X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=recctrl%2Frecgrs.c;h=1c34c2e72718193962ef3f0de2a09a18e62bf532;hp=568774f10502eb0f4867df52ce2099f498dbcded;hb=4eb3b54bb2ca9af74f39f000d3d40dba99ded887;hpb=bdf6fff005b3a0821fd6673bd441f79ffc97e81b

diff --git a/recctrl/recgrs.c b/recctrl/recgrs.c
index 568774f..1c34c2e 100644
--- a/recctrl/recgrs.c
+++ b/recctrl/recgrs.c
@@ -1,5 +1,5 @@
-/* $Id: recgrs.c,v 1.78 2003-04-25 08:57:36 adam Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
+/* $Id: recgrs.c,v 1.95 2004-12-13 20:51:32 adam Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
 This file is part of the Zebra server.
@@ -25,102 +25,147 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 #include <sys/types.h>
 #ifndef WIN32
 #include <unistd.h>
+#include <dlfcn.h>
 #endif
 
 #include <yaz/log.h>
 #include <yaz/oid.h>
 
-#include <recctrl.h>
-#include "grsread.h"
+#include <d1_absyn.h>
+#include <idzebra/recgrs.h>
 
 #define GRS_MAX_WORD 512
 
-struct grs_handler {
-    RecTypeGrs type;
-    void *clientData;
-    int initFlag;
-    struct grs_handler *next;
+struct source_parser {
+    int len;
+    const char *tok;
+    const char *src;
+    int lookahead;
 };
 
-struct grs_handlers {
-    struct grs_handler *handlers;
-};
-
-static int read_grs_type (struct grs_handlers *h,
-			  struct grs_read_info *p, const char *type,
-			  data1_node **root)
+static int sp_lex(struct source_parser *sp)
 {
-    struct grs_handler *gh = h->handlers;
-    const char *cp = strchr (type, '.');
-
-    if (cp == NULL || cp == type)
+    while (*sp->src == ' ')
+	(sp->src)++;
+    sp->tok = sp->src;
+    sp->len = 0;
+    while (*sp->src && !strchr("<>();,-: ", *sp->src))
     {
-        cp = strlen(type) + type;
-        *p->type = 0;
+	sp->src++;
+	sp->len++;
     }
+    if (sp->len)
+	sp->lookahead = 't';
     else
-        strcpy (p->type, cp+1);
-    for (gh = h->handlers; gh; gh = gh->next)
     {
-        if (!memcmp (type, gh->type->type, cp-type))
+	sp->lookahead = *sp->src;
+	if (*sp->src)
+	    sp->src++;
+    }
+    return sp->lookahead;
+}
+
+
+static int sp_expr(struct source_parser *sp, data1_node *n, RecWord *wrd)
+{
+    if (sp->lookahead != 't')
+	return 0;
+    if (sp->len == 4 && !memcmp(sp->tok, "data", sp->len))
+    {
+	if (n->which == DATA1N_data)
+	{
+	    wrd->string = n->u.data.data;
+	    wrd->length = n->u.data.len;
+	}
+	sp_lex(sp);
+    }
+    else if (sp->len == 3 && !memcmp(sp->tok, "tag", sp->len))
+    {
+	if (n->which == DATA1N_tag)
+	{		
+	    wrd->string = n->u.tag.tag;
+	    wrd->length = strlen(n->u.tag.tag);
+	}
+	sp_lex(sp);
+    }
+    else if (sp->len == 4 && !memcmp(sp->tok, "attr", sp->len))
+    {
+	sp_lex(sp);
+	if (sp->lookahead != '(')
+	    return 0;
+	sp_lex(sp);
+	if (sp->lookahead != 't')
+	    return 0;
+	
+	if (n->which == DATA1N_tag)
 	{
-	    if (!gh->initFlag)
+	    data1_xattr *p = n->u.tag.attributes;
+	    while (p && strlen(p->name) != sp->len && 
+		   memcmp (p->name, sp->tok, sp->len))
+		p = p->next;
+	    if (p)
 	    {
-		gh->initFlag = 1;
-		gh->clientData = (*gh->type->init)();
+		wrd->string = p->value;
+		wrd->length = strlen(p->value);
 	    }
-	    p->clientData = gh->clientData;
-            *root = (gh->type->read)(p);
-	    gh->clientData = p->clientData;
+	}
+	sp_lex(sp);
+	if (sp->lookahead != ')')
+	    return 0;
+	sp_lex(sp);
+    }
+    else if (sp->len == 5 && !memcmp(sp->tok, "range", sp->len))
+    {
+	int start, len;
+	sp_lex(sp);
+	if (sp->lookahead != '(')
+	    return 0;
+	
+	sp_lex(sp);
+	sp_expr(sp, n, wrd);
+	if (sp->lookahead != ',')
+	    return 0;
+	
+	sp_lex(sp);
+	if (sp->lookahead != 't')
 	    return 0;
+	start = atoi_n(sp->tok, sp->len);
+	
+	sp_lex(sp);
+	if (sp->lookahead != ',')
+	    return 0;
+	
+	sp_lex(sp);
+	if (sp->lookahead != 't')
+	    return 0;
+	len = atoi_n(sp->tok, sp->len);
+	
+	sp_lex(sp);
+	if (sp->lookahead != ')')
+	    return 0;
+	
+	sp_lex(sp);
+	if (wrd->string && wrd->length)
+	{
+	    wrd->string += start;
+	    wrd->length -= start;
+	    if (wrd->length > len)
+		wrd->length = len;
 	}
     }
     return 1;
 }
 
-static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t)
-{
-    struct grs_handler *gh = (struct grs_handler *) xmalloc (sizeof(*gh));
-    gh->next = h->handlers;
-    h->handlers = gh;
-    gh->initFlag = 0;
-    gh->clientData = 0;
-    gh->type = t;
-}
-
-static void *grs_init(RecType recType)
-{
-    struct grs_handlers *h = (struct grs_handlers *) xmalloc (sizeof(*h));
-    h->handlers = 0;
-
-    grs_add_handler (h, recTypeGrs_sgml);
-    grs_add_handler (h, recTypeGrs_regx);
-#if HAVE_TCL_H
-    grs_add_handler (h, recTypeGrs_tcl);
-#endif
-    grs_add_handler (h, recTypeGrs_marc);
-#if HAVE_EXPAT_H
-    grs_add_handler (h, recTypeGrs_xml);
-#endif
-#if HAVE_PERL
-    grs_add_handler (h, recTypeGrs_perl);
-#endif
-    return h;
-}
-
-static void grs_destroy(void *clientData)
+static int sp_parse(data1_node *n, RecWord *wrd, const char *src)
 {
-    struct grs_handlers *h = (struct grs_handlers *) clientData;
-    struct grs_handler *gh = h->handlers, *gh_next;
-    while (gh)
-    {
-	gh_next = gh->next;
-	if (gh->initFlag)
-	    (*gh->type->destroy)(gh->clientData);
-	xfree (gh);
-	gh = gh_next;
-    }
-    xfree (h);
+    struct source_parser sp;
+    sp.len = 0;
+    sp.tok = 0;
+    sp.src = src;
+    sp.lookahead = 0;
+    sp_lex(&sp);
+
+    return sp_expr(&sp, n, wrd);
 }
 
 int d1_check_xpath_predicate(data1_node *n, struct xpath_predicate *p)
@@ -135,26 +180,26 @@ int d1_check_xpath_predicate(data1_node *n, struct xpath_predicate *p)
         if (p->which == XPATH_PREDICATE_RELATION) {
             if (p->u.relation.name[0]) {
                 if (*p->u.relation.name != '@') {
-                    logf(LOG_WARN, 
+                    yaz_log(YLOG_WARN, 
                          "  Only attributes (@) are supported in xelm xpath predicates");
-                    logf(LOG_WARN, "predicate %s ignored", p->u.relation.name);
+                    yaz_log(YLOG_WARN, "predicate %s ignored", p->u.relation.name);
                     return (1);
                 }
                 attname = p->u.relation.name + 1;
                 res = 0;
                 /* looking for the attribute with a specified name */
                 for (attr = n->u.tag.attributes; attr; attr = attr->next) {
-                    logf(LOG_DEBUG,"  - attribute %s <-> %s", attname, attr->name );
+                    yaz_log(YLOG_DEBUG,"  - attribute %s <-> %s", attname, attr->name );
                     
                     if (!strcmp(attr->name, attname)) {
                         if (p->u.relation.op[0]) {
                             if (*p->u.relation.op != '=') {
-                                logf(LOG_WARN, 
+                                yaz_log(YLOG_WARN, 
                                      "Only '=' relation is supported (%s)",p->u.relation.op);
-                                logf(LOG_WARN, "predicate %s ignored", p->u.relation.name);
+                                yaz_log(YLOG_WARN, "predicate %s ignored", p->u.relation.name);
                                 res = 1; break;
                             } else {
-                                logf(LOG_DEBUG,"    - value %s <-> %s", 
+                                yaz_log(YLOG_DEBUG,"    - value %s <-> %s", 
                                      p->u.relation.value, attr->value );
                                 if (!strcmp(attr->value, p->u.relation.value)) {
                                     res = 1; break;
@@ -166,6 +211,7 @@ int d1_check_xpath_predicate(data1_node *n, struct xpath_predicate *p)
                         }
                     }
                 }
+		yaz_log(YLOG_DEBUG, "return %d", res);
                 return res;
             } else {
                 return 1;
@@ -180,7 +226,7 @@ int d1_check_xpath_predicate(data1_node *n, struct xpath_predicate *p)
                 return (d1_check_xpath_predicate(n, p->u.boolean.left) 
                         || d1_check_xpath_predicate(n, p->u.boolean.right)); 
             } else {
-                logf(LOG_WARN, "Unknown boolean relation %s, ignored",p->u.boolean.op);
+                yaz_log(YLOG_WARN, "Unknown boolean relation %s, ignored",p->u.boolean.op);
                 return 1;
             }
         }
@@ -219,7 +265,7 @@ data1_termlist *xpath_termlist_by_tagpath(char *tagpath, data1_node *n)
     int ok = 0;
     
     sprintf (pexpr, "%s\n", tagpath);
-    logf(LOG_DEBUG,"Checking tagpath %s",tagpath);
+    yaz_log(YLOG_DEBUG,"Checking tagpath %s",tagpath);
     while (xpe) 
     {
         struct DFA_state **dfaar = xpe->dfa->states;
@@ -230,8 +276,6 @@ data1_termlist *xpath_termlist_by_tagpath(char *tagpath, data1_node *n)
         unsigned char c;
         int start_line = 1;
 
-	logf(LOG_DEBUG," - xpath %s",xpe->xpath_expr);
-
         c = *pexpr++; t = s->trans; i = s->tran_no;
 	if ((c >= t->ch[0] && c <= t->ch[1]) || (!t->ch[0])) {
             p = pexpr;
@@ -248,6 +292,11 @@ data1_termlist *xpath_termlist_by_tagpath(char *tagpath, data1_node *n)
                 p++;
             } while (i >= 0);
 	}
+	if (ok)
+	    yaz_log(YLOG_DEBUG," xpath match %s",xpe->xpath_expr);
+	else
+	    yaz_log(YLOG_DEBUG," xpath no match %s",xpe->xpath_expr);
+
         pexpr--;
         if (ok) {
 #ifdef ENHANCED_XELM 
@@ -263,11 +312,11 @@ data1_termlist *xpath_termlist_by_tagpath(char *tagpath, data1_node *n)
                backwards trough xpath location steps ... */
             for (i=xpe->xpath_len - 1; i>0; i--) {
                 
-                logf(LOG_DEBUG,"Checking step %d: %s on tag %s",
+                yaz_log(YLOG_DEBUG,"Checking step %d: %s on tag %s",
 		     i,xp[i].part,nn->u.tag.tag);
                 
                 if (!d1_check_xpath_predicate(nn, xp[i].predicate)) {
-                    logf(LOG_DEBUG,"  Predicates didn't match");
+                    yaz_log(YLOG_DEBUG,"  Predicates didn't match");
                     ok = 0;
                     break;
                 }
@@ -287,7 +336,7 @@ data1_termlist *xpath_termlist_by_tagpath(char *tagpath, data1_node *n)
     xfree(pexpr);
     
     if (ok) {
-      logf(LOG_DEBUG,"Got it");
+      yaz_log(YLOG_DEBUG,"Got it");
         return xpe->termlists;
     } else {
         return NULL;
@@ -348,7 +397,10 @@ static void index_xpath (data1_node *n, struct recExtractCtrl *p,
     size_t flen = 0;
     data1_node *nn;
     int termlist_only = 1;
+    data1_termlist *tl;
+    int xpdone = 0;
 
+    yaz_log(YLOG_DEBUG, "index_xpath level=%d use=%d", level, use);
     if ((!n->root->u.root.absyn) ||
 	(n->root->u.root.absyn->enable_xpath_indexing)) {
       termlist_only = 0;
@@ -359,60 +411,88 @@ static void index_xpath (data1_node *n, struct recExtractCtrl *p,
     case DATA1N_data:
         wrd->string = n->u.data.data;
         wrd->length = n->u.data.len;
-        if (p->flagShowRecords)
-        {
-            printf("%*s data=", (level + 1) * 4, "");
-            for (i = 0; i<wrd->length && i < 8; i++)
-                fputc (wrd->string[i], stdout);
-            printf("\n");
-        }  
-        else  {
-            data1_termlist *tl;
-            int xpdone = 0;
-            flen = 0;
-            
-	    //	    logf (LOG_LOG,"Index value, %s",wrd->string);
-
-            /* we have to fetch the whole path to the data tag */
-            for (nn = n; nn; nn = nn->parent) {
-                if (nn->which == DATA1N_tag) {
-                    size_t tlen = strlen(nn->u.tag.tag);
-                    if (tlen + flen > (sizeof(tag_path_full)-2)) return;
-                    memcpy (tag_path_full + flen, nn->u.tag.tag, tlen);
-                    flen += tlen;
-                    tag_path_full[flen++] = '/';
-                }
-                else if (nn->which == DATA1N_root)  break;
-            }
-
-            tag_path_full[flen] = 0;
+        xpdone = 0;
+        flen = 0;
             
-            /* If we have a matching termlist... */
-            if ((tl = xpath_termlist_by_tagpath(tag_path_full, n))) {
-                for (; tl; tl = tl->next) {
-                    wrd->reg_type = *tl->structure;
-                    /* this is the ! case, so structure is for the xpath index */
-                    if (!tl->att) {
-                        wrd->attrSet = VAL_IDXPATH;
-                        wrd->attrUse = use;
-                        (*p->tokenAdd)(wrd);
-                        xpdone = 1;
-                        /* this is just the old fashioned attribute based index */
-                    } else {
-                        wrd->attrSet = (int) (tl->att->parent->reference);
-                        wrd->attrUse = tl->att->locals->local;
-                        (*p->tokenAdd)(wrd);
-                    }
-                }
-            }
-            /* xpath indexing is done, if there was no termlist given, 
-               or no ! in the termlist, and default indexing is enabled... */
-            if ((!xpdone) && (!termlist_only)) {
-                wrd->attrSet = VAL_IDXPATH;
-                wrd->attrUse = use;
-                wrd->reg_type = 'w';
-                (*p->tokenAdd)(wrd);
-            }
+	/* we have to fetch the whole path to the data tag */
+	for (nn = n; nn; nn = nn->parent) {
+	    if (nn->which == DATA1N_tag) {
+		size_t tlen = strlen(nn->u.tag.tag);
+		if (tlen + flen > (sizeof(tag_path_full)-2)) return;
+		memcpy (tag_path_full + flen, nn->u.tag.tag, tlen);
+		flen += tlen;
+		tag_path_full[flen++] = '/';
+	    }
+	    else if (nn->which == DATA1N_root)  break;
+	}
+	
+	tag_path_full[flen] = 0;
+	
+	/* If we have a matching termlist... */
+	if (n->root->u.root.absyn && 
+	    (tl = xpath_termlist_by_tagpath(tag_path_full, n)))
+	{
+	    for (; tl; tl = tl->next)
+	    {
+		/* need to copy recword because it may be changed */
+		RecWord wrd_tl;
+		wrd->reg_type = *tl->structure;
+		/* this is the ! case, so structure is for the xpath index */
+		memcpy (&wrd_tl, wrd, sizeof(*wrd));
+		if (tl->source)
+		    sp_parse(n, &wrd_tl, tl->source);
+		if (!tl->att) {
+		    wrd_tl.attrSet = VAL_IDXPATH;
+		    wrd_tl.attrUse = use;
+		    if (p->flagShowRecords)
+		    {
+			int i;
+		        printf("%*sXPath index", (level + 1) * 4, "");
+			printf (" XData:\"");
+			for (i = 0; i<wrd_tl.length && i < 40; i++)
+			    fputc (wrd_tl.string[i], stdout);
+			fputc ('"', stdout);
+			if (wrd_tl.length > 40)
+			    printf (" ...");
+			fputc ('\n', stdout);
+		    }
+		    else
+			(*p->tokenAdd)(&wrd_tl);
+		    xpdone = 1;
+		} else {
+		    /* this is just the old fashioned attribute based index */
+		    wrd_tl.attrSet = (int) (tl->att->parent->reference);
+		    wrd_tl.attrUse = tl->att->locals->local;
+		    if (p->flagShowRecords)
+		    {
+			int i;
+			printf("%*sIdx: [%s]", (level + 1) * 4, "",
+			       tl->structure);
+			printf("%s:%s [%d] %s",
+			       tl->att->parent->name,
+			       tl->att->name, tl->att->value,
+			       tl->source);
+			printf (" XData:\"");
+			for (i = 0; i<wrd_tl.length && i < 40; i++)
+			    fputc (wrd_tl.string[i], stdout);
+			fputc ('"', stdout);
+			if (wrd_tl.length > 40)
+			    printf (" ...");
+			fputc ('\n', stdout);
+		    }
+		    else
+			(*p->tokenAdd)(&wrd_tl);
+		}
+	    }
+	}
+	/* xpath indexing is done, if there was no termlist given, 
+	   or no ! in the termlist, and default indexing is enabled... */
+	if (!p->flagShowRecords && !xpdone && !termlist_only)
+	{
+	    wrd->attrSet = VAL_IDXPATH;
+	    wrd->attrUse = use;
+	    wrd->reg_type = 'w';
+	    (*p->tokenAdd)(wrd);
 	}
         break;
     case DATA1N_tag:
@@ -452,136 +532,134 @@ static void index_xpath (data1_node *n, struct recExtractCtrl *p,
             data1_xattr *xp;
             data1_termlist *tl;
 	    int do_xpindex;
- 
+            
 	    tag_path_full[flen] = 0;
- 
+            
             /* Add tag start/end xpath index, only when there is a ! in the apropriate xelm
-	      directive, or default xpath indexing is enabled */
+               directive, or default xpath indexing is enabled */
 	    if (!(do_xpindex = 1 - termlist_only)) {
-	      if ((tl = xpath_termlist_by_tagpath(tag_path_full, n))) {
-		for (; tl; tl = tl->next) { if (!tl->att) {do_xpindex = 1;} }
-	      }
+                if ((tl = xpath_termlist_by_tagpath(tag_path_full, n))) {
+                    for (; tl; tl = tl->next) { if (!tl->att) {do_xpindex = 1;} }
+                }
 	    }
 	    if (do_xpindex) {
-	      (*p->tokenAdd)(wrd);   /* index element pag (AKA tag path) */
+                (*p->tokenAdd)(wrd);   /* index element pag (AKA tag path) */
 	    }
-
+            
             if (use == 1) /* only for the starting tag... */
             {
-
 #define MAX_ATTR_COUNT 50
-
-	      data1_termlist *tll[MAX_ATTR_COUNT];
-	      
-	      int i = 0;
-
-	      /* get termlists for attributes, and find out, if we have to do xpath indexing */
-	      for (xp = n->u.tag.attributes; xp; xp = xp->next) {
-                  i++;
-	      }
-              
-	      i = 0;
-	      for (xp = n->u.tag.attributes; xp; xp = xp->next) {
-                  char comb[512];
-                  int do_xpindex = 1 - termlist_only;
-                  data1_termlist *tl;
-                  char attr_tag_path_full[1024]; 
-                  int int_len = flen;
-                  
-                  /* this could be cached as well */
-                  sprintf (attr_tag_path_full, "@%s/%.*s",
-                           xp->name, int_len, tag_path_full);
-                  
-                  tll[i] = xpath_termlist_by_tagpath(attr_tag_path_full,n);
-                  
-                  /* if there is a ! in the xelm termlist, or default indexing is on, 
-                     proceed with xpath idx */
-                  if ((tl = tll[i]))
-                  {
-                      for (; tl; tl = tl->next)
-                      { 
-                          if (!tl->att)
-                              do_xpindex = 1;
-                      }
-                  }
-                  
-                  if (do_xpindex) {
-                      
-                      /* attribute  (no value) */
-                      wrd->reg_type = '0';
-                      wrd->attrUse = 3;
-                      wrd->string = xp->name;
-                      wrd->length = strlen(xp->name);
-                      
-                      wrd->seqno--;
-                      (*p->tokenAdd)(wrd);
-                      
-                      if (xp->value &&
-                          strlen(xp->name) + strlen(xp->value) < sizeof(comb)-2) {
-                          
-                          /* attribute value exact */
-                          strcpy (comb, xp->name);
-                          strcat (comb, "=");
-                          strcat (comb, xp->value);
-                          
-                          wrd->attrUse = 3;
-                          wrd->reg_type = '0';
-                          wrd->string = comb;
-                          wrd->length = strlen(comb);
-                          wrd->seqno--;
-                          
-                          (*p->tokenAdd)(wrd);
-                      }
-                  }                
-                  i++;
-	      }
-              
-	      i = 0;
-	      for (xp = n->u.tag.attributes; xp; xp = xp->next) {
-                  data1_termlist *tl;
-                  char attr_tag_path_full[1024];
-                  int int_len = flen;
-                  int xpdone = 0;
-                  
-                  sprintf (attr_tag_path_full, "@%s/%.*s",
-                           xp->name, int_len, tag_path_full);
-                  
-                  if ((tl = tll[i]))
-                  {
-                      /* If there is a termlist given (=xelm directive) */
-                      for (; tl; tl = tl->next)
-                      {
-                          if (!tl->att) {
-                              /* add xpath index for the attribute */
-                              index_xpath_attr (attr_tag_path_full, xp->name,
-                                                xp->value, tl->structure,
-                                                p, wrd);
-                              xpdone = 1;
-                          } else {
-                              /* add attribute based index for the attribute */
-                              if (xp->value) {
-                                  wrd->attrSet = (int) 
-                                      (tl->att->parent->reference);
-                                  wrd->attrUse = tl->att->locals->local;
-                                  wrd->reg_type = *tl->structure;
-                                  wrd->string = xp->value;
-                                  wrd->length = strlen(xp->value);
-                                  (*p->tokenAdd)(wrd);
-                              }
-                          }
-                      }
-                  }
-                  /* if there was no termlist for the given path, 
-                     or the termlist didn't have a ! element, index 
-                     the attribute as "w" */
-                  if ((!xpdone) && (!termlist_only))
-                  {
-                      index_xpath_attr (attr_tag_path_full, xp->name,
-                                        xp->value,  "w", p, wrd);
-                  }
-                  i++;
-	      }
-	    }
+                data1_termlist *tll[MAX_ATTR_COUNT];
+                
+                int i = 0;
+                
+                /* get termlists for attributes, and find out, if we have to do xpath indexing */
+                for (xp = n->u.tag.attributes; xp; xp = xp->next) {
+                    i++;
+                }
+                
+                i = 0;
+                for (xp = n->u.tag.attributes; xp; xp = xp->next) {
+                    char comb[512];
+                    int do_xpindex = 1 - termlist_only;
+                    data1_termlist *tl;
+                    char attr_tag_path_full[1024]; 
+                    int int_len = flen;
+                    
+                    /* this could be cached as well */
+                    sprintf (attr_tag_path_full, "@%s/%.*s",
+                             xp->name, int_len, tag_path_full);
+                    
+                    tll[i] = xpath_termlist_by_tagpath(attr_tag_path_full,n);
+                    
+                    /* if there is a ! in the xelm termlist, or default indexing is on, 
+                       proceed with xpath idx */
+                    if ((tl = tll[i]))
+                    {
+                        for (; tl; tl = tl->next)
+                        { 
+                            if (!tl->att)
+                                do_xpindex = 1;
+                        }
+                    }
+                    
+                    if (do_xpindex) {
+                        
+                        /* attribute  (no value) */
+                        wrd->reg_type = '0';
+                        wrd->attrUse = 3;
+                        wrd->string = xp->name;
+                        wrd->length = strlen(xp->name);
+                        
+                        wrd->seqno--;
+                        (*p->tokenAdd)(wrd);
+                        
+                        if (xp->value &&
+                            strlen(xp->name) + strlen(xp->value) < sizeof(comb)-2) {
+                            
+                            /* attribute value exact */
+                            strcpy (comb, xp->name);
+                            strcat (comb, "=");
+                            strcat (comb, xp->value);
+                            
+                            wrd->attrUse = 3;
+                            wrd->reg_type = '0';
+                            wrd->string = comb;
+                            wrd->length = strlen(comb);
+                            wrd->seqno--;
+                            
+                            (*p->tokenAdd)(wrd);
+                        }
+                    }                
+                    i++;
+                }
+                
+                i = 0;
+                for (xp = n->u.tag.attributes; xp; xp = xp->next) {
+                    data1_termlist *tl;
+                    char attr_tag_path_full[1024];
+                    int int_len = flen;
+                    int xpdone = 0;
+                    
+                    sprintf (attr_tag_path_full, "@%s/%.*s",
+                             xp->name, int_len, tag_path_full);
+                    
+                    if ((tl = tll[i]))
+                    {
+                        /* If there is a termlist given (=xelm directive) */
+                        for (; tl; tl = tl->next)
+                        {
+                            if (!tl->att) {
+                                /* add xpath index for the attribute */
+                                index_xpath_attr (attr_tag_path_full, xp->name,
+                                                  xp->value, tl->structure,
+                                                  p, wrd);
+                                xpdone = 1;
+                            } else {
+                                /* add attribute based index for the attribute */
+                                if (xp->value) {
+                                    wrd->attrSet = (int) 
+                                        (tl->att->parent->reference);
+                                    wrd->attrUse = tl->att->locals->local;
+                                    wrd->reg_type = *tl->structure;
+                                    wrd->string = xp->value;
+                                    wrd->length = strlen(xp->value);
+                                    (*p->tokenAdd)(wrd);
+                                }
+                            }
+                        }
+                    }
+                    /* if there was no termlist for the given path, 
+                       or the termlist didn't have a ! element, index 
+                       the attribute as "w" */
+                    if ((!xpdone) && (!termlist_only))
+                    {
+                        index_xpath_attr (attr_tag_path_full, xp->name,
+                                          xp->value,  "w", p, wrd);
+                    }
+                    i++;
+                }
+            }
 	}
     }
 }
@@ -608,33 +686,11 @@ static void index_termlist (data1_node *par, data1_node *n,
     
     for (; tlist; tlist = tlist->next)
     {
-
-	char xattr[512];
 	/* consider source */
 	wrd->string = 0;
+	assert(tlist->source);
+	sp_parse(n, wrd, tlist->source);
 
-	if (!strcmp (tlist->source, "data") && n->which == DATA1N_data)
-	{
-	    wrd->string = n->u.data.data;
-	    wrd->length = n->u.data.len;
-	}
-	else if (!strcmp (tlist->source, "tag") && n->which == DATA1N_tag)
-        {
-	    wrd->string = n->u.tag.tag;
-	    wrd->length = strlen(n->u.tag.tag);
-	}
-	else if (sscanf (tlist->source, "attr(%511[^)])", xattr) == 1 &&
-	    n->which == DATA1N_tag)
-	{
-	    data1_xattr *p = n->u.tag.attributes;
-	    while (p && strcmp (p->name, xattr))
-		p = p->next;
-	    if (p)
-	    {
-		wrd->string = p->value;
-		wrd->length = strlen(p->value);
-	    }
-	}
 	if (wrd->string)
 	{
 	    if (p->flagShowRecords)
@@ -646,7 +702,7 @@ static void index_termlist (data1_node *par, data1_node *n,
 		       tlist->att->parent->name,
 		       tlist->att->name, tlist->att->value,
 		       tlist->source);
-		printf (" data=\"");
+		printf (" XData:\"");
 		for (i = 0; i<wrd->length && i < 40; i++)
 		    fputc (wrd->string[i], stdout);
 		fputc ('"', stdout);
@@ -711,8 +767,8 @@ static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level,
 	{
             index_termlist (n, n, p, level, wrd);
             /* index start tag */
-            assert (n->root->u.root.absyn);
-      	    index_xpath (n, p, level, wrd, 1);
+	    if (n->root->u.root.absyn)
+      	        index_xpath (n, p, level, wrd, 1);
  	}
 
 	if (n->child)
@@ -729,8 +785,8 @@ static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level,
 		printf("%*s", level * 4, "");
 		printf("Data: ");
 		if (n->u.data.len > 256)
-		    printf("'%.240s ... %.6s'\n", n->u.data.data,
-			   n->u.data.data + n->u.data.len-6);
+		    printf("'%.170s ... %.70s'\n", n->u.data.data,
+			   n->u.data.data + n->u.data.len-70);
 		else if (n->u.data.len > 0)
 		    printf("'%.*s'\n", n->u.data.len, n->u.data.data);
 		else
@@ -777,8 +833,9 @@ int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
     return dumpkeys(n, p, 0, &wrd);
 }
 
-static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
-			   NMEM mem)
+static int grs_extract_sub(void *clientData, struct recExtractCtrl *p,
+			   NMEM mem,
+			   data1_node *(*grs_read)(struct grs_read_info *))
 {
     data1_node *n;
     struct grs_read_info gri;
@@ -794,9 +851,9 @@ static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
     gri.offset = p->offset;
     gri.mem = mem;
     gri.dh = p->dh;
+    gri.clientData = clientData;
 
-    if (read_grs_type (h, &gri, p->subType, &n))
-	return RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER;
+    n = (*grs_read)(&gri);
     if (!n)
         return RECCTRL_EXTRACT_EOF;
     oe.proto = PROTO_Z3950;
@@ -811,6 +868,7 @@ static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
         if ((oid_ent_to_oid (&oe, oidtmp)))
             (*p->schemaAdd)(p, oidtmp);
     }
+    data1_concat_text(p->dh, mem, n);
 
     /* ensure our data1 tree is UTF-8 */
     data1_iconv (p->dh, mem, n, "UTF-8", data1_get_encoding(p->dh, n));
@@ -829,13 +887,12 @@ static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
     return RECCTRL_EXTRACT_OK;
 }
 
-static int grs_extract(void *clientData, struct recExtractCtrl *p)
+int zebra_grs_extract(void *clientData, struct recExtractCtrl *p,
+		      data1_node *(*grs_read)(struct grs_read_info *))
 {
     int ret;
     NMEM mem = nmem_create ();
-    struct grs_handlers *h = (struct grs_handlers *) clientData;
-
-    ret = grs_extract_sub(h, p, mem);
+    ret = grs_extract_sub(clientData, p, mem, grs_read);
     nmem_destroy(mem);
     return ret;
 }
@@ -843,7 +900,8 @@ static int grs_extract(void *clientData, struct recExtractCtrl *p)
 /*
  * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
  */
-static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
+static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c,
+			char **addinfo, ODR o)
 {
     data1_esetname *eset;
     Z_Espec1 *espec = 0;
@@ -857,10 +915,11 @@ static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
 	if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
 					 c->u.simple->u.generic)))
 	{
-	    logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
+	    yaz_log(YLOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
+	    *addinfo = odr_strdup(o, c->u.simple->u.generic);
 	    return 25; /* invalid esetname */
 	}
-	logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
+	yaz_log(YLOG_DEBUG, "Esetname '%s' in simple compspec",
 	     c->u.simple->u.generic);
 	espec = eset->spec;
 	break;
@@ -877,23 +936,24 @@ static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
 			  data1_getesetbyname(dh, n->u.root.absyn,
 					      p->u.elementSetName)))
 		    {
-			logf(LOG_LOG, "Unknown esetname '%s'",
+			yaz_log(YLOG_DEBUG, "Unknown esetname '%s'",
 			     p->u.elementSetName);
+			*addinfo = odr_strdup(o, p->u.elementSetName);
 			return 25; /* invalid esetname */
 		    }
-		    logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
+		    yaz_log(YLOG_DEBUG, "Esetname '%s' in complex compspec",
 			 p->u.elementSetName);
 		    espec = eset->spec;
 		    break;
 		case Z_ElementSpec_externalSpec:
 		    if (p->u.externalSpec->which == Z_External_espec1)
 		    {
-			logf(LOG_DEBUG, "Got Espec-1");
+			yaz_log(YLOG_DEBUG, "Got Espec-1");
 			espec = p->u.externalSpec-> u.espec1;
 		    }
 		    else
 		    {
-			logf(LOG_LOG, "Unknown external espec.");
+			yaz_log(YLOG_LOG, "Unknown external espec.");
 			return 25; /* bad. what is proper diagnostic? */
 		    }
 		    break;
@@ -905,12 +965,12 @@ static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
     }
     if (espec)
     {
-        logf (LOG_DEBUG, "Element: Espec-1 match");
+        yaz_log(YLOG_DEBUG, "Element: Espec-1 match");
 	return data1_doespec1(dh, n, espec);
     }
     else
     {
-	logf (LOG_DEBUG, "Element: all match");
+	yaz_log(YLOG_DEBUG, "Element: all match");
 	return -1;
     }
 }
@@ -954,7 +1014,7 @@ static void zebra_xml_metadata (struct recRetrieveCtrl *p, data1_node *top,
         data1_mk_tag_data_int (p->dh, n, "score", p->score, mem);
     }
     data1_mk_text (p->dh, mem, i4, n);
-    data1_mk_tag_data_int (p->dh, n, "localnumber", p->localno, mem);
+    data1_mk_tag_data_zint (p->dh, n, "localnumber", p->localno, mem);
     if (p->fname)
     {
         data1_mk_text (p->dh, mem, i4, n);
@@ -963,7 +1023,8 @@ static void zebra_xml_metadata (struct recRetrieveCtrl *p, data1_node *top,
     data1_mk_text (p->dh, mem, i2, n);
 }
 
-static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
+int zebra_grs_retrieve(void *clientData, struct recRetrieveCtrl *p,
+		       data1_node *(*grs_read)(struct grs_read_info *))
 {
     data1_node *node = 0, *onode = 0, *top;
     data1_node *dnew;
@@ -972,7 +1033,7 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
     NMEM mem;
     struct grs_read_info gri;
     const char *tagname;
-    struct grs_handlers *h = (struct grs_handlers *) clientData;
+
     int requested_schema = VAL_NONE;
     data1_marctab *marctab;
     int dummy;
@@ -986,20 +1047,18 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
     gri.offset = 0;
     gri.mem = mem;
     gri.dh = p->dh;
+    gri.clientData = clientData;
 
-    logf (LOG_DEBUG, "grs_retrieve");
-    if (read_grs_type (h, &gri, p->subType, &node))
-    {
-	p->diagnostic = 14;
-        nmem_destroy (mem);
-	return 0;
-    }
+    yaz_log(YLOG_DEBUG, "grs_retrieve");
+    node = (*grs_read)(&gri);
     if (!node)
     {
 	p->diagnostic = 14;
         nmem_destroy (mem);
 	return 0;
     }
+    data1_concat_text(p->dh, mem, node);
+
     /* ensure our data1 tree is UTF-8 */
     data1_iconv (p->dh, mem, node, "UTF-8", data1_get_encoding(p->dh, node));
 
@@ -1008,7 +1067,7 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
 #endif
     top = data1_get_root_tag (p->dh, node);
 
-    logf (LOG_DEBUG, "grs_retrieve: size");
+    yaz_log(YLOG_DEBUG, "grs_retrieve: size");
     tagname = data1_systag_lookup(node->u.root.absyn, "size", "size");
     if (tagname &&
         (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem)))
@@ -1023,7 +1082,7 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
     if (tagname && p->score >= 0 &&
 	(dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem)))
     {
-        logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
+        yaz_log(YLOG_DEBUG, "grs_retrieve: %s", tagname);
 	dnew->u.data.what = DATA1I_num;
 	dnew->u.data.data = dnew->lbuf;
 	sprintf(dnew->u.data.data, "%d", p->score);
@@ -1035,13 +1094,17 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
     if (tagname && p->localno > 0 &&
         (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem)))
     {
-        logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
+        yaz_log(YLOG_DEBUG, "grs_retrieve: %s", tagname);
 	dnew->u.data.what = DATA1I_text;
 	dnew->u.data.data = dnew->lbuf;
         
-	sprintf(dnew->u.data.data, "%d", p->localno);
+	sprintf(dnew->u.data.data, ZINT_FORMAT, p->localno);
 	dnew->u.data.len = strlen(dnew->u.data.data);
     }
+
+    if (p->input_format == VAL_TEXT_XML)
+       zebra_xml_metadata (p, top, mem);
+
 #if 0
     data1_pr_tree (p->dh, node, stdout);
 #endif
@@ -1070,7 +1133,7 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
      */
     if (requested_schema != VAL_NONE)
     {
-	logf (LOG_DEBUG, "grs_retrieve: schema mapping");
+	yaz_log(YLOG_DEBUG, "grs_retrieve: schema mapping");
 	for (map = node->u.root.absyn->maptabs; map; map = map->next)
 	{
 	    if (map->target_absyn_ref == requested_schema)
@@ -1098,7 +1161,7 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
      * the overlap of schema and formatting which is inherent in the MARC
      * family)
      */
-    yaz_log (LOG_DEBUG, "grs_retrieve: syntax mapping");
+    yaz_log(YLOG_DEBUG, "grs_retrieve: syntax mapping");
     if (node->u.root.absyn)
         for (map = node->u.root.absyn->maptabs; map; map = map->next)
         {
@@ -1114,7 +1177,7 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
                 break;
             }
         }
-    yaz_log (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
+    yaz_log(YLOG_DEBUG, "grs_retrieve: schemaIdentifier");
     if (node->u.root.absyn &&
 	node->u.root.absyn->reference != VAL_NONE &&
 	p->input_format == VAL_GRS1)
@@ -1152,8 +1215,9 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
 	}
     }
 
-    logf (LOG_DEBUG, "grs_retrieve: element spec");
-    if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
+    yaz_log(YLOG_DEBUG, "grs_retrieve: element spec");
+    if (p->comp && (res = process_comp(p->dh, node, p->comp, &p->addinfo,
+				       p->odr)) > 0)
     {
 	p->diagnostic = res;
 	if (onode)
@@ -1168,12 +1232,11 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
 #if 0
     data1_pr_tree (p->dh, node, stdout);
 #endif
-    logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
+    yaz_log(YLOG_DEBUG, "grs_retrieve: transfer syntax mapping");
     switch (p->output_format = (p->input_format != VAL_NONE ?
 				p->input_format : VAL_SUTRS))
     {
     case VAL_TEXT_XML:
-        zebra_xml_metadata (p, top, mem);
 
 #if 0
         data1_pr_tree (p->dh, node, stdout);
@@ -1273,13 +1336,3 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
     return 0;
 }
 
-static struct recType grs_type =
-{
-    "grs",
-    grs_init,
-    grs_destroy,
-    grs_extract,
-    grs_retrieve
-};
-
-RecType recTypeGrs = &grs_type;