+ 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;
+
+ /* 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 ((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 ! attribute... */
+ if (!xpdone) {
+ wrd->attrSet = VAL_IDXPATH;
+ wrd->attrUse = use;
+ wrd->reg_type = 'w';
+ (*p->tokenAdd)(wrd);
+ }
+ }
+ break;
+ case DATA1N_tag:
+ flen = 0;
+ 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;
+ }
+
+
+ wrd->reg_type = '0';
+ wrd->string = tag_path_full;
+ wrd->length = flen;
+ wrd->attrSet = VAL_IDXPATH;
+ wrd->attrUse = use;
+ if (p->flagShowRecords)
+ {
+ printf("%*s tag=", (level + 1) * 4, "");
+ for (i = 0; i<wrd->length && i < 40; i++)
+ fputc (wrd->string[i], stdout);
+ if (i == 40)
+ printf (" ..");
+ printf("\n");
+ }
+ else
+ {
+ data1_xattr *xp;
+ (*p->tokenAdd)(wrd); /* index element pag (AKA tag path) */
+ if (use == 1)
+ {
+ for (xp = n->u.tag.attributes; xp; xp = xp->next)
+ {
+ char comb[512];
+ /* 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);
+ }
+ }
+ for (xp = n->u.tag.attributes; xp; xp = xp->next)
+ {
+ char attr_tag_path_full[1024];
+ int int_len = flen;
+
+ sprintf (attr_tag_path_full, "@%s/%.*s",
+ xp->name, int_len, tag_path_full);
+ wrd->reg_type = '0';
+ wrd->attrUse = 1;
+ wrd->string = attr_tag_path_full;
+ wrd->length = strlen(attr_tag_path_full);
+ (*p->tokenAdd)(wrd);
+
+ if (xp->value)
+ {
+ /* the same jokes, as with the data nodes ... */
+ data1_termlist *tl;
+ int xpdone = 0;
+
+ wrd->string = xp->value;
+ wrd->length = strlen(xp->value);
+ wrd->reg_type = 'w';
+
+ if ((tl = xpath_termlist_by_tagpath(attr_tag_path_full,
+ n))) {
+ for (; tl; tl = tl->next) {
+ wrd->reg_type = *tl->structure;
+ if (!tl->att) {
+ wrd->attrSet = VAL_IDXPATH;
+ wrd->attrUse = 1015;
+ (*p->tokenAdd)(wrd);
+ xpdone = 1;
+ } else {
+ wrd->attrSet = (int) (tl->att->parent->reference);
+ wrd->attrUse = tl->att->locals->local;
+ (*p->tokenAdd)(wrd);
+ }
+ }
+
+ }
+ if (!xpdone) {
+ wrd->attrSet = VAL_IDXPATH;
+ wrd->attrUse = 1015;
+ wrd->reg_type = 'w';
+ (*p->tokenAdd)(wrd);
+ }
+ }
+
+ wrd->attrSet = VAL_IDXPATH;
+ wrd->reg_type = '0';
+ wrd->attrUse = 2;
+ wrd->string = attr_tag_path_full;
+ wrd->length = strlen(attr_tag_path_full);
+ (*p->tokenAdd)(wrd);
+ }
+ }
+ }
+ }
+}
+
+static void index_termlist (data1_node *par, data1_node *n,
+ struct recExtractCtrl *p, int level, RecWord *wrd)
+{
+ data1_termlist *tlist = 0;
+ data1_datatype dtype = DATA1K_string;
+
+ /*
+ * cycle up towards the root until we find a tag with an att..
+ * this has the effect of indexing locally defined tags with
+ * the attribute of their ancestor in the record.
+ */
+
+ while (!par->u.tag.element)
+ if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
+ break;
+ if (!par || !(tlist = par->u.tag.element->termlists))
+ return;
+ if (par->u.tag.element->tag)
+ dtype = par->u.tag.element->tag->kind;
+
+ for (; tlist; tlist = tlist->next)
+ {
+
+ char xattr[512];
+ /* consider source */
+ wrd->string = 0;
+
+ if (!strcmp (tlist->source, "data") && n->which == DATA1N_data)