Merge from head the facility that removes leading articles.

[idzebra-moved-to-github.git] / index / extract.c

diff --git a/index/extract.c b/index/extract.c
index 9183bbc..1b8fdb7 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.156 2004-07-28 08:15:45 adam Exp $
+/* $Id: extract.c,v 1.157.2.1 2004-09-16 14:07:50 adam Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -1602,7 +1602,7 @@ static void extract_add_incomplete_field (RecWord *p)
     const char **map = 0;
 
     if (remain > 0)
-       map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+       map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
 
     while (map)
     {
@@ -1614,7 +1614,7 @@ static void extract_add_incomplete_field (RecWord *p)
        {
            remain = p->length - (b - p->string);
            if (remain > 0)
-               map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+               map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
            else
                map = 0;
        }
@@ -1629,7 +1629,7 @@ static void extract_add_incomplete_field (RecWord *p)
                buf[i++] = *(cp++);
            remain = p->length - (b - p->string);
            if (remain > 0)
-               map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+               map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
            else
                map = 0;
        }
@@ -1646,17 +1646,24 @@ static void extract_add_complete_field (RecWord *p)
     char buf[IT_MAX_WORD+1];
     const char **map = 0;
     int i = 0, remain = p->length;
+    int first; /* first position */
+
+yaz_log(LOG_DEBUG, "Complete field, w='%s'", p->string);
 
     if (remain > 0)
-       map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
+       map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1);
 
     while (remain > 0 && i < IT_MAX_WORD)
     {
        while (map && *map && **map == *CHR_SPACE)
        {
            remain = p->length - (b - p->string);
+
            if (remain > 0)
-               map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+           {
+               first = i ? 0 : 1;
+               map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first);
+           }
            else
                map = 0;
        }
@@ -1669,14 +1676,24 @@ static void extract_add_complete_field (RecWord *p)
        {
            const char *cp = *map;
 
-           if (i >= IT_MAX_WORD)
-               break;
-           while (i < IT_MAX_WORD && *cp)
-               buf[i++] = *(cp++);
+           if (**map == *CHR_CUT)
+           {
+               i = 0;
+           }
+           else
+           {
+               if (i >= IT_MAX_WORD)
+                   break;
+yaz_log(LOG_DEBUG, "Adding string to index '%d'", *map);
+               while (i < IT_MAX_WORD && *cp)
+                   buf[i++] = *(cp++);
+           }
            remain = p->length  - (b - p->string);
            if (remain > 0)
+           {
                map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
-                                       remain);
+                                       remain, 0);
+           }
            else
                map = 0;
        }