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 "gthreadprivate.h"
36 #ifdef G_PLATFORM_WIN32
44 #if defined(USE_LIBICONV_GNU) && !defined (_LIBICONV_H)
45 #error GNU libiconv in use but included iconv.h not from libiconv
47 #if !defined(USE_LIBICONV_GNU) && defined (_LIBICONV_H)
48 #error GNU libiconv not in use but included iconv.h is from libiconv
54 g_convert_error_quark (void)
56 return g_quark_from_static_string ("g_convert_error");
60 try_conversion (const char *to_codeset,
61 const char *from_codeset,
64 *cd = iconv_open (to_codeset, from_codeset);
66 if (*cd == (iconv_t)-1 && errno == EINVAL)
73 try_to_aliases (const char **to_aliases,
74 const char *from_codeset,
79 const char **p = to_aliases;
82 if (try_conversion (*p, from_codeset, cd))
92 G_GNUC_INTERNAL extern const char **
93 _g_charset_get_aliases (const char *canonical_name);
97 * @to_codeset: destination codeset
98 * @from_codeset: source codeset
100 * Same as the standard UNIX routine iconv_open(), but
101 * may be implemented via libiconv on UNIX flavors that lack
102 * a native implementation.
104 * GLib provides g_convert() and g_locale_to_utf8() which are likely
105 * more convenient than the raw iconv wrappers.
107 * Return value: a "conversion descriptor", or (GIConv)-1 if
108 * opening the converter failed.
111 g_iconv_open (const gchar *to_codeset,
112 const gchar *from_codeset)
116 if (!try_conversion (to_codeset, from_codeset, &cd))
118 const char **to_aliases = _g_charset_get_aliases (to_codeset);
119 const char **from_aliases = _g_charset_get_aliases (from_codeset);
123 const char **p = from_aliases;
126 if (try_conversion (to_codeset, *p, &cd))
129 if (try_to_aliases (to_aliases, *p, &cd))
136 if (try_to_aliases (to_aliases, from_codeset, &cd))
141 return (cd == (iconv_t)-1) ? (GIConv)-1 : (GIConv)cd;
146 * @converter: conversion descriptor from g_iconv_open()
147 * @inbuf: bytes to convert
148 * @inbytes_left: inout parameter, bytes remaining to convert in @inbuf
149 * @outbuf: converted output bytes
150 * @outbytes_left: inout parameter, bytes available to fill in @outbuf
152 * Same as the standard UNIX routine iconv(), but
153 * may be implemented via libiconv on UNIX flavors that lack
154 * a native implementation.
156 * GLib provides g_convert() and g_locale_to_utf8() which are likely
157 * more convenient than the raw iconv wrappers.
159 * Return value: count of non-reversible conversions, or -1 on error
162 g_iconv (GIConv converter,
166 gsize *outbytes_left)
168 iconv_t cd = (iconv_t)converter;
170 return iconv (cd, inbuf, inbytes_left, outbuf, outbytes_left);
175 * @converter: a conversion descriptor from g_iconv_open()
177 * Same as the standard UNIX routine iconv_close(), but
178 * may be implemented via libiconv on UNIX flavors that lack
179 * a native implementation. Should be called to clean up
180 * the conversion descriptor from g_iconv_open() when
181 * you are done converting things.
183 * GLib provides g_convert() and g_locale_to_utf8() which are likely
184 * more convenient than the raw iconv wrappers.
186 * Return value: -1 on error, 0 on success
189 g_iconv_close (GIConv converter)
191 iconv_t cd = (iconv_t)converter;
193 return iconv_close (cd);
197 #ifdef NEED_ICONV_CACHE
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 * This assumes ownership of @key.
241 * Returns a pointer to the newly allocated cache bucket.
243 static struct _iconv_cache_bucket *
244 iconv_cache_bucket_new (gchar *key, GIConv cd)
246 struct _iconv_cache_bucket *bucket;
248 bucket = g_new (struct _iconv_cache_bucket, 1);
250 bucket->refcount = 1;
254 g_hash_table_insert (iconv_cache, bucket->key, bucket);
256 /* FIXME: if we sorted the list so items with few refcounts were
257 first, then we could expire them faster in iconv_cache_expire_unused () */
258 iconv_cache_list = g_list_prepend (iconv_cache_list, bucket);
267 * iconv_cache_bucket_expire:
268 * @node: cache bucket's node
269 * @bucket: cache bucket
271 * Expires a single cache bucket @bucket. This should only ever be
272 * called on a bucket that currently has no used iconv descriptors
275 * @node is not a required argument. If @node is not supplied, we
276 * search for it ourselves.
279 iconv_cache_bucket_expire (GList *node, struct _iconv_cache_bucket *bucket)
281 g_hash_table_remove (iconv_cache, bucket->key);
284 node = g_list_find (iconv_cache_list, bucket);
286 g_assert (node != NULL);
290 node->prev->next = node->next;
292 node->next->prev = node->prev;
296 iconv_cache_list = node->next;
298 node->next->prev = NULL;
301 g_list_free_1 (node);
303 g_free (bucket->key);
304 g_iconv_close (bucket->cd);
312 * iconv_cache_expire_unused:
314 * Expires as many unused cache buckets as it needs to in order to get
315 * the total number of buckets < ICONV_CACHE_SIZE.
318 iconv_cache_expire_unused (void)
320 struct _iconv_cache_bucket *bucket;
323 node = iconv_cache_list;
324 while (node && iconv_cache_size >= ICONV_CACHE_SIZE)
329 if (bucket->refcount == 0)
330 iconv_cache_bucket_expire (node, bucket);
337 open_converter (const gchar *to_codeset,
338 const gchar *from_codeset,
341 struct _iconv_cache_bucket *bucket;
342 gchar *key, *dyn_key, auto_key[80];
344 gsize len_from_codeset, len_to_codeset;
347 len_from_codeset = strlen (from_codeset);
348 len_to_codeset = strlen (to_codeset);
349 if (len_from_codeset + len_to_codeset + 2 < sizeof (auto_key))
355 key = dyn_key = g_malloc (len_from_codeset + len_to_codeset + 2);
356 memcpy (key, from_codeset, len_from_codeset);
357 key[len_from_codeset] = ':';
358 strcpy (key + len_from_codeset + 1, to_codeset);
360 G_LOCK (iconv_cache_lock);
362 /* make sure the cache has been initialized */
365 bucket = g_hash_table_lookup (iconv_cache, key);
372 cd = g_iconv_open (to_codeset, from_codeset);
373 if (cd == (GIConv) -1)
378 /* Apparently iconv on Solaris <= 7 segfaults if you pass in
379 * NULL for anything but inbuf; work around that. (NULL outbuf
380 * or NULL *outbuf is allowed by Unix98.)
382 gsize inbytes_left = 0;
383 gchar *outbuf = NULL;
384 gsize outbytes_left = 0;
389 /* reset the descriptor */
390 g_iconv (cd, NULL, &inbytes_left, &outbuf, &outbytes_left);
397 cd = g_iconv_open (to_codeset, from_codeset);
398 if (cd == (GIConv) -1)
404 iconv_cache_expire_unused ();
406 bucket = iconv_cache_bucket_new (dyn_key ? dyn_key : g_strdup (key), cd);
409 g_hash_table_insert (iconv_open_hash, cd, bucket->key);
411 G_UNLOCK (iconv_cache_lock);
417 G_UNLOCK (iconv_cache_lock);
419 /* Something went wrong. */
423 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NO_CONVERSION,
424 _("Conversion from character set '%s' to '%s' is not supported"),
425 from_codeset, to_codeset);
427 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
428 _("Could not open converter from '%s' to '%s'"),
429 from_codeset, to_codeset);
436 close_converter (GIConv converter)
438 struct _iconv_cache_bucket *bucket;
444 if (cd == (GIConv) -1)
447 G_LOCK (iconv_cache_lock);
449 key = g_hash_table_lookup (iconv_open_hash, cd);
452 g_hash_table_remove (iconv_open_hash, cd);
454 bucket = g_hash_table_lookup (iconv_cache, key);
459 if (cd == bucket->cd)
460 bucket->used = FALSE;
464 if (!bucket->refcount && iconv_cache_size > ICONV_CACHE_SIZE)
466 /* expire this cache bucket */
467 iconv_cache_bucket_expire (NULL, bucket);
472 G_UNLOCK (iconv_cache_lock);
474 g_warning ("This iconv context wasn't opened using open_converter");
476 return g_iconv_close (converter);
479 G_UNLOCK (iconv_cache_lock);
484 #else /* !NEED_ICONV_CACHE */
487 open_converter (const gchar *to_codeset,
488 const gchar *from_codeset,
493 cd = g_iconv_open (to_codeset, from_codeset);
495 if (cd == (GIConv) -1)
497 /* Something went wrong. */
501 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NO_CONVERSION,
502 _("Conversion from character set '%s' to '%s' is not supported"),
503 from_codeset, to_codeset);
505 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
506 _("Could not open converter from '%s' to '%s'"),
507 from_codeset, to_codeset);
515 close_converter (GIConv cd)
517 if (cd == (GIConv) -1)
520 return g_iconv_close (cd);
523 #endif /* NEED_ICONV_CACHE */
526 * g_convert_with_iconv:
527 * @str: the string to convert
528 * @len: the length of the string, or -1 if the string is
529 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
530 * @converter: conversion descriptor from g_iconv_open()
531 * @bytes_read: location to store the number of bytes in the
532 * input string that were successfully converted, or %NULL.
533 * Even if the conversion was successful, this may be
534 * less than @len if there were partial characters
535 * at the end of the input. If the error
536 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
537 * stored will the byte offset after the last valid
539 * @bytes_written: the number of bytes stored in the output buffer (not
540 * including the terminating nul).
541 * @error: location to store the error occuring, or %NULL to ignore
542 * errors. Any of the errors in #GConvertError may occur.
544 * Converts a string from one character set to another.
546 * Note that you should use g_iconv() for streaming
547 * conversions<footnote id="streaming-state">
549 * Despite the fact that @byes_read can return information about partial
550 * characters, the <literal>g_convert_...</literal> functions
551 * are not generally suitable for streaming. If the underlying converter
552 * being used maintains internal state, then this won't be preserved
553 * across successive calls to g_convert(), g_convert_with_iconv() or
554 * g_convert_with_fallback(). (An example of this is the GNU C converter
555 * for CP1255 which does not emit a base character until it knows that
556 * the next character is not a mark that could combine with the base
561 * Return value: If the conversion was successful, a newly allocated
562 * nul-terminated string, which must be freed with
563 * g_free(). Otherwise %NULL and @error will be set.
566 g_convert_with_iconv (const gchar *str,
570 gsize *bytes_written,
576 gsize inbytes_remaining;
577 gsize outbytes_remaining;
580 gboolean have_error = FALSE;
581 gboolean done = FALSE;
582 gboolean reset = FALSE;
584 g_return_val_if_fail (converter != (GIConv) -1, NULL);
590 inbytes_remaining = len;
591 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
593 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
594 outp = dest = g_malloc (outbuf_size);
596 while (!done && !have_error)
599 err = g_iconv (converter, NULL, &inbytes_remaining, &outp, &outbytes_remaining);
601 err = g_iconv (converter, (char **)&p, &inbytes_remaining, &outp, &outbytes_remaining);
603 if (err == (gsize) -1)
608 /* Incomplete text, do not report an error */
613 gsize used = outp - dest;
616 dest = g_realloc (dest, outbuf_size);
619 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
624 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
625 _("Invalid byte sequence in conversion input"));
630 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
631 _("Error during conversion: %s"),
641 /* call g_iconv with NULL inbuf to cleanup shift state */
643 inbytes_remaining = 0;
653 *bytes_read = p - str;
656 if ((p - str) != len)
661 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_PARTIAL_INPUT,
662 _("Partial character sequence at end of input"));
669 *bytes_written = outp - dest; /* Doesn't include '\0' */
682 * @str: the string to convert
683 * @len: the length of the string, or -1 if the string is
684 * nul-terminated<footnote id="nul-unsafe">
686 Note that some encodings may allow nul bytes to
687 occur inside strings. In that case, using -1 for
688 the @len parameter is unsafe.
691 * @to_codeset: name of character set into which to convert @str
692 * @from_codeset: character set of @str.
693 * @bytes_read: location to store the number of bytes in the
694 * input string that were successfully converted, or %NULL.
695 * Even if the conversion was successful, this may be
696 * less than @len if there were partial characters
697 * at the end of the input. If the error
698 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
699 * stored will the byte offset after the last valid
701 * @bytes_written: the number of bytes stored in the output buffer (not
702 * including the terminating nul).
703 * @error: location to store the error occuring, or %NULL to ignore
704 * errors. Any of the errors in #GConvertError may occur.
706 * Converts a string from one character set to another.
708 * Note that you should use g_iconv() for streaming
709 * conversions<footnoteref linkend="streaming-state"/>.
711 * Return value: If the conversion was successful, a newly allocated
712 * nul-terminated string, which must be freed with
713 * g_free(). Otherwise %NULL and @error will be set.
716 g_convert (const gchar *str,
718 const gchar *to_codeset,
719 const gchar *from_codeset,
721 gsize *bytes_written,
727 g_return_val_if_fail (str != NULL, NULL);
728 g_return_val_if_fail (to_codeset != NULL, NULL);
729 g_return_val_if_fail (from_codeset != NULL, NULL);
731 cd = open_converter (to_codeset, from_codeset, error);
733 if (cd == (GIConv) -1)
744 res = g_convert_with_iconv (str, len, cd,
745 bytes_read, bytes_written,
748 close_converter (cd);
754 * g_convert_with_fallback:
755 * @str: the string to convert
756 * @len: the length of the string, or -1 if the string is
757 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
758 * @to_codeset: name of character set into which to convert @str
759 * @from_codeset: character set of @str.
760 * @fallback: UTF-8 string to use in place of character not
761 * present in the target encoding. (The string must be
762 * representable in the target encoding).
763 If %NULL, characters not in the target encoding will
764 be represented as Unicode escapes \uxxxx or \Uxxxxyyyy.
765 * @bytes_read: location to store the number of bytes in the
766 * input string that were successfully converted, or %NULL.
767 * Even if the conversion was successful, this may be
768 * less than @len if there were partial characters
769 * at the end of the input.
770 * @bytes_written: the number of bytes stored in the output buffer (not
771 * including the terminating nul).
772 * @error: location to store the error occuring, or %NULL to ignore
773 * errors. Any of the errors in #GConvertError may occur.
775 * Converts a string from one character set to another, possibly
776 * including fallback sequences for characters not representable
777 * in the output. Note that it is not guaranteed that the specification
778 * for the fallback sequences in @fallback will be honored. Some
779 * systems may do a approximate conversion from @from_codeset
780 * to @to_codeset in their iconv() functions,
781 * in which case GLib will simply return that approximate conversion.
783 * Note that you should use g_iconv() for streaming
784 * conversions<footnoteref linkend="streaming-state"/>.
786 * Return value: If the conversion was successful, a newly allocated
787 * nul-terminated string, which must be freed with
788 * g_free(). Otherwise %NULL and @error will be set.
791 g_convert_with_fallback (const gchar *str,
793 const gchar *to_codeset,
794 const gchar *from_codeset,
797 gsize *bytes_written,
803 const gchar *insert_str = NULL;
805 gsize inbytes_remaining;
806 const gchar *save_p = NULL;
807 gsize save_inbytes = 0;
808 gsize outbytes_remaining;
812 gboolean have_error = FALSE;
813 gboolean done = FALSE;
815 GError *local_error = NULL;
817 g_return_val_if_fail (str != NULL, NULL);
818 g_return_val_if_fail (to_codeset != NULL, NULL);
819 g_return_val_if_fail (from_codeset != NULL, NULL);
824 /* Try an exact conversion; we only proceed if this fails
825 * due to an illegal sequence in the input string.
827 dest = g_convert (str, len, to_codeset, from_codeset,
828 bytes_read, bytes_written, &local_error);
832 if (!g_error_matches (local_error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE))
834 g_propagate_error (error, local_error);
838 g_error_free (local_error);
842 /* No go; to proceed, we need a converter from "UTF-8" to
843 * to_codeset, and the string as UTF-8.
845 cd = open_converter (to_codeset, "UTF-8", error);
846 if (cd == (GIConv) -1)
857 utf8 = g_convert (str, len, "UTF-8", from_codeset,
858 bytes_read, &inbytes_remaining, error);
861 close_converter (cd);
867 /* Now the heart of the code. We loop through the UTF-8 string, and
868 * whenever we hit an offending character, we form fallback, convert
869 * the fallback to the target codeset, and then go back to
870 * converting the original string after finishing with the fallback.
872 * The variables save_p and save_inbytes store the input state
873 * for the original string while we are converting the fallback
877 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
878 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
879 outp = dest = g_malloc (outbuf_size);
881 while (!done && !have_error)
883 gsize inbytes_tmp = inbytes_remaining;
884 err = g_iconv (cd, (char **)&p, &inbytes_tmp, &outp, &outbytes_remaining);
885 inbytes_remaining = inbytes_tmp;
887 if (err == (gsize) -1)
892 g_assert_not_reached();
896 gsize used = outp - dest;
899 dest = g_realloc (dest, outbuf_size);
902 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
909 /* Error converting fallback string - fatal
911 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
912 _("Cannot convert fallback '%s' to codeset '%s'"),
913 insert_str, to_codeset);
921 gunichar ch = g_utf8_get_char (p);
922 insert_str = g_strdup_printf (ch < 0x10000 ? "\\u%04x" : "\\U%08x",
926 insert_str = fallback;
928 save_p = g_utf8_next_char (p);
929 save_inbytes = inbytes_remaining - (save_p - p);
931 inbytes_remaining = strlen (p);
934 /* fall thru if p is NULL */
936 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
937 _("Error during conversion: %s"),
948 g_free ((gchar *)insert_str);
950 inbytes_remaining = save_inbytes;
955 /* call g_iconv with NULL inbuf to cleanup shift state */
957 inbytes_remaining = 0;
968 close_converter (cd);
971 *bytes_written = outp - dest; /* Doesn't include '\0' */
977 if (save_p && !fallback)
978 g_free ((gchar *)insert_str);
993 strdup_len (const gchar *string,
995 gsize *bytes_written,
1002 if (!g_utf8_validate (string, len, NULL))
1009 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
1010 _("Invalid byte sequence in conversion input"));
1015 real_len = strlen (string);
1020 while (real_len < len && string[real_len])
1025 *bytes_read = real_len;
1027 *bytes_written = real_len;
1029 return g_strndup (string, real_len);
1034 * @opsysstring: a string in the encoding of the current locale. On Windows
1035 * this means the system codepage.
1036 * @len: the length of the string, or -1 if the string is
1037 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
1038 * @bytes_read: location to store the number of bytes in the
1039 * input string that were successfully converted, or %NULL.
1040 * Even if the conversion was successful, this may be
1041 * less than @len if there were partial characters
1042 * at the end of the input. If the error
1043 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1044 * stored will the byte offset after the last valid
1046 * @bytes_written: the number of bytes stored in the output buffer (not
1047 * including the terminating nul).
1048 * @error: location to store the error occuring, or %NULL to ignore
1049 * errors. Any of the errors in #GConvertError may occur.
1051 * Converts a string which is in the encoding used for strings by
1052 * the C runtime (usually the same as that used by the operating
1053 * system) in the <link linkend="setlocale">current locale</link> into a
1056 * Return value: The converted string, or %NULL on an error.
1059 g_locale_to_utf8 (const gchar *opsysstring,
1062 gsize *bytes_written,
1065 const char *charset;
1067 if (g_get_charset (&charset))
1068 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1070 return g_convert (opsysstring, len,
1071 "UTF-8", charset, bytes_read, bytes_written, error);
1075 * g_locale_from_utf8:
1076 * @utf8string: a UTF-8 encoded string
1077 * @len: the length of the string, or -1 if the string is
1078 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
1079 * @bytes_read: location to store the number of bytes in the
1080 * input string that were successfully converted, or %NULL.
1081 * Even if the conversion was successful, this may be
1082 * less than @len if there were partial characters
1083 * at the end of the input. If the error
1084 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1085 * stored will the byte offset after the last valid
1087 * @bytes_written: the number of bytes stored in the output buffer (not
1088 * including the terminating nul).
1089 * @error: location to store the error occuring, or %NULL to ignore
1090 * errors. Any of the errors in #GConvertError may occur.
1092 * Converts a string from UTF-8 to the encoding used for strings by
1093 * the C runtime (usually the same as that used by the operating
1094 * system) in the <link linkend="setlocale">current locale</link>.
1096 * Return value: The converted string, or %NULL on an error.
1099 g_locale_from_utf8 (const gchar *utf8string,
1102 gsize *bytes_written,
1105 const gchar *charset;
1107 if (g_get_charset (&charset))
1108 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1110 return g_convert (utf8string, len,
1111 charset, "UTF-8", bytes_read, bytes_written, error);
1114 #ifndef G_PLATFORM_WIN32
1116 typedef struct _GFilenameCharsetCache GFilenameCharsetCache;
1118 struct _GFilenameCharsetCache {
1121 gchar **filename_charsets;
1125 filename_charset_cache_free (gpointer data)
1127 GFilenameCharsetCache *cache = data;
1128 g_free (cache->charset);
1129 g_strfreev (cache->filename_charsets);
1134 * g_get_filename_charsets:
1135 * @charsets: return location for the %NULL-terminated list of encoding names
1137 * Determines the preferred character sets used for filenames.
1138 * The first character set from the @charsets is the filename encoding, the
1139 * subsequent character sets are used when trying to generate a displayable
1140 * representation of a filename, see g_filename_display_name().
1142 * On Unix, the character sets are determined by consulting the
1143 * environment variables <envar>G_FILENAME_ENCODING</envar> and
1144 * <envar>G_BROKEN_FILENAMES</envar>. On Windows, the character set
1145 * used in the GLib API is always UTF-8 and said environment variables
1148 * <envar>G_FILENAME_ENCODING</envar> may be set to a comma-separated list
1149 * of character set names. The special token "@locale" is taken to
1150 * mean the character set for the <link linkend="setlocale">current
1151 * locale</link>. If <envar>G_FILENAME_ENCODING</envar> is not set, but
1152 * <envar>G_BROKEN_FILENAMES</envar> is, the character set of the current
1153 * locale is taken as the filename encoding. If neither environment variable
1154 * is set, UTF-8 is taken as the filename encoding, but the character
1155 * set of the current locale is also put in the list of encodings.
1157 * The returned @charsets belong to GLib and must not be freed.
1159 * Note that on Unix, regardless of the locale character set or
1160 * <envar>G_FILENAME_ENCODING</envar> value, the actual file names present
1161 * on a system might be in any random encoding or just gibberish.
1163 * Return value: %TRUE if the filename encoding is UTF-8.
1168 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1170 static GStaticPrivate cache_private = G_STATIC_PRIVATE_INIT;
1171 GFilenameCharsetCache *cache = g_static_private_get (&cache_private);
1172 const gchar *charset;
1176 cache = g_new0 (GFilenameCharsetCache, 1);
1177 g_static_private_set (&cache_private, cache, filename_charset_cache_free);
1180 g_get_charset (&charset);
1182 if (!(cache->charset && strcmp (cache->charset, charset) == 0))
1184 const gchar *new_charset;
1188 g_free (cache->charset);
1189 g_strfreev (cache->filename_charsets);
1190 cache->charset = g_strdup (charset);
1192 p = getenv ("G_FILENAME_ENCODING");
1193 if (p != NULL && p[0] != '\0')
1195 cache->filename_charsets = g_strsplit (p, ",", 0);
1196 cache->is_utf8 = (strcmp (cache->filename_charsets[0], "UTF-8") == 0);
1198 for (i = 0; cache->filename_charsets[i]; i++)
1200 if (strcmp ("@locale", cache->filename_charsets[i]) == 0)
1202 g_get_charset (&new_charset);
1203 g_free (cache->filename_charsets[i]);
1204 cache->filename_charsets[i] = g_strdup (new_charset);
1208 else if (getenv ("G_BROKEN_FILENAMES") != NULL)
1210 cache->filename_charsets = g_new0 (gchar *, 2);
1211 cache->is_utf8 = g_get_charset (&new_charset);
1212 cache->filename_charsets[0] = g_strdup (new_charset);
1216 cache->filename_charsets = g_new0 (gchar *, 3);
1217 cache->is_utf8 = TRUE;
1218 cache->filename_charsets[0] = g_strdup ("UTF-8");
1219 if (!g_get_charset (&new_charset))
1220 cache->filename_charsets[1] = g_strdup (new_charset);
1224 if (filename_charsets)
1225 *filename_charsets = (const gchar **)cache->filename_charsets;
1227 return cache->is_utf8;
1230 #else /* G_PLATFORM_WIN32 */
1233 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1235 static const gchar *charsets[] = {
1241 /* On Windows GLib pretends that the filename charset is UTF-8 */
1242 if (filename_charsets)
1243 *filename_charsets = charsets;
1249 /* Cygwin works like before */
1250 result = g_get_charset (&(charsets[0]));
1252 if (filename_charsets)
1253 *filename_charsets = charsets;
1259 #endif /* G_PLATFORM_WIN32 */
1262 get_filename_charset (const gchar **filename_charset)
1264 const gchar **charsets;
1267 is_utf8 = g_get_filename_charsets (&charsets);
1269 if (filename_charset)
1270 *filename_charset = charsets[0];
1275 /* This is called from g_thread_init(). It's used to
1276 * initialize some static data in a threadsafe way.
1279 _g_convert_thread_init (void)
1281 const gchar **dummy;
1282 (void) g_get_filename_charsets (&dummy);
1286 * g_filename_to_utf8:
1287 * @opsysstring: a string in the encoding for filenames
1288 * @len: the length of the string, or -1 if the string is
1289 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
1290 * @bytes_read: location to store the number of bytes in the
1291 * input string that were successfully converted, or %NULL.
1292 * Even if the conversion was successful, this may be
1293 * less than @len if there were partial characters
1294 * at the end of the input. If the error
1295 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1296 * stored will the byte offset after the last valid
1298 * @bytes_written: the number of bytes stored in the output buffer (not
1299 * including the terminating nul).
1300 * @error: location to store the error occuring, or %NULL to ignore
1301 * errors. Any of the errors in #GConvertError may occur.
1303 * Converts a string which is in the encoding used by GLib for
1304 * filenames into a UTF-8 string. Note that on Windows GLib uses UTF-8
1305 * for filenames; on other platforms, this function indirectly depends on
1306 * the <link linkend="setlocale">current locale</link>.
1308 * Return value: The converted string, or %NULL on an error.
1311 g_filename_to_utf8 (const gchar *opsysstring,
1314 gsize *bytes_written,
1317 const gchar *charset;
1319 if (get_filename_charset (&charset))
1320 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1322 return g_convert (opsysstring, len,
1323 "UTF-8", charset, bytes_read, bytes_written, error);
1328 #undef g_filename_to_utf8
1330 /* Binary compatibility version. Not for newly compiled code. */
1333 g_filename_to_utf8 (const gchar *opsysstring,
1336 gsize *bytes_written,
1339 const gchar *charset;
1341 if (g_get_charset (&charset))
1342 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1344 return g_convert (opsysstring, len,
1345 "UTF-8", charset, bytes_read, bytes_written, error);
1351 * g_filename_from_utf8:
1352 * @utf8string: a UTF-8 encoded string.
1353 * @len: the length of the string, or -1 if the string is
1355 * @bytes_read: location to store the number of bytes in the
1356 * input string that were successfully converted, or %NULL.
1357 * Even if the conversion was successful, this may be
1358 * less than @len if there were partial characters
1359 * at the end of the input. If the error
1360 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1361 * stored will the byte offset after the last valid
1363 * @bytes_written: the number of bytes stored in the output buffer (not
1364 * including the terminating nul).
1365 * @error: location to store the error occuring, or %NULL to ignore
1366 * errors. Any of the errors in #GConvertError may occur.
1368 * Converts a string from UTF-8 to the encoding GLib uses for
1369 * filenames. Note that on Windows GLib uses UTF-8 for filenames;
1370 * on other platforms, this function indirectly depends on the
1371 * <link linkend="setlocale">current locale</link>.
1373 * Return value: The converted string, or %NULL on an error.
1376 g_filename_from_utf8 (const gchar *utf8string,
1379 gsize *bytes_written,
1382 const gchar *charset;
1384 if (get_filename_charset (&charset))
1385 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1387 return g_convert (utf8string, len,
1388 charset, "UTF-8", bytes_read, bytes_written, error);
1393 #undef g_filename_from_utf8
1395 /* Binary compatibility version. Not for newly compiled code. */
1398 g_filename_from_utf8 (const gchar *utf8string,
1401 gsize *bytes_written,
1404 const gchar *charset;
1406 if (g_get_charset (&charset))
1407 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1409 return g_convert (utf8string, len,
1410 charset, "UTF-8", bytes_read, bytes_written, error);
1415 /* Test of haystack has the needle prefix, comparing case
1416 * insensitive. haystack may be UTF-8, but needle must
1417 * contain only ascii. */
1419 has_case_prefix (const gchar *haystack, const gchar *needle)
1423 /* Eat one character at a time. */
1428 g_ascii_tolower (*n) == g_ascii_tolower (*h))
1438 UNSAFE_ALL = 0x1, /* Escape all unsafe characters */
1439 UNSAFE_ALLOW_PLUS = 0x2, /* Allows '+' */
1440 UNSAFE_PATH = 0x8, /* Allows '/', '&', '=', ':', '@', '+', '$' and ',' */
1441 UNSAFE_HOST = 0x10, /* Allows '/' and ':' and '@' */
1442 UNSAFE_SLASHES = 0x20 /* Allows all characters except for '/' and '%' */
1443 } UnsafeCharacterSet;
1445 static const guchar acceptable[96] = {
1446 /* A table of the ASCII chars from space (32) to DEL (127) */
1447 /* ! " # $ % & ' ( ) * + , - . / */
1448 0x00,0x3F,0x20,0x20,0x28,0x00,0x2C,0x3F,0x3F,0x3F,0x3F,0x2A,0x28,0x3F,0x3F,0x1C,
1449 /* 0 1 2 3 4 5 6 7 8 9 : ; < = > ? */
1450 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x38,0x20,0x20,0x2C,0x20,0x20,
1451 /* @ A B C D E F G H I J K L M N O */
1452 0x38,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1453 /* P Q R S T U V W X Y Z [ \ ] ^ _ */
1454 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x20,0x3F,
1455 /* ` a b c d e f g h i j k l m n o */
1456 0x20,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1457 /* p q r s t u v w x y z { | } ~ DEL */
1458 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x3F,0x20
1461 static const gchar hex[16] = "0123456789ABCDEF";
1463 /* Note: This escape function works on file: URIs, but if you want to
1464 * escape something else, please read RFC-2396 */
1466 g_escape_uri_string (const gchar *string,
1467 UnsafeCharacterSet mask)
1469 #define ACCEPTABLE(a) ((a)>=32 && (a)<128 && (acceptable[(a)-32] & use_mask))
1476 UnsafeCharacterSet use_mask;
1478 g_return_val_if_fail (mask == UNSAFE_ALL
1479 || mask == UNSAFE_ALLOW_PLUS
1480 || mask == UNSAFE_PATH
1481 || mask == UNSAFE_HOST
1482 || mask == UNSAFE_SLASHES, NULL);
1486 for (p = string; *p != '\0'; p++)
1489 if (!ACCEPTABLE (c))
1493 result = g_malloc (p - string + unacceptable * 2 + 1);
1496 for (q = result, p = string; *p != '\0'; p++)
1500 if (!ACCEPTABLE (c))
1502 *q++ = '%'; /* means hex coming */
1517 g_escape_file_uri (const gchar *hostname,
1518 const gchar *pathname)
1520 char *escaped_hostname = NULL;
1525 char *p, *backslash;
1527 /* Turn backslashes into forward slashes. That's what Netscape
1528 * does, and they are actually more or less equivalent in Windows.
1531 pathname = g_strdup (pathname);
1532 p = (char *) pathname;
1534 while ((backslash = strchr (p, '\\')) != NULL)
1541 if (hostname && *hostname != '\0')
1543 escaped_hostname = g_escape_uri_string (hostname, UNSAFE_HOST);
1546 escaped_path = g_escape_uri_string (pathname, UNSAFE_PATH);
1548 res = g_strconcat ("file://",
1549 (escaped_hostname) ? escaped_hostname : "",
1550 (*escaped_path != '/') ? "/" : "",
1555 g_free ((char *) pathname);
1558 g_free (escaped_hostname);
1559 g_free (escaped_path);
1565 unescape_character (const char *scanner)
1570 first_digit = g_ascii_xdigit_value (scanner[0]);
1571 if (first_digit < 0)
1574 second_digit = g_ascii_xdigit_value (scanner[1]);
1575 if (second_digit < 0)
1578 return (first_digit << 4) | second_digit;
1582 g_unescape_uri_string (const char *escaped,
1584 const char *illegal_escaped_characters,
1585 gboolean ascii_must_not_be_escaped)
1587 const gchar *in, *in_end;
1588 gchar *out, *result;
1591 if (escaped == NULL)
1595 len = strlen (escaped);
1597 result = g_malloc (len + 1);
1600 for (in = escaped, in_end = escaped + len; in < in_end; in++)
1606 /* catch partial escape sequences past the end of the substring */
1607 if (in + 3 > in_end)
1610 c = unescape_character (in + 1);
1612 /* catch bad escape sequences and NUL characters */
1616 /* catch escaped ASCII */
1617 if (ascii_must_not_be_escaped && c <= 0x7F)
1620 /* catch other illegal escaped characters */
1621 if (strchr (illegal_escaped_characters, c) != NULL)
1630 g_assert (out - result <= len);
1643 is_asciialphanum (gunichar c)
1645 return c <= 0x7F && g_ascii_isalnum (c);
1649 is_asciialpha (gunichar c)
1651 return c <= 0x7F && g_ascii_isalpha (c);
1654 /* allows an empty string */
1656 hostname_validate (const char *hostname)
1659 gunichar c, first_char, last_char;
1666 /* read in a label */
1667 c = g_utf8_get_char (p);
1668 p = g_utf8_next_char (p);
1669 if (!is_asciialphanum (c))
1675 c = g_utf8_get_char (p);
1676 p = g_utf8_next_char (p);
1678 while (is_asciialphanum (c) || c == '-');
1679 if (last_char == '-')
1682 /* if that was the last label, check that it was a toplabel */
1683 if (c == '\0' || (c == '.' && *p == '\0'))
1684 return is_asciialpha (first_char);
1691 * g_filename_from_uri:
1692 * @uri: a uri describing a filename (escaped, encoded in ASCII).
1693 * @hostname: Location to store hostname for the URI, or %NULL.
1694 * If there is no hostname in the URI, %NULL will be
1695 * stored in this location.
1696 * @error: location to store the error occuring, or %NULL to ignore
1697 * errors. Any of the errors in #GConvertError may occur.
1699 * Converts an escaped ASCII-encoded URI to a local filename in the
1700 * encoding used for filenames.
1702 * Return value: a newly-allocated string holding the resulting
1703 * filename, or %NULL on an error.
1706 g_filename_from_uri (const gchar *uri,
1710 const char *path_part;
1711 const char *host_part;
1712 char *unescaped_hostname;
1723 if (!has_case_prefix (uri, "file:/"))
1725 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1726 _("The URI '%s' is not an absolute URI using the \"file\" scheme"),
1731 path_part = uri + strlen ("file:");
1733 if (strchr (path_part, '#') != NULL)
1735 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1736 _("The local file URI '%s' may not include a '#'"),
1741 if (has_case_prefix (path_part, "///"))
1743 else if (has_case_prefix (path_part, "//"))
1746 host_part = path_part;
1748 path_part = strchr (path_part, '/');
1750 if (path_part == NULL)
1752 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1753 _("The URI '%s' is invalid"),
1758 unescaped_hostname = g_unescape_uri_string (host_part, path_part - host_part, "", TRUE);
1760 if (unescaped_hostname == NULL ||
1761 !hostname_validate (unescaped_hostname))
1763 g_free (unescaped_hostname);
1764 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1765 _("The hostname of the URI '%s' is invalid"),
1771 *hostname = unescaped_hostname;
1773 g_free (unescaped_hostname);
1776 filename = g_unescape_uri_string (path_part, -1, "/", FALSE);
1778 if (filename == NULL)
1780 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1781 _("The URI '%s' contains invalidly escaped characters"),
1788 /* Drop localhost */
1789 if (hostname && *hostname != NULL &&
1790 g_ascii_strcasecmp (*hostname, "localhost") == 0)
1796 /* Turn slashes into backslashes, because that's the canonical spelling */
1798 while ((slash = strchr (p, '/')) != NULL)
1804 /* Windows URIs with a drive letter can be like "file://host/c:/foo"
1805 * or "file://host/c|/foo" (some Netscape versions). In those cases, start
1806 * the filename from the drive letter.
1808 if (g_ascii_isalpha (filename[1]))
1810 if (filename[2] == ':')
1812 else if (filename[2] == '|')
1820 result = g_strdup (filename + offs);
1828 #undef g_filename_from_uri
1831 g_filename_from_uri (const gchar *uri,
1835 gchar *utf8_filename;
1836 gchar *retval = NULL;
1838 utf8_filename = g_filename_from_uri_utf8 (uri, hostname, error);
1841 retval = g_locale_from_utf8 (utf8_filename, -1, NULL, NULL, error);
1842 g_free (utf8_filename);
1850 * g_filename_to_uri:
1851 * @filename: an absolute filename specified in the GLib file name encoding,
1852 * which is the on-disk file name bytes on Unix, and UTF-8 on
1854 * @hostname: A UTF-8 encoded hostname, or %NULL for none.
1855 * @error: location to store the error occuring, or %NULL to ignore
1856 * errors. Any of the errors in #GConvertError may occur.
1858 * Converts an absolute filename to an escaped ASCII-encoded URI, with the path
1859 * component following Section 3.3. of RFC 2396.
1861 * Return value: a newly-allocated string holding the resulting
1862 * URI, or %NULL on an error.
1865 g_filename_to_uri (const gchar *filename,
1866 const gchar *hostname,
1871 g_return_val_if_fail (filename != NULL, NULL);
1873 if (!g_path_is_absolute (filename))
1875 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NOT_ABSOLUTE_PATH,
1876 _("The pathname '%s' is not an absolute path"),
1882 !(g_utf8_validate (hostname, -1, NULL)
1883 && hostname_validate (hostname)))
1885 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
1886 _("Invalid hostname"));
1891 /* Don't use localhost unnecessarily */
1892 if (hostname && g_ascii_strcasecmp (hostname, "localhost") == 0)
1896 escaped_uri = g_escape_file_uri (hostname, filename);
1903 #undef g_filename_to_uri
1906 g_filename_to_uri (const gchar *filename,
1907 const gchar *hostname,
1910 gchar *utf8_filename;
1911 gchar *retval = NULL;
1913 utf8_filename = g_locale_to_utf8 (filename, -1, NULL, NULL, error);
1917 retval = g_filename_to_uri_utf8 (utf8_filename, hostname, error);
1918 g_free (utf8_filename);
1927 * g_uri_list_extract_uris:
1928 * @uri_list: an URI list
1930 * Splits an URI list conforming to the text/uri-list
1931 * mime type defined in RFC 2483 into individual URIs,
1932 * discarding any comments. The URIs are not validated.
1934 * Returns: a newly allocated %NULL-terminated list of
1935 * strings holding the individual URIs. The array should
1936 * be freed with g_strfreev().
1941 g_uri_list_extract_uris (const gchar *uri_list)
1952 /* We don't actually try to validate the URI according to RFC
1953 * 2396, or even check for allowed characters - we just ignore
1954 * comments and trim whitespace off the ends. We also
1955 * allow LF delimination as well as the specified CRLF.
1957 * We do allow comments like specified in RFC 2483.
1963 while (g_ascii_isspace (*p))
1967 while (*q && (*q != '\n') && (*q != '\r'))
1973 while (q > p && g_ascii_isspace (*q))
1978 uris = g_slist_prepend (uris, g_strndup (p, q - p + 1));
1983 p = strchr (p, '\n');
1988 result = g_new (gchar *, n_uris + 1);
1990 result[n_uris--] = NULL;
1991 for (u = uris; u; u = u->next)
1992 result[n_uris--] = u->data;
1994 g_slist_free (uris);
2000 * g_filename_display_basename:
2001 * @filename: an absolute pathname in the GLib file name encoding
2003 * Returns the display basename for the particular filename, guaranteed
2004 * to be valid UTF-8. The display name might not be identical to the filename,
2005 * for instance there might be problems converting it to UTF-8, and some files
2006 * can be translated in the display.
2008 * If GLib can not make sense of the encoding of @filename, as a last resort it
2009 * replaces unknown characters with U+FFFD, the Unicode replacement character.
2010 * You can search the result for the UTF-8 encoding of this character (which is
2011 * "\357\277\275" in octal notation) to find out if @filename was in an invalid
2014 * You must pass the whole absolute pathname to this functions so that
2015 * translation of well known locations can be done.
2017 * This function is preferred over g_filename_display_name() if you know the
2018 * whole path, as it allows translation.
2020 * Return value: a newly allocated string containing
2021 * a rendition of the basename of the filename in valid UTF-8
2026 g_filename_display_basename (const gchar *filename)
2031 g_return_val_if_fail (filename != NULL, NULL);
2033 basename = g_path_get_basename (filename);
2034 display_name = g_filename_display_name (basename);
2036 return display_name;
2040 * g_filename_display_name:
2041 * @filename: a pathname hopefully in the GLib file name encoding
2043 * Converts a filename into a valid UTF-8 string. The conversion is
2044 * not necessarily reversible, so you should keep the original around
2045 * and use the return value of this function only for display purposes.
2046 * Unlike g_filename_to_utf8(), the result is guaranteed to be non-%NULL
2047 * even if the filename actually isn't in the GLib file name encoding.
2049 * If GLib can not make sense of the encoding of @filename, as a last resort it
2050 * replaces unknown characters with U+FFFD, the Unicode replacement character.
2051 * You can search the result for the UTF-8 encoding of this character (which is
2052 * "\357\277\275" in octal notation) to find out if @filename was in an invalid
2055 * If you know the whole pathname of the file you should use
2056 * g_filename_display_basename(), since that allows location-based
2057 * translation of filenames.
2059 * Return value: a newly allocated string containing
2060 * a rendition of the filename in valid UTF-8
2065 g_filename_display_name (const gchar *filename)
2068 const gchar **charsets;
2069 gchar *display_name = NULL;
2072 is_utf8 = g_get_filename_charsets (&charsets);
2076 if (g_utf8_validate (filename, -1, NULL))
2077 display_name = g_strdup (filename);
2082 /* Try to convert from the filename charsets to UTF-8.
2083 * Skip the first charset if it is UTF-8.
2085 for (i = is_utf8 ? 1 : 0; charsets[i]; i++)
2087 display_name = g_convert (filename, -1, "UTF-8", charsets[i],
2095 /* if all conversions failed, we replace invalid UTF-8
2096 * by a question mark
2099 display_name = _g_utf8_make_valid (filename);
2101 return display_name;
2104 #define __G_CONVERT_C__
2105 #include "galiasdef.c"