1 /* $Id: zrpn.c,v 1.187 2005-05-03 13:57:44 adam Exp $
2 Copyright (C) 1995-2005
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
32 #include <yaz/diagbib1.h>
34 #include <zebra_xpath.h>
39 struct rpn_char_map_info
50 Z_AttributesPlusTerm *zapt;
54 static int log_level_set = 0;
55 static int log_level_rpn = 0;
57 static const char **rpn_char_map_handler(void *vp, const char **from, int len)
59 struct rpn_char_map_info *p = (struct rpn_char_map_info *) vp;
60 const char **out = zebra_maps_input(p->zm, p->reg_type, from, len, 0);
64 const char *outp = *out;
65 yaz_log(YLOG_LOG, "---");
68 yaz_log(YLOG_LOG, "%02X", *outp);
76 static void rpn_char_map_prepare(struct zebra_register *reg, int reg_type,
77 struct rpn_char_map_info *map_info)
79 map_info->zm = reg->zebra_maps;
80 map_info->reg_type = reg_type;
81 dict_grep_cmap(reg->dict, map_info, rpn_char_map_handler);
84 static int attr_find_ex(AttrType *src, oid_value *attributeSetP,
85 const char **string_value)
89 num_attributes = src->zapt->attributes->num_attributes;
90 while (src->major < num_attributes)
92 Z_AttributeElement *element;
94 element = src->zapt->attributes->attributes[src->major];
95 if (src->type == *element->attributeType)
97 switch (element->which)
99 case Z_AttributeValue_numeric:
101 if (element->attributeSet && attributeSetP)
105 attrset = oid_getentbyoid(element->attributeSet);
106 *attributeSetP = attrset->value;
108 return *element->value.numeric;
110 case Z_AttributeValue_complex:
111 if (src->minor >= element->value.complex->num_list)
113 if (element->attributeSet && attributeSetP)
117 attrset = oid_getentbyoid(element->attributeSet);
118 *attributeSetP = attrset->value;
120 if (element->value.complex->list[src->minor]->which ==
121 Z_StringOrNumeric_numeric)
125 *element->value.complex->list[src->minor-1]->u.numeric;
127 else if (element->value.complex->list[src->minor]->which ==
128 Z_StringOrNumeric_string)
134 element->value.complex->list[src->minor-1]->u.string;
148 static int attr_find(AttrType *src, oid_value *attributeSetP)
150 return attr_find_ex(src, attributeSetP, 0);
153 static void attr_init(AttrType *src, Z_AttributesPlusTerm *zapt,
176 static void term_untrans(ZebraHandle zh, int reg_type,
177 char *dst, const char *src)
182 const char *cp = zebra_maps_output(zh->reg->zebra_maps,
184 if (!cp && len < IT_MAX_WORD-1)
187 while (*cp && len < IT_MAX_WORD-1)
193 static void add_isam_p(const char *name, const char *info,
198 log_level_rpn = yaz_log_module_level("rpn");
201 if (p->isam_p_indx == p->isam_p_size)
203 ISAM_P *new_isam_p_buf;
207 p->isam_p_size = 2*p->isam_p_size + 100;
208 new_isam_p_buf = (ISAM_P *) xmalloc(sizeof(*new_isam_p_buf) *
212 memcpy(new_isam_p_buf, p->isam_p_buf,
213 p->isam_p_indx * sizeof(*p->isam_p_buf));
214 xfree(p->isam_p_buf);
216 p->isam_p_buf = new_isam_p_buf;
219 new_term_no = (int *) xmalloc(sizeof(*new_term_no) * p->isam_p_size);
222 memcpy(new_term_no, p->isam_p_buf,
223 p->isam_p_indx * sizeof(*p->term_no));
226 p->term_no = new_term_no;
229 assert(*info == sizeof(*p->isam_p_buf));
230 memcpy(p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
237 char term_tmp[IT_MAX_WORD];
239 int len = key_SU_decode (&su_code, name);
241 term_untrans (p->zh, p->reg_type, term_tmp, name+len+1);
242 yaz_log(log_level_rpn, "grep: %d %c %s", su_code, name[len], term_tmp);
243 zebraExplain_lookup_ord (p->zh->reg->zei,
244 su_code, &db, &set, &use);
245 yaz_log(log_level_rpn, "grep: set=%d use=%d db=%s", set, use, db);
247 resultSetAddTerm(p->zh, p->termset, name[len], db,
254 static int grep_handle(char *name, const char *info, void *p)
256 add_isam_p(name, info, (struct grep_info *) p);
260 static int term_pre(ZebraMaps zebra_maps, int reg_type, const char **src,
261 const char *ct1, const char *ct2, int first)
263 const char *s1, *s0 = *src;
266 /* skip white space */
269 if (ct1 && strchr(ct1, *s0))
271 if (ct2 && strchr(ct2, *s0))
274 map = zebra_maps_input(zebra_maps, reg_type, &s1, strlen(s1), first);
275 if (**map != *CHR_SPACE)
284 static void esc_str(char *out_buf, int out_size,
285 const char *in_buf, int in_size)
291 assert(out_size > 20);
293 for (k = 0; k<in_size; k++)
295 int c = in_buf[k] & 0xff;
297 if (c < 32 || c > 126)
301 sprintf(out_buf +strlen(out_buf), "%02X:%c ", c, pc);
302 if (strlen(out_buf) > out_size-20)
304 strcat(out_buf, "..");
310 #define REGEX_CHARS " []()|.*+?!"
312 /* term_100: handle term, where trunc = none(no operators at all) */
313 static int term_100(ZebraMaps zebra_maps, int reg_type,
314 const char **src, char *dst, int space_split,
322 const char *space_start = 0;
323 const char *space_end = 0;
325 if (!term_pre(zebra_maps, reg_type, src, NULL, NULL, !space_split))
332 map = zebra_maps_search(zebra_maps, reg_type, &s0, strlen(s0),
336 if (**map == *CHR_SPACE)
339 else /* complete subfield only. */
341 if (**map == *CHR_SPACE)
342 { /* save space mapping for later .. */
347 else if (space_start)
348 { /* reload last space */
349 while (space_start < space_end)
351 if (strchr(REGEX_CHARS, *space_start))
353 dst_term[j++] = *space_start;
354 dst[i++] = *space_start++;
357 space_start = space_end = 0;
360 /* add non-space char */
361 memcpy(dst_term+j, s1, s0 - s1);
367 if (strchr(REGEX_CHARS, *s1))
375 esc_str(tmpbuf, sizeof(tmpbuf), map[0], strlen(map[0]));
377 strcpy(dst + i, map[0]);
387 /* term_101: handle term, where trunc = Process # */
388 static int term_101(ZebraMaps zebra_maps, int reg_type,
389 const char **src, char *dst, int space_split,
397 if (!term_pre(zebra_maps, reg_type, src, "#", "#", !space_split))
406 dst_term[j++] = *s0++;
412 map = zebra_maps_search(zebra_maps, reg_type, &s0, strlen(s0),
414 if (space_split && **map == *CHR_SPACE)
417 /* add non-space char */
418 memcpy(dst_term+j, s1, s0 - s1);
424 if (strchr(REGEX_CHARS, *s1))
432 esc_str(tmpbuf, sizeof(tmpbuf), map[0], strlen(map[0]));
434 strcpy(dst + i, map[0]);
440 dst_term[j++] = '\0';
445 /* term_103: handle term, where trunc = re-2 (regular expressions) */
446 static int term_103(ZebraMaps zebra_maps, int reg_type, const char **src,
447 char *dst, int *errors, int space_split,
455 if (!term_pre(zebra_maps, reg_type, src, "^\\()[].*+?|", "(", !space_split))
458 if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
459 isdigit(((const unsigned char *)s0)[1]))
461 *errors = s0[1] - '0';
468 if (strchr("^\\()[].*+?|-", *s0))
477 map = zebra_maps_search(zebra_maps, reg_type, &s0, strlen(s0),
479 if (space_split && **map == *CHR_SPACE)
482 /* add non-space char */
483 memcpy(dst_term+j, s1, s0 - s1);
489 if (strchr(REGEX_CHARS, *s1))
497 esc_str(tmpbuf, sizeof(tmpbuf), map[0], strlen(map[0]));
499 strcpy(dst + i, map[0]);
511 /* term_103: handle term, where trunc = re-1 (regular expressions) */
512 static int term_102(ZebraMaps zebra_maps, int reg_type, const char **src,
513 char *dst, int space_split, char *dst_term)
515 return term_103(zebra_maps, reg_type, src, dst, NULL, space_split,
520 /* term_104: handle term, where trunc = Process # and ! */
521 static int term_104(ZebraMaps zebra_maps, int reg_type,
522 const char **src, char *dst, int space_split,
530 if (!term_pre(zebra_maps, reg_type, src, "?*#", "?*#", !space_split))
537 dst_term[j++] = *s0++;
538 if (*s0 >= '0' && *s0 <= '9')
541 while (*s0 >= '0' && *s0 <= '9')
543 limit = limit * 10 + (*s0 - '0');
544 dst_term[j++] = *s0++;
564 dst_term[j++] = *s0++;
569 dst_term[j++] = *s0++;
574 map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0);
575 if (space_split && **map == *CHR_SPACE)
579 if (strchr(REGEX_CHARS, *s1))
587 dst_term[j++] = '\0';
592 /* term_105/106: handle term, where trunc = Process * and ! and right trunc */
593 static int term_105(ZebraMaps zebra_maps, int reg_type,
594 const char **src, char *dst, int space_split,
595 char *dst_term, int right_truncate)
602 if (!term_pre(zebra_maps, reg_type, src, "*!", "*!", !space_split))
611 dst_term[j++] = *s0++;
616 dst_term[j++] = *s0++;
621 map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0);
622 if (space_split && **map == *CHR_SPACE)
626 if (strchr(REGEX_CHARS, *s1))
640 dst_term[j++] = '\0';
646 /* gen_regular_rel - generate regular expression from relation
647 * val: border value (inclusive)
648 * islt: 1 if <=; 0 if >=.
650 static void gen_regular_rel(char *dst, int val, int islt)
657 yaz_log(YLOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt);
661 strcpy(dst, "(-[0-9]+|(");
669 strcpy(dst, "([0-9]+|-(");
681 sprintf(numstr, "%d", val);
682 for (w = strlen(numstr); --w >= 0; pos++)
701 strcpy(dst + dst_p, numstr);
702 dst_p = strlen(dst) - pos - 1;
730 for (i = 0; i<pos; i++)
743 /* match everything less than 10^(pos-1) */
745 for (i = 1; i<pos; i++)
746 strcat(dst, "[0-9]?");
750 /* match everything greater than 10^pos */
751 for (i = 0; i <= pos; i++)
752 strcat(dst, "[0-9]");
753 strcat(dst, "[0-9]*");
758 void string_rel_add_char(char **term_p, const char *src, int *indx)
760 if (src[*indx] == '\\')
761 *(*term_p)++ = src[(*indx)++];
762 *(*term_p)++ = src[(*indx)++];
766 * > abc ([b-].*|a[c-].*|ab[d-].*|abc.+)
767 * ([^-a].*|a[^-b].*ab[^-c].*|abc.+)
768 * >= abc ([b-].*|a[c-].*|ab[c-].*)
769 * ([^-a].*|a[^-b].*|ab[c-].*)
770 * < abc ([-0].*|a[-a].*|ab[-b].*)
771 * ([^a-].*|a[^b-].*|ab[^c-].*)
772 * <= abc ([-0].*|a[-a].*|ab[-b].*|abc)
773 * ([^a-].*|a[^b-].*|ab[^c-].*|abc)
775 static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
776 const char **term_sub, char *term_dict,
777 oid_value attributeSet,
778 int reg_type, int space_split, char *term_dst,
784 char *term_tmp = term_dict + strlen(term_dict);
785 char term_component[2*IT_MAX_WORD+20];
787 attr_init(&relation, zapt, 2);
788 relation_value = attr_find(&relation, NULL);
791 yaz_log(YLOG_DEBUG, "string relation value=%d", relation_value);
792 switch (relation_value)
795 if (!term_100(zh->reg->zebra_maps, reg_type,
796 term_sub, term_component,
797 space_split, term_dst))
799 yaz_log(log_level_rpn, "Relation <");
802 for (i = 0; term_component[i]; )
809 string_rel_add_char(&term_tmp, term_component, &j);
814 string_rel_add_char(&term_tmp, term_component, &i);
821 if ((term_tmp - term_dict) > IT_MAX_WORD)
828 if (!term_100(zh->reg->zebra_maps, reg_type,
829 term_sub, term_component,
830 space_split, term_dst))
832 yaz_log(log_level_rpn, "Relation <=");
835 for (i = 0; term_component[i]; )
840 string_rel_add_char(&term_tmp, term_component, &j);
844 string_rel_add_char(&term_tmp, term_component, &i);
853 if ((term_tmp - term_dict) > IT_MAX_WORD)
856 for (i = 0; term_component[i]; )
857 string_rel_add_char(&term_tmp, term_component, &i);
862 if (!term_100 (zh->reg->zebra_maps, reg_type,
863 term_sub, term_component, space_split, term_dst))
865 yaz_log(log_level_rpn, "Relation >");
868 for (i = 0; term_component[i];)
873 string_rel_add_char(&term_tmp, term_component, &j);
878 string_rel_add_char(&term_tmp, term_component, &i);
886 if ((term_tmp - term_dict) > IT_MAX_WORD)
889 for (i = 0; term_component[i];)
890 string_rel_add_char(&term_tmp, term_component, &i);
897 if (!term_100(zh->reg->zebra_maps, reg_type, term_sub,
898 term_component, space_split, term_dst))
900 yaz_log(log_level_rpn, "Relation >=");
903 for (i = 0; term_component[i];)
910 string_rel_add_char(&term_tmp, term_component, &j);
913 if (term_component[i+1])
917 string_rel_add_char(&term_tmp, term_component, &i);
921 string_rel_add_char(&term_tmp, term_component, &i);
928 if ((term_tmp - term_dict) > IT_MAX_WORD)
937 yaz_log(log_level_rpn, "Relation =");
938 if (!term_100(zh->reg->zebra_maps, reg_type, term_sub,
939 term_component, space_split, term_dst))
941 strcat(term_tmp, "(");
942 strcat(term_tmp, term_component);
943 strcat(term_tmp, ")");
952 static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
953 const char **term_sub,
954 oid_value attributeSet, NMEM stream,
955 struct grep_info *grep_info,
956 int reg_type, int complete_flag,
957 int num_bases, char **basenames,
958 char *term_dst, int xpath_use);
960 static ZEBRA_RES term_trunc(ZebraHandle zh,
961 Z_AttributesPlusTerm *zapt,
962 const char **term_sub,
963 oid_value attributeSet, NMEM stream,
964 struct grep_info *grep_info,
965 int reg_type, int complete_flag,
966 int num_bases, char **basenames,
968 const char *rank_type, int xpath_use,
971 struct rset_key_control *kc)
975 grep_info->isam_p_indx = 0;
976 res = string_term(zh, zapt, term_sub, attributeSet, stream, grep_info,
977 reg_type, complete_flag, num_bases, basenames,
978 term_dst, xpath_use);
981 if (!*term_sub) /* no more terms ? */
983 yaz_log(log_level_rpn, "term: %s", term_dst);
984 *rset = rset_trunc(zh, grep_info->isam_p_buf,
985 grep_info->isam_p_indx, term_dst,
986 strlen(term_dst), rank_type, 1 /* preserve pos */,
987 zapt->term->which, rset_nmem,
994 static char *nmem_strdup_i(NMEM nmem, int v)
997 sprintf(val_str, "%d", v);
998 return nmem_strdup(nmem, val_str);
1001 static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1002 const char **term_sub,
1003 oid_value attributeSet, NMEM stream,
1004 struct grep_info *grep_info,
1005 int reg_type, int complete_flag,
1006 int num_bases, char **basenames,
1007 char *term_dst, int xpath_use)
1009 char term_dict[2*IT_MAX_WORD+4000];
1011 AttrType truncation;
1012 int truncation_value;
1015 const char *use_string = 0;
1016 oid_value curAttributeSet = attributeSet;
1018 struct rpn_char_map_info rcmi;
1019 int space_split = complete_flag ? 0 : 1;
1021 int bases_ok = 0; /* no of databases with OK attribute */
1022 int errCode = 0; /* err code (if any is not OK) */
1023 char *errString = 0; /* addinfo */
1025 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
1026 attr_init(&use, zapt, 1);
1027 use_value = attr_find_ex(&use, &curAttributeSet, &use_string);
1028 yaz_log(log_level_rpn, "string_term, use value %d", use_value);
1029 attr_init(&truncation, zapt, 5);
1030 truncation_value = attr_find(&truncation, NULL);
1031 yaz_log(log_level_rpn, "truncation value %d", truncation_value);
1033 if (use_value == -1) /* no attribute - assumy "any" */
1035 for (base_no = 0; base_no < num_bases; base_no++)
1039 int regex_range = 0;
1042 data1_local_attribute id_xpath_attr;
1043 data1_local_attribute *local_attr;
1044 int max_pos, prefix_len = 0;
1049 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
1051 zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE;
1052 zh->errString = basenames[base_no];
1055 if (xpath_use > 0 && use_value == -2)
1057 /* xpath mode and we have a string attribute */
1058 attp.local_attributes = &id_xpath_attr;
1059 attp.attset_ordinal = VAL_IDXPATH;
1060 id_xpath_attr.next = 0;
1062 use_value = xpath_use; /* xpath_use as use-attribute now */
1063 id_xpath_attr.local = use_value;
1065 else if (curAttributeSet == VAL_IDXPATH && use_value >= 0)
1067 /* X-Path attribute, use numeric value directly */
1068 attp.local_attributes = &id_xpath_attr;
1069 attp.attset_ordinal = VAL_IDXPATH;
1070 id_xpath_attr.next = 0;
1071 id_xpath_attr.local = use_value;
1073 else if (use_string &&
1074 (ord = zebraExplain_lookup_attr_str(zh->reg->zei,
1077 /* we have a match for a raw string attribute */
1082 term_dict[prefix_len++] = '|';
1084 term_dict[prefix_len++] = '(';
1086 ord_len = key_SU_encode (ord, ord_buf);
1087 for (i = 0; i<ord_len; i++)
1089 term_dict[prefix_len++] = 1;
1090 term_dict[prefix_len++] = ord_buf[i];
1092 attp.local_attributes = 0; /* no more attributes */
1096 /* lookup in the .att files . Allow string as well */
1097 if ((r = att_getentbyatt (zh, &attp, curAttributeSet, use_value,
1100 yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
1101 curAttributeSet, use_value, r);
1104 /* set was found, but value wasn't defined */
1105 errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
1107 errString = nmem_strdup(stream, use_string);
1109 errString = nmem_strdup_i (stream, use_value);
1114 struct oident oident;
1116 oident.proto = PROTO_Z3950;
1117 oident.oclass = CLASS_ATTSET;
1118 oident.value = curAttributeSet;
1119 oid_ent_to_oid (&oident, oid);
1121 errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET;
1122 errString = nmem_strdup(stream, oident.desc);
1127 for (local_attr = attp.local_attributes; local_attr;
1128 local_attr = local_attr->next)
1133 ord = zebraExplain_lookup_attr_su(zh->reg->zei,
1134 attp.attset_ordinal,
1139 term_dict[prefix_len++] = '|';
1141 term_dict[prefix_len++] = '(';
1143 ord_len = key_SU_encode (ord, ord_buf);
1144 for (i = 0; i<ord_len; i++)
1146 term_dict[prefix_len++] = 1;
1147 term_dict[prefix_len++] = ord_buf[i];
1154 term_dict[prefix_len++] = ')';
1155 term_dict[prefix_len++] = 1;
1156 term_dict[prefix_len++] = reg_type;
1157 yaz_log(log_level_rpn, "reg_type = %d", term_dict[prefix_len-1]);
1158 term_dict[prefix_len] = '\0';
1160 switch (truncation_value)
1162 case -1: /* not specified */
1163 case 100: /* do not truncate */
1164 if (!string_relation (zh, zapt, &termp, term_dict,
1166 reg_type, space_split, term_dst,
1171 zh->errCode = relation_error;
1178 case 1: /* right truncation */
1179 term_dict[j++] = '(';
1180 if (!term_100(zh->reg->zebra_maps, reg_type,
1181 &termp, term_dict + j, space_split, term_dst))
1186 strcat(term_dict, ".*)");
1188 case 2: /* keft truncation */
1189 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1190 if (!term_100(zh->reg->zebra_maps, reg_type,
1191 &termp, term_dict + j, space_split, term_dst))
1196 strcat(term_dict, ")");
1198 case 3: /* left&right truncation */
1199 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1200 if (!term_100(zh->reg->zebra_maps, reg_type,
1201 &termp, term_dict + j, space_split, term_dst))
1206 strcat(term_dict, ".*)");
1208 case 101: /* process # in term */
1209 term_dict[j++] = '(';
1210 if (!term_101(zh->reg->zebra_maps, reg_type,
1211 &termp, term_dict + j, space_split, term_dst))
1216 strcat(term_dict, ")");
1218 case 102: /* Regexp-1 */
1219 term_dict[j++] = '(';
1220 if (!term_102(zh->reg->zebra_maps, reg_type,
1221 &termp, term_dict + j, space_split, term_dst))
1226 strcat(term_dict, ")");
1228 case 103: /* Regexp-2 */
1230 term_dict[j++] = '(';
1232 if (!term_103(zh->reg->zebra_maps, reg_type,
1233 &termp, term_dict + j, ®ex_range,
1234 space_split, term_dst))
1239 strcat(term_dict, ")");
1241 case 104: /* process # and ! in term */
1242 term_dict[j++] = '(';
1243 if (!term_104(zh->reg->zebra_maps, reg_type,
1244 &termp, term_dict + j, space_split, term_dst))
1249 strcat(term_dict, ")");
1251 case 105: /* process * and ! in term */
1252 term_dict[j++] = '(';
1253 if (!term_105(zh->reg->zebra_maps, reg_type,
1254 &termp, term_dict + j, space_split, term_dst, 1))
1259 strcat(term_dict, ")");
1261 case 106: /* process * and ! in term */
1262 term_dict[j++] = '(';
1263 if (!term_105(zh->reg->zebra_maps, reg_type,
1264 &termp, term_dict + j, space_split, term_dst, 0))
1269 strcat(term_dict, ")");
1272 zh->errCode = YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE;
1273 zh->errString = nmem_strdup_i(stream, truncation_value);
1279 const char *input = term_dict + prefix_len;
1280 esc_str(buf, sizeof(buf), input, strlen(input));
1284 yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_dict+prefix_len);
1285 r = dict_lookup_grep(zh->reg->dict, term_dict, regex_range,
1286 grep_info, &max_pos, init_pos,
1289 yaz_log(YLOG_WARN, "dict_lookup_grep fail %d", r);
1294 zh->errCode = errCode;
1295 zh->errString = errString;
1299 yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1304 /* convert APT search term to UTF8 */
1305 static ZEBRA_RES zapt_term_to_utf8(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1309 Z_Term *term = zapt->term;
1311 switch (term->which)
1313 case Z_Term_general:
1314 if (zh->iconv_to_utf8 != 0)
1316 char *inbuf = term->u.general->buf;
1317 size_t inleft = term->u.general->len;
1318 char *outbuf = termz;
1319 size_t outleft = IT_MAX_WORD-1;
1322 ret = yaz_iconv(zh->iconv_to_utf8, &inbuf, &inleft,
1324 if (ret == (size_t)(-1))
1326 ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0);
1328 YAZ_BIB1_QUERY_TERM_INCLUDES_CHARS_THAT_DO_NOT_TRANSLATE_INTO_;
1335 sizez = term->u.general->len;
1336 if (sizez > IT_MAX_WORD-1)
1337 sizez = IT_MAX_WORD-1;
1338 memcpy (termz, term->u.general->buf, sizez);
1339 termz[sizez] = '\0';
1342 case Z_Term_characterString:
1343 sizez = strlen(term->u.characterString);
1344 if (sizez > IT_MAX_WORD-1)
1345 sizez = IT_MAX_WORD-1;
1346 memcpy (termz, term->u.characterString, sizez);
1347 termz[sizez] = '\0';
1350 zh->errCode = YAZ_BIB1_UNSUPP_CODED_VALUE_FOR_TERM;
1356 /* convert APT SCAN term to internal cmap */
1357 static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1358 char *termz, int reg_type)
1360 char termz0[IT_MAX_WORD];
1362 if (zapt_term_to_utf8(zh, zapt, termz0) == ZEBRA_FAIL)
1363 return ZEBRA_FAIL; /* error */
1367 const char *cp = (const char *) termz0;
1368 const char *cp_end = cp + strlen(cp);
1371 const char *space_map = NULL;
1374 while ((len = (cp_end - cp)) > 0)
1376 map = zebra_maps_input(zh->reg->zebra_maps, reg_type, &cp, len, 0);
1377 if (**map == *CHR_SPACE)
1382 for (src = space_map; *src; src++)
1385 for (src = *map; *src; src++)
1394 char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1395 const char *termz, NMEM stream, unsigned reg_id)
1398 AttrType truncation;
1399 int truncation_value;
1402 attr_init(&truncation, zapt, 5);
1403 truncation_value = attr_find(&truncation, NULL);
1405 switch (truncation_value)
1425 wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list,
1426 termz, strlen(termz));
1428 return nmem_strdup(stream, termz);
1431 char *buf = (char*) nmem_malloc(stream, wrbuf_len(wrbuf)+1);
1432 memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
1433 buf[wrbuf_len(wrbuf)] = '\0';
1438 static void grep_info_delete(struct grep_info *grep_info)
1441 xfree(grep_info->term_no);
1443 xfree(grep_info->isam_p_buf);
1446 static int grep_info_prepare(ZebraHandle zh,
1447 Z_AttributesPlusTerm *zapt,
1448 struct grep_info *grep_info,
1453 int termset_value_numeric;
1454 const char *termset_value_string;
1457 grep_info->term_no = 0;
1459 grep_info->isam_p_size = 0;
1460 grep_info->isam_p_buf = NULL;
1462 grep_info->reg_type = reg_type;
1463 grep_info->termset = 0;
1467 attr_init(&termset, zapt, 8);
1468 termset_value_numeric =
1469 attr_find_ex(&termset, NULL, &termset_value_string);
1470 if (termset_value_numeric != -1)
1473 const char *termset_name = 0;
1474 if (termset_value_numeric != -2)
1477 sprintf(resname, "%d", termset_value_numeric);
1478 termset_name = resname;
1481 termset_name = termset_value_string;
1482 yaz_log(log_level_rpn, "creating termset set %s", termset_name);
1483 grep_info->termset = resultSetAdd(zh, termset_name, 1);
1484 if (!grep_info->termset)
1486 zh->errCode = YAZ_BIB1_ILLEGAL_RESULT_SET_NAME;
1487 zh->errString = nmem_strdup(stream, termset_name);
1495 static ZEBRA_RES term_list_trunc(ZebraHandle zh,
1496 Z_AttributesPlusTerm *zapt,
1497 const char *termz_org,
1498 oid_value attributeSet,
1500 int reg_type, int complete_flag,
1501 const char *rank_type, int xpath_use,
1502 int num_bases, char **basenames,
1504 RSET **result_sets, int *num_result_sets,
1505 struct rset_key_control *kc)
1507 char term_dst[IT_MAX_WORD+1];
1508 struct grep_info grep_info;
1509 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1510 const char *termp = termz;
1513 *num_result_sets = 0;
1515 if (grep_info_prepare(zh, zapt, &grep_info, reg_type, stream))
1521 if (alloc_sets == *num_result_sets)
1524 RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) *
1527 memcpy(rnew, *result_sets, alloc_sets * sizeof(*rnew));
1528 alloc_sets = alloc_sets + add;
1529 *result_sets = rnew;
1531 res = term_trunc(zh, zapt, &termp, attributeSet,
1533 reg_type, complete_flag,
1534 num_bases, basenames,
1535 term_dst, rank_type,
1536 xpath_use, rset_nmem,
1537 &(*result_sets)[*num_result_sets],
1539 if (res != ZEBRA_OK)
1542 for (i = 0; i < *num_result_sets; i++)
1543 rset_delete((*result_sets)[i]);
1544 grep_info_delete (&grep_info);
1547 if ((*result_sets)[*num_result_sets] == 0)
1549 (*num_result_sets)++;
1551 grep_info_delete(&grep_info);
1555 static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh,
1556 Z_AttributesPlusTerm *zapt,
1557 const char *termz_org,
1558 oid_value attributeSet,
1560 int reg_type, int complete_flag,
1561 const char *rank_type, int xpath_use,
1562 int num_bases, char **basenames,
1565 struct rset_key_control *kc)
1567 RSET *result_sets = 0;
1568 int num_result_sets = 0;
1570 term_list_trunc(zh, zapt, termz_org, attributeSet,
1571 stream, reg_type, complete_flag,
1572 rank_type, xpath_use,
1573 num_bases, basenames,
1575 &result_sets, &num_result_sets, kc);
1576 if (res != ZEBRA_OK)
1578 if (num_result_sets == 0)
1579 *rset = rsnull_create (rset_nmem, kc);
1580 else if (num_result_sets == 1)
1581 *rset = result_sets[0];
1583 *rset = rsprox_create(rset_nmem, kc, kc->scope,
1584 num_result_sets, result_sets,
1585 1 /* ordered */, 0 /* exclusion */,
1586 3 /* relation */, 1 /* distance */);
1592 static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh,
1593 Z_AttributesPlusTerm *zapt,
1594 const char *termz_org,
1595 oid_value attributeSet,
1597 int reg_type, int complete_flag,
1598 const char *rank_type,
1600 int num_bases, char **basenames,
1603 struct rset_key_control *kc)
1605 RSET *result_sets = 0;
1606 int num_result_sets = 0;
1608 term_list_trunc(zh, zapt, termz_org, attributeSet,
1609 stream, reg_type, complete_flag,
1610 rank_type, xpath_use,
1611 num_bases, basenames,
1613 &result_sets, &num_result_sets, kc);
1614 if (res != ZEBRA_OK)
1616 if (num_result_sets == 0)
1617 *rset = rsnull_create (rset_nmem, kc);
1618 else if (num_result_sets == 1)
1619 *rset = result_sets[0];
1621 *rset = rsmulti_or_create(rset_nmem, kc, kc->scope,
1622 num_result_sets, result_sets);
1628 static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh,
1629 Z_AttributesPlusTerm *zapt,
1630 const char *termz_org,
1631 oid_value attributeSet,
1633 int reg_type, int complete_flag,
1634 const char *rank_type,
1636 int num_bases, char **basenames,
1639 struct rset_key_control *kc)
1641 RSET *result_sets = 0;
1642 int num_result_sets = 0;
1644 term_list_trunc(zh, zapt, termz_org, attributeSet,
1645 stream, reg_type, complete_flag,
1646 rank_type, xpath_use,
1647 num_bases, basenames,
1649 &result_sets, &num_result_sets,
1651 if (res != ZEBRA_OK)
1653 if (num_result_sets == 0)
1654 *rset = rsnull_create (rset_nmem, kc);
1655 else if (num_result_sets == 1)
1656 *rset = result_sets[0];
1658 *rset = rsmulti_and_create(rset_nmem, kc, kc->scope,
1659 num_result_sets, result_sets);
1665 static int numeric_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1666 const char **term_sub,
1668 oid_value attributeSet,
1669 struct grep_info *grep_info,
1679 char *term_tmp = term_dict + strlen(term_dict);
1682 attr_init(&relation, zapt, 2);
1683 relation_value = attr_find(&relation, NULL);
1685 yaz_log(log_level_rpn, "numeric relation value=%d", relation_value);
1687 if (!term_100(zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1,
1690 term_value = atoi (term_tmp);
1691 switch (relation_value)
1694 yaz_log(log_level_rpn, "Relation <");
1695 gen_regular_rel(term_tmp, term_value-1, 1);
1698 yaz_log(log_level_rpn, "Relation <=");
1699 gen_regular_rel(term_tmp, term_value, 1);
1702 yaz_log(log_level_rpn, "Relation >=");
1703 gen_regular_rel(term_tmp, term_value, 0);
1706 yaz_log(log_level_rpn, "Relation >");
1707 gen_regular_rel(term_tmp, term_value+1, 0);
1711 yaz_log(log_level_rpn, "Relation =");
1712 sprintf(term_tmp, "(0*%d)", term_value);
1718 yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_tmp);
1719 r = dict_lookup_grep(zh->reg->dict, term_dict, 0, grep_info, max_pos,
1722 yaz_log(YLOG_WARN, "dict_lookup_grep fail, rel = gt: %d", r);
1723 yaz_log(log_level_rpn, "%d positions", grep_info->isam_p_indx);
1727 static ZEBRA_RES numeric_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1728 const char **term_sub,
1729 oid_value attributeSet,
1730 struct grep_info *grep_info,
1731 int reg_type, int complete_flag,
1732 int num_bases, char **basenames,
1733 char *term_dst, int xpath_use, NMEM stream)
1735 char term_dict[2*IT_MAX_WORD+2];
1739 const char *use_string = 0;
1740 oid_value curAttributeSet = attributeSet;
1742 struct rpn_char_map_info rcmi;
1744 int bases_ok = 0; /* no of databases with OK attribute */
1745 int errCode = 0; /* err code (if any is not OK) */
1746 char *errString = 0; /* addinfo */
1748 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
1749 attr_init(&use, zapt, 1);
1750 use_value = attr_find_ex(&use, &curAttributeSet, &use_string);
1752 if (use_value == -1)
1755 for (base_no = 0; base_no < num_bases; base_no++)
1758 data1_local_attribute id_xpath_attr;
1759 data1_local_attribute *local_attr;
1760 int max_pos, prefix_len = 0;
1761 int relation_error = 0;
1764 if (use_value == -2) /* string attribute (assume IDXPATH/any) */
1766 use_value = xpath_use;
1767 attp.local_attributes = &id_xpath_attr;
1768 attp.attset_ordinal = VAL_IDXPATH;
1769 id_xpath_attr.next = 0;
1770 id_xpath_attr.local = use_value;
1772 else if (curAttributeSet == VAL_IDXPATH)
1774 attp.local_attributes = &id_xpath_attr;
1775 attp.attset_ordinal = VAL_IDXPATH;
1776 id_xpath_attr.next = 0;
1777 id_xpath_attr.local = use_value;
1781 if ((r = att_getentbyatt (zh, &attp, curAttributeSet, use_value,
1784 yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
1785 curAttributeSet, use_value, r);
1788 errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
1790 errString = nmem_strdup(stream, use_string);
1792 errString = nmem_strdup_i (stream, use_value);
1795 errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET;
1799 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
1801 zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE;
1802 zh->errString = basenames[base_no];
1805 for (local_attr = attp.local_attributes; local_attr;
1806 local_attr = local_attr->next)
1812 ord = zebraExplain_lookup_attr_su(zh->reg->zei,
1813 attp.attset_ordinal,
1818 term_dict[prefix_len++] = '|';
1820 term_dict[prefix_len++] = '(';
1822 ord_len = key_SU_encode (ord, ord_buf);
1823 for (i = 0; i<ord_len; i++)
1825 term_dict[prefix_len++] = 1;
1826 term_dict[prefix_len++] = ord_buf[i];
1831 errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
1832 errString = nmem_strdup_i(stream, use_value);
1836 term_dict[prefix_len++] = ')';
1837 term_dict[prefix_len++] = 1;
1838 term_dict[prefix_len++] = reg_type;
1839 yaz_log(YLOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
1840 term_dict[prefix_len] = '\0';
1841 if (!numeric_relation(zh, zapt, &termp, term_dict,
1842 attributeSet, grep_info, &max_pos, reg_type,
1843 term_dst, &relation_error))
1847 zh->errCode = relation_error;
1857 zh->errCode = errCode;
1858 zh->errString = errString;
1862 yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1866 static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh,
1867 Z_AttributesPlusTerm *zapt,
1869 oid_value attributeSet,
1871 int reg_type, int complete_flag,
1872 const char *rank_type, int xpath_use,
1873 int num_bases, char **basenames,
1876 struct rset_key_control *kc)
1878 char term_dst[IT_MAX_WORD+1];
1879 const char *termp = termz;
1880 RSET *result_sets = 0;
1881 int num_result_sets = 0;
1883 struct grep_info grep_info;
1886 yaz_log(log_level_rpn, "APT_numeric t='%s'", termz);
1887 if (grep_info_prepare(zh, zapt, &grep_info, reg_type, stream))
1891 if (alloc_sets == num_result_sets)
1894 RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) *
1897 memcpy(rnew, result_sets, alloc_sets * sizeof(*rnew));
1898 alloc_sets = alloc_sets + add;
1901 yaz_log(YLOG_DEBUG, "APT_numeric termp=%s", termp);
1902 grep_info.isam_p_indx = 0;
1903 res = numeric_term(zh, zapt, &termp, attributeSet, &grep_info,
1904 reg_type, complete_flag, num_bases, basenames,
1905 term_dst, xpath_use,
1907 if (res == ZEBRA_FAIL || termp == 0)
1909 yaz_log(YLOG_DEBUG, "term: %s", term_dst);
1910 result_sets[num_result_sets] =
1911 rset_trunc(zh, grep_info.isam_p_buf,
1912 grep_info.isam_p_indx, term_dst,
1913 strlen(term_dst), rank_type,
1914 0 /* preserve position */,
1915 zapt->term->which, rset_nmem,
1917 if (!result_sets[num_result_sets])
1921 grep_info_delete(&grep_info);
1925 for (i = 0; i<num_result_sets; i++)
1926 rset_delete(result_sets[i]);
1929 if (num_result_sets == 0)
1930 *rset = rsnull_create(rset_nmem, kc);
1931 if (num_result_sets == 1)
1932 *rset = result_sets[0];
1934 *rset = rsmulti_and_create(rset_nmem, kc, kc->scope,
1935 num_result_sets, result_sets);
1941 static ZEBRA_RES rpn_search_APT_local(ZebraHandle zh,
1942 Z_AttributesPlusTerm *zapt,
1944 oid_value attributeSet,
1946 const char *rank_type, NMEM rset_nmem,
1948 struct rset_key_control *kc)
1953 *rset = rstemp_create(rset_nmem, kc, kc->scope,
1954 res_get (zh->res, "setTmpDir"),0 );
1955 rsfd = rset_open(*rset, RSETF_WRITE);
1963 rset_write (rsfd, &key);
1968 static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1969 oid_value attributeSet, NMEM stream,
1970 Z_SortKeySpecList *sort_sequence,
1971 const char *rank_type,
1973 struct rset_key_control *kc)
1976 int sort_relation_value;
1977 AttrType sort_relation_type;
1982 Z_AttributeElement *ae;
1987 attr_init(&sort_relation_type, zapt, 7);
1988 sort_relation_value = attr_find(&sort_relation_type, &attributeSet);
1990 attr_init(&use_type, zapt, 1);
1991 use_value = attr_find(&use_type, &attributeSet);
1993 if (!sort_sequence->specs)
1995 sort_sequence->num_specs = 10;
1996 sort_sequence->specs = (Z_SortKeySpec **)
1997 nmem_malloc(stream, sort_sequence->num_specs *
1998 sizeof(*sort_sequence->specs));
1999 for (i = 0; i<sort_sequence->num_specs; i++)
2000 sort_sequence->specs[i] = 0;
2002 if (zapt->term->which != Z_Term_general)
2005 i = atoi_n ((char *) zapt->term->u.general->buf,
2006 zapt->term->u.general->len);
2007 if (i >= sort_sequence->num_specs)
2009 sprintf(termz, "%d", i);
2011 oe.proto = PROTO_Z3950;
2012 oe.oclass = CLASS_ATTSET;
2013 oe.value = attributeSet;
2014 if (!oid_ent_to_oid (&oe, oid))
2017 sks = (Z_SortKeySpec *) nmem_malloc(stream, sizeof(*sks));
2018 sks->sortElement = (Z_SortElement *)
2019 nmem_malloc(stream, sizeof(*sks->sortElement));
2020 sks->sortElement->which = Z_SortElement_generic;
2021 sk = sks->sortElement->u.generic = (Z_SortKey *)
2022 nmem_malloc(stream, sizeof(*sk));
2023 sk->which = Z_SortKey_sortAttributes;
2024 sk->u.sortAttributes = (Z_SortAttributes *)
2025 nmem_malloc(stream, sizeof(*sk->u.sortAttributes));
2027 sk->u.sortAttributes->id = oid;
2028 sk->u.sortAttributes->list = (Z_AttributeList *)
2029 nmem_malloc(stream, sizeof(*sk->u.sortAttributes->list));
2030 sk->u.sortAttributes->list->num_attributes = 1;
2031 sk->u.sortAttributes->list->attributes = (Z_AttributeElement **)
2032 nmem_malloc(stream, sizeof(*sk->u.sortAttributes->list->attributes));
2033 ae = *sk->u.sortAttributes->list->attributes = (Z_AttributeElement *)
2034 nmem_malloc(stream, sizeof(**sk->u.sortAttributes->list->attributes));
2035 ae->attributeSet = 0;
2036 ae->attributeType = (int *)
2037 nmem_malloc(stream, sizeof(*ae->attributeType));
2038 *ae->attributeType = 1;
2039 ae->which = Z_AttributeValue_numeric;
2040 ae->value.numeric = (int *)
2041 nmem_malloc(stream, sizeof(*ae->value.numeric));
2042 *ae->value.numeric = use_value;
2044 sks->sortRelation = (int *)
2045 nmem_malloc(stream, sizeof(*sks->sortRelation));
2046 if (sort_relation_value == 1)
2047 *sks->sortRelation = Z_SortKeySpec_ascending;
2048 else if (sort_relation_value == 2)
2049 *sks->sortRelation = Z_SortKeySpec_descending;
2051 *sks->sortRelation = Z_SortKeySpec_ascending;
2053 sks->caseSensitivity = (int *)
2054 nmem_malloc(stream, sizeof(*sks->caseSensitivity));
2055 *sks->caseSensitivity = 0;
2057 sks->which = Z_SortKeySpec_null;
2058 sks->u.null = odr_nullval ();
2059 sort_sequence->specs[i] = sks;
2060 *rset = rsnull_create (NULL, kc);
2065 static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2066 oid_value attributeSet,
2067 struct xpath_location_step *xpath, int max, NMEM mem)
2069 oid_value curAttributeSet = attributeSet;
2071 const char *use_string = 0;
2073 attr_init(&use, zapt, 1);
2074 attr_find_ex(&use, &curAttributeSet, &use_string);
2076 if (!use_string || *use_string != '/')
2079 return zebra_parse_xpath_str(use_string, xpath, max, mem);
2084 static RSET xpath_trunc(ZebraHandle zh, NMEM stream,
2085 int reg_type, const char *term, int use,
2086 oid_value curAttributeSet, NMEM rset_nmem,
2087 struct rset_key_control *kc)
2090 struct grep_info grep_info;
2091 char term_dict[2048];
2094 int ord = zebraExplain_lookup_attr_su(zh->reg->zei, curAttributeSet, use);
2095 int ord_len, i, r, max_pos;
2096 int term_type = Z_Term_characterString;
2097 const char *flags = "void";
2099 if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, '0', stream))
2100 return rsnull_create(rset_nmem, kc);
2103 return rsnull_create(rset_nmem, kc);
2105 term_dict[prefix_len++] = '|';
2107 term_dict[prefix_len++] = '(';
2109 ord_len = key_SU_encode (ord, ord_buf);
2110 for (i = 0; i<ord_len; i++)
2112 term_dict[prefix_len++] = 1;
2113 term_dict[prefix_len++] = ord_buf[i];
2115 term_dict[prefix_len++] = ')';
2116 term_dict[prefix_len++] = 1;
2117 term_dict[prefix_len++] = reg_type;
2119 strcpy(term_dict+prefix_len, term);
2121 grep_info.isam_p_indx = 0;
2122 r = dict_lookup_grep(zh->reg->dict, term_dict, 0,
2123 &grep_info, &max_pos, 0, grep_handle);
2124 yaz_log(YLOG_DEBUG, "%s %d positions", term,
2125 grep_info.isam_p_indx);
2126 rset = rset_trunc(zh, grep_info.isam_p_buf,
2127 grep_info.isam_p_indx, term, strlen(term),
2128 flags, 1, term_type,rset_nmem,
2130 grep_info_delete(&grep_info);
2134 static RSET rpn_search_xpath(ZebraHandle zh,
2135 oid_value attributeSet,
2136 int num_bases, char **basenames,
2137 NMEM stream, const char *rank_type, RSET rset,
2138 int xpath_len, struct xpath_location_step *xpath,
2140 struct rset_key_control *kc)
2142 oid_value curAttributeSet = attributeSet;
2149 yaz_log(YLOG_DEBUG, "xpath len=%d", xpath_len);
2150 for (i = 0; i<xpath_len; i++)
2152 yaz_log(log_level_rpn, "XPATH %d %s", i, xpath[i].part);
2156 curAttributeSet = VAL_IDXPATH;
2166 a[@attr = value]/b[@other = othervalue]
2168 /e/@a val range(e/,range(@a,freetext(w,1015,val),@a),e/)
2169 /a/b val range(b/a/,freetext(w,1016,val),b/a/)
2170 /a/b/@c val range(b/a/,range(@c,freetext(w,1016,val),@c),b/a/)
2171 /a/b[@c = y] val range(b/a/,freetext(w,1016,val),b/a/,@c = y)
2172 /a[@c = y]/b val range(a/,range(b/a/,freetext(w,1016,val),b/a/),a/,@c = y)
2173 /a[@c = x]/b[@c = y] range(a/,range(b/a/,freetext(w,1016,val),b/a/,@c = y),a/,@c = x)
2177 dict_grep_cmap (zh->reg->dict, 0, 0);
2179 for (base_no = 0; base_no < num_bases; base_no++)
2181 int level = xpath_len;
2184 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2186 zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE;
2187 zh->errString = basenames[base_no];
2190 while (--level >= 0)
2192 char xpath_rev[128];
2194 RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0;
2198 for (i = level; i >= 1; --i)
2200 const char *cp = xpath[i].part;
2206 memcpy (xpath_rev + len, "[^/]*", 5);
2209 else if (*cp == ' ')
2212 xpath_rev[len++] = 1;
2213 xpath_rev[len++] = ' ';
2217 xpath_rev[len++] = *cp;
2218 xpath_rev[len++] = '/';
2220 else if (i == 1) /* // case */
2222 xpath_rev[len++] = '.';
2223 xpath_rev[len++] = '*';
2228 if (xpath[level].predicate &&
2229 xpath[level].predicate->which == XPATH_PREDICATE_RELATION &&
2230 xpath[level].predicate->u.relation.name[0])
2232 WRBUF wbuf = wrbuf_alloc();
2233 wrbuf_puts(wbuf, xpath[level].predicate->u.relation.name+1);
2234 if (xpath[level].predicate->u.relation.value)
2236 const char *cp = xpath[level].predicate->u.relation.value;
2237 wrbuf_putc(wbuf, '=');
2241 if (strchr(REGEX_CHARS, *cp))
2242 wrbuf_putc(wbuf, '\\');
2243 wrbuf_putc(wbuf, *cp);
2247 wrbuf_puts(wbuf, "");
2248 rset_attr = xpath_trunc(
2249 zh, stream, '0', wrbuf_buf(wbuf), 3,
2250 curAttributeSet, rset_nmem, kc);
2251 wrbuf_free(wbuf, 1);
2258 yaz_log(log_level_rpn, "xpath_rev (%d) = %s", level, xpath_rev);
2259 if (strlen(xpath_rev))
2261 rset_start_tag = xpath_trunc(zh, stream, '0',
2262 xpath_rev, 1, curAttributeSet, rset_nmem, kc);
2264 rset_end_tag = xpath_trunc(zh, stream, '0',
2265 xpath_rev, 2, curAttributeSet, rset_nmem, kc);
2267 rset = rsbetween_create(rset_nmem, kc, kc->scope,
2268 rset_start_tag, rset,
2269 rset_end_tag, rset_attr);
2278 static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2279 oid_value attributeSet, NMEM stream,
2280 Z_SortKeySpecList *sort_sequence,
2281 int num_bases, char **basenames,
2284 struct rset_key_control *kc)
2286 ZEBRA_RES res = ZEBRA_OK;
2288 char *search_type = NULL;
2289 char rank_type[128];
2292 char termz[IT_MAX_WORD+1];
2295 struct xpath_location_step xpath[10];
2299 log_level_rpn = yaz_log_module_level("rpn");
2302 zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type,
2303 rank_type, &complete_flag, &sort_flag);
2305 yaz_log(YLOG_DEBUG, "reg_id=%c", reg_id);
2306 yaz_log(YLOG_DEBUG, "complete_flag=%d", complete_flag);
2307 yaz_log(YLOG_DEBUG, "search_type=%s", search_type);
2308 yaz_log(YLOG_DEBUG, "rank_type=%s", rank_type);
2310 if (zapt_term_to_utf8(zh, zapt, termz) == ZEBRA_FAIL)
2314 return rpn_sort_spec(zh, zapt, attributeSet, stream, sort_sequence,
2315 rank_type, rset, kc);
2316 xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream);
2320 if (xpath[xpath_len-1].part[0] == '@')
2324 if (!strcmp(search_type, "phrase"))
2326 res = rpn_search_APT_phrase(zh, zapt, termz, attributeSet, stream,
2327 reg_id, complete_flag, rank_type,
2329 num_bases, basenames, rset_nmem,
2332 else if (!strcmp(search_type, "and-list"))
2334 res = rpn_search_APT_and_list(zh, zapt, termz, attributeSet, stream,
2335 reg_id, complete_flag, rank_type,
2337 num_bases, basenames, rset_nmem,
2340 else if (!strcmp(search_type, "or-list"))
2342 res = rpn_search_APT_or_list(zh, zapt, termz, attributeSet, stream,
2343 reg_id, complete_flag, rank_type,
2345 num_bases, basenames, rset_nmem,
2348 else if (!strcmp(search_type, "local"))
2350 res = rpn_search_APT_local(zh, zapt, termz, attributeSet, stream,
2351 rank_type, rset_nmem, rset, kc);
2353 else if (!strcmp(search_type, "numeric"))
2355 res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, stream,
2356 reg_id, complete_flag, rank_type,
2358 num_bases, basenames, rset_nmem,
2363 zh->errCode = YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE;
2366 if (res != ZEBRA_OK)
2370 *rset = rpn_search_xpath(zh, attributeSet, num_bases, basenames,
2371 stream, rank_type, *rset,
2372 xpath_len, xpath, rset_nmem, kc);
2378 static ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs,
2379 oid_value attributeSet,
2380 NMEM stream, NMEM rset_nmem,
2381 Z_SortKeySpecList *sort_sequence,
2382 int num_bases, char **basenames,
2383 RSET **result_sets, int *num_result_sets,
2384 Z_Operator *parent_op,
2385 struct rset_key_control *kc);
2387 ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs,
2388 oid_value attributeSet,
2389 NMEM stream, NMEM rset_nmem,
2390 Z_SortKeySpecList *sort_sequence,
2391 int num_bases, char **basenames,
2394 RSET *result_sets = 0;
2395 int num_result_sets = 0;
2397 struct rset_key_control *kc = zebra_key_control_create(zh);
2399 res = rpn_search_structure(zh, zs, attributeSet,
2402 num_bases, basenames,
2403 &result_sets, &num_result_sets,
2404 0 /* no parent op */,
2406 if (res != ZEBRA_OK)
2409 for (i = 0; i<num_result_sets; i++)
2410 rset_delete(result_sets[i]);
2414 assert(num_result_sets == 1);
2415 assert(result_sets);
2416 assert(*result_sets);
2417 *result_set = *result_sets;
2423 ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs,
2424 oid_value attributeSet,
2425 NMEM stream, NMEM rset_nmem,
2426 Z_SortKeySpecList *sort_sequence,
2427 int num_bases, char **basenames,
2428 RSET **result_sets, int *num_result_sets,
2429 Z_Operator *parent_op,
2430 struct rset_key_control *kc)
2432 *num_result_sets = 0;
2433 if (zs->which == Z_RPNStructure_complex)
2436 Z_Operator *zop = zs->u.complex->roperator;
2437 RSET *result_sets_l = 0;
2438 int num_result_sets_l = 0;
2439 RSET *result_sets_r = 0;
2440 int num_result_sets_r = 0;
2442 res = rpn_search_structure(zh, zs->u.complex->s1,
2443 attributeSet, stream, rset_nmem,
2445 num_bases, basenames,
2446 &result_sets_l, &num_result_sets_l,
2448 if (res != ZEBRA_OK)
2451 for (i = 0; i<num_result_sets_l; i++)
2452 rset_delete(result_sets_l[i]);
2455 res = rpn_search_structure(zh, zs->u.complex->s2,
2456 attributeSet, stream, rset_nmem,
2458 num_bases, basenames,
2459 &result_sets_r, &num_result_sets_r,
2461 if (res != ZEBRA_OK)
2464 for (i = 0; i<num_result_sets_l; i++)
2465 rset_delete(result_sets_l[i]);
2466 for (i = 0; i<num_result_sets_r; i++)
2467 rset_delete(result_sets_r[i]);
2471 /* make a new list of result for all children */
2472 *num_result_sets = num_result_sets_l + num_result_sets_r;
2473 *result_sets = nmem_malloc(stream, *num_result_sets *
2474 sizeof(**result_sets));
2475 memcpy(*result_sets, result_sets_l,
2476 num_result_sets_l * sizeof(**result_sets));
2477 memcpy(*result_sets + num_result_sets_l, result_sets_r,
2478 num_result_sets_r * sizeof(**result_sets));
2480 if (!parent_op || parent_op->which != zop->which
2481 || (zop->which != Z_Operator_and &&
2482 zop->which != Z_Operator_or))
2484 /* parent node different from this one (or non-present) */
2485 /* we must combine result sets now */
2489 case Z_Operator_and:
2490 rset = rsmulti_and_create(rset_nmem, kc,
2492 *num_result_sets, *result_sets);
2495 rset = rsmulti_or_create(rset_nmem, kc,
2497 *num_result_sets, *result_sets);
2499 case Z_Operator_and_not:
2500 rset = rsbool_create_not(rset_nmem, kc,
2505 case Z_Operator_prox:
2506 if (zop->u.prox->which != Z_ProximityOperator_known)
2508 zh->errCode = YAZ_BIB1_UNSUPP_PROX_UNIT_CODE;
2511 if (*zop->u.prox->u.known != Z_ProxUnit_word)
2513 char *val = (char *) nmem_malloc(stream, 16);
2514 zh->errCode = YAZ_BIB1_UNSUPP_PROX_UNIT_CODE;
2515 zh->errString = val;
2516 sprintf(val, "%d", *zop->u.prox->u.known);
2521 rset = rsprox_create(rset_nmem, kc,
2523 *num_result_sets, *result_sets,
2524 *zop->u.prox->ordered,
2525 (!zop->u.prox->exclusion ?
2526 0 : *zop->u.prox->exclusion),
2527 *zop->u.prox->relationType,
2528 *zop->u.prox->distance );
2532 zh->errCode = YAZ_BIB1_OPERATOR_UNSUPP;
2535 *num_result_sets = 1;
2536 *result_sets = nmem_malloc(stream, *num_result_sets *
2537 sizeof(**result_sets));
2538 (*result_sets)[0] = rset;
2541 else if (zs->which == Z_RPNStructure_simple)
2546 if (zs->u.simple->which == Z_Operand_APT)
2548 yaz_log(YLOG_DEBUG, "rpn_search_APT");
2549 res = rpn_search_APT(zh, zs->u.simple->u.attributesPlusTerm,
2550 attributeSet, stream, sort_sequence,
2551 num_bases, basenames, rset_nmem, &rset,
2553 if (res != ZEBRA_OK)
2556 else if (zs->u.simple->which == Z_Operand_resultSetId)
2558 yaz_log(YLOG_DEBUG, "rpn_search_ref");
2559 rset = resultSetRef(zh, zs->u.simple->u.resultSetId);
2562 zh->errCode = YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST;
2564 nmem_strdup(stream, zs->u.simple->u.resultSetId);
2571 zh->errCode = YAZ_BIB1_UNSUPP_SEARCH;
2574 *num_result_sets = 1;
2575 *result_sets = nmem_malloc(stream, *num_result_sets *
2576 sizeof(**result_sets));
2577 (*result_sets)[0] = rset;
2581 zh->errCode = YAZ_BIB1_UNSUPP_SEARCH;
2587 struct scan_info_entry {
2593 struct scan_info_entry *list;
2599 static int scan_handle (char *name, const char *info, int pos, void *client)
2601 int len_prefix, idx;
2602 struct scan_info *scan_info = (struct scan_info *) client;
2604 len_prefix = strlen(scan_info->prefix);
2605 if (memcmp (name, scan_info->prefix, len_prefix))
2608 idx = scan_info->after - pos + scan_info->before;
2614 scan_info->list[idx].term = (char *)
2615 odr_malloc(scan_info->odr, strlen(name + len_prefix)+1);
2616 strcpy(scan_info->list[idx].term, name + len_prefix);
2617 assert (*info == sizeof(ISAM_P));
2618 memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
2622 static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type,
2623 char **dst, const char *src)
2625 char term_src[IT_MAX_WORD];
2626 char term_dst[IT_MAX_WORD];
2628 term_untrans (zh, reg_type, term_src, src);
2630 if (zh->iconv_from_utf8 != 0)
2633 char *inbuf = term_src;
2634 size_t inleft = strlen(term_src);
2635 char *outbuf = term_dst;
2636 size_t outleft = sizeof(term_dst)-1;
2639 ret = yaz_iconv (zh->iconv_from_utf8, &inbuf, &inleft,
2641 if (ret == (size_t)(-1))
2644 len = outbuf - term_dst;
2645 *dst = nmem_malloc(stream, len + 1);
2647 memcpy (*dst, term_dst, len);
2651 *dst = nmem_strdup(stream, term_src);
2654 static void count_set (RSET r, int *count)
2661 yaz_log(YLOG_DEBUG, "count_set");
2664 rfd = rset_open (r, RSETF_READ);
2665 while (rset_read (rfd, &key,0 /* never mind terms */))
2667 if (key.mem[0] != psysno)
2669 psysno = key.mem[0];
2675 yaz_log(YLOG_DEBUG, "%d keys, %d records", kno, *count);
2678 ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
2679 oid_value attributeset,
2680 int num_bases, char **basenames,
2681 int *position, int *num_entries, ZebraScanEntry **list,
2682 int *is_partial, RSET limit_set, int return_zero)
2685 int pos = *position;
2686 int num = *num_entries;
2690 char termz[IT_MAX_WORD+20];
2693 const char *use_string = 0;
2694 struct scan_info *scan_info_array;
2695 ZebraScanEntry *glist;
2696 int ords[32], ord_no = 0;
2699 int bases_ok = 0; /* no of databases with OK attribute */
2700 int errCode = 0; /* err code (if any is not OK) */
2701 char *errString = 0; /* addinfo */
2704 char *search_type = NULL;
2705 char rank_type[128];
2708 NMEM rset_nmem = NULL;
2709 struct rset_key_control *kc = 0;
2714 if (attributeset == VAL_NONE)
2715 attributeset = VAL_BIB1;
2720 int termset_value_numeric;
2721 const char *termset_value_string;
2722 attr_init(&termset, zapt, 8);
2723 termset_value_numeric =
2724 attr_find_ex(&termset, NULL, &termset_value_string);
2725 if (termset_value_numeric != -1)
2728 const char *termset_name = 0;
2730 if (termset_value_numeric != -2)
2733 sprintf(resname, "%d", termset_value_numeric);
2734 termset_name = resname;
2737 termset_name = termset_value_string;
2739 limit_set = resultSetRef (zh, termset_name);
2743 yaz_log(YLOG_DEBUG, "position = %d, num = %d set=%d",
2744 pos, num, attributeset);
2746 attr_init(&use, zapt, 1);
2747 use_value = attr_find_ex(&use, &attributeset, &use_string);
2749 if (zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type,
2750 rank_type, &complete_flag, &sort_flag))
2753 zh->errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_TYPE;
2756 yaz_log(YLOG_DEBUG, "use_value = %d", use_value);
2758 if (use_value == -1)
2760 for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
2762 data1_local_attribute *local_attr;
2766 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2768 zh->errString = basenames[base_no];
2769 zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE;
2775 (ord = zebraExplain_lookup_attr_str(zh->reg->zei,
2778 /* we have a match for a raw string attribute */
2780 ords[ord_no++] = ord;
2781 attp.local_attributes = 0; /* no more attributes */
2787 if ((r = att_getentbyatt (zh, &attp, attributeset, use_value,
2790 yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
2791 attributeset, use_value);
2794 errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
2796 errString = odr_strdup(stream, use_string);
2800 sprintf(val_str, "%d", use_value);
2801 errString = odr_strdup(stream, val_str);
2805 errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET;
2810 for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
2811 local_attr = local_attr->next)
2813 ord = zebraExplain_lookup_attr_su(zh->reg->zei,
2814 attp.attset_ordinal,
2817 ords[ord_no++] = ord;
2820 if (!bases_ok && errCode)
2822 zh->errCode = errCode;
2823 zh->errString = errString;
2832 /* prepare dictionary scanning */
2844 yaz_log(YLOG_DEBUG, "rpn_scan pos=%d num=%d before=%d "
2845 "after=%d before+after=%d",
2846 pos, num, before, after, before+after);
2847 scan_info_array = (struct scan_info *)
2848 odr_malloc(stream, ord_no * sizeof(*scan_info_array));
2849 for (i = 0; i < ord_no; i++)
2851 int j, prefix_len = 0;
2852 int before_tmp = before, after_tmp = after;
2853 struct scan_info *scan_info = scan_info_array + i;
2854 struct rpn_char_map_info rcmi;
2856 rpn_char_map_prepare (zh->reg, reg_id, &rcmi);
2858 scan_info->before = before;
2859 scan_info->after = after;
2860 scan_info->odr = stream;
2862 scan_info->list = (struct scan_info_entry *)
2863 odr_malloc(stream, (before+after) * sizeof(*scan_info->list));
2864 for (j = 0; j<before+after; j++)
2865 scan_info->list[j].term = NULL;
2867 prefix_len += key_SU_encode (ords[i], termz + prefix_len);
2868 termz[prefix_len++] = reg_id;
2869 termz[prefix_len] = 0;
2870 strcpy(scan_info->prefix, termz);
2872 if (trans_scan_term(zh, zapt, termz+prefix_len, reg_id) == ZEBRA_FAIL)
2875 dict_scan(zh->reg->dict, termz, &before_tmp, &after_tmp,
2876 scan_info, scan_handle);
2878 glist = (ZebraScanEntry *)
2879 odr_malloc(stream, (before+after)*sizeof(*glist));
2881 rset_nmem = nmem_create();
2882 kc = zebra_key_control_create(zh);
2884 /* consider terms after main term */
2885 for (i = 0; i < ord_no; i++)
2889 for (i = 0; i<after; i++)
2892 const char *mterm = NULL;
2895 int lo = i + pos-1; /* offset in result list */
2897 /* find: j0 is the first of the minimal values */
2898 for (j = 0; j < ord_no; j++)
2900 if (ptr[j] < before+after && ptr[j] >= 0 &&
2901 (tst = scan_info_array[j].list[ptr[j]].term) &&
2902 (!mterm || strcmp (tst, mterm) < 0))
2909 break; /* no value found, stop */
2911 /* get result set for first one , but only if it's within bounds */
2914 /* get result set for first term */
2915 scan_term_untrans(zh, stream->mem, reg_id,
2916 &glist[lo].term, mterm);
2917 rset = rset_trunc(zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
2918 glist[lo].term, strlen(glist[lo].term),
2919 NULL, 0, zapt->term->which, rset_nmem,
2922 ptr[j0]++; /* move index for this set .. */
2923 /* get result set for remaining scan terms */
2924 for (j = j0+1; j<ord_no; j++)
2926 if (ptr[j] < before+after && ptr[j] >= 0 &&
2927 (tst = scan_info_array[j].list[ptr[j]].term) &&
2928 !strcmp (tst, mterm))
2937 zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
2939 strlen(glist[lo].term), NULL, 0,
2940 zapt->term->which,rset_nmem,
2942 rset = rsmulti_or_create(rset_nmem, kc,
2943 2, kc->scope, rsets);
2950 /* merge with limit_set if given */
2955 rsets[1] = rset_dup(limit_set);
2957 rset = rsmulti_and_create(rset_nmem, kc,
2958 kc->scope, 2, rsets);
2961 count_set(rset, &glist[lo].occurrences);
2967 *num_entries -= (after-i);
2969 if (*num_entries < 0)
2972 nmem_destroy(rset_nmem);
2977 /* consider terms before main term */
2978 for (i = 0; i<ord_no; i++)
2981 for (i = 0; i<before; i++)
2984 const char *mterm = NULL;
2987 int lo = before-1-i; /* offset in result list */
2989 for (j = 0; j <ord_no; j++)
2991 if (ptr[j] < before && ptr[j] >= 0 &&
2992 (tst = scan_info_array[j].list[before-1-ptr[j]].term) &&
2993 (!mterm || strcmp (tst, mterm) > 0))
3002 scan_term_untrans (zh, stream->mem, reg_id,
3003 &glist[lo].term, mterm);
3006 (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
3007 glist[lo].term, strlen(glist[lo].term),
3008 NULL, 0, zapt->term->which,rset_nmem,
3013 for (j = j0+1; j<ord_no; j++)
3015 if (ptr[j] < before && ptr[j] >= 0 &&
3016 (tst = scan_info_array[j].list[before-1-ptr[j]].term) &&
3017 !strcmp (tst, mterm))
3022 rsets[1] = rset_trunc(
3024 &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
3026 strlen(glist[lo].term), NULL, 0,
3027 zapt->term->which, rset_nmem,
3029 rset = rsmulti_or_create(rset_nmem, kc,
3030 2, kc->scope, rsets);
3039 rsets[1] = rset_dup(limit_set);
3041 rset = rsmulti_and_create(rset_nmem, kc,
3042 kc->scope, 2, rsets);
3044 count_set (rset, &glist[lo].occurrences);
3048 nmem_destroy(rset_nmem);
3055 if (*num_entries <= 0)
3062 *list = glist + i; /* list is set to first 'real' entry */
3064 yaz_log(YLOG_DEBUG, "position = %d, num_entries = %d",
3065 *position, *num_entries);