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);
469 * @str: the string to convert
470 * @len: the length of the string, or -1 if the string is
471 * nul-terminated<footnote id="nul-unsafe">
473 Note that some encodings may allow nul bytes to
474 occur inside strings. In that case, using -1 for
475 the @len parameter is unsafe.
478 * @to_codeset: name of character set into which to convert @str
479 * @from_codeset: character set of @str.
480 * @bytes_read: location to store the number of bytes in the
481 * input string that were successfully converted, or %NULL.
482 * Even if the conversion was successful, this may be
483 * less than @len if there were partial characters
484 * at the end of the input. If the error
485 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
486 * stored will the byte offset after the last valid
488 * @bytes_written: the number of bytes stored in the output buffer (not
489 * including the terminating nul).
490 * @error: location to store the error occuring, or %NULL to ignore
491 * errors. Any of the errors in #GConvertError may occur.
493 * Converts a string from one character set to another.
495 * Return value: If the conversion was successful, a newly allocated
496 * nul-terminated string, which must be freed with
497 * g_free(). Otherwise %NULL and @error will be set.
500 g_convert (const gchar *str,
502 const gchar *to_codeset,
503 const gchar *from_codeset,
505 gsize *bytes_written,
511 g_return_val_if_fail (str != NULL, NULL);
512 g_return_val_if_fail (to_codeset != NULL, NULL);
513 g_return_val_if_fail (from_codeset != NULL, NULL);
515 cd = open_converter (to_codeset, from_codeset, error);
517 if (cd == (GIConv) -1)
528 res = g_convert_with_iconv (str, len, cd,
529 bytes_read, bytes_written,
532 close_converter (cd);
538 * g_convert_with_iconv:
539 * @str: the string to convert
540 * @len: the length of the string, or -1 if the string is
541 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
542 * @converter: conversion descriptor from g_iconv_open()
543 * @bytes_read: location to store the number of bytes in the
544 * input string that were successfully converted, or %NULL.
545 * Even if the conversion was successful, this may be
546 * less than @len if there were partial characters
547 * at the end of the input. If the error
548 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
549 * stored will the byte offset after the last valid
551 * @bytes_written: the number of bytes stored in the output buffer (not
552 * including the terminating nul).
553 * @error: location to store the error occuring, or %NULL to ignore
554 * errors. Any of the errors in #GConvertError may occur.
556 * Converts a string from one character set to another.
558 * Return value: If the conversion was successful, a newly allocated
559 * nul-terminated string, which must be freed with
560 * g_free(). Otherwise %NULL and @error will be set.
563 g_convert_with_iconv (const gchar *str,
567 gsize *bytes_written,
573 gsize inbytes_remaining;
574 gsize outbytes_remaining;
577 gboolean have_error = FALSE;
579 g_return_val_if_fail (str != NULL, NULL);
580 g_return_val_if_fail (converter != (GIConv) -1, NULL);
586 inbytes_remaining = len;
587 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
589 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
590 outp = dest = g_malloc (outbuf_size);
594 err = g_iconv (converter, (char **)&p, &inbytes_remaining, &outp, &outbytes_remaining);
596 if (err == (size_t) -1)
601 /* Incomplete text, do not report an error */
605 size_t used = outp - dest;
608 dest = g_realloc (dest, outbuf_size);
611 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
617 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
618 _("Invalid byte sequence in conversion input"));
623 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
624 _("Error during conversion: %s"),
634 *bytes_read = p - str;
637 if ((p - str) != len)
642 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_PARTIAL_INPUT,
643 _("Partial character sequence at end of input"));
650 *bytes_written = outp - dest; /* Doesn't include '\0' */
662 * g_convert_with_fallback:
663 * @str: the string to convert
664 * @len: the length of the string, or -1 if the string is
665 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
666 * @to_codeset: name of character set into which to convert @str
667 * @from_codeset: character set of @str.
668 * @fallback: UTF-8 string to use in place of character not
669 * present in the target encoding. (The string must be
670 * representable in the target encoding).
671 If %NULL, characters not in the target encoding will
672 be represented as Unicode escapes \uxxxx or \Uxxxxyyyy.
673 * @bytes_read: location to store the number of bytes in the
674 * input string that were successfully converted, or %NULL.
675 * Even if the conversion was successful, this may be
676 * less than @len if there were partial characters
677 * at the end of the input.
678 * @bytes_written: the number of bytes stored in the output buffer (not
679 * including the terminating nul).
680 * @error: location to store the error occuring, or %NULL to ignore
681 * errors. Any of the errors in #GConvertError may occur.
683 * Converts a string from one character set to another, possibly
684 * including fallback sequences for characters not representable
685 * in the output. Note that it is not guaranteed that the specification
686 * for the fallback sequences in @fallback will be honored. Some
687 * systems may do a approximate conversion from @from_codeset
688 * to @to_codeset in their iconv() functions,
689 * in which case GLib will simply return that approximate conversion.
691 * Return value: If the conversion was successful, a newly allocated
692 * nul-terminated string, which must be freed with
693 * g_free(). Otherwise %NULL and @error will be set.
696 g_convert_with_fallback (const gchar *str,
698 const gchar *to_codeset,
699 const gchar *from_codeset,
702 gsize *bytes_written,
708 const gchar *insert_str = NULL;
710 gsize inbytes_remaining;
711 const gchar *save_p = NULL;
712 gsize save_inbytes = 0;
713 gsize outbytes_remaining;
717 gboolean have_error = FALSE;
718 gboolean done = FALSE;
720 GError *local_error = NULL;
722 g_return_val_if_fail (str != NULL, NULL);
723 g_return_val_if_fail (to_codeset != NULL, NULL);
724 g_return_val_if_fail (from_codeset != NULL, NULL);
729 /* Try an exact conversion; we only proceed if this fails
730 * due to an illegal sequence in the input string.
732 dest = g_convert (str, len, to_codeset, from_codeset,
733 bytes_read, bytes_written, &local_error);
737 if (!g_error_matches (local_error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE))
739 g_propagate_error (error, local_error);
743 g_error_free (local_error);
747 /* No go; to proceed, we need a converter from "UTF-8" to
748 * to_codeset, and the string as UTF-8.
750 cd = open_converter (to_codeset, "UTF-8", error);
751 if (cd == (GIConv) -1)
762 utf8 = g_convert (str, len, "UTF-8", from_codeset,
763 bytes_read, &inbytes_remaining, error);
766 close_converter (cd);
772 /* Now the heart of the code. We loop through the UTF-8 string, and
773 * whenever we hit an offending character, we form fallback, convert
774 * the fallback to the target codeset, and then go back to
775 * converting the original string after finishing with the fallback.
777 * The variables save_p and save_inbytes store the input state
778 * for the original string while we are converting the fallback
782 outbuf_size = len + 1; /* + 1 for nul in case len == 1 */
783 outbytes_remaining = outbuf_size - 1; /* -1 for nul */
784 outp = dest = g_malloc (outbuf_size);
786 while (!done && !have_error)
788 size_t inbytes_tmp = inbytes_remaining;
789 err = g_iconv (cd, (char **)&p, &inbytes_tmp, &outp, &outbytes_remaining);
790 inbytes_remaining = inbytes_tmp;
792 if (err == (size_t) -1)
797 g_assert_not_reached();
801 size_t used = outp - dest;
804 dest = g_realloc (dest, outbuf_size);
807 outbytes_remaining = outbuf_size - used - 1; /* -1 for nul */
814 /* Error converting fallback string - fatal
816 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
817 _("Cannot convert fallback '%s' to codeset '%s'"),
818 insert_str, to_codeset);
826 gunichar ch = g_utf8_get_char (p);
827 insert_str = g_strdup_printf (ch < 0x10000 ? "\\u%04x" : "\\U%08x",
831 insert_str = fallback;
833 save_p = g_utf8_next_char (p);
834 save_inbytes = inbytes_remaining - (save_p - p);
836 inbytes_remaining = strlen (p);
840 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
841 _("Error during conversion: %s"),
852 g_free ((gchar *)insert_str);
854 inbytes_remaining = save_inbytes;
866 close_converter (cd);
869 *bytes_written = outp - dest; /* Doesn't include '\0' */
875 if (save_p && !fallback)
876 g_free ((gchar *)insert_str);
891 strdup_len (const gchar *string,
893 gsize *bytes_written,
900 if (!g_utf8_validate (string, len, NULL))
907 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
908 _("Invalid byte sequence in conversion input"));
913 real_len = strlen (string);
918 while (real_len < len && string[real_len])
923 *bytes_read = real_len;
925 *bytes_written = real_len;
927 return g_strndup (string, real_len);
932 * @opsysstring: a string in the encoding of the current locale. On Windows
933 * this means the system codepage.
934 * @len: the length of the string, or -1 if the string is
935 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
936 * @bytes_read: location to store the number of bytes in the
937 * input string that were successfully converted, or %NULL.
938 * Even if the conversion was successful, this may be
939 * less than @len if there were partial characters
940 * at the end of the input. If the error
941 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
942 * stored will the byte offset after the last valid
944 * @bytes_written: the number of bytes stored in the output buffer (not
945 * including the terminating nul).
946 * @error: location to store the error occuring, or %NULL to ignore
947 * errors. Any of the errors in #GConvertError may occur.
949 * Converts a string which is in the encoding used for strings by
950 * the C runtime (usually the same as that used by the operating
951 * system) in the current locale into a UTF-8 string.
953 * Return value: The converted string, or %NULL on an error.
956 g_locale_to_utf8 (const gchar *opsysstring,
959 gsize *bytes_written,
964 if (g_get_charset (&charset))
965 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
967 return g_convert (opsysstring, len,
968 "UTF-8", charset, bytes_read, bytes_written, error);
972 * g_locale_from_utf8:
973 * @utf8string: a UTF-8 encoded string
974 * @len: the length of the string, or -1 if the string is
975 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
976 * @bytes_read: location to store the number of bytes in the
977 * input string that were successfully converted, or %NULL.
978 * Even if the conversion was successful, this may be
979 * less than @len if there were partial characters
980 * at the end of the input. If the error
981 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
982 * stored will the byte offset after the last valid
984 * @bytes_written: the number of bytes stored in the output buffer (not
985 * including the terminating nul).
986 * @error: location to store the error occuring, or %NULL to ignore
987 * errors. Any of the errors in #GConvertError may occur.
989 * Converts a string from UTF-8 to the encoding used for strings by
990 * the C runtime (usually the same as that used by the operating
991 * system) in the current locale.
993 * Return value: The converted string, or %NULL on an error.
996 g_locale_from_utf8 (const gchar *utf8string,
999 gsize *bytes_written,
1002 const gchar *charset;
1004 if (g_get_charset (&charset))
1005 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1007 return g_convert (utf8string, len,
1008 charset, "UTF-8", bytes_read, bytes_written, error);
1011 #ifndef G_PLATFORM_WIN32
1013 typedef struct _GFilenameCharsetCache GFilenameCharsetCache;
1015 struct _GFilenameCharsetCache {
1018 gchar **filename_charsets;
1022 filename_charset_cache_free (gpointer data)
1024 GFilenameCharsetCache *cache = data;
1025 g_free (cache->charset);
1026 g_strfreev (cache->filename_charsets);
1031 * g_get_filename_charsets:
1032 * @charsets: return location for the %NULL-terminated list of encoding names
1034 * Determines the preferred character sets used for filenames.
1035 * The first character set from the @charsets is the filename encoding, the
1036 * subsequent character sets are used when trying to generate a displayable
1037 * representation of a filename, see g_filename_display_name().
1039 * On Unix, the character sets are determined by consulting the
1040 * environment variables <envar>G_FILENAME_ENCODING</envar> and
1041 * <envar>G_BROKEN_FILENAMES</envar>. On Windows, the character set
1042 * used in the GLib API is always UTF-8 and said environment variables
1045 * <envar>G_FILENAME_ENCODING</envar> may be set to a comma-separated list
1046 * of character set names. The special token "@locale" is taken to mean the
1047 * character set for the current locale. If <envar>G_FILENAME_ENCODING</envar>
1048 * is not set, but <envar>G_BROKEN_FILENAMES</envar> is, the character set of
1049 * the current locale is taken as the filename encoding. If neither environment
1050 * variable is set, UTF-8 is taken as the filename encoding, but the character
1051 * set of the current locale is also put in the list of encodings.
1053 * The returned @charsets belong to GLib and must not be freed.
1055 * Note that on Unix, regardless of the locale character set or
1056 * <envar>G_FILENAME_ENCODING</envar> value, the actual file names present on a
1057 * system might be in any random encoding or just gibberish.
1059 * Return value: %TRUE if the filename encoding is UTF-8.
1064 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1066 static GStaticPrivate cache_private = G_STATIC_PRIVATE_INIT;
1067 GFilenameCharsetCache *cache = g_static_private_get (&cache_private);
1068 const gchar *charset;
1072 cache = g_new0 (GFilenameCharsetCache, 1);
1073 g_static_private_set (&cache_private, cache, filename_charset_cache_free);
1076 g_get_charset (&charset);
1078 if (!(cache->charset && strcmp (cache->charset, charset) == 0))
1080 const gchar *new_charset;
1084 g_free (cache->charset);
1085 g_strfreev (cache->filename_charsets);
1086 cache->charset = g_strdup (charset);
1088 p = getenv ("G_FILENAME_ENCODING");
1089 if (p != NULL && p[0] != '\0')
1091 cache->filename_charsets = g_strsplit (p, ",", 0);
1092 cache->is_utf8 = (strcmp (cache->filename_charsets[0], "UTF-8") == 0);
1094 for (i = 0; cache->filename_charsets[i]; i++)
1096 if (strcmp ("@locale", cache->filename_charsets[i]) == 0)
1098 g_get_charset (&new_charset);
1099 g_free (cache->filename_charsets[i]);
1100 cache->filename_charsets[i] = g_strdup (new_charset);
1104 else if (getenv ("G_BROKEN_FILENAMES") != NULL)
1106 cache->filename_charsets = g_new0 (gchar *, 2);
1107 cache->is_utf8 = g_get_charset (&new_charset);
1108 cache->filename_charsets[0] = g_strdup (new_charset);
1112 cache->filename_charsets = g_new0 (gchar *, 3);
1113 cache->is_utf8 = TRUE;
1114 cache->filename_charsets[0] = g_strdup ("UTF-8");
1115 if (!g_get_charset (&new_charset))
1116 cache->filename_charsets[1] = g_strdup (new_charset);
1120 if (filename_charsets)
1121 *filename_charsets = (const gchar **)cache->filename_charsets;
1123 return cache->is_utf8;
1126 #else /* G_PLATFORM_WIN32 */
1129 g_get_filename_charsets (G_CONST_RETURN gchar ***filename_charsets)
1131 static const gchar *charsets[] = {
1137 /* On Windows GLib pretends that the filename charset is UTF-8 */
1138 if (filename_charsets)
1139 *filename_charsets = charsets;
1145 /* Cygwin works like before */
1146 result = g_get_charset (&(charsets[0]));
1148 if (filename_charsets)
1149 *filename_charsets = charsets;
1155 #endif /* G_PLATFORM_WIN32 */
1158 get_filename_charset (const gchar **filename_charset)
1160 const gchar **charsets;
1163 is_utf8 = g_get_filename_charsets (&charsets);
1165 if (filename_charset)
1166 *filename_charset = charsets[0];
1171 /* This is called from g_thread_init(). It's used to
1172 * initialize some static data in a threadsafe way.
1175 _g_convert_thread_init (void)
1177 const gchar **dummy;
1178 (void) g_get_filename_charsets (&dummy);
1182 * g_filename_to_utf8:
1183 * @opsysstring: a string in the encoding for filenames
1184 * @len: the length of the string, or -1 if the string is
1185 * nul-terminated<footnoteref linkend="nul-unsafe"/>.
1186 * @bytes_read: location to store the number of bytes in the
1187 * input string that were successfully converted, or %NULL.
1188 * Even if the conversion was successful, this may be
1189 * less than @len if there were partial characters
1190 * at the end of the input. If the error
1191 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1192 * stored will the byte offset after the last valid
1194 * @bytes_written: the number of bytes stored in the output buffer (not
1195 * including the terminating nul).
1196 * @error: location to store the error occuring, or %NULL to ignore
1197 * errors. Any of the errors in #GConvertError may occur.
1199 * Converts a string which is in the encoding used by GLib for
1200 * filenames into a UTF-8 string. Note that on Windows GLib uses UTF-8
1203 * Return value: The converted string, or %NULL on an error.
1206 g_filename_to_utf8 (const gchar *opsysstring,
1209 gsize *bytes_written,
1212 const gchar *charset;
1214 if (get_filename_charset (&charset))
1215 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1217 return g_convert (opsysstring, len,
1218 "UTF-8", charset, bytes_read, bytes_written, error);
1223 #undef g_filename_to_utf8
1225 /* Binary compatibility version. Not for newly compiled code. */
1228 g_filename_to_utf8 (const gchar *opsysstring,
1231 gsize *bytes_written,
1234 const gchar *charset;
1236 if (g_get_charset (&charset))
1237 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1239 return g_convert (opsysstring, len,
1240 "UTF-8", charset, bytes_read, bytes_written, error);
1246 * g_filename_from_utf8:
1247 * @utf8string: a UTF-8 encoded string.
1248 * @len: the length of the string, or -1 if the string is
1250 * @bytes_read: location to store the number of bytes in the
1251 * input string that were successfully converted, or %NULL.
1252 * Even if the conversion was successful, this may be
1253 * less than @len if there were partial characters
1254 * at the end of the input. If the error
1255 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1256 * stored will the byte offset after the last valid
1258 * @bytes_written: the number of bytes stored in the output buffer (not
1259 * including the terminating nul).
1260 * @error: location to store the error occuring, or %NULL to ignore
1261 * errors. Any of the errors in #GConvertError may occur.
1263 * Converts a string from UTF-8 to the encoding GLib uses for
1264 * filenames. Note that on Windows GLib uses UTF-8 for filenames.
1266 * Return value: The converted string, or %NULL on an error.
1269 g_filename_from_utf8 (const gchar *utf8string,
1272 gsize *bytes_written,
1275 const gchar *charset;
1277 if (get_filename_charset (&charset))
1278 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1280 return g_convert (utf8string, len,
1281 charset, "UTF-8", bytes_read, bytes_written, error);
1286 #undef g_filename_from_utf8
1288 /* Binary compatibility version. Not for newly compiled code. */
1291 g_filename_from_utf8 (const gchar *utf8string,
1294 gsize *bytes_written,
1297 const gchar *charset;
1299 if (g_get_charset (&charset))
1300 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1302 return g_convert (utf8string, len,
1303 charset, "UTF-8", bytes_read, bytes_written, error);
1308 /* Test of haystack has the needle prefix, comparing case
1309 * insensitive. haystack may be UTF-8, but needle must
1310 * contain only ascii. */
1312 has_case_prefix (const gchar *haystack, const gchar *needle)
1316 /* Eat one character at a time. */
1321 g_ascii_tolower (*n) == g_ascii_tolower (*h))
1331 UNSAFE_ALL = 0x1, /* Escape all unsafe characters */
1332 UNSAFE_ALLOW_PLUS = 0x2, /* Allows '+' */
1333 UNSAFE_PATH = 0x8, /* Allows '/', '&', '=', ':', '@', '+', '$' and ',' */
1334 UNSAFE_HOST = 0x10, /* Allows '/' and ':' and '@' */
1335 UNSAFE_SLASHES = 0x20 /* Allows all characters except for '/' and '%' */
1336 } UnsafeCharacterSet;
1338 static const guchar acceptable[96] = {
1339 /* A table of the ASCII chars from space (32) to DEL (127) */
1340 /* ! " # $ % & ' ( ) * + , - . / */
1341 0x00,0x3F,0x20,0x20,0x28,0x00,0x2C,0x3F,0x3F,0x3F,0x3F,0x2A,0x28,0x3F,0x3F,0x1C,
1342 /* 0 1 2 3 4 5 6 7 8 9 : ; < = > ? */
1343 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x38,0x20,0x20,0x2C,0x20,0x20,
1344 /* @ A B C D E F G H I J K L M N O */
1345 0x38,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1346 /* P Q R S T U V W X Y Z [ \ ] ^ _ */
1347 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x20,0x3F,
1348 /* ` a b c d e f g h i j k l m n o */
1349 0x20,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1350 /* p q r s t u v w x y z { | } ~ DEL */
1351 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x3F,0x20
1354 static const gchar hex[16] = "0123456789ABCDEF";
1356 /* Note: This escape function works on file: URIs, but if you want to
1357 * escape something else, please read RFC-2396 */
1359 g_escape_uri_string (const gchar *string,
1360 UnsafeCharacterSet mask)
1362 #define ACCEPTABLE(a) ((a)>=32 && (a)<128 && (acceptable[(a)-32] & use_mask))
1369 UnsafeCharacterSet use_mask;
1371 g_return_val_if_fail (mask == UNSAFE_ALL
1372 || mask == UNSAFE_ALLOW_PLUS
1373 || mask == UNSAFE_PATH
1374 || mask == UNSAFE_HOST
1375 || mask == UNSAFE_SLASHES, NULL);
1379 for (p = string; *p != '\0'; p++)
1382 if (!ACCEPTABLE (c))
1386 result = g_malloc (p - string + unacceptable * 2 + 1);
1389 for (q = result, p = string; *p != '\0'; p++)
1393 if (!ACCEPTABLE (c))
1395 *q++ = '%'; /* means hex coming */
1410 g_escape_file_uri (const gchar *hostname,
1411 const gchar *pathname)
1413 char *escaped_hostname = NULL;
1418 char *p, *backslash;
1420 /* Turn backslashes into forward slashes. That's what Netscape
1421 * does, and they are actually more or less equivalent in Windows.
1424 pathname = g_strdup (pathname);
1425 p = (char *) pathname;
1427 while ((backslash = strchr (p, '\\')) != NULL)
1434 if (hostname && *hostname != '\0')
1436 escaped_hostname = g_escape_uri_string (hostname, UNSAFE_HOST);
1439 escaped_path = g_escape_uri_string (pathname, UNSAFE_PATH);
1441 res = g_strconcat ("file://",
1442 (escaped_hostname) ? escaped_hostname : "",
1443 (*escaped_path != '/') ? "/" : "",
1448 g_free ((char *) pathname);
1451 g_free (escaped_hostname);
1452 g_free (escaped_path);
1458 unescape_character (const char *scanner)
1463 first_digit = g_ascii_xdigit_value (scanner[0]);
1464 if (first_digit < 0)
1467 second_digit = g_ascii_xdigit_value (scanner[1]);
1468 if (second_digit < 0)
1471 return (first_digit << 4) | second_digit;
1475 g_unescape_uri_string (const char *escaped,
1477 const char *illegal_escaped_characters,
1478 gboolean ascii_must_not_be_escaped)
1480 const gchar *in, *in_end;
1481 gchar *out, *result;
1484 if (escaped == NULL)
1488 len = strlen (escaped);
1490 result = g_malloc (len + 1);
1493 for (in = escaped, in_end = escaped + len; in < in_end; in++)
1499 /* catch partial escape sequences past the end of the substring */
1500 if (in + 3 > in_end)
1503 c = unescape_character (in + 1);
1505 /* catch bad escape sequences and NUL characters */
1509 /* catch escaped ASCII */
1510 if (ascii_must_not_be_escaped && c <= 0x7F)
1513 /* catch other illegal escaped characters */
1514 if (strchr (illegal_escaped_characters, c) != NULL)
1523 g_assert (out - result <= len);
1536 is_asciialphanum (gunichar c)
1538 return c <= 0x7F && g_ascii_isalnum (c);
1542 is_asciialpha (gunichar c)
1544 return c <= 0x7F && g_ascii_isalpha (c);
1547 /* allows an empty string */
1549 hostname_validate (const char *hostname)
1552 gunichar c, first_char, last_char;
1559 /* read in a label */
1560 c = g_utf8_get_char (p);
1561 p = g_utf8_next_char (p);
1562 if (!is_asciialphanum (c))
1568 c = g_utf8_get_char (p);
1569 p = g_utf8_next_char (p);
1571 while (is_asciialphanum (c) || c == '-');
1572 if (last_char == '-')
1575 /* if that was the last label, check that it was a toplabel */
1576 if (c == '\0' || (c == '.' && *p == '\0'))
1577 return is_asciialpha (first_char);
1584 * g_filename_from_uri:
1585 * @uri: a uri describing a filename (escaped, encoded in ASCII).
1586 * @hostname: Location to store hostname for the URI, or %NULL.
1587 * If there is no hostname in the URI, %NULL will be
1588 * stored in this location.
1589 * @error: location to store the error occuring, or %NULL to ignore
1590 * errors. Any of the errors in #GConvertError may occur.
1592 * Converts an escaped ASCII-encoded URI to a local filename in the
1593 * encoding used for filenames.
1595 * Return value: a newly-allocated string holding the resulting
1596 * filename, or %NULL on an error.
1599 g_filename_from_uri (const gchar *uri,
1603 const char *path_part;
1604 const char *host_part;
1605 char *unescaped_hostname;
1616 if (!has_case_prefix (uri, "file:/"))
1618 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1619 _("The URI '%s' is not an absolute URI using the \"file\" scheme"),
1624 path_part = uri + strlen ("file:");
1626 if (strchr (path_part, '#') != NULL)
1628 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1629 _("The local file URI '%s' may not include a '#'"),
1634 if (has_case_prefix (path_part, "///"))
1636 else if (has_case_prefix (path_part, "//"))
1639 host_part = path_part;
1641 path_part = strchr (path_part, '/');
1643 if (path_part == NULL)
1645 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1646 _("The URI '%s' is invalid"),
1651 unescaped_hostname = g_unescape_uri_string (host_part, path_part - host_part, "", TRUE);
1653 if (unescaped_hostname == NULL ||
1654 !hostname_validate (unescaped_hostname))
1656 g_free (unescaped_hostname);
1657 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1658 _("The hostname of the URI '%s' is invalid"),
1664 *hostname = unescaped_hostname;
1666 g_free (unescaped_hostname);
1669 filename = g_unescape_uri_string (path_part, -1, "/", FALSE);
1671 if (filename == NULL)
1673 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1674 _("The URI '%s' contains invalidly escaped characters"),
1681 /* Drop localhost */
1682 if (hostname && *hostname != NULL &&
1683 g_ascii_strcasecmp (*hostname, "localhost") == 0)
1689 /* Turn slashes into backslashes, because that's the canonical spelling */
1691 while ((slash = strchr (p, '/')) != NULL)
1697 /* Windows URIs with a drive letter can be like "file://host/c:/foo"
1698 * or "file://host/c|/foo" (some Netscape versions). In those cases, start
1699 * the filename from the drive letter.
1701 if (g_ascii_isalpha (filename[1]))
1703 if (filename[2] == ':')
1705 else if (filename[2] == '|')
1713 result = g_strdup (filename + offs);
1721 #undef g_filename_from_uri
1724 g_filename_from_uri (const gchar *uri,
1728 gchar *utf8_filename;
1729 gchar *retval = NULL;
1731 utf8_filename = g_filename_from_uri_utf8 (uri, hostname, error);
1734 retval = g_locale_from_utf8 (utf8_filename, -1, NULL, NULL, error);
1735 g_free (utf8_filename);
1743 * g_filename_to_uri:
1744 * @filename: an absolute filename specified in the GLib file name encoding,
1745 * which is the on-disk file name bytes on Unix, and UTF-8 on
1747 * @hostname: A UTF-8 encoded hostname, or %NULL for none.
1748 * @error: location to store the error occuring, or %NULL to ignore
1749 * errors. Any of the errors in #GConvertError may occur.
1751 * Converts an absolute filename to an escaped ASCII-encoded URI.
1753 * Return value: a newly-allocated string holding the resulting
1754 * URI, or %NULL on an error.
1757 g_filename_to_uri (const gchar *filename,
1758 const gchar *hostname,
1763 g_return_val_if_fail (filename != NULL, NULL);
1765 if (!g_path_is_absolute (filename))
1767 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NOT_ABSOLUTE_PATH,
1768 _("The pathname '%s' is not an absolute path"),
1774 !(g_utf8_validate (hostname, -1, NULL)
1775 && hostname_validate (hostname)))
1777 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
1778 _("Invalid hostname"));
1783 /* Don't use localhost unnecessarily */
1784 if (hostname && g_ascii_strcasecmp (hostname, "localhost") == 0)
1788 escaped_uri = g_escape_file_uri (hostname, filename);
1795 #undef g_filename_to_uri
1798 g_filename_to_uri (const gchar *filename,
1799 const gchar *hostname,
1802 gchar *utf8_filename;
1803 gchar *retval = NULL;
1805 utf8_filename = g_locale_to_utf8 (filename, -1, NULL, NULL, error);
1809 retval = g_filename_to_uri_utf8 (utf8_filename, hostname, error);
1810 g_free (utf8_filename);
1819 * g_uri_list_extract_uris:
1820 * @uri_list: an URI list
1822 * Splits an URI list conforming to the text/uri-list
1823 * mime type defined in RFC 2483 into individual URIs,
1824 * discarding any comments. The URIs are not validated.
1826 * Returns: a newly allocated %NULL-terminated list of
1827 * strings holding the individual URIs. The array should
1828 * be freed with g_strfreev().
1833 g_uri_list_extract_uris (const gchar *uri_list)
1844 /* We don't actually try to validate the URI according to RFC
1845 * 2396, or even check for allowed characters - we just ignore
1846 * comments and trim whitespace off the ends. We also
1847 * allow LF delimination as well as the specified CRLF.
1849 * We do allow comments like specified in RFC 2483.
1855 while (g_ascii_isspace (*p))
1859 while (*q && (*q != '\n') && (*q != '\r'))
1865 while (q > p && g_ascii_isspace (*q))
1870 uris = g_slist_prepend (uris, g_strndup (p, q - p + 1));
1875 p = strchr (p, '\n');
1880 result = g_new (gchar *, n_uris + 1);
1882 result[n_uris--] = NULL;
1883 for (u = uris; u; u = u->next)
1884 result[n_uris--] = u->data;
1886 g_slist_free (uris);
1892 make_valid_utf8 (const gchar *name)
1895 const gchar *remainder, *invalid;
1896 gint remaining_bytes, valid_bytes;
1900 remaining_bytes = strlen (name);
1902 while (remaining_bytes != 0)
1904 if (g_utf8_validate (remainder, remaining_bytes, &invalid))
1906 valid_bytes = invalid - remainder;
1909 string = g_string_sized_new (remaining_bytes);
1911 g_string_append_len (string, remainder, valid_bytes);
1912 g_string_append_c (string, '?');
1914 remaining_bytes -= valid_bytes + 1;
1915 remainder = invalid + 1;
1919 return g_strdup (name);
1921 g_string_append (string, remainder);
1922 g_string_append (string, " (invalid encoding)");
1924 g_assert (g_utf8_validate (string->str, -1, NULL));
1926 return g_string_free (string, FALSE);
1930 * g_filename_display_basename:
1931 * @filename: an absolute pathname in the GLib file name encoding
1933 * Returns the display basename for the particular filename, guaranteed
1934 * to be valid UTF-8. The display name might not be identical to the filename,
1935 * for instance there might be problems converting it to UTF-8, and some files
1936 * can be translated in the display
1938 * You must pass the whole absolute pathname to this functions so that
1939 * translation of well known locations can be done.
1941 * This function is preferred over g_filename_display_name() if you know the
1942 * whole path, as it allows translation.
1944 * Return value: a newly allocated string containing
1945 * a rendition of the basename of the filename in valid UTF-8
1950 g_filename_display_basename (const gchar *filename)
1955 g_return_val_if_fail (filename != NULL, NULL);
1957 basename = g_path_get_basename (filename);
1958 display_name = g_filename_display_name (basename);
1960 return display_name;
1964 * g_filename_display_name:
1965 * @filename: a pathname hopefully in the GLib file name encoding
1967 * Converts a filename into a valid UTF-8 string. The
1968 * conversion is not necessarily reversible, so you
1969 * should keep the original around and use the return
1970 * value of this function only for display purposes.
1971 * Unlike g_filename_to_utf8(), the result is guaranteed
1972 * to be non-NULL even if the filename actually isn't in the GLib
1973 * file name encoding.
1975 * If you know the whole pathname of the file you should use
1976 * g_filename_display_basename(), since that allows location-based
1977 * translation of filenames.
1979 * Return value: a newly allocated string containing
1980 * a rendition of the filename in valid UTF-8
1985 g_filename_display_name (const gchar *filename)
1988 const gchar **charsets;
1989 gchar *display_name = NULL;
1992 is_utf8 = g_get_filename_charsets (&charsets);
1996 if (g_utf8_validate (filename, -1, NULL))
1997 display_name = g_strdup (filename);
2002 /* Try to convert from the filename charsets to UTF-8.
2003 * Skip the first charset if it is UTF-8.
2005 for (i = is_utf8 ? 1 : 0; charsets[i]; i++)
2007 display_name = g_convert (filename, -1, "UTF-8", charsets[i],
2015 /* if all conversions failed, we replace invalid UTF-8
2016 * by a question mark
2019 display_name = make_valid_utf8 (filename);
2021 return display_name;
2024 #define __G_CONVERT_C__
2025 #include "galiasdef.c"