1 /* $Id: zrpn.c,v 1.192 2005-05-24 11:35:42 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++;
575 map = zebra_maps_search(zebra_maps, reg_type, &s0, strlen(s0),
577 if (space_split && **map == *CHR_SPACE)
580 /* add non-space char */
581 memcpy(dst_term+j, s1, s0 - s1);
587 if (strchr(REGEX_CHARS, *s1))
595 esc_str(tmpbuf, sizeof(tmpbuf), map[0], strlen(map[0]));
597 strcpy(dst + i, map[0]);
603 dst_term[j++] = '\0';
608 /* term_105/106: handle term, where trunc = Process * and ! and right trunc */
609 static int term_105(ZebraMaps zebra_maps, int reg_type,
610 const char **src, char *dst, int space_split,
611 char *dst_term, int right_truncate)
618 if (!term_pre(zebra_maps, reg_type, src, "*!", "*!", !space_split))
627 dst_term[j++] = *s0++;
632 dst_term[j++] = *s0++;
638 map = zebra_maps_search(zebra_maps, reg_type, &s0, strlen(s0),
640 if (space_split && **map == *CHR_SPACE)
643 /* add non-space char */
644 memcpy(dst_term+j, s1, s0 - s1);
650 if (strchr(REGEX_CHARS, *s1))
658 esc_str(tmpbuf, sizeof(tmpbuf), map[0], strlen(map[0]));
660 strcpy(dst + i, map[0]);
672 dst_term[j++] = '\0';
678 /* gen_regular_rel - generate regular expression from relation
679 * val: border value (inclusive)
680 * islt: 1 if <=; 0 if >=.
682 static void gen_regular_rel(char *dst, int val, int islt)
689 yaz_log(YLOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt);
693 strcpy(dst, "(-[0-9]+|(");
701 strcpy(dst, "([0-9]+|-(");
713 sprintf(numstr, "%d", val);
714 for (w = strlen(numstr); --w >= 0; pos++)
733 strcpy(dst + dst_p, numstr);
734 dst_p = strlen(dst) - pos - 1;
762 for (i = 0; i<pos; i++)
775 /* match everything less than 10^(pos-1) */
777 for (i = 1; i<pos; i++)
778 strcat(dst, "[0-9]?");
782 /* match everything greater than 10^pos */
783 for (i = 0; i <= pos; i++)
784 strcat(dst, "[0-9]");
785 strcat(dst, "[0-9]*");
790 void string_rel_add_char(char **term_p, const char *src, int *indx)
792 if (src[*indx] == '\\')
793 *(*term_p)++ = src[(*indx)++];
794 *(*term_p)++ = src[(*indx)++];
798 * > abc ([b-].*|a[c-].*|ab[d-].*|abc.+)
799 * ([^-a].*|a[^-b].*ab[^-c].*|abc.+)
800 * >= abc ([b-].*|a[c-].*|ab[c-].*)
801 * ([^-a].*|a[^-b].*|ab[c-].*)
802 * < abc ([-0].*|a[-a].*|ab[-b].*)
803 * ([^a-].*|a[^b-].*|ab[^c-].*)
804 * <= abc ([-0].*|a[-a].*|ab[-b].*|abc)
805 * ([^a-].*|a[^b-].*|ab[^c-].*|abc)
807 static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
808 const char **term_sub, char *term_dict,
809 oid_value attributeSet,
810 int reg_type, int space_split, char *term_dst,
816 char *term_tmp = term_dict + strlen(term_dict);
817 char term_component[2*IT_MAX_WORD+20];
819 attr_init(&relation, zapt, 2);
820 relation_value = attr_find(&relation, NULL);
823 yaz_log(YLOG_DEBUG, "string relation value=%d", relation_value);
824 switch (relation_value)
827 if (!term_100(zh->reg->zebra_maps, reg_type,
828 term_sub, term_component,
829 space_split, term_dst))
831 yaz_log(log_level_rpn, "Relation <");
834 for (i = 0; term_component[i]; )
841 string_rel_add_char(&term_tmp, term_component, &j);
846 string_rel_add_char(&term_tmp, term_component, &i);
853 if ((term_tmp - term_dict) > IT_MAX_WORD)
860 if (!term_100(zh->reg->zebra_maps, reg_type,
861 term_sub, term_component,
862 space_split, term_dst))
864 yaz_log(log_level_rpn, "Relation <=");
867 for (i = 0; term_component[i]; )
872 string_rel_add_char(&term_tmp, term_component, &j);
876 string_rel_add_char(&term_tmp, term_component, &i);
885 if ((term_tmp - term_dict) > IT_MAX_WORD)
888 for (i = 0; term_component[i]; )
889 string_rel_add_char(&term_tmp, term_component, &i);
894 if (!term_100 (zh->reg->zebra_maps, reg_type,
895 term_sub, term_component, space_split, term_dst))
897 yaz_log(log_level_rpn, "Relation >");
900 for (i = 0; term_component[i];)
905 string_rel_add_char(&term_tmp, term_component, &j);
910 string_rel_add_char(&term_tmp, term_component, &i);
918 if ((term_tmp - term_dict) > IT_MAX_WORD)
921 for (i = 0; term_component[i];)
922 string_rel_add_char(&term_tmp, term_component, &i);
929 if (!term_100(zh->reg->zebra_maps, reg_type, term_sub,
930 term_component, space_split, term_dst))
932 yaz_log(log_level_rpn, "Relation >=");
935 for (i = 0; term_component[i];)
942 string_rel_add_char(&term_tmp, term_component, &j);
945 if (term_component[i+1])
949 string_rel_add_char(&term_tmp, term_component, &i);
953 string_rel_add_char(&term_tmp, term_component, &i);
960 if ((term_tmp - term_dict) > IT_MAX_WORD)
969 yaz_log(log_level_rpn, "Relation =");
970 if (!term_100(zh->reg->zebra_maps, reg_type, term_sub,
971 term_component, space_split, term_dst))
973 strcat(term_tmp, "(");
974 strcat(term_tmp, term_component);
975 strcat(term_tmp, ")");
978 *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE;
984 static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
985 const char **term_sub,
986 oid_value attributeSet, NMEM stream,
987 struct grep_info *grep_info,
988 int reg_type, int complete_flag,
989 int num_bases, char **basenames,
990 char *term_dst, int xpath_use);
992 static ZEBRA_RES term_trunc(ZebraHandle zh,
993 Z_AttributesPlusTerm *zapt,
994 const char **term_sub,
995 oid_value attributeSet, NMEM stream,
996 struct grep_info *grep_info,
997 int reg_type, int complete_flag,
998 int num_bases, char **basenames,
1000 const char *rank_type, int xpath_use,
1003 struct rset_key_control *kc)
1007 grep_info->isam_p_indx = 0;
1008 res = string_term(zh, zapt, term_sub, attributeSet, stream, grep_info,
1009 reg_type, complete_flag, num_bases, basenames,
1010 term_dst, xpath_use);
1011 if (res != ZEBRA_OK)
1013 if (!*term_sub) /* no more terms ? */
1015 yaz_log(log_level_rpn, "term: %s", term_dst);
1016 *rset = rset_trunc(zh, grep_info->isam_p_buf,
1017 grep_info->isam_p_indx, term_dst,
1018 strlen(term_dst), rank_type, 1 /* preserve pos */,
1019 zapt->term->which, rset_nmem,
1026 static char *nmem_strdup_i(NMEM nmem, int v)
1029 sprintf(val_str, "%d", v);
1030 return nmem_strdup(nmem, val_str);
1033 static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1034 const char **term_sub,
1035 oid_value attributeSet, NMEM stream,
1036 struct grep_info *grep_info,
1037 int reg_type, int complete_flag,
1038 int num_bases, char **basenames,
1039 char *term_dst, int xpath_use)
1041 char term_dict[2*IT_MAX_WORD+4000];
1043 AttrType truncation;
1044 int truncation_value;
1047 const char *use_string = 0;
1048 oid_value curAttributeSet = attributeSet;
1050 struct rpn_char_map_info rcmi;
1051 int space_split = complete_flag ? 0 : 1;
1053 int bases_ok = 0; /* no of databases with OK attribute */
1054 int errCode = 0; /* err code (if any is not OK) */
1055 char *errString = 0; /* addinfo */
1057 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
1058 attr_init(&use, zapt, 1);
1059 use_value = attr_find_ex(&use, &curAttributeSet, &use_string);
1060 yaz_log(log_level_rpn, "string_term, use value %d", use_value);
1061 attr_init(&truncation, zapt, 5);
1062 truncation_value = attr_find(&truncation, NULL);
1063 yaz_log(log_level_rpn, "truncation value %d", truncation_value);
1065 if (use_value == -1) /* no attribute - assumy "any" */
1067 for (base_no = 0; base_no < num_bases; base_no++)
1071 int regex_range = 0;
1074 data1_local_attribute id_xpath_attr;
1075 data1_local_attribute *local_attr;
1076 int max_pos, prefix_len = 0;
1081 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
1083 zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
1084 basenames[base_no]);
1087 if (xpath_use > 0 && use_value == -2)
1089 /* xpath mode and we have a string attribute */
1090 attp.local_attributes = &id_xpath_attr;
1091 attp.attset_ordinal = VAL_IDXPATH;
1092 id_xpath_attr.next = 0;
1094 use_value = xpath_use; /* xpath_use as use-attribute now */
1095 id_xpath_attr.local = use_value;
1097 else if (curAttributeSet == VAL_IDXPATH && use_value >= 0)
1099 /* X-Path attribute, use numeric value directly */
1100 attp.local_attributes = &id_xpath_attr;
1101 attp.attset_ordinal = VAL_IDXPATH;
1102 id_xpath_attr.next = 0;
1103 id_xpath_attr.local = use_value;
1105 else if (use_string &&
1106 (ord = zebraExplain_lookup_attr_str(zh->reg->zei,
1109 /* we have a match for a raw string attribute */
1114 term_dict[prefix_len++] = '|';
1116 term_dict[prefix_len++] = '(';
1118 ord_len = key_SU_encode (ord, ord_buf);
1119 for (i = 0; i<ord_len; i++)
1121 term_dict[prefix_len++] = 1;
1122 term_dict[prefix_len++] = ord_buf[i];
1124 attp.local_attributes = 0; /* no more attributes */
1128 /* lookup in the .att files . Allow string as well */
1129 if ((r = att_getentbyatt (zh, &attp, curAttributeSet, use_value,
1132 yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
1133 curAttributeSet, use_value, r);
1136 /* set was found, but value wasn't defined */
1137 errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
1139 errString = nmem_strdup(stream, use_string);
1141 errString = nmem_strdup_i (stream, use_value);
1146 struct oident oident;
1148 oident.proto = PROTO_Z3950;
1149 oident.oclass = CLASS_ATTSET;
1150 oident.value = curAttributeSet;
1151 oid_ent_to_oid (&oident, oid);
1153 errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET;
1154 errString = nmem_strdup(stream, oident.desc);
1159 for (local_attr = attp.local_attributes; local_attr;
1160 local_attr = local_attr->next)
1165 ord = zebraExplain_lookup_attr_su(zh->reg->zei,
1166 attp.attset_ordinal,
1171 term_dict[prefix_len++] = '|';
1173 term_dict[prefix_len++] = '(';
1175 ord_len = key_SU_encode (ord, ord_buf);
1176 for (i = 0; i<ord_len; i++)
1178 term_dict[prefix_len++] = 1;
1179 term_dict[prefix_len++] = ord_buf[i];
1186 term_dict[prefix_len++] = ')';
1187 term_dict[prefix_len++] = 1;
1188 term_dict[prefix_len++] = reg_type;
1189 yaz_log(log_level_rpn, "reg_type = %d", term_dict[prefix_len-1]);
1190 term_dict[prefix_len] = '\0';
1192 switch (truncation_value)
1194 case -1: /* not specified */
1195 case 100: /* do not truncate */
1196 if (!string_relation (zh, zapt, &termp, term_dict,
1198 reg_type, space_split, term_dst,
1203 zebra_setError(zh, relation_error, 0);
1210 case 1: /* right truncation */
1211 term_dict[j++] = '(';
1212 if (!term_100(zh->reg->zebra_maps, reg_type,
1213 &termp, term_dict + j, space_split, term_dst))
1218 strcat(term_dict, ".*)");
1220 case 2: /* keft truncation */
1221 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1222 if (!term_100(zh->reg->zebra_maps, reg_type,
1223 &termp, term_dict + j, space_split, term_dst))
1228 strcat(term_dict, ")");
1230 case 3: /* left&right truncation */
1231 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1232 if (!term_100(zh->reg->zebra_maps, reg_type,
1233 &termp, term_dict + j, space_split, term_dst))
1238 strcat(term_dict, ".*)");
1240 case 101: /* process # in term */
1241 term_dict[j++] = '(';
1242 if (!term_101(zh->reg->zebra_maps, reg_type,
1243 &termp, term_dict + j, space_split, term_dst))
1248 strcat(term_dict, ")");
1250 case 102: /* Regexp-1 */
1251 term_dict[j++] = '(';
1252 if (!term_102(zh->reg->zebra_maps, reg_type,
1253 &termp, term_dict + j, space_split, term_dst))
1258 strcat(term_dict, ")");
1260 case 103: /* Regexp-2 */
1262 term_dict[j++] = '(';
1264 if (!term_103(zh->reg->zebra_maps, reg_type,
1265 &termp, term_dict + j, ®ex_range,
1266 space_split, term_dst))
1271 strcat(term_dict, ")");
1273 case 104: /* process # and ! in term */
1274 term_dict[j++] = '(';
1275 if (!term_104(zh->reg->zebra_maps, reg_type,
1276 &termp, term_dict + j, space_split, term_dst))
1281 strcat(term_dict, ")");
1283 case 105: /* process * and ! in term */
1284 term_dict[j++] = '(';
1285 if (!term_105(zh->reg->zebra_maps, reg_type,
1286 &termp, term_dict + j, space_split, term_dst, 1))
1291 strcat(term_dict, ")");
1293 case 106: /* process * and ! in term */
1294 term_dict[j++] = '(';
1295 if (!term_105(zh->reg->zebra_maps, reg_type,
1296 &termp, term_dict + j, space_split, term_dst, 0))
1301 strcat(term_dict, ")");
1304 zebra_setError_zint(zh,
1305 YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE,
1312 const char *input = term_dict + prefix_len;
1313 esc_str(buf, sizeof(buf), input, strlen(input));
1317 yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_dict+prefix_len);
1318 r = dict_lookup_grep(zh->reg->dict, term_dict, regex_range,
1319 grep_info, &max_pos, init_pos,
1322 yaz_log(YLOG_WARN, "dict_lookup_grep fail %d", r);
1327 zebra_setError(zh, errCode, errString);
1331 yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1336 /* convert APT search term to UTF8 */
1337 static ZEBRA_RES zapt_term_to_utf8(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1341 Z_Term *term = zapt->term;
1343 switch (term->which)
1345 case Z_Term_general:
1346 if (zh->iconv_to_utf8 != 0)
1348 char *inbuf = term->u.general->buf;
1349 size_t inleft = term->u.general->len;
1350 char *outbuf = termz;
1351 size_t outleft = IT_MAX_WORD-1;
1354 ret = yaz_iconv(zh->iconv_to_utf8, &inbuf, &inleft,
1356 if (ret == (size_t)(-1))
1358 ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0);
1361 YAZ_BIB1_QUERY_TERM_INCLUDES_CHARS_THAT_DO_NOT_TRANSLATE_INTO_,
1369 sizez = term->u.general->len;
1370 if (sizez > IT_MAX_WORD-1)
1371 sizez = IT_MAX_WORD-1;
1372 memcpy (termz, term->u.general->buf, sizez);
1373 termz[sizez] = '\0';
1376 case Z_Term_characterString:
1377 sizez = strlen(term->u.characterString);
1378 if (sizez > IT_MAX_WORD-1)
1379 sizez = IT_MAX_WORD-1;
1380 memcpy (termz, term->u.characterString, sizez);
1381 termz[sizez] = '\0';
1384 zebra_setError(zh, YAZ_BIB1_UNSUPP_CODED_VALUE_FOR_TERM, 0);
1390 /* convert APT SCAN term to internal cmap */
1391 static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1392 char *termz, int reg_type)
1394 char termz0[IT_MAX_WORD];
1396 if (zapt_term_to_utf8(zh, zapt, termz0) == ZEBRA_FAIL)
1397 return ZEBRA_FAIL; /* error */
1401 const char *cp = (const char *) termz0;
1402 const char *cp_end = cp + strlen(cp);
1405 const char *space_map = NULL;
1408 while ((len = (cp_end - cp)) > 0)
1410 map = zebra_maps_input(zh->reg->zebra_maps, reg_type, &cp, len, 0);
1411 if (**map == *CHR_SPACE)
1416 for (src = space_map; *src; src++)
1419 for (src = *map; *src; src++)
1428 char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1429 const char *termz, NMEM stream, unsigned reg_id)
1432 AttrType truncation;
1433 int truncation_value;
1436 attr_init(&truncation, zapt, 5);
1437 truncation_value = attr_find(&truncation, NULL);
1439 switch (truncation_value)
1459 wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list,
1460 termz, strlen(termz));
1462 return nmem_strdup(stream, termz);
1465 char *buf = (char*) nmem_malloc(stream, wrbuf_len(wrbuf)+1);
1466 memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
1467 buf[wrbuf_len(wrbuf)] = '\0';
1472 static void grep_info_delete(struct grep_info *grep_info)
1475 xfree(grep_info->term_no);
1477 xfree(grep_info->isam_p_buf);
1480 static ZEBRA_RES grep_info_prepare(ZebraHandle zh,
1481 Z_AttributesPlusTerm *zapt,
1482 struct grep_info *grep_info,
1486 int termset_value_numeric;
1487 const char *termset_value_string;
1490 grep_info->term_no = 0;
1492 grep_info->isam_p_size = 0;
1493 grep_info->isam_p_buf = NULL;
1495 grep_info->reg_type = reg_type;
1496 grep_info->termset = 0;
1500 attr_init(&termset, zapt, 8);
1501 termset_value_numeric =
1502 attr_find_ex(&termset, NULL, &termset_value_string);
1503 if (termset_value_numeric != -1)
1506 const char *termset_name = 0;
1507 if (termset_value_numeric != -2)
1510 sprintf(resname, "%d", termset_value_numeric);
1511 termset_name = resname;
1514 termset_name = termset_value_string;
1515 yaz_log(log_level_rpn, "creating termset set %s", termset_name);
1516 grep_info->termset = resultSetAdd(zh, termset_name, 1);
1517 if (!grep_info->termset)
1519 zebra_setError(zh, YAZ_BIB1_ILLEGAL_RESULT_SET_NAME, termset_name);
1527 static ZEBRA_RES term_list_trunc(ZebraHandle zh,
1528 Z_AttributesPlusTerm *zapt,
1529 const char *termz_org,
1530 oid_value attributeSet,
1532 int reg_type, int complete_flag,
1533 const char *rank_type, int xpath_use,
1534 int num_bases, char **basenames,
1536 RSET **result_sets, int *num_result_sets,
1537 struct rset_key_control *kc)
1539 char term_dst[IT_MAX_WORD+1];
1540 struct grep_info grep_info;
1541 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1542 const char *termp = termz;
1545 *num_result_sets = 0;
1547 if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL)
1553 if (alloc_sets == *num_result_sets)
1556 RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) *
1559 memcpy(rnew, *result_sets, alloc_sets * sizeof(*rnew));
1560 alloc_sets = alloc_sets + add;
1561 *result_sets = rnew;
1563 res = term_trunc(zh, zapt, &termp, attributeSet,
1565 reg_type, complete_flag,
1566 num_bases, basenames,
1567 term_dst, rank_type,
1568 xpath_use, rset_nmem,
1569 &(*result_sets)[*num_result_sets],
1571 if (res != ZEBRA_OK)
1574 for (i = 0; i < *num_result_sets; i++)
1575 rset_delete((*result_sets)[i]);
1576 grep_info_delete (&grep_info);
1579 if ((*result_sets)[*num_result_sets] == 0)
1581 (*num_result_sets)++;
1583 grep_info_delete(&grep_info);
1587 static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh,
1588 Z_AttributesPlusTerm *zapt,
1589 const char *termz_org,
1590 oid_value attributeSet,
1592 int reg_type, int complete_flag,
1593 const char *rank_type, int xpath_use,
1594 int num_bases, char **basenames,
1597 struct rset_key_control *kc)
1599 RSET *result_sets = 0;
1600 int num_result_sets = 0;
1602 term_list_trunc(zh, zapt, termz_org, attributeSet,
1603 stream, reg_type, complete_flag,
1604 rank_type, xpath_use,
1605 num_bases, basenames,
1607 &result_sets, &num_result_sets, kc);
1608 if (res != ZEBRA_OK)
1610 if (num_result_sets == 0)
1611 *rset = rsnull_create (rset_nmem, kc, 0);
1612 else if (num_result_sets == 1)
1613 *rset = result_sets[0];
1615 *rset = rsprox_create(rset_nmem, kc, kc->scope,
1616 num_result_sets, result_sets,
1617 1 /* ordered */, 0 /* exclusion */,
1618 3 /* relation */, 1 /* distance */);
1624 static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh,
1625 Z_AttributesPlusTerm *zapt,
1626 const char *termz_org,
1627 oid_value attributeSet,
1629 int reg_type, int complete_flag,
1630 const char *rank_type,
1632 int num_bases, char **basenames,
1635 struct rset_key_control *kc)
1637 RSET *result_sets = 0;
1638 int num_result_sets = 0;
1640 term_list_trunc(zh, zapt, termz_org, attributeSet,
1641 stream, reg_type, complete_flag,
1642 rank_type, xpath_use,
1643 num_bases, basenames,
1645 &result_sets, &num_result_sets, kc);
1646 if (res != ZEBRA_OK)
1648 if (num_result_sets == 0)
1649 *rset = rsnull_create (rset_nmem, kc, 0);
1650 else if (num_result_sets == 1)
1651 *rset = result_sets[0];
1653 *rset = rsmulti_or_create(rset_nmem, kc, kc->scope, 0 /* termid */,
1654 num_result_sets, result_sets);
1660 static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh,
1661 Z_AttributesPlusTerm *zapt,
1662 const char *termz_org,
1663 oid_value attributeSet,
1665 int reg_type, int complete_flag,
1666 const char *rank_type,
1668 int num_bases, char **basenames,
1671 struct rset_key_control *kc)
1673 RSET *result_sets = 0;
1674 int num_result_sets = 0;
1676 term_list_trunc(zh, zapt, termz_org, attributeSet,
1677 stream, reg_type, complete_flag,
1678 rank_type, xpath_use,
1679 num_bases, basenames,
1681 &result_sets, &num_result_sets,
1683 if (res != ZEBRA_OK)
1685 if (num_result_sets == 0)
1686 *rset = rsnull_create (rset_nmem, kc, 0);
1687 else if (num_result_sets == 1)
1688 *rset = result_sets[0];
1690 *rset = rsmulti_and_create(rset_nmem, kc, kc->scope,
1691 num_result_sets, result_sets);
1697 static int numeric_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1698 const char **term_sub,
1700 oid_value attributeSet,
1701 struct grep_info *grep_info,
1711 char *term_tmp = term_dict + strlen(term_dict);
1714 attr_init(&relation, zapt, 2);
1715 relation_value = attr_find(&relation, NULL);
1717 yaz_log(log_level_rpn, "numeric relation value=%d", relation_value);
1719 if (!term_100(zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1,
1722 term_value = atoi (term_tmp);
1723 switch (relation_value)
1726 yaz_log(log_level_rpn, "Relation <");
1727 gen_regular_rel(term_tmp, term_value-1, 1);
1730 yaz_log(log_level_rpn, "Relation <=");
1731 gen_regular_rel(term_tmp, term_value, 1);
1734 yaz_log(log_level_rpn, "Relation >=");
1735 gen_regular_rel(term_tmp, term_value, 0);
1738 yaz_log(log_level_rpn, "Relation >");
1739 gen_regular_rel(term_tmp, term_value+1, 0);
1743 yaz_log(log_level_rpn, "Relation =");
1744 sprintf(term_tmp, "(0*%d)", term_value);
1747 *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE;
1750 yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_tmp);
1751 r = dict_lookup_grep(zh->reg->dict, term_dict, 0, grep_info, max_pos,
1754 yaz_log(YLOG_WARN, "dict_lookup_grep fail, rel = gt: %d", r);
1755 yaz_log(log_level_rpn, "%d positions", grep_info->isam_p_indx);
1759 static ZEBRA_RES numeric_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1760 const char **term_sub,
1761 oid_value attributeSet,
1762 struct grep_info *grep_info,
1763 int reg_type, int complete_flag,
1764 int num_bases, char **basenames,
1765 char *term_dst, int xpath_use, NMEM stream)
1767 char term_dict[2*IT_MAX_WORD+2];
1771 const char *use_string = 0;
1772 oid_value curAttributeSet = attributeSet;
1774 struct rpn_char_map_info rcmi;
1776 int bases_ok = 0; /* no of databases with OK attribute */
1777 int errCode = 0; /* err code (if any is not OK) */
1778 char *errString = 0; /* addinfo */
1780 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
1781 attr_init(&use, zapt, 1);
1782 use_value = attr_find_ex(&use, &curAttributeSet, &use_string);
1784 if (use_value == -1)
1787 for (base_no = 0; base_no < num_bases; base_no++)
1790 data1_local_attribute id_xpath_attr;
1791 data1_local_attribute *local_attr;
1792 int max_pos, prefix_len = 0;
1793 int relation_error = 0;
1796 if (use_value == -2) /* string attribute (assume IDXPATH/any) */
1798 use_value = xpath_use;
1799 attp.local_attributes = &id_xpath_attr;
1800 attp.attset_ordinal = VAL_IDXPATH;
1801 id_xpath_attr.next = 0;
1802 id_xpath_attr.local = use_value;
1804 else if (curAttributeSet == VAL_IDXPATH)
1806 attp.local_attributes = &id_xpath_attr;
1807 attp.attset_ordinal = VAL_IDXPATH;
1808 id_xpath_attr.next = 0;
1809 id_xpath_attr.local = use_value;
1813 if ((r = att_getentbyatt (zh, &attp, curAttributeSet, use_value,
1816 yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
1817 curAttributeSet, use_value, r);
1820 errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
1822 errString = nmem_strdup(stream, use_string);
1824 errString = nmem_strdup_i (stream, use_value);
1827 errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET;
1831 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
1833 zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
1834 basenames[base_no]);
1837 for (local_attr = attp.local_attributes; local_attr;
1838 local_attr = local_attr->next)
1844 ord = zebraExplain_lookup_attr_su(zh->reg->zei,
1845 attp.attset_ordinal,
1850 term_dict[prefix_len++] = '|';
1852 term_dict[prefix_len++] = '(';
1854 ord_len = key_SU_encode (ord, ord_buf);
1855 for (i = 0; i<ord_len; i++)
1857 term_dict[prefix_len++] = 1;
1858 term_dict[prefix_len++] = ord_buf[i];
1863 zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, use_value);
1867 term_dict[prefix_len++] = ')';
1868 term_dict[prefix_len++] = 1;
1869 term_dict[prefix_len++] = reg_type;
1870 yaz_log(YLOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
1871 term_dict[prefix_len] = '\0';
1872 if (!numeric_relation(zh, zapt, &termp, term_dict,
1873 attributeSet, grep_info, &max_pos, reg_type,
1874 term_dst, &relation_error))
1878 zebra_setError(zh, relation_error, 0);
1887 zebra_setError(zh, errCode, errString);
1891 yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1895 static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh,
1896 Z_AttributesPlusTerm *zapt,
1898 oid_value attributeSet,
1900 int reg_type, int complete_flag,
1901 const char *rank_type, int xpath_use,
1902 int num_bases, char **basenames,
1905 struct rset_key_control *kc)
1907 char term_dst[IT_MAX_WORD+1];
1908 const char *termp = termz;
1909 RSET *result_sets = 0;
1910 int num_result_sets = 0;
1912 struct grep_info grep_info;
1915 yaz_log(log_level_rpn, "APT_numeric t='%s'", termz);
1916 if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL)
1920 if (alloc_sets == num_result_sets)
1923 RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) *
1926 memcpy(rnew, result_sets, alloc_sets * sizeof(*rnew));
1927 alloc_sets = alloc_sets + add;
1930 yaz_log(YLOG_DEBUG, "APT_numeric termp=%s", termp);
1931 grep_info.isam_p_indx = 0;
1932 res = numeric_term(zh, zapt, &termp, attributeSet, &grep_info,
1933 reg_type, complete_flag, num_bases, basenames,
1934 term_dst, xpath_use,
1936 if (res == ZEBRA_FAIL || termp == 0)
1938 yaz_log(YLOG_DEBUG, "term: %s", term_dst);
1939 result_sets[num_result_sets] =
1940 rset_trunc(zh, grep_info.isam_p_buf,
1941 grep_info.isam_p_indx, term_dst,
1942 strlen(term_dst), rank_type,
1943 0 /* preserve position */,
1944 zapt->term->which, rset_nmem,
1946 if (!result_sets[num_result_sets])
1950 grep_info_delete(&grep_info);
1954 for (i = 0; i<num_result_sets; i++)
1955 rset_delete(result_sets[i]);
1958 if (num_result_sets == 0)
1959 *rset = rsnull_create(rset_nmem, kc, 0);
1960 if (num_result_sets == 1)
1961 *rset = result_sets[0];
1963 *rset = rsmulti_and_create(rset_nmem, kc, kc->scope,
1964 num_result_sets, result_sets);
1970 static ZEBRA_RES rpn_search_APT_local(ZebraHandle zh,
1971 Z_AttributesPlusTerm *zapt,
1973 oid_value attributeSet,
1975 const char *rank_type, NMEM rset_nmem,
1977 struct rset_key_control *kc)
1982 *rset = rstemp_create(rset_nmem, kc, kc->scope,
1983 res_get (zh->res, "setTmpDir"),0 );
1984 rsfd = rset_open(*rset, RSETF_WRITE);
1992 rset_write (rsfd, &key);
1997 static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1998 oid_value attributeSet, NMEM stream,
1999 Z_SortKeySpecList *sort_sequence,
2000 const char *rank_type,
2003 struct rset_key_control *kc)
2006 int sort_relation_value;
2007 AttrType sort_relation_type;
2014 attr_init(&sort_relation_type, zapt, 7);
2015 sort_relation_value = attr_find(&sort_relation_type, &attributeSet);
2017 if (!sort_sequence->specs)
2019 sort_sequence->num_specs = 10;
2020 sort_sequence->specs = (Z_SortKeySpec **)
2021 nmem_malloc(stream, sort_sequence->num_specs *
2022 sizeof(*sort_sequence->specs));
2023 for (i = 0; i<sort_sequence->num_specs; i++)
2024 sort_sequence->specs[i] = 0;
2026 if (zapt->term->which != Z_Term_general)
2029 i = atoi_n ((char *) zapt->term->u.general->buf,
2030 zapt->term->u.general->len);
2031 if (i >= sort_sequence->num_specs)
2033 sprintf(termz, "%d", i);
2035 oe.proto = PROTO_Z3950;
2036 oe.oclass = CLASS_ATTSET;
2037 oe.value = attributeSet;
2038 if (!oid_ent_to_oid (&oe, oid))
2041 sks = (Z_SortKeySpec *) nmem_malloc(stream, sizeof(*sks));
2042 sks->sortElement = (Z_SortElement *)
2043 nmem_malloc(stream, sizeof(*sks->sortElement));
2044 sks->sortElement->which = Z_SortElement_generic;
2045 sk = sks->sortElement->u.generic = (Z_SortKey *)
2046 nmem_malloc(stream, sizeof(*sk));
2047 sk->which = Z_SortKey_sortAttributes;
2048 sk->u.sortAttributes = (Z_SortAttributes *)
2049 nmem_malloc(stream, sizeof(*sk->u.sortAttributes));
2051 sk->u.sortAttributes->id = oid;
2052 sk->u.sortAttributes->list = zapt->attributes;
2054 sks->sortRelation = (int *)
2055 nmem_malloc(stream, sizeof(*sks->sortRelation));
2056 if (sort_relation_value == 1)
2057 *sks->sortRelation = Z_SortKeySpec_ascending;
2058 else if (sort_relation_value == 2)
2059 *sks->sortRelation = Z_SortKeySpec_descending;
2061 *sks->sortRelation = Z_SortKeySpec_ascending;
2063 sks->caseSensitivity = (int *)
2064 nmem_malloc(stream, sizeof(*sks->caseSensitivity));
2065 *sks->caseSensitivity = 0;
2067 sks->which = Z_SortKeySpec_null;
2068 sks->u.null = odr_nullval ();
2069 sort_sequence->specs[i] = sks;
2070 *rset = rsnull_create (rset_nmem, kc, 0);
2075 static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2076 oid_value attributeSet,
2077 struct xpath_location_step *xpath, int max, NMEM mem)
2079 oid_value curAttributeSet = attributeSet;
2081 const char *use_string = 0;
2083 attr_init(&use, zapt, 1);
2084 attr_find_ex(&use, &curAttributeSet, &use_string);
2086 if (!use_string || *use_string != '/')
2089 return zebra_parse_xpath_str(use_string, xpath, max, mem);
2094 static RSET xpath_trunc(ZebraHandle zh, NMEM stream,
2095 int reg_type, const char *term, int use,
2096 oid_value curAttributeSet, NMEM rset_nmem,
2097 struct rset_key_control *kc)
2100 struct grep_info grep_info;
2101 char term_dict[2048];
2104 int ord = zebraExplain_lookup_attr_su(zh->reg->zei, curAttributeSet, use);
2105 int ord_len, i, r, max_pos;
2106 int term_type = Z_Term_characterString;
2107 const char *flags = "void";
2109 if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, '0') == ZEBRA_FAIL)
2110 return rsnull_create(rset_nmem, kc, 0);
2113 return rsnull_create(rset_nmem, kc, 0);
2115 term_dict[prefix_len++] = '|';
2117 term_dict[prefix_len++] = '(';
2119 ord_len = key_SU_encode (ord, ord_buf);
2120 for (i = 0; i<ord_len; i++)
2122 term_dict[prefix_len++] = 1;
2123 term_dict[prefix_len++] = ord_buf[i];
2125 term_dict[prefix_len++] = ')';
2126 term_dict[prefix_len++] = 1;
2127 term_dict[prefix_len++] = reg_type;
2129 strcpy(term_dict+prefix_len, term);
2131 grep_info.isam_p_indx = 0;
2132 r = dict_lookup_grep(zh->reg->dict, term_dict, 0,
2133 &grep_info, &max_pos, 0, grep_handle);
2134 yaz_log(YLOG_DEBUG, "%s %d positions", term,
2135 grep_info.isam_p_indx);
2136 rset = rset_trunc(zh, grep_info.isam_p_buf,
2137 grep_info.isam_p_indx, term, strlen(term),
2138 flags, 1, term_type,rset_nmem,
2140 grep_info_delete(&grep_info);
2145 ZEBRA_RES rpn_search_xpath(ZebraHandle zh,
2146 oid_value attributeSet,
2147 int num_bases, char **basenames,
2148 NMEM stream, const char *rank_type, RSET rset,
2149 int xpath_len, struct xpath_location_step *xpath,
2152 struct rset_key_control *kc)
2154 oid_value curAttributeSet = attributeSet;
2164 yaz_log(YLOG_DEBUG, "xpath len=%d", xpath_len);
2165 for (i = 0; i<xpath_len; i++)
2167 yaz_log(log_level_rpn, "XPATH %d %s", i, xpath[i].part);
2171 curAttributeSet = VAL_IDXPATH;
2181 a[@attr = value]/b[@other = othervalue]
2183 /e/@a val range(e/,range(@a,freetext(w,1015,val),@a),e/)
2184 /a/b val range(b/a/,freetext(w,1016,val),b/a/)
2185 /a/b/@c val range(b/a/,range(@c,freetext(w,1016,val),@c),b/a/)
2186 /a/b[@c = y] val range(b/a/,freetext(w,1016,val),b/a/,@c = y)
2187 /a[@c = y]/b val range(a/,range(b/a/,freetext(w,1016,val),b/a/),a/,@c = y)
2188 /a[@c = x]/b[@c = y] range(a/,range(b/a/,freetext(w,1016,val),b/a/,@c = y),a/,@c = x)
2192 dict_grep_cmap (zh->reg->dict, 0, 0);
2194 for (base_no = 0; base_no < num_bases; base_no++)
2196 int level = xpath_len;
2199 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2201 zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
2202 basenames[base_no]);
2206 while (--level >= 0)
2208 char xpath_rev[128];
2210 RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0;
2214 for (i = level; i >= 1; --i)
2216 const char *cp = xpath[i].part;
2222 memcpy (xpath_rev + len, "[^/]*", 5);
2225 else if (*cp == ' ')
2228 xpath_rev[len++] = 1;
2229 xpath_rev[len++] = ' ';
2233 xpath_rev[len++] = *cp;
2234 xpath_rev[len++] = '/';
2236 else if (i == 1) /* // case */
2238 xpath_rev[len++] = '.';
2239 xpath_rev[len++] = '*';
2244 if (xpath[level].predicate &&
2245 xpath[level].predicate->which == XPATH_PREDICATE_RELATION &&
2246 xpath[level].predicate->u.relation.name[0])
2248 WRBUF wbuf = wrbuf_alloc();
2249 wrbuf_puts(wbuf, xpath[level].predicate->u.relation.name+1);
2250 if (xpath[level].predicate->u.relation.value)
2252 const char *cp = xpath[level].predicate->u.relation.value;
2253 wrbuf_putc(wbuf, '=');
2257 if (strchr(REGEX_CHARS, *cp))
2258 wrbuf_putc(wbuf, '\\');
2259 wrbuf_putc(wbuf, *cp);
2263 wrbuf_puts(wbuf, "");
2264 rset_attr = xpath_trunc(
2265 zh, stream, '0', wrbuf_buf(wbuf), 3,
2266 curAttributeSet, rset_nmem, kc);
2267 wrbuf_free(wbuf, 1);
2274 yaz_log(log_level_rpn, "xpath_rev (%d) = %s", level, xpath_rev);
2275 if (strlen(xpath_rev))
2277 rset_start_tag = xpath_trunc(zh, stream, '0',
2278 xpath_rev, 1, curAttributeSet, rset_nmem, kc);
2280 rset_end_tag = xpath_trunc(zh, stream, '0',
2281 xpath_rev, 2, curAttributeSet, rset_nmem, kc);
2283 rset = rsbetween_create(rset_nmem, kc, kc->scope,
2284 rset_start_tag, rset,
2285 rset_end_tag, rset_attr);
2294 static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2295 oid_value attributeSet, NMEM stream,
2296 Z_SortKeySpecList *sort_sequence,
2297 int num_bases, char **basenames,
2300 struct rset_key_control *kc)
2302 ZEBRA_RES res = ZEBRA_OK;
2304 char *search_type = NULL;
2305 char rank_type[128];
2308 char termz[IT_MAX_WORD+1];
2311 struct xpath_location_step xpath[10];
2315 log_level_rpn = yaz_log_module_level("rpn");
2318 zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type,
2319 rank_type, &complete_flag, &sort_flag);
2321 yaz_log(YLOG_DEBUG, "reg_id=%c", reg_id);
2322 yaz_log(YLOG_DEBUG, "complete_flag=%d", complete_flag);
2323 yaz_log(YLOG_DEBUG, "search_type=%s", search_type);
2324 yaz_log(YLOG_DEBUG, "rank_type=%s", rank_type);
2326 if (zapt_term_to_utf8(zh, zapt, termz) == ZEBRA_FAIL)
2330 return rpn_sort_spec(zh, zapt, attributeSet, stream, sort_sequence,
2331 rank_type, rset_nmem, rset, kc);
2332 xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream);
2336 if (xpath[xpath_len-1].part[0] == '@')
2340 if (!strcmp(search_type, "phrase"))
2342 res = rpn_search_APT_phrase(zh, zapt, termz, attributeSet, stream,
2343 reg_id, complete_flag, rank_type,
2345 num_bases, basenames, rset_nmem,
2348 else if (!strcmp(search_type, "and-list"))
2350 res = rpn_search_APT_and_list(zh, zapt, termz, attributeSet, stream,
2351 reg_id, complete_flag, rank_type,
2353 num_bases, basenames, rset_nmem,
2356 else if (!strcmp(search_type, "or-list"))
2358 res = rpn_search_APT_or_list(zh, zapt, termz, attributeSet, stream,
2359 reg_id, complete_flag, rank_type,
2361 num_bases, basenames, rset_nmem,
2364 else if (!strcmp(search_type, "local"))
2366 res = rpn_search_APT_local(zh, zapt, termz, attributeSet, stream,
2367 rank_type, rset_nmem, rset, kc);
2369 else if (!strcmp(search_type, "numeric"))
2371 res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, stream,
2372 reg_id, complete_flag, rank_type,
2374 num_bases, basenames, rset_nmem,
2379 zebra_setError(zh, YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE, 0);
2382 if (res != ZEBRA_OK)
2386 return rpn_search_xpath(zh, attributeSet, num_bases, basenames,
2387 stream, rank_type, *rset,
2388 xpath_len, xpath, rset_nmem, rset, kc);
2391 static ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs,
2392 oid_value attributeSet,
2393 NMEM stream, NMEM rset_nmem,
2394 Z_SortKeySpecList *sort_sequence,
2395 int num_bases, char **basenames,
2396 RSET **result_sets, int *num_result_sets,
2397 Z_Operator *parent_op,
2398 struct rset_key_control *kc);
2400 ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs,
2401 oid_value attributeSet,
2402 NMEM stream, NMEM rset_nmem,
2403 Z_SortKeySpecList *sort_sequence,
2404 int num_bases, char **basenames,
2407 RSET *result_sets = 0;
2408 int num_result_sets = 0;
2410 struct rset_key_control *kc = zebra_key_control_create(zh);
2412 res = rpn_search_structure(zh, zs, attributeSet,
2415 num_bases, basenames,
2416 &result_sets, &num_result_sets,
2417 0 /* no parent op */,
2419 if (res != ZEBRA_OK)
2422 for (i = 0; i<num_result_sets; i++)
2423 rset_delete(result_sets[i]);
2428 assert(num_result_sets == 1);
2429 assert(result_sets);
2430 assert(*result_sets);
2431 *result_set = *result_sets;
2437 ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs,
2438 oid_value attributeSet,
2439 NMEM stream, NMEM rset_nmem,
2440 Z_SortKeySpecList *sort_sequence,
2441 int num_bases, char **basenames,
2442 RSET **result_sets, int *num_result_sets,
2443 Z_Operator *parent_op,
2444 struct rset_key_control *kc)
2446 *num_result_sets = 0;
2447 if (zs->which == Z_RPNStructure_complex)
2450 Z_Operator *zop = zs->u.complex->roperator;
2451 RSET *result_sets_l = 0;
2452 int num_result_sets_l = 0;
2453 RSET *result_sets_r = 0;
2454 int num_result_sets_r = 0;
2456 res = rpn_search_structure(zh, zs->u.complex->s1,
2457 attributeSet, stream, rset_nmem,
2459 num_bases, basenames,
2460 &result_sets_l, &num_result_sets_l,
2462 if (res != ZEBRA_OK)
2465 for (i = 0; i<num_result_sets_l; i++)
2466 rset_delete(result_sets_l[i]);
2469 res = rpn_search_structure(zh, zs->u.complex->s2,
2470 attributeSet, stream, rset_nmem,
2472 num_bases, basenames,
2473 &result_sets_r, &num_result_sets_r,
2475 if (res != ZEBRA_OK)
2478 for (i = 0; i<num_result_sets_l; i++)
2479 rset_delete(result_sets_l[i]);
2480 for (i = 0; i<num_result_sets_r; i++)
2481 rset_delete(result_sets_r[i]);
2485 /* make a new list of result for all children */
2486 *num_result_sets = num_result_sets_l + num_result_sets_r;
2487 *result_sets = nmem_malloc(stream, *num_result_sets *
2488 sizeof(**result_sets));
2489 memcpy(*result_sets, result_sets_l,
2490 num_result_sets_l * sizeof(**result_sets));
2491 memcpy(*result_sets + num_result_sets_l, result_sets_r,
2492 num_result_sets_r * sizeof(**result_sets));
2494 if (!parent_op || parent_op->which != zop->which
2495 || (zop->which != Z_Operator_and &&
2496 zop->which != Z_Operator_or))
2498 /* parent node different from this one (or non-present) */
2499 /* we must combine result sets now */
2503 case Z_Operator_and:
2504 rset = rsmulti_and_create(rset_nmem, kc,
2506 *num_result_sets, *result_sets);
2509 rset = rsmulti_or_create(rset_nmem, kc,
2510 kc->scope, 0, /* termid */
2511 *num_result_sets, *result_sets);
2513 case Z_Operator_and_not:
2514 rset = rsbool_create_not(rset_nmem, kc,
2519 case Z_Operator_prox:
2520 if (zop->u.prox->which != Z_ProximityOperator_known)
2523 YAZ_BIB1_UNSUPP_PROX_UNIT_CODE,
2527 if (*zop->u.prox->u.known != Z_ProxUnit_word)
2529 zebra_setError_zint(zh,
2530 YAZ_BIB1_UNSUPP_PROX_UNIT_CODE,
2531 *zop->u.prox->u.known);
2536 rset = rsprox_create(rset_nmem, kc,
2538 *num_result_sets, *result_sets,
2539 *zop->u.prox->ordered,
2540 (!zop->u.prox->exclusion ?
2541 0 : *zop->u.prox->exclusion),
2542 *zop->u.prox->relationType,
2543 *zop->u.prox->distance );
2547 zebra_setError(zh, YAZ_BIB1_OPERATOR_UNSUPP, 0);
2550 *num_result_sets = 1;
2551 *result_sets = nmem_malloc(stream, *num_result_sets *
2552 sizeof(**result_sets));
2553 (*result_sets)[0] = rset;
2556 else if (zs->which == Z_RPNStructure_simple)
2561 if (zs->u.simple->which == Z_Operand_APT)
2563 yaz_log(YLOG_DEBUG, "rpn_search_APT");
2564 res = rpn_search_APT(zh, zs->u.simple->u.attributesPlusTerm,
2565 attributeSet, stream, sort_sequence,
2566 num_bases, basenames, rset_nmem, &rset,
2568 if (res != ZEBRA_OK)
2571 else if (zs->u.simple->which == Z_Operand_resultSetId)
2573 yaz_log(YLOG_DEBUG, "rpn_search_ref");
2574 rset = resultSetRef(zh, zs->u.simple->u.resultSetId);
2578 YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
2579 zs->u.simple->u.resultSetId);
2586 zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0);
2589 *num_result_sets = 1;
2590 *result_sets = nmem_malloc(stream, *num_result_sets *
2591 sizeof(**result_sets));
2592 (*result_sets)[0] = rset;
2596 zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0);
2602 struct scan_info_entry {
2608 struct scan_info_entry *list;
2614 static int scan_handle (char *name, const char *info, int pos, void *client)
2616 int len_prefix, idx;
2617 struct scan_info *scan_info = (struct scan_info *) client;
2619 len_prefix = strlen(scan_info->prefix);
2620 if (memcmp (name, scan_info->prefix, len_prefix))
2623 idx = scan_info->after - pos + scan_info->before;
2629 scan_info->list[idx].term = (char *)
2630 odr_malloc(scan_info->odr, strlen(name + len_prefix)+1);
2631 strcpy(scan_info->list[idx].term, name + len_prefix);
2632 assert (*info == sizeof(ISAM_P));
2633 memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
2637 static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type,
2638 char **dst, const char *src)
2640 char term_src[IT_MAX_WORD];
2641 char term_dst[IT_MAX_WORD];
2643 term_untrans (zh, reg_type, term_src, src);
2645 if (zh->iconv_from_utf8 != 0)
2648 char *inbuf = term_src;
2649 size_t inleft = strlen(term_src);
2650 char *outbuf = term_dst;
2651 size_t outleft = sizeof(term_dst)-1;
2654 ret = yaz_iconv (zh->iconv_from_utf8, &inbuf, &inleft,
2656 if (ret == (size_t)(-1))
2659 len = outbuf - term_dst;
2660 *dst = nmem_malloc(stream, len + 1);
2662 memcpy (*dst, term_dst, len);
2666 *dst = nmem_strdup(stream, term_src);
2669 static void count_set (RSET r, int *count)
2676 yaz_log(YLOG_DEBUG, "count_set");
2679 rfd = rset_open (r, RSETF_READ);
2680 while (rset_read (rfd, &key,0 /* never mind terms */))
2682 if (key.mem[0] != psysno)
2684 psysno = key.mem[0];
2690 yaz_log(YLOG_DEBUG, "%d keys, %d records", kno, *count);
2693 ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
2694 oid_value attributeset,
2695 int num_bases, char **basenames,
2696 int *position, int *num_entries, ZebraScanEntry **list,
2697 int *is_partial, RSET limit_set, int return_zero)
2700 int pos = *position;
2701 int num = *num_entries;
2705 char termz[IT_MAX_WORD+20];
2708 const char *use_string = 0;
2709 struct scan_info *scan_info_array;
2710 ZebraScanEntry *glist;
2711 int ords[32], ord_no = 0;
2714 int bases_ok = 0; /* no of databases with OK attribute */
2715 int errCode = 0; /* err code (if any is not OK) */
2716 char *errString = 0; /* addinfo */
2719 char *search_type = NULL;
2720 char rank_type[128];
2723 NMEM rset_nmem = NULL;
2724 struct rset_key_control *kc = 0;
2729 if (attributeset == VAL_NONE)
2730 attributeset = VAL_BIB1;
2735 int termset_value_numeric;
2736 const char *termset_value_string;
2737 attr_init(&termset, zapt, 8);
2738 termset_value_numeric =
2739 attr_find_ex(&termset, NULL, &termset_value_string);
2740 if (termset_value_numeric != -1)
2743 const char *termset_name = 0;
2745 if (termset_value_numeric != -2)
2748 sprintf(resname, "%d", termset_value_numeric);
2749 termset_name = resname;
2752 termset_name = termset_value_string;
2754 limit_set = resultSetRef (zh, termset_name);
2758 yaz_log(YLOG_DEBUG, "position = %d, num = %d set=%d",
2759 pos, num, attributeset);
2761 attr_init(&use, zapt, 1);
2762 use_value = attr_find_ex(&use, &attributeset, &use_string);
2764 if (zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type,
2765 rank_type, &complete_flag, &sort_flag))
2768 zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_TYPE, 0);
2771 yaz_log(YLOG_DEBUG, "use_value = %d", use_value);
2773 if (use_value == -1)
2775 for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
2777 data1_local_attribute *local_attr;
2781 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2783 zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE,
2784 basenames[base_no]);
2790 (ord = zebraExplain_lookup_attr_str(zh->reg->zei,
2793 /* we have a match for a raw string attribute */
2795 ords[ord_no++] = ord;
2796 attp.local_attributes = 0; /* no more attributes */
2802 if ((r = att_getentbyatt (zh, &attp, attributeset, use_value,
2805 yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
2806 attributeset, use_value);
2809 errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
2811 zebra_setError(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
2814 zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
2819 zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, 0);
2825 for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
2826 local_attr = local_attr->next)
2828 ord = zebraExplain_lookup_attr_su(zh->reg->zei,
2829 attp.attset_ordinal,
2832 ords[ord_no++] = ord;
2835 if (!bases_ok && errCode)
2837 zebra_setError(zh, errCode, errString);
2846 /* prepare dictionary scanning */
2858 yaz_log(YLOG_DEBUG, "rpn_scan pos=%d num=%d before=%d "
2859 "after=%d before+after=%d",
2860 pos, num, before, after, before+after);
2861 scan_info_array = (struct scan_info *)
2862 odr_malloc(stream, ord_no * sizeof(*scan_info_array));
2863 for (i = 0; i < ord_no; i++)
2865 int j, prefix_len = 0;
2866 int before_tmp = before, after_tmp = after;
2867 struct scan_info *scan_info = scan_info_array + i;
2868 struct rpn_char_map_info rcmi;
2870 rpn_char_map_prepare (zh->reg, reg_id, &rcmi);
2872 scan_info->before = before;
2873 scan_info->after = after;
2874 scan_info->odr = stream;
2876 scan_info->list = (struct scan_info_entry *)
2877 odr_malloc(stream, (before+after) * sizeof(*scan_info->list));
2878 for (j = 0; j<before+after; j++)
2879 scan_info->list[j].term = NULL;
2881 prefix_len += key_SU_encode (ords[i], termz + prefix_len);
2882 termz[prefix_len++] = reg_id;
2883 termz[prefix_len] = 0;
2884 strcpy(scan_info->prefix, termz);
2886 if (trans_scan_term(zh, zapt, termz+prefix_len, reg_id) == ZEBRA_FAIL)
2889 dict_scan(zh->reg->dict, termz, &before_tmp, &after_tmp,
2890 scan_info, scan_handle);
2892 glist = (ZebraScanEntry *)
2893 odr_malloc(stream, (before+after)*sizeof(*glist));
2895 rset_nmem = nmem_create();
2896 kc = zebra_key_control_create(zh);
2898 /* consider terms after main term */
2899 for (i = 0; i < ord_no; i++)
2903 for (i = 0; i<after; i++)
2906 const char *mterm = NULL;
2909 int lo = i + pos-1; /* offset in result list */
2911 /* find: j0 is the first of the minimal values */
2912 for (j = 0; j < ord_no; j++)
2914 if (ptr[j] < before+after && ptr[j] >= 0 &&
2915 (tst = scan_info_array[j].list[ptr[j]].term) &&
2916 (!mterm || strcmp (tst, mterm) < 0))
2923 break; /* no value found, stop */
2925 /* get result set for first one , but only if it's within bounds */
2928 /* get result set for first term */
2929 scan_term_untrans(zh, stream->mem, reg_id,
2930 &glist[lo].term, mterm);
2931 rset = rset_trunc(zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
2932 glist[lo].term, strlen(glist[lo].term),
2933 NULL, 0, zapt->term->which, rset_nmem,
2936 ptr[j0]++; /* move index for this set .. */
2937 /* get result set for remaining scan terms */
2938 for (j = j0+1; j<ord_no; j++)
2940 if (ptr[j] < before+after && ptr[j] >= 0 &&
2941 (tst = scan_info_array[j].list[ptr[j]].term) &&
2942 !strcmp (tst, mterm))
2951 zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
2953 strlen(glist[lo].term), NULL, 0,
2954 zapt->term->which,rset_nmem,
2956 rset = rsmulti_or_create(rset_nmem, kc,
2957 kc->scope, 0 /* termid */,
2965 /* merge with limit_set if given */
2970 rsets[1] = rset_dup(limit_set);
2972 rset = rsmulti_and_create(rset_nmem, kc,
2977 count_set(rset, &glist[lo].occurrences);
2983 *num_entries -= (after-i);
2985 if (*num_entries < 0)
2988 nmem_destroy(rset_nmem);
2993 /* consider terms before main term */
2994 for (i = 0; i<ord_no; i++)
2997 for (i = 0; i<before; i++)
3000 const char *mterm = NULL;
3003 int lo = before-1-i; /* offset in result list */
3005 for (j = 0; j <ord_no; j++)
3007 if (ptr[j] < before && ptr[j] >= 0 &&
3008 (tst = scan_info_array[j].list[before-1-ptr[j]].term) &&
3009 (!mterm || strcmp (tst, mterm) > 0))
3018 scan_term_untrans (zh, stream->mem, reg_id,
3019 &glist[lo].term, mterm);
3022 (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
3023 glist[lo].term, strlen(glist[lo].term),
3024 NULL, 0, zapt->term->which,rset_nmem,
3029 for (j = j0+1; j<ord_no; j++)
3031 if (ptr[j] < before && ptr[j] >= 0 &&
3032 (tst = scan_info_array[j].list[before-1-ptr[j]].term) &&
3033 !strcmp (tst, mterm))
3038 rsets[1] = rset_trunc(
3040 &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
3042 strlen(glist[lo].term), NULL, 0,
3043 zapt->term->which, rset_nmem,
3045 rset = rsmulti_or_create(rset_nmem, kc,
3046 kc->scope, 0 /* termid */, 2, rsets);
3055 rsets[1] = rset_dup(limit_set);
3057 rset = rsmulti_and_create(rset_nmem, kc,
3058 kc->scope, 2, rsets);
3060 count_set (rset, &glist[lo].occurrences);
3064 nmem_destroy(rset_nmem);
3071 if (*num_entries <= 0)
3078 *list = glist + i; /* list is set to first 'real' entry */
3080 yaz_log(YLOG_DEBUG, "position = %d, num_entries = %d",
3081 *position, *num_entries);