Work on xpath-like queries

[idzebra-moved-to-github.git] / recctrl / recgrs.c

diff --git a/recctrl/recgrs.c b/recctrl/recgrs.c
index 1e0c72a..0b21dea 100644 (file)
--- a/recctrl/recgrs.c
+++ b/recctrl/recgrs.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1994-2002, Index Data
  * All rights reserved.
  *
- * $Id: recgrs.c,v 1.53 2002-07-03 10:05:19 adam Exp $
+ * $Id: recgrs.c,v 1.56 2002-08-01 08:53:35 adam Exp $
  */
 
 #include <stdio.h>
@@ -105,6 +105,16 @@ static void grs_destroy(void *clientData)
     xfree (h);
 }
 
+/* use
+     1   start element (tag)
+     2   end element
+     3   start attr (and attr-exact)
+     4   end attr
+
+  1016   cdata
+  1015   attr data
+*/
+
 static void index_xpath (data1_node *n, struct recExtractCtrl *p,
                          int level, RecWord *wrd, int use)
 {
@@ -164,7 +174,59 @@ static void index_xpath (data1_node *n, struct recExtractCtrl *p,
         }
         else
         {
+            data1_xattr *xp;
             (*p->tokenAdd)(wrd);
+
+            for (xp = n->u.tag.attributes; xp; xp = xp->next)
+            {
+                char comb[512];
+                
+                if (use == 1)
+                {   /* attribute start */
+                    wrd->reg_type = '0';
+                    wrd->attrUse = 3;
+                    wrd->string = xp->name;
+                    wrd->length = strlen(xp->name);
+                    
+                    wrd->seqno--;
+                    (*p->tokenAdd)(wrd);
+                }
+                
+                if (use == 1 && 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);
+
+                    /* attribute value phrase */
+
+                    wrd->attrUse = 1015;
+                    wrd->reg_type = 'w';
+                    wrd->string = xp->value;
+                    wrd->length = strlen(xp->value);
+
+                    (*p->tokenAdd)(wrd);
+                }
+                if (use == 2)
+                {
+                    wrd->reg_type = '0';
+                    wrd->attrUse = 4;
+                    wrd->string = xp->name;
+                    wrd->length = strlen(xp->name);
+                    
+                    (*p->tokenAdd)(wrd);
+                }
+            }
         }
         break;
     }
@@ -358,6 +420,7 @@ int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
             (*p->schemaAdd)(p, oidtmp);
     }
     (*p->init)(p, &wrd);
+
     return dumpkeys(n, p, 0, &wrd);
 }
 
@@ -395,9 +458,14 @@ static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
         if ((oid_ent_to_oid (&oe, oidtmp)))
             (*p->schemaAdd)(p, oidtmp);
     }
+
+    /* ensure our data1 tree is UTF-8 */
+    data1_iconv (p->dh, mem, n, "UTF-8", data1_get_encoding(p->dh, n));
+
 #if 0
     data1_pr_tree (p->dh, n, stdout);
 #endif
+
     (*p->init)(p, &wrd);
     if (dumpkeys(n, p, 0, &wrd) < 0)
     {
@@ -555,6 +623,9 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
         nmem_destroy (mem);
        return 0;
     }
+    /* ensure our data1 tree is UTF-8 */
+    data1_iconv (p->dh, mem, node, "UTF-8", data1_get_encoding(p->dh, node));
+
 #if 0
     data1_pr_tree (p->dh, node, stdout);
 #endif
@@ -705,17 +776,19 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
     else if (p->comp && !res)
        selected = 1;
 
-#if 0
+#if 1
     data1_pr_tree (p->dh, node, stdout);
 #endif
     logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
     switch (p->output_format = (p->input_format != VAL_NONE ?
                                p->input_format : VAL_SUTRS))
     {
-       
     case VAL_TEXT_XML:
         add_idzebra_info (p, top, mem);
 
+        if (p->encoding)
+            data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
+
        if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
                                              &p->rec_len)))
            p->diagnostic = 238;
@@ -749,6 +822,8 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
            p->rec_len = (size_t) (-1);
        break;
     case VAL_SUTRS:
+        if (p->encoding)
+            data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
        if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
                                           &p->rec_len)))
            p->diagnostic = 238;
@@ -785,6 +860,8 @@ static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
            p->diagnostic = 238;
            break;
        }
+        if (p->encoding)
+            data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
        if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
                                        selected, &p->rec_len)))
            p->diagnostic = 238;