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. */
403 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NO_CONVERSION,
404 _("Conversion from character set '%s' to '%s' is not supported"),
405 from_codeset, to_codeset);
407 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
408 _("Could not open converter from '%s' to '%s'"),
409 from_codeset, to_codeset);
415 close_converter (GIConv converter)
417 struct _iconv_cache_bucket *bucket;
423 if (cd == (GIConv) -1)
426 G_LOCK (iconv_cache_lock);
428 key = g_hash_table_lookup (iconv_open_hash, cd);
431 g_hash_table_remove (iconv_open_hash, cd);
433 bucket = g_hash_table_lookup (iconv_cache, key);
438 if (cd == bucket->cd)
439 bucket->used = FALSE;
443 if (!bucket->refcount && iconv_cache_size > ICONV_CACHE_SIZE)
445 /* expire this cache bucket */
446 iconv_cache_bucket_expire (NULL, bucket);
451 G_UNLOCK (iconv_cache_lock);
453 g_warning ("This iconv context wasn't opened using open_converter");
455 return g_iconv_close (converter);
458 G_UNLOCK (iconv_cache_lock);
466 * @str: the string to convert
467 * @len: the length of the string
468 * @to_codeset: name of character set into which to convert @str
469 * @from_codeset: character set of @str.
470 * @bytes_read: location to store the number of bytes in the
471 * input string that were successfully converted, or %NULL.
472 * Even if the conversion was successful, this may be
473 * less than @len if there were partial characters
474 * at the end of the input. If the error
475 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
476 * stored will the byte offset after the last valid
478 * @bytes_written: the number of bytes stored in the output buffer (not
479 * including the terminating nul).
480 * @error: location to store the error occuring, or %NULL to ignore
481 * errors. Any of the errors in #GConvertError may occur.
483 * Converts a string from one character set to another.
485 * Return value: If the conversion was successful, a newly allocated
486 * nul-terminated string, which must be freed with
487 * g_free(). Otherwise %NULL and @error will be set.
490 g_convert (const gchar *str,
492 const gchar *to_codeset,
493 const gchar *from_codeset,
495 gsize *bytes_written,
501 g_return_val_if_fail (str != NULL, NULL);
502 g_return_val_if_fail (to_codeset != NULL, NULL);
503 g_return_val_if_fail (from_codeset != NULL, NULL);
505 cd = open_converter (to_codeset, from_codeset, error);
507 if (cd == (GIConv) -1)
518 res = g_convert_with_iconv (str, len, cd,
519 bytes_read, bytes_written,
522 close_converter (cd);
528 * g_convert_with_iconv:
529 * @str: the string to convert
530 * @len: the length of the string
531 * @converter: conversion descriptor from g_iconv_open()
532 * @bytes_read: location to store the number of bytes in the
533 * input string that were successfully converted, or %NULL.
534 * Even if the conversion was successful, this may be
535 * less than @len if there were partial characters
536 * at the end of the input. If the error
537 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
538 * stored will the byte offset after the last valid
540 * @bytes_written: the number of bytes stored in the output buffer (not
541 * including the terminating nul).
542 * @error: location to store the error occuring, or %NULL to ignore
543 * errors. Any of the errors in #GConvertError may occur.
545 * Converts a string from one character set to another.
547 * Return value: If the conversion was successful, a newly allocated
548 * nul-terminated string, which must be freed with
549 * g_free(). Otherwise %NULL and @error will be set.
552 g_convert_with_iconv (const gchar *str,
556 gsize *bytes_written,
562 gsize inbytes_remaining;
563 gsize outbytes_remaining;
566 gboolean have_error = FALSE;
568 g_return_val_if_fail (str != NULL, NULL);
569 g_return_val_if_fail (converter != (GIConv) -1, NULL);
575 inbytes_remaining = len;
576 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
578 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
579 outp = dest = g_malloc (outbuf_size);
583 err = g_iconv (converter, (char **)&p, &inbytes_remaining, &outp, &outbytes_remaining);
585 if (err == (size_t) -1)
590 /* Incomplete text, do not report an error */
594 size_t used = outp - dest;
597 dest = g_realloc (dest, outbuf_size);
600 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
605 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
606 _("Invalid byte sequence in conversion input"));
610 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
611 _("Error during conversion: %s"),
621 *bytes_read = p - str;
624 if ((p - str) != len)
628 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_PARTIAL_INPUT,
629 _("Partial character sequence at end of input"));
636 *bytes_written = outp - dest; /* Doesn't include '\0' */
648 * g_convert_with_fallback:
649 * @str: the string to convert
650 * @len: the length of the string
651 * @to_codeset: name of character set into which to convert @str
652 * @from_codeset: character set of @str.
653 * @fallback: UTF-8 string to use in place of character not
654 * present in the target encoding. (This must be
655 * in the target encoding), if %NULL, characters
656 * not in the target encoding will be represented
657 * as Unicode escapes \uxxxx or \Uxxxxyyyy.
658 * @bytes_read: location to store the number of bytes in the
659 * input string that were successfully converted, or %NULL.
660 * Even if the conversion was successful, this may be
661 * less than @len if there were partial characters
662 * at the end of the input.
663 * @bytes_written: the number of bytes stored in the output buffer (not
664 * including the terminating nul).
665 * @error: location to store the error occuring, or %NULL to ignore
666 * errors. Any of the errors in #GConvertError may occur.
668 * Converts a string from one character set to another, possibly
669 * including fallback sequences for characters not representable
670 * in the output. Note that it is not guaranteed that the specification
671 * for the fallback sequences in @fallback will be honored. Some
672 * systems may do a approximate conversion from @from_codeset
673 * to @to_codeset in their iconv() functions,
674 * in which case GLib will simply return that approximate conversion.
676 * Return value: If the conversion was successful, a newly allocated
677 * nul-terminated string, which must be freed with
678 * g_free(). Otherwise %NULL and @error will be set.
681 g_convert_with_fallback (const gchar *str,
683 const gchar *to_codeset,
684 const gchar *from_codeset,
687 gsize *bytes_written,
693 const gchar *insert_str = NULL;
695 gsize inbytes_remaining;
696 const gchar *save_p = NULL;
697 gsize save_inbytes = 0;
698 gsize outbytes_remaining;
702 gboolean have_error = FALSE;
703 gboolean done = FALSE;
705 GError *local_error = NULL;
707 g_return_val_if_fail (str != NULL, NULL);
708 g_return_val_if_fail (to_codeset != NULL, NULL);
709 g_return_val_if_fail (from_codeset != NULL, NULL);
714 /* Try an exact conversion; we only proceed if this fails
715 * due to an illegal sequence in the input string.
717 dest = g_convert (str, len, to_codeset, from_codeset,
718 bytes_read, bytes_written, &local_error);
722 if (!g_error_matches (local_error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE))
724 g_propagate_error (error, local_error);
728 g_error_free (local_error);
732 /* No go; to proceed, we need a converter from "UTF-8" to
733 * to_codeset, and the string as UTF-8.
735 cd = open_converter (to_codeset, "UTF-8", error);
736 if (cd == (GIConv) -1)
747 utf8 = g_convert (str, len, "UTF-8", from_codeset,
748 bytes_read, &inbytes_remaining, error);
751 close_converter (cd);
757 /* Now the heart of the code. We loop through the UTF-8 string, and
758 * whenever we hit an offending character, we form fallback, convert
759 * the fallback to the target codeset, and then go back to
760 * converting the original string after finishing with the fallback.
762 * The variables save_p and save_inbytes store the input state
763 * for the original string while we are converting the fallback
767 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
768 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
769 outp = dest = g_malloc (outbuf_size);
771 while (!done && !have_error)
773 size_t inbytes_tmp = inbytes_remaining;
774 err = g_iconv (cd, (char **)&p, &inbytes_tmp, &outp, &outbytes_remaining);
775 inbytes_remaining = inbytes_tmp;
777 if (err == (size_t) -1)
782 g_assert_not_reached();
786 size_t used = outp - dest;
789 dest = g_realloc (dest, outbuf_size);
792 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
799 /* Error converting fallback string - fatal
801 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
802 _("Cannot convert fallback '%s' to codeset '%s'"),
803 insert_str, to_codeset);
811 gunichar ch = g_utf8_get_char (p);
812 insert_str = g_strdup_printf (ch < 0x10000 ? "\\u%04x" : "\\U%08x",
816 insert_str = fallback;
818 save_p = g_utf8_next_char (p);
819 save_inbytes = inbytes_remaining - (save_p - p);
821 inbytes_remaining = strlen (p);
825 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
826 _("Error during conversion: %s"),
837 g_free ((gchar *)insert_str);
839 inbytes_remaining = save_inbytes;
851 close_converter (cd);
854 *bytes_written = outp - dest; /* Doesn't include '\0' */
860 if (save_p && !fallback)
861 g_free ((gchar *)insert_str);
876 strdup_len (const gchar *string,
878 gsize *bytes_written,
885 if (!g_utf8_validate (string, len, NULL))
892 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
893 _("Invalid byte sequence in conversion input"));
898 real_len = strlen (string);
903 while (real_len < len && string[real_len])
908 *bytes_read = real_len;
910 *bytes_written = real_len;
912 return g_strndup (string, real_len);
917 * @opsysstring: a string in the encoding of the current locale. On Windows
918 * this means the system codepage.
919 * @len: the length of the string, or -1 if the string is
921 * @bytes_read: location to store the number of bytes in the
922 * input string that were successfully converted, or %NULL.
923 * Even if the conversion was successful, this may be
924 * less than @len if there were partial characters
925 * at the end of the input. If the error
926 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
927 * stored will the byte offset after the last valid
929 * @bytes_written: the number of bytes stored in the output buffer (not
930 * including the terminating nul).
931 * @error: location to store the error occuring, or %NULL to ignore
932 * errors. Any of the errors in #GConvertError may occur.
934 * Converts a string which is in the encoding used for strings by
935 * the C runtime (usually the same as that used by the operating
936 * system) in the current locale into a UTF-8 string.
938 * Return value: The converted string, or %NULL on an error.
941 g_locale_to_utf8 (const gchar *opsysstring,
944 gsize *bytes_written,
949 if (g_get_charset (&charset))
950 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
952 return g_convert (opsysstring, len,
953 "UTF-8", charset, bytes_read, bytes_written, error);
957 * g_locale_from_utf8:
958 * @utf8string: a UTF-8 encoded string
959 * @len: the length of the string, or -1 if the string is
961 * @bytes_read: location to store the number of bytes in the
962 * input string that were successfully converted, or %NULL.
963 * Even if the conversion was successful, this may be
964 * less than @len if there were partial characters
965 * at the end of the input. If the error
966 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
967 * stored will the byte offset after the last valid
969 * @bytes_written: the number of bytes stored in the output buffer (not
970 * including the terminating nul).
971 * @error: location to store the error occuring, or %NULL to ignore
972 * errors. Any of the errors in #GConvertError may occur.
974 * Converts a string from UTF-8 to the encoding used for strings by
975 * the C runtime (usually the same as that used by the operating
976 * system) in the current locale.
978 * Return value: The converted string, or %NULL on an error.
981 g_locale_from_utf8 (const gchar *utf8string,
984 gsize *bytes_written,
987 const gchar *charset;
989 if (g_get_charset (&charset))
990 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
992 return g_convert (utf8string, len,
993 charset, "UTF-8", bytes_read, bytes_written, error);
996 #ifndef G_PLATFORM_WIN32
998 typedef struct _GFilenameCharsetCache GFilenameCharsetCache;
1000 struct _GFilenameCharsetCache {
1003 gchar **filename_charsets;
1007 filename_charset_cache_free (gpointer data)
1009 GFilenameCharsetCache *cache = data;
1010 g_free (cache->charset);
1011 g_strfreev (cache->filename_charsets);
1016 * g_get_filename_charsets:
1017 * @charsets: return location for the %NULL-terminated list of encoding names
1019 * Determines the preferred character sets used for filenames.
1020 * The first character set from the @charsets is the filename encoding, the
1021 * subsequent character sets are used when trying to generate a displayable
1022 * representation of a filename, see g_filename_display_name().
1024 * On Unix, the character sets are determined by consulting the
1025 * environment variables <envar>G_FILENAME_ENCODING</envar> and
1026 * <envar>G_BROKEN_FILENAMES</envar>. On Windows, the character set
1027 * used in the GLib API is always UTF-8 and said environment variables
1030 * <envar>G_FILENAME_ENCODING</envar> may be set to a comma-separated list
1031 * of character set names. The special token "@locale" is taken to mean the
1032 * character set for the current locale. If <envar>G_FILENAME_ENCODING</envar>
1033 * is not set, but <envar>G_BROKEN_FILENAMES</envar> is, the character set of
1034 * the current locale is taken as the filename encoding. If neither environment
1035 * variable is set, UTF-8 is taken as the filename encoding, but the character
1036 * set of the current locale is also put in the list of encodings.
1038 * The returned @charsets belong to GLib and must not be freed.
1040 * Note that on Unix, regardless of the locale character set or
1041 * <envar>G_FILENAME_ENCODING</envar> value, the actual file names present on a
1042 * system might be in any random encoding or just gibberish.
1044 * Return value: %TRUE if the filename encoding is UTF-8.
1049 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1051 static GStaticPrivate cache_private = G_STATIC_PRIVATE_INIT;
1052 GFilenameCharsetCache *cache = g_static_private_get (&cache_private);
1053 const gchar *charset;
1057 cache = g_new0 (GFilenameCharsetCache, 1);
1058 g_static_private_set (&cache_private, cache, filename_charset_cache_free);
1061 g_get_charset (&charset);
1063 if (!(cache->charset && strcmp (cache->charset, charset) == 0))
1065 const gchar *new_charset;
1069 g_free (cache->charset);
1070 g_strfreev (cache->filename_charsets);
1071 cache->charset = g_strdup (charset);
1073 p = getenv ("G_FILENAME_ENCODING");
1074 if (p != NULL && p[0] != '\0')
1076 cache->filename_charsets = g_strsplit (p, ",", 0);
1077 cache->is_utf8 = (strcmp (cache->filename_charsets[0], "UTF-8") == 0);
1079 for (i = 0; cache->filename_charsets[i]; i++)
1081 if (strcmp ("@locale", cache->filename_charsets[i]) == 0)
1083 g_get_charset (&new_charset);
1084 g_free (cache->filename_charsets[i]);
1085 cache->filename_charsets[i] = g_strdup (new_charset);
1089 else if (getenv ("G_BROKEN_FILENAMES") != NULL)
1091 cache->filename_charsets = g_new0 (gchar *, 2);
1092 cache->is_utf8 = g_get_charset (&new_charset);
1093 cache->filename_charsets[0] = g_strdup (new_charset);
1097 cache->filename_charsets = g_new0 (gchar *, 3);
1098 cache->is_utf8 = TRUE;
1099 cache->filename_charsets[0] = g_strdup ("UTF-8");
1100 if (!g_get_charset (&new_charset))
1101 cache->filename_charsets[1] = g_strdup (new_charset);
1105 if (filename_charsets)
1106 *filename_charsets = (const gchar **)cache->filename_charsets;
1108 return cache->is_utf8;
1111 #else /* G_PLATFORM_WIN32 */
1114 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1116 static gchar *charsets[] = {
1122 /* On Windows GLib pretends that the filename charset is UTF-8 */
1123 if (filename_charsets)
1124 *filename_charsets = charsets;
1130 /* Cygwin works like before */
1131 result = g_get_charset (&(charsets[0]));
1133 if (filename_charsets)
1134 *filename_charsets = charsets;
1140 #endif /* G_PLATFORM_WIN32 */
1143 get_filename_charset (const gchar **filename_charset)
1145 const gchar **charsets;
1148 is_utf8 = g_get_filename_charsets (&charsets);
1150 if (filename_charset)
1151 *filename_charset = charsets[0];
1156 /* This is called from g_thread_init(). It's used to
1157 * initialize some static data in a threadsafe way.
1160 _g_convert_thread_init (void)
1162 const gchar **dummy;
1163 (void) g_get_filename_charsets (&dummy);
1167 * g_filename_to_utf8:
1168 * @opsysstring: a string in the encoding for filenames
1169 * @len: the length of the string, or -1 if the string is
1171 * @bytes_read: location to store the number of bytes in the
1172 * input string that were successfully converted, or %NULL.
1173 * Even if the conversion was successful, this may be
1174 * less than @len if there were partial characters
1175 * at the end of the input. If the error
1176 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1177 * stored will the byte offset after the last valid
1179 * @bytes_written: the number of bytes stored in the output buffer (not
1180 * including the terminating nul).
1181 * @error: location to store the error occuring, or %NULL to ignore
1182 * errors. Any of the errors in #GConvertError may occur.
1184 * Converts a string which is in the encoding used by GLib for filenames
1185 * into a UTF-8 string.
1187 * Return value: The converted string, or %NULL on an error.
1190 g_filename_to_utf8 (const gchar *opsysstring,
1193 gsize *bytes_written,
1196 const gchar *charset;
1198 if (get_filename_charset (&charset))
1199 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1201 return g_convert (opsysstring, len,
1202 "UTF-8", charset, bytes_read, bytes_written, error);
1207 #undef g_filename_to_utf8
1209 /* Binary compatibility version. Not for newly compiled code. */
1212 g_filename_to_utf8 (const gchar *opsysstring,
1215 gsize *bytes_written,
1218 const gchar *charset;
1220 if (g_get_charset (&charset))
1221 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1223 return g_convert (opsysstring, len,
1224 "UTF-8", charset, bytes_read, bytes_written, error);
1230 * g_filename_from_utf8:
1231 * @utf8string: a UTF-8 encoded string.
1232 * @len: the length of the string, or -1 if the string is
1234 * @bytes_read: location to store the number of bytes in the
1235 * input string that were successfully converted, or %NULL.
1236 * Even if the conversion was successful, this may be
1237 * less than @len if there were partial characters
1238 * at the end of the input. If the error
1239 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1240 * stored will the byte offset after the last valid
1242 * @bytes_written: the number of bytes stored in the output buffer (not
1243 * including the terminating nul).
1244 * @error: location to store the error occuring, or %NULL to ignore
1245 * errors. Any of the errors in #GConvertError may occur.
1247 * Converts a string from UTF-8 to the encoding used for filenames.
1249 * Return value: The converted string, or %NULL on an error.
1252 g_filename_from_utf8 (const gchar *utf8string,
1255 gsize *bytes_written,
1258 const gchar *charset;
1260 if (get_filename_charset (&charset))
1261 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1263 return g_convert (utf8string, len,
1264 charset, "UTF-8", bytes_read, bytes_written, error);
1269 #undef g_filename_from_utf8
1271 /* Binary compatibility version. Not for newly compiled code. */
1274 g_filename_from_utf8 (const gchar *utf8string,
1277 gsize *bytes_written,
1280 const gchar *charset;
1282 if (g_get_charset (&charset))
1283 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1285 return g_convert (utf8string, len,
1286 charset, "UTF-8", bytes_read, bytes_written, error);
1291 /* Test of haystack has the needle prefix, comparing case
1292 * insensitive. haystack may be UTF-8, but needle must
1293 * contain only ascii. */
1295 has_case_prefix (const gchar *haystack, const gchar *needle)
1299 /* Eat one character at a time. */
1304 g_ascii_tolower (*n) == g_ascii_tolower (*h))
1314 UNSAFE_ALL = 0x1, /* Escape all unsafe characters */
1315 UNSAFE_ALLOW_PLUS = 0x2, /* Allows '+' */
1316 UNSAFE_PATH = 0x8, /* Allows '/', '&', '=', ':', '@', '+', '$' and ',' */
1317 UNSAFE_HOST = 0x10, /* Allows '/' and ':' and '@' */
1318 UNSAFE_SLASHES = 0x20 /* Allows all characters except for '/' and '%' */
1319 } UnsafeCharacterSet;
1321 static const guchar acceptable[96] = {
1322 /* A table of the ASCII chars from space (32) to DEL (127) */
1323 /* ! " # $ % & ' ( ) * + , - . / */
1324 0x00,0x3F,0x20,0x20,0x28,0x00,0x2C,0x3F,0x3F,0x3F,0x3F,0x2A,0x28,0x3F,0x3F,0x1C,
1325 /* 0 1 2 3 4 5 6 7 8 9 : ; < = > ? */
1326 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x38,0x20,0x20,0x2C,0x20,0x20,
1327 /* @ A B C D E F G H I J K L M N O */
1328 0x38,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1329 /* P Q R S T U V W X Y Z [ \ ] ^ _ */
1330 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x20,0x3F,
1331 /* ` a b c d e f g h i j k l m n o */
1332 0x20,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1333 /* p q r s t u v w x y z { | } ~ DEL */
1334 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x3F,0x20
1337 static const gchar hex[16] = "0123456789ABCDEF";
1339 /* Note: This escape function works on file: URIs, but if you want to
1340 * escape something else, please read RFC-2396 */
1342 g_escape_uri_string (const gchar *string,
1343 UnsafeCharacterSet mask)
1345 #define ACCEPTABLE(a) ((a)>=32 && (a)<128 && (acceptable[(a)-32] & use_mask))
1352 UnsafeCharacterSet use_mask;
1354 g_return_val_if_fail (mask == UNSAFE_ALL
1355 || mask == UNSAFE_ALLOW_PLUS
1356 || mask == UNSAFE_PATH
1357 || mask == UNSAFE_HOST
1358 || mask == UNSAFE_SLASHES, NULL);
1362 for (p = string; *p != '\0'; p++)
1365 if (!ACCEPTABLE (c))
1369 result = g_malloc (p - string + unacceptable * 2 + 1);
1372 for (q = result, p = string; *p != '\0'; p++)
1376 if (!ACCEPTABLE (c))
1378 *q++ = '%'; /* means hex coming */
1393 g_escape_file_uri (const gchar *hostname,
1394 const gchar *pathname)
1396 char *escaped_hostname = NULL;
1401 char *p, *backslash;
1403 /* Turn backslashes into forward slashes. That's what Netscape
1404 * does, and they are actually more or less equivalent in Windows.
1407 pathname = g_strdup (pathname);
1408 p = (char *) pathname;
1410 while ((backslash = strchr (p, '\\')) != NULL)
1417 if (hostname && *hostname != '\0')
1419 escaped_hostname = g_escape_uri_string (hostname, UNSAFE_HOST);
1422 escaped_path = g_escape_uri_string (pathname, UNSAFE_PATH);
1424 res = g_strconcat ("file://",
1425 (escaped_hostname) ? escaped_hostname : "",
1426 (*escaped_path != '/') ? "/" : "",
1431 g_free ((char *) pathname);
1434 g_free (escaped_hostname);
1435 g_free (escaped_path);
1441 unescape_character (const char *scanner)
1446 first_digit = g_ascii_xdigit_value (scanner[0]);
1447 if (first_digit < 0)
1450 second_digit = g_ascii_xdigit_value (scanner[1]);
1451 if (second_digit < 0)
1454 return (first_digit << 4) | second_digit;
1458 g_unescape_uri_string (const char *escaped,
1460 const char *illegal_escaped_characters,
1461 gboolean ascii_must_not_be_escaped)
1463 const gchar *in, *in_end;
1464 gchar *out, *result;
1467 if (escaped == NULL)
1471 len = strlen (escaped);
1473 result = g_malloc (len + 1);
1476 for (in = escaped, in_end = escaped + len; in < in_end; in++)
1482 /* catch partial escape sequences past the end of the substring */
1483 if (in + 3 > in_end)
1486 c = unescape_character (in + 1);
1488 /* catch bad escape sequences and NUL characters */
1492 /* catch escaped ASCII */
1493 if (ascii_must_not_be_escaped && c <= 0x7F)
1496 /* catch other illegal escaped characters */
1497 if (strchr (illegal_escaped_characters, c) != NULL)
1506 g_assert (out - result <= len);
1519 is_asciialphanum (gunichar c)
1521 return c <= 0x7F && g_ascii_isalnum (c);
1525 is_asciialpha (gunichar c)
1527 return c <= 0x7F && g_ascii_isalpha (c);
1530 /* allows an empty string */
1532 hostname_validate (const char *hostname)
1535 gunichar c, first_char, last_char;
1542 /* read in a label */
1543 c = g_utf8_get_char (p);
1544 p = g_utf8_next_char (p);
1545 if (!is_asciialphanum (c))
1551 c = g_utf8_get_char (p);
1552 p = g_utf8_next_char (p);
1554 while (is_asciialphanum (c) || c == '-');
1555 if (last_char == '-')
1558 /* if that was the last label, check that it was a toplabel */
1559 if (c == '\0' || (c == '.' && *p == '\0'))
1560 return is_asciialpha (first_char);
1567 * g_filename_from_uri:
1568 * @uri: a uri describing a filename (escaped, encoded in ASCII).
1569 * @hostname: Location to store hostname for the URI, or %NULL.
1570 * If there is no hostname in the URI, %NULL will be
1571 * stored in this location.
1572 * @error: location to store the error occuring, or %NULL to ignore
1573 * errors. Any of the errors in #GConvertError may occur.
1575 * Converts an escaped ASCII-encoded URI to a local filename in the
1576 * encoding used for filenames.
1578 * Return value: a newly-allocated string holding the resulting
1579 * filename, or %NULL on an error.
1582 g_filename_from_uri (const gchar *uri,
1586 const char *path_part;
1587 const char *host_part;
1588 char *unescaped_hostname;
1599 if (!has_case_prefix (uri, "file:/"))
1601 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1602 _("The URI '%s' is not an absolute URI using the \"file\" scheme"),
1607 path_part = uri + strlen ("file:");
1609 if (strchr (path_part, '#') != NULL)
1611 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1612 _("The local file URI '%s' may not include a '#'"),
1617 if (has_case_prefix (path_part, "///"))
1619 else if (has_case_prefix (path_part, "//"))
1622 host_part = path_part;
1624 path_part = strchr (path_part, '/');
1626 if (path_part == NULL)
1628 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1629 _("The URI '%s' is invalid"),
1634 unescaped_hostname = g_unescape_uri_string (host_part, path_part - host_part, "", TRUE);
1636 if (unescaped_hostname == NULL ||
1637 !hostname_validate (unescaped_hostname))
1639 g_free (unescaped_hostname);
1640 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1641 _("The hostname of the URI '%s' is invalid"),
1647 *hostname = unescaped_hostname;
1649 g_free (unescaped_hostname);
1652 filename = g_unescape_uri_string (path_part, -1, "/", FALSE);
1654 if (filename == NULL)
1656 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1657 _("The URI '%s' contains invalidly escaped characters"),
1664 /* Drop localhost */
1665 if (hostname && *hostname != NULL &&
1666 g_ascii_strcasecmp (*hostname, "localhost") == 0)
1672 /* Turn slashes into backslashes, because that's the canonical spelling */
1674 while ((slash = strchr (p, '/')) != NULL)
1680 /* Windows URIs with a drive letter can be like "file://host/c:/foo"
1681 * or "file://host/c|/foo" (some Netscape versions). In those cases, start
1682 * the filename from the drive letter.
1684 if (g_ascii_isalpha (filename[1]))
1686 if (filename[2] == ':')
1688 else if (filename[2] == '|')
1696 result = g_strdup (filename + offs);
1704 #undef g_filename_from_uri
1707 g_filename_from_uri (const gchar *uri,
1711 gchar *utf8_filename;
1712 gchar *retval = NULL;
1714 utf8_filename = g_filename_from_uri_utf8 (uri, hostname, error);
1717 retval = g_locale_from_utf8 (utf8_filename, -1, NULL, NULL, error);
1718 g_free (utf8_filename);
1726 * g_filename_to_uri:
1727 * @filename: an absolute filename specified in the GLib file name encoding,
1728 * which is the on-disk file name bytes on Unix, and UTF-8 on
1730 * @hostname: A UTF-8 encoded hostname, or %NULL for none.
1731 * @error: location to store the error occuring, or %NULL to ignore
1732 * errors. Any of the errors in #GConvertError may occur.
1734 * Converts an absolute filename to an escaped ASCII-encoded URI.
1736 * Return value: a newly-allocated string holding the resulting
1737 * URI, or %NULL on an error.
1740 g_filename_to_uri (const gchar *filename,
1741 const gchar *hostname,
1746 g_return_val_if_fail (filename != NULL, NULL);
1748 if (!g_path_is_absolute (filename))
1750 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NOT_ABSOLUTE_PATH,
1751 _("The pathname '%s' is not an absolute path"),
1757 !(g_utf8_validate (hostname, -1, NULL)
1758 && hostname_validate (hostname)))
1760 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
1761 _("Invalid hostname"));
1766 /* Don't use localhost unnecessarily */
1767 if (hostname && g_ascii_strcasecmp (hostname, "localhost") == 0)
1771 escaped_uri = g_escape_file_uri (hostname, filename);
1778 #undef g_filename_to_uri
1781 g_filename_to_uri (const gchar *filename,
1782 const gchar *hostname,
1785 gchar *utf8_filename;
1786 gchar *retval = NULL;
1788 utf8_filename = g_locale_to_utf8 (filename, -1, NULL, NULL, error);
1792 retval = g_filename_to_uri_utf8 (utf8_filename, hostname, error);
1793 g_free (utf8_filename);
1802 * g_uri_list_extract_uris:
1803 * @uri_list: an URI list
1805 * Splits an URI list conforming to the text/uri-list
1806 * mime type defined in RFC 2483 into individual URIs,
1807 * discarding any comments. The URIs are not validated.
1809 * Returns: a newly allocated %NULL-terminated list of
1810 * strings holding the individual URIs. The array should
1811 * be freed with g_strfreev().
1816 g_uri_list_extract_uris (const gchar *uri_list)
1827 /* We don't actually try to validate the URI according to RFC
1828 * 2396, or even check for allowed characters - we just ignore
1829 * comments and trim whitespace off the ends. We also
1830 * allow LF delimination as well as the specified CRLF.
1832 * We do allow comments like specified in RFC 2483.
1838 while (g_ascii_isspace (*p))
1842 while (*q && (*q != '\n') && (*q != '\r'))
1848 while (q > p && g_ascii_isspace (*q))
1853 uris = g_slist_prepend (uris, g_strndup (p, q - p + 1));
1858 p = strchr (p, '\n');
1863 result = g_new (gchar *, n_uris + 1);
1865 result[n_uris--] = NULL;
1866 for (u = uris; u; u = u->next)
1867 result[n_uris--] = u->data;
1869 g_slist_free (uris);
1875 make_valid_utf8 (const gchar *name)
1878 const gchar *remainder, *invalid;
1879 gint remaining_bytes, valid_bytes;
1883 remaining_bytes = strlen (name);
1885 while (remaining_bytes != 0)
1887 if (g_utf8_validate (remainder, remaining_bytes, &invalid))
1889 valid_bytes = invalid - remainder;
1892 string = g_string_sized_new (remaining_bytes);
1894 g_string_append_len (string, remainder, valid_bytes);
1895 g_string_append_c (string, '?');
1897 remaining_bytes -= valid_bytes + 1;
1898 remainder = invalid + 1;
1902 return g_strdup (name);
1904 g_string_append (string, remainder);
1905 g_string_append (string, " (invalid encoding)");
1907 g_assert (g_utf8_validate (string->str, -1, NULL));
1909 return g_string_free (string, FALSE);
1913 * g_filename_display_basename:
1914 * @filename: an absolute pathname in the GLib file name encoding
1916 * Returns the display basename for the particular filename, guaranteed
1917 * to be valid UTF-8. The display name might not be identical to the filename,
1918 * for instance there might be problems converting it to UTF-8, and some files
1919 * can be translated in the display
1921 * You must pass the whole absolute pathname to this functions so that
1922 * translation of well known locations can be done.
1924 * This function is preferred over g_filename_display_name() if you know the
1925 * whole path, as it allows translation.
1927 * Return value: a newly allocated string containing
1928 * a rendition of the basename of the filename in valid UTF-8
1933 g_filename_display_basename (const gchar *filename)
1938 g_return_val_if_fail (filename != NULL, NULL);
1940 basename = g_path_get_basename (filename);
1941 display_name = g_filename_display_name (basename);
1943 return display_name;
1947 * g_filename_display_name:
1948 * @filename: a pathname hopefully in the GLib file name encoding
1950 * Converts a filename into a valid UTF-8 string. The
1951 * conversion is not necessarily reversible, so you
1952 * should keep the original around and use the return
1953 * value of this function only for display purposes.
1954 * Unlike g_filename_to_utf8(), the result is guaranteed
1955 * to be non-NULL even if the filename actually isn't in the GLib
1956 * file name encoding.
1958 * If you know the whole pathname of the file you should use
1959 * g_filename_display_basename(), since that allows location-based
1960 * translation of filenames.
1962 * Return value: a newly allocated string containing
1963 * a rendition of the filename in valid UTF-8
1968 g_filename_display_name (const gchar *filename)
1971 const gchar **charsets;
1972 gchar *display_name = NULL;
1975 is_utf8 = g_get_filename_charsets (&charsets);
1979 if (g_utf8_validate (filename, -1, NULL))
1980 display_name = g_strdup (filename);
1985 /* Try to convert from the filename charsets to UTF-8.
1986 * Skip the first charset if it is UTF-8.
1988 for (i = is_utf8 ? 1 : 0; charsets[i]; i++)
1990 display_name = g_convert (filename, -1, "UTF-8", charsets[i],
1998 /* if all conversions failed, we replace invalid UTF-8
1999 * by a question mark
2002 display_name = make_valid_utf8 (filename);
2004 return display_name;
2007 #define __G_CONVERT_C__
2008 #include "galiasdef.c"