1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2006, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19 #include "libpspp/i18n.h"
32 #include "libpspp/assertion.h"
33 #include "libpspp/compiler.h"
34 #include "libpspp/hmapx.h"
35 #include "libpspp/hash-functions.h"
36 #include "libpspp/pool.h"
37 #include "libpspp/str.h"
38 #include "libpspp/version.h"
40 #include "gl/c-ctype.h"
41 #include "gl/c-strcase.h"
42 #include "gl/localcharset.h"
43 #include <gl/localename.h>
44 #include "gl/minmax.h"
45 #include "gl/xalloc.h"
46 #include "gl/relocatable.h"
47 #include "gl/xstrndup.h"
50 #define _(msgid) gettext (msgid)
60 static char *default_encoding;
61 static struct hmapx map;
63 /* A wrapper around iconv_open */
64 static struct converter *
65 create_iconv (const char* tocode, const char* fromcode)
68 struct hmapx_node *node;
69 struct converter *converter;
72 hash = hash_string (tocode, hash_string (fromcode, 0));
73 HMAPX_FOR_EACH_WITH_HASH (converter, node, hash, &map)
78 if (!strcmp (tocode, converter->tocode)
79 && !strcmp (fromcode, converter->fromcode))
83 converter = xmalloc (sizeof *converter);
84 converter->tocode = xstrdup (tocode);
85 converter->fromcode = xstrdup (fromcode);
86 converter->conv = iconv_open (tocode, fromcode);
87 int error = converter->conv == (iconv_t) ~0 ? errno : 0;
88 /* I don't think it's safe to translate this string or to use messaging
89 as the converters have not yet been set up */
90 if (error && strcmp (tocode, fromcode))
94 "cannot create a converter for `%s' to `%s': %s\n",
95 fromcode, tocode, strerror (error));
97 free (converter->tocode);
98 free (converter->fromcode);
101 hmapx_insert (&map, NULL, hash);
105 /* Find out how many bytes there are in a null char in the target
107 iconv_t bconv = iconv_open (tocode, "ASCII");
108 if (bconv != (iconv_t) -1)
110 ICONV_CONST char *nullstr = strdup ("");
111 ICONV_CONST char *outbuf = strdup ("XXXXXXXX");
112 ICONV_CONST char *snullstr = nullstr;
113 ICONV_CONST char *soutbuf = outbuf;
116 const size_t bytes = 8;
117 size_t outbytes = bytes;
118 if (-1 != iconv (bconv, &nullstr, &inbytes, &outbuf, &outbytes))
119 converter->null_char_width = bytes - outbytes;
125 hmapx_insert (&map, converter, hash);
131 /* Converts the single byte C from encoding FROM to TO, returning the first
134 This function probably shouldn't be used at all, but some code still does
137 recode_byte (const char *to, const char *from, char c)
140 char *s = recode_string (to, from, &c, 1);
146 /* Similar to recode_string_pool, but allocates the returned value on the heap
147 instead of in a pool. It is the caller's responsibility to free the
150 recode_string (const char *to, const char *from,
151 const char *text, int length)
153 return recode_string_pool (to, from, text, length, NULL);
156 /* Returns the length, in bytes, of the string that a similar recode_string()
157 call would return. */
159 recode_string_len (const char *to, const char *from,
160 const char *text, int length)
162 char *s = recode_string (to, from, text, length);
163 size_t len = strlen (s);
168 /* Uses CONV to convert the INBYTES starting at IP into the OUTBYTES starting
169 at OP, and appends a null terminator to the output.
171 Returns the output length if successful, -1 if the output buffer is too
174 try_recode (struct converter *cvtr, char fallbackchar,
175 const char *in, size_t inbytes,
176 char *out_, size_t outbytes)
181 int null_bytes = cvtr->null_char_width;
183 /* Put the converter into the initial shift state, in case there was any
184 state information left over from its last usage. */
185 iconv (cvtr->conv, NULL, 0, NULL, 0);
187 /* Do two rounds of iconv() calls:
189 - The first round does the bulk of the conversion using the
190 caller-supplied input data..
192 - The second round flushes any leftover output. This has a real effect
193 with input encodings that use combining diacritics, e.g. without the
194 second round the last character tends to gets dropped when converting
195 from windows-1258 to other encodings.
197 for (i = 0; i < 2; i++)
199 ICONV_CONST char **inp = i ? NULL : (ICONV_CONST char **) ∈
200 size_t *inbytesp = i ? NULL : &inbytes;
202 while (iconv (cvtr->conv, inp, inbytesp, &out, &outbytes) == -1)
206 if (outbytes < null_bytes + 1)
210 *out++ = fallbackchar;
211 for (j = 0 ; j < null_bytes ; ++j)
213 return out - 1 - out_;
220 *out++ = fallbackchar;
233 /* should never happen */
234 fprintf (stderr, "Character conversion error: %s\n",
241 if (outbytes <= null_bytes - 1)
244 for (i = 0 ; i < null_bytes ; ++i)
247 return out - 1 - out_;
250 /* Converts the string TEXT, which should be encoded in FROM-encoding, to a
251 dynamically allocated string in TO-encoding. Any characters which cannot be
252 converted will be represented by '?'.
254 LENGTH should be the length of the string or -1, if null terminated.
256 The returned string will be allocated on POOL.
258 This function's behaviour differs from that of g_convert_with_fallback
259 provided by GLib. The GLib function will fail (returns NULL) if any part of
260 the input string is not valid in the declared input encoding. This function
261 however perseveres even in the presence of badly encoded input. */
263 recode_string_pool (const char *to, const char *from,
264 const char *text, int length, struct pool *pool)
266 struct substring out;
272 length = strlen (text);
274 out = recode_substring_pool (to, from, ss_buffer (text, length), pool);
278 /* Returns the name of the encoding that should be used for file names.
280 This is meant to be the same encoding used by g_filename_from_uri() and
281 g_filename_to_uri() in GLib. */
283 filename_encoding (void)
285 #if defined _WIN32 || defined __WIN32__
288 return locale_charset ();
293 xconcat2 (const char *a, size_t a_len,
294 const char *b, size_t b_len)
296 char *s = xmalloc (a_len + b_len + 1);
297 memcpy (s, a, a_len);
298 memcpy (s + a_len, b, b_len);
299 s[a_len + b_len] = '\0';
303 /* Conceptually, this function concatenates HEAD_LEN-byte string HEAD and
304 TAIL_LEN-byte string TAIL, both encoded in UTF-8, then converts them to
305 ENCODING. If the re-encoded result is no more than MAX_LEN bytes long, then
306 it returns HEAD_LEN. Otherwise, it drops one character[*] from the end of
307 HEAD and tries again, repeating as necessary until the concatenated result
308 fits or until HEAD_LEN reaches 0.
310 [*] Actually this function drops grapheme clusters instead of characters, so
311 that, e.g. a Unicode character followed by a combining accent character
312 is either completely included or completely excluded from HEAD_LEN. See
313 UAX #29 at http://unicode.org/reports/tr29/ for more information on
316 A null ENCODING is treated as UTF-8.
318 Sometimes this function has to actually construct the concatenated string to
319 measure its length. When this happens, it sets *RESULTP to that
320 null-terminated string, allocated with malloc(), for the caller to use if it
321 needs it. Otherwise, it sets *RESULTP to NULL.
323 Simple examples for encoding="UTF-8", max_len=6:
325 head="abc", tail="xyz" => 3
326 head="abcd", tail="xyz" => 3 ("d" dropped).
327 head="abc", tail="uvwxyz" => 0 ("abc" dropped).
328 head="abc", tail="tuvwxyz" => 0 ("abc" dropped).
330 Examples for encoding="ISO-8859-1", max_len=6:
332 head="éèä", tail="xyz" => 6
333 (each letter in head is only 1 byte in ISO-8859-1 even though they
334 each take 2 bytes in UTF-8 encoding)
337 utf8_encoding_concat__ (const char *head, size_t head_len,
338 const char *tail, size_t tail_len,
339 const char *encoding, size_t max_len,
345 else if (encoding == NULL || !c_strcasecmp (encoding, "UTF-8"))
347 if (head_len + tail_len <= max_len)
349 else if (tail_len >= max_len)
359 for (ofs = u8_mbtouc (&prev, CHAR_CAST (const uint8_t *, head),
361 ofs <= max_len - tail_len;
366 mblen = u8_mbtouc (&next,
367 CHAR_CAST (const uint8_t *, head + ofs),
369 if (uc_is_grapheme_break (prev, next))
382 result = (tail_len > 0
383 ? xconcat2 (head, head_len, tail, tail_len)
384 : CONST_CAST (char *, head));
385 if (recode_string_len (encoding, "UTF-8", result,
386 head_len + tail_len) <= max_len)
388 *resultp = result != head ? result : NULL;
393 bool correct_result = false;
400 for (ofs = u8_mbtouc (&prev, CHAR_CAST (const uint8_t *, head),
407 mblen = u8_mbtouc (&next,
408 CHAR_CAST (const uint8_t *, head + ofs),
410 if (uc_is_grapheme_break (prev, next))
414 memcpy (result, head, ofs);
415 memcpy (result + ofs, tail, tail_len);
416 result[ofs + tail_len] = '\0';
419 if (recode_string_len (encoding, "UTF-8", result,
420 ofs + tail_len) <= max_len)
422 correct_result = true;
426 correct_result = false;
445 /* Concatenates a prefix of HEAD with all of TAIL and returns the result as a
446 null-terminated string owned by the caller. HEAD, TAIL, and the returned
447 string are all encoded in UTF-8. As many characters[*] from the beginning
448 of HEAD are included as will fit within MAX_LEN bytes supposing that the
449 resulting string were to be re-encoded in ENCODING. All of TAIL is always
450 included, even if TAIL by itself is longer than MAX_LEN in ENCODING.
452 [*] Actually this function drops grapheme clusters instead of characters, so
453 that, e.g. a Unicode character followed by a combining accent character
454 is either completely included or completely excluded from the returned
455 string. See UAX #29 at http://unicode.org/reports/tr29/ for more
456 information on grapheme clusters.
458 A null ENCODING is treated as UTF-8.
460 Simple examples for encoding="UTF-8", max_len=6:
462 head="abc", tail="xyz" => "abcxyz"
463 head="abcd", tail="xyz" => "abcxyz"
464 head="abc", tail="uvwxyz" => "uvwxyz"
465 head="abc", tail="tuvwxyz" => "tuvwxyz"
467 Examples for encoding="ISO-8859-1", max_len=6:
469 head="éèä", tail="xyz" => "éèäxyz"
470 (each letter in HEAD is only 1 byte in ISO-8859-1 even though they
471 each take 2 bytes in UTF-8 encoding)
474 utf8_encoding_concat (const char *head, const char *tail,
475 const char *encoding, size_t max_len)
477 size_t tail_len = strlen (tail);
481 prefix_len = utf8_encoding_concat__ (head, strlen (head), tail, tail_len,
482 encoding, max_len, &result);
483 return (result != NULL
485 : xconcat2 (head, prefix_len, tail, tail_len));
488 /* Returns the length, in bytes, of the string that would be returned by
489 utf8_encoding_concat() if passed the same arguments, but the implementation
490 is often more efficient. */
492 utf8_encoding_concat_len (const char *head, const char *tail,
493 const char *encoding, size_t max_len)
495 size_t tail_len = strlen (tail);
499 prefix_len = utf8_encoding_concat__ (head, strlen (head), tail, tail_len,
500 encoding, max_len, &result);
502 return prefix_len + tail_len;
505 /* Returns an allocated, null-terminated string, owned by the caller,
506 containing as many characters[*] from the beginning of S that would fit
507 within MAX_LEN bytes if the returned string were to be re-encoded in
508 ENCODING. Both S and the returned string are encoded in UTF-8.
510 [*] Actually this function drops grapheme clusters instead of characters, so
511 that, e.g. a Unicode character followed by a combining accent character
512 is either completely included or completely excluded from the returned
513 string. See UAX #29 at http://unicode.org/reports/tr29/ for more
514 information on grapheme clusters.
516 A null ENCODING is treated as UTF-8.
519 utf8_encoding_trunc (const char *s, const char *encoding, size_t max_len)
521 return utf8_encoding_concat (s, "", encoding, max_len);
524 /* Returns the length, in bytes, of the string that would be returned by
525 utf8_encoding_trunc() if passed the same arguments, but the implementation
526 is often more efficient. */
528 utf8_encoding_trunc_len (const char *s, const char *encoding, size_t max_len)
530 return utf8_encoding_concat_len (s, "", encoding, max_len);
533 /* Returns FILENAME converted from UTF-8 to the filename encoding.
534 On Windows the filename encoding is UTF-8; elsewhere it is based on the
537 utf8_to_filename (const char *filename)
539 return recode_string (filename_encoding (), "UTF-8", filename, -1);
542 /* Returns FILENAME converted from the filename encoding to UTF-8.
543 On Windows the filename encoding is UTF-8; elsewhere it is based on the
546 filename_to_utf8 (const char *filename)
548 return recode_string ("UTF-8", filename_encoding (), filename, -1);
552 recode_substring_pool__ (const char *to, const char *from,
553 struct substring text, char fallbackchar,
554 struct pool *pool, struct substring *out)
557 struct converter *conv;
560 to = default_encoding;
563 from = default_encoding;
565 conv = create_iconv (to, from);
571 out->string = pool_malloc (pool, text.length + 1);
572 out->length = text.length;
573 memcpy (out->string, text.string, text.length);
574 out->string[out->length] = '\0';
581 for (bufsize = text.length + 1; bufsize > text.length; bufsize *= 2)
583 char *output = pool_malloc (pool, bufsize);
586 retval = try_recode (conv, fallbackchar, text.string, text.length,
590 *out = ss_buffer (output, retval);
593 pool_free (pool, output);
595 if (retval != -E2BIG)
602 /* Converts the string TEXT, which should be encoded in FROM-encoding, to a
603 dynamically allocated string in TO-encoding. Any characters which cannot be
604 converted will be represented by '?'.
606 The returned string will be null-terminated and allocated on POOL with
609 This function's behaviour differs from that of g_convert_with_fallback
610 provided by GLib. The GLib function will fail (returns NULL) if any part of
611 the input string is not valid in the declared input encoding. This function
612 however perseveres even in the presence of badly encoded input. */
614 recode_substring_pool (const char *to, const char *from,
615 struct substring text, struct pool *pool)
617 struct substring out;
619 recode_substring_pool__ (to, from, text, '?', pool, &out);
623 /* Converts the string TEXT, which should be encoded in FROM-encoding, to a
624 dynamically allocated string in TO-encoding. On success, returns 0, and the
625 converted null-terminated string, allocated from POOL with pool_malloc(), is
626 stored in *OUT. On failure, returns a positive errno value.
628 The function fails with an error if any part of the input string is not
629 valid in the declared input encoding. */
631 recode_pedantically (const char *to, const char *from,
632 struct substring text, struct pool *pool,
633 struct substring *out)
637 error = recode_substring_pool__ (to, from, text, 0, pool, out);
646 setlocale (LC_ALL, "");
647 bindtextdomain (PACKAGE, relocate(locale_dir));
648 textdomain (PACKAGE);
650 assert (default_encoding == NULL);
651 default_encoding = xstrdup (locale_charset ());
657 get_default_encoding (void)
659 return default_encoding;
663 set_default_encoding (const char *enc)
665 free (default_encoding);
666 default_encoding = xstrdup (enc);
669 /* Return the ISO two letter code for the current LC_MESSAGES
674 const char *localename = gl_locale_name (LC_MESSAGES, "LC_MESSAGES");
675 if (0 == strcmp (localename, "C"))
677 char *ln = xstrdup (localename);
678 char *end = strchr (ln, '_');
685 /* Attempts to set the encoding from a locale name
686 returns true if successful.
687 This function does not (should not!) alter the current locale.
690 set_encoding_from_locale (const char *loc)
695 char *tmp = xstrdup (setlocale (LC_CTYPE, NULL));
697 setlocale (LC_CTYPE, "C");
698 c_encoding = xstrdup (locale_charset ());
700 setlocale (LC_CTYPE, loc);
701 loc_encoding = xstrdup (locale_charset ());
704 if (0 == strcmp (loc_encoding, c_encoding))
709 setlocale (LC_CTYPE, tmp);
715 free (default_encoding);
716 default_encoding = loc_encoding;
729 struct hmapx_node *node;
730 struct converter *cvtr;
732 HMAPX_FOR_EACH (cvtr, node, &map)
737 free (cvtr->fromcode);
738 if (cvtr->conv != (iconv_t) -1)
739 iconv_close (cvtr->conv);
743 hmapx_destroy (&map);
745 free (default_encoding);
746 default_encoding = NULL;
752 valid_encoding (const char *enc)
754 iconv_t conv = iconv_open (UTF8, enc);
756 if (conv == (iconv_t) -1)
765 /* Return the system local's idea of the
766 decimal separator character */
768 get_system_decimal (void)
773 radix_char = nl_langinfo (RADIXCHAR)[0];
777 snprintf (buf, sizeof buf, "%f", 2.5);
786 uc_name (ucs4_t uc, char buffer[16])
788 if (uc >= 0x20 && uc < 0x7f)
789 snprintf (buffer, 16, "`%c'", uc);
791 snprintf (buffer, 16, "U+%04X", uc);
795 /* UTF-8 functions that deal with uppercase/lowercase distinctions. */
797 /* Returns a hash value for the N bytes of UTF-8 encoded data starting at S,
798 with lowercase and uppercase letters treated as equal, starting from
801 utf8_hash_case_bytes (const char *s, size_t n, unsigned int basis)
803 uint8_t folded_buf[2048];
804 size_t folded_len = sizeof folded_buf;
808 folded_s = u8_casefold (CHAR_CAST (const uint8_t *, s), n,
809 NULL, UNINORM_NFKD, folded_buf, &folded_len);
810 if (folded_s != NULL)
812 hash = hash_bytes (folded_s, folded_len, basis);
813 if (folded_s != folded_buf)
820 hash = hash_bytes (s, n, basis);
826 /* Returns a hash value for null-terminated UTF-8 string S, with lowercase and
827 uppercase letters treated as equal, starting from BASIS. */
829 utf8_hash_case_string (const char *s, unsigned int basis)
831 return utf8_hash_case_bytes (s, strlen (s), basis);
834 /* Compares UTF-8 strings A and B case-insensitively.
835 Returns a negative value if A < B, zero if A == B, positive if A > B. */
837 utf8_strcasecmp (const char *a, const char *b)
839 return utf8_strncasecmp (a, strlen (a), b, strlen (b));
842 /* Compares UTF-8 strings A (with length AN) and B (with length BN)
844 Returns a negative value if A < B, zero if A == B, positive if A > B. */
846 utf8_strncasecmp (const char *a, size_t an, const char *b, size_t bn)
850 if (u8_casecmp (CHAR_CAST (const uint8_t *, a), an,
851 CHAR_CAST (const uint8_t *, b), bn,
852 NULL, UNINORM_NFKD, &result))
857 result = memcmp (a, b, MIN (an, bn));
859 result = an < bn ? -1 : an > bn;
866 is_all_digits (const uint8_t *s, size_t len)
868 for (size_t i = 0; i < len; i++)
869 if (!c_isdigit (s[i]))
874 /* Compares UTF-8 strings A and B case-insensitively. If the strings end in a
875 number, then they are compared numerically. Returns a negative value if A <
876 B, zero if A == B, positive if A > B. */
878 utf8_strverscasecmp (const char *a, const char *b)
882 size_t a_len = sizeof a_stub;
883 uint8_t *a_norm = u8_casefold (CHAR_CAST (uint8_t *, a), strlen (a), NULL,
884 UNINORM_NFKD, a_stub, &a_len);
888 size_t b_len = sizeof b_stub;
889 uint8_t *b_norm = u8_casefold (CHAR_CAST (uint8_t *, b), strlen (b), NULL,
890 UNINORM_NFKD, b_stub, &b_len);
893 if (!a_norm || !b_norm)
895 result = strcmp (a, b);
899 size_t len = MIN (a_len, b_len);
900 for (size_t i = 0; i < len; i++)
901 if (a_norm[i] != b_norm[i])
903 /* If both strings end in digits, compare them numerically. */
904 if (is_all_digits (&a_norm[i], a_len - i)
905 && is_all_digits (&b_norm[i], b_len - i))
907 /* Start by stripping leading zeros, since those don't matter for
908 numerical comparison. */
910 for (ap = i; ap < a_len; ap++)
911 if (a_norm[ap] != '0')
913 for (bp = i; bp < b_len; bp++)
914 if (b_norm[bp] != '0')
917 /* The number with more digits, if there is one, is larger. */
918 size_t a_digits = a_len - ap;
919 size_t b_digits = b_len - bp;
920 if (a_digits != b_digits)
921 result = a_digits > b_digits ? 1 : -1;
923 result = memcmp (&a_norm[ap], &b_norm[bp], a_digits);
926 result = a_norm[i] > b_norm[i] ? 1 : -1;
929 result = a_len < b_len ? -1 : a_len > b_len;
932 if (a_norm != a_stub)
934 if (b_norm != b_stub)
940 utf8_casemap (const char *s,
941 uint8_t *(*f) (const uint8_t *, size_t, const char *, uninorm_t,
942 uint8_t *, size_t *))
947 result = CHAR_CAST (char *,
948 f (CHAR_CAST (const uint8_t *, s), strlen (s) + 1,
949 NULL, NULL, NULL, &size));
955 result = xstrdup (s);
961 utf8_to_upper (const char *s)
963 return utf8_casemap (s, u8_toupper);
967 utf8_to_lower (const char *s)
969 return utf8_casemap (s, u8_tolower);
973 utf8_to_title (const char *s)
975 return utf8_casemap (s, u8_totitle);
979 get_encoding_info (struct encoding_info *e, const char *name)
981 const struct substring in = SS_LITERAL_INITIALIZER (
983 "!\"#$%&'()*+,-./0123456789:;<=>?@"
984 "ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`"
985 "abcdefghijklmnopqrstuvwxyz{|}~");
987 struct substring out, cr, lf, space;
990 memset (e, 0, sizeof *e);
992 cr = recode_substring_pool (name, "UTF-8", ss_cstr ("\r"), NULL);
993 lf = recode_substring_pool (name, "UTF-8", ss_cstr ("\n"), NULL);
994 space = recode_substring_pool (name, "UTF-8", ss_cstr (" "), NULL);
996 && cr.length <= MAX_UNIT
997 && cr.length == lf.length
998 && cr.length == space.length);
1001 fprintf (stderr, "warning: encoding `%s' is not supported.\n", name);
1004 ss_dealloc (&space);
1005 ss_alloc_substring (&cr, ss_cstr ("\r"));
1006 ss_alloc_substring (&lf, ss_cstr ("\n"));
1007 ss_alloc_substring (&space, ss_cstr (" "));
1010 e->unit = cr.length;
1011 memcpy (e->cr, cr.string, e->unit);
1012 memcpy (e->lf, lf.string, e->unit);
1013 memcpy (e->space, space.string, e->unit);
1017 ss_dealloc (&space);
1019 out = recode_substring_pool ("UTF-8", name, in, NULL);
1020 e->is_ascii_compatible = ss_equals (in, out);
1023 if (!e->is_ascii_compatible && e->unit == 1)
1025 out = recode_substring_pool ("UTF-8", name, ss_cstr ("A"), NULL);
1026 e->is_ebcdic_compatible = (out.length == 1
1027 && (uint8_t) out.string[0] == 0xc1);
1031 e->is_ebcdic_compatible = false;
1037 is_encoding_ascii_compatible (const char *encoding)
1039 struct encoding_info e;
1041 get_encoding_info (&e, encoding);
1042 return e.is_ascii_compatible;
1046 is_encoding_ebcdic_compatible (const char *encoding)
1048 struct encoding_info e;
1050 get_encoding_info (&e, encoding);
1051 return e.is_ebcdic_compatible;
1054 /* Returns true if iconv can convert ENCODING to and from UTF-8,
1057 is_encoding_supported (const char *encoding)
1059 return (create_iconv ("UTF-8", encoding)
1060 && create_iconv (encoding, "UTF-8"));
1063 /* Returns true if E is the name of a UTF-8 encoding.
1065 XXX Possibly we should test not E as a string but its properties via
1068 is_encoding_utf8 (const char *e)
1070 return ((e[0] == 'u' || e[0] == 'U')
1071 && (e[1] == 't' || e[1] == 'T')
1072 && (e[2] == 'f' || e[2] == 'F')
1073 && ((e[3] == '8' && e[4] == '\0')
1074 || (e[3] == '-' && e[4] == '8' && e[5] == '\0')));
1077 static struct encoding_category *categories;
1078 static int n_categories;
1080 static void SENTINEL (0)
1081 add_category (size_t *allocated_categories, const char *category, ...)
1083 struct encoding_category *c;
1084 const char *encodings[16];
1088 /* Count encoding arguments. */
1089 va_start (args, category);
1091 while ((encodings[n] = va_arg (args, const char *)) != NULL)
1093 const char *encoding = encodings[n];
1094 if (!strcmp (encoding, "Auto") || is_encoding_supported (encoding))
1097 assert (n < sizeof encodings / sizeof *encodings);
1103 if (n_categories >= *allocated_categories)
1104 categories = x2nrealloc (categories,
1105 allocated_categories, sizeof *categories);
1107 c = &categories[n_categories++];
1108 c->category = category;
1109 c->encodings = xmalloc (n * sizeof *c->encodings);
1110 for (i = 0; i < n; i++)
1111 c->encodings[i] = encodings[i];
1116 init_encoding_categories (void)
1126 add_category (&alloc, "Unicode", "UTF-8", "UTF-16", "UTF-16BE", "UTF-16LE",
1127 "UTF-32", "UTF-32BE", "UTF-32LE", NULL_SENTINEL);
1128 add_category (&alloc, _("Arabic"), "IBM864", "ISO-8859-6", "Windows-1256",
1130 add_category (&alloc, _("Armenian"), "ARMSCII-8", NULL_SENTINEL);
1131 add_category (&alloc, _("Baltic"), "ISO-8859-13", "ISO-8859-4",
1132 "Windows-1257", NULL_SENTINEL);
1133 add_category (&alloc, _("Celtic"), "ISO-8859-14", NULL_SENTINEL);
1134 add_category (&alloc, _("Central European"), "IBM852", "ISO-8859-2",
1135 "Mac-CentralEurope", "Windows-1250", NULL_SENTINEL);
1136 add_category (&alloc, _("Chinese Simplified"), "GB18030", "GB2312", "GBK",
1137 "HZ-GB-2312", "ISO-2022-CN", NULL_SENTINEL);
1138 add_category (&alloc, _("Chinese Traditional"), "Big5", "Big5-HKSCS",
1139 "EUC-TW", NULL_SENTINEL);
1140 add_category (&alloc, _("Croatian"), "MacCroatian", NULL_SENTINEL);
1141 add_category (&alloc, _("Cyrillic"), "IBM855", "ISO-8859-5", "ISO-IR-111",
1142 "KOI8-R", "MacCyrillic", NULL_SENTINEL);
1143 add_category (&alloc, _("Cyrillic/Russian"), "IBM866", NULL_SENTINEL);
1144 add_category (&alloc, _("Cyrillic/Ukrainian"), "KOI8-U", "MacUkrainian",
1146 add_category (&alloc, _("Georgian"), "GEOSTD8", NULL_SENTINEL);
1147 add_category (&alloc, _("Greek"), "ISO-8859-7", "MacGreek", NULL_SENTINEL);
1148 add_category (&alloc, _("Gujarati"), "MacGujarati", NULL_SENTINEL);
1149 add_category (&alloc, _("Gurmukhi"), "MacGurmukhi", NULL_SENTINEL);
1150 add_category (&alloc, _("Hebrew"), "IBM862", "ISO-8859-8-I", "Windows-1255",
1152 add_category (&alloc, _("Hebrew Visual"), "ISO-8859-8", NULL_SENTINEL);
1153 add_category (&alloc, _("Hindi"), "MacDevangari", NULL_SENTINEL);
1154 add_category (&alloc, _("Icelandic"), "MacIcelandic", NULL_SENTINEL);
1155 add_category (&alloc, _("Japanese"), "EUC-JP", "ISO-2022-JP", "Shift_JIS",
1157 add_category (&alloc, _("Korean"), "EUC-KR", "ISO-2022-KR", "JOHAB", "UHC",
1159 add_category (&alloc, _("Nordic"), "ISO-8859-10", NULL_SENTINEL);
1160 add_category (&alloc, _("Romanian"), "ISO-8859-16", "MacRomanian",
1162 add_category (&alloc, _("South European"), "ISO-8859-3", NULL_SENTINEL);
1163 add_category (&alloc, _("Thai"), "ISO-8859-11", "TIS-620", "Windows-874",
1165 add_category (&alloc, _("Turkish"), "IBM857", "ISO-8859-9", "Windows-1254",
1167 add_category (&alloc, _("Vietnamese"), "TVCN", "VISCII", "VPS",
1168 "Windows-1258", NULL_SENTINEL);
1169 add_category (&alloc, _("Western European"), "ISO-8859-1", "ISO-8859-15",
1170 "Windows-1252", "IBM850", "MacRoman", NULL_SENTINEL);
1173 /* Returns an array of "struct encoding_category" that contains only the
1174 categories and encodings that the system supports. */
1175 struct encoding_category *
1176 get_encoding_categories (void)
1178 init_encoding_categories ();
1182 /* Returns the number of elements in the array returned by
1183 get_encoding_categories(). */
1185 get_n_encoding_categories (void)
1187 init_encoding_categories ();
1188 return n_categories;