yaz-icu option -o writes original string as well

This changes the output and so test results must be updated. yaz-icu
now uses new facilities of YAZ-836.

diff --git a/test/test_icu.0.output b/test/test_icu.0.output
index ba0c37c..c34bcb1 100644 (file)
--- a/test/test_icu.0.output
+++ b/test/test_icu.0.output
@@ -1,7 +1,7 @@
-1 1 'børn' 'Børn' 0+5
-2 2 'le' 'Le' 0+3
-3 2 'carré' 'Carré' 3+6
-4 3 'le' 'Le' 0+3
-5 3 'carre' 'Carre' 3+5
-6 4 'le' 'Le' 0+3
-7 4 'carré' 'Carré' 3+7
+1 1 'børn' 'Børn' 0+5 *Børn*
+2 2 'le' 'Le' 0+3 *Le *Carré
+3 2 'carré' 'Carré' 3+6 Le *Carré*
+4 3 'le' 'Le' 0+3 *Le *Carre
+5 3 'carre' 'Carre' 3+5 Le *Carre*
+6 4 'le' 'Le' 0+3 *Le *Carré
+7 4 'carré' 'Carré' 3+7 Le *Carré*

diff --git a/test/test_icu.1.output b/test/test_icu.1.output
index ba8e613..8a0b32c 100644 (file)
--- a/test/test_icu.1.output
+++ b/test/test_icu.1.output
@@ -1,7 +1,7 @@
-1 1 'børn' 'Børn' 0+5
-2 2 'le' 'Le' 0+3
-3 2 'carre' 'Carre' 3+6
-4 3 'le' 'Le' 0+3
-5 3 'carre' 'Carre' 3+5
-6 4 'le' 'Le' 0+3
-7 4 'carre' 'Carre' 3+7
+1 1 'børn' 'Børn' 0+5 *Børn*
+2 2 'le' 'Le' 0+3 *Le *Carré
+3 2 'carre' 'Carre' 3+6 Le *Carré*
+4 3 'le' 'Le' 0+3 *Le *Carre
+5 3 'carre' 'Carre' 3+5 Le *Carre*
+6 4 'le' 'Le' 0+3 *Le *Carré
+7 4 'carre' 'Carre' 3+7 Le *Carré*

diff --git a/test/test_icu.2.output b/test/test_icu.2.output
index d8bf7c6..fdff316 100644 (file)
--- a/test/test_icu.2.output
+++ b/test/test_icu.2.output
@@ -1,4 +1,6 @@
 1 1 '‘a'
-' '' 0+4
+' '' 0+4 *`a'
+*
 2 2 '“a''
-' '' 0+6
+' '' 0+6 *``a''
+*

diff --git a/test/test_icu.3.output b/test/test_icu.3.output
index 3fe68cc..b85056b 100644 (file)
--- a/test/test_icu.3.output
+++ b/test/test_icu.3.output
@@ -1,4 +1,4 @@
-1 1 'men' '' 0+4
-2 1 'run' '' 4+7
-3 2 'he' '' 0+3
-4 2 'run' '' 3+5
+1 1 'men' '' 0+4 *men *running
+2 1 'run' '' 4+7 men *running*
+3 2 'he' '' 0+3 *he *runs 
+4 2 'run' '' 3+5 he *runs *

diff --git a/test/test_icu.4.output b/test/test_icu.4.output
index 5bbdaed..01dd145 100644 (file)
--- a/test/test_icu.4.output
+++ b/test/test_icu.4.output
@@ -1 +1 @@
-1 1 'Andersen, H.C' '' 0+15
+1 1 'Andersen, H.C' '' 0+15 *Andersen, H.C.,*

diff --git a/util/yaz-icu.c b/util/yaz-icu.c
index 1386a16..ac7806a 100644 (file)
--- a/util/yaz-icu.c
+++ b/util/yaz-icu.c
@@ -479,9 +479,11 @@ static void process_text_file(struct config_t *p_config)
             else
             {
                 size_t start, len;
+                const char *org_string = 0;
                 const char *sortkey = icu_chain_token_sortkey(p_config->chain);
 
-                icu_chain_get_org_info(p_config->chain, &start, &len);
+                icu_chain_get_org_info2(p_config->chain, &start, &len,
+                                        &org_string);
                 wrbuf_rewind(sw);
                 wrbuf_puts_escaped(sw, sortkey);
                 token_count++;
@@ -525,6 +527,12 @@ static void process_text_file(struct config_t *p_config)
                     {
                         fprintf(p_config->outfile, " %ld+%ld",
                                 (long) start, (long) len);
+                        fputc(' ', p_config->outfile);
+                        fwrite(org_string, 1, start, p_config->outfile);
+                        fputc('*', p_config->outfile);
+                        fwrite(org_string + start, 1, len, p_config->outfile);
+                        fputc('*', p_config->outfile);
+                        fputs(org_string + start + len, p_config->outfile);
                     }
                     fprintf(p_config->outfile, "\n");
                 }