+/*
+ * > abc ([b-].*|a[c-].*|ab[d-].*|abc.+)
+ * ([^-a].*|a[^-b].*ab[^-c].*|abc.+)
+ * >= abc ([b-].*|a[c-].*|ab[c-].*)
+ * ([^-a].*|a[^-b].*|ab[c-].*)
+ * < abc ([-0].*|a[-a].*|ab[-b].*)
+ * ([^a-].*|a[^b-].*|ab[^c-].*)
+ * <= abc ([-0].*|a[-a].*|ab[-b].*|abc)
+ * ([^a-].*|a[^b-].*|ab[^c-].*|abc)
+ */
+static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
+ const char **term_sub, char *term_dict,
+ oid_value attributeSet,
+ int reg_type, int space_split, char *term_dst,
+ int *error_code)
+{
+ AttrType relation;
+ int relation_value;
+ int i;
+ char *term_tmp = term_dict + strlen(term_dict);
+ char term_component[2*IT_MAX_WORD+20];
+
+ attr_init(&relation, zapt, 2);
+ relation_value = attr_find(&relation, NULL);
+
+ *error_code = 0;
+ yaz_log(YLOG_DEBUG, "string relation value=%d", relation_value);
+ switch (relation_value)
+ {
+ case 1:
+ 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++ = '[';
+
+ *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:
+ 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 = 117;
+ 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);
+
+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)
+{
+ ZEBRA_RES res;
+ *rset = 0;
+ 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);
+ 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,
+ key_it_ctrl, key_it_ctrl->scope);
+ if (!*rset)
+ return ZEBRA_FAIL;
+ return ZEBRA_OK;
+}
+
+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 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)
+{
+ 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;
+ int relation_error;
+
+ termp = *term_sub;
+
+ if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
+ {
+ zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE;
+ zh->errString = 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,
+ 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 = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
+ 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 = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET;
+ 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];
+ }
+ }
+ bases_ok++;
+ if (prefix_len)
+ 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,
+ &relation_error))
+ {
+ if (relation_error)
+ {
+ zh->errCode = relation_error;
+ 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 */
+ 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))
+ {
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ 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))
+ {
+ *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:
+ zh->errCode = YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE;
+ zh->errString = nmem_strdup_i(stream, 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)
+ {
+ zh->errCode = errCode;
+ zh->errString = errString;
+ 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 = 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 =
+ YAZ_BIB1_QUERY_TERM_INCLUDES_CHARS_THAT_DO_NOT_TRANSLATE_INTO_;
+ 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 = YAZ_BIB1_UNSUPP_CODED_VALUE_FOR_TERM;
+ 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;
+}
+
+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 = YAZ_BIB1_ILLEGAL_RESULT_SET_NAME;
+ zh->errString = nmem_strdup(stream, termset_name);
+ return -1;
+ }
+ }
+ return 0;
+}
+
+
+static ZEBRA_RES term_list_trunc(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 **result_sets, int *num_result_sets)
+{
+ char term_dst[IT_MAX_WORD+1];
+ struct grep_info grep_info;
+ char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
+ 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, stream))
+ 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]);
+ 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)
+{
+ 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);
+ if (res != ZEBRA_OK)
+ return res;
+ if (num_result_sets == 0)
+ *rset = rsnull_create (rset_nmem, key_it_ctrl);
+ else if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsprox_create(rset_nmem, key_it_ctrl, key_it_ctrl->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)
+{
+ 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);
+ if (res != ZEBRA_OK)
+ return res;
+ if (num_result_sets == 0)
+ *rset = rsnull_create (rset_nmem, key_it_ctrl);
+ else if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsmulti_or_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope,
+ 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)
+{
+ 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);
+ if (res != ZEBRA_OK)
+ return res;
+ if (num_result_sets == 0)
+ *rset = rsnull_create (rset_nmem, key_it_ctrl);
+ else if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsmulti_and_create(rset_nmem, key_it_ctrl, key_it_ctrl->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 = 117;
+ 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 */
+ 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;
+ 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)
+ {
+ errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
+ if (use_string)
+ errString = nmem_strdup(stream, use_string);
+ else
+ errString = nmem_strdup_i (stream, use_value);
+ }
+ else
+ errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET;
+ continue;
+ }
+ }
+ if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
+ {
+ zh->errCode = YAZ_BIB1_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 = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
+ 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, &relation_error))
+ {
+ if (relation_error)
+ {
+ zh->errCode = relation_error;
+ zh->errString = 0;
+ return ZEBRA_FAIL;
+ }
+ *term_sub = 0;
+ return ZEBRA_OK;
+ }
+ }
+ if (!bases_ok)
+ {
+ zh->errCode = errCode;
+ zh->errString = errString;
+ 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)
+{
+ 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;
+
+ yaz_log(log_level_rpn, "APT_numeric t='%s'", termz);
+ if (grep_info_prepare(zh, zapt, &grep_info, reg_type, stream))
+ 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,
+ key_it_ctrl,key_it_ctrl->scope);
+ 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, key_it_ctrl);
+ if (num_result_sets == 1)
+ *rset = result_sets[0];
+ else
+ *rset = rsmulti_and_create(rset_nmem, key_it_ctrl, key_it_ctrl->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)
+{
+ RSFD rsfd;
+ struct it_key key;
+ int sys;
+ *rset = rstemp_create(rset_nmem,key_it_ctrl,key_it_ctrl->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,
+ RSET *rset)
+{
+ 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 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 = (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;
+ *rset = rsnull_create (NULL, key_it_ctrl);
+ 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)