1 /* GLIB - Library of useful routines for C programming
3 * gconvert.c: Convert between character sets using iconv
4 * Copyright Red Hat Inc., 2000
5 * Authors: Havoc Pennington <hp@redhat.com>, Owen Taylor <otaylor@redhat.com
7 * This library is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2 of the License, or (at your option) any later version.
12 * This library is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with this library; if not, write to the
19 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
20 * Boston, MA 02111-1307, USA.
32 #include "gprintfint.h"
33 #include "gthreadinit.h"
35 #ifdef G_PLATFORM_WIN32
43 #if defined(USE_LIBICONV_GNU) && !defined (_LIBICONV_H)
44 #error GNU libiconv in use but included iconv.h not from libiconv
46 #if !defined(USE_LIBICONV_GNU) && defined (_LIBICONV_H)
47 #error GNU libiconv not in use but included iconv.h is from libiconv
53 g_convert_error_quark (void)
57 quark = g_quark_from_static_string ("g_convert_error");
63 try_conversion (const char *to_codeset,
64 const char *from_codeset,
67 *cd = iconv_open (to_codeset, from_codeset);
69 if (*cd == (iconv_t)-1 && errno == EINVAL)
76 try_to_aliases (const char **to_aliases,
77 const char *from_codeset,
82 const char **p = to_aliases;
85 if (try_conversion (*p, from_codeset, cd))
95 extern const char **_g_charset_get_aliases (const char *canonical_name) G_GNUC_INTERNAL;
99 * @to_codeset: destination codeset
100 * @from_codeset: source codeset
102 * Same as the standard UNIX routine iconv_open(), but
103 * may be implemented via libiconv on UNIX flavors that lack
104 * a native implementation.
106 * GLib provides g_convert() and g_locale_to_utf8() which are likely
107 * more convenient than the raw iconv wrappers.
109 * Return value: a "conversion descriptor", or (GIConv)-1 if
110 * opening the converter failed.
113 g_iconv_open (const gchar *to_codeset,
114 const gchar *from_codeset)
118 if (!try_conversion (to_codeset, from_codeset, &cd))
120 const char **to_aliases = _g_charset_get_aliases (to_codeset);
121 const char **from_aliases = _g_charset_get_aliases (from_codeset);
125 const char **p = from_aliases;
128 if (try_conversion (to_codeset, *p, &cd))
131 if (try_to_aliases (to_aliases, *p, &cd))
138 if (try_to_aliases (to_aliases, from_codeset, &cd))
143 return (cd == (iconv_t)-1) ? (GIConv)-1 : (GIConv)cd;
148 * @converter: conversion descriptor from g_iconv_open()
149 * @inbuf: bytes to convert
150 * @inbytes_left: inout parameter, bytes remaining to convert in @inbuf
151 * @outbuf: converted output bytes
152 * @outbytes_left: inout parameter, bytes available to fill in @outbuf
154 * Same as the standard UNIX routine iconv(), but
155 * may be implemented via libiconv on UNIX flavors that lack
156 * a native implementation.
158 * GLib provides g_convert() and g_locale_to_utf8() which are likely
159 * more convenient than the raw iconv wrappers.
161 * Return value: count of non-reversible conversions, or -1 on error
164 g_iconv (GIConv converter,
168 gsize *outbytes_left)
170 iconv_t cd = (iconv_t)converter;
172 return iconv (cd, inbuf, inbytes_left, outbuf, outbytes_left);
177 * @converter: a conversion descriptor from g_iconv_open()
179 * Same as the standard UNIX routine iconv_close(), but
180 * may be implemented via libiconv on UNIX flavors that lack
181 * a native implementation. Should be called to clean up
182 * the conversion descriptor from g_iconv_open() when
183 * you are done converting things.
185 * GLib provides g_convert() and g_locale_to_utf8() which are likely
186 * more convenient than the raw iconv wrappers.
188 * Return value: -1 on error, 0 on success
191 g_iconv_close (GIConv converter)
193 iconv_t cd = (iconv_t)converter;
195 return iconv_close (cd);
199 #define ICONV_CACHE_SIZE (16)
201 struct _iconv_cache_bucket {
208 static GList *iconv_cache_list;
209 static GHashTable *iconv_cache;
210 static GHashTable *iconv_open_hash;
211 static guint iconv_cache_size = 0;
212 G_LOCK_DEFINE_STATIC (iconv_cache_lock);
214 /* caller *must* hold the iconv_cache_lock */
216 iconv_cache_init (void)
218 static gboolean initialized = FALSE;
223 iconv_cache_list = NULL;
224 iconv_cache = g_hash_table_new (g_str_hash, g_str_equal);
225 iconv_open_hash = g_hash_table_new (g_direct_hash, g_direct_equal);
232 * iconv_cache_bucket_new:
234 * @cd: iconv descriptor
236 * Creates a new cache bucket, inserts it into the cache and
237 * increments the cache size.
239 * Returns a pointer to the newly allocated cache bucket.
241 static struct _iconv_cache_bucket *
242 iconv_cache_bucket_new (const gchar *key, GIConv cd)
244 struct _iconv_cache_bucket *bucket;
246 bucket = g_new (struct _iconv_cache_bucket, 1);
247 bucket->key = g_strdup (key);
248 bucket->refcount = 1;
252 g_hash_table_insert (iconv_cache, bucket->key, bucket);
254 /* FIXME: if we sorted the list so items with few refcounts were
255 first, then we could expire them faster in iconv_cache_expire_unused () */
256 iconv_cache_list = g_list_prepend (iconv_cache_list, bucket);
265 * iconv_cache_bucket_expire:
266 * @node: cache bucket's node
267 * @bucket: cache bucket
269 * Expires a single cache bucket @bucket. This should only ever be
270 * called on a bucket that currently has no used iconv descriptors
273 * @node is not a required argument. If @node is not supplied, we
274 * search for it ourselves.
277 iconv_cache_bucket_expire (GList *node, struct _iconv_cache_bucket *bucket)
279 g_hash_table_remove (iconv_cache, bucket->key);
282 node = g_list_find (iconv_cache_list, bucket);
284 g_assert (node != NULL);
288 node->prev->next = node->next;
290 node->next->prev = node->prev;
294 iconv_cache_list = node->next;
296 node->next->prev = NULL;
299 g_list_free_1 (node);
301 g_free (bucket->key);
302 g_iconv_close (bucket->cd);
310 * iconv_cache_expire_unused:
312 * Expires as many unused cache buckets as it needs to in order to get
313 * the total number of buckets < ICONV_CACHE_SIZE.
316 iconv_cache_expire_unused (void)
318 struct _iconv_cache_bucket *bucket;
321 node = iconv_cache_list;
322 while (node && iconv_cache_size >= ICONV_CACHE_SIZE)
327 if (bucket->refcount == 0)
328 iconv_cache_bucket_expire (node, bucket);
335 open_converter (const gchar *to_codeset,
336 const gchar *from_codeset,
339 struct _iconv_cache_bucket *bucket;
344 key = g_alloca (strlen (from_codeset) + strlen (to_codeset) + 2);
345 _g_sprintf (key, "%s:%s", from_codeset, to_codeset);
347 G_LOCK (iconv_cache_lock);
349 /* make sure the cache has been initialized */
352 bucket = g_hash_table_lookup (iconv_cache, key);
357 cd = g_iconv_open (to_codeset, from_codeset);
358 if (cd == (GIConv) -1)
363 /* Apparently iconv on Solaris <= 7 segfaults if you pass in
364 * NULL for anything but inbuf; work around that. (NULL outbuf
365 * or NULL *outbuf is allowed by Unix98.)
367 gsize inbytes_left = 0;
368 gchar *outbuf = NULL;
369 gsize outbytes_left = 0;
374 /* reset the descriptor */
375 g_iconv (cd, NULL, &inbytes_left, &outbuf, &outbytes_left);
382 cd = g_iconv_open (to_codeset, from_codeset);
383 if (cd == (GIConv) -1)
386 iconv_cache_expire_unused ();
388 bucket = iconv_cache_bucket_new (key, cd);
391 g_hash_table_insert (iconv_open_hash, cd, bucket->key);
393 G_UNLOCK (iconv_cache_lock);
399 G_UNLOCK (iconv_cache_lock);
401 /* Something went wrong. */
405 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NO_CONVERSION,
406 _("Conversion from character set '%s' to '%s' is not supported"),
407 from_codeset, to_codeset);
409 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
410 _("Could not open converter from '%s' to '%s'"),
411 from_codeset, to_codeset);
418 close_converter (GIConv converter)
420 struct _iconv_cache_bucket *bucket;
426 if (cd == (GIConv) -1)
429 G_LOCK (iconv_cache_lock);
431 key = g_hash_table_lookup (iconv_open_hash, cd);
434 g_hash_table_remove (iconv_open_hash, cd);
436 bucket = g_hash_table_lookup (iconv_cache, key);
441 if (cd == bucket->cd)
442 bucket->used = FALSE;
446 if (!bucket->refcount && iconv_cache_size > ICONV_CACHE_SIZE)
448 /* expire this cache bucket */
449 iconv_cache_bucket_expire (NULL, bucket);
454 G_UNLOCK (iconv_cache_lock);
456 g_warning ("This iconv context wasn't opened using open_converter");
458 return g_iconv_close (converter);
461 G_UNLOCK (iconv_cache_lock);
467 * g_convert_with_iconv:
468 * @str: the string to convert
469 * @len: the length of the string, or -1 if the string is
470 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
471 * @converter: conversion descriptor from g_iconv_open()
472 * @bytes_read: location to store the number of bytes in the
473 * input string that were successfully converted, or %NULL.
474 * Even if the conversion was successful, this may be
475 * less than @len if there were partial characters
476 * at the end of the input. If the error
477 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
478 * stored will the byte offset after the last valid
480 * @bytes_written: the number of bytes stored in the output buffer (not
481 * including the terminating nul).
482 * @error: location to store the error occuring, or %NULL to ignore
483 * errors. Any of the errors in #GConvertError may occur.
485 * Converts a string from one character set to another.
487 * Note that you should use g_iconv() for streaming
488 * conversions<footnote id="streaming-state">
490 * Despite the fact that @byes_read can return information about partial
491 * characters, the <literal>g_convert_...</literal> functions
492 * are not generally suitable for streaming. If the underlying converter
493 * being used maintains internal state, then this won't be preserved
494 * across successive calls to g_convert(), g_convert_with_iconv() or
495 * g_convert_with_fallback(). (An example of this is the GNU C converter
496 * for CP1255 which does not emit a base character until it knows that
497 * the next character is not a mark that could combine with the base
502 * Return value: If the conversion was successful, a newly allocated
503 * nul-terminated string, which must be freed with
504 * g_free(). Otherwise %NULL and @error will be set.
507 g_convert_with_iconv (const gchar *str,
511 gsize *bytes_written,
517 const gchar *shift_p = NULL;
518 gsize inbytes_remaining;
519 gsize outbytes_remaining;
522 gboolean have_error = FALSE;
523 gboolean done = FALSE;
525 g_return_val_if_fail (converter != (GIConv) -1, NULL);
531 inbytes_remaining = len;
532 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
534 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
535 outp = dest = g_malloc (outbuf_size);
537 while (!done && !have_error)
539 err = g_iconv (converter, (char **)&p, &inbytes_remaining, &outp, &outbytes_remaining);
541 if (err == (size_t) -1)
546 /* Incomplete text, do not report an error */
550 size_t used = outp - dest;
553 dest = g_realloc (dest, outbuf_size);
556 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
561 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
562 _("Invalid byte sequence in conversion input"));
567 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
568 _("Error during conversion: %s"),
578 /* call g_iconv with NULL inbuf to cleanup shift state */
581 inbytes_remaining = 0;
594 *bytes_read = p - str;
597 if ((p - str) != len)
602 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_PARTIAL_INPUT,
603 _("Partial character sequence at end of input"));
610 *bytes_written = outp - dest; /* Doesn't include '\0' */
623 * @str: the string to convert
624 * @len: the length of the string, or -1 if the string is
625 * nul-terminated<footnote id="nul-unsafe">
627 Note that some encodings may allow nul bytes to
628 occur inside strings. In that case, using -1 for
629 the @len parameter is unsafe.
632 * @to_codeset: name of character set into which to convert @str
633 * @from_codeset: character set of @str.
634 * @bytes_read: location to store the number of bytes in the
635 * input string that were successfully converted, or %NULL.
636 * Even if the conversion was successful, this may be
637 * less than @len if there were partial characters
638 * at the end of the input. If the error
639 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
640 * stored will the byte offset after the last valid
642 * @bytes_written: the number of bytes stored in the output buffer (not
643 * including the terminating nul).
644 * @error: location to store the error occuring, or %NULL to ignore
645 * errors. Any of the errors in #GConvertError may occur.
647 * Converts a string from one character set to another.
649 * Note that you should use g_iconv() for streaming
650 * conversions<footnoteref linkend="streaming-state"/>.
652 * Return value: If the conversion was successful, a newly allocated
653 * nul-terminated string, which must be freed with
654 * g_free(). Otherwise %NULL and @error will be set.
657 g_convert (const gchar *str,
659 const gchar *to_codeset,
660 const gchar *from_codeset,
662 gsize *bytes_written,
668 g_return_val_if_fail (str != NULL, NULL);
669 g_return_val_if_fail (to_codeset != NULL, NULL);
670 g_return_val_if_fail (from_codeset != NULL, NULL);
672 cd = open_converter (to_codeset, from_codeset, error);
674 if (cd == (GIConv) -1)
685 res = g_convert_with_iconv (str, len, cd,
686 bytes_read, bytes_written,
689 close_converter (cd);
695 * g_convert_with_fallback:
696 * @str: the string to convert
697 * @len: the length of the string, or -1 if the string is
698 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
699 * @to_codeset: name of character set into which to convert @str
700 * @from_codeset: character set of @str.
701 * @fallback: UTF-8 string to use in place of character not
702 * present in the target encoding. (The string must be
703 * representable in the target encoding).
704 If %NULL, characters not in the target encoding will
705 be represented as Unicode escapes \uxxxx or \Uxxxxyyyy.
706 * @bytes_read: location to store the number of bytes in the
707 * input string that were successfully converted, or %NULL.
708 * Even if the conversion was successful, this may be
709 * less than @len if there were partial characters
710 * at the end of the input.
711 * @bytes_written: the number of bytes stored in the output buffer (not
712 * including the terminating nul).
713 * @error: location to store the error occuring, or %NULL to ignore
714 * errors. Any of the errors in #GConvertError may occur.
716 * Converts a string from one character set to another, possibly
717 * including fallback sequences for characters not representable
718 * in the output. Note that it is not guaranteed that the specification
719 * for the fallback sequences in @fallback will be honored. Some
720 * systems may do a approximate conversion from @from_codeset
721 * to @to_codeset in their iconv() functions,
722 * in which case GLib will simply return that approximate conversion.
724 * Note that you should use g_iconv() for streaming
725 * conversions<footnoteref linkend="streaming-state"/>.
727 * Return value: If the conversion was successful, a newly allocated
728 * nul-terminated string, which must be freed with
729 * g_free(). Otherwise %NULL and @error will be set.
732 g_convert_with_fallback (const gchar *str,
734 const gchar *to_codeset,
735 const gchar *from_codeset,
738 gsize *bytes_written,
744 const gchar *insert_str = NULL;
746 gsize inbytes_remaining;
747 const gchar *save_p = NULL;
748 gsize save_inbytes = 0;
749 gsize outbytes_remaining;
753 gboolean have_error = FALSE;
754 gboolean done = FALSE;
756 GError *local_error = NULL;
758 g_return_val_if_fail (str != NULL, NULL);
759 g_return_val_if_fail (to_codeset != NULL, NULL);
760 g_return_val_if_fail (from_codeset != NULL, NULL);
765 /* Try an exact conversion; we only proceed if this fails
766 * due to an illegal sequence in the input string.
768 dest = g_convert (str, len, to_codeset, from_codeset,
769 bytes_read, bytes_written, &local_error);
773 if (!g_error_matches (local_error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE))
775 g_propagate_error (error, local_error);
779 g_error_free (local_error);
783 /* No go; to proceed, we need a converter from "UTF-8" to
784 * to_codeset, and the string as UTF-8.
786 cd = open_converter (to_codeset, "UTF-8", error);
787 if (cd == (GIConv) -1)
798 utf8 = g_convert (str, len, "UTF-8", from_codeset,
799 bytes_read, &inbytes_remaining, error);
802 close_converter (cd);
808 /* Now the heart of the code. We loop through the UTF-8 string, and
809 * whenever we hit an offending character, we form fallback, convert
810 * the fallback to the target codeset, and then go back to
811 * converting the original string after finishing with the fallback.
813 * The variables save_p and save_inbytes store the input state
814 * for the original string while we are converting the fallback
818 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
819 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
820 outp = dest = g_malloc (outbuf_size);
822 while (!done && !have_error)
824 size_t inbytes_tmp = inbytes_remaining;
825 err = g_iconv (cd, (char **)&p, &inbytes_tmp, &outp, &outbytes_remaining);
826 inbytes_remaining = inbytes_tmp;
828 if (err == (size_t) -1)
833 g_assert_not_reached();
837 size_t used = outp - dest;
840 dest = g_realloc (dest, outbuf_size);
843 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
850 /* Error converting fallback string - fatal
852 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
853 _("Cannot convert fallback '%s' to codeset '%s'"),
854 insert_str, to_codeset);
862 gunichar ch = g_utf8_get_char (p);
863 insert_str = g_strdup_printf (ch < 0x10000 ? "\\u%04x" : "\\U%08x",
867 insert_str = fallback;
869 save_p = g_utf8_next_char (p);
870 save_inbytes = inbytes_remaining - (save_p - p);
872 inbytes_remaining = strlen (p);
875 /* fall thru if p is NULL */
877 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
878 _("Error during conversion: %s"),
889 g_free ((gchar *)insert_str);
891 inbytes_remaining = save_inbytes;
896 /* call g_iconv with NULL inbuf to cleanup shift state */
898 inbytes_remaining = 0;
909 close_converter (cd);
912 *bytes_written = outp - dest; /* Doesn't include '\0' */
918 if (save_p && !fallback)
919 g_free ((gchar *)insert_str);
934 strdup_len (const gchar *string,
936 gsize *bytes_written,
943 if (!g_utf8_validate (string, len, NULL))
950 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
951 _("Invalid byte sequence in conversion input"));
956 real_len = strlen (string);
961 while (real_len < len && string[real_len])
966 *bytes_read = real_len;
968 *bytes_written = real_len;
970 return g_strndup (string, real_len);
975 * @opsysstring: a string in the encoding of the current locale. On Windows
976 * this means the system codepage.
977 * @len: the length of the string, or -1 if the string is
978 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
979 * @bytes_read: location to store the number of bytes in the
980 * input string that were successfully converted, or %NULL.
981 * Even if the conversion was successful, this may be
982 * less than @len if there were partial characters
983 * at the end of the input. If the error
984 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
985 * stored will the byte offset after the last valid
987 * @bytes_written: the number of bytes stored in the output buffer (not
988 * including the terminating nul).
989 * @error: location to store the error occuring, or %NULL to ignore
990 * errors. Any of the errors in #GConvertError may occur.
992 * Converts a string which is in the encoding used for strings by
993 * the C runtime (usually the same as that used by the operating
994 * system) in the current locale into a UTF-8 string.
996 * Return value: The converted string, or %NULL on an error.
999 g_locale_to_utf8 (const gchar *opsysstring,
1002 gsize *bytes_written,
1005 const char *charset;
1007 if (g_get_charset (&charset))
1008 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1010 return g_convert (opsysstring, len,
1011 "UTF-8", charset, bytes_read, bytes_written, error);
1015 * g_locale_from_utf8:
1016 * @utf8string: a UTF-8 encoded string
1017 * @len: the length of the string, or -1 if the string is
1018 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
1019 * @bytes_read: location to store the number of bytes in the
1020 * input string that were successfully converted, or %NULL.
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 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1025 * stored will the byte offset after the last valid
1027 * @bytes_written: the number of bytes stored in the output buffer (not
1028 * including the terminating nul).
1029 * @error: location to store the error occuring, or %NULL to ignore
1030 * errors. Any of the errors in #GConvertError may occur.
1032 * Converts a string from UTF-8 to the encoding used for strings by
1033 * the C runtime (usually the same as that used by the operating
1034 * system) in the current locale.
1036 * Return value: The converted string, or %NULL on an error.
1039 g_locale_from_utf8 (const gchar *utf8string,
1042 gsize *bytes_written,
1045 const gchar *charset;
1047 if (g_get_charset (&charset))
1048 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1050 return g_convert (utf8string, len,
1051 charset, "UTF-8", bytes_read, bytes_written, error);
1054 #ifndef G_PLATFORM_WIN32
1056 typedef struct _GFilenameCharsetCache GFilenameCharsetCache;
1058 struct _GFilenameCharsetCache {
1061 gchar **filename_charsets;
1065 filename_charset_cache_free (gpointer data)
1067 GFilenameCharsetCache *cache = data;
1068 g_free (cache->charset);
1069 g_strfreev (cache->filename_charsets);
1074 * g_get_filename_charsets:
1075 * @charsets: return location for the %NULL-terminated list of encoding names
1077 * Determines the preferred character sets used for filenames.
1078 * The first character set from the @charsets is the filename encoding, the
1079 * subsequent character sets are used when trying to generate a displayable
1080 * representation of a filename, see g_filename_display_name().
1082 * On Unix, the character sets are determined by consulting the
1083 * environment variables <envar>G_FILENAME_ENCODING</envar> and
1084 * <envar>G_BROKEN_FILENAMES</envar>. On Windows, the character set
1085 * used in the GLib API is always UTF-8 and said environment variables
1088 * <envar>G_FILENAME_ENCODING</envar> may be set to a comma-separated list
1089 * of character set names. The special token "@locale" is taken to mean the
1090 * character set for the current locale. If <envar>G_FILENAME_ENCODING</envar>
1091 * is not set, but <envar>G_BROKEN_FILENAMES</envar> is, the character set of
1092 * the current locale is taken as the filename encoding. If neither environment
1093 * variable is set, UTF-8 is taken as the filename encoding, but the character
1094 * set of the current locale is also put in the list of encodings.
1096 * The returned @charsets belong to GLib and must not be freed.
1098 * Note that on Unix, regardless of the locale character set or
1099 * <envar>G_FILENAME_ENCODING</envar> value, the actual file names present on a
1100 * system might be in any random encoding or just gibberish.
1102 * Return value: %TRUE if the filename encoding is UTF-8.
1107 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1109 static GStaticPrivate cache_private = G_STATIC_PRIVATE_INIT;
1110 GFilenameCharsetCache *cache = g_static_private_get (&cache_private);
1111 const gchar *charset;
1115 cache = g_new0 (GFilenameCharsetCache, 1);
1116 g_static_private_set (&cache_private, cache, filename_charset_cache_free);
1119 g_get_charset (&charset);
1121 if (!(cache->charset && strcmp (cache->charset, charset) == 0))
1123 const gchar *new_charset;
1127 g_free (cache->charset);
1128 g_strfreev (cache->filename_charsets);
1129 cache->charset = g_strdup (charset);
1131 p = getenv ("G_FILENAME_ENCODING");
1132 if (p != NULL && p[0] != '\0')
1134 cache->filename_charsets = g_strsplit (p, ",", 0);
1135 cache->is_utf8 = (strcmp (cache->filename_charsets[0], "UTF-8") == 0);
1137 for (i = 0; cache->filename_charsets[i]; i++)
1139 if (strcmp ("@locale", cache->filename_charsets[i]) == 0)
1141 g_get_charset (&new_charset);
1142 g_free (cache->filename_charsets[i]);
1143 cache->filename_charsets[i] = g_strdup (new_charset);
1147 else if (getenv ("G_BROKEN_FILENAMES") != NULL)
1149 cache->filename_charsets = g_new0 (gchar *, 2);
1150 cache->is_utf8 = g_get_charset (&new_charset);
1151 cache->filename_charsets[0] = g_strdup (new_charset);
1155 cache->filename_charsets = g_new0 (gchar *, 3);
1156 cache->is_utf8 = TRUE;
1157 cache->filename_charsets[0] = g_strdup ("UTF-8");
1158 if (!g_get_charset (&new_charset))
1159 cache->filename_charsets[1] = g_strdup (new_charset);
1163 if (filename_charsets)
1164 *filename_charsets = (const gchar **)cache->filename_charsets;
1166 return cache->is_utf8;
1169 #else /* G_PLATFORM_WIN32 */
1172 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1174 static const gchar *charsets[] = {
1180 /* On Windows GLib pretends that the filename charset is UTF-8 */
1181 if (filename_charsets)
1182 *filename_charsets = charsets;
1188 /* Cygwin works like before */
1189 result = g_get_charset (&(charsets[0]));
1191 if (filename_charsets)
1192 *filename_charsets = charsets;
1198 #endif /* G_PLATFORM_WIN32 */
1201 get_filename_charset (const gchar **filename_charset)
1203 const gchar **charsets;
1206 is_utf8 = g_get_filename_charsets (&charsets);
1208 if (filename_charset)
1209 *filename_charset = charsets[0];
1214 /* This is called from g_thread_init(). It's used to
1215 * initialize some static data in a threadsafe way.
1218 _g_convert_thread_init (void)
1220 const gchar **dummy;
1221 (void) g_get_filename_charsets (&dummy);
1225 * g_filename_to_utf8:
1226 * @opsysstring: a string in the encoding for filenames
1227 * @len: the length of the string, or -1 if the string is
1228 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
1229 * @bytes_read: location to store the number of bytes in the
1230 * input string that were successfully converted, or %NULL.
1231 * Even if the conversion was successful, this may be
1232 * less than @len if there were partial characters
1233 * at the end of the input. If the error
1234 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1235 * stored will the byte offset after the last valid
1237 * @bytes_written: the number of bytes stored in the output buffer (not
1238 * including the terminating nul).
1239 * @error: location to store the error occuring, or %NULL to ignore
1240 * errors. Any of the errors in #GConvertError may occur.
1242 * Converts a string which is in the encoding used by GLib for
1243 * filenames into a UTF-8 string. Note that on Windows GLib uses UTF-8
1246 * Return value: The converted string, or %NULL on an error.
1249 g_filename_to_utf8 (const gchar *opsysstring,
1252 gsize *bytes_written,
1255 const gchar *charset;
1257 if (get_filename_charset (&charset))
1258 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1260 return g_convert (opsysstring, len,
1261 "UTF-8", charset, bytes_read, bytes_written, error);
1266 #undef g_filename_to_utf8
1268 /* Binary compatibility version. Not for newly compiled code. */
1271 g_filename_to_utf8 (const gchar *opsysstring,
1274 gsize *bytes_written,
1277 const gchar *charset;
1279 if (g_get_charset (&charset))
1280 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1282 return g_convert (opsysstring, len,
1283 "UTF-8", charset, bytes_read, bytes_written, error);
1289 * g_filename_from_utf8:
1290 * @utf8string: a UTF-8 encoded string.
1291 * @len: the length of the string, or -1 if the string is
1293 * @bytes_read: location to store the number of bytes in the
1294 * input string that were successfully converted, or %NULL.
1295 * Even if the conversion was successful, this may be
1296 * less than @len if there were partial characters
1297 * at the end of the input. If the error
1298 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1299 * stored will the byte offset after the last valid
1301 * @bytes_written: the number of bytes stored in the output buffer (not
1302 * including the terminating nul).
1303 * @error: location to store the error occuring, or %NULL to ignore
1304 * errors. Any of the errors in #GConvertError may occur.
1306 * Converts a string from UTF-8 to the encoding GLib uses for
1307 * filenames. Note that on Windows GLib uses UTF-8 for filenames.
1309 * Return value: The converted string, or %NULL on an error.
1312 g_filename_from_utf8 (const gchar *utf8string,
1315 gsize *bytes_written,
1318 const gchar *charset;
1320 if (get_filename_charset (&charset))
1321 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1323 return g_convert (utf8string, len,
1324 charset, "UTF-8", bytes_read, bytes_written, error);
1329 #undef g_filename_from_utf8
1331 /* Binary compatibility version. Not for newly compiled code. */
1334 g_filename_from_utf8 (const gchar *utf8string,
1337 gsize *bytes_written,
1340 const gchar *charset;
1342 if (g_get_charset (&charset))
1343 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1345 return g_convert (utf8string, len,
1346 charset, "UTF-8", bytes_read, bytes_written, error);
1351 /* Test of haystack has the needle prefix, comparing case
1352 * insensitive. haystack may be UTF-8, but needle must
1353 * contain only ascii. */
1355 has_case_prefix (const gchar *haystack, const gchar *needle)
1359 /* Eat one character at a time. */
1364 g_ascii_tolower (*n) == g_ascii_tolower (*h))
1374 UNSAFE_ALL = 0x1, /* Escape all unsafe characters */
1375 UNSAFE_ALLOW_PLUS = 0x2, /* Allows '+' */
1376 UNSAFE_PATH = 0x8, /* Allows '/', '&', '=', ':', '@', '+', '$' and ',' */
1377 UNSAFE_HOST = 0x10, /* Allows '/' and ':' and '@' */
1378 UNSAFE_SLASHES = 0x20 /* Allows all characters except for '/' and '%' */
1379 } UnsafeCharacterSet;
1381 static const guchar acceptable[96] = {
1382 /* A table of the ASCII chars from space (32) to DEL (127) */
1383 /* ! " # $ % & ' ( ) * + , - . / */
1384 0x00,0x3F,0x20,0x20,0x28,0x00,0x2C,0x3F,0x3F,0x3F,0x3F,0x2A,0x28,0x3F,0x3F,0x1C,
1385 /* 0 1 2 3 4 5 6 7 8 9 : ; < = > ? */
1386 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x38,0x20,0x20,0x2C,0x20,0x20,
1387 /* @ A B C D E F G H I J K L M N O */
1388 0x38,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1389 /* P Q R S T U V W X Y Z [ \ ] ^ _ */
1390 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x20,0x3F,
1391 /* ` a b c d e f g h i j k l m n o */
1392 0x20,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1393 /* p q r s t u v w x y z { | } ~ DEL */
1394 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x3F,0x20
1397 static const gchar hex[16] = "0123456789ABCDEF";
1399 /* Note: This escape function works on file: URIs, but if you want to
1400 * escape something else, please read RFC-2396 */
1402 g_escape_uri_string (const gchar *string,
1403 UnsafeCharacterSet mask)
1405 #define ACCEPTABLE(a) ((a)>=32 && (a)<128 && (acceptable[(a)-32] & use_mask))
1412 UnsafeCharacterSet use_mask;
1414 g_return_val_if_fail (mask == UNSAFE_ALL
1415 || mask == UNSAFE_ALLOW_PLUS
1416 || mask == UNSAFE_PATH
1417 || mask == UNSAFE_HOST
1418 || mask == UNSAFE_SLASHES, NULL);
1422 for (p = string; *p != '\0'; p++)
1425 if (!ACCEPTABLE (c))
1429 result = g_malloc (p - string + unacceptable * 2 + 1);
1432 for (q = result, p = string; *p != '\0'; p++)
1436 if (!ACCEPTABLE (c))
1438 *q++ = '%'; /* means hex coming */
1453 g_escape_file_uri (const gchar *hostname,
1454 const gchar *pathname)
1456 char *escaped_hostname = NULL;
1461 char *p, *backslash;
1463 /* Turn backslashes into forward slashes. That's what Netscape
1464 * does, and they are actually more or less equivalent in Windows.
1467 pathname = g_strdup (pathname);
1468 p = (char *) pathname;
1470 while ((backslash = strchr (p, '\\')) != NULL)
1477 if (hostname && *hostname != '\0')
1479 escaped_hostname = g_escape_uri_string (hostname, UNSAFE_HOST);
1482 escaped_path = g_escape_uri_string (pathname, UNSAFE_PATH);
1484 res = g_strconcat ("file://",
1485 (escaped_hostname) ? escaped_hostname : "",
1486 (*escaped_path != '/') ? "/" : "",
1491 g_free ((char *) pathname);
1494 g_free (escaped_hostname);
1495 g_free (escaped_path);
1501 unescape_character (const char *scanner)
1506 first_digit = g_ascii_xdigit_value (scanner[0]);
1507 if (first_digit < 0)
1510 second_digit = g_ascii_xdigit_value (scanner[1]);
1511 if (second_digit < 0)
1514 return (first_digit << 4) | second_digit;
1518 g_unescape_uri_string (const char *escaped,
1520 const char *illegal_escaped_characters,
1521 gboolean ascii_must_not_be_escaped)
1523 const gchar *in, *in_end;
1524 gchar *out, *result;
1527 if (escaped == NULL)
1531 len = strlen (escaped);
1533 result = g_malloc (len + 1);
1536 for (in = escaped, in_end = escaped + len; in < in_end; in++)
1542 /* catch partial escape sequences past the end of the substring */
1543 if (in + 3 > in_end)
1546 c = unescape_character (in + 1);
1548 /* catch bad escape sequences and NUL characters */
1552 /* catch escaped ASCII */
1553 if (ascii_must_not_be_escaped && c <= 0x7F)
1556 /* catch other illegal escaped characters */
1557 if (strchr (illegal_escaped_characters, c) != NULL)
1566 g_assert (out - result <= len);
1579 is_asciialphanum (gunichar c)
1581 return c <= 0x7F && g_ascii_isalnum (c);
1585 is_asciialpha (gunichar c)
1587 return c <= 0x7F && g_ascii_isalpha (c);
1590 /* allows an empty string */
1592 hostname_validate (const char *hostname)
1595 gunichar c, first_char, last_char;
1602 /* read in a label */
1603 c = g_utf8_get_char (p);
1604 p = g_utf8_next_char (p);
1605 if (!is_asciialphanum (c))
1611 c = g_utf8_get_char (p);
1612 p = g_utf8_next_char (p);
1614 while (is_asciialphanum (c) || c == '-');
1615 if (last_char == '-')
1618 /* if that was the last label, check that it was a toplabel */
1619 if (c == '\0' || (c == '.' && *p == '\0'))
1620 return is_asciialpha (first_char);
1627 * g_filename_from_uri:
1628 * @uri: a uri describing a filename (escaped, encoded in ASCII).
1629 * @hostname: Location to store hostname for the URI, or %NULL.
1630 * If there is no hostname in the URI, %NULL will be
1631 * stored in this location.
1632 * @error: location to store the error occuring, or %NULL to ignore
1633 * errors. Any of the errors in #GConvertError may occur.
1635 * Converts an escaped ASCII-encoded URI to a local filename in the
1636 * encoding used for filenames.
1638 * Return value: a newly-allocated string holding the resulting
1639 * filename, or %NULL on an error.
1642 g_filename_from_uri (const gchar *uri,
1646 const char *path_part;
1647 const char *host_part;
1648 char *unescaped_hostname;
1659 if (!has_case_prefix (uri, "file:/"))
1661 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1662 _("The URI '%s' is not an absolute URI using the \"file\" scheme"),
1667 path_part = uri + strlen ("file:");
1669 if (strchr (path_part, '#') != NULL)
1671 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1672 _("The local file URI '%s' may not include a '#'"),
1677 if (has_case_prefix (path_part, "///"))
1679 else if (has_case_prefix (path_part, "//"))
1682 host_part = path_part;
1684 path_part = strchr (path_part, '/');
1686 if (path_part == NULL)
1688 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1689 _("The URI '%s' is invalid"),
1694 unescaped_hostname = g_unescape_uri_string (host_part, path_part - host_part, "", TRUE);
1696 if (unescaped_hostname == NULL ||
1697 !hostname_validate (unescaped_hostname))
1699 g_free (unescaped_hostname);
1700 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1701 _("The hostname of the URI '%s' is invalid"),
1707 *hostname = unescaped_hostname;
1709 g_free (unescaped_hostname);
1712 filename = g_unescape_uri_string (path_part, -1, "/", FALSE);
1714 if (filename == NULL)
1716 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1717 _("The URI '%s' contains invalidly escaped characters"),
1724 /* Drop localhost */
1725 if (hostname && *hostname != NULL &&
1726 g_ascii_strcasecmp (*hostname, "localhost") == 0)
1732 /* Turn slashes into backslashes, because that's the canonical spelling */
1734 while ((slash = strchr (p, '/')) != NULL)
1740 /* Windows URIs with a drive letter can be like "file://host/c:/foo"
1741 * or "file://host/c|/foo" (some Netscape versions). In those cases, start
1742 * the filename from the drive letter.
1744 if (g_ascii_isalpha (filename[1]))
1746 if (filename[2] == ':')
1748 else if (filename[2] == '|')
1756 result = g_strdup (filename + offs);
1764 #undef g_filename_from_uri
1767 g_filename_from_uri (const gchar *uri,
1771 gchar *utf8_filename;
1772 gchar *retval = NULL;
1774 utf8_filename = g_filename_from_uri_utf8 (uri, hostname, error);
1777 retval = g_locale_from_utf8 (utf8_filename, -1, NULL, NULL, error);
1778 g_free (utf8_filename);
1786 * g_filename_to_uri:
1787 * @filename: an absolute filename specified in the GLib file name encoding,
1788 * which is the on-disk file name bytes on Unix, and UTF-8 on
1790 * @hostname: A UTF-8 encoded hostname, or %NULL for none.
1791 * @error: location to store the error occuring, or %NULL to ignore
1792 * errors. Any of the errors in #GConvertError may occur.
1794 * Converts an absolute filename to an escaped ASCII-encoded URI.
1796 * Return value: a newly-allocated string holding the resulting
1797 * URI, or %NULL on an error.
1800 g_filename_to_uri (const gchar *filename,
1801 const gchar *hostname,
1806 g_return_val_if_fail (filename != NULL, NULL);
1808 if (!g_path_is_absolute (filename))
1810 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NOT_ABSOLUTE_PATH,
1811 _("The pathname '%s' is not an absolute path"),
1817 !(g_utf8_validate (hostname, -1, NULL)
1818 && hostname_validate (hostname)))
1820 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
1821 _("Invalid hostname"));
1826 /* Don't use localhost unnecessarily */
1827 if (hostname && g_ascii_strcasecmp (hostname, "localhost") == 0)
1831 escaped_uri = g_escape_file_uri (hostname, filename);
1838 #undef g_filename_to_uri
1841 g_filename_to_uri (const gchar *filename,
1842 const gchar *hostname,
1845 gchar *utf8_filename;
1846 gchar *retval = NULL;
1848 utf8_filename = g_locale_to_utf8 (filename, -1, NULL, NULL, error);
1852 retval = g_filename_to_uri_utf8 (utf8_filename, hostname, error);
1853 g_free (utf8_filename);
1862 * g_uri_list_extract_uris:
1863 * @uri_list: an URI list
1865 * Splits an URI list conforming to the text/uri-list
1866 * mime type defined in RFC 2483 into individual URIs,
1867 * discarding any comments. The URIs are not validated.
1869 * Returns: a newly allocated %NULL-terminated list of
1870 * strings holding the individual URIs. The array should
1871 * be freed with g_strfreev().
1876 g_uri_list_extract_uris (const gchar *uri_list)
1887 /* We don't actually try to validate the URI according to RFC
1888 * 2396, or even check for allowed characters - we just ignore
1889 * comments and trim whitespace off the ends. We also
1890 * allow LF delimination as well as the specified CRLF.
1892 * We do allow comments like specified in RFC 2483.
1898 while (g_ascii_isspace (*p))
1902 while (*q && (*q != '\n') && (*q != '\r'))
1908 while (q > p && g_ascii_isspace (*q))
1913 uris = g_slist_prepend (uris, g_strndup (p, q - p + 1));
1918 p = strchr (p, '\n');
1923 result = g_new (gchar *, n_uris + 1);
1925 result[n_uris--] = NULL;
1926 for (u = uris; u; u = u->next)
1927 result[n_uris--] = u->data;
1929 g_slist_free (uris);
1935 make_valid_utf8 (const gchar *name)
1938 const gchar *remainder, *invalid;
1939 gint remaining_bytes, valid_bytes;
1943 remaining_bytes = strlen (name);
1945 while (remaining_bytes != 0)
1947 if (g_utf8_validate (remainder, remaining_bytes, &invalid))
1949 valid_bytes = invalid - remainder;
1952 string = g_string_sized_new (remaining_bytes);
1954 g_string_append_len (string, remainder, valid_bytes);
1955 g_string_append_c (string, '?');
1957 remaining_bytes -= valid_bytes + 1;
1958 remainder = invalid + 1;
1962 return g_strdup (name);
1964 g_string_append (string, remainder);
1965 g_string_append (string, " (invalid encoding)");
1967 g_assert (g_utf8_validate (string->str, -1, NULL));
1969 return g_string_free (string, FALSE);
1973 * g_filename_display_basename:
1974 * @filename: an absolute pathname in the GLib file name encoding
1976 * Returns the display basename for the particular filename, guaranteed
1977 * to be valid UTF-8. The display name might not be identical to the filename,
1978 * for instance there might be problems converting it to UTF-8, and some files
1979 * can be translated in the display
1981 * You must pass the whole absolute pathname to this functions so that
1982 * translation of well known locations can be done.
1984 * This function is preferred over g_filename_display_name() if you know the
1985 * whole path, as it allows translation.
1987 * Return value: a newly allocated string containing
1988 * a rendition of the basename of the filename in valid UTF-8
1993 g_filename_display_basename (const gchar *filename)
1998 g_return_val_if_fail (filename != NULL, NULL);
2000 basename = g_path_get_basename (filename);
2001 display_name = g_filename_display_name (basename);
2003 return display_name;
2007 * g_filename_display_name:
2008 * @filename: a pathname hopefully in the GLib file name encoding
2010 * Converts a filename into a valid UTF-8 string. The
2011 * conversion is not necessarily reversible, so you
2012 * should keep the original around and use the return
2013 * value of this function only for display purposes.
2014 * Unlike g_filename_to_utf8(), the result is guaranteed
2015 * to be non-NULL even if the filename actually isn't in the GLib
2016 * file name encoding.
2018 * If you know the whole pathname of the file you should use
2019 * g_filename_display_basename(), since that allows location-based
2020 * translation of filenames.
2022 * Return value: a newly allocated string containing
2023 * a rendition of the filename in valid UTF-8
2028 g_filename_display_name (const gchar *filename)
2031 const gchar **charsets;
2032 gchar *display_name = NULL;
2035 is_utf8 = g_get_filename_charsets (&charsets);
2039 if (g_utf8_validate (filename, -1, NULL))
2040 display_name = g_strdup (filename);
2045 /* Try to convert from the filename charsets to UTF-8.
2046 * Skip the first charset if it is UTF-8.
2048 for (i = is_utf8 ? 1 : 0; charsets[i]; i++)
2050 display_name = g_convert (filename, -1, "UTF-8", charsets[i],
2058 /* if all conversions failed, we replace invalid UTF-8
2059 * by a question mark
2062 display_name = make_valid_utf8 (filename);
2064 return display_name;
2067 #define __G_CONVERT_C__
2068 #include "galiasdef.c"