/* The input word in this method MUST be normalized in NFKD form,
* and given in UTF-8, where str_length is the byte-length */
gboolean
tracker_parser_unaccent_nfkd_string (gpointer str,
gsize *str_length)
{
gchar *word;
gsize word_length;
gsize i;
gsize j;
g_return_val_if_fail (str != NULL, FALSE);
g_return_val_if_fail (str_length != NULL, FALSE);
g_return_val_if_fail (*str_length > 0, FALSE);
word = (gchar *)str;
word_length = *str_length;
i = 0;
j = 0;
while (i < word_length) {
ucs4_t unichar;
gint utf8_len;
/* Get next character of the word as UCS4 */
utf8_len = u8_strmbtouc (&unichar, &word[i]);
/* Invalid UTF-8 character or end of original string. */
if (utf8_len <= 0) {
break;
}
/* If the given unichar is a combining diacritical mark,
* just update the original index, not the output one */
if (IS_CDM_UCS4 ((guint32) unichar)) {
i += utf8_len;
continue;
}
/* If already found a previous combining
* diacritical mark, indexes are different so
* need to copy characters. As output and input
* buffers may overlap, need to use memmove
* instead of memcpy */
if (i != j) {
memmove (&word[j], &word[i], utf8_len);
}
/* Update both indexes */
i += utf8_len;
j += utf8_len;
}
/* Force proper string end */
word[j] = '\0';
/* Set new output length */
*str_length = j;
return TRUE;
}
/* Copied from tracker/src/libtracker-fts/tracker-parser-glib.c under the GPL
* And then from gnome-shell/src/shell-util.c
*
* Originally written by Aleksander Morgado <*****@*****.**>
*/
char *
cc_util_normalize_casefold_and_unaccent (const char *str)
{
char *normalized, *tmp;
int i = 0, j = 0, ilen;
if (str == NULL)
return NULL;
normalized = g_utf8_normalize (str, -1, G_NORMALIZE_NFKD);
tmp = g_utf8_casefold (normalized, -1);
g_free (normalized);
ilen = strlen (tmp);
while (i < ilen)
{
gunichar unichar;
gchar *next_utf8;
gint utf8_len;
/* Get next character of the word as UCS4 */
unichar = g_utf8_get_char_validated (&tmp[i], -1);
/* Invalid UTF-8 character or end of original string. */
if (unichar == (gunichar) -1 ||
unichar == (gunichar) -2)
{
break;
}
/* Find next UTF-8 character */
next_utf8 = g_utf8_next_char (&tmp[i]);
utf8_len = next_utf8 - &tmp[i];
if (IS_CDM_UCS4 ((guint32) unichar))
{
/* If the given unichar is a combining diacritical mark,
* just update the original index, not the output one */
i += utf8_len;
continue;
}
/* If already found a previous combining
* diacritical mark, indexes are different so
* need to copy characters. As output and input
* buffers may overlap, need to use memmove
* instead of memcpy */
if (i != j)
{
memmove (&tmp[j], &tmp[i], utf8_len);
}
/* Update both indexes */
i += utf8_len;
j += utf8_len;
}
/* Force proper string end */
tmp[j] = '\0';
return tmp;
}