+static void trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ char *termz)
+{
+ size_t sizez;
+ Z_Term *term = zapt->term;
+
+ sizez = term->u.general->len;
+ if (sizez > IT_MAX_WORD-1)
+ sizez = IT_MAX_WORD-1;
+ memcpy (termz, term->u.general->buf, sizez);
+ termz[sizez] = '\0';
+}
+
+static void trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ char *termz, int reg_type)
+{
+ Z_Term *term = zapt->term;
+ const char **map;
+ const char *cp = (const char *) term->u.general->buf;
+ const char *cp_end = cp + term->u.general->len;
+ const char *src;
+ int i = 0;
+ const char *space_map = NULL;
+ int len;
+
+ while ((len = (cp_end - cp)) > 0)
+ {
+ map = zebra_maps_input (zh->zebra_maps, reg_type, &cp, len);
+ if (**map == *CHR_SPACE)
+ space_map = *map;
+ else
+ {
+ if (i && space_map)
+ for (src = space_map; *src; src++)
+ termz[i++] = *src;
+ space_map = NULL;
+ for (src = *map; *src; src++)
+ termz[i++] = *src;
+ }
+ }
+ termz[i] = '\0';
+}
+
+static RSET rpn_proximity (ZebraHandle zh, RSET rset1, RSET rset2,
+ int ordered,
+ int exclusion, int relation, int distance)
+{
+ int i;
+ RSFD rsfd1, rsfd2;
+ int more1, more2;
+ struct it_key buf1, buf2;
+ RSFD rsfd_result;
+ RSET result;
+ rset_temp_parms parms;
+ int term_index;
+
+ rsfd1 = rset_open (rset1, RSETF_READ);
+ more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
+
+ rsfd2 = rset_open (rset2, RSETF_READ);
+ more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
+
+ parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zh->res, "setTmpDir");
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd_result = rset_open (result, RSETF_WRITE);
+
+ logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d",
+ exclusion, ordered, relation, distance);
+ while (more1 && more2)
+ {
+ int cmp = key_compare_it (&buf1, &buf2);
+ if (cmp < -1)
+ more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
+ else if (cmp > 1)
+ more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
+ else
+ {
+ int sysno = buf1.sysno;
+ int seqno[500];
+ int n = 0;
+
+ seqno[n++] = buf1.seqno;
+ while ((more1 = rset_read (rset1, rsfd1, &buf1, &term_index)) &&
+ sysno == buf1.sysno)
+ if (n < 500)
+ seqno[n++] = buf1.seqno;
+ do
+ {
+ for (i = 0; i<n; i++)
+ {
+ int diff = buf2.seqno - seqno[i];
+ int excl = exclusion;
+ if (!ordered && diff < 0)
+ diff = -diff;
+ switch (relation)
+ {
+ case 1: /* < */
+ if (diff < distance)
+ excl = !excl;
+ break;
+ case 2: /* <= */
+ if (diff <= distance)
+ excl = !excl;
+ break;
+ case 3: /* == */
+ if (diff == distance)
+ excl = !excl;
+ break;
+ case 4: /* >= */
+ if (diff >= distance)
+ excl = !excl;
+ break;
+ case 5: /* > */
+ if (diff > distance)
+ excl = !excl;
+ break;
+ case 6: /* != */
+ if (diff != distance)
+ excl = !excl;
+ break;
+ }
+ if (excl)
+ rset_write (result, rsfd_result, &buf2);
+ }
+ } while ((more2 = rset_read (rset2, rsfd2, &buf2, &term_index)) &&
+ sysno == buf2.sysno);
+ }
+ }
+ rset_close (result, rsfd_result);
+ rset_close (rset1, rsfd1);
+ rset_close (rset2, rsfd2);
+ return result;
+}
+
+static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no)
+{
+ int i;
+ RSFD *rsfd;
+ int *more;
+ struct it_key **buf;
+ RSET result;
+ char prox_term[1024];
+ int length_prox_term = 0;
+ int min_nn = 10000000;
+ int term_index;
+ const char *flags = NULL;
+
+ rsfd = xmalloc (sizeof(*rsfd)*rset_no);
+ more = xmalloc (sizeof(*more)*rset_no);
+ buf = xmalloc (sizeof(*buf)*rset_no);
+
+ for (i = 0; i<rset_no; i++)
+ {
+ int j;
+ buf[i] = xmalloc (sizeof(**buf));
+ rsfd[i] = rset_open (rset[i], RSETF_READ);
+ if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i], &term_index)))
+ break;
+ for (j = 0; j<rset[i]->no_rset_terms; j++)
+ {
+ const char *nflags = rset[i]->rset_terms[j]->flags;
+ char *term = rset[i]->rset_terms[j]->name;
+ int lterm = strlen(term);
+ if (length_prox_term)
+ prox_term[length_prox_term++] = ' ';
+ strcpy (prox_term + length_prox_term, term);
+ length_prox_term += lterm;
+ if (min_nn > rset[i]->rset_terms[j]->nn)
+ min_nn = rset[i]->rset_terms[j]->nn;
+ flags = nflags;
+ }
+ }
+ if (i != rset_no)
+ {
+ rset_null_parms parms;
+
+ while (i >= 0)
+ {
+ rset_close (rset[i], rsfd[i]);
+ xfree (buf[i]);
+ --i;
+ }
+ parms.rset_term = rset_term_create (prox_term, -1, flags);
+ parms.rset_term->nn = 0;
+ result = rset_create (rset_kind_null, &parms);
+ }
+ else
+ {
+ rset_temp_parms parms;
+ RSFD rsfd_result;
+
+ parms.rset_term = rset_term_create (prox_term, -1, flags);
+ parms.rset_term->nn = min_nn;
+ parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zh->res, "setTmpDir");
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd_result = rset_open (result, RSETF_WRITE);
+
+ while (*more)
+ {
+ for (i = 1; i<rset_no; i++)
+ {
+ int cmp;
+
+ if (!more[i])
+ {
+ *more = 0;
+ break;
+ }
+ cmp = key_compare_it (buf[i], buf[i-1]);
+ if (cmp > 1)
+ {
+ more[i-1] = rset_read (rset[i-1], rsfd[i-1],
+ buf[i-1], &term_index);
+ break;
+ }
+ else if (cmp == 1)
+ {
+ if (buf[i-1]->seqno+1 != buf[i]->seqno)
+ {
+ more[i-1] = rset_read (rset[i-1], rsfd[i-1],
+ buf[i-1], &term_index);
+ break;
+ }
+ }
+ else
+ {
+ more[i] = rset_read (rset[i], rsfd[i], buf[i],
+ &term_index);
+ break;
+ }
+ }
+ if (i == rset_no)
+ {
+ rset_write (result, rsfd_result, buf[0]);
+ more[0] = rset_read (*rset, *rsfd, *buf, &term_index);
+ }
+ }
+
+ for (i = 0; i<rset_no; i++)
+ {
+ rset_close (rset[i], rsfd[i]);
+ xfree (buf[i]);
+ }
+ rset_close (result, rsfd_result);
+ }
+ xfree (buf);
+ xfree (more);
+ xfree (rsfd);
+ return result;
+}
+
+static RSET rpn_search_APT_phrase (ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ const char *termz,
+ oid_value attributeSet,
+ int reg_type, int complete_flag,
+ const char *rank_type,
+ int num_bases, char **basenames)
+{
+ char term_dst[IT_MAX_WORD+1];
+ const char *termp = termz;
+ RSET rset[60], result;
+ int i, r, rset_no = 0;
+ struct grep_info grep_info;
+
+#ifdef TERM_COUNT
+ grep_info.term_no = 0;
+#endif
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
+ grep_info.zh = zh;
+ grep_info.reg_type = reg_type;
+
+ while (1)
+ {
+ logf (LOG_DEBUG, "APT_phrase termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ r = string_term (zh, zapt, &termp, attributeSet, &grep_info,
+ reg_type, complete_flag, num_bases, basenames,
+ term_dst);
+ if (r < 1)
+ break;
+ logf (LOG_DEBUG, "term: %s", term_dst);
+ rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
+ grep_info.isam_p_indx, term_dst,
+ strlen(term_dst), rank_type);
+ assert (rset[rset_no]);
+ if (++rset_no >= sizeof(rset)/sizeof(*rset))
+ break;
+ }
+#ifdef TERM_COUNT
+ xfree(grep_info.term_no);
+#endif
+ xfree (grep_info.isam_p_buf);
+ if (rset_no == 0)
+ {
+ rset_null_parms parms;
+
+ parms.rset_term = rset_term_create (term_dst, -1, rank_type);
+ return rset_create (rset_kind_null, &parms);
+ }
+ else if (rset_no == 1)
+ return (rset[0]);
+ result = rpn_prox (zh, rset, rset_no);
+ for (i = 0; i<rset_no; i++)
+ rset_delete (rset[i]);
+ return result;
+}
+
+static RSET rpn_search_APT_or_list (ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ const char *termz,
+ oid_value attributeSet,
+ int reg_type, int complete_flag,
+ const char *rank_type,
+ int num_bases, char **basenames)
+{
+ char term_dst[IT_MAX_WORD+1];
+ const char *termp = termz;
+ RSET rset[60], result;
+ int i, r, rset_no = 0;
+ struct grep_info grep_info;
+
+#ifdef TERM_COUNT
+ grep_info.term_no = 0;
+#endif
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
+ grep_info.zh = zh;
+ grep_info.reg_type = reg_type;
+
+ while (1)
+ {
+ logf (LOG_DEBUG, "APT_or_list termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ r = string_term (zh, zapt, &termp, attributeSet, &grep_info,
+ reg_type, complete_flag, num_bases, basenames,
+ term_dst);
+ if (r < 1)
+ break;
+ logf (LOG_DEBUG, "term: %s", term_dst);
+ rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
+ grep_info.isam_p_indx, term_dst,
+ strlen(term_dst), rank_type);
+ assert (rset[rset_no]);
+ if (++rset_no >= sizeof(rset)/sizeof(*rset))
+ break;
+ }
+#ifdef TERM_COUNT
+ xfree(grep_info.term_no);
+#endif
+ xfree (grep_info.isam_p_buf);
+ if (rset_no == 0)
+ {
+ rset_null_parms parms;
+
+ parms.rset_term = rset_term_create (term_dst, -1, rank_type);
+ return rset_create (rset_kind_null, &parms);
+ }
+ result = rset[0];
+ for (i = 1; i<rset_no; i++)
+ {
+ rset_bool_parms bool_parms;
+
+ bool_parms.rset_l = result;
+ bool_parms.rset_r = rset[i];
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare_it;
+ result = rset_create (rset_kind_or, &bool_parms);
+ }
+ return result;
+}
+
+static RSET rpn_search_APT_and_list (ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ const char *termz,
+ oid_value attributeSet,
+ int reg_type, int complete_flag,
+ const char *rank_type,
+ int num_bases, char **basenames)
+{
+ char term_dst[IT_MAX_WORD+1];
+ const char *termp = termz;
+ RSET rset[60], result;
+ int i, r, rset_no = 0;
+ struct grep_info grep_info;
+
+#ifdef TERM_COUNT
+ grep_info.term_no = 0;
+#endif
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
+ grep_info.zh = zh;
+ grep_info.reg_type = reg_type;
+
+ while (1)
+ {
+ logf (LOG_DEBUG, "APT_and_list termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ r = string_term (zh, zapt, &termp, attributeSet, &grep_info,
+ reg_type, complete_flag, num_bases, basenames,
+ term_dst);
+ if (r < 1)
+ break;
+ logf (LOG_DEBUG, "term: %s", term_dst);
+ rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
+ grep_info.isam_p_indx, term_dst,
+ strlen(term_dst), rank_type);
+ assert (rset[rset_no]);
+ if (++rset_no >= sizeof(rset)/sizeof(*rset))
+ break;
+ }
+#ifdef TERM_COUNT
+ xfree(grep_info.term_no);
+#endif
+ xfree (grep_info.isam_p_buf);
+ if (rset_no == 0)
+ {
+ rset_null_parms parms;
+
+ parms.rset_term = rset_term_create (term_dst, -1, rank_type);
+ return rset_create (rset_kind_null, &parms);
+ }
+ result = rset[0];
+ for (i = 1; i<rset_no; i++)
+ {
+ rset_bool_parms bool_parms;
+
+ bool_parms.rset_l = result;
+ bool_parms.rset_r = rset[i];
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare_it;
+ result = rset_create (rset_kind_and, &bool_parms);
+ }
+ return result;
+}
+
+static int numeric_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char **term_sub,
+ char *term_dict,
+ oid_value attributeSet,
+ struct grep_info *grep_info,
+ int *max_pos,
+ int reg_type,
+ char *term_dst)
+{
+ AttrType relation;
+ int relation_value;
+ int term_value;
+ int r;
+ char *term_tmp = term_dict + strlen(term_dict);
+
+ attr_init (&relation, zapt, 2);
+ relation_value = attr_find (&relation, NULL);
+
+ logf (LOG_DEBUG, "numeric relation value=%d", relation_value);
+
+ if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
+ term_dst))
+ return 0;
+ term_value = atoi (term_tmp);
+ switch (relation_value)
+ {
+ case 1:
+ logf (LOG_DEBUG, "Relation <");
+ gen_regular_rel (term_tmp, term_value-1, 1);
+ break;
+ case 2:
+ logf (LOG_DEBUG, "Relation <=");
+ gen_regular_rel (term_tmp, term_value, 1);
+ break;
+ case 4:
+ logf (LOG_DEBUG, "Relation >=");
+ gen_regular_rel (term_tmp, term_value, 0);
+ break;
+ case 5:
+ logf (LOG_DEBUG, "Relation >");
+ gen_regular_rel (term_tmp, term_value+1, 0);
+ break;
+ case 3:
+ default:
+ logf (LOG_DEBUG, "Relation =");
+ sprintf (term_tmp, "(0*%d)", term_value);
+ }
+ logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp);
+ r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info, max_pos,
+ 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
+ logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
+ return 1;
+}
+
+static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char **term_sub,
+ oid_value attributeSet, struct grep_info *grep_info,
+ int reg_type, int complete_flag,
+ int num_bases, char **basenames,
+ char *term_dst)
+{
+ char term_dict[2*IT_MAX_WORD+2];
+ int r, base_no;
+ AttrType use;
+ int use_value;
+ oid_value curAttributeSet = attributeSet;
+ const char *termp;
+ struct rpn_char_map_info rcmi;
+
+ rpn_char_map_prepare (zh, reg_type, &rcmi);
+ attr_init (&use, zapt, 1);
+ use_value = attr_find (&use, &curAttributeSet);
+ logf (LOG_DEBUG, "numeric_term, use value %d", use_value);
+
+ if (use_value == -1)
+ use_value = 1016;
+
+ for (base_no = 0; base_no < num_bases; base_no++)
+ {
+ attent attp;
+ data1_local_attribute *local_attr;
+ int max_pos, prefix_len = 0;
+
+ termp = *term_sub;
+ if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value)))
+ {
+ logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
+ curAttributeSet, use_value, r);
+ if (r == -1)
+ zh->errCode = 114;
+ else
+ zh->errCode = 121;
+ return -1;
+ }
+ if (zebraExplain_curDatabase (zh->zei, basenames[base_no]))
+ {
+ zh->errCode = 109; /* Database unavailable */
+ zh->errString = basenames[base_no];
+ return -1;
+ }
+ for (local_attr = attp.local_attributes; local_attr;
+ local_attr = local_attr->next)
+ {
+ int ord;
+ char ord_buf[32];
+ int i, ord_len;
+
+ ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal,
+ local_attr->local);
+ if (ord < 0)
+ continue;
+ if (prefix_len)
+ term_dict[prefix_len++] = '|';
+ else
+ term_dict[prefix_len++] = '(';
+
+ ord_len = key_SU_code (ord, ord_buf);
+ for (i = 0; i<ord_len; i++)
+ {
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = ord_buf[i];
+ }
+ }
+ if (!prefix_len)
+ {
+ zh->errCode = 114;
+ return -1;
+ }
+ term_dict[prefix_len++] = ')';
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = reg_type;
+ logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
+ term_dict[prefix_len] = '\0';
+ if (!numeric_relation (zh, zapt, &termp, term_dict,
+ attributeSet, grep_info, &max_pos, reg_type,
+ term_dst))
+ return 0;
+ }
+ *term_sub = termp;
+ logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
+ return 1;
+}
+
+static RSET rpn_search_APT_numeric (ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ const char *termz,
+ oid_value attributeSet,
+ int reg_type, int complete_flag,
+ const char *rank_type,
+ int num_bases, char **basenames)
+{
+ char term_dst[IT_MAX_WORD+1];
+ const char *termp = termz;
+ RSET rset[60], result;
+ int i, r, rset_no = 0;
+ struct grep_info grep_info;
+
+#ifdef TERM_COUNT
+ grep_info.term_no = 0;
+#endif
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
+ grep_info.zh = zh;
+ grep_info.reg_type = reg_type;
+
+ while (1)
+ {
+ logf (LOG_DEBUG, "APT_numeric termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ r = numeric_term (zh, zapt, &termp, attributeSet, &grep_info,
+ reg_type, complete_flag, num_bases, basenames,
+ term_dst);
+ if (r < 1)
+ break;
+ logf (LOG_DEBUG, "term: %s", term_dst);
+ rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
+ grep_info.isam_p_indx, term_dst,
+ strlen(term_dst), rank_type);
+ assert (rset[rset_no]);
+ if (++rset_no >= sizeof(rset)/sizeof(*rset))
+ break;
+ }
+#ifdef TERM_COUNT
+ xfree(grep_info.term_no);
+#endif
+ xfree (grep_info.isam_p_buf);
+ if (rset_no == 0)
+ {
+ rset_null_parms parms;
+
+ parms.rset_term = rset_term_create (term_dst, -1, rank_type);
+ return rset_create (rset_kind_null, &parms);
+ }
+ result = rset[0];
+ for (i = 1; i<rset_no; i++)
+ {
+ rset_bool_parms bool_parms;
+
+ bool_parms.rset_l = result;
+ bool_parms.rset_r = rset[i];
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare_it;
+ result = rset_create (rset_kind_and, &bool_parms);
+ }
+ return result;
+}
+
+static RSET rpn_search_APT_local (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char *termz,
+ oid_value attributeSet,
+ const char *rank_type)
+{
+ RSET result;
+ RSFD rsfd;
+ struct it_key key;
+ rset_temp_parms parms;
+
+ parms.rset_term = rset_term_create (termz, -1, rank_type);
+ parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zh->res, "setTmpDir");
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd = rset_open (result, RSETF_WRITE);
+
+ key.sysno = atoi (termz);
+ key.seqno = 1;
+ if (key.sysno <= 0)
+ key.sysno = 1;
+ rset_write (result, rsfd, &key);
+ rset_close (result, rsfd);
+ return result;
+}
+
+static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet, NMEM stream,
+ Z_SortKeySpecList *sort_sequence,
+ const char *rank_type)
+{
+ rset_null_parms parms;
+ int i;
+ int sort_relation_value;
+ AttrType sort_relation_type;
+ int use_value;
+ AttrType use_type;
+ Z_SortKeySpec *sks;
+ Z_SortKey *sk;
+ Z_AttributeElement *ae;
+ int oid[OID_SIZE];
+ oident oe;
+
+ attr_init (&sort_relation_type, zapt, 7);
+ sort_relation_value = attr_find (&sort_relation_type, &attributeSet);
+
+ attr_init (&use_type, zapt, 1);
+ use_value = attr_find (&use_type, &attributeSet);
+
+ if (!sort_sequence->specs)
+ {
+ sort_sequence->num_specs = 10;
+ sort_sequence->specs = nmem_malloc (stream, sort_sequence->num_specs *
+ sizeof(*sort_sequence->specs));
+ for (i = 0; i<sort_sequence->num_specs; i++)
+ sort_sequence->specs[i] = 0;
+ }
+ if (zapt->term->which != Z_Term_general)
+ i = 0;
+ else
+ i = atoi_n (zapt->term->u.general->buf, zapt->term->u.general->len);
+ if (i >= sort_sequence->num_specs)
+ i = 0;
+
+ oe.proto = PROTO_Z3950;
+ oe.oclass = CLASS_ATTSET;
+ oe.value = attributeSet;
+ if (!oid_ent_to_oid (&oe, oid))
+ return 0;
+
+ sks = nmem_malloc (stream, sizeof(*sks));
+ sks->sortElement = nmem_malloc (stream, sizeof(*sks->sortElement));
+ sks->sortElement->which = Z_SortElement_generic;
+ sk = sks->sortElement->u.generic = nmem_malloc (stream, sizeof(*sk));
+ sk->which = Z_SortKey_sortAttributes;
+ sk->u.sortAttributes = nmem_malloc (stream, sizeof(*sk->u.sortAttributes));
+
+ sk->u.sortAttributes->id = oid;
+ sk->u.sortAttributes->list =
+ nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list));
+ sk->u.sortAttributes->list->num_attributes = 1;
+ sk->u.sortAttributes->list->attributes =
+ nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list->attributes));
+ ae = *sk->u.sortAttributes->list->attributes =
+ nmem_malloc (stream, sizeof(**sk->u.sortAttributes->list->attributes));
+ ae->attributeSet = 0;
+ ae->attributeType = nmem_malloc (stream, sizeof(*ae->attributeType));
+ *ae->attributeType = 1;
+ ae->which = Z_AttributeValue_numeric;
+ ae->value.numeric = nmem_malloc (stream, sizeof(*ae->value.numeric));
+ *ae->value.numeric = use_value;
+
+ sks->sortRelation = nmem_malloc (stream, sizeof(*sks->sortRelation));
+ if (sort_relation_value == 1)
+ *sks->sortRelation = Z_SortRelation_ascending;
+ else if (sort_relation_value == 2)
+ *sks->sortRelation = Z_SortRelation_descending;
+ else
+ *sks->sortRelation = Z_SortRelation_ascending;
+
+ sks->caseSensitivity = nmem_malloc (stream, sizeof(*sks->caseSensitivity));
+ *sks->caseSensitivity = 0;
+
+ sks->missingValueAction = 0;
+
+ sort_sequence->specs[i] = sks;
+
+ parms.rset_term = rset_term_create ("", -1, rank_type);
+ return rset_create (rset_kind_null, &parms);
+}
+
+
+static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet, NMEM stream,
+ Z_SortKeySpecList *sort_sequence,
+ int num_bases, char **basenames)
+{
+ unsigned reg_id;
+ char *search_type = NULL;
+ char *rank_type = NULL;
+ int complete_flag;
+ int sort_flag;
+ char termz[IT_MAX_WORD+1];
+
+ zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type,
+ &rank_type, &complete_flag, &sort_flag);
+
+ logf (LOG_DEBUG, "reg_id=%c", reg_id);
+ logf (LOG_DEBUG, "complete_flag=%d", complete_flag);
+ logf (LOG_DEBUG, "search_type=%s", search_type);
+ logf (LOG_DEBUG, "rank_type=%s", rank_type);
+
+ if (zapt->term->which != Z_Term_general)
+ {
+ zh->errCode = 124;
+ return NULL;
+ }
+ trans_term (zh, zapt, termz);
+
+ if (sort_flag)
+ return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence,
+ rank_type);
+
+ if (!strcmp (search_type, "phrase"))
+ {
+ return rpn_search_APT_phrase (zh, zapt, termz, attributeSet,
+ reg_id, complete_flag, rank_type,
+ num_bases, basenames);
+ }
+ else if (!strcmp (search_type, "and-list"))
+ {
+ return rpn_search_APT_and_list (zh, zapt, termz, attributeSet,
+ reg_id, complete_flag, rank_type,
+ num_bases, basenames);
+ }
+ else if (!strcmp (search_type, "or-list"))
+ {
+ return rpn_search_APT_or_list (zh, zapt, termz, attributeSet,
+ reg_id, complete_flag, rank_type,
+ num_bases, basenames);
+ }
+ else if (!strcmp (search_type, "local"))
+ {
+ return rpn_search_APT_local (zh, zapt, termz, attributeSet,
+ rank_type);
+ }
+ else if (!strcmp (search_type, "numeric"))
+ {
+ return rpn_search_APT_numeric (zh, zapt, termz, attributeSet,
+ reg_id, complete_flag, rank_type,
+ num_bases, basenames);
+ }
+ zh->errCode = 118;
+ return NULL;
+}
+
+static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs,
+ oid_value attributeSet, NMEM stream,
+ Z_SortKeySpecList *sort_sequence,
+ int num_bases, char **basenames)
+{
+ RSET r = NULL;
+ if (zs->which == Z_RPNStructure_complex)
+ {
+ Z_Operator *zop = zs->u.complex->roperator;
+ rset_bool_parms bool_parms;
+
+ bool_parms.rset_l = rpn_search_structure (zh, zs->u.complex->s1,
+ attributeSet, stream,
+ sort_sequence,
+ num_bases, basenames);
+ if (bool_parms.rset_l == NULL)
+ return NULL;
+ bool_parms.rset_r = rpn_search_structure (zh, zs->u.complex->s2,
+ attributeSet, stream,
+ sort_sequence,
+ num_bases, basenames);
+ if (bool_parms.rset_r == NULL)
+ {
+ rset_delete (bool_parms.rset_l);
+ return NULL;
+ }
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare_it;
+
+ switch (zop->which)
+ {
+ case Z_Operator_and:
+ r = rset_create (rset_kind_and, &bool_parms);
+ break;
+ case Z_Operator_or:
+ r = rset_create (rset_kind_or, &bool_parms);
+ break;
+ case Z_Operator_and_not:
+ r = rset_create (rset_kind_not, &bool_parms);
+ break;
+ case Z_Operator_prox:
+#ifdef ASN_COMPILED
+ if (zop->u.prox->which != Z_ProximityOperator_known)
+ {
+ zh->errCode = 132;
+ return NULL;
+ }
+#else
+ if (zop->u.prox->which != Z_ProxCode_known)
+ {
+ zh->errCode = 132;
+ return NULL;
+ }
+#endif
+
+#ifdef ASN_COMPILED
+ if (*zop->u.prox->u.known != Z_ProxUnit_word)
+ {
+ char *val = nmem_malloc (stream, 16);
+ zh->errCode = 132;
+ zh->errString = val;
+ sprintf (val, "%d", *zop->u.prox->u.known);
+ return NULL;
+ }
+#else
+ if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word)
+ {
+ char *val = nmem_malloc (stream, 16);
+ zh->errCode = 132;
+ zh->errString = val;
+ sprintf (val, "%d", *zop->u.prox->proximityUnitCode);
+ return NULL;
+ }
+#endif
+ r = rpn_proximity (zh, bool_parms.rset_l, bool_parms.rset_r,
+ *zop->u.prox->ordered,
+ (!zop->u.prox->exclusion ? 0 :
+ *zop->u.prox->exclusion),
+ *zop->u.prox->relationType,
+ *zop->u.prox->distance);
+ break;
+ default:
+ zh->errCode = 110;
+ return NULL;
+ }
+ }
+ else if (zs->which == Z_RPNStructure_simple)
+ {
+ if (zs->u.simple->which == Z_Operand_APT)
+ {
+ logf (LOG_DEBUG, "rpn_search_APT");
+ r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm,
+ attributeSet, stream, sort_sequence,
+ num_bases, basenames);
+ }
+ else if (zs->u.simple->which == Z_Operand_resultSetId)
+ {
+ logf (LOG_DEBUG, "rpn_search_ref");
+ r = resultSetRef (zh, zs->u.simple->u.resultSetId);
+ if (!r)
+ r = rset_create (rset_kind_null, NULL);
+ }
+ else
+ {
+ zh->errCode = 3;
+ return NULL;
+ }
+ }
+ else
+ {
+ zh->errCode = 3;
+ return NULL;
+ }
+ return r;
+}
+
+
+RSET rpn_search (ZebraHandle zh, NMEM nmem,
+ Z_RPNQuery *rpn, int num_bases, char **basenames,
+ const char *setname,
+ ZebraSet sset)
+{
+ RSET rset;
+ oident *attrset;
+ oid_value attributeSet;
+ Z_SortKeySpecList *sort_sequence;
+ int sort_status, i;
+
+ zh->errCode = 0;
+ zh->errString = NULL;
+ zh->hits = 0;
+
+ sort_sequence = nmem_malloc (nmem, sizeof(*sort_sequence));
+ sort_sequence->num_specs = 10;
+ sort_sequence->specs = nmem_malloc (nmem, sort_sequence->num_specs *
+ sizeof(*sort_sequence->specs));
+ for (i = 0; i<sort_sequence->num_specs; i++)
+ sort_sequence->specs[i] = 0;
+
+ attrset = oid_getentbyoid (rpn->attributeSetId);
+ attributeSet = attrset->value;
+ rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet,
+ nmem, sort_sequence, num_bases, basenames);
+ if (!rset)
+ return 0;
+
+ if (zh->errCode)
+ logf (LOG_DEBUG, "search error: %d", zh->errCode);
+
+ for (i = 0; sort_sequence->specs[i]; i++)
+ ;
+ sort_sequence->num_specs = i;
+ if (!i)
+ resultSetRank (zh, sset, rset);
+ else
+ {
+ logf (LOG_DEBUG, "resultSetSortSingle in rpn_search");
+ resultSetSortSingle (zh, nmem, sset, rset,
+ sort_sequence, &sort_status);
+ if (zh->errCode)
+ {
+ logf (LOG_DEBUG, "resultSetSortSingle status = %d", zh->errCode);
+ }
+ }
+ return rset;
+}
+
+struct scan_info_entry {
+ char *term;
+ ISAM_P isam_p;
+};
+
+struct scan_info {
+ struct scan_info_entry *list;
+ ODR odr;
+ int before, after;
+ char prefix[20];
+};
+
+static int scan_handle (char *name, const char *info, int pos, void *client)
+{
+ int len_prefix, idx;
+ struct scan_info *scan_info = client;
+
+ len_prefix = strlen(scan_info->prefix);
+ if (memcmp (name, scan_info->prefix, len_prefix))
+ return 1;
+ if (pos > 0)
+ idx = scan_info->after - pos + scan_info->before;
+ else
+ idx = - pos - 1;
+ scan_info->list[idx].term = odr_malloc (scan_info->odr,
+ strlen(name + len_prefix)+1);
+ strcpy (scan_info->list[idx].term, name + len_prefix);
+ assert (*info == sizeof(ISAM_P));
+ memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
+ return 0;
+}
+
+static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type,
+ char **dst, const char *src)
+{
+ char term_dst[1024];
+
+ term_untrans (zh, reg_type, term_dst, src);
+
+ *dst = nmem_malloc (stream, strlen(term_dst)+1);
+ strcpy (*dst, term_dst);
+}
+
+static void count_set (RSET r, int *count)
+{
+ int psysno = 0;
+ int kno = 0;
+ struct it_key key;
+ RSFD rfd;
+ int term_index;
+
+ logf (LOG_DEBUG, "count_set");
+
+ *count = 0;
+ rfd = rset_open (r, RSETF_READ);
+ while (rset_read (r, rfd, &key, &term_index))
+ {
+ if (key.sysno != psysno)
+ {
+ psysno = key.sysno;
+ (*count)++;
+ }
+ kno++;
+ }
+ rset_close (r, rfd);
+ logf (LOG_DEBUG, "%d keys, %d records", kno, *count);
+}
+
+void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
+ oid_value attributeset,
+ int num_bases, char **basenames,
+ int *position, int *num_entries, ZebraScanEntry **list,
+ int *is_partial)
+{
+ int i;
+ int pos = *position;
+ int num = *num_entries;
+ int before;
+ int after;
+ int base_no;
+ char termz[IT_MAX_WORD+20];
+ AttrType use;
+ int use_value;
+ struct scan_info *scan_info_array;
+ ZebraScanEntry *glist;
+ int ords[32], ord_no = 0;
+ int ptr[32];
+
+ unsigned reg_id;
+ char *search_type = NULL;
+ char *rank_type = NULL;
+ int complete_flag;
+ int sort_flag;
+
+ if (attributeset == VAL_NONE)
+ attributeset = VAL_BIB1;
+
+ zlog_scan (zapt, attributeset);
+ logf (LOG_DEBUG, "position = %d, num = %d", pos, num);
+
+ attr_init (&use, zapt, 1);
+ use_value = attr_find (&use, &attributeset);
+
+ if (zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type,
+ &rank_type, &complete_flag, &sort_flag))
+ {
+ zh->errCode = 113;
+ return ;
+ }
+
+ if (use_value == -1)
+ use_value = 1016;
+ for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
+ {
+ int r;
+ attent attp;
+ data1_local_attribute *local_attr;
+
+ if ((r=att_getentbyatt (zh, &attp, attributeset, use_value)))
+ {
+ logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
+ attributeset, use_value);
+ if (r == -1)
+ zh->errCode = 114;
+ else
+ zh->errCode = 121;
+ }
+ if (zebraExplain_curDatabase (zh->zei, basenames[base_no]))
+ {
+ zh->errString = basenames[base_no];
+ zh->errCode = 109; /* Database unavailable */
+ return;
+ }
+ for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
+ local_attr = local_attr->next)
+ {
+ int ord;
+
+ ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal,
+ local_attr->local);
+ if (ord > 0)
+ ords[ord_no++] = ord;
+ }
+ }
+ if (ord_no == 0)
+ {
+ zh->errCode = 113;
+ return;
+ }
+ /* prepare dictionary scanning */
+ before = pos-1;
+ after = 1+num-pos;
+ scan_info_array = odr_malloc (stream, ord_no * sizeof(*scan_info_array));
+ for (i = 0; i < ord_no; i++)
+ {
+ int j, prefix_len = 0;
+ int before_tmp = before, after_tmp = after;
+ struct scan_info *scan_info = scan_info_array + i;
+ struct rpn_char_map_info rcmi;
+
+ rpn_char_map_prepare (zh, reg_id, &rcmi);
+
+ scan_info->before = before;
+ scan_info->after = after;
+ scan_info->odr = stream;
+
+ scan_info->list = odr_malloc (stream, (before+after)*
+ sizeof(*scan_info->list));
+ for (j = 0; j<before+after; j++)
+ scan_info->list[j].term = NULL;
+
+ prefix_len += key_SU_code (ords[i], termz + prefix_len);
+ termz[prefix_len++] = reg_id;
+ termz[prefix_len] = 0;
+ strcpy (scan_info->prefix, termz);
+
+ trans_scan_term (zh, zapt, termz+prefix_len, reg_id);
+
+ dict_scan (zh->dict, termz, &before_tmp, &after_tmp, scan_info,
+ scan_handle);
+ }
+ glist = odr_malloc (stream, (before+after)*sizeof(*glist));
+
+ /* consider terms after main term */
+ for (i = 0; i < ord_no; i++)
+ ptr[i] = before;
+
+ *is_partial = 0;
+ for (i = 0; i<after; i++)
+ {
+ int j, j0 = -1;
+ const char *mterm = NULL;
+ const char *tst;
+ RSET rset;
+
+ for (j = 0; j < ord_no; j++)
+ {
+ if (ptr[j] < before+after &&
+ (tst=scan_info_array[j].list[ptr[j]].term) &&
+ (!mterm || strcmp (tst, mterm) < 0))
+ {
+ j0 = j;
+ mterm = tst;
+ }
+ }
+ if (j0 == -1)
+ break;
+ scan_term_untrans (zh, stream->mem, reg_id,
+ &glist[i+before].term, mterm);
+ rset = rset_trunc (zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
+ glist[i+before].term, strlen(glist[i+before].term),
+ NULL);
+
+ ptr[j0]++;
+ for (j = j0+1; j<ord_no; j++)
+ {
+ if (ptr[j] < before+after &&
+ (tst=scan_info_array[j].list[ptr[j]].term) &&
+ !strcmp (tst, mterm))
+ {
+ rset_bool_parms bool_parms;
+ RSET rset2;
+
+ rset2 =
+ rset_trunc (zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
+ glist[i+before].term,
+ strlen(glist[i+before].term), NULL);
+
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare_it;
+ bool_parms.rset_l = rset;
+ bool_parms.rset_r = rset2;
+
+ rset = rset_create (rset_kind_or, &bool_parms);
+
+ ptr[j]++;
+ }
+ }
+ count_set (rset, &glist[i+before].occurrences);
+ rset_delete (rset);
+ }
+ if (i < after)
+ {
+ *num_entries -= (after-i);
+ *is_partial = 1;
+ }
+
+ /* consider terms before main term */
+ for (i = 0; i<ord_no; i++)
+ ptr[i] = 0;
+
+ for (i = 0; i<before; i++)
+ {
+ int j, j0 = -1;
+ const char *mterm = NULL;
+ const char *tst;
+ RSET rset;
+
+ for (j = 0; j <ord_no; j++)
+ {
+ if (ptr[j] < before &&
+ (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
+ (!mterm || strcmp (tst, mterm) > 0))
+ {
+ j0 = j;
+ mterm = tst;
+ }
+ }
+ if (j0 == -1)
+ break;
+
+ scan_term_untrans (zh, stream->mem, reg_id,
+ &glist[before-1-i].term, mterm);
+
+ rset = rset_trunc
+ (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
+ glist[before-1-i].term, strlen(glist[before-1-i].term),
+ NULL);
+
+ ptr[j0]++;
+
+ for (j = j0+1; j<ord_no; j++)
+ {
+ if (ptr[j] < before &&
+ (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
+ !strcmp (tst, mterm))
+ {
+ rset_bool_parms bool_parms;
+ RSET rset2;
+
+ rset2 = rset_trunc (zh,
+ &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
+ glist[before-1-i].term,
+ strlen(glist[before-1-i].term), NULL);
+
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare_it;
+ bool_parms.rset_l = rset;
+ bool_parms.rset_r = rset2;
+
+ rset = rset_create (rset_kind_or, &bool_parms);
+
+ ptr[j]++;
+ }
+ }
+ count_set (rset, &glist[before-1-i].occurrences);
+ rset_delete (rset);
+ }
+ i = before-i;
+ if (i)
+ {
+ *is_partial = 1;
+ *position -= i;
+ *num_entries -= i;
+ }
+ *list = glist + i; /* list is set to first 'real' entry */
+
+ logf (LOG_DEBUG, "position = %d, num_entries = %d",
+ *position, *num_entries);
+ if (zh->errCode)
+ logf (LOG_DEBUG, "scan error: %d", zh->errCode);
+}
+