X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Ficu_chain.c;h=b672975293c24ce880e6d18a0f7cbc15833e6a1e;hp=730eddea1a96217d6b4a7b0f08479fcea3c7b12a;hb=0b121738c22cd67b153792d645746447d8a81767;hpb=d05e5d54fcac20824e0efa3d939e5fbb1505964a

diff --git a/src/icu_chain.c b/src/icu_chain.c
index 730edde..b672975 100644
--- a/src/icu_chain.c
+++ b/src/icu_chain.c
@@ -1,5 +1,5 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2013 Index Data
+ * Copyright (C) Index Data
  * See the file LICENSE for details.
  */
 
@@ -291,7 +291,7 @@ struct icu_chain *icu_chain_xml_config(const xmlNode *xml_node,
 
         if (!rule && strcmp((const char *) node->name, "display"))
         {
-            yaz_log(YLOG_WARN, "Missing attribute rule for element %s",
+            yaz_log(YLOG_WARN, "Missing attribute 'rule' for element %s",
                     (const char *) node->name);
             no_errors++;
             continue;
@@ -346,6 +346,9 @@ struct icu_chain *icu_chain_xml_config(const xmlNode *xml_node,
         }
         if (step && U_FAILURE(*status))
         {
+            yaz_log(YLOG_WARN, "ICU Error %d %s for element %s, rule %s",
+                    *status, u_errorName(*status), node->name, rule ?
+                    rule : "");
             no_errors++;
             break;
         }
@@ -370,6 +373,8 @@ struct icu_iter {
     int token_count;
     size_t org_start;
     size_t org_len;
+    size_t utf8_base;
+    size_t utf16_base;
     struct icu_chain_step *steps;
 };
 
@@ -501,6 +506,7 @@ void icu_iter_first(yaz_icu_iter_t iter, const char *src8cstr)
     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);
 }
@@ -563,25 +569,32 @@ int icu_iter_get_token_number(yaz_icu_iter_t iter)
 
 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;
+    int32_t len1 = 0, len2 = 0;
+    UErrorCode status = U_ZERO_ERROR;
 
-    struct icu_buf_utf8 *tmp = icu_buf_utf8_create(0);
-    UErrorCode status;
+    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);
 
-    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;
-    iter->org->utf16_len = save_len;
+    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,