Refactor ICU WRT locale.

The locale parameter is no longer a parameter for icu_chain_xml_config
and locale is read as an attribute for top XML node. The icu_casemap
locale is no longer a member and a parater for icu_casemap_casemap
instead. This saves a copy of a locale in each casemap.

diff --git a/include/yaz/icu.h b/include/yaz/icu.h
index 19ab4ce..7389d60 100644 (file)
--- a/include/yaz/icu.h
+++ b/include/yaz/icu.h
@@ -50,7 +50,6 @@ yaz_icu_chain_t icu_chain_create(const char * locale,
 void icu_chain_destroy(yaz_icu_chain_t chain);
 
 yaz_icu_chain_t icu_chain_xml_config(const xmlNode *xml_node,
 void icu_chain_destroy(yaz_icu_chain_t chain);
 
 yaz_icu_chain_t icu_chain_xml_config(const xmlNode *xml_node,

-                                     const char *locale,

                                      int sort,
                                      UErrorCode * status);
 
                                      int sort,
                                      UErrorCode * status);
 

diff --git a/include/yaz/icu_I18N.h b/include/yaz/icu_I18N.h
index 0b06dbd..2a1364d 100644 (file)
--- a/include/yaz/icu_I18N.h
+++ b/include/yaz/icu_I18N.h
@@ -102,7 +102,6 @@ UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 * dest8,
 
 struct icu_casemap
 {
 
 struct icu_casemap
 {

-    char locale[16];

     char action;
 };
 
     char action;
 };
 

@@ -114,7 +113,8 @@ void icu_casemap_destroy(struct icu_casemap * casemap);
 int icu_casemap_casemap(struct icu_casemap * casemap,
                         struct icu_buf_utf16 * dest16,
                         struct icu_buf_utf16 * src16,
 int icu_casemap_casemap(struct icu_casemap * casemap,
                         struct icu_buf_utf16 * dest16,
                         struct icu_buf_utf16 * src16,

-                        UErrorCode *status);
+                        UErrorCode *status,
+                        const char *locale);

 
 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
                       struct icu_buf_utf16 * src16,
 
 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
                       struct icu_buf_utf16 * src16,

@@ -226,7 +226,7 @@ void icu_chain_step_destroy(struct icu_chain_step * step);
 
 struct icu_chain
 {
 
 struct icu_chain
 {

-    uint8_t locale[16];
+    char locale[16];

     int sort;
 
     const char * src8cstr;
     int sort;
 
     const char * src8cstr;

diff --git a/src/icu_I18N.c b/src/icu_I18N.c
index 2259b46..4e9dcf7 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.
  *
  * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *

- * $Id: icu_I18N.c,v 1.13 2007-11-08 08:17:17 adam Exp $
+ * $Id: icu_I18N.c,v 1.14 2007-11-08 13:35:13 adam Exp $

  */
 
 #if HAVE_CONFIG_H
  */
 
 #if HAVE_CONFIG_H

@@ -333,7 +333,6 @@ struct icu_casemap * icu_casemap_create(const char *locale, char action,
 {    
     struct icu_casemap * casemap
         = (struct icu_casemap *) malloc(sizeof(struct icu_casemap));
 {    
     struct icu_casemap * casemap
         = (struct icu_casemap *) malloc(sizeof(struct icu_casemap));

-    strcpy(casemap->locale, locale);

     casemap->action = action;
 
     switch(casemap->action) {    
     casemap->action = action;
 
     switch(casemap->action) {    

@@ -356,7 +355,7 @@ struct icu_casemap * icu_casemap_create(const char *locale, char action,
 
 void icu_casemap_destroy(struct icu_casemap * casemap)
 {
 
 void icu_casemap_destroy(struct icu_casemap * casemap)
 {

-    if (casemap) 
+    if (casemap)

         free(casemap);
 }
 
         free(casemap);
 }
 

@@ -364,13 +363,14 @@ void icu_casemap_destroy(struct icu_casemap * casemap)
 int icu_casemap_casemap(struct icu_casemap * casemap,
                         struct icu_buf_utf16 * dest16,
                         struct icu_buf_utf16 * src16,
 int icu_casemap_casemap(struct icu_casemap * casemap,
                         struct icu_buf_utf16 * dest16,
                         struct icu_buf_utf16 * src16,

-                        UErrorCode *status)
+                        UErrorCode *status,
+                        const char *locale)

 {
     if(!casemap)
         return 0;
     
 {
     if(!casemap)
         return 0;
     

-    return icu_utf16_casemap(dest16, src16,
-                             casemap->locale, casemap->action, status);
+    return icu_utf16_casemap(dest16, src16, locale,
+                             casemap->action, status);

 }
 
 
 }
 
 

@@ -909,28 +909,31 @@ void icu_chain_destroy(struct icu_chain * chain)
 
 
 struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node, 
 
 
 struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node, 

-                                        const char *locale, 

                                         int sort,
                                         int sort,

-                                        UErrorCode * status){
-
+                                        UErrorCode * status)
+{

     xmlNode *node = 0;
     struct icu_chain * chain = 0;
    
     *status = U_ZERO_ERROR;
 
     xmlNode *node = 0;
     struct icu_chain * chain = 0;
    
     *status = U_ZERO_ERROR;
 

-    if (!xml_node 
-        ||xml_node->type != XML_ELEMENT_NODE 
-        // || strcmp((const char *) xml_node->name, "icu_chain")
-        )
-
+    if (!xml_node ||xml_node->type != XML_ELEMENT_NODE)

         return 0;
         return 0;

-
-        chain = icu_chain_create(locale, sort, status);
-
+    
+    {
+        xmlChar * xml_locale = xmlGetProp((xmlNode *) xml_node, 
+                                          (xmlChar *) "locale");
+        
+        if (xml_locale)
+        {
+            chain = icu_chain_create((const char *) xml_locale, sort, status);
+            xmlFree(xml_locale);
+        }
+        
+    }

     if (!chain)
         return 0;
 
     if (!chain)
         return 0;
 

-        

     for (node = xml_node->children; node; node = node->next)
     {
         xmlChar *xml_rule;
     for (node = xml_node->children; node; node = node->next)
     {
         xmlChar *xml_rule;

@@ -1072,7 +1075,8 @@ int icu_chain_step_next_token(struct icu_chain * chain,
         break;
     case ICU_chain_step_type_casemap:
         icu_casemap_casemap(step->u.casemap,
         break;
     case ICU_chain_step_type_casemap:
         icu_casemap_casemap(step->u.casemap,

-                            step->buf16, src16, status);
+                            step->buf16, src16, status,
+                            chain->locale);

         break;
     case ICU_chain_step_type_normalize:
         icu_normalizer_normalize(step->u.normalizer,
         break;
     case ICU_chain_step_type_normalize:
         icu_normalizer_normalize(step->u.normalizer,

diff --git a/test/tst_icu_I18N.c b/test/tst_icu_I18N.c
index 17c5a69..c58ac85 100644 (file)
--- a/test/tst_icu_I18N.c
+++ b/test/tst_icu_I18N.c
@@ -2,7 +2,7 @@
  * Copyright (C) 2005-2007, Index Data ApS
  * See the file LICENSE for details.
  *
  * Copyright (C) 2005-2007, Index Data ApS
  * See the file LICENSE for details.
  *

- * $Id: tst_icu_I18N.c,v 1.13 2007-11-08 09:03:11 adam Exp $
+ * $Id: tst_icu_I18N.c,v 1.14 2007-11-08 13:35:14 adam Exp $

  *
  */
 
  *
  */
 

@@ -480,7 +480,7 @@ void test_icu_I18N_chain(int argc, char **argv)
     struct icu_chain * chain = 0;
     
 
     struct icu_chain * chain = 0;
     
 

-    const char * xml_str = "<icu>"
+    const char * xml_str = "<icu locale=\"en\">"

         "<normalize rule=\"[:Control:] Any-Remove\"/>"
         "<tokenize rule=\"l\"/>"
         "<normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
         "<normalize rule=\"[:Control:] Any-Remove\"/>"
         "<tokenize rule=\"l\"/>"
         "<normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"

@@ -493,8 +493,7 @@ void test_icu_I18N_chain(int argc, char **argv)
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 

-
-    chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", 0, &status);
+    chain = icu_chain_xml_config(xml_node, 0, &status);

 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);
 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);

@@ -534,7 +533,7 @@ void test_bug_1140(void)
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
     
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
     

-    const char * xml_str = "<icu>"
+    const char * xml_str = "<icu locale=\"en\">"

 
         /* if the first rule is normalize instead. Then it works */
 #if 0
 
         /* if the first rule is normalize instead. Then it works */
 #if 0

@@ -551,7 +550,7 @@ void test_bug_1140(void)
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 

-    chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", 0, &status);
+    chain = icu_chain_xml_config(xml_node, 0, &status);

 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);
 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);

@@ -595,7 +594,7 @@ void test_chain_empty_token(void)
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
 
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
 

-    const char * xml_str = "<icu>"
+    const char * xml_str = "<icu locale=\"en\">"

         "<tokenize rule=\"w\"/>"
         "<normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
         "</icu>";
         "<tokenize rule=\"w\"/>"
         "<normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
         "</icu>";

@@ -604,7 +603,7 @@ void test_chain_empty_token(void)
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 

-    chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", 0, &status);
+    chain = icu_chain_xml_config(xml_node, 0, &status);

 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);
 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);

@@ -631,7 +630,7 @@ void test_chain_empty_chain(void)
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
 
     UErrorCode status = U_ZERO_ERROR;
     struct icu_chain * chain = 0;
 

-    const char * xml_str = "<icu>"
+    const char * xml_str = "<icu locale=\"en\">"

         "</icu>";
     
     const char * src8 = "some 5487 weired !¤%&(/& sTuFf";
         "</icu>";
     
     const char * src8 = "some 5487 weired !¤%&(/& sTuFf";

@@ -641,7 +640,7 @@ void test_chain_empty_chain(void)
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 
     xmlNode *xml_node = xmlDocGetRootElement(doc);
     YAZ_CHECK(xml_node);
 

-    chain = icu_chain_xml_config(xml_node, (uint8_t *) "en", 0, &status);
+    chain = icu_chain_xml_config(xml_node, 0, &status);

 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);
 
     xmlFreeDoc(doc);
     YAZ_CHECK(chain);

diff --git a/util/yaz-icu.c b/util/yaz-icu.c
index c4fb91f..f41b93c 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.
  *
  * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *

- * $Id: yaz-icu.c,v 1.10 2007-11-08 09:29:39 adam Exp $
+ * $Id: yaz-icu.c,v 1.11 2007-11-08 13:35:14 adam Exp $

  */
 
 #if HAVE_CONFIG_H
  */
 
 #if HAVE_CONFIG_H

@@ -428,7 +428,6 @@ static void process_text_file(const struct config_t *p_config)
  
     xmlDoc *doc = xmlParseFile(config.conffile);  
     xmlNode *xml_node = xmlDocGetRootElement(doc);
  
     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;    
 
     long unsigned int token_count = 0;    
     long unsigned int line_count = 0;    

@@ -442,14 +441,7 @@ static void process_text_file(const struct config_t *p_config)
         exit (1);
     }
 
         exit (1);
     }
 

-    if (!xml_locale || !strlen((const char *) xml_locale))
-        return;        
-    
-    config.chain = icu_chain_xml_config(xml_node, (const char *) xml_locale, 0,
-                                        &status);
-
-    xmlFree(xml_locale);
-
+    config.chain = icu_chain_xml_config(xml_node, 0, &status);

 
     if (config.chain && U_SUCCESS(status))
         success = 1;
 
     if (config.chain && U_SUCCESS(status))
         success = 1;