      inbytes_remaining = 0;
	    }
940 941 942 943 944 945 946 947 948
	  else
	    done = TRUE;
	}
    }

  /* Cleanup
   */
  *outp = '\0';
  
949
  close_converter (cd);
950 951

  if (bytes_written)
952
    *bytes_written = outp - dest;	/* Doesn't include '\0' */
953 954 955 956 957 958 959 960 961 962 963 964 965

  g_free (utf8);

  if (have_error)
    {
      if (save_p && !fallback)
	g_free ((gchar *)insert_str);
      g_free (dest);
      return NULL;
    }
  else
    return dest;
}
966 967 968 969

/*
 * g_locale_to_utf8
 *
970 971 972
 * 
 */

973 974 975 976
static gchar *
strdup_len (const gchar *string,
	    gssize       len,
	    gsize       *bytes_written,
977 978
	    gsize       *bytes_read,
	    GError      **error)
979 980 981 982
	 
{
  gsize real_len;

983
  if (!g_utf8_validate (string, len, NULL))
984 985 986 987 988 989 990 991 992 993 994
    {
      if (bytes_read)
	*bytes_read = 0;
      if (bytes_written)
	*bytes_written = 0;

      g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
		   _("Invalid byte sequence in conversion input"));
      return NULL;
    }
  
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
  if (len < 0)
    real_len = strlen (string);
  else
    {
      real_len = 0;
      
      while (real_len < len && string[real_len])
	real_len++;
    }
  
  if (bytes_read)
    *bytes_read = real_len;
  if (bytes_written)
    *bytes_written = real_len;

  return g_strndup (string, real_len);
}

1013 1014
/**
 * g_locale_to_utf8:
Tor Lillqvist's avatar
Tor Lillqvist committed
1015 1016
 * @opsysstring:   a string in the encoding of the current locale. On Windows
 *                 this means the system codepage.
1017
 * @len:           the length of the string, or -1 if the string is
1018
 *                 nul-terminated<footnoteref linkend="nul-unsafe"/>. 
1019 1020
 * @bytes_read:    location to store the number of bytes in the
 *                 input string that were successfully converted, or %NULL.
Matthias Clasen's avatar
Matthias Clasen committed
1021
 *                 Even if the conversion was successful, this may be 
1022
 *                 less than @len if there were partial characters
1023
 *                 at the end of the input. If the error
1024 1025
 *                 #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
 *                 stored will the byte offset after the last valid
1026
 *                 input sequence.
1027
 * @bytes_written: the number of bytes stored in the output buffer (not 
Matthias Clasen's avatar
Matthias Clasen committed
1028
 *                 including the terminating nul).
1029
 * @error:         location to store the error occuring, or %NULL to ignore
1030 1031
 *                 errors. Any of the errors in #GConvertError may occur.
 * 
1032 1033 1034
 * Converts a string which is in the encoding used for strings by
 * the C runtime (usually the same as that used by the operating
 * system) in the current locale into a UTF-8 string.
1035 1036 1037
 * 
 * Return value: The converted string, or %NULL on an error.
 **/
1038
gchar *
1039
g_locale_to_utf8 (const gchar  *opsysstring,
1040 1041 1042
		  gssize        len,            
		  gsize        *bytes_read,    
		  gsize        *bytes_written,
1043
		  GError      **error)
1044
{
1045
  const char *charset;
1046 1047

  if (g_get_charset (&charset))
1048
    return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1049 1050 1051
  else
    return g_convert (opsysstring, len, 
		      "UTF-8", charset, bytes_read, bytes_written, error);
1052 1053
}

1054 1055 1056 1057
/**
 * g_locale_from_utf8:
 * @utf8string:    a UTF-8 encoded string 
 * @len:           the length of the string, or -1 if the string is
1058
 *                 nul-terminated<footnoteref linkend="nul-unsafe"/>. 
1059 1060
 * @bytes_read:    location to store the number of bytes in the
 *                 input string that were successfully converted, or %NULL.
Matthias Clasen's avatar
Matthias Clasen committed
1061
 *                 Even if the conversion was successful, this may be 
1062
 *                 less than @len if there were partial characters
1063
 *                 at the end of the input. If the error
1064 1065
 *                 #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
 *                 stored will the byte offset after the last valid
1066
 *                 input sequence.
1067
 * @bytes_written: the number of bytes stored in the output buffer (not 
Matthias Clasen's avatar
Matthias Clasen committed
1068
 *                 including the terminating nul).
1069
 * @error:         location to store the error occuring, or %NULL to ignore
1070 1071 1072 1073 1074 1075 1076 1077
 *                 errors. Any of the errors in #GConvertError may occur.
 * 
 * Converts a string from UTF-8 to the encoding used for strings by
 * the C runtime (usually the same as that used by the operating
 * system) in the current locale.
 * 
 * Return value: The converted string, or %NULL on an error.
 **/
1078
gchar *
1079
g_locale_from_utf8 (const gchar *utf8string,
1080 1081 1082
		    gssize       len,            
		    gsize       *bytes_read,    
		    gsize       *bytes_written,
1083
		    GError     **error)
1084
{
1085
  const gchar *charset;
1086 1087

  if (g_get_charset (&charset))
1088
    return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1089 1090 1091
  else
    return g_convert (utf8string, len,
		      charset, "UTF-8", bytes_read, bytes_written, error);
1092 1093
}

1094
#ifndef G_PLATFORM_WIN32
1095

1096 1097 1098 1099 1100
typedef struct _GFilenameCharsetCache GFilenameCharsetCache;

struct _GFilenameCharsetCache {
  gboolean is_utf8;
  gchar *charset;
1101
  gchar **filename_charsets;
1102 1103 1104 1105 1106 1107 1108
};

static void
filename_charset_cache_free (gpointer data)
{
  GFilenameCharsetCache *cache = data;
  g_free (cache->charset);
1109
  g_strfreev (cache->filename_charsets);
1110 1111 1112
  g_free (cache);
}

Matthias Clasen's avatar
Matthias Clasen committed
1113
/**
1114 1115
 * g_get_filename_charsets:
 * @charsets: return location for the %NULL-terminated list of encoding names
1116
 *
1117 1118 1119
 * Determines the preferred character sets used for filenames.
 * The first character set from the @charsets is the filename encoding, the
 * subsequent character sets are used when trying to generate a displayable
Tor Lillqvist's avatar
Tor Lillqvist committed
1120
 * representation of a filename, see g_filename_display_name().
1121
 *
1122 1123 1124 1125 1126
 * On Unix, the character sets are determined by consulting the
 * environment variables <envar>G_FILENAME_ENCODING</envar> and
 * <envar>G_BROKEN_FILENAMES</envar>. On Windows, the character set
 * used in the GLib API is always UTF-8 and said environment variables
 * have no effect.
1127
 *
1128 1129 1130 1131 1132 1133 1134 1135 1136
 * <envar>G_FILENAME_ENCODING</envar> may be set to a comma-separated list 
 * of character set names. The special token "@locale" is taken to mean the 
 * character set for the current locale. If <envar>G_FILENAME_ENCODING</envar> 
 * is not set, but <envar>G_BROKEN_FILENAMES</envar> is, the character set of 
 * the current locale is taken as the filename encoding. If neither environment
 * variable is set, UTF-8 is taken as the filename encoding, but the character
 * set of the current locale is also put in the list of encodings.
 *
 * The returned @charsets belong to GLib and must not be freed.
1137 1138
 *
 * Note that on Unix, regardless of the locale character set or
1139
 * <envar>G_FILENAME_ENCODING</envar> value, the actual file names present on a
1140 1141
 * system might be in any random encoding or just gibberish.
 *
1142 1143 1144
 * Return value: %TRUE if the filename encoding is UTF-8.
 * 
 * Since: 2.6
1145
 */
1146 1147
gboolean
g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1148
{
1149 1150 1151
  static GStaticPrivate cache_private = G_STATIC_PRIVATE_INIT;
  GFilenameCharsetCache *cache = g_static_private_get (&cache_private);
  const gchar *charset;
1152

1153
  if (!cache)
1154
    {
1155 1156 1157 1158 1159 1160 1161 1162 1163
      cache = g_new0 (GFilenameCharsetCache, 1);
      g_static_private_set (&cache_private, cache, filename_charset_cache_free);
    }

  g_get_charset (&charset);

  if (!(cache->charset && strcmp (cache->charset, charset) == 0))
    {
      const gchar *new_charset;
1164 1165
      gchar *p;
      gint i;
1166

1167
      g_free (cache->charset);
1168
      g_strfreev (cache->filename_charsets);
1169
      cache->charset = g_strdup (charset);
1170 1171
      
      p = getenv ("G_FILENAME_ENCODING");
1172
      if (p != NULL && p[0] != '\0') 
1173
	{
1174 1175
	  cache->filename_charsets = g_strsplit (p, ",", 0);
	  cache->is_utf8 = (strcmp (cache->filename_charsets[0], "UTF-8") == 0);
1176

1177
	  for (i = 0; cache->filename_charsets[i]; i++)
1178
	    {
1179 1180 1181 1182 1183 1184
	      if (strcmp ("@locale", cache->filename_charsets[i]) == 0)
		{
		  g_get_charset (&new_charset);
		  g_free (cache->filename_charsets[i]);
		  cache->filename_charsets[i] = g_strdup (new_charset);
		}
1185 1186 1187
	    }
	}
      else if (getenv ("G_BROKEN_FILENAMES") != NULL)
1188
	{
1189
	  cache->filename_charsets = g_new0 (gchar *, 2);
1190
	  cache->is_utf8 = g_get_charset (&new_charset);
1191
	  cache->filename_charsets[0] = g_strdup (new_charset);
1192
	}
1193 1194
      else 
	{
1195
	  cache->filename_charsets = g_new0 (gchar *, 3);
1196
	  cache->is_utf8 = TRUE;
1197 1198 1199
	  cache->filename_charsets[0] = g_strdup ("UTF-8");
	  if (!g_get_charset (&new_charset))
	    cache->filename_charsets[1] = g_strdup (new_charset);
1200 1201
	}
    }
1202

1203 1204
  if (filename_charsets)
    *filename_charsets = (const gchar **)cache->filename_charsets;
1205

1206
  return cache->is_utf8;
1207
}
1208

1209
#else /* G_PLATFORM_WIN32 */
1210

1211 1212
gboolean
g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets) 
1213
{
1214
  static const gchar *charsets[] = {
1215 1216 1217 1218
    "UTF-8",
    NULL
  };

1219 1220
#ifdef G_OS_WIN32
  /* On Windows GLib pretends that the filename charset is UTF-8 */
1221 1222 1223
  if (filename_charsets)
    *filename_charsets = charsets;

1224 1225
  return TRUE;
#else
1226 1227
  gboolean result;

1228
  /* Cygwin works like before */
1229 1230 1231 1232 1233 1234
  result = g_get_charset (&(charsets[0]));

  if (filename_charsets)
    *filename_charsets = charsets;

  return result;
1235 1236 1237
#endif
}

1238
#endif /* G_PLATFORM_WIN32 */
1239 1240

static gboolean
1241
get_filename_charset (const gchar **filename_charset)
1242
{
1243 1244 1245 1246
  const gchar **charsets;
  gboolean is_utf8;
  
  is_utf8 = g_get_filename_charsets (&charsets);
1247

1248 1249 1250 1251 1252
  if (filename_charset)
    *filename_charset = charsets[0];
  
  return is_utf8;
}
1253 1254 1255 1256 1257

/* This is called from g_thread_init(). It's used to
 * initialize some static data in a threadsafe way.
 */
void 
1258
_g_convert_thread_init (void)
1259
{
1260
  const gchar **dummy;
Matthias Clasen's avatar
Matthias Clasen committed
1261
  (void) g_get_filename_charsets (&dummy);
1262 1263
}

1264 1265 1266 1267
/**
 * g_filename_to_utf8:
 * @opsysstring:   a string in the encoding for filenames
 * @len:           the length of the string, or -1 if the string is
1268
 *                 nul-terminated<footnoteref linkend="nul-unsafe"/>. 
1269 1270
 * @bytes_read:    location to store the number of bytes in the
 *                 input string that were successfully converted, or %NULL.
Matthias Clasen's avatar
Matthias Clasen committed
1271
 *                 Even if the conversion was successful, this may be 
1272
 *                 less than @len if there were partial characters
1273
 *                 at the end of the input. If the error
1274 1275
 *                 #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
 *                 stored will the byte offset after the last valid
1276
 *                 input sequence.
1277
 * @bytes_written: the number of bytes stored in the output buffer (not 
Matthias Clasen's avatar
Matthias Clasen committed
1278
 *                 including the terminating nul).
1279
 * @error:         location to store the error occuring, or %NULL to ignore
1280 1281
 *                 errors. Any of the errors in #GConvertError may occur.
 * 
1282 1283 1284
 * Converts a string which is in the encoding used by GLib for
 * filenames into a UTF-8 string. Note that on Windows GLib uses UTF-8
 * for filenames.
1285 1286 1287
 * 
 * Return value: The converted string, or %NULL on an error.
 **/
1288
gchar*
1289
g_filename_to_utf8 (const gchar *opsysstring, 
1290 1291 1292
		    gssize       len,           
		    gsize       *bytes_read,   
		    gsize       *bytes_written,
1293
		    GError     **error)
1294
{
1295 1296 1297
  const gchar *charset;

  if (get_filename_charset (&charset))
1298
    return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1299 1300 1301
  else
    return g_convert (opsysstring, len, 
		      "UTF-8", charset, bytes_read, bytes_written, error);
1302 1303
}

1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318
#ifdef G_OS_WIN32

#undef g_filename_to_utf8

/* Binary compatibility version. Not for newly compiled code. */

gchar*
g_filename_to_utf8 (const gchar *opsysstring, 
		    gssize       len,           
		    gsize       *bytes_read,   
		    gsize       *bytes_written,
		    GError     **error)
{
  const gchar *charset;

1319
  if (g_get_charset (&charset))
1320 1321 1322 1323 1324 1325 1326 1327
    return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
  else
    return g_convert (opsysstring, len, 
		      "UTF-8", charset, bytes_read, bytes_written, error);
}

#endif

1328 1329
/**
 * g_filename_from_utf8:
1330
 * @utf8string:    a UTF-8 encoded string.
1331
 * @len:           the length of the string, or -1 if the string is
1332
 *                 nul-terminated.
1333 1334
 * @bytes_read:    location to store the number of bytes in the
 *                 input string that were successfully converted, or %NULL.
Matthias Clasen's avatar
Matthias Clasen committed
1335
 *                 Even if the conversion was successful, this may be 
1336
 *                 less than @len if there were partial characters
1337
 *                 at the end of the input. If the error
1338 1339
 *                 #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
 *                 stored will the byte offset after the last valid
1340
 *                 input sequence.
1341
 * @bytes_written: the number of bytes stored in the output buffer (not 
Matthias Clasen's avatar
Matthias Clasen committed
1342
 *                 including the terminating nul).
1343
 * @error:         location to store the error occuring, or %NULL to ignore
1344 1345
 *                 errors. Any of the errors in #GConvertError may occur.
 * 
1346 1347
 * Converts a string from UTF-8 to the encoding GLib uses for
 * filenames. Note that on Windows GLib uses UTF-8 for filenames.
1348 1349 1350
 * 
 * Return value: The converted string, or %NULL on an error.
 **/
1351
gchar*
1352
g_filename_from_utf8 (const gchar *utf8string,
1353 1354 1355
		      gssize       len,            
		      gsize       *bytes_read,    
		      gsize       *bytes_written,
1356
		      GError     **error)
1357
{
1358 1359 1360
  const gchar *charset;

  if (get_filename_charset (&charset))
1361
    return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1362 1363 1364
  else
    return g_convert (utf8string, len,
		      charset, "UTF-8", bytes_read, bytes_written, error);
1365
}
1366

1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381
#ifdef G_OS_WIN32

#undef g_filename_from_utf8

/* Binary compatibility version. Not for newly compiled code. */

gchar*
g_filename_from_utf8 (const gchar *utf8string,
		      gssize       len,            
		      gsize       *bytes_read,    
		      gsize       *bytes_written,
		      GError     **error)
{
  const gchar *charset;