+static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no,
+ int ordered, int exclusion, int relation, int distance)
+{
+ 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 = (RSFD *) xmalloc (sizeof(*rsfd)*rset_no);
+ more = (int *) xmalloc (sizeof(*more)*rset_no);
+ buf = (struct it_key **) xmalloc (sizeof(*buf)*rset_no);
+
+ *prox_term = '\0';
+ for (i = 0; i<rset_no; i++)
+ {
+ int j;
+ 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 (lterm + length_prox_term < sizeof(prox_term)-1)
+ {
+ 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;
+ }
+ }
+ for (i = 0; i<rset_no; i++)
+ {
+ buf[i] = 0;
+ rsfd[i] = 0;
+ }
+ for (i = 0; i<rset_no; i++)
+ {
+ buf[i] = (struct it_key *) 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;
+ }
+ if (i != rset_no)
+ {
+ /* at least one is empty ... return null set */
+ rset_null_parms parms;
+
+ parms.rset_term = rset_term_create (prox_term, length_prox_term,
+ flags);
+ parms.rset_term->nn = 0;
+ result = rset_create (rset_kind_null, &parms);
+ }
+ else if (ordered && relation == 3 && exclusion == 0 && distance == 1)
+ {
+ /* special proximity case = phrase search ... */
+ rset_temp_parms parms;
+ RSFD rsfd_result;
+
+ parms.rset_term = rset_term_create (prox_term, length_prox_term,
+ flags);
+ parms.rset_term->nn = min_nn;
+ parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zh->service->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);
+ }
+ }
+ rset_close (result, rsfd_result);
+ }
+ else if (rset_no == 2)
+ {
+ /* generic proximity case (two input sets only) ... */
+ rset_temp_parms parms;
+ RSFD rsfd_result;
+
+ logf (LOG_LOG, "generic prox, dist = %d, relation = %d, ordered =%d, exclusion=%d",
+ distance, relation, ordered, exclusion);
+ parms.rset_term = rset_term_create (prox_term, length_prox_term,
+ flags);
+ parms.rset_term->nn = min_nn;
+ parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zh->service->res, "setTmpDir");
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd_result = rset_open (result, RSETF_WRITE);
+
+ while (more[0] && more[1])
+ {
+ int cmp = key_compare_it (buf[0], buf[1]);
+ if (cmp < -1)
+ more[0] = rset_read (rset[0], rsfd[0], buf[0], &term_index);
+ else if (cmp > 1)
+ more[1] = rset_read (rset[1], rsfd[1], buf[1], &term_index);
+ else
+ {
+ int sysno = buf[0]->sysno;
+ int seqno[500];
+ int n = 0;
+
+ seqno[n++] = buf[0]->seqno;
+ while ((more[0] = rset_read (rset[0], rsfd[0], buf[0],
+ &term_index)) &&
+ sysno == buf[0]->sysno)
+ if (n < 500)
+ seqno[n++] = buf[0]->seqno;
+ do
+ {
+ for (i = 0; i<n; i++)
+ {
+ int diff = buf[1]->seqno - seqno[i];
+ int excl = exclusion;
+ if (!ordered && diff < 0)
+ diff = -diff;
+ switch (relation)
+ {
+ case 1: /* < */
+ if (diff < distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 2: /* <= */
+ if (diff <= distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 3: /* == */
+ if (diff == distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 4: /* >= */
+ if (diff >= distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 5: /* > */
+ if (diff > distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 6: /* != */
+ if (diff != distance && diff >= 0)
+ excl = !excl;
+ break;
+ }
+ if (excl)
+ {
+ rset_write (result, rsfd_result, buf[1]);
+ break;
+ }
+ }
+ } while ((more[1] = rset_read (rset[1], rsfd[1], buf[1],
+ &term_index)) &&
+ sysno == buf[1]->sysno);
+ }
+ }
+ rset_close (result, rsfd_result);
+ }
+ else
+ {
+ rset_null_parms parms;
+
+ parms.rset_term = rset_term_create (prox_term, length_prox_term,
+ flags);
+ parms.rset_term->nn = 0;
+ result = rset_create (rset_kind_null, &parms);
+ }
+ for (i = 0; i<rset_no; i++)
+ {
+ if (rsfd[i])
+ rset_close (rset[i], rsfd[i]);
+ xfree (buf[i]);
+ }
+ xfree (buf);
+ xfree (more);
+ xfree (rsfd);
+ return result;
+}
+
+
+char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char *termz, NMEM stream, unsigned reg_id)
+{
+ WRBUF wrbuf = 0;
+ AttrType truncation;
+ int truncation_value;
+ char *ex_list = 0;
+
+ attr_init (&truncation, zapt, 5);
+ truncation_value = attr_find (&truncation, NULL);
+
+ switch (truncation_value)
+ {
+ default:
+ ex_list = "";
+ break;
+ case 101:
+ ex_list = "#";
+ break;
+ case 102:
+ case 103:
+ ex_list = 0;
+ break;
+ case 104:
+ ex_list = "!#";
+ break;
+ case 105:
+ ex_list = "!*";
+ break;
+ }
+ if (ex_list)
+ wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, ex_list,
+ termz, strlen(termz));
+ if (!wrbuf)
+ return nmem_strdup(stream, termz);
+ else
+ {
+ char *buf = (char*) nmem_malloc (stream, wrbuf_len(wrbuf)+1);
+ memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
+ buf[wrbuf_len(wrbuf)] = '\0';
+ return buf;
+ }
+}
+
+static int grep_info_prepare (ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ struct grep_info *grep_info,
+ int reg_type,
+ NMEM stream)
+{
+ AttrType termset;
+ int termset_value_numeric;
+ const char *termset_value_string;
+
+#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;
+ grep_info->termset = 0;
+
+ attr_init (&termset, zapt, 8);
+ termset_value_numeric =
+ attr_find_ex (&termset, NULL, &termset_value_string);
+ if (termset_value_numeric != -1)
+ {
+ char resname[32];
+ const char *termset_name = 0;
+ if (termset_value_numeric != -2)
+ {
+
+ sprintf (resname, "%d", termset_value_numeric);
+ termset_name = resname;
+ }
+ else
+ termset_name = termset_value_string;
+ logf (LOG_LOG, "creating termset set %s", termset_name);
+ grep_info->termset = resultSetAdd (zh, termset_name, 1);
+ if (!grep_info->termset)
+ {
+ zh->errCode = 128;
+ zh->errString = nmem_strdup (stream, termset_name);
+ return -1;
+ }
+ }
+ return 0;
+}
+
+
+static RSET rpn_search_APT_phrase (ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ const char *termz_org,
+ oid_value attributeSet,
+ NMEM stream,
+ int reg_type, int complete_flag,
+ const char *rank_type,
+ int num_bases, char **basenames)
+{
+ char term_dst[IT_MAX_WORD+1];
+ RSET rset[60], result;
+ int i, r, rset_no = 0;
+ struct grep_info grep_info;
+ char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
+ const char *termp = termz;
+
+ if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
+ return 0;
+ while (1)
+ {
+ logf (LOG_DEBUG, "APT_phrase termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ r = string_term (zh, zapt, &termp, attributeSet, stream, &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 >= (int) (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, 1, 0, 3, 1);
+ 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_org,
+ oid_value attributeSet,
+ NMEM stream,
+ int reg_type, int complete_flag,
+ const char *rank_type,
+ int num_bases, char **basenames)
+{
+ char term_dst[IT_MAX_WORD+1];
+ RSET rset[60], result;
+ int i, r, rset_no = 0;
+ struct grep_info grep_info;
+ char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
+ const char *termp = termz;
+
+ if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
+ return 0;
+ while (1)
+ {
+ logf (LOG_DEBUG, "APT_or_list termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ r = string_term (zh, zapt, &termp, attributeSet, stream, &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 >= (int) (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_org,
+ oid_value attributeSet,
+ NMEM stream,
+ int reg_type, int complete_flag,
+ const char *rank_type,
+ int num_bases, char **basenames)
+{
+ char term_dst[IT_MAX_WORD+1];
+ RSET rset[60], result;
+ int i, r, rset_no = 0;
+ struct grep_info grep_info;
+ char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
+ const char *termp = termz;
+
+ if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
+ return 0;
+ while (1)
+ {
+ logf (LOG_DEBUG, "APT_and_list termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ r = string_term (zh, zapt, &termp, attributeSet, stream, &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 >= (int) (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)