+ *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:
+ case 102:
+ case -1:
+ 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, ")");
+ break;
+ default:
+ *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE;
+ return 0;
+ }
+ return 1;
+}
+
+static ZEBRA_RES 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,
+ struct ord_list **ol);
+
+static ZEBRA_RES term_limits_APT(ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ zint *hits_limit_value,
+ const char **term_ref_id_str,
+ NMEM nmem)
+{
+ AttrType term_ref_id_attr;
+ AttrType hits_limit_attr;
+ int term_ref_id_int;
+
+ attr_init(&hits_limit_attr, zapt, 9);
+ *hits_limit_value = attr_find(&hits_limit_attr, NULL);
+
+ attr_init(&term_ref_id_attr, zapt, 10);
+ term_ref_id_int = attr_find_ex(&term_ref_id_attr, NULL, term_ref_id_str);
+ if (term_ref_id_int != -1)
+ {
+ char *res = nmem_malloc(nmem, 20);
+ sprintf(res, "%d", term_ref_id_int);
+ *term_ref_id_str = res;
+ }
+
+ /* no limit given ? */
+ if (*hits_limit_value == -1)
+ {
+ if (*term_ref_id_str)
+ {
+ /* use global if term_ref is present */
+ *hits_limit_value = zh->approx_limit;
+ }
+ else
+ {
+ /* no counting if term_ref is not present */
+ *hits_limit_value = 0;
+ }
+ }
+ else if (*hits_limit_value == 0)
+ {
+ /* 0 is the same as global limit */
+ *hits_limit_value = zh->approx_limit;
+ }
+ yaz_log(YLOG_DEBUG, "term_limits_APT ref_id=%s limit=" ZINT_FORMAT,
+ *term_ref_id_str ? *term_ref_id_str : "none",
+ *hits_limit_value);
+ return ZEBRA_OK;
+}
+
+static ZEBRA_RES 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,
+ RSET *rset,
+ struct rset_key_control *kc)
+{
+ ZEBRA_RES res;
+ struct ord_list *ol;
+ zint hits_limit_value;
+ const char *term_ref_id_str = 0;
+ *rset = 0;
+
+ term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str,
+ stream);
+ grep_info->isam_p_indx = 0;
+ res = string_term(zh, zapt, term_sub, attributeSet, stream, grep_info,
+ reg_type, complete_flag, num_bases, basenames,
+ term_dst, xpath_use, &ol);
+ if (res != ZEBRA_OK)
+ return res;
+ if (!*term_sub) /* no more terms ? */
+ return res;
+ yaz_log(log_level_rpn, "term: %s", term_dst);
+ *rset = 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,
+ kc, kc->scope, ol, reg_type, hits_limit_value,
+ term_ref_id_str);
+ if (!*rset)
+ return ZEBRA_FAIL;
+ return ZEBRA_OK;
+}
+
+static ZEBRA_RES 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,
+ struct ord_list **ol)
+{
+ 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 */
+
+ *ol = ord_list_create(stream);
+
+ 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;
+ int relation_error;
+
+ termp = *term_sub;
+
+ if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
+ {
+ zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
+ basenames[base_no]);
+ return ZEBRA_FAIL;
+ }
+ 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,
+ reg_type,
+ 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 */
+ *ol = ord_list_append(stream, *ol, ord);
+ }
+ 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 */
+ if (use_string)
+ zebra_setError(zh,
+ YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
+ use_string);
+ else
+ zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
+ 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);
+
+ zebra_setError(zh,
+ YAZ_BIB1_UNSUPP_ATTRIBUTE_SET,
+ 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,
+ reg_type,
+ attp.attset_ordinal,
+ local_attr->local);
+ if (ord < 0)
+ continue;
+ *ol = ord_list_append(stream, *ol, ord);
+ 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];
+ }
+ }
+ bases_ok++;
+ if (prefix_len)
+ attr_ok = 1;
+
+ term_dict[prefix_len++] = ')';
+#if REG_TYPE_PREFIX
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = reg_type;
+ yaz_log(log_level_rpn, "reg_type = %d", term_dict[prefix_len-1]);
+#endif
+ 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,
+ &relation_error))
+ {
+ if (relation_error)
+ {
+ zebra_setError(zh, relation_error, 0);
+ return ZEBRA_FAIL;
+ }
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ strcat(term_dict, ")");
+ break;
+ case 103: /* Regexp-2 */
+ regex_range = 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ strcat(term_dict, ")");
+ break;
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ strcat(term_dict, ")");
+ break;
+ default:
+ zebra_setError_zint(zh,
+ YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE,
+ truncation_value);
+ return ZEBRA_FAIL;
+ }
+ 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)
+ return ZEBRA_FAIL;
+ *term_sub = termp;
+ yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx);
+ return ZEBRA_OK;
+}
+
+
+/* convert APT search term to UTF8 */
+static ZEBRA_RES 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 = (char *) 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);
+ zebra_setError(
+ zh,
+ YAZ_BIB1_QUERY_TERM_INCLUDES_CHARS_THAT_DO_NOT_TRANSLATE_INTO_,
+ 0);
+ return ZEBRA_FAIL;
+ }
+ *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:
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_CODED_VALUE_FOR_TERM, 0);
+ return ZEBRA_FAIL;
+ }
+ return ZEBRA_OK;
+}
+
+/* convert APT SCAN term to internal cmap */
+static ZEBRA_RES 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) == ZEBRA_FAIL)
+ return ZEBRA_FAIL; /* 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 ZEBRA_OK;
+}
+
+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 ZEBRA_RES grep_info_prepare(ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ struct grep_info *grep_info,
+ int reg_type)
+{
+ 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 ZEBRA_OK;
+ 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)
+ {
+ zebra_setError(zh, YAZ_BIB1_ILLEGAL_RESULT_SET_NAME, termset_name);
+ return ZEBRA_FAIL;
+ }
+ }
+ return ZEBRA_OK;
+}
+
+/**
+ \brief Create result set(s) for list of terms
+ \param zh Zebra Handle
+ \param termz term as used in query but converted to UTF-8
+ \param attributeSet default attribute set
+ \param stream memory for result
+ \param reg_type register type ('w', 'p',..)
+ \param complete_flag whether it's phrases or not
+ \param rank_type term flags for ranking
+ \param xpath_use use attribute for X-Path (-1 for no X-path)
+ \param num_bases number of databases
+ \param basenames array of databases
+ \param rset_mem memory for result sets
+ \param result_sets output result set for each term in list (output)
+ \param number number of output result sets
+ \param kc rset key control to be used for created result sets
+*/
+static ZEBRA_RES term_list_trunc(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,
+ RSET **result_sets, int *num_result_sets,
+ struct rset_key_control *kc)
+{
+ char term_dst[IT_MAX_WORD+1];
+ struct grep_info grep_info;
+ const char *termp = termz;
+ int alloc_sets = 0;
+
+ *num_result_sets = 0;
+ *term_dst = 0;
+ if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL)
+ return ZEBRA_FAIL;
+ while(1)
+ {
+ ZEBRA_RES res;
+
+ if (alloc_sets == *num_result_sets)
+ {
+ int add = 10;
+ RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) *
+ sizeof(*rnew));
+ if (alloc_sets)
+ memcpy(rnew, *result_sets, alloc_sets * sizeof(*rnew));
+ alloc_sets = alloc_sets + add;
+ *result_sets = rnew;
+ }
+ res = term_trunc(zh, zapt, &termp, attributeSet,
+ stream, &grep_info,
+ reg_type, complete_flag,
+ num_bases, basenames,
+ term_dst, rank_type,
+ xpath_use, rset_nmem,
+ &(*result_sets)[*num_result_sets],
+ kc);
+ if (res != ZEBRA_OK)
+ {
+ int i;
+ for (i = 0; i < *num_result_sets; i++)
+ rset_delete((*result_sets)[i]);
+ grep_info_delete (&grep_info);
+ return res;
+ }
+ if ((*result_sets)[*num_result_sets] == 0)
+ break;
+ (*num_result_sets)++;
+ }
+ grep_info_delete(&grep_info);
+ return ZEBRA_OK;
+}
+
+static ZEBRA_RES 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,
+ RSET *rset,
+ struct rset_key_control *kc)
+{
+ RSET *result_sets = 0;
+ int num_result_sets = 0;
+ ZEBRA_RES res =
+ term_list_trunc(zh, zapt, termz_org, attributeSet,
+ stream, reg_type, complete_flag,
+ rank_type, xpath_use,
+ num_bases, basenames,
+ rset_nmem,
+ &result_sets, &num_result_sets, kc);
+ if (res != ZEBRA_OK)
+ return res;
+ if (num_result_sets == 0)
+ *rset = rsnull_create (rset_nmem, kc, 0);
+ else if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsprox_create(rset_nmem, kc, kc->scope,
+ num_result_sets, result_sets,
+ 1 /* ordered */, 0 /* exclusion */,
+ 3 /* relation */, 1 /* distance */);
+ if (!*rset)
+ return ZEBRA_FAIL;
+ return ZEBRA_OK;
+}
+
+static ZEBRA_RES 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,
+ RSET *rset,
+ struct rset_key_control *kc)
+{
+ RSET *result_sets = 0;
+ int num_result_sets = 0;
+ ZEBRA_RES res =
+ term_list_trunc(zh, zapt, termz_org, attributeSet,
+ stream, reg_type, complete_flag,
+ rank_type, xpath_use,
+ num_bases, basenames,
+ rset_nmem,
+ &result_sets, &num_result_sets, kc);
+ if (res != ZEBRA_OK)
+ return res;
+ if (num_result_sets == 0)
+ *rset = rsnull_create (rset_nmem, kc, 0);
+ else if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsmulti_or_create(rset_nmem, kc, kc->scope, 0 /* termid */,
+ num_result_sets, result_sets);
+ if (!*rset)
+ return ZEBRA_FAIL;
+ return ZEBRA_OK;
+}
+
+static ZEBRA_RES 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,
+ RSET *rset,
+ struct rset_key_control *kc)
+{
+ RSET *result_sets = 0;
+ int num_result_sets = 0;
+ ZEBRA_RES res =
+ term_list_trunc(zh, zapt, termz_org, attributeSet,
+ stream, reg_type, complete_flag,
+ rank_type, xpath_use,
+ num_bases, basenames,
+ rset_nmem,
+ &result_sets, &num_result_sets,
+ kc);
+ if (res != ZEBRA_OK)
+ return res;
+ if (num_result_sets == 0)
+ *rset = rsnull_create (rset_nmem, kc, 0);
+ else if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsmulti_and_create(rset_nmem, kc, kc->scope,
+ num_result_sets, result_sets);
+ if (!*rset)
+ return ZEBRA_FAIL;
+ return ZEBRA_OK;
+}
+
+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,
+ int *error_code)
+{
+ AttrType relation;
+ int relation_value;
+ int term_value;
+ int r;
+ char *term_tmp = term_dict + strlen(term_dict);
+
+ *error_code = 0;
+ 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 -1:
+ case 3:
+ yaz_log(log_level_rpn, "Relation =");
+ sprintf(term_tmp, "(0*%d)", term_value);
+ break;
+ default:
+ *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE;
+ return 0;
+ }
+ 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 ZEBRA_RES 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 */
+
+ 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;
+ int relation_error = 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)
+ {
+ if (use_string)
+ zebra_setError(zh,
+ YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
+ use_string);
+ else
+ zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
+ use_value);
+ }
+ else
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, 0);
+ continue;
+ }
+ }
+ if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
+ {
+ zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
+ basenames[base_no]);
+ return ZEBRA_FAIL;
+ }
+ 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,
+ reg_type,
+ 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)
+ {
+ zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, use_value);
+ continue;
+ }
+ bases_ok++;
+ term_dict[prefix_len++] = ')';
+#if REG_TYPE_PREFIX
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = reg_type;
+ yaz_log(YLOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
+#endif
+ term_dict[prefix_len] = '\0';
+ if (!numeric_relation(zh, zapt, &termp, term_dict,
+ attributeSet, grep_info, &max_pos, reg_type,
+ term_dst, &relation_error))
+ {
+ if (relation_error)
+ {
+ zebra_setError(zh, relation_error, 0);
+ return ZEBRA_FAIL;
+ }
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ }
+ if (!bases_ok)
+ return ZEBRA_FAIL;
+ *term_sub = termp;
+ yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx);
+ return ZEBRA_OK;
+}
+
+
+static ZEBRA_RES 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,
+ RSET *rset,
+ struct rset_key_control *kc)
+{
+ char term_dst[IT_MAX_WORD+1];
+ const char *termp = termz;
+ RSET *result_sets = 0;
+ int num_result_sets = 0;
+ ZEBRA_RES res;
+ struct grep_info grep_info;
+ int alloc_sets = 0;
+ zint hits_limit_value;
+ const char *term_ref_id_str = 0;
+
+ term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, stream);
+
+ yaz_log(log_level_rpn, "APT_numeric t='%s'", termz);
+ if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL)
+ return ZEBRA_FAIL;
+ while (1)
+ {
+ if (alloc_sets == num_result_sets)
+ {
+ int add = 10;
+ RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) *
+ sizeof(*rnew));
+ if (alloc_sets)
+ memcpy(rnew, result_sets, alloc_sets * sizeof(*rnew));
+ alloc_sets = alloc_sets + add;
+ result_sets = rnew;
+ }
+ yaz_log(YLOG_DEBUG, "APT_numeric termp=%s", termp);
+ grep_info.isam_p_indx = 0;
+ res = numeric_term(zh, zapt, &termp, attributeSet, &grep_info,
+ reg_type, complete_flag, num_bases, basenames,
+ term_dst, xpath_use,
+ stream);
+ if (res == ZEBRA_FAIL || termp == 0)
+ break;
+ yaz_log(YLOG_DEBUG, "term: %s", term_dst);
+ result_sets[num_result_sets] =
+ 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,
+ kc, kc->scope, 0, reg_type,
+ hits_limit_value,
+ term_ref_id_str);
+ if (!result_sets[num_result_sets])
+ break;
+ num_result_sets++;
+ }
+ grep_info_delete(&grep_info);
+ if (termp)
+ {
+ int i;
+ for (i = 0; i<num_result_sets; i++)
+ rset_delete(result_sets[i]);
+ return ZEBRA_FAIL;
+ }
+ if (num_result_sets == 0)
+ *rset = rsnull_create(rset_nmem, kc, 0);
+ if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsmulti_and_create(rset_nmem, kc, kc->scope,
+ num_result_sets, result_sets);
+ if (!*rset)
+ return ZEBRA_FAIL;
+ return ZEBRA_OK;
+}
+
+static ZEBRA_RES 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 *rset,
+ struct rset_key_control *kc)
+{
+ RSFD rsfd;
+ struct it_key key;
+ int sys;
+ *rset = rstemp_create(rset_nmem, kc, kc->scope,
+ res_get (zh->res, "setTmpDir"),0 );
+ rsfd = rset_open(*rset, 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 ZEBRA_OK;
+}
+
+static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet, NMEM stream,
+ Z_SortKeySpecList *sort_sequence,
+ const char *rank_type,
+ NMEM rset_nmem,
+ RSET *rset,
+ struct rset_key_control *kc)
+{
+ int i;
+ int sort_relation_value;
+ AttrType sort_relation_type;
+ Z_SortKeySpec *sks;
+ Z_SortKey *sk;
+ 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);
+
+ 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 ZEBRA_FAIL;
+
+ 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 = zapt->attributes;
+
+ 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;
+ *rset = rsnull_create (rset_nmem, kc, 0);
+ return ZEBRA_OK;
+}
+
+
+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,
+ struct rset_key_control *kc)
+{
+ 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, reg_type,
+ 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') == ZEBRA_FAIL)
+ return rsnull_create(rset_nmem, kc, 0);
+
+ if (ord < 0)
+ return rsnull_create(rset_nmem, kc, 0);
+ 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++] = ')';
+#if REG_TYPE_PREFIX
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = reg_type;
+#endif
+ 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,
+ kc, kc->scope, 0, reg_type, 0 /* hits_limit */,
+ 0 /* term_ref_id_str */);
+ grep_info_delete(&grep_info);
+ return rset;
+}
+
+static
+ZEBRA_RES 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,
+ RSET *rset_out,
+ struct rset_key_control *kc)
+{
+ oid_value curAttributeSet = attributeSet;
+ int base_no;
+ int i;
+
+ if (xpath_len < 0)
+ {
+ *rset_out = rset;
+ return ZEBRA_OK;
+ }
+
+ 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]))
+ {
+ zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
+ basenames[base_no]);
+ *rset_out = rset;
+ return ZEBRA_FAIL;
+ }
+ 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, kc);
+ 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, kc);
+
+ rset_end_tag = xpath_trunc(zh, stream, '0',
+ xpath_rev, 2, curAttributeSet, rset_nmem, kc);
+
+ rset = rsbetween_create(rset_nmem, kc, kc->scope,
+ rset_start_tag, rset,
+ rset_end_tag, rset_attr);
+ }
+ first_path = 0;
+ }
+ }
+ *rset_out = rset;
+ return ZEBRA_OK;
+}
+
+static ZEBRA_RES 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,
+ RSET *rset,
+ struct rset_key_control *kc)
+{
+ ZEBRA_RES res = ZEBRA_OK;
+ unsigned reg_id;
+ char *search_type = NULL;
+ char rank_type[128];
+ int complete_flag;
+ int sort_flag;
+ char termz[IT_MAX_WORD+1];
+ 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) == ZEBRA_FAIL)
+ return ZEBRA_FAIL;
+
+ if (sort_flag)
+ return rpn_sort_spec(zh, zapt, attributeSet, stream, sort_sequence,
+ rank_type, rset_nmem, rset, kc);
+ /* consider if an X-Path query is used */
+ xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream);
+ if (xpath_len >= 0)
+ {
+ xpath_use = 1016; /* searching for element by default */
+ if (xpath[xpath_len-1].part[0] == '@')
+ xpath_use = 1015; /* last step an attribute .. */
+ }
+
+ /* search using one of the various search type strategies
+ termz is our UTF-8 search term
+ attributeSet is top-level default attribute set
+ stream is ODR for search
+ reg_id is the register type
+ complete_flag is 1 for complete subfield, 0 for incomplete
+ xpath_use is use-attribute to be used for X-Path search, 0 for none
+ */
+ if (!strcmp(search_type, "phrase"))
+ {
+ res = rpn_search_APT_phrase(zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem,
+ rset, kc);
+ }
+ else if (!strcmp(search_type, "and-list"))
+ {
+ res = rpn_search_APT_and_list(zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem,
+ rset, kc);
+ }
+ else if (!strcmp(search_type, "or-list"))
+ {
+ res = rpn_search_APT_or_list(zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem,
+ rset, kc);
+ }
+ else if (!strcmp(search_type, "local"))
+ {
+ res = rpn_search_APT_local(zh, zapt, termz, attributeSet, stream,
+ rank_type, rset_nmem, rset, kc);
+ }
+ else if (!strcmp(search_type, "numeric"))
+ {
+ res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, stream,
+ reg_id, complete_flag, rank_type,
+ xpath_use,
+ num_bases, basenames, rset_nmem,
+ rset, kc);
+ }
+ else
+ {
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE, 0);
+ res = ZEBRA_FAIL;
+ }
+ if (res != ZEBRA_OK)
+ return res;
+ if (!*rset)
+ return ZEBRA_FAIL;
+ return rpn_search_xpath(zh, attributeSet, num_bases, basenames,
+ stream, rank_type, *rset,
+ xpath_len, xpath, rset_nmem, rset, kc);
+}
+
+static ZEBRA_RES 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 **result_sets, int *num_result_sets,
+ Z_Operator *parent_op,
+ struct rset_key_control *kc);
+
+ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs,
+ oid_value attributeSet,
+ NMEM stream, NMEM rset_nmem,
+ Z_SortKeySpecList *sort_sequence,
+ int num_bases, char **basenames,
+ RSET *result_set)
+{
+ RSET *result_sets = 0;
+ int num_result_sets = 0;
+ ZEBRA_RES res;
+ struct rset_key_control *kc = zebra_key_control_create(zh);
+
+ res = rpn_search_structure(zh, zs, attributeSet,
+ stream, rset_nmem,
+ sort_sequence,
+ num_bases, basenames,
+ &result_sets, &num_result_sets,
+ 0 /* no parent op */,
+ kc);
+ if (res != ZEBRA_OK)
+ {
+ int i;
+ for (i = 0; i<num_result_sets; i++)
+ rset_delete(result_sets[i]);
+ *result_set = 0;
+ }
+ else
+ {
+ assert(num_result_sets == 1);
+ assert(result_sets);
+ assert(*result_sets);
+ *result_set = *result_sets;
+ }
+ (*kc->dec)(kc);
+ return res;
+}
+
+ZEBRA_RES 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 **result_sets, int *num_result_sets,
+ Z_Operator *parent_op,
+ struct rset_key_control *kc)
+{
+ *num_result_sets = 0;
+ if (zs->which == Z_RPNStructure_complex)
+ {
+ ZEBRA_RES res;
+ Z_Operator *zop = zs->u.complex->roperator;
+ RSET *result_sets_l = 0;
+ int num_result_sets_l = 0;
+ RSET *result_sets_r = 0;
+ int num_result_sets_r = 0;
+
+ res = rpn_search_structure(zh, zs->u.complex->s1,
+ attributeSet, stream, rset_nmem,
+ sort_sequence,
+ num_bases, basenames,
+ &result_sets_l, &num_result_sets_l,
+ zop, kc);
+ if (res != ZEBRA_OK)
+ {
+ int i;
+ for (i = 0; i<num_result_sets_l; i++)
+ rset_delete(result_sets_l[i]);
+ return res;
+ }
+ res = rpn_search_structure(zh, zs->u.complex->s2,
+ attributeSet, stream, rset_nmem,
+ sort_sequence,
+ num_bases, basenames,
+ &result_sets_r, &num_result_sets_r,
+ zop, kc);
+ if (res != ZEBRA_OK)
+ {
+ int i;
+ for (i = 0; i<num_result_sets_l; i++)
+ rset_delete(result_sets_l[i]);
+ for (i = 0; i<num_result_sets_r; i++)
+ rset_delete(result_sets_r[i]);
+ return res;
+ }
+
+ /* make a new list of result for all children */
+ *num_result_sets = num_result_sets_l + num_result_sets_r;
+ *result_sets = nmem_malloc(stream, *num_result_sets *
+ sizeof(**result_sets));
+ memcpy(*result_sets, result_sets_l,
+ num_result_sets_l * sizeof(**result_sets));
+ memcpy(*result_sets + num_result_sets_l, result_sets_r,
+ num_result_sets_r * sizeof(**result_sets));
+
+ if (!parent_op || parent_op->which != zop->which
+ || (zop->which != Z_Operator_and &&
+ zop->which != Z_Operator_or))
+ {
+ /* parent node different from this one (or non-present) */
+ /* we must combine result sets now */
+ RSET rset;
+ switch (zop->which)
+ {
+ case Z_Operator_and:
+ rset = rsmulti_and_create(rset_nmem, kc,
+ kc->scope,
+ *num_result_sets, *result_sets);
+ break;
+ case Z_Operator_or:
+ rset = rsmulti_or_create(rset_nmem, kc,
+ kc->scope, 0, /* termid */
+ *num_result_sets, *result_sets);
+ break;
+ case Z_Operator_and_not:
+ rset = rsbool_create_not(rset_nmem, kc,
+ kc->scope,
+ (*result_sets)[0],
+ (*result_sets)[1]);
+ break;
+ case Z_Operator_prox:
+ if (zop->u.prox->which != Z_ProximityOperator_known)
+ {
+ zebra_setError(zh,
+ YAZ_BIB1_UNSUPP_PROX_UNIT_CODE,
+ 0);
+ return ZEBRA_FAIL;
+ }
+ if (*zop->u.prox->u.known != Z_ProxUnit_word)
+ {
+ zebra_setError_zint(zh,
+ YAZ_BIB1_UNSUPP_PROX_UNIT_CODE,
+ *zop->u.prox->u.known);
+ return ZEBRA_FAIL;
+ }
+ else
+ {
+ rset = rsprox_create(rset_nmem, kc,
+ kc->scope,
+ *num_result_sets, *result_sets,
+ *zop->u.prox->ordered,
+ (!zop->u.prox->exclusion ?
+ 0 : *zop->u.prox->exclusion),
+ *zop->u.prox->relationType,
+ *zop->u.prox->distance );
+ }
+ break;
+ default:
+ zebra_setError(zh, YAZ_BIB1_OPERATOR_UNSUPP, 0);
+ return ZEBRA_FAIL;
+ }
+ *num_result_sets = 1;
+ *result_sets = nmem_malloc(stream, *num_result_sets *
+ sizeof(**result_sets));
+ (*result_sets)[0] = rset;
+ }
+ }
+ else if (zs->which == Z_RPNStructure_simple)
+ {
+ RSET rset;
+ ZEBRA_RES res;
+
+ if (zs->u.simple->which == Z_Operand_APT)
+ {
+ yaz_log(YLOG_DEBUG, "rpn_search_APT");
+ res = rpn_search_APT(zh, zs->u.simple->u.attributesPlusTerm,
+ attributeSet, stream, sort_sequence,
+ num_bases, basenames, rset_nmem, &rset,
+ kc);
+ if (res != ZEBRA_OK)
+ return res;
+ }
+ else if (zs->u.simple->which == Z_Operand_resultSetId)
+ {
+ yaz_log(YLOG_DEBUG, "rpn_search_ref");
+ rset = resultSetRef(zh, zs->u.simple->u.resultSetId);
+ if (!rset)
+ {
+ zebra_setError(zh,
+ YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
+ zs->u.simple->u.resultSetId);
+ return ZEBRA_FAIL;
+ }
+ rset_dup(rset);
+ }
+ else
+ {
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0);
+ return ZEBRA_FAIL;
+ }
+ *num_result_sets = 1;
+ *result_sets = nmem_malloc(stream, *num_result_sets *
+ sizeof(**result_sets));
+ (*result_sets)[0] = rset;
+ }
+ else
+ {
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0);
+ return ZEBRA_FAIL;
+ }
+ return ZEBRA_OK;
+}
+
+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;
+ else
+ idx = - pos - 1;
+
+ if (idx < 0)
+ return 0;
+ scan_info->list[idx].term = (char *)
+ 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;
+}
+
+void zebra_term_untrans_iconv(ZebraHandle zh, NMEM stream, int reg_type,
+ char **dst, const char *src)
+{
+ char term_src[IT_MAX_WORD];
+ char term_dst[IT_MAX_WORD];
+
+ zebra_term_untrans (zh, reg_type, term_src, src);
+
+ if (zh->iconv_from_utf8 != 0)
+ {
+ int len;
+ char *inbuf = term_src;
+ size_t inleft = strlen(term_src);
+ char *outbuf = term_dst;
+ size_t outleft = sizeof(term_dst)-1;
+ size_t ret;
+
+ ret = yaz_iconv (zh->iconv_from_utf8, &inbuf, &inleft,
+ &outbuf, &outleft);
+ if (ret == (size_t)(-1))
+ len = 0;
+ else
+ len = outbuf - term_dst;
+ *dst = nmem_malloc(stream, len + 1);
+ if (len > 0)
+ memcpy (*dst, term_dst, len);
+ (*dst)[len] = '\0';
+ }
+ else
+ *dst = nmem_strdup(stream, term_src);
+}
+
+static void count_set(ZebraHandle zh, RSET rset, zint *count)
+{
+ zint psysno = 0;
+ struct it_key key;
+ RSFD rfd;
+
+ yaz_log(YLOG_DEBUG, "count_set");
+
+ rset->hits_limit = zh->approx_limit;
+
+ *count = 0;
+ rfd = rset_open(rset, RSETF_READ);
+ while (rset_read(rfd, &key,0 /* never mind terms */))
+ {
+ if (key.mem[0] != psysno)
+ {
+ psysno = key.mem[0];
+ if (rfd->counted_items >= rset->hits_limit)
+ break;
+ }
+ }
+ rset_close (rfd);
+ *count = rset->hits_count;
+}
+
+ZEBRA_RES 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, RSET limit_set, int return_zero)
+{
+ 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;
+ const char *use_string = 0;
+ struct scan_info *scan_info_array;
+ ZebraScanEntry *glist;
+ int ords[32], ord_no = 0;
+ int ptr[32];
+
+ int bases_ok = 0; /* no of databases with OK attribute */
+ int errCode = 0; /* err code (if any is not OK) */
+ char *errString = 0; /* addinfo */
+
+ unsigned reg_id;
+ char *search_type = NULL;
+ char rank_type[128];
+ int complete_flag;
+ int sort_flag;
+ NMEM rset_nmem = NULL;
+ struct rset_key_control *kc = 0;
+
+ *list = 0;
+ *is_partial = 0;
+
+ if (attributeset == VAL_NONE)
+ attributeset = VAL_BIB1;
+
+ if (!limit_set)
+ {
+ AttrType termset;
+ int termset_value_numeric;
+ const char *termset_value_string;
+ 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;
+
+ limit_set = resultSetRef (zh, termset_name);
+ }
+ }
+
+ yaz_log(YLOG_DEBUG, "position = %d, num = %d set=%d",
+ pos, num, attributeset);
+
+ attr_init(&use, zapt, 1);
+ use_value = attr_find_ex(&use, &attributeset, &use_string);
+
+ if (zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type,
+ rank_type, &complete_flag, &sort_flag))
+ {
+ *num_entries = 0;
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_TYPE, 0);
+ return ZEBRA_FAIL;
+ }
+ yaz_log(YLOG_DEBUG, "use_value = %d", use_value);
+
+ if (use_value == -1)
+ use_value = 1016;
+ for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
+ {
+ data1_local_attribute *local_attr;
+ attent attp;
+ int ord;
+
+ if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
+ {
+ zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
+ basenames[base_no]);
+ *num_entries = 0;
+ return ZEBRA_FAIL;
+ }
+
+ if (use_string &&
+ (ord = zebraExplain_lookup_attr_str(zh->reg->zei, reg_id,
+ use_string)) >= 0)
+ {
+ /* we have a match for a raw string attribute */
+ if (ord > 0)
+ ords[ord_no++] = ord;
+ attp.local_attributes = 0; /* no more attributes */
+ }
+ else
+ {
+ int r;
+
+ if ((r = att_getentbyatt (zh, &attp, attributeset, use_value,
+ use_string)))
+ {
+ yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
+ attributeset, use_value);
+ if (r == -1)
+ {
+ errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
+ if (use_string)
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
+ use_string);
+ else
+ zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
+ use_value);
+ }
+ else
+ {
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, 0);
+ }
+ continue;
+ }
+ }
+ bases_ok++;
+ for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
+ local_attr = local_attr->next)
+ {
+ ord = zebraExplain_lookup_attr_su(zh->reg->zei, reg_id,
+ attp.attset_ordinal,
+ local_attr->local);
+ if (ord > 0)
+ ords[ord_no++] = ord;
+ }
+ }
+ if (!bases_ok && errCode)
+ {
+ zebra_setError(zh, errCode, errString);
+ *num_entries = 0;
+ return ZEBRA_FAIL;
+ }
+ if (ord_no == 0)
+ {
+ *num_entries = 0;
+ return ZEBRA_OK;
+ }
+ /* prepare dictionary scanning */
+ if (num < 1)
+ {
+ *num_entries = 0;
+ return ZEBRA_OK;
+ }
+ before = pos-1;
+ if (before < 0)
+ before = 0;
+ after = 1+num-pos;
+ if (after < 0)
+ after = 0;
+ yaz_log(YLOG_DEBUG, "rpn_scan pos=%d num=%d before=%d "
+ "after=%d before+after=%d",
+ pos, num, before, after, before+after);
+ scan_info_array = (struct scan_info *)
+ 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, reg_id, &rcmi);
+
+ scan_info->before = before;
+ scan_info->after = after;
+ scan_info->odr = stream;
+
+ scan_info->list = (struct scan_info_entry *)
+ 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_encode (ords[i], termz + prefix_len);
+#if REG_TYPE_PREFIX
+ termz[prefix_len++] = reg_id;
+#endif
+ termz[prefix_len] = 0;
+ strcpy(scan_info->prefix, termz);
+
+ if (trans_scan_term(zh, zapt, termz+prefix_len, reg_id) == ZEBRA_FAIL)
+ return ZEBRA_FAIL;
+
+ dict_scan(zh->reg->dict, termz, &before_tmp, &after_tmp,
+ scan_info, scan_handle);
+ }
+ glist = (ZebraScanEntry *)
+ odr_malloc(stream, (before+after)*sizeof(*glist));
+
+ rset_nmem = nmem_create();
+ kc = zebra_key_control_create(zh);
+
+ /* 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 = 0;
+ int lo = i + pos-1; /* offset in result list */
+
+ /* find: j0 is the first of the minimal values */
+ for (j = 0; j < ord_no; j++)
+ {
+ if (ptr[j] < before+after && ptr[j] >= 0 &&
+ (tst = scan_info_array[j].list[ptr[j]].term) &&
+ (!mterm || strcmp (tst, mterm) < 0))
+ {
+ j0 = j;
+ mterm = tst;
+ }
+ }
+ if (j0 == -1)
+ break; /* no value found, stop */
+
+ /* get result set for first one , but only if it's within bounds */
+ if (lo >= 0)
+ {
+ /* get result set for first term */
+ zebra_term_untrans_iconv(zh, stream->mem, reg_id,
+ &glist[lo].term, mterm);
+ rset = rset_trunc(zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
+ glist[lo].term, strlen(glist[lo].term),
+ NULL, 0, zapt->term->which, rset_nmem,
+ kc, kc->scope, 0, reg_id, 0 /* hits_limit */,
+ 0 /* term_ref_id_str */);
+ }
+ ptr[j0]++; /* move index for this set .. */
+ /* get result set for remaining scan terms */
+ for (j = j0+1; j<ord_no; j++)
+ {
+ if (ptr[j] < before+after && ptr[j] >= 0 &&
+ (tst = scan_info_array[j].list[ptr[j]].term) &&
+ !strcmp (tst, mterm))
+ {
+ if (lo >= 0)
+ {
+ RSET rsets[2];
+
+ rsets[0] = rset;
+ rsets[1] =
+ rset_trunc(
+ zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
+ glist[lo].term,
+ strlen(glist[lo].term), NULL, 0,
+ zapt->term->which,rset_nmem,
+ kc, kc->scope, 0, reg_id, 0 /* hits_limit */,
+ 0 /* term_ref_id_str */ );
+ rset = rsmulti_or_create(rset_nmem, kc,
+ kc->scope, 0 /* termid */,
+ 2, rsets);
+ }
+ ptr[j]++;
+ }
+ }
+ if (lo >= 0)
+ {
+ zint count;
+ /* merge with limit_set if given */
+ if (limit_set)
+ {
+ RSET rsets[2];
+ rsets[0] = rset;
+ rsets[1] = rset_dup(limit_set);
+
+ rset = rsmulti_and_create(rset_nmem, kc,
+ kc->scope,
+ 2, rsets);
+ }
+ /* count it */
+ count_set(zh, rset, &count);
+ glist[lo].occurrences = count;
+ rset_delete(rset);
+ }
+ }
+ if (i < after)
+ {
+ *num_entries -= (after-i);
+ *is_partial = 1;
+ if (*num_entries < 0)
+ {
+ (*kc->dec)(kc);
+ nmem_destroy(rset_nmem);
+ *num_entries = 0;
+ return ZEBRA_OK;
+ }
+ }
+ /* 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;
+ int lo = before-1-i; /* offset in result list */
+ zint count;
+
+ for (j = 0; j <ord_no; j++)
+ {
+ if (ptr[j] < before && ptr[j] >= 0 &&
+ (tst = scan_info_array[j].list[before-1-ptr[j]].term) &&
+ (!mterm || strcmp (tst, mterm) > 0))
+ {
+ j0 = j;
+ mterm = tst;
+ }
+ }
+ if (j0 == -1)
+ break;
+
+ zebra_term_untrans_iconv(zh, stream->mem, reg_id,
+ &glist[lo].term, mterm);
+
+ rset = rset_trunc
+ (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
+ glist[lo].term, strlen(glist[lo].term),
+ NULL, 0, zapt->term->which, rset_nmem,
+ kc, kc->scope, 0, reg_id, 0 /* hits_limit */,
+ 0 /* term_ref_id_str */);
+
+ ptr[j0]++;
+
+ for (j = j0+1; j<ord_no; j++)
+ {
+ if (ptr[j] < before && ptr[j] >= 0 &&
+ (tst = scan_info_array[j].list[before-1-ptr[j]].term) &&
+ !strcmp (tst, mterm))
+ {
+ RSET rsets[2];
+
+ rsets[0] = rset;
+ rsets[1] = rset_trunc(
+ zh,
+ &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
+ glist[lo].term,
+ strlen(glist[lo].term), NULL, 0,
+ zapt->term->which, rset_nmem,
+ kc, kc->scope, 0, reg_id, 0 /* hits_limit */,
+ 0 /* term_ref_id_str */);
+ rset = rsmulti_or_create(rset_nmem, kc,
+ kc->scope, 0 /* termid */, 2, rsets);
+
+ ptr[j]++;
+ }
+ }
+ if (limit_set)
+ {
+ RSET rsets[2];
+ rsets[0] = rset;
+ rsets[1] = rset_dup(limit_set);
+
+ rset = rsmulti_and_create(rset_nmem, kc,
+ kc->scope, 2, rsets);
+ }
+ count_set(zh, rset, &count);
+ glist[lo].occurrences = count;
+ rset_delete (rset);
+ }
+ (*kc->dec)(kc);
+ nmem_destroy(rset_nmem);
+ i = before-i;
+ if (i)
+ {
+ *is_partial = 1;
+ *position -= i;
+ *num_entries -= i;
+ if (*num_entries <= 0)
+ {
+ *num_entries = 0;
+ return ZEBRA_OK;
+ }
+ }
+
+ *list = glist + i; /* list is set to first 'real' entry */
+
+ yaz_log(YLOG_DEBUG, "position = %d, num_entries = %d",
+ *position, *num_entries);
+ return ZEBRA_OK;