X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=util%2Fzebramap.c;h=f462ef9dfab91fa23b323fb923406f4927860140;hb=306231655249a6c67f850f6b0e907ed931585b59;hp=d2cd37f8fceb7b19118a10e181655da699be2d29;hpb=896b30853daabb6294afe8b0a2f74fa6d6e397d8;p=idzebra-moved-to-github.git

diff --git a/util/zebramap.c b/util/zebramap.c
index d2cd37f..f462ef9 100644
--- a/util/zebramap.c
+++ b/util/zebramap.c
@@ -1,4 +1,4 @@
-/* $Id: zebramap.c,v 1.73 2007-12-07 14:09:10 adam Exp $
+/* $Id: zebramap.c,v 1.78 2008-01-24 16:13:12 adam Exp $
    Copyright (C) 1995-2007
    Index Data ApS
 
@@ -230,8 +230,16 @@ static int parse_command(zebra_maps_t zms, int argc, char **argv,
     }
     else if (!yaz_matchstr(argv[0], "icuchain"))
     {
+        char full_path[1024];
+        if (!yaz_filepath_resolve(argv[1], zms->tabpath, zms->tabroot,
+                                  full_path))
+        {
+            yaz_log(YLOG_WARN, "%s:%d: Could not locate icuchain config '%s'",
+                    fname, lineno, argv[1]);
+            return -1;
+        }
 #if YAZ_HAVE_XML2
-        zm->doc = xmlParseFile(argv[1]);
+        zm->doc = xmlParseFile(full_path);
         if (!zm->doc)
         {
             yaz_log(YLOG_WARN, "%s:%d: Could not load icuchain config '%s'",
@@ -641,7 +649,8 @@ static int tokenize_simple(zebra_map_t zm,
 
 
 int zebra_map_tokenize_next(zebra_map_t zm,
-                            const char **result_buf, size_t *result_len)
+                            const char **result_buf, size_t *result_len,
+                            const char **display_buf, size_t *display_len)
 {
     assert(zm->use_chain);
 
@@ -653,12 +662,19 @@ int zebra_map_tokenize_next(zebra_map_t zm,
         UErrorCode status;
         while (icu_chain_next_token(zm->icu_chain, &status))
         {
-            assert(U_SUCCESS(status));
+            if (!U_SUCCESS(status))
+                return 0;
             *result_buf = icu_chain_token_sortkey(zm->icu_chain);
             assert(*result_buf);
 
             *result_len = strlen(*result_buf);
 
+            if (display_buf)
+            {
+                *display_buf = icu_chain_token_display(zm->icu_chain);
+                if (display_len)
+                    *display_len = strlen(*display_buf);
+            }
             if (zm->debug)
             {
                 wrbuf_rewind(zm->print_str);
@@ -669,7 +685,6 @@ int zebra_map_tokenize_next(zebra_map_t zm,
             if (**result_buf != '\0')
                 return 1;
         }
-        assert(U_SUCCESS(status));
     }
     return 0;
 #else
@@ -701,77 +716,22 @@ int zebra_map_tokenize_start(zebra_map_t zm,
         icu_chain_assign_cstr(zm->icu_chain,
                               wrbuf_cstr(zm->input_str),
                               &status);
-        assert(U_SUCCESS(status));
-    }
-#endif
-    return 0;
-}
-
-#if 0
-int zebra_map_tokenize(zebra_map_t zm,
-                       const char *buf, size_t len,
-                       const char **result_buf, size_t *result_len)
-{
-    assert(zm->use_chain);
-
-    if (buf)
-    {
-        wrbuf_rewind(zm->input_str);
-        wrbuf_write(zm->input_str, buf, len);
-        zm->simple_off = 0;
-    }
-
-#if YAZ_HAVE_ICU
-    if (!zm->icu_chain)
-        return tokenize_simple(zm, result_buf, result_len);
-    else
-    {
-        UErrorCode status;
-        if (buf)
+        if (!U_SUCCESS(status))
         {
             if (zm->debug)
             {
-                wrbuf_rewind(zm->print_str);
-                wrbuf_write_escaped(zm->print_str, wrbuf_buf(zm->input_str),
-                                    wrbuf_len(zm->input_str));
-                
-                yaz_log(YLOG_LOG, "input %s", 
-                        wrbuf_cstr(zm->print_str)); 
+                yaz_log(YLOG_WARN, "bad encoding for input");
             }
-            icu_chain_assign_cstr(zm->icu_chain,
-                                  wrbuf_cstr(zm->input_str),
-                                  &status);
-            assert(U_SUCCESS(status));
-        }
-        while (icu_chain_next_token(zm->icu_chain, &status))
-        {
-            assert(U_SUCCESS(status));
-            *result_buf = icu_chain_token_sortkey(zm->icu_chain);
-            assert(*result_buf);
-
-            *result_len = strlen(*result_buf);
-
-            if (zm->debug)
-            {
-                wrbuf_rewind(zm->print_str);
-                wrbuf_write_escaped(zm->print_str, *result_buf, *result_len);
-                yaz_log(YLOG_LOG, "output %s", wrbuf_cstr(zm->print_str));
-            }
-
-            if (**result_buf != '\0')
-                return 1;
+            return -1;
         }
-        assert(U_SUCCESS(status));
     }
-    return 0;
-#else
-    return tokenize_simple(zm, result_buf, result_len);
 #endif
+    return 0;
 }
-#endif
 
 int zebra_maps_is_icu(zebra_map_t zm)
 {
+    assert(zm);
 #if YAZ_HAVE_ICU
     return zm->use_chain;
 #else