X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=src%2Ficu_I18N.c;h=1d08d5a11acdee1526db009beeaa75cd240bf143;hb=2fdbd5de9185e926401609c22b328f07af0248bd;hp=29b9edda834905110d9b4777d18dcec3617e0715;hpb=968c591d366c50df78a3dda223fdb3491cbcb6cd;p=pazpar2-moved-to-github.git

diff --git a/src/icu_I18N.c b/src/icu_I18N.c
index 29b9edd..1d08d5a 100644
--- a/src/icu_I18N.c
+++ b/src/icu_I18N.c
@@ -1,4 +1,4 @@
-/* $Id: icu_I18N.c,v 1.17 2007-05-20 19:00:17 marc Exp $
+/* $Id: icu_I18N.c,v 1.22 2007-05-25 13:27:21 marc Exp $
    Copyright (c) 2006-2007, Index Data.
 
    This file is part of Pazpar2.
@@ -687,6 +687,7 @@ struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
                            UTRANS_FORWARD,
                            0, 0, 
                            normalizer->parse_error, status);
+        // yaz_log(YLOG_LOG, "utrans_open %p", normalizer->trans);
         break;
     case 'r':
         normalizer->trans
@@ -695,6 +696,7 @@ struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
                            UTRANS_REVERSE ,
                            0, 0,
                            normalizer->parse_error, status);
+        // yaz_log(YLOG_LOG, "utrans_open %p", normalizer->trans);
         break;
     default:
         *status = U_UNSUPPORTED_ERROR;
@@ -716,7 +718,10 @@ void icu_normalizer_destroy(struct icu_normalizer * normalizer){
         if (normalizer->rules16) 
             icu_buf_utf16_destroy(normalizer->rules16);
         if (normalizer->trans)
+        {
+            // yaz_log(YLOG_LOG, "utrans_close %p", normalizer->trans);
             utrans_close(normalizer->trans);
+        }
         free(normalizer);
     }
 };
@@ -764,21 +769,16 @@ struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
     step = (struct icu_chain_step *) malloc(sizeof(struct icu_chain_step));
 
     step->type = type;
-    step->more_tokens = 0;
-    step->need_new_token = 1;
 
-    if (buf16)
-        step->buf16 = buf16;
-    else
-        step->buf16 = 0;
+    step->buf16 = buf16;
 
     // create auxilary objects
     switch(step->type) {
     case ICU_chain_step_type_display:
         break;
-    case ICU_chain_step_type_norm:
+    case ICU_chain_step_type_index:
         break;
-    case ICU_chain_step_type_sort:
+    case ICU_chain_step_type_sortkey:
         break;
     case ICU_chain_step_type_casemap:
         step->u.casemap = icu_casemap_create((char *) chain->locale, 
@@ -809,9 +809,9 @@ void icu_chain_step_destroy(struct icu_chain_step * step){
     switch(step->type) {
     case ICU_chain_step_type_display:
         break;
-    case ICU_chain_step_type_norm:
+    case ICU_chain_step_type_index:
         break;
-    case ICU_chain_step_type_sort:
+    case ICU_chain_step_type_sortkey:
         break;
     case ICU_chain_step_type_casemap:
         icu_casemap_destroy(step->u.casemap);
@@ -828,8 +828,7 @@ void icu_chain_step_destroy(struct icu_chain_step * step){
     default:
         break;
     }
-
-
+    free(step);
 };
 
 
@@ -862,16 +861,100 @@ struct icu_chain * icu_chain_create(const uint8_t * identifier,
 
 void icu_chain_destroy(struct icu_chain * chain)
 {
-    icu_buf_utf8_destroy(chain->display8);
-    icu_buf_utf8_destroy(chain->norm8);
-    icu_buf_utf8_destroy(chain->sort8);
+    if (chain){
+        icu_buf_utf8_destroy(chain->display8);
+        icu_buf_utf8_destroy(chain->norm8);
+        icu_buf_utf8_destroy(chain->sort8);
+        
+        icu_buf_utf16_destroy(chain->src16);
+    
+        icu_chain_step_destroy(chain->steps);
+        free(chain);
+    }
+};
+
+
+
+struct icu_chain * icu_chain_xml_config(xmlNode *xml_node, 
+                                        UErrorCode * status){
+
+    xmlNode *node = 0;
+    struct icu_chain * chain = 0;
+   
+    if (!xml_node 
+        ||xml_node->type != XML_ELEMENT_NODE 
+        || strcmp((const char *) xml_node->name, "icu_chain"))
+
+        return 0;
+    
+    xmlChar *xml_id = xmlGetProp(xml_node, (xmlChar *) "id");
+    xmlChar *xml_locale = xmlGetProp(xml_node, (xmlChar *) "locale");
+
+    if (!xml_id || !strlen((const char *) xml_id) 
+        || !xml_locale || !strlen((const char *) xml_locale))
+        return 0;
+
+    chain = icu_chain_create((const uint8_t *) xml_id, 
+                             (const uint8_t *) xml_locale);
+    
+    xmlFree(xml_id);
+    xmlFree(xml_locale);
+    if (!chain)
+        return 0;
+        
+    for (node = xml_node->children; node; node = node->next)
+    {
+        if (node->type != XML_ELEMENT_NODE)
+            continue;
+
+        xmlChar *xml_rule = xmlGetProp(node, (xmlChar *) "rule");
+        struct icu_chain_step * step = 0;
+
+        if (!strcmp((const char *) node->name, 
+                    (const char *) "casemap")){
+            step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap, 
+                                         (const uint8_t *) xml_rule, status);
+        }
+        else if (!strcmp((const char *) node->name,
+                         (const char *) "normalize")){
+            step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize, 
+                                         (const uint8_t *) xml_rule, status);
+        }
+        else if (!strcmp((const char *) node->name,
+                         (const char *) "tokenize")){
+            step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize, 
+                                         (const uint8_t *) xml_rule, status);
+        }
+        else if (!strcmp((const char *) node->name,
+                         (const char *) "display")){
+            step = icu_chain_insert_step(chain, ICU_chain_step_type_display, 
+                                         (const uint8_t *) "", status);
+        }
+        else if (!strcmp((const char *) node->name,
+                         (const char *) "index")){
+            step = icu_chain_insert_step(chain, ICU_chain_step_type_index, 
+                                         (const uint8_t *) "", status);
+        }
+        else if (!strcmp((const char *) node->name,
+                         (const char *) "sortkey")){
+            step = icu_chain_insert_step(chain, ICU_chain_step_type_sortkey, 
+                                         (const uint8_t *) "", status);
+        }
 
-    icu_buf_utf16_destroy(chain->src16);
+        xmlFree(xml_rule);
+        if (!step || U_FAILURE(*status)){
+            icu_chain_destroy(chain);
+            return 0;
+        }
+        
 
-    icu_chain_step_destroy(chain->steps);
+    }
+
+    return chain;
 };
 
 
+
 struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
                                               enum icu_chain_step_type type,
                                               const uint8_t * rule,
@@ -898,10 +981,10 @@ struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
     case ICU_chain_step_type_display:
         buf16 = src16;
         break;
-    case ICU_chain_step_type_norm:
+    case ICU_chain_step_type_index:
         buf16 = src16;
         break;
-    case ICU_chain_step_type_sort:
+    case ICU_chain_step_type_sortkey:
         buf16 = src16;
         break;
     case ICU_chain_step_type_casemap:
@@ -966,10 +1049,10 @@ int icu_chain_step_next_token(struct icu_chain * chain,
     case ICU_chain_step_type_display:
         icu_utf16_to_utf8(chain->display8, src16, status);
         break;
-    case ICU_chain_step_type_norm:
+    case ICU_chain_step_type_index:
         icu_utf16_to_utf8(chain->norm8, src16, status);
         break;
-    case ICU_chain_step_type_sort:
+    case ICU_chain_step_type_sortkey:
         icu_utf16_to_utf8(chain->sort8, src16, status);
         break;
     case ICU_chain_step_type_casemap:
@@ -1031,10 +1114,12 @@ int icu_chain_assign_cstr(struct icu_chain * chain,
                           const char * src8cstr, 
                           UErrorCode *status)
 {
-    struct icu_chain_step * stp = chain->steps;
+    struct icu_chain_step * stp = 0; 
 
     if (!chain || !src8cstr)
         return 0;
+
+    stp = chain->steps;
     
     // clear token count
     chain->token_count = 0;
@@ -1043,6 +1128,7 @@ int icu_chain_assign_cstr(struct icu_chain * chain,
 
     while (stp){
         stp->more_tokens = 1;
+        stp->need_new_token = 1;
         stp = stp->previous;
     }