X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=util%2Fyaz-icu.c;h=10fa2f599ffa53da33e09ada8850af81893c0f07;hp=c6e7a0e2eb11fb0bfebe497194f211737f581391;hb=36f0d152f5f2b4d6ef4bcb01a041f4c16c1deaf8;hpb=479969ddd4d243955f2e64ef8571926429ad4f37

diff --git a/util/yaz-icu.c b/util/yaz-icu.c
index c6e7a0e..10fa2f5 100644
--- a/util/yaz-icu.c
+++ b/util/yaz-icu.c
@@ -1,8 +1,6 @@
-/*
- * Copyright (C) 1995-2007, Index Data ApS
+/* This file is part of the YAZ toolkit.
+ * Copyright (C) 1995-2009 Index Data
  * See the file LICENSE for details.
- *
- * $Id: yaz-icu.c,v 1.9 2007-11-08 08:17:18 adam Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -16,20 +14,24 @@
 
 #include <yaz/options.h>
 
-
 #if YAZ_HAVE_ICU
 
 #include <unicode/ucnv.h>
 #include <unicode/ustring.h>
+#include <unicode/ucol.h> 
+#include <unicode/ubrk.h>
+#include <unicode/utrans.h>
 
-#include <yaz/icu_I18N.h>
+#include <yaz/icu.h>
+#include <yaz/wrbuf.h>
 
 /* commando line and config parameters */
 static struct config_t { 
     char conffile[1024];
     char print[1024];
     int xmloutput;
-    struct icu_chain * chain;
+    int sortoutput;
+    yaz_icu_chain_t chain;
     FILE * infile;
     FILE * outfile;
 } config;
@@ -42,6 +44,7 @@ void print_option_error(const struct config_t *p_config)
     fprintf(stderr, "yaz-icu\n"
             "   [-c (path/to/config/file.xml)]\n"
             "   [-p (a|c|l|t)] print ICU info \n"
+            "   [-s] Show sort normalization key\n"
             "   [-x] XML output\n"
             "\n"
             "Examples:\n"
@@ -51,10 +54,10 @@ void print_option_error(const struct config_t *p_config)
             "./yaz-icu -p t -x\n"
             "\n"
             "Example ICU chain XML configuration file:\n"
-            "<icu_chain id=\"en:word\" locale=\"en\">\n"
-            "  <normalize rule=\"[:Control:] Any-Remove\"/>\n"
+            "<icu_chain locale=\"en\">\n"
+            "  <transform rule=\"[:Control:] Any-Remove\"/>\n"
             "  <tokenize rule=\"l\"/>\n"
-            "  <normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>\n"
+            "  <transform rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>\n"
             "  <casemap rule=\"l\"/>\n"
             "</icu_chain>\n"
           );
@@ -70,13 +73,14 @@ void read_params(int argc, char **argv, struct config_t *p_config)
     p_config->conffile[0] = 0;
     p_config->print[0] = 0;
     p_config->xmloutput = 0;
+    p_config->sortoutput = 0;
     p_config->chain = 0;
     p_config->infile = stdin;
     p_config->outfile = stdout;
     
     /* set up command line parameters */
     
-    while ((ret = options("c:p:x", argv, argc, &arg)) != -2)
+    while ((ret = options("c:p:xs", argv, argc, &arg)) != -2)
     {
         switch (ret)
         {
@@ -86,10 +90,14 @@ void read_params(int argc, char **argv, struct config_t *p_config)
         case 'p':
             strcpy(p_config->print, arg);
             break;
+        case 's':
+            p_config->sortoutput = 1;
+            break;
         case 'x':
             p_config->xmloutput = 1;
             break;
         default:
+            printf("Got %d\n", ret);
             print_option_error(p_config);
         }
     }
@@ -136,7 +144,8 @@ static void print_icu_converters(const struct config_t *p_config)
                 ucnv_getDefaultName());
     }
     
-    for(i=0;i<count;i++){
+    for(i=0;i<count;i++)
+    {
         if (p_config->xmloutput)
             fprintf(config.outfile, "<converter id=\"%s\"/>\n", 
                     ucnv_getAvailableName(i));
@@ -153,7 +162,7 @@ static void print_icu_converters(const struct config_t *p_config)
 static void print_icu_transliterators(const struct config_t *p_config)
 {
     int32_t buf_cap = 128;
-    char buf[buf_cap];
+    char buf[128];
     int32_t i;
     int32_t count = utrans_countAvailableIDs();
     
@@ -171,7 +180,8 @@ static void print_icu_transliterators(const struct config_t *p_config)
             fprintf(config.outfile, " %s", buf);
     }
     
-    if (p_config->xmloutput){
+    if (p_config->xmloutput)
+    {
         fprintf(config.outfile, "</transliterators>\n");
     }
     else
@@ -268,8 +278,8 @@ static void print_icu_xml_locales(const struct config_t *p_config)
 
     count = uloc_countAvailable() ;
 
-    if (p_config->xmloutput){
-    
+    if (p_config->xmloutput)
+    {
         fprintf(config.outfile, "<locales count=\"%d\" default=\"%s\" collations=\"%d\">\n", 
                 count, uloc_getDefault(), ucol_countAvailable());
     }
@@ -343,7 +353,8 @@ static void print_icu_xml_locales(const struct config_t *p_config)
                     &status);
 
 
-        if (p_config->xmloutput){
+        if (p_config->xmloutput)
+        {
             fprintf(config.outfile, "<locale id=\"%s\"", uloc_getAvailable(i)); 
             /* fprintf(config.outfile, " locale=\"%s\"", uloc_getAvailable(i)); */
             /* if (strlen(keyword_str)) */
@@ -367,7 +378,8 @@ static void print_icu_xml_locales(const struct config_t *p_config)
                 fprintf(config.outfile, "%s", localname_str);
             fprintf(config.outfile, "</locale>\n"); 
         }
-        else if (1 == p_config->xmloutput){
+        else if (1 == p_config->xmloutput)
+        {
             fprintf(config.outfile, "%s", uloc_getAvailable(i)); 
             fprintf(config.outfile, " | ");
             if (strlen(name_str))
@@ -385,9 +397,10 @@ static void print_icu_xml_locales(const struct config_t *p_config)
     else
         fprintf(config.outfile, "\n");
 
-    if(U_FAILURE(status)) {
+    if(U_FAILURE(status))
+    {
         fprintf(stderr, "ICU Error: %d %s\n", status, u_errorName(status));
-        exit(status);
+        exit(2);
     }
 }
 
@@ -425,38 +438,29 @@ 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;    
     
     UErrorCode status = U_ZERO_ERROR;
-    int success = 0;
     
-    if (! xml_node) {   
+    if (!xml_node)
+    {   
         printf("Could not parse XML config file '%s' \n",
                 config.conffile);
-        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);
+    config.chain = icu_chain_xml_config(xml_node, 1, &status);
 
-    xmlFree(xml_locale);
-
-
-    if (config.chain && U_SUCCESS(status))
-        success = 1;
-    else {   
+    if (!config.chain || !U_SUCCESS(status))
+    {   
         printf("Could not set up ICU chain from config file '%s' \n",
                 config.conffile);
-        exit (1);
+        if (!U_SUCCESS(status))
+            printf("ICU Error: %d %s\n", status, u_errorName(status));
+        exit(1);
     }
-    
-
 
     if (p_config->xmloutput)
         fprintf(config.outfile,
@@ -467,38 +471,70 @@ static void process_text_file(const struct config_t *p_config)
     /* read input lines for processing */
     while ((line=fgets(linebuf, sizeof(linebuf)-1, config.infile)))
     {
-        success = icu_chain_assign_cstr(config.chain, line, &status);
+        WRBUF sw = wrbuf_alloc();
+        WRBUF cdata = wrbuf_alloc();
+        int success = icu_chain_assign_cstr(config.chain, line, &status);
         line_count++;
 
-        while (success && icu_chain_next_token(config.chain, &status)){
+        while (success && icu_chain_next_token(config.chain, &status))
+        {
             if (U_FAILURE(status))
                 success = 0;
-            else {
+            else
+            {
+                const char *sortkey = icu_chain_token_sortkey(config.chain);
+                wrbuf_rewind(sw);
+                wrbuf_puts_escaped(sw, sortkey);
                 token_count++;
                 if (p_config->xmloutput)                    
+                {
                     fprintf(config.outfile, 
-                            "<token id=\%lu\" line=\"%lu\""
-                            " norm=\"%s\" display=\"%s\"/>\n",
-                            token_count,
-                            line_count,
-                            icu_chain_token_norm(config.chain),
-                            icu_chain_token_display(config.chain));
+                            "<token id=\"%lu\" line=\"%lu\"",
+                            token_count, line_count);
+
+                    wrbuf_rewind(cdata);
+                    wrbuf_xmlputs(cdata, icu_chain_token_norm(config.chain));
+                    fprintf(config.outfile, " norm=\"%s\"",
+                            wrbuf_cstr(cdata));
+
+                    wrbuf_rewind(cdata);
+                    wrbuf_xmlputs(cdata, icu_chain_token_display(config.chain));
+                    fprintf(config.outfile, " display=\"%s\"",
+                            wrbuf_cstr(cdata));
+                    
+                    if (p_config->sortoutput)
+                    {
+                        wrbuf_rewind(cdata);
+                        wrbuf_xmlputs(cdata, wrbuf_cstr(sw));
+                        fprintf(config.outfile, " sortkey=\"%s\"",
+                                wrbuf_cstr(cdata));
+                    }
+                    fprintf(config.outfile, "/>\n");
+                }
                 else
-                    fprintf(config.outfile, "%lu %lu '%s' '%s'\n",
+                {
+                    fprintf(config.outfile, "%lu %lu '%s' '%s'",
                             token_count,
                             line_count,
                             icu_chain_token_norm(config.chain),
                             icu_chain_token_display(config.chain));
+                    if (p_config->sortoutput)
+                    {
+                        fprintf(config.outfile, " '%s'", wrbuf_cstr(sw));
+                    }
+                    fprintf(config.outfile, "\n");
+                }
             }
         }
-        
+        wrbuf_destroy(sw);
+        wrbuf_destroy(cdata);
     }
 
     if (p_config->xmloutput)
-        fprintf(config.outfile, 
+        fprintf(config.outfile,
                 "</tokens>\n"
                 "</icu>\n");
-
+    
     icu_chain_destroy(config.chain);
     xmlFreeDoc(doc);
     if (line)
@@ -528,15 +564,17 @@ int main(int argc, char **argv)
            "re-configure and re-compile\n");
 
 
+    exit(3);
 #endif /* YAZ_HAVE_ICU */
 
-    return(0);
+    return 0;
 }
 
 
 /*
  * Local variables:
  * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
  * indent-tabs-mode: nil
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab