*
* You should have received a copy of the GNU Lesser General Public
* License along with the Gnome Library; see the file COPYING.LIB. If not,
- * write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
- * Boston, MA 02111-1307, USA.
+ * see <http://www.gnu.org/licenses/>.
+ */
+
+/**
+ * SECTION:unicode
+ * @Title: Unicode Manipulation
+ * @Short_description: functions operating on Unicode characters and
+ * UTF-8 strings
+ * @See_also: g_locale_to_utf8(), g_locale_from_utf8()
+ *
+ * This section describes a number of functions for dealing with
+ * Unicode characters and strings. There are analogues of the
+ * traditional `ctype.h` character classification and case conversion
+ * functions, UTF-8 analogues of some string utility functions,
+ * functions to perform normalization, case conversion and collation
+ * on UTF-8 strings and finally functions to convert between the UTF-8,
+ * UTF-16 and UCS-4 encodings of Unicode.
+ *
+ * The implementations of the Unicode functions in GLib are based
+ * on the Unicode Character Data tables, which are available from
+ * [www.unicode.org](http://www.unicode.org/).
+ * GLib 2.8 supports Unicode 4.0, GLib 2.10 supports Unicode 4.1,
+ * GLib 2.12 supports Unicode 5.0, GLib 2.16.3 supports Unicode 5.1,
+ * GLib 2.30 supports Unicode 6.0.
*/
#include "config.h"
*
* Determines the canonical combining class of a Unicode character.
*
- * Return value: the combining class of the character
+ * Returns: the combining class of the character
*
* Since: 2.14
**/
* only calculate the result_len; however, a buffer with space for three
* characters will always be big enough. */
static void
-decompose_hangul (gunichar s,
+decompose_hangul (gunichar s,
gunichar *r,
gsize *result_len)
{
gint SIndex = s - SBase;
+ gint TIndex = SIndex % TCount;
- /* not a hangul syllable */
- if (SIndex < 0 || SIndex >= SCount)
+ if (r)
{
- if (r)
- r[0] = s;
- *result_len = 1;
+ r[0] = LBase + SIndex / NCount;
+ r[1] = VBase + (SIndex % NCount) / TCount;
}
- else
- {
- gunichar L = LBase + SIndex / NCount;
- gunichar V = VBase + (SIndex % NCount) / TCount;
- gunichar T = TBase + SIndex % TCount;
+ if (TIndex)
+ {
if (r)
- {
- r[0] = L;
- r[1] = V;
- }
-
- if (T != TBase)
- {
- if (r)
- r[2] = T;
- *result_len = 3;
- }
- else
- *result_len = 2;
+ r[2] = TBase + TIndex;
+ *result_len = 3;
}
+ else
+ *result_len = 2;
}
/* returns a pointer to a null-terminated UTF-8 string */
*
* Computes the canonical decomposition of a Unicode character.
*
- * Return value: a newly allocated string of Unicode characters.
+ * Returns: a newly allocated string of Unicode characters.
* @result_len is set to the resulting length of the string.
+ *
+ * Deprecated: 2.30: Use the more flexible g_unichar_fully_decompose()
+ * instead.
**/
gunichar *
g_unicode_canonical_decomposition (gunichar ch,
gunichar *r;
/* Hangul syllable */
- if (ch >= 0xac00 && ch <= 0xd7a3)
+ if (ch >= SBase && ch < SBase + SCount)
{
decompose_hangul (ch, NULL, result_len);
r = g_malloc (*result_len * sizeof (gunichar));
const gchar *decomp;
gunichar wc = g_utf8_get_char (p);
- if (wc >= 0xac00 && wc <= 0xd7a3)
+ if (wc >= SBase && wc < SBase + SCount)
{
gsize result_len;
decompose_hangul (wc, NULL, &result_len);
int cc;
gsize old_n_wc = n_wc;
- if (wc >= 0xac00 && wc <= 0xd7a3)
+ if (wc >= SBase && wc < SBase + SCount)
{
gsize result_len;
decompose_hangul (wc, wc_buffer + n_wc, &result_len);
* a legacy encoding or pass it to a system with
* less capable Unicode handling.
*
- * Return value: a newly allocated string, that is the
+ * Returns: a newly allocated string, that is the
* normalized form of @str, or %NULL if @str is not
* valid UTF-8.
**/
gunichar *a,
gunichar *b)
{
- gint SIndex;
- gunichar L, V, T;
+ gint SIndex, TIndex;
- SIndex = ch - SBase;
-
- if (SIndex < 0 || SIndex >= SCount)
+ if (ch < SBase || ch >= SBase + SCount)
return FALSE; /* not a hangul syllable */
- L = LBase + SIndex / NCount;
- V = VBase + (SIndex % NCount) / TCount;
- T = TBase + SIndex % TCount;
+ SIndex = ch - SBase;
+ TIndex = SIndex % TCount;
- if (T != TBase)
+ if (TIndex)
{
- gint LIndex, VIndex;
- gunichar LV;
-
/* split LVT -> LV,T */
- LIndex = L - LBase;
- VIndex = V - VBase;
- LV = SBase + (LIndex * VCount + VIndex) * TCount;
-
- *a = LV;
- *b = T;
+ *a = ch - TIndex;
+ *b = TBase + TIndex;
}
else
{
/* split LV -> L,V */
- *a = L;
- *b = V;
+ *a = LBase + SIndex / NCount;
+ *b = VBase + (SIndex % NCount) / TCount;
}
return TRUE;
}
-static gboolean
-compose_hangul_step (gunichar a,
- gunichar b,
- gunichar *ch)
-{
- gint LIndex, SIndex;
-
- /* first try L,V -> LV */
- LIndex = a - LBase;
- if (0 <= LIndex && LIndex < LCount)
- {
- gint VIndex;
-
- VIndex = b - VBase;
- if (0 <= VIndex && VIndex < VCount)
- {
- *ch = SBase + (LIndex * VCount + VIndex) * TCount;
- return TRUE;
- }
- }
-
- /* next try LV,T -> LVT */
- SIndex = a - SBase;
- if (0 <= SIndex && SIndex < SCount && (SIndex % TCount) == 0)
- {
- gint TIndex;
-
- TIndex = b - TBase;
- if (0 < TIndex && TIndex < TCount)
- {
- *ch = a + TIndex;
- return TRUE;
- }
- }
-
- return FALSE;
-}
-
/**
* g_unichar_decompose:
* @ch: a Unicode character
* If @ch is not decomposable, *@a is set to @ch and *@b
* is set to zero.
*
- * See <ulink url="http://unicode.org/reports/tr15/">UAX#15</ulink>
+ * Note that the way Unicode decomposition pairs are
+ * defined, it is guaranteed that @b would not decompose
+ * further, but @a may itself decompose. To get the full
+ * canonical decomposition for @ch, one would need to
+ * recursively call this function on @a. Or use
+ * g_unichar_fully_decompose().
+ *
+ * See
+ * [UAX#15](http://unicode.org/reports/tr15/)
* for details.
*
* Returns: %TRUE if the character could be decomposed
if (decompose_hangul_step (ch, a, b))
return TRUE;
+ /* TODO use bsearch() */
if (ch >= decomp_step_table[start].ch &&
ch <= decomp_step_table[end - 1].ch)
{
* Performs a single composition step of the
* Unicode canonical composition algorithm.
*
- * This function does not perform algorithmic composition
- * for Hangul characters, and does not include compatibility
- * compositions. It does, however, include 'singleton'
- * compositions which replace a character by a single
- * other character. To obtain these, pass zero for @b.
- *
- * This function includes algorithmic Hangul Jamo composition.
+ * This function includes algorithmic Hangul Jamo composition,
+ * but it is not exactly the inverse of g_unichar_decompose().
+ * No composition can have either of @a or @b equal to zero.
+ * To be precise, this function composes if and only if
+ * there exists a Primary Composite P which is canonically
+ * equivalent to the sequence <@a,@b>. See the Unicode
+ * Standard for the definition of Primary Composite.
*
* If @a and @b do not compose a new character, @ch is set to zero.
*
- * See <ulink url="http://unicode.org/reports/tr15/">UAX#15</ulink>
+ * See
+ * [UAX#15](http://unicode.org/reports/tr15/)
* for details.
*
* Returns: %TRUE if the characters could be composed
gunichar b,
gunichar *ch)
{
- gint start = 0;
- gint end = G_N_ELEMENTS (comp_step_table);
-
- if (compose_hangul_step (a, b, ch))
+ if (combine (a, b, ch))
return TRUE;
- if (a >= comp_step_table[start].a &&
- a <= comp_step_table[end - 1].a)
+ *ch = 0;
+ return FALSE;
+}
+
+/**
+ * g_unichar_fully_decompose:
+ * @ch: a Unicode character.
+ * @compat: whether perform canonical or compatibility decomposition
+ * @result: (allow-none): location to store decomposed result, or %NULL
+ * @result_len: length of @result
+ *
+ * Computes the canonical or compatibility decomposition of a
+ * Unicode character. For compatibility decomposition,
+ * pass %TRUE for @compat; for canonical decomposition
+ * pass %FALSE for @compat.
+ *
+ * The decomposed sequence is placed in @result. Only up to
+ * @result_len characters are written into @result. The length
+ * of the full decomposition (irrespective of @result_len) is
+ * returned by the function. For canonical decomposition,
+ * currently all decompositions are of length at most 4, but
+ * this may change in the future (very unlikely though).
+ * At any rate, Unicode does guarantee that a buffer of length
+ * 18 is always enough for both compatibility and canonical
+ * decompositions, so that is the size recommended. This is provided
+ * as %G_UNICHAR_MAX_DECOMPOSITION_LENGTH.
+ *
+ * See
+ * [UAX#15](http://unicode.org/reports/tr15/)
+ * for details.
+ *
+ * Returns: the length of the full decomposition.
+ *
+ * Since: 2.30
+ **/
+gsize
+g_unichar_fully_decompose (gunichar ch,
+ gboolean compat,
+ gunichar *result,
+ gsize result_len)
+{
+ const gchar *decomp;
+ const gchar *p;
+
+ /* Hangul syllable */
+ if (ch >= SBase && ch < SBase + SCount)
{
- while (TRUE)
- {
- gint half = (start + end) / 2;
- const decomposition_step *p = &(comp_step_table[half]);
- if (a == p->a && b == p->b)
- {
- *ch = p->ch;
- return TRUE;
- }
- else if (half == start)
- break;
- else if (a > p->a || (a == p->a && b > p->b))
- start = half;
- else
- end = half;
- }
+ gsize len, i;
+ gunichar buffer[3];
+ decompose_hangul (ch, result ? buffer : NULL, &len);
+ if (result)
+ for (i = 0; i < len && i < result_len; i++)
+ result[i] = buffer[i];
+ return len;
}
+ else if ((decomp = find_decomposition (ch, compat)) != NULL)
+ {
+ /* Found it. */
+ gsize len, i;
- *ch = 0;
+ len = g_utf8_strlen (decomp, -1);
- return FALSE;
+ for (p = decomp, i = 0; i < len && i < result_len; p = g_utf8_next_char (p), i++)
+ result[i] = g_utf8_get_char (p);
+
+ return len;
+ }
+
+ /* Does not decompose */
+ if (result && result_len >= 1)
+ *result = ch;
+ return 1;
}