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(Page, Char) \
32 ((combining_class_table[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
33 ? (combining_class_table[Page] - G_UNICODE_MAX_TABLE_INDEX) \
34 : (cclass_data[combining_class_table[Page]][Char]))
36 #define COMBINING_CLASS(Char) \
37 (((Char) > (G_UNICODE_LAST_CHAR)) ? 0 : CC((Char) >> 8, (Char) & 0xff))
40 * g_unicode_canonical_ordering:
41 * @string: a UCS-4 encoded string.
42 * @len: the maximum length of @string to use.
44 * Computes the canonical ordering of a string in-place.
45 * This rearranges decomposed characters in the string
46 * according to their combining classes. See the Unicode
47 * manual for more information.
50 g_unicode_canonical_ordering (gunichar *string,
60 last = COMBINING_CLASS (string[0]);
61 for (i = 0; i < len - 1; ++i)
63 int next = COMBINING_CLASS (string[i + 1]);
64 if (next != 0 && last > next)
67 /* Percolate item leftward through string. */
68 for (j = i + 1; j > 0; --j)
71 if (COMBINING_CLASS (string[j - 1]) <= next)
74 string[j] = string[j - 1];
78 /* We're re-entering the loop looking at the old
88 find_decomposition (gunichar ch,
92 int end = G_N_ELEMENTS (decomp_table);
94 if (ch >= decomp_table[start].ch &&
95 ch <= decomp_table[end - 1].ch)
99 int half = (start + end) / 2;
100 if (ch == decomp_table[half].ch)
106 offset = decomp_table[half].compat_offset;
108 offset = decomp_table[half].canon_offset;
112 offset = decomp_table[half].canon_offset;
117 return &(decomp_expansion_string[decomp_table[half].expansion_offset + offset]);
119 else if (half == start)
121 else if (ch > decomp_table[half].ch)
132 * g_unicode_canonical_decomposition:
133 * @ch: a Unicode character.
134 * @result_len: location to store the length of the return value.
136 * Computes the canonical decomposition of a Unicode character.
138 * Return value: a newly allocated string of Unicode characters.
139 * @result_len is set to the resulting length of the string.
142 g_unicode_canonical_decomposition (gunichar ch,
145 const guchar *decomp = find_decomposition (ch, FALSE);
152 /* We store as a double-nul terminated string. */
153 for (len = 0; (decomp[len] || decomp[len + 1]);
157 /* We've counted twice as many bytes as there are
159 *result_len = len / 2;
160 r = g_malloc (len / 2 * sizeof (gunichar));
162 for (i = 0; i < len; i += 2)
164 r[i / 2] = (decomp[i] << 8 | decomp[i + 1]);
169 /* Not in our table. */
170 r = g_malloc (sizeof (gunichar));
175 /* Supposedly following the Unicode 2.1.9 table means that the
176 decompositions come out in canonical order. I haven't tested
177 this, but we rely on it here. */
181 #define CI(Page, Char) \
182 ((compose_table[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
183 ? (compose_table[Page] - G_UNICODE_MAX_TABLE_INDEX) \
184 : (compose_data[compose_table[Page]][Char]))
186 #define COMPOSE_INDEX(Char) \
187 (((Char) > (G_UNICODE_LAST_CHAR)) ? 0 : CI((Char) >> 8, (Char) & 0xff))
194 gushort index_a, index_b;
196 index_a = COMPOSE_INDEX(a);
197 if (index_a >= COMPOSE_FIRST_SINGLE_START && index_a < COMPOSE_SECOND_START)
199 if (b == compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][0])
201 *result = compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][1];
208 index_b = COMPOSE_INDEX(b);
209 if (index_b >= COMPOSE_SECOND_SINGLE_START)
211 if (a == compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][0])
213 *result = compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][1];
220 if (index_a >= COMPOSE_FIRST_START && index_a < COMPOSE_FIRST_SINGLE_START &&
221 index_b >= COMPOSE_SECOND_START && index_a < COMPOSE_SECOND_SINGLE_START)
223 gunichar res = compose_array[index_a - COMPOSE_FIRST_START][index_b - COMPOSE_SECOND_START];
236 _g_utf8_normalize_wc (const gchar *str,
244 gboolean do_compat = (mode == G_NORMALIZE_NFKC ||
245 mode == G_NORMALIZE_NFKD);
246 gboolean do_compose = (mode == G_NORMALIZE_NFC ||
247 mode == G_NORMALIZE_NFKC);
251 while ((max_len < 0 || p < str + max_len) && *p)
253 gunichar wc = g_utf8_get_char (p);
255 const guchar *decomp = find_decomposition (wc, do_compat);
260 /* We store as a double-nul terminated string. */
261 for (len = 0; (decomp[len] || decomp[len + 1]);
269 p = g_utf8_next_char (p);
272 wc_buffer = g_new (gunichar, n_wc + 1);
277 while ((max_len < 0 || p < str + max_len) && *p)
279 gunichar wc = g_utf8_get_char (p);
280 const guchar *decomp;
282 gsize old_n_wc = n_wc;
284 decomp = find_decomposition (wc, do_compat);
289 /* We store as a double-nul terminated string. */
290 for (len = 0; (decomp[len] || decomp[len + 1]);
292 wc_buffer[n_wc++] = (decomp[len] << 8 | decomp[len + 1]);
295 wc_buffer[n_wc++] = wc;
299 cc = COMBINING_CLASS (wc_buffer[old_n_wc]);
303 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
304 last_start = old_n_wc;
308 p = g_utf8_next_char (p);
313 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
319 /* All decomposed and reordered */
322 if (do_compose && n_wc > 0)
328 for (i = 0; i < n_wc; i++)
330 int cc = COMBINING_CLASS (wc_buffer[i]);
333 (last_cc == 0 || last_cc != cc) &&
334 combine (wc_buffer[last_start], wc_buffer[i],
335 &wc_buffer[last_start]))
337 for (j = i + 1; j < n_wc; j++)
338 wc_buffer[j-1] = wc_buffer[j];
345 last_cc = COMBINING_CLASS (wc_buffer[i-1]);
364 * @str: a UTF-8 encoded string.
365 * @len: length of @str, in bytes, or -1 if @str is nul-terminated.
366 * @mode: the type of normalization to perform.
368 * Converts a string into canonical form, standardizing
369 * such issues as whether a character with an accent
370 * is represented as a base character and combining
371 * accent or as a single precomposed character. You
372 * should generally call g_utf8_normalize() before
373 * comparing two Unicode strings.
375 * The normalization mode %G_NORMALIZE_DEFAULT only
376 * standardizes differences that do not affect the
377 * text content, such as the above-mentioned accent
378 * representation. %G_NORMALIZE_ALL also standardizes
379 * the "compatibility" characters in Unicode, such
380 * as SUPERSCRIPT THREE to the standard forms
381 * (in this case DIGIT THREE). Formatting information
382 * may be lost but for most text operations such
383 * characters should be considered the same.
384 * For example, g_utf8_collate() normalizes
385 * with %G_NORMALIZE_ALL as its first step.
387 * %G_NORMALIZE_DEFAULT_COMPOSE and %G_NORMALIZE_ALL_COMPOSE
388 * are like %G_NORMALIZE_DEFAULT and %G_NORMALIZE_ALL,
389 * but returned a result with composed forms rather
390 * than a maximally decomposed form. This is often
391 * useful if you intend to convert the string to
392 * a legacy encoding or pass it to a system with
393 * less capable Unicode handling.
395 * Return value: a newly allocated string, that is the
396 * normalized form of @str.
399 g_utf8_normalize (const gchar *str,
403 gunichar *result_wc = _g_utf8_normalize_wc (str, len, mode);
406 result = g_ucs4_to_utf8 (result_wc, -1, NULL, NULL, NULL);