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"
32 #define CC_PART1(Page, Char) \
33 ((combining_class_table_part1[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
34 ? (combining_class_table_part1[Page] - G_UNICODE_MAX_TABLE_INDEX) \
35 : (cclass_data[combining_class_table_part1[Page]][Char]))
37 #define CC_PART2(Page, Char) \
38 ((combining_class_table_part2[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
39 ? (combining_class_table_part2[Page] - G_UNICODE_MAX_TABLE_INDEX) \
40 : (cclass_data[combining_class_table_part2[Page]][Char]))
42 #define COMBINING_CLASS(Char) \
43 (((Char) <= G_UNICODE_LAST_CHAR_PART1) \
44 ? CC_PART1 ((Char) >> 8, (Char) & 0xff) \
45 : (((Char) >= 0xe0000 && (Char) <= G_UNICODE_LAST_CHAR) \
46 ? CC_PART2 (((Char) - 0xe0000) >> 8, (Char) & 0xff) \
50 _g_unichar_combining_class (gunichar uc)
52 return COMBINING_CLASS (uc);
55 /* constants for hangul syllable [de]composition */
63 #define NCount (VCount * TCount)
64 #define SCount (LCount * NCount)
67 * g_unicode_canonical_ordering:
68 * @string: a UCS-4 encoded string.
69 * @len: the maximum length of @string to use.
71 * Computes the canonical ordering of a string in-place.
72 * This rearranges decomposed characters in the string
73 * according to their combining classes. See the Unicode
74 * manual for more information.
77 g_unicode_canonical_ordering (gunichar *string,
87 last = COMBINING_CLASS (string[0]);
88 for (i = 0; i < len - 1; ++i)
90 int next = COMBINING_CLASS (string[i + 1]);
91 if (next != 0 && last > next)
94 /* Percolate item leftward through string. */
95 for (j = i + 1; j > 0; --j)
98 if (COMBINING_CLASS (string[j - 1]) <= next)
101 string[j] = string[j - 1];
105 /* We're re-entering the loop looking at the old
114 /* http://www.unicode.org/unicode/reports/tr15/#Hangul
115 * r should be null or have sufficient space. Calling with r == NULL will
116 * only calculate the result_len; however, a buffer with space for three
117 * characters will always be big enough. */
119 decompose_hangul (gunichar s,
123 gint SIndex = s - SBase;
125 /* not a hangul syllable */
126 if (SIndex < 0 || SIndex >= SCount)
134 gunichar L = LBase + SIndex / NCount;
135 gunichar V = VBase + (SIndex % NCount) / TCount;
136 gunichar T = TBase + SIndex % TCount;
155 /* returns a pointer to a null-terminated UTF-8 string */
157 find_decomposition (gunichar ch,
161 int end = G_N_ELEMENTS (decomp_table);
163 if (ch >= decomp_table[start].ch &&
164 ch <= decomp_table[end - 1].ch)
168 int half = (start + end) / 2;
169 if (ch == decomp_table[half].ch)
175 offset = decomp_table[half].compat_offset;
176 if (offset == G_UNICODE_NOT_PRESENT_OFFSET)
177 offset = decomp_table[half].canon_offset;
181 offset = decomp_table[half].canon_offset;
182 if (offset == G_UNICODE_NOT_PRESENT_OFFSET)
186 return &(decomp_expansion_string[offset]);
188 else if (half == start)
190 else if (ch > decomp_table[half].ch)
201 * g_unicode_canonical_decomposition:
202 * @ch: a Unicode character.
203 * @result_len: location to store the length of the return value.
205 * Computes the canonical decomposition of a Unicode character.
207 * Return value: a newly allocated string of Unicode characters.
208 * @result_len is set to the resulting length of the string.
211 g_unicode_canonical_decomposition (gunichar ch,
218 /* Hangul syllable */
219 if (ch >= 0xac00 && ch <= 0xd7af)
221 decompose_hangul (ch, NULL, result_len);
222 r = g_malloc (*result_len * sizeof (gunichar));
223 decompose_hangul (ch, r, result_len);
225 else if ((decomp = find_decomposition (ch, FALSE)) != NULL)
230 *result_len = g_utf8_strlen (decomp, -1);
231 r = g_malloc (*result_len * sizeof (gunichar));
233 for (p = decomp, i = 0; *p != '\0'; p = g_utf8_next_char (p), i++)
234 r[i] = g_utf8_get_char (p);
238 /* Not in our table. */
239 r = g_malloc (sizeof (gunichar));
244 /* Supposedly following the Unicode 2.1.9 table means that the
245 decompositions come out in canonical order. I haven't tested
246 this, but we rely on it here. */
250 /* L,V => LV and LV,T => LVT */
252 combine_hangul (gunichar a,
256 gint LIndex = a - LBase;
257 gint SIndex = a - SBase;
259 gint VIndex = b - VBase;
260 gint TIndex = b - TBase;
262 if (0 <= LIndex && LIndex < LCount
263 && 0 <= VIndex && VIndex < VCount)
265 *result = SBase + (LIndex * VCount + VIndex) * TCount;
268 else if (0 <= SIndex && SIndex < SCount && (SIndex % TCount) == 0
269 && 0 <= TIndex && TIndex <= TCount)
271 *result = a + TIndex;
278 #define CI(Page, Char) \
279 ((compose_table[Page] >= G_UNICODE_MAX_TABLE_INDEX) \
280 ? (compose_table[Page] - G_UNICODE_MAX_TABLE_INDEX) \
281 : (compose_data[compose_table[Page]][Char]))
283 #define COMPOSE_INDEX(Char) \
284 (((Char >> 8) > (COMPOSE_TABLE_LAST)) ? 0 : CI((Char) >> 8, (Char) & 0xff))
291 gushort index_a, index_b;
293 if (combine_hangul (a, b, result))
296 index_a = COMPOSE_INDEX(a);
298 if (index_a >= COMPOSE_FIRST_SINGLE_START && index_a < COMPOSE_SECOND_START)
300 if (b == compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][0])
302 *result = compose_first_single[index_a - COMPOSE_FIRST_SINGLE_START][1];
309 index_b = COMPOSE_INDEX(b);
311 if (index_b >= COMPOSE_SECOND_SINGLE_START)
313 if (a == compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][0])
315 *result = compose_second_single[index_b - COMPOSE_SECOND_SINGLE_START][1];
322 if (index_a >= COMPOSE_FIRST_START && index_a < COMPOSE_FIRST_SINGLE_START &&
323 index_b >= COMPOSE_SECOND_START && index_b < COMPOSE_SECOND_SINGLE_START)
325 gunichar res = compose_array[index_a - COMPOSE_FIRST_START][index_b - COMPOSE_SECOND_START];
338 _g_utf8_normalize_wc (const gchar *str,
346 gboolean do_compat = (mode == G_NORMALIZE_NFKC ||
347 mode == G_NORMALIZE_NFKD);
348 gboolean do_compose = (mode == G_NORMALIZE_NFC ||
349 mode == G_NORMALIZE_NFKC);
353 while ((max_len < 0 || p < str + max_len) && *p)
356 gunichar wc = g_utf8_get_char (p);
358 if (wc >= 0xac00 && wc <= 0xd7af)
361 decompose_hangul (wc, NULL, &result_len);
366 decomp = find_decomposition (wc, do_compat);
369 n_wc += g_utf8_strlen (decomp, -1);
374 p = g_utf8_next_char (p);
377 wc_buffer = g_new (gunichar, n_wc + 1);
382 while ((max_len < 0 || p < str + max_len) && *p)
384 gunichar wc = g_utf8_get_char (p);
387 gsize old_n_wc = n_wc;
389 if (wc >= 0xac00 && wc <= 0xd7af)
392 decompose_hangul (wc, wc_buffer + n_wc, &result_len);
397 decomp = find_decomposition (wc, do_compat);
402 for (pd = decomp; *pd != '\0'; pd = g_utf8_next_char (pd))
403 wc_buffer[n_wc++] = g_utf8_get_char (pd);
406 wc_buffer[n_wc++] = wc;
411 cc = COMBINING_CLASS (wc_buffer[old_n_wc]);
415 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
416 last_start = old_n_wc;
420 p = g_utf8_next_char (p);
425 g_unicode_canonical_ordering (wc_buffer + last_start, n_wc - last_start);
431 /* All decomposed and reordered */
433 if (do_compose && n_wc > 0)
439 for (i = 0; i < n_wc; i++)
441 int cc = COMBINING_CLASS (wc_buffer[i]);
444 (last_cc == 0 || last_cc != cc) &&
445 combine (wc_buffer[last_start], wc_buffer[i],
446 &wc_buffer[last_start]))
448 for (j = i + 1; j < n_wc; j++)
449 wc_buffer[j-1] = wc_buffer[j];
456 last_cc = COMBINING_CLASS (wc_buffer[i-1]);
475 * @str: a UTF-8 encoded string.
476 * @len: length of @str, in bytes, or -1 if @str is nul-terminated.
477 * @mode: the type of normalization to perform.
479 * Converts a string into canonical form, standardizing
480 * such issues as whether a character with an accent
481 * is represented as a base character and combining
482 * accent or as a single precomposed character. You
483 * should generally call g_utf8_normalize() before
484 * comparing two Unicode strings.
486 * The normalization mode %G_NORMALIZE_DEFAULT only
487 * standardizes differences that do not affect the
488 * text content, such as the above-mentioned accent
489 * representation. %G_NORMALIZE_ALL also standardizes
490 * the "compatibility" characters in Unicode, such
491 * as SUPERSCRIPT THREE to the standard forms
492 * (in this case DIGIT THREE). Formatting information
493 * may be lost but for most text operations such
494 * characters should be considered the same.
495 * For example, g_utf8_collate() normalizes
496 * with %G_NORMALIZE_ALL as its first step.
498 * %G_NORMALIZE_DEFAULT_COMPOSE and %G_NORMALIZE_ALL_COMPOSE
499 * are like %G_NORMALIZE_DEFAULT and %G_NORMALIZE_ALL,
500 * but returned a result with composed forms rather
501 * than a maximally decomposed form. This is often
502 * useful if you intend to convert the string to
503 * a legacy encoding or pass it to a system with
504 * less capable Unicode handling.
506 * Return value: a newly allocated string, that is the
507 * normalized form of @str.
510 g_utf8_normalize (const gchar *str,
514 gunichar *result_wc = _g_utf8_normalize_wc (str, len, mode);
517 result = g_ucs4_to_utf8 (result_wc, -1, NULL, NULL, NULL);