+ *term_tmp++ = '[';
+
+ *term_tmp++ = '^';
+ string_rel_add_char (&term_tmp, term_component, &i);
+ *term_tmp++ = '-';
+
+ *term_tmp++ = ']';
+ *term_tmp++ = '.';
+ *term_tmp++ = '*';
+
+ if ((term_tmp - term_dict) > IT_MAX_WORD)
+ break;
+ }
+ *term_tmp++ = ')';
+ *term_tmp = '\0';
+ break;
+ case 2:
+ if (!term_100 (zh->reg->zebra_maps, reg_type,
+ term_sub, term_component,
+ space_split, term_dst))
+ return 0;
+ yaz_log(log_level_rpn, "Relation <=");
+
+ *term_tmp++ = '(';
+ for (i = 0; term_component[i]; )
+ {
+ int j = 0;
+
+ while (j < i)
+ string_rel_add_char (&term_tmp, term_component, &j);
+ *term_tmp++ = '[';
+
+ *term_tmp++ = '^';
+ string_rel_add_char (&term_tmp, term_component, &i);
+ *term_tmp++ = '-';
+
+ *term_tmp++ = ']';
+ *term_tmp++ = '.';
+ *term_tmp++ = '*';
+
+ *term_tmp++ = '|';
+
+ if ((term_tmp - term_dict) > IT_MAX_WORD)
+ break;
+ }
+ for (i = 0; term_component[i]; )
+ string_rel_add_char (&term_tmp, term_component, &i);
+ *term_tmp++ = ')';
+ *term_tmp = '\0';
+ break;
+ case 5:
+ if (!term_100 (zh->reg->zebra_maps, reg_type,
+ term_sub, term_component, space_split, term_dst))
+ return 0;
+ yaz_log(log_level_rpn, "Relation >");
+
+ *term_tmp++ = '(';
+ for (i = 0; term_component[i];)
+ {
+ int j = 0;
+
+ while (j < i)
+ string_rel_add_char (&term_tmp, term_component, &j);
+ *term_tmp++ = '[';
+
+ *term_tmp++ = '^';
+ *term_tmp++ = '-';
+ string_rel_add_char (&term_tmp, term_component, &i);
+
+ *term_tmp++ = ']';
+ *term_tmp++ = '.';
+ *term_tmp++ = '*';
+
+ *term_tmp++ = '|';
+
+ if ((term_tmp - term_dict) > IT_MAX_WORD)
+ break;
+ }
+ for (i = 0; term_component[i];)
+ string_rel_add_char (&term_tmp, term_component, &i);
+ *term_tmp++ = '.';
+ *term_tmp++ = '+';
+ *term_tmp++ = ')';
+ *term_tmp = '\0';
+ break;
+ case 4:
+ if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub,
+ term_component, space_split, term_dst))
+ return 0;
+ yaz_log(log_level_rpn, "Relation >=");
+
+ *term_tmp++ = '(';
+ for (i = 0; term_component[i];)
+ {
+ int j = 0;
+
+ if (i)
+ *term_tmp++ = '|';
+ while (j < i)
+ string_rel_add_char (&term_tmp, term_component, &j);
+ *term_tmp++ = '[';
+
+ if (term_component[i+1])
+ {
+ *term_tmp++ = '^';
+ *term_tmp++ = '-';
+ string_rel_add_char (&term_tmp, term_component, &i);
+ }
+ else
+ {
+ string_rel_add_char (&term_tmp, term_component, &i);
+ *term_tmp++ = '-';
+ }
+ *term_tmp++ = ']';
+ *term_tmp++ = '.';
+ *term_tmp++ = '*';
+
+ if ((term_tmp - term_dict) > IT_MAX_WORD)
+ break;
+ }
+ *term_tmp++ = ')';
+ *term_tmp = '\0';
+ break;
+ case 3:
+ default:
+ yaz_log(log_level_rpn, "Relation =");
+ if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub,
+ term_component, space_split, term_dst))
+ return 0;
+ strcat(term_tmp, "(");
+ strcat(term_tmp, term_component);
+ strcat(term_tmp, ")");
+ }
+ return 1;
+}
+
+static int string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char **term_sub,
+ oid_value attributeSet, NMEM stream,
+ struct grep_info *grep_info,
+ int reg_type, int complete_flag,
+ int num_bases, char **basenames,
+ char *term_dst, int xpath_use);
+
+static RSET term_trunc(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char **term_sub,
+ oid_value attributeSet, NMEM stream,
+ struct grep_info *grep_info,
+ int reg_type, int complete_flag,
+ int num_bases, char **basenames,
+ char *term_dst,
+ const char *rank_type, int xpath_use,
+ NMEM rset_nmem)
+{
+ int r;
+ grep_info->isam_p_indx = 0;
+ r = string_term(zh, zapt, term_sub, attributeSet, stream, grep_info,
+ reg_type, complete_flag, num_bases, basenames,
+ term_dst, xpath_use);
+ if (r < 1)
+ return 0;
+ yaz_log(log_level_rpn, "term: %s", term_dst);
+ return rset_trunc(zh, grep_info->isam_p_buf,
+ grep_info->isam_p_indx, term_dst,
+ strlen(term_dst), rank_type, 1 /* preserve pos */,
+ zapt->term->which, rset_nmem,
+ key_it_ctrl,key_it_ctrl->scope);
+}
+static char *nmem_strdup_i(NMEM nmem, int v)
+{
+ char val_str[64];
+ sprintf (val_str, "%d", v);
+ return nmem_strdup(nmem, val_str);
+}
+
+static int string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char **term_sub,
+ oid_value attributeSet, NMEM stream,
+ struct grep_info *grep_info,
+ int reg_type, int complete_flag,
+ int num_bases, char **basenames,
+ char *term_dst, int xpath_use)
+{
+ char term_dict[2*IT_MAX_WORD+4000];
+ int j, r, base_no;
+ AttrType truncation;
+ int truncation_value;
+ AttrType use;
+ int use_value;
+ const char *use_string = 0;
+ oid_value curAttributeSet = attributeSet;
+ const char *termp;
+ struct rpn_char_map_info rcmi;
+ int space_split = complete_flag ? 0 : 1;
+
+ int bases_ok = 0; /* no of databases with OK attribute */
+ int errCode = 0; /* err code (if any is not OK) */
+ char *errString = 0; /* addinfo */
+
+ rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
+ attr_init (&use, zapt, 1);
+ use_value = attr_find_ex (&use, &curAttributeSet, &use_string);
+ yaz_log(log_level_rpn, "string_term, use value %d", use_value);
+ attr_init (&truncation, zapt, 5);
+ truncation_value = attr_find (&truncation, NULL);
+ yaz_log(log_level_rpn, "truncation value %d", truncation_value);
+
+ if (use_value == -1) /* no attribute - assumy "any" */
+ use_value = 1016;
+ for (base_no = 0; base_no < num_bases; base_no++)
+ {
+ int ord = -1;
+ int attr_ok = 0;
+ int regex_range = 0;
+ int init_pos = 0;
+ attent attp;
+ data1_local_attribute id_xpath_attr;
+ data1_local_attribute *local_attr;
+ int max_pos, prefix_len = 0;
+
+ termp = *term_sub;
+
+ if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
+ {
+ zh->errCode = 109; /* Database unavailable */
+ zh->errString = basenames[base_no];
+ return -1;
+ }
+ if (xpath_use > 0 && use_value == -2)
+ {
+ /* xpath mode and we have a string attribute */
+ attp.local_attributes = &id_xpath_attr;
+ attp.attset_ordinal = VAL_IDXPATH;
+ id_xpath_attr.next = 0;
+
+ use_value = xpath_use; /* xpath_use as use-attribute now */
+ id_xpath_attr.local = use_value;
+ }
+ else if (curAttributeSet == VAL_IDXPATH && use_value >= 0)
+ {
+ /* X-Path attribute, use numeric value directly */
+ attp.local_attributes = &id_xpath_attr;
+ attp.attset_ordinal = VAL_IDXPATH;
+ id_xpath_attr.next = 0;
+ id_xpath_attr.local = use_value;
+ }
+ else if (use_string &&
+ (ord = zebraExplain_lookup_attr_str(zh->reg->zei,
+ use_string)) >= 0)
+ {
+ /* we have a match for a raw string attribute */
+ char ord_buf[32];
+ int i, ord_len;
+
+ if (prefix_len)
+ term_dict[prefix_len++] = '|';
+ else
+ term_dict[prefix_len++] = '(';
+
+ ord_len = key_SU_encode (ord, ord_buf);
+ for (i = 0; i<ord_len; i++)
+ {
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = ord_buf[i];
+ }
+ attp.local_attributes = 0; /* no more attributes */
+ }
+ else
+ {
+ /* lookup in the .att files . Allow string as well */
+ if ((r = att_getentbyatt (zh, &attp, curAttributeSet, use_value,
+ use_string)))
+ {
+ yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
+ curAttributeSet, use_value, r);
+ if (r == -1)
+ {
+ /* set was found, but value wasn't defined */
+ errCode = 114;
+ if (use_string)
+ errString = nmem_strdup(stream, use_string);
+ else
+ errString = nmem_strdup_i (stream, use_value);
+ }
+ else
+ {
+ int oid[OID_SIZE];
+ struct oident oident;
+
+ oident.proto = PROTO_Z3950;
+ oident.oclass = CLASS_ATTSET;
+ oident.value = curAttributeSet;
+ oid_ent_to_oid (&oident, oid);
+
+ errCode = 121;
+ errString = nmem_strdup (stream, oident.desc);
+ }
+ continue;
+ }
+ }
+ for (local_attr = attp.local_attributes; local_attr;
+ local_attr = local_attr->next)
+ {
+ char ord_buf[32];
+ int i, ord_len;
+
+ ord = zebraExplain_lookup_attr_su(zh->reg->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_encode (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)
+ {
+#if 1
+ bases_ok++;
+#else
+ errCode = 114;
+ errString = nmem_strdup_i(stream, use_value);
+ continue;
+#endif
+ }
+ else
+ {
+ bases_ok++; /* this has OK attributes */
+ attr_ok = 1;
+ }
+
+ term_dict[prefix_len++] = ')';
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = reg_type;
+ yaz_log(log_level_rpn, "reg_type = %d", term_dict[prefix_len-1]);
+ term_dict[prefix_len] = '\0';
+ j = prefix_len;
+ switch (truncation_value)
+ {
+ case -1: /* not specified */
+ case 100: /* do not truncate */
+ if (!string_relation (zh, zapt, &termp, term_dict,
+ attributeSet,
+ reg_type, space_split, term_dst))
+ return 0;
+ break;
+ case 1: /* right truncation */
+ term_dict[j++] = '(';
+ if (!term_100(zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst))
+ return 0;
+ strcat(term_dict, ".*)");
+ break;
+ case 2: /* keft truncation */
+ term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
+ if (!term_100(zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst))
+ return 0;
+ strcat(term_dict, ")");
+ break;
+ case 3: /* left&right truncation */
+ term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
+ if (!term_100(zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst))
+ return 0;
+ strcat(term_dict, ".*)");
+ break;
+ case 101: /* process # in term */
+ term_dict[j++] = '(';
+ if (!term_101(zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst))
+ return 0;
+ strcat(term_dict, ")");
+ break;
+ case 102: /* Regexp-1 */
+ term_dict[j++] = '(';
+ if (!term_102(zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst))
+ return 0;
+ strcat(term_dict, ")");
+ break;
+ case 103: /* Regexp-2 */
+ r = 1;
+ term_dict[j++] = '(';
+ init_pos = 2;
+ if (!term_103 (zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, ®ex_range,
+ space_split, term_dst))
+ return 0;
+ strcat(term_dict, ")");
+ case 104: /* process # and ! in term */
+ term_dict[j++] = '(';
+ if (!term_104 (zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst))
+ return 0;
+ strcat(term_dict, ")");
+ break;
+ case 105: /* process * and ! in term */
+ term_dict[j++] = '(';
+ if (!term_105 (zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst, 1))
+ return 0;
+ strcat(term_dict, ")");
+ break;
+ case 106: /* process * and ! in term */
+ term_dict[j++] = '(';
+ if (!term_105 (zh->reg->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst, 0))
+ return 0;
+ strcat(term_dict, ")");
+ break;
+ default:
+ zh->errCode = 120;
+ zh->errString = nmem_strdup_i(stream, truncation_value);
+ return -1;
+ }
+ if (attr_ok)
+ {
+ char buf[80];
+ const char *input = term_dict + prefix_len;
+ esc_str(buf, sizeof(buf), input, strlen(input));
+ }
+ if (attr_ok)
+ {
+ yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_dict+prefix_len);
+ r = dict_lookup_grep(zh->reg->dict, term_dict, regex_range,
+ grep_info, &max_pos, init_pos,
+ grep_handle);
+ if (r)
+ yaz_log(YLOG_WARN, "dict_lookup_grep fail %d", r);
+ }
+ }
+ if (!bases_ok)
+ {
+ zh->errCode = errCode;
+ zh->errString = errString;
+ return -1;
+ }
+ *term_sub = termp;
+ yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx);
+ return 1;
+}
+
+
+/* convert APT search term to UTF8 */
+static int zapt_term_to_utf8 (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ char *termz)
+{
+ size_t sizez;
+ Z_Term *term = zapt->term;
+
+ switch (term->which)
+ {
+ case Z_Term_general:
+ if (zh->iconv_to_utf8 != 0)
+ {
+ char *inbuf = term->u.general->buf;
+ size_t inleft = term->u.general->len;
+ char *outbuf = termz;
+ size_t outleft = IT_MAX_WORD-1;
+ size_t ret;
+
+ ret = yaz_iconv(zh->iconv_to_utf8, &inbuf, &inleft,
+ &outbuf, &outleft);
+ if (ret == (size_t)(-1))
+ {
+ ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0);
+ zh->errCode = 125;
+ return -1;
+ }
+ *outbuf = 0;
+ }
+ else
+ {
+ 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';
+ }
+ break;
+ case Z_Term_characterString:
+ sizez = strlen(term->u.characterString);
+ if (sizez > IT_MAX_WORD-1)
+ sizez = IT_MAX_WORD-1;
+ memcpy (termz, term->u.characterString, sizez);
+ termz[sizez] = '\0';
+ break;
+ default:
+ zh->errCode = 124;
+ return -1;
+ }
+ return 0;
+}
+
+/* convert APT SCAN term to internal cmap */
+static int trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ char *termz, int reg_type)
+{
+ char termz0[IT_MAX_WORD];
+
+ if (zapt_term_to_utf8(zh, zapt, termz0))
+ return -1; /* error */
+ else
+ {
+ const char **map;
+ const char *cp = (const char *) termz0;
+ const char *cp_end = cp + strlen(cp);
+ const char *src;
+ int i = 0;
+ const char *space_map = NULL;
+ int len;
+
+ while ((len = (cp_end - cp)) > 0)
+ {
+ map = zebra_maps_input (zh->reg->zebra_maps, reg_type, &cp, len, 0);
+ 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';
+ }
+ return 0;
+}
+
+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->reg->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 void grep_info_delete (struct grep_info *grep_info)
+{
+#ifdef TERM_COUNT
+ xfree(grep_info->term_no);
+#endif
+ xfree (grep_info->isam_p_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;
+
+ if (!zapt)
+ return 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;
+ yaz_log(log_level_rpn, "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 xpath_use,
+ int num_bases, char **basenames,
+ NMEM rset_nmem)
+{
+ char term_dst[IT_MAX_WORD+1];
+ RSET rset[TERM_LIST_LENGTH_MAX], result;
+ size_t rset_no = 0;
+ struct grep_info grep_info;
+ char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
+ const char *termp = termz;
+
+ *term_dst = 0;
+ if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
+ return 0;
+ for (; rset_no < sizeof(rset)/sizeof(*rset); rset_no++)
+ {
+ yaz_log(log_level_rpn, "APT_phrase termp=%s", termp);
+ rset[rset_no] = term_trunc(zh, zapt, &termp, attributeSet,
+ stream, &grep_info,
+ reg_type, complete_flag,
+ num_bases, basenames,
+ term_dst, rank_type,
+ xpath_use,rset_nmem);
+ if (!rset[rset_no])
+ break;
+ }
+ grep_info_delete (&grep_info);
+ if (rset_no == 0)
+ return rsnull_create (rset_nmem,key_it_ctrl);
+ else if (rset_no == 1)
+ return (rset[0]);
+ else
+ result = rsprox_create( rset_nmem, key_it_ctrl, key_it_ctrl->scope,
+ rset_no, rset,
+ 1 /* ordered */, 0 /* exclusion */,
+ 3 /* relation */, 1 /* distance */);
+ 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 xpath_use,
+ int num_bases, char **basenames,
+ NMEM rset_nmem)
+{
+ char term_dst[IT_MAX_WORD+1];
+ RSET rset[TERM_LIST_LENGTH_MAX];
+ size_t 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;
+ for (; rset_no < sizeof(rset)/sizeof(*rset); rset_no++)
+ {
+ yaz_log(log_level_rpn, "APT_or_list termp=%s", termp);
+ rset[rset_no] = term_trunc(zh, zapt, &termp, attributeSet,
+ stream, &grep_info,
+ reg_type, complete_flag,
+ num_bases, basenames,
+ term_dst, rank_type,
+ xpath_use,rset_nmem);
+ if (!rset[rset_no])
+ break;
+ }
+ grep_info_delete (&grep_info);
+ if (rset_no == 0)
+ return rsnull_create (rset_nmem,key_it_ctrl);
+ return rsmulti_or_create(rset_nmem, key_it_ctrl,key_it_ctrl->scope,
+ rset_no, rset);
+}
+
+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 xpath_use,
+ int num_bases, char **basenames,
+ NMEM rset_nmem)
+{
+ char term_dst[IT_MAX_WORD+1];
+ RSET rset[TERM_LIST_LENGTH_MAX];
+ size_t 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;
+ for (; rset_no < sizeof(rset)/sizeof(*rset); rset_no++)
+ {
+ yaz_log(log_level_rpn, "APT_and_list termp=%s", termp);
+ rset[rset_no] = term_trunc(zh, zapt, &termp, attributeSet,
+ stream, &grep_info,
+ reg_type, complete_flag,
+ num_bases, basenames,
+ term_dst, rank_type,
+ xpath_use, rset_nmem);
+ if (!rset[rset_no])
+ break;
+ }
+ grep_info_delete (&grep_info);
+ if (rset_no == 0)
+ return rsnull_create(rset_nmem,key_it_ctrl);
+
+ return rsmulti_and_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope,
+ rset_no, rset);
+}
+
+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);
+
+ yaz_log(log_level_rpn, "numeric relation value=%d", relation_value);
+
+ if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1,
+ term_dst))
+ return 0;
+ term_value = atoi (term_tmp);
+ switch (relation_value)
+ {
+ case 1:
+ yaz_log(log_level_rpn, "Relation <");
+ gen_regular_rel (term_tmp, term_value-1, 1);
+ break;
+ case 2:
+ yaz_log(log_level_rpn, "Relation <=");
+ gen_regular_rel (term_tmp, term_value, 1);
+ break;
+ case 4:
+ yaz_log(log_level_rpn, "Relation >=");
+ gen_regular_rel (term_tmp, term_value, 0);
+ break;
+ case 5:
+ yaz_log(log_level_rpn, "Relation >");
+ gen_regular_rel (term_tmp, term_value+1, 0);
+ break;
+ case 3:
+ default:
+ yaz_log(log_level_rpn, "Relation =");
+ sprintf (term_tmp, "(0*%d)", term_value);
+ }
+ yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_tmp);
+ r = dict_lookup_grep(zh->reg->dict, term_dict, 0, grep_info, max_pos,
+ 0, grep_handle);
+ if (r)
+ yaz_log(YLOG_WARN, "dict_lookup_grep fail, rel = gt: %d", r);
+ yaz_log(log_level_rpn, "%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, int xpath_use, NMEM stream)
+{
+ char term_dict[2*IT_MAX_WORD+2];
+ int r, base_no;
+ AttrType use;
+ int use_value;
+ const char *use_string = 0;
+ oid_value curAttributeSet = attributeSet;
+ const char *termp;
+ struct rpn_char_map_info rcmi;
+
+ int bases_ok = 0; /* no of databases with OK attribute */
+ int errCode = 0; /* err code (if any is not OK) */
+ char *errString = 0; /* addinfo */
+
+ rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
+ attr_init (&use, zapt, 1);
+ use_value = attr_find_ex (&use, &curAttributeSet, &use_string);
+
+ if (use_value == -1)
+ use_value = 1016;
+
+ for (base_no = 0; base_no < num_bases; base_no++)
+ {
+ attent attp;
+ data1_local_attribute id_xpath_attr;
+ data1_local_attribute *local_attr;
+ int max_pos, prefix_len = 0;
+
+ termp = *term_sub;
+ if (use_value == -2) /* string attribute (assume IDXPATH/any) */
+ {
+ use_value = xpath_use;
+ attp.local_attributes = &id_xpath_attr;
+ attp.attset_ordinal = VAL_IDXPATH;
+ id_xpath_attr.next = 0;
+ id_xpath_attr.local = use_value;
+ }
+ else if (curAttributeSet == VAL_IDXPATH)
+ {
+ attp.local_attributes = &id_xpath_attr;
+ attp.attset_ordinal = VAL_IDXPATH;
+ id_xpath_attr.next = 0;
+ id_xpath_attr.local = use_value;
+ }
+ else
+ {
+ if ((r = att_getentbyatt (zh, &attp, curAttributeSet, use_value,
+ use_string)))
+ {
+ yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
+ curAttributeSet, use_value, r);
+ if (r == -1)
+ {
+ errString = nmem_strdup_i(stream, use_value);
+ errCode = 114;
+ }
+ else
+ errCode = 121;
+ continue;
+ }
+ }
+ if (zebraExplain_curDatabase (zh->reg->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_lookup_attr_su(zh->reg->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_encode (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)
+ {
+ errCode = 114;
+ errString = nmem_strdup_i(stream, use_value);
+ continue;
+ }
+ bases_ok++;
+ term_dict[prefix_len++] = ')';
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = reg_type;
+ yaz_log(YLOG_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;
+ }
+ if (!bases_ok)
+ {
+ zh->errCode = errCode;
+ zh->errString = errString;
+ return -1;
+ }
+ *term_sub = termp;
+ yaz_log(YLOG_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,
+ NMEM stream,
+ int reg_type, int complete_flag,
+ const char *rank_type, int xpath_use,
+ int num_bases, char **basenames,
+ NMEM rset_nmem)
+{
+ char term_dst[IT_MAX_WORD+1];
+ const char *termp = termz;
+ RSET rset[TERM_LIST_LENGTH_MAX];
+ int r;
+ size_t rset_no = 0;
+ struct grep_info grep_info;
+
+ yaz_log(log_level_rpn, "APT_numeric t='%s'",termz);
+ if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
+ return 0;
+ for (; rset_no < sizeof(rset)/sizeof(*rset); rset_no++)
+ {
+ yaz_log(YLOG_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, xpath_use,
+ stream);
+ if (r < 1)
+ break;
+ yaz_log(YLOG_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,
+ 0 /* preserve position */,
+ zapt->term->which, rset_nmem,
+ key_it_ctrl,key_it_ctrl->scope);
+ if (!rset[rset_no])
+ break;
+ }
+ grep_info_delete (&grep_info);
+ if (rset_no == 0)
+ return rsnull_create(rset_nmem,key_it_ctrl);
+ if (rset_no == 1)
+ return rset[0];
+ return rsmulti_and_create(rset_nmem,key_it_ctrl,key_it_ctrl->scope,
+ rset_no, rset);
+}
+
+static RSET rpn_search_APT_local (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char *termz,
+ oid_value attributeSet,
+ NMEM stream,
+ const char *rank_type, NMEM rset_nmem)
+{
+ RSET result;
+ RSFD rsfd;
+ struct it_key key;
+ int sys;
+ result = rstemp_create( rset_nmem,key_it_ctrl,key_it_ctrl->scope,
+ res_get (zh->res, "setTmpDir"),0 );
+ rsfd = rset_open (result, RSETF_WRITE);
+
+ sys = atoi(termz);
+ if (sys <= 0)
+ sys = 1;
+ key.mem[0] = sys;
+ key.mem[1] = 1;
+ key.len = 2;
+ rset_write (rsfd, &key);
+ rset_close (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)
+{
+ 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;
+ char termz[20];
+
+ 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 = (Z_SortKeySpec **)
+ 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 ((char *) zapt->term->u.general->buf,
+ zapt->term->u.general->len);
+ if (i >= sort_sequence->num_specs)
+ i = 0;
+ sprintf (termz, "%d", i);
+
+ oe.proto = PROTO_Z3950;
+ oe.oclass = CLASS_ATTSET;
+ oe.value = attributeSet;
+ if (!oid_ent_to_oid (&oe, oid))
+ return 0;
+
+ sks = (Z_SortKeySpec *) nmem_malloc(stream, sizeof(*sks));
+ sks->sortElement = (Z_SortElement *)
+ nmem_malloc(stream, sizeof(*sks->sortElement));
+ sks->sortElement->which = Z_SortElement_generic;
+ sk = sks->sortElement->u.generic = (Z_SortKey *)
+ nmem_malloc(stream, sizeof(*sk));
+ sk->which = Z_SortKey_sortAttributes;
+ sk->u.sortAttributes = (Z_SortAttributes *)
+ nmem_malloc(stream, sizeof(*sk->u.sortAttributes));
+
+ sk->u.sortAttributes->id = oid;
+ sk->u.sortAttributes->list = (Z_AttributeList *)
+ nmem_malloc(stream, sizeof(*sk->u.sortAttributes->list));
+ sk->u.sortAttributes->list->num_attributes = 1;
+ sk->u.sortAttributes->list->attributes = (Z_AttributeElement **)
+ nmem_malloc(stream, sizeof(*sk->u.sortAttributes->list->attributes));
+ ae = *sk->u.sortAttributes->list->attributes = (Z_AttributeElement *)
+ nmem_malloc(stream, sizeof(**sk->u.sortAttributes->list->attributes));
+ ae->attributeSet = 0;
+ ae->attributeType = (int *)
+ nmem_malloc(stream, sizeof(*ae->attributeType));
+ *ae->attributeType = 1;
+ ae->which = Z_AttributeValue_numeric;
+ ae->value.numeric = (int *)
+ nmem_malloc(stream, sizeof(*ae->value.numeric));
+ *ae->value.numeric = use_value;
+
+ sks->sortRelation = (int *)
+ nmem_malloc(stream, sizeof(*sks->sortRelation));
+ if (sort_relation_value == 1)
+ *sks->sortRelation = Z_SortKeySpec_ascending;
+ else if (sort_relation_value == 2)
+ *sks->sortRelation = Z_SortKeySpec_descending;
+ else
+ *sks->sortRelation = Z_SortKeySpec_ascending;
+
+ sks->caseSensitivity = (int *)
+ nmem_malloc(stream, sizeof(*sks->caseSensitivity));
+ *sks->caseSensitivity = 0;
+
+ sks->which = Z_SortKeySpec_null;
+ sks->u.null = odr_nullval ();
+ sort_sequence->specs[i] = sks;
+ return rsnull_create (NULL,key_it_ctrl);
+ /* FIXME - nmem?? */
+}
+
+
+static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet,
+ struct xpath_location_step *xpath, int max, NMEM mem)
+{
+ oid_value curAttributeSet = attributeSet;
+ AttrType use;
+ const char *use_string = 0;
+
+ attr_init (&use, zapt, 1);
+ attr_find_ex (&use, &curAttributeSet, &use_string);
+
+ if (!use_string || *use_string != '/')
+ return -1;
+
+ return zebra_parse_xpath_str(use_string, xpath, max, mem);
+}
+
+
+
+static RSET xpath_trunc(ZebraHandle zh, NMEM stream,
+ int reg_type, const char *term, int use,
+ oid_value curAttributeSet, NMEM rset_nmem)
+{
+ RSET rset;
+ struct grep_info grep_info;
+ char term_dict[2048];
+ char ord_buf[32];
+ int prefix_len = 0;
+ int ord = zebraExplain_lookup_attr_su(zh->reg->zei, curAttributeSet, use);
+ int ord_len, i, r, max_pos;
+ int term_type = Z_Term_characterString;
+ const char *flags = "void";
+
+ if (grep_info_prepare (zh, 0 /* zapt */, &grep_info, '0', stream))
+ return rsnull_create (rset_nmem,key_it_ctrl);
+
+ if (ord < 0)
+ return rsnull_create (rset_nmem,key_it_ctrl);
+ if (prefix_len)
+ term_dict[prefix_len++] = '|';
+ else
+ term_dict[prefix_len++] = '(';
+
+ ord_len = key_SU_encode (ord, ord_buf);
+ for (i = 0; i<ord_len; i++)
+ {
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = ord_buf[i];
+ }
+ term_dict[prefix_len++] = ')';
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = reg_type;
+
+ strcpy(term_dict+prefix_len, term);
+
+ grep_info.isam_p_indx = 0;
+ r = dict_lookup_grep(zh->reg->dict, term_dict, 0,
+ &grep_info, &max_pos, 0, grep_handle);
+ yaz_log (YLOG_DEBUG, "%s %d positions", term,
+ grep_info.isam_p_indx);
+ rset = rset_trunc(zh, grep_info.isam_p_buf,
+ grep_info.isam_p_indx, term, strlen(term),
+ flags, 1, term_type,rset_nmem,
+ key_it_ctrl, key_it_ctrl->scope);
+ grep_info_delete (&grep_info);
+ return rset;
+}
+
+static RSET rpn_search_xpath (ZebraHandle zh,
+ oid_value attributeSet,
+ int num_bases, char **basenames,
+ NMEM stream, const char *rank_type, RSET rset,
+ int xpath_len, struct xpath_location_step *xpath,
+ NMEM rset_nmem)
+{
+ oid_value curAttributeSet = attributeSet;
+ int base_no;
+ int i;
+
+ if (xpath_len < 0)
+ return rset;
+
+ yaz_log (YLOG_DEBUG, "xpath len=%d", xpath_len);
+ for (i = 0; i<xpath_len; i++)
+ {
+ yaz_log (log_level_rpn, "XPATH %d %s", i, xpath[i].part);
+
+ }
+
+ curAttributeSet = VAL_IDXPATH;
+
+ /*
+ //a -> a/.*
+ //a/b -> b/a/.*
+ /a -> a/
+ /a/b -> b/a/
+
+ / -> none
+
+ a[@attr = value]/b[@other = othervalue]
+
+ /e/@a val range(e/,range(@a,freetext(w,1015,val),@a),e/)
+ /a/b val range(b/a/,freetext(w,1016,val),b/a/)
+ /a/b/@c val range(b/a/,range(@c,freetext(w,1016,val),@c),b/a/)
+ /a/b[@c = y] val range(b/a/,freetext(w,1016,val),b/a/,@c = y)
+ /a[@c = y]/b val range(a/,range(b/a/,freetext(w,1016,val),b/a/),a/,@c = y)
+ /a[@c = x]/b[@c = y] range(a/,range(b/a/,freetext(w,1016,val),b/a/,@c = y),a/,@c = x)
+
+ */
+
+ dict_grep_cmap (zh->reg->dict, 0, 0);
+
+ for (base_no = 0; base_no < num_bases; base_no++)
+ {
+ int level = xpath_len;
+ int first_path = 1;
+
+ if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
+ {
+ zh->errCode = 109; /* Database unavailable */
+ zh->errString = basenames[base_no];
+ return rset;
+ }
+ while (--level >= 0)
+ {
+ char xpath_rev[128];
+ int i, len;
+ RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0;
+
+ *xpath_rev = 0;
+ len = 0;
+ for (i = level; i >= 1; --i)
+ {
+ const char *cp = xpath[i].part;
+ if (*cp)
+ {
+ for (;*cp; cp++)
+ if (*cp == '*')
+ {
+ memcpy (xpath_rev + len, "[^/]*", 5);
+ len += 5;
+ }
+ else if (*cp == ' ')
+ {
+
+ xpath_rev[len++] = 1;
+ xpath_rev[len++] = ' ';
+ }
+
+ else
+ xpath_rev[len++] = *cp;
+ xpath_rev[len++] = '/';
+ }
+ else if (i == 1) /* // case */
+ {
+ xpath_rev[len++] = '.';
+ xpath_rev[len++] = '*';
+ }
+ }
+ xpath_rev[len] = 0;
+
+ if (xpath[level].predicate &&
+ xpath[level].predicate->which == XPATH_PREDICATE_RELATION &&
+ xpath[level].predicate->u.relation.name[0])
+ {
+ WRBUF wbuf = wrbuf_alloc();
+ wrbuf_puts(wbuf, xpath[level].predicate->u.relation.name+1);
+ if (xpath[level].predicate->u.relation.value)
+ {
+ const char *cp = xpath[level].predicate->u.relation.value;
+ wrbuf_putc(wbuf, '=');
+
+ while (*cp)
+ {
+ if (strchr(REGEX_CHARS, *cp))
+ wrbuf_putc(wbuf, '\\');
+ wrbuf_putc(wbuf, *cp);
+ cp++;
+ }
+ }
+ wrbuf_puts(wbuf, "");
+ rset_attr = xpath_trunc(
+ zh, stream, '0', wrbuf_buf(wbuf), 3,
+ curAttributeSet,rset_nmem);
+ wrbuf_free(wbuf, 1);
+ }
+ else
+ {
+ if (!first_path)
+ continue;
+ }
+ yaz_log (log_level_rpn, "xpath_rev (%d) = %s", level, xpath_rev);
+ if (strlen(xpath_rev))
+ {
+ rset_start_tag = xpath_trunc(zh, stream, '0',
+ xpath_rev, 1, curAttributeSet, rset_nmem);
+
+ rset_end_tag = xpath_trunc(zh, stream, '0',
+ xpath_rev, 2, curAttributeSet, rset_nmem);
+
+ rset = rsbetween_create(rset_nmem, key_it_ctrl,
+ key_it_ctrl->scope,
+ rset_start_tag, rset,
+ rset_end_tag, rset_attr);
+ }
+ first_path = 0;
+ }
+ }
+
+ return rset;
+}
+
+
+
+static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet, NMEM stream,
+ Z_SortKeySpecList *sort_sequence,
+ int num_bases, char **basenames,
+ NMEM rset_nmem)
+{
+ unsigned reg_id;
+ char *search_type = NULL;
+ char rank_type[128];
+ int complete_flag;
+ int sort_flag;
+ char termz[IT_MAX_WORD+1];
+ RSET rset = 0;
+ int xpath_len;
+ int xpath_use = 0;
+ struct xpath_location_step xpath[10];
+
+ if (!log_level_set)
+ {
+ log_level_rpn = yaz_log_module_level("rpn");
+ log_level_set = 1;
+ }
+ zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type,
+ rank_type, &complete_flag, &sort_flag);
+
+ yaz_log(YLOG_DEBUG, "reg_id=%c", reg_id);
+ yaz_log(YLOG_DEBUG, "complete_flag=%d", complete_flag);
+ yaz_log(YLOG_DEBUG, "search_type=%s", search_type);
+ yaz_log(YLOG_DEBUG, "rank_type=%s", rank_type);
+
+ if (zapt_term_to_utf8(zh, zapt, termz))
+ return 0;
+
+ if (sort_flag)
+ return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence,
+ rank_type);
+ xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream);
+ if (xpath_len >= 0)
+ {
+ xpath_use = 1016;
+ if (xpath[xpath_len-1].part[0] == '@')
+ xpath_use = 1015;
+ }
+
+ if (!strcmp (search_type, "phrase"))
+ {
+ rset = rpn_search_APT_phrase (zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem);
+ }
+ else if (!strcmp (search_type, "and-list"))
+ {
+ rset = rpn_search_APT_and_list (zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem);
+ }
+ else if (!strcmp (search_type, "or-list"))
+ {
+ rset = rpn_search_APT_or_list (zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem);
+ }
+ else if (!strcmp (search_type, "local"))
+ {
+ rset = rpn_search_APT_local (zh, zapt, termz, attributeSet, stream,
+ rank_type, rset_nmem);
+ }
+ else if (!strcmp (search_type, "numeric"))
+ {
+ rset = rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem);
+ }
+ else if (!strcmp (search_type, "always"))
+ {
+ rset = 0;
+ }
+ else
+ zh->errCode = 118;
+ return rpn_search_xpath (zh, attributeSet, num_bases, basenames,
+ stream, rank_type, rset,
+ xpath_len, xpath, rset_nmem);
+}
+
+static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs,
+ oid_value attributeSet,
+ NMEM stream, NMEM rset_nmem,
+ 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 rsets[2]; /* l and r argument */
+
+ rsets[0]=rpn_search_structure (zh, zs->u.complex->s1,
+ attributeSet, stream, rset_nmem,
+ sort_sequence,
+ num_bases, basenames);
+ if (rsets[0] == NULL)
+ return NULL;
+ rsets[1]=rpn_search_structure (zh, zs->u.complex->s2,
+ attributeSet, stream, rset_nmem,
+ sort_sequence,
+ num_bases, basenames);
+ if (rsets[1] == NULL)
+ {
+ rset_delete (rsets[0]);
+ return NULL;
+ }
+
+ switch (zop->which)
+ {
+ case Z_Operator_and:
+ r = rsmulti_and_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope,
+ 2, rsets);
+ break;
+ case Z_Operator_or:
+ r = rsmulti_or_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope,
+ 2, rsets);
+ break;
+ case Z_Operator_and_not:
+ r = rsbool_create_not(rset_nmem,key_it_ctrl, key_it_ctrl->scope,
+ rsets[0],rsets[1]);
+ break;
+ case Z_Operator_prox:
+ if (zop->u.prox->which != Z_ProximityOperator_known)
+ {
+ zh->errCode = 132;
+ return NULL;
+ }
+ if (*zop->u.prox->u.known != Z_ProxUnit_word)
+ {
+ char *val = (char *) nmem_malloc(stream, 16);
+ zh->errCode = 132;
+ zh->errString = val;
+ sprintf (val, "%d", *zop->u.prox->u.known);
+ return NULL;
+ }
+ else
+ {
+ /* new / old prox */
+ r = rsprox_create(rset_nmem,key_it_ctrl,key_it_ctrl->scope,
+ 2, rsets,
+ *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)
+ {
+ yaz_log(YLOG_DEBUG, "rpn_search_APT");
+ r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm,
+ attributeSet, stream, sort_sequence,
+ num_bases, basenames,rset_nmem);
+ }
+ else if (zs->u.simple->which == Z_Operand_resultSetId)
+ {
+ yaz_log(YLOG_DEBUG, "rpn_search_ref");
+ r = resultSetRef (zh, zs->u.simple->u.resultSetId);
+ if (!r)
+ {
+ r = rsnull_create (rset_nmem,key_it_ctrl);
+ zh->errCode = 30;
+ zh->errString =
+ nmem_strdup (stream, zs->u.simple->u.resultSetId);
+ return 0;
+ }
+ else
+ rset_dup(r);
+ }
+ else
+ {
+ zh->errCode = 3;
+ return 0;
+ }
+ }
+ else
+ {
+ zh->errCode = 3;
+ return 0;
+ }
+ return r;
+}
+
+
+RSET rpn_search(ZebraHandle zh, NMEM nmem, NMEM rset_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 = (Z_SortKeySpecList *)
+ nmem_malloc(nmem, sizeof(*sort_sequence));
+ sort_sequence->num_specs = 10; /* FIXME - Hard-coded number */
+ sort_sequence->specs = (Z_SortKeySpec **)
+ 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, rset_nmem,
+ sort_sequence, num_bases, basenames);
+ if (!rset)
+ return 0;
+
+ if (zh->errCode)
+ yaz_log(YLOG_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, rset_nmem);
+ else
+ {
+ yaz_log(YLOG_DEBUG, "resultSetSortSingle in rpn_search");
+ resultSetSortSingle (zh, nmem, sset, rset,
+ sort_sequence, &sort_status);
+ if (zh->errCode)
+ {
+ yaz_log(YLOG_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 = (struct 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;