/* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2013 Index Data
+ * Copyright (C) Index Data
* See the file LICENSE for details.
*/
int token_count;
size_t org_start;
size_t org_len;
+ size_t utf8_base;
+ size_t utf16_base;
struct icu_chain_step *steps;
};
icu_buf_utf16_copy(iter->org, src);
iter->token_count = 0;
iter->org_start = 0;
+ iter->utf8_base = iter->utf16_base = 0;
iter->org_len = src->utf16_len;
iter->last = icu_iter_invoke(iter, iter->steps, src);
}
void icu_iter_get_org_info(yaz_icu_iter_t iter, size_t *start, size_t *len)
{
- /* save full length of org since we're gonna cut it */
- int32_t save_len = iter->org->utf16_len;
-
- struct icu_buf_utf8 *tmp = icu_buf_utf8_create(0);
+ int32_t len1 = 0, len2 = 0;
UErrorCode status = U_ZERO_ERROR;
- iter->org->utf16_len = iter->org_start;
- icu_utf16_to_utf8(tmp, iter->org, &status);
- if (U_SUCCESS(status))
- *start = tmp->utf8_len;
- else
- *start = 0;
- iter->org->utf16_len = iter->org_start + iter->org_len;
- icu_utf16_to_utf8(tmp, iter->org, &status);
- if (U_SUCCESS(status))
- *len = tmp->utf8_len - *start;
- else
- *len = 0;
- icu_buf_utf8_destroy(tmp);
- iter->org->utf16_len = save_len;
+ if (iter->org_start < iter->utf16_base)
+ {
+ iter->utf8_base = 0;
+ iter->utf16_base = 0;
+ }
+ u_strToUTF8(0, 0, &len1,
+ iter->org->utf16 + iter->utf16_base,
+ iter->org_start - iter->utf16_base,
+ &status);
+
+ status = U_ZERO_ERROR;
+
+ *start = len1 + iter->utf8_base;
+
+ u_strToUTF8(0, 0, &len2,
+ iter->org->utf16 + iter->utf16_base,
+ iter->org_start - iter->utf16_base + iter->org_len,
+ &status);
+
+ *len = len2 - len1;
+
+ iter->utf8_base = *start;
+ iter->utf16_base = iter->org_start;
}
int icu_chain_assign_cstr(struct icu_chain *chain, const char *src8cstr,