+ data1_absyn *abs = n->root->u.root.absyn;
+ data1_xpelement *xpe = abs->xp_elements;
+ data1_node *nn;
+#ifdef ENHANCED_XELM
+ struct xpath_location_step *xp;
+
+#endif
+ char *pexpr = malloc(strlen(tagpath)+2);
+ int ok = 0;
+
+ sprintf (pexpr, "%s\n", tagpath);
+ while (xpe)
+ {
+ struct DFA_state **dfaar = xpe->dfa->states;
+ struct DFA_state *s=dfaar[0];
+ struct DFA_tran *t;
+ const char *p;
+ int i;
+ unsigned char c;
+ int start_line = 1;
+
+ c = *pexpr++; t = s->trans; i = s->tran_no;
+ if (c >= t->ch[0] && c <= t->ch[1]) {
+ p = pexpr;
+ do {
+ if ((s = dfaar[t->to])->rule_no &&
+ (start_line || s->rule_nno)) {
+ ok = 1;
+ break;
+ }
+ for (t=s->trans, i=s->tran_no; --i >= 0; t++) {
+ if ((unsigned) *p >= t->ch[0] && (unsigned) *p <= t->ch[1])
+ break;
+ }
+ p++;
+ } while (i >= 0);
+ }
+ pexpr--;
+ if (ok) {
+#ifdef ENHANCED_XELM
+ /* we have to check the perdicates up to the root node */
+ xp = xpe->xpath;
+
+ /* find the first tag up in the node structure */
+ nn = n; while (nn && nn->which != DATA1N_tag) {
+ nn = nn->parent;
+ }
+
+ /* go from inside out in the node structure, while going
+ backwards trough xpath location steps ... */
+ for (i=xpe->xpath_len - 1; i>0; i--) {
+
+ logf(LOG_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");
+ ok = 0;
+ break;
+ }
+
+ if (nn->which == DATA1N_tag) {
+ nn = nn->parent;
+ }
+ }
+#endif
+ if (ok) {
+ break;
+ }
+ }
+ xpe = xpe->next;
+ }
+
+ if (ok) {
+ logf(LOG_DEBUG,"Got it");
+ return xpe->termlists;
+ } else {
+ return NULL;
+ }
+}
+
+/* use
+ 1 start element (tag)
+ 2 end element
+ 3 start attr (and attr-exact)
+ 4 end attr
+
+ 1016 cdata
+ 1015 attr data
+
+ *ostrich*
+
+ Now, if there is a matching xelm described in abs, for the
+ indexed element or the attribute, then the data is handled according
+ to those definitions...
+
+ modified by pop, 2002-12-13
+*/
+
+static void index_xpath (data1_node *n, struct recExtractCtrl *p,
+ int level, RecWord *wrd, int use)
+{
+ int i;
+ char tag_path_full[1024];
+ size_t flen = 0;
+ data1_node *nn;
+
+ switch (n->which)
+ {
+ 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);