Introduce icu_iter_first

diff --git a/include/yaz/icu_I18N.h b/include/yaz/icu_I18N.h
index c6b8672..932dab6 100644 (file)
--- a/include/yaz/icu_I18N.h
+++ b/include/yaz/icu_I18N.h
@@ -161,8 +161,8 @@ yaz_icu_chain_t icu_chain_create(const char * locale,
 
 
 struct icu_iter;
-struct icu_iter *icu_iter_create(struct icu_chain *chain,
-                                 const char *src8cstr);
+struct icu_iter *icu_iter_create(struct icu_chain *chain);
+void icu_iter_first(struct icu_iter *iter, const char *src8cstr);
 void icu_iter_destroy(struct icu_iter *iter);
 int icu_iter_next(struct icu_iter *iter, struct icu_buf_utf8 *result);
 const char *icu_iter_get_sortkey(struct icu_iter *iter);

diff --git a/src/icu_chain.c b/src/icu_chain.c
index eb52978..4abbf68 100644 (file)
--- a/src/icu_chain.c
+++ b/src/icu_chain.c
@@ -409,28 +409,29 @@ struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter,
     }
 }
 
-struct icu_iter *icu_iter_create(struct icu_chain *chain,
-                                 const char *src8cstr)
+struct icu_iter *icu_iter_create(struct icu_chain *chain)
 {
-    if (!src8cstr)
-        return 0;
-    else
-    {
-        struct icu_iter *iter = xmalloc(sizeof(*iter));
-        iter->chain = chain;
-        iter->status = U_ZERO_ERROR;
-        iter->display = icu_buf_utf8_create(0);
-        iter->sort8 = icu_buf_utf8_create(0);
-        iter->token_count = 0;
-        iter->last = 0; /* no last returned string (yet) */
-        iter->steps = icu_chain_step_clone(chain->csteps);
-
-        /* fill and assign input string.. It will be 0 after
-           first iteration */
-        iter->input =  icu_buf_utf16_create(0);
-        icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status);
-        return iter;
-    }
+    struct icu_iter *iter = xmalloc(sizeof(*iter));
+    iter->chain = chain;
+    iter->status = U_ZERO_ERROR;
+    iter->display = icu_buf_utf8_create(0);
+    iter->sort8 = icu_buf_utf8_create(0);
+    iter->last = 0; /* no last returned string (yet) */
+    iter->steps = icu_chain_step_clone(chain->csteps);
+    iter->input = 0;
+
+    return iter;
+}
+
+void icu_iter_first(struct icu_iter *iter, const char *src8cstr)
+{
+    if (iter->input)
+        icu_buf_utf16_destroy(iter->input);
+    iter->input = icu_buf_utf16_create(0);
+    iter->token_count = 0;
+    /* fill and assign input string.. It will be 0 after
+       first iteration */
+    icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status);
 }
 
 void icu_iter_destroy(struct icu_iter *iter)
@@ -491,7 +492,8 @@ int icu_chain_assign_cstr(struct icu_chain * chain, const char * src8cstr,
 {
     if (chain->iter)
         icu_iter_destroy(chain->iter);
-    chain->iter = icu_iter_create(chain, src8cstr);
+    chain->iter = icu_iter_create(chain);
+    icu_iter_first(chain->iter, src8cstr);
     return 1;
 }
 

diff --git a/src/icu_transform.c b/src/icu_transform.c
index 0d75ff7..4e3036d 100644 (file)
--- a/src/icu_transform.c
+++ b/src/icu_transform.c
@@ -60,7 +60,10 @@ struct icu_transform * icu_transform_create(const char *id, char action,
     transform->trans = 0;
 
     if (id)
+    {
         icu_utf16_from_utf8_cstr(id16, id, status);
+        id16->utf16[id16->utf16_len] = 0;
+    }
     if (rules)
         icu_utf16_from_utf8_cstr(rules16, rules, status);
 

diff --git a/test/tst_icu_I18N.c b/test/tst_icu_I18N.c
index d4dc7c1..4ffc9de 100644 (file)
--- a/test/tst_icu_I18N.c
+++ b/test/tst_icu_I18N.c
@@ -647,7 +647,8 @@ static void check_icu_iter1(void)
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);
     
-    iter = icu_iter_create(chain, "a string with 15 tokens and 8 displays");
+    iter = icu_iter_create(chain);
+    icu_iter_first(iter, "a string with 15 tokens and 8 displays");
     YAZ_CHECK(iter);
     if (!iter)
         return;
@@ -665,8 +666,9 @@ static void check_icu_iter1(void)
 static int test_iter(struct icu_chain *chain, const char *input,
                      const char *expected)
 {
-    struct icu_iter *iter = icu_iter_create(chain, input);
-    WRBUF result;
+    struct icu_iter *iter = icu_iter_create(chain);
+    WRBUF result, second;
+    int success = 1;
     struct icu_buf_utf8 *token;
 
     if (!iter)
@@ -676,26 +678,52 @@ static int test_iter(struct icu_chain *chain, const char *input,
     }
 
     token = icu_buf_utf8_create(0);
+
+    if (icu_iter_next(iter, token))
+    {
+        yaz_log(YLOG_WARN, "test_iter: expecting 0 before icu_iter_first");
+        return 0;
+    }
+
     result = wrbuf_alloc();
+    icu_iter_first(iter, input);
     while (icu_iter_next(iter, token))
     {
         wrbuf_puts(result, "[");
         wrbuf_write(result, (const char *) token->utf8, (int) token->utf8_len);
         wrbuf_puts(result, "]");
     }
-    icu_buf_utf8_destroy(token);
 
+
+    second = wrbuf_alloc();
+    icu_iter_first(iter, input);
+    while (icu_iter_next(iter, token))
+    {
+        wrbuf_puts(second, "[");
+        wrbuf_write(second, (const char *) token->utf8, (int) token->utf8_len);
+        wrbuf_puts(second, "]");
+    }
+
+    icu_buf_utf8_destroy(token);
     icu_iter_destroy(iter);
 
     if (strcmp(expected, wrbuf_cstr(result)))
     {
         yaz_log(YLOG_WARN, "test_iter: input=%s expected=%s got=%s",
                 input, expected, wrbuf_cstr(result));
-        wrbuf_destroy(result);
-        return 0;
+        success = 0;
+    }
+
+    if (strcmp(expected, wrbuf_cstr(second)))
+    {
+        yaz_log(YLOG_WARN, "test_iter: input=%s expected=%s got=%s (2nd)",
+                input, expected, wrbuf_cstr(second));
+        success = 0;
     }
+
     wrbuf_destroy(result);
-    return 1;
+    wrbuf_destroy(second);
+    return success;
 }
 
 static void *iter_thread(void *p)