/* PSPP - a program for statistical analysis.
- Copyright (C) 2006, 2009 Free Software Foundation, Inc.
+ Copyright (C) 2006, 2009, 2010 Free Software Foundation, Inc.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
#include <libintl.h>
#include <iconv.h>
#include <errno.h>
+#include <relocatable.h>
#include "assertion.h"
#include "hmapx.h"
#include "hash-functions.h"
+#include "pool.h"
#include "i18n.h"
#include <langinfo.h>
#endif
+struct converter
+ {
+ char *tocode;
+ char *fromcode;
+ iconv_t conv;
+ };
+
static char *default_encoding;
static struct hmapx map;
static iconv_t
create_iconv (const char* tocode, const char* fromcode)
{
- iconv_t conv;
+ size_t hash;
struct hmapx_node *node;
- size_t hash ;
- char *key = alloca (strlen (tocode) + strlen (fromcode) + 2);
-
- strcpy (key, tocode);
- strcat (key, "\n"); /* hopefully no encoding names contain '\n' */
- strcat (key, fromcode);
-
- hash = hsh_hash_string (key);
-
- node = hmapx_first_with_hash (&map, hash);
-
- if (!node)
- {
- conv = iconv_open (tocode, fromcode);
-
- /* I don't think it's safe to translate this string or to use messaging
- as the convertors have not yet been set up */
- if ( (iconv_t) -1 == conv && 0 != strcmp (tocode, fromcode))
- {
- const int err = errno;
- fprintf (stderr,
- "Warning: "
- "cannot create a convertor for \"%s\" to \"%s\": %s\n",
- fromcode, tocode, strerror (err));
- }
-
- hmapx_insert (&map, conv, hash);
- }
- else
+ struct converter *converter;
+ assert (fromcode);
+
+ hash = hash_string (tocode, hash_string (fromcode, 0));
+ HMAPX_FOR_EACH_WITH_HASH (converter, node, hash, &map)
+ if (!strcmp (tocode, converter->tocode)
+ && !strcmp (fromcode, converter->fromcode))
+ return converter->conv;
+
+ converter = xmalloc (sizeof *converter);
+ converter->tocode = xstrdup (tocode);
+ converter->fromcode = xstrdup (fromcode);
+ converter->conv = iconv_open (tocode, fromcode);
+ hmapx_insert (&map, converter, hash);
+
+ /* I don't think it's safe to translate this string or to use messaging
+ as the converters have not yet been set up */
+ if ( (iconv_t) -1 == converter->conv && 0 != strcmp (tocode, fromcode))
{
- conv = hmapx_node_data (node);
+ const int err = errno;
+ fprintf (stderr,
+ "Warning: "
+ "cannot create a converter for `%s' to `%s': %s\n",
+ fromcode, tocode, strerror (err));
}
- return conv;
+ return converter->conv;
}
-/* Return a string based on TEXT converted according to HOW.
- If length is not -1, then it must be the number of bytes in TEXT.
- The returned string must be freed when no longer required.
-*/
+
+/* Similar to recode_string_pool, but allocates the returned value on the heap
+ instead of in a pool. It is the caller's responsibility to free the
+ returned value. */
char *
recode_string (const char *to, const char *from,
const char *text, int length)
+{
+ return recode_string_pool (to, from, text, length, NULL);
+}
+
+
+/* Converts the string TEXT, which should be encoded in FROM-encoding, to a
+ dynamically allocated string in TO-encoding. Any characters which cannot be
+ converted will be represented by '?'.
+
+ LENGTH should be the length of the string or -1, if null terminated.
+
+ The returned string will be allocated on POOL.
+
+ This function's behaviour differs from that of g_convert_with_fallback
+ provided by GLib. The GLib function will fail (returns NULL) if any part of
+ the input string is not valid in the declared input encoding. This function
+ however perseveres even in the presence of badly encoded input. */
+char *
+recode_string_pool (const char *to, const char *from,
+ const char *text, int length, struct pool *pool)
{
char *outbuf = 0;
size_t outbufferlength;
size_t result;
+ char *ip;
char *op ;
size_t inbytes = 0;
size_t outbytes ;
if ( length == -1 )
length = strlen(text);
-
if (to == NULL)
to = default_encoding;
if (from == NULL)
from = default_encoding;
+ conv = create_iconv (to, from);
+
+ if ( (iconv_t) -1 == conv )
+ return xstrdup (text);
+
+ /* Put the converter into the initial shift state, in case there was any
+ state information left over from its last usage. */
+ iconv (conv, NULL, 0, NULL, 0);
+
for ( outbufferlength = 1 ; outbufferlength != 0; outbufferlength <<= 1 )
if ( outbufferlength > length)
break;
- outbuf = xmalloc(outbufferlength);
+ ip = text;
+
+ outbuf = pool_malloc (pool, outbufferlength);
op = outbuf;
outbytes = outbufferlength;
inbytes = length;
- conv = create_iconv (to, from);
-
do {
- const char *ip = text;
- result = iconv (conv, (ICONV_CONST char **) &text, &inbytes,
+ result = iconv (conv, (ICONV_CONST char **) &ip, &inbytes,
&op, &outbytes);
if ( -1 == result )
}
/* Fall through */
case E2BIG:
- free (outbuf);
+ iconv (conv, NULL, 0, NULL, 0);
+ pool_free (pool, outbuf);
outbufferlength <<= 1;
- outbuf = xmalloc (outbufferlength);
+ outbuf = pool_malloc (pool, outbufferlength);
op = outbuf;
outbytes = outbufferlength;
inbytes = length;
- text = ip;
+ ip = text;
break;
default:
/* should never happen */
+ fprintf (stderr, "Character conversion error: %s\n",
+ strerror (the_error));
NOT_REACHED ();
break;
}
if (outbytes == 0 )
{
char *const oldaddr = outbuf;
- outbuf = xrealloc (outbuf, outbufferlength + 1);
+ outbuf = pool_realloc (pool, outbuf, outbufferlength + 1);
op += (outbuf - oldaddr) ;
}
{
#if ENABLE_NLS
setlocale (LC_CTYPE, "");
-#if HAVE_LC_MESSAGES
+#ifdef LC_MESSAGES
setlocale (LC_MESSAGES, "");
#endif
#if HAVE_LC_PAPER
setlocale (LC_PAPER, "");
#endif
- bindtextdomain (PACKAGE, locale_dir);
+ bindtextdomain (PACKAGE, relocate(locale_dir));
textdomain (PACKAGE);
#endif /* ENABLE_NLS */
assert (default_encoding == NULL);
- default_encoding = strdup (locale_charset ());
+ default_encoding = xstrdup (locale_charset ());
hmapx_init (&map);
}
+const char *
+get_default_encoding (void)
+{
+ return default_encoding;
+}
+
+void
+set_default_encoding (const char *enc)
+{
+ free (default_encoding);
+ default_encoding = xstrdup (enc);
+}
+
+
+/* Attempts to set the encoding from a locale name
+ returns true if successfull.
+ This function does not (should not!) alter the current locale.
+*/
+bool
+set_encoding_from_locale (const char *loc)
+{
+ bool ok = true;
+ char *c_encoding;
+ char *loc_encoding;
+ char *tmp = xstrdup (setlocale (LC_CTYPE, NULL));
+
+ setlocale (LC_CTYPE, "C");
+ c_encoding = xstrdup (locale_charset ());
+
+ setlocale (LC_CTYPE, loc);
+ loc_encoding = xstrdup (locale_charset ());
+
+
+ if ( 0 == strcmp (loc_encoding, c_encoding))
+ {
+ ok = false;
+ }
+
+
+ setlocale (LC_CTYPE, tmp);
+
+ free (tmp);
+
+ if (ok)
+ {
+ free (default_encoding);
+ default_encoding = loc_encoding;
+ }
+ else
+ free (loc_encoding);
+
+ free (c_encoding);
+
+ return ok;
+}
+
void
i18n_done (void)
{
struct hmapx_node *node;
- iconv_t conv;
- HMAPX_FOR_EACH (conv, node, &map)
- iconv_close (conv);
+ struct converter *cvtr;
+
+ HMAPX_FOR_EACH (cvtr, node, &map)
+ {
+ free (cvtr->tocode);
+ free (cvtr->fromcode);
+ iconv_close (cvtr->conv);
+ free (cvtr);
+ }
hmapx_destroy (&map);
+bool
+valid_encoding (const char *enc)
+{
+ iconv_t conv = iconv_open ("UTF8", enc);
+
+ if ( conv == (iconv_t) -1)
+ return false;
+
+ iconv_close (conv);
+
+ return true;
+}
+
/* Return the system local's idea of the
decimal seperator character */
{
char radix_char;
- char *ol = strdup (setlocale (LC_NUMERIC, NULL));
+ char *ol = xstrdup (setlocale (LC_NUMERIC, NULL));
setlocale (LC_NUMERIC, "");
#if HAVE_NL_LANGINFO
projects / pspp / blobdiff