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"
31 #define CC_PART1(Page, Char) \
32 ((combining_class_table_part1[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
33 ? (combining_class_table_part1[Page] - G_UNICODE_MAX_TABLE_INDEX) \
34 : (cclass_data[combining_class_table_part1[Page]][Char]))
36 #define CC_PART2(Page, Char) \
37 ((combining_class_table_part2[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
38 ? (combining_class_table_part2[Page] - G_UNICODE_MAX_TABLE_INDEX) \
39 : (cclass_data[combining_class_table_part2[Page]][Char]))
41 #define COMBINING_CLASS(Char) \
42 (((Char) <= G_UNICODE_LAST_CHAR_PART1) \
43 ? CC_PART1 ((Char) >> 8, (Char) & 0xff) \
44 : (((Char) >= 0xe0000 && (Char) <= G_UNICODE_LAST_CHAR) \
45 ? CC_PART2 (((Char) - 0xe0000) >> 8, (Char) & 0xff) \
49 * g_unicode_canonical_ordering:
50 * @string: a UCS-4 encoded string.
51 * @len: the maximum length of @string to use.
53 * Computes the canonical ordering of a string in-place.
54 * This rearranges decomposed characters in the string
55 * according to their combining classes. See the Unicode
56 * manual for more information.
59 g_unicode_canonical_ordering (gunichar *string,
69 last = COMBINING_CLASS (string[0]);
70 for (i = 0; i < len - 1; ++i)
72 int next = COMBINING_CLASS (string[i + 1]);
73 if (next != 0 && last > next)
76 /* Percolate item leftward through string. */
77 for (j = i + 1; j > 0; --j)
80 if (COMBINING_CLASS (string[j - 1]) <= next)
83 string[j] = string[j - 1];
87 /* We're re-entering the loop looking at the old
96 /* returns a pointer to a null-terminated UTF-8 string */
98 find_decomposition (gunichar ch,
102 int end = G_N_ELEMENTS (decomp_table);
104 if (ch >= decomp_table[start].ch &&
105 ch <= decomp_table[end - 1].ch)
109 int half = (start + end) / 2;
110 if (ch == decomp_table[half].ch)
116 offset = decomp_table[half].compat_offset;
117 if (offset == G_UNICODE_NOT_PRESENT_OFFSET)
118 offset = decomp_table[half].canon_offset;
122 offset = decomp_table[half].canon_offset;
123 if (offset == G_UNICODE_NOT_PRESENT_OFFSET)
127 return &(decomp_expansion_string[offset]);
129 else if (half == start)
131 else if (ch > decomp_table[half].ch)
142 * g_unicode_canonical_decomposition:
143 * @ch: a Unicode character.
144 * @result_len: location to store the length of the return value.
146 * Computes the canonical decomposition of a Unicode character.
148 * Return value: a newly allocated string of Unicode characters.
149 * @result_len is set to the resulting length of the string.
152 g_unicode_canonical_decomposition (gunichar ch,
155 const gchar *decomp = find_decomposition (ch, FALSE);
164 *result_len = g_utf8_strlen (decomp, -1);
165 r = g_malloc (*result_len * sizeof (gunichar));
167 for (p = decomp, i = 0; *p != '\0'; p = g_utf8_next_char (p), i++)
168 r[i] = g_utf8_get_char (p);
172 /* Not in our table. */
173 r = g_malloc (sizeof (gunichar));
178 /* Supposedly following the Unicode 2.1.9 table means that the
179 decompositions come out in canonical order. I haven't tested
180 this, but we rely on it here. */
184 #define CI(Page, Char) \
185 ((compose_table[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
186 ? (compose_table[Page] - G_UNICODE_MAX_TABLE_INDEX) \
187 : (compose_data[compose_table[Page]][Char]))
189 #define COMPOSE_INDEX(Char) \
190 (((Char) > (G_UNICODE_LAST_CHAR)) ? 0 : CI((Char) >> 8, (Char) & 0xff))
197 gushort index_a, index_b;
199 index_a = COMPOSE_INDEX(a);
201 if (index_a >= COMPOSE_FIRST_SINGLE_START && index_a < COMPOSE_SECOND_START)
203 if (b == compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][0])
205 *result = compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][1];
212 index_b = COMPOSE_INDEX(b);
214 if (index_b >= COMPOSE_SECOND_SINGLE_START)
216 if (a == compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][0])
218 *result = compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][1];
225 if (index_a >= COMPOSE_FIRST_START && index_a < COMPOSE_FIRST_SINGLE_START &&
226 index_b >= COMPOSE_SECOND_START && index_b < COMPOSE_SECOND_SINGLE_START)
228 gunichar res = compose_array[index_a - COMPOSE_FIRST_START][index_b - COMPOSE_SECOND_START];
241 _g_utf8_normalize_wc (const gchar *str,
249 gboolean do_compat = (mode == G_NORMALIZE_NFKC ||
250 mode == G_NORMALIZE_NFKD);
251 gboolean do_compose = (mode == G_NORMALIZE_NFC ||
252 mode == G_NORMALIZE_NFKC);
256 while ((max_len < 0 || p < str + max_len) && *p)
258 gunichar wc = g_utf8_get_char (p);
260 const gchar *decomp = find_decomposition (wc, do_compat);
263 n_wc += g_utf8_strlen (decomp, -1);
267 p = g_utf8_next_char (p);
270 wc_buffer = g_new (gunichar, n_wc + 1);
275 while ((max_len < 0 || p < str + max_len) && *p)
277 gunichar wc = g_utf8_get_char (p);
280 gsize old_n_wc = n_wc;
282 decomp = find_decomposition (wc, do_compat);
287 for (pd = decomp; *pd != '\0'; pd = g_utf8_next_char (pd))
288 wc_buffer[n_wc++] = g_utf8_get_char (pd);
291 wc_buffer[n_wc++] = wc;
295 cc = COMBINING_CLASS (wc_buffer[old_n_wc]);
299 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
300 last_start = old_n_wc;
304 p = g_utf8_next_char (p);
309 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
315 /* All decomposed and reordered */
317 if (do_compose && n_wc > 0)
323 for (i = 0; i < n_wc; i++)
325 int cc = COMBINING_CLASS (wc_buffer[i]);
328 (last_cc == 0 || last_cc != cc) &&
329 combine (wc_buffer[last_start], wc_buffer[i],
330 &wc_buffer[last_start]))
332 for (j = i + 1; j < n_wc; j++)
333 wc_buffer[j-1] = wc_buffer[j];
340 last_cc = COMBINING_CLASS (wc_buffer[i-1]);
359 * @str: a UTF-8 encoded string.
360 * @len: length of @str, in bytes, or -1 if @str is nul-terminated.
361 * @mode: the type of normalization to perform.
363 * Converts a string into canonical form, standardizing
364 * such issues as whether a character with an accent
365 * is represented as a base character and combining
366 * accent or as a single precomposed character. You
367 * should generally call g_utf8_normalize() before
368 * comparing two Unicode strings.
370 * The normalization mode %G_NORMALIZE_DEFAULT only
371 * standardizes differences that do not affect the
372 * text content, such as the above-mentioned accent
373 * representation. %G_NORMALIZE_ALL also standardizes
374 * the "compatibility" characters in Unicode, such
375 * as SUPERSCRIPT THREE to the standard forms
376 * (in this case DIGIT THREE). Formatting information
377 * may be lost but for most text operations such
378 * characters should be considered the same.
379 * For example, g_utf8_collate() normalizes
380 * with %G_NORMALIZE_ALL as its first step.
382 * %G_NORMALIZE_DEFAULT_COMPOSE and %G_NORMALIZE_ALL_COMPOSE
383 * are like %G_NORMALIZE_DEFAULT and %G_NORMALIZE_ALL,
384 * but returned a result with composed forms rather
385 * than a maximally decomposed form. This is often
386 * useful if you intend to convert the string to
387 * a legacy encoding or pass it to a system with
388 * less capable Unicode handling.
390 * Return value: a newly allocated string, that is the
391 * normalized form of @str.
394 g_utf8_normalize (const gchar *str,
398 gunichar *result_wc = _g_utf8_normalize_wc (str, len, mode);
401 result = g_ucs4_to_utf8 (result_wc, -1, NULL, NULL, NULL);