Isolated ICU locale out of XML config file. Locale is now the second argument to...

diff --git a/include/yaz/icu_I18N.h b/include/yaz/icu_I18N.h
index 7436b2e..e4f5c8c 100644 (file)
--- a/include/yaz/icu_I18N.h
+++ b/include/yaz/icu_I18N.h
@@ -243,12 +243,13 @@ struct icu_chain
     struct icu_chain_step * steps;
 };
 
-struct icu_chain * icu_chain_create(const uint8_t * identifier, 
+struct icu_chain * icu_chain_create( // const uint8_t * identifier, 
                                     const uint8_t * locale);
 
 void icu_chain_destroy(struct icu_chain * chain);
 
-struct icu_chain * icu_chain_xml_config(xmlNode *xml_node, 
+struct icu_chain * icu_chain_xml_config(xmlNode *xml_node,
+                                        const uint8_t * locale, 
                                         UErrorCode * status);
 
 

diff --git a/src/icu_I18N.c b/src/icu_I18N.c
index c3b0680..230d59f 100644 (file)
--- a/src/icu_I18N.c
+++ b/src/icu_I18N.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: icu_I18N.c,v 1.4 2007-10-24 13:23:34 marc Exp $
+ * $Id: icu_I18N.c,v 1.5 2007-10-24 14:48:17 marc Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -858,15 +858,15 @@ void icu_chain_step_destroy(struct icu_chain_step * step){
 
 
 
-struct icu_chain * icu_chain_create(const uint8_t * identifier,
+struct icu_chain * icu_chain_create( //const uint8_t * identifier,
                                     const uint8_t * locale)
 {
 
     struct icu_chain * chain 
         = (struct icu_chain *) malloc(sizeof(struct icu_chain));
 
-    strncpy((char *) chain->identifier, (const char *) identifier, 128);
-    chain->identifier[128 - 1] = '\0';
+    //strncpy((char *) chain->identifier, (const char *) identifier, 128);
+    //chain->identifier[128 - 1] = '\0';
     strncpy((char *) chain->locale, (const char *) locale, 16);    
     chain->locale[16 - 1] = '\0';
 
@@ -901,6 +901,7 @@ void icu_chain_destroy(struct icu_chain * chain)
 
 
 struct icu_chain * icu_chain_xml_config(xmlNode *xml_node, 
+                                        const uint8_t * locale, 
                                         UErrorCode * status){
 
     xmlNode *node = 0;
@@ -908,26 +909,17 @@ struct icu_chain * icu_chain_xml_config(xmlNode *xml_node,
    
     if (!xml_node 
         ||xml_node->type != XML_ELEMENT_NODE 
-        || strcmp((const char *) xml_node->name, "icu_chain"))
+        // || 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);
-    }
+        chain = icu_chain_create( // (const uint8_t *) xml_id, 
+                                 locale);
+
     if (!chain)
         return 0;
+
         
     for (node = xml_node->children; node; node = node->next)
     {

diff --git a/test/tst_icu_I18N.c b/test/tst_icu_I18N.c
index 717e760..35b4c56 100644 (file)
--- a/test/tst_icu_I18N.c
+++ b/test/tst_icu_I18N.c
@@ -1,4 +1,4 @@
-/* $Id: tst_icu_I18N.c,v 1.5 2007-10-24 13:23:34 marc Exp $
+/* $Id: tst_icu_I18N.c,v 1.6 2007-10-24 14:48:17 marc Exp $
    Copyright (c) 2006-2007, Index Data.
 
    This file is part of Pazpar2.
@@ -493,7 +493,7 @@ void test_icu_I18N_chain(int argc, char **argv)
     struct icu_chain * chain = 0;
     
 
-    const char * xml_str = "<icu_chain id=\"en:word\" locale=\"en\">"
+    const char * xml_str = "<icu_chain>"
         "<normalize rule=\"[:Control:] Any-Remove\"/>"
         "<tokenize rule=\"l\"/>"
         "<normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
@@ -511,7 +511,7 @@ void test_icu_I18N_chain(int argc, char **argv)
     // printf("ICU chain:\ninput: '%s'\n", en_str);
 
 
-    chain = icu_chain_xml_config(xml_node, &status);
+    chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", &status);
 
 #if 0
     chain  = icu_chain_create((uint8_t *) "en:word", (uint8_t *) "en");
@@ -584,7 +584,7 @@ void test_bug_1140(void)
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
     
-    const char * xml_str = "<icu_chain id=\"en:word\" locale=\"en\">"
+    const char * xml_str = "<icu_chain>"
 
         /* if the first rule is normalize instead. Then it works */
 #if 0
@@ -603,7 +603,7 @@ void test_bug_1140(void)
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 
-    chain = icu_chain_xml_config(xml_node, &status);
+    chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", &status);
 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);
@@ -647,7 +647,7 @@ void test_chain_empty_token(void)
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
 
-    const char * xml_str = "<icu_chain id=\"en:word\" locale=\"en\">"
+    const char * xml_str = "<icu_chain>"
         "<tokenize rule=\"w\"/>"
         "<normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
         "<index/>"
@@ -657,7 +657,7 @@ void test_chain_empty_token(void)
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 
-    chain = icu_chain_xml_config(xml_node, &status);
+    chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", &status);
 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);

diff --git a/util/yaz-icu.c b/util/yaz-icu.c
index d929229..a876186 100644 (file)
--- a/util/yaz-icu.c
+++ b/util/yaz-icu.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: yaz-icu.c,v 1.2 2007-10-22 17:32:08 adam Exp $
+ * $Id: yaz-icu.c,v 1.3 2007-10-24 14:48:18 marc Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -428,6 +428,7 @@ static void process_text_file(const struct config_t *p_config)
  
     xmlDoc *doc = xmlParseFile(config.conffile);  
     xmlNode *xml_node = xmlDocGetRootElement(doc);
+    xmlChar *xml_locale = xmlGetProp(xml_node, (xmlChar *) "locale");
 
     long unsigned int token_count = 0;    
     long unsigned int line_count = 0;    
@@ -441,8 +442,13 @@ static void process_text_file(const struct config_t *p_config)
         exit (1);
     }
 
+    if (!xml_locale || !strlen((const char *) xml_locale))
+        return;        
     
-    config.chain = icu_chain_xml_config(xml_node, &status);
+    config.chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", &status);
+
+    xmlFree(xml_locale);
+
 
     if (config.chain && U_SUCCESS(status))
         success = 1;
@@ -452,6 +458,8 @@ static void process_text_file(const struct config_t *p_config)
         exit (1);
     }
     
+
+
     if (p_config->xmloutput)
         fprintf(config.outfile,
                 "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"