1 /* decomp.c - Character decomposition.
3 * Copyright (C) 1999, 2000 Tom Tromey
4 * Copyright 2000 Red Hat, Inc.
6 * The Gnome Library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public License as
8 * published by the Free Software Foundation; either version 2 of the
9 * License, or (at your option) any later version.
11 * The Gnome Library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with the Gnome Library; see the file COPYING.LIB. If not,
18 * write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19 * Boston, MA 02111-1307, USA.
27 #include "gunidecomp.h"
29 #include "gunicodeprivate.h"
33 #define CC_PART1(Page, Char) \
34 ((combining_class_table_part1[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
35 ? (combining_class_table_part1[Page] - G_UNICODE_MAX_TABLE_INDEX) \
36 : (cclass_data[combining_class_table_part1[Page]][Char]))
38 #define CC_PART2(Page, Char) \
39 ((combining_class_table_part2[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
40 ? (combining_class_table_part2[Page] - G_UNICODE_MAX_TABLE_INDEX) \
41 : (cclass_data[combining_class_table_part2[Page]][Char]))
43 #define COMBINING_CLASS(Char) \
44 (((Char) <= G_UNICODE_LAST_CHAR_PART1) \
45 ? CC_PART1 ((Char) >> 8, (Char) & 0xff) \
46 : (((Char) >= 0xe0000 && (Char) <= G_UNICODE_LAST_CHAR) \
47 ? CC_PART2 (((Char) - 0xe0000) >> 8, (Char) & 0xff) \
51 _g_unichar_combining_class (gunichar uc)
53 return COMBINING_CLASS (uc);
56 /* constants for hangul syllable [de]composition */
64 #define NCount (VCount * TCount)
65 #define SCount (LCount * NCount)
68 * g_unicode_canonical_ordering:
69 * @string: a UCS-4 encoded string.
70 * @len: the maximum length of @string to use.
72 * Computes the canonical ordering of a string in-place.
73 * This rearranges decomposed characters in the string
74 * according to their combining classes. See the Unicode
75 * manual for more information.
78 g_unicode_canonical_ordering (gunichar *string,
88 last = COMBINING_CLASS (string[0]);
89 for (i = 0; i < len - 1; ++i)
91 int next = COMBINING_CLASS (string[i + 1]);
92 if (next != 0 && last > next)
95 /* Percolate item leftward through string. */
96 for (j = i + 1; j > 0; --j)
99 if (COMBINING_CLASS (string[j - 1]) <= next)
102 string[j] = string[j - 1];
106 /* We're re-entering the loop looking at the old
115 /* http://www.unicode.org/unicode/reports/tr15/#Hangul
116 * r should be null or have sufficient space. Calling with r == NULL will
117 * only calculate the result_len; however, a buffer with space for three
118 * characters will always be big enough. */
120 decompose_hangul (gunichar s,
124 gint SIndex = s - SBase;
126 /* not a hangul syllable */
127 if (SIndex < 0 || SIndex >= SCount)
135 gunichar L = LBase + SIndex / NCount;
136 gunichar V = VBase + (SIndex % NCount) / TCount;
137 gunichar T = TBase + SIndex % TCount;
156 /* returns a pointer to a null-terminated UTF-8 string */
158 find_decomposition (gunichar ch,
162 int end = G_N_ELEMENTS (decomp_table);
164 if (ch >= decomp_table[start].ch &&
165 ch <= decomp_table[end - 1].ch)
169 int half = (start + end) / 2;
170 if (ch == decomp_table[half].ch)
176 offset = decomp_table[half].compat_offset;
177 if (offset == G_UNICODE_NOT_PRESENT_OFFSET)
178 offset = decomp_table[half].canon_offset;
182 offset = decomp_table[half].canon_offset;
183 if (offset == G_UNICODE_NOT_PRESENT_OFFSET)
187 return &(decomp_expansion_string[offset]);
189 else if (half == start)
191 else if (ch > decomp_table[half].ch)
202 * g_unicode_canonical_decomposition:
203 * @ch: a Unicode character.
204 * @result_len: location to store the length of the return value.
206 * Computes the canonical decomposition of a Unicode character.
208 * Return value: a newly allocated string of Unicode characters.
209 * @result_len is set to the resulting length of the string.
212 g_unicode_canonical_decomposition (gunichar ch,
219 /* Hangul syllable */
220 if (ch >= 0xac00 && ch <= 0xd7a3)
222 decompose_hangul (ch, NULL, result_len);
223 r = g_malloc (*result_len * sizeof (gunichar));
224 decompose_hangul (ch, r, result_len);
226 else if ((decomp = find_decomposition (ch, FALSE)) != NULL)
231 *result_len = g_utf8_strlen (decomp, -1);
232 r = g_malloc (*result_len * sizeof (gunichar));
234 for (p = decomp, i = 0; *p != '\0'; p = g_utf8_next_char (p), i++)
235 r[i] = g_utf8_get_char (p);
239 /* Not in our table. */
240 r = g_malloc (sizeof (gunichar));
245 /* Supposedly following the Unicode 2.1.9 table means that the
246 decompositions come out in canonical order. I haven't tested
247 this, but we rely on it here. */
251 /* L,V => LV and LV,T => LVT */
253 combine_hangul (gunichar a,
257 gint LIndex = a - LBase;
258 gint SIndex = a - SBase;
260 gint VIndex = b - VBase;
261 gint TIndex = b - TBase;
263 if (0 <= LIndex && LIndex < LCount
264 && 0 <= VIndex && VIndex < VCount)
266 *result = SBase + (LIndex * VCount + VIndex) * TCount;
269 else if (0 <= SIndex && SIndex < SCount && (SIndex % TCount) == 0
270 && 0 < TIndex && TIndex < TCount)
272 *result = a + TIndex;
279 #define CI(Page, Char) \
280 ((compose_table[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
281 ? (compose_table[Page] - G_UNICODE_MAX_TABLE_INDEX) \
282 : (compose_data[compose_table[Page]][Char]))
284 #define COMPOSE_INDEX(Char) \
285 (((Char >> 8) > (COMPOSE_TABLE_LAST)) ? 0 : CI((Char) >> 8, (Char) & 0xff))
292 gushort index_a, index_b;
294 if (combine_hangul (a, b, result))
297 index_a = COMPOSE_INDEX(a);
299 if (index_a >= COMPOSE_FIRST_SINGLE_START && index_a < COMPOSE_SECOND_START)
301 if (b == compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][0])
303 *result = compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][1];
310 index_b = COMPOSE_INDEX(b);
312 if (index_b >= COMPOSE_SECOND_SINGLE_START)
314 if (a == compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][0])
316 *result = compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][1];
323 if (index_a >= COMPOSE_FIRST_START && index_a < COMPOSE_FIRST_SINGLE_START &&
324 index_b >= COMPOSE_SECOND_START && index_b < COMPOSE_SECOND_SINGLE_START)
326 gunichar res = compose_array[index_a - COMPOSE_FIRST_START][index_b - COMPOSE_SECOND_START];
339 _g_utf8_normalize_wc (const gchar *str,
347 gboolean do_compat = (mode == G_NORMALIZE_NFKC ||
348 mode == G_NORMALIZE_NFKD);
349 gboolean do_compose = (mode == G_NORMALIZE_NFC ||
350 mode == G_NORMALIZE_NFKC);
354 while ((max_len < 0 || p < str + max_len) && *p)
357 gunichar wc = g_utf8_get_char (p);
359 if (wc >= 0xac00 && wc <= 0xd7a3)
362 decompose_hangul (wc, NULL, &result_len);
367 decomp = find_decomposition (wc, do_compat);
370 n_wc += g_utf8_strlen (decomp, -1);
375 p = g_utf8_next_char (p);
378 wc_buffer = g_new (gunichar, n_wc + 1);
383 while ((max_len < 0 || p < str + max_len) && *p)
385 gunichar wc = g_utf8_get_char (p);
388 gsize old_n_wc = n_wc;
390 if (wc >= 0xac00 && wc <= 0xd7a3)
393 decompose_hangul (wc, wc_buffer + n_wc, &result_len);
398 decomp = find_decomposition (wc, do_compat);
403 for (pd = decomp; *pd != '\0'; pd = g_utf8_next_char (pd))
404 wc_buffer[n_wc++] = g_utf8_get_char (pd);
407 wc_buffer[n_wc++] = wc;
412 cc = COMBINING_CLASS (wc_buffer[old_n_wc]);
416 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
417 last_start = old_n_wc;
421 p = g_utf8_next_char (p);
426 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
432 /* All decomposed and reordered */
434 if (do_compose && n_wc > 0)
440 for (i = 0; i < n_wc; i++)
442 int cc = COMBINING_CLASS (wc_buffer[i]);
445 (last_cc == 0 || last_cc != cc) &&
446 combine (wc_buffer[last_start], wc_buffer[i],
447 &wc_buffer[last_start]))
449 for (j = i + 1; j < n_wc; j++)
450 wc_buffer[j-1] = wc_buffer[j];
457 last_cc = COMBINING_CLASS (wc_buffer[i-1]);
476 * @str: a UTF-8 encoded string.
477 * @len: length of @str, in bytes, or -1 if @str is nul-terminated.
478 * @mode: the type of normalization to perform.
480 * Converts a string into canonical form, standardizing
481 * such issues as whether a character with an accent
482 * is represented as a base character and combining
483 * accent or as a single precomposed character. You
484 * should generally call g_utf8_normalize() before
485 * comparing two Unicode strings.
487 * The normalization mode %G_NORMALIZE_DEFAULT only
488 * standardizes differences that do not affect the
489 * text content, such as the above-mentioned accent
490 * representation. %G_NORMALIZE_ALL also standardizes
491 * the "compatibility" characters in Unicode, such
492 * as SUPERSCRIPT THREE to the standard forms
493 * (in this case DIGIT THREE). Formatting information
494 * may be lost but for most text operations such
495 * characters should be considered the same.
496 * For example, g_utf8_collate() normalizes
497 * with %G_NORMALIZE_ALL as its first step.
499 * %G_NORMALIZE_DEFAULT_COMPOSE and %G_NORMALIZE_ALL_COMPOSE
500 * are like %G_NORMALIZE_DEFAULT and %G_NORMALIZE_ALL,
501 * but returned a result with composed forms rather
502 * than a maximally decomposed form. This is often
503 * useful if you intend to convert the string to
504 * a legacy encoding or pass it to a system with
505 * less capable Unicode handling.
507 * Return value: a newly allocated string, that is the
508 * normalized form of @str.
511 g_utf8_normalize (const gchar *str,
515 gunichar *result_wc = _g_utf8_normalize_wc (str, len, mode);
518 result = g_ucs4_to_utf8 (result_wc, -1, NULL, NULL, NULL);
524 #define __G_UNIDECOMP_C__
525 #include "galiasdef.c"