Fixup data1 tags which makes XML non-wellformed.. This effectively

but not elegantly fixes records produced by the grs.marc filter.
If a tag t is bad, it is changed to from <t> to <tag value="t">.

diff --git a/data1/d1_handle.c b/data1/d1_handle.c
index 723a081..2c5f853 100644 (file)
--- a/data1/d1_handle.c
+++ b/data1/d1_handle.c
@@ -1,4 +1,4 @@
-/* $Id: d1_handle.c,v 1.7 2005-01-15 19:38:18 adam Exp $
+/* $Id: d1_handle.c,v 1.8 2006-03-29 10:43:23 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -26,6 +26,8 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 #include <yaz/log.h>
 #include <idzebra/data1.h>
 
+#define DATA1_FLAG_XML  1
+
 struct data1_handle_info {
     WRBUF wrbuf;
     char *tab_path;
@@ -41,15 +43,9 @@ struct data1_handle_info {
     int map_len;
 
     NMEM mem;
-    int flags;
 };
 
-data1_handle data1_create (void)
-{
-    return data1_createx(0);
-}
-
-data1_handle data1_createx (int flags)
+data1_handle data1_create(void)
 {
     data1_handle p = (data1_handle)xmalloc (sizeof(*p));
     if (!p)
@@ -64,7 +60,6 @@ data1_handle data1_createx (int flags)
     p->absyn_cache = NULL;
     p->attset_cache = NULL;
     p->mem = nmem_create ();
-    p->flags = flags;
     return p;
 }
 
@@ -162,5 +157,5 @@ FILE *data1_path_fopen (data1_handle dh, const char *file, const char *mode)
 
 int data1_is_xmlmode(data1_handle dh)
 {
-    return dh->flags & DATA1_FLAG_XML;
+    return 1;
 }

diff --git a/data1/d1_write.c b/data1/d1_write.c
index cecc25f..cdaf1b8 100644 (file)
--- a/data1/d1_write.c
+++ b/data1/d1_write.c
@@ -1,4 +1,4 @@
-/* $Id: d1_write.c,v 1.6 2005-01-15 19:38:18 adam Exp $
+/* $Id: d1_write.c,v 1.7 2006-03-29 10:43:23 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -69,6 +69,31 @@ static void wrbuf_put_xattr(WRBUF b, data1_xattr *p)
     }
 }
 
+static void wrbuf_write_tag(WRBUF b, const char *tag, int opening)
+{
+    int i, fixup = 0;
+
+    /* see if we must fix the tag.. The grs.marc filter produces
+       a data1 tree with not well-formed XML */
+    if (*tag >= '0' && *tag <= '9')
+       fixup = 1;
+    for (i = 0; tag[i]; i++)
+       if (strchr( " <>$,()[]", tag[i]))
+           fixup = 1;
+    if (fixup)
+    {
+       wrbuf_puts(b, "tag");
+       if (opening)
+       {
+           wrbuf_puts(b, " value=\"");
+           wrbuf_xmlputs(b, tag);
+           wrbuf_puts(b, "\"");
+       }
+    }
+    else
+       wrbuf_puts(b, tag);
+}
+
 static int nodetoidsgml(data1_node *n, int select, WRBUF b, int col,
                         int pretty_format)
 {
@@ -106,8 +131,8 @@ static int nodetoidsgml(data1_node *n, int select, WRBUF b, int col,
            {
                 if (pretty_format)
                     indent (b, col);
-               wrbuf_puts (b, "<");    
-               wrbuf_xmlputs (b, tag);
+               wrbuf_puts(b, "<");
+               wrbuf_write_tag(b, tag, 1);
                 wrbuf_put_xattr (b, c->u.tag.attributes);
                wrbuf_puts(b, ">");
                 if (pretty_format)
@@ -118,7 +143,7 @@ static int nodetoidsgml(data1_node *n, int select, WRBUF b, int col,
                 if (pretty_format)
                     indent (b, col);
                wrbuf_puts(b, "</");
-               wrbuf_xmlputs(b, tag);
+               wrbuf_write_tag(b, tag, 0);
                wrbuf_puts(b, ">");
                 if (pretty_format)
                     wrbuf_puts (b, "\n");
@@ -211,7 +236,7 @@ char *data1_nodetoidsgml (data1_handle dh, data1_node *n, int select, int *len)
     if (!data1_is_xmlmode (dh))
     {
         wrbuf_puts (b, "<");
-        wrbuf_puts (b, n->u.root.type);
+        wrbuf_write_tag(b, n->u.root.type, 1);
         wrbuf_puts (b, ">\n");
     }
     if (nodetoidsgml(n, select, b, 0, 0 /* no pretty format */))
@@ -219,7 +244,7 @@ char *data1_nodetoidsgml (data1_handle dh, data1_node *n, int select, int *len)
     if (!data1_is_xmlmode (dh))
     {
         wrbuf_puts (b, "</");
-        wrbuf_puts (b, n->u.root.type);
+        wrbuf_write_tag(b, n->u.root.type, 0);
         wrbuf_puts (b, ">\n");
     }
     *len = wrbuf_len(b);

diff --git a/include/idzebra/data1.h b/include/idzebra/data1.h
index a470854..dfdcd03 100644 (file)
--- a/include/idzebra/data1.h
+++ b/include/idzebra/data1.h
@@ -1,4 +1,4 @@
-/* $Id: data1.h,v 1.10 2005-06-23 06:45:46 adam Exp $
+/* $Id: data1.h,v 1.11 2006-03-29 10:43:24 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -367,8 +367,6 @@ typedef struct data1_node
 
 YAZ_EXPORT data1_handle data1_create (void);
 
-#define DATA1_FLAG_XML  1
-YAZ_EXPORT data1_handle data1_createx (int flags);
 
 YAZ_EXPORT void data1_destroy(data1_handle dh);
 YAZ_EXPORT data1_node *get_parent_tag(data1_handle dh, data1_node *n);

diff --git a/index/zebraapi.c b/index/zebraapi.c
index e32e590..c91b5b2 100644 (file)
--- a/index/zebraapi.c
+++ b/index/zebraapi.c
@@ -1,4 +1,4 @@
-/* $Id: zebraapi.c,v 1.207 2006-03-28 12:39:07 adam Exp $
+/* $Id: zebraapi.c,v 1.208 2006-03-29 10:43:24 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -297,7 +297,7 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name,
     yaz_log (YLOG_DEBUG, "zebra_register_open rw=%d useshadow=%d p=%p n=%s rp=%s",
              rw, useshadow, reg, name, reg_path ? reg_path : "(none)");
     
-    reg->dh = data1_createx (DATA1_FLAG_XML);
+    reg->dh = data1_create();
     if (!reg->dh)
     {
        xfree(reg->name);