1 /* $Id: zrpn.c,v 1.138 2004-05-26 13:52:26 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
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
34 #include <zebra_xpath.h>
40 #include <rsbetween.h>
42 struct rpn_char_map_info {
51 Z_AttributesPlusTerm *zapt;
55 static const char **rpn_char_map_handler (void *vp, const char **from, int len)
57 struct rpn_char_map_info *p = (struct rpn_char_map_info *) vp;
58 const char **out = zebra_maps_input (p->zm, p->reg_type, from, len);
62 const char *outp = *out;
63 yaz_log (LOG_LOG, "---");
66 yaz_log (LOG_LOG, "%02X", *outp);
74 static void rpn_char_map_prepare (struct zebra_register *reg, int reg_type,
75 struct rpn_char_map_info *map_info)
77 map_info->zm = reg->zebra_maps;
78 map_info->reg_type = reg_type;
79 dict_grep_cmap (reg->dict, map_info, rpn_char_map_handler);
82 static int attr_find_ex (AttrType *src, oid_value *attributeSetP,
83 const char **string_value)
87 num_attributes = src->zapt->attributes->num_attributes;
88 while (src->major < num_attributes)
90 Z_AttributeElement *element;
92 element = src->zapt->attributes->attributes[src->major];
93 if (src->type == *element->attributeType)
95 switch (element->which)
97 case Z_AttributeValue_numeric:
99 if (element->attributeSet && attributeSetP)
103 attrset = oid_getentbyoid (element->attributeSet);
104 *attributeSetP = attrset->value;
106 return *element->value.numeric;
108 case Z_AttributeValue_complex:
109 if (src->minor >= element->value.complex->num_list)
111 if (element->attributeSet && attributeSetP)
115 attrset = oid_getentbyoid (element->attributeSet);
116 *attributeSetP = attrset->value;
118 if (element->value.complex->list[src->minor]->which ==
119 Z_StringOrNumeric_numeric)
123 *element->value.complex->list[src->minor-1]->u.numeric;
125 else if (element->value.complex->list[src->minor]->which ==
126 Z_StringOrNumeric_string)
132 element->value.complex->list[src->minor-1]->u.string;
146 static int attr_find (AttrType *src, oid_value *attributeSetP)
148 return attr_find_ex (src, attributeSetP, 0);
151 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
174 static void term_untrans (ZebraHandle zh, int reg_type,
175 char *dst, const char *src)
180 const char *cp = zebra_maps_output (zh->reg->zebra_maps,
182 if (!cp && len < IT_MAX_WORD-1)
185 while (*cp && len < IT_MAX_WORD-1)
191 static void add_isam_p (const char *name, const char *info,
194 if (p->isam_p_indx == p->isam_p_size)
196 ISAMS_P *new_isam_p_buf;
200 p->isam_p_size = 2*p->isam_p_size + 100;
201 new_isam_p_buf = (ISAMS_P *) xmalloc (sizeof(*new_isam_p_buf) *
205 memcpy (new_isam_p_buf, p->isam_p_buf,
206 p->isam_p_indx * sizeof(*p->isam_p_buf));
207 xfree (p->isam_p_buf);
209 p->isam_p_buf = new_isam_p_buf;
212 new_term_no = (int *) xmalloc (sizeof(*new_term_no) *
216 memcpy (new_term_no, p->isam_p_buf,
217 p->isam_p_indx * sizeof(*p->term_no));
220 p->term_no = new_term_no;
223 assert (*info == sizeof(*p->isam_p_buf));
224 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
231 char term_tmp[IT_MAX_WORD];
233 int len = key_SU_decode (&su_code, name);
235 term_untrans (p->zh, p->reg_type, term_tmp, name+len+1);
236 logf (LOG_LOG, "grep: %d %c %s", su_code, name[len], term_tmp);
237 zebraExplain_lookup_ord (p->zh->reg->zei,
238 su_code, &db, &set, &use);
239 logf (LOG_LOG, "grep: set=%d use=%d db=%s", set, use, db);
241 resultSetAddTerm (p->zh, p->termset, name[len], db,
248 static int grep_handle (char *name, const char *info, void *p)
250 add_isam_p (name, info, (struct grep_info *) p);
254 static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src,
255 const char *ct1, const char *ct2)
257 const char *s1, *s0 = *src;
260 /* skip white space */
263 if (ct1 && strchr (ct1, *s0))
265 if (ct2 && strchr (ct2, *s0))
268 map = zebra_maps_input (zebra_maps, reg_type, &s1, strlen(s1));
269 if (**map != *CHR_SPACE)
277 #define REGEX_CHARS " []()|.*+?!"
279 /* term_100: handle term, where trunc=none (no operators at all) */
280 static int term_100 (ZebraMaps zebra_maps, int reg_type,
281 const char **src, char *dst, int space_split,
289 const char *space_start = 0;
290 const char *space_end = 0;
292 if (!term_pre (zebra_maps, reg_type, src, NULL, NULL))
298 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
301 if (**map == *CHR_SPACE)
304 else /* complete subfield only. */
306 if (**map == *CHR_SPACE)
307 { /* save space mapping for later .. */
312 else if (space_start)
313 { /* reload last space */
314 while (space_start < space_end)
316 if (strchr (REGEX_CHARS, *space_start))
318 dst_term[j++] = *space_start;
319 dst[i++] = *space_start++;
322 space_start = space_end = 0;
325 /* add non-space char */
328 if (strchr(REGEX_CHARS, *s1))
340 /* term_101: handle term, where trunc=Process # */
341 static int term_101 (ZebraMaps zebra_maps, int reg_type,
342 const char **src, char *dst, int space_split,
350 if (!term_pre (zebra_maps, reg_type, src, "#", "#"))
359 dst_term[j++] = *s0++;
364 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
365 if (space_split && **map == *CHR_SPACE)
369 if (strchr(REGEX_CHARS, *s1))
377 dst_term[j++] = '\0';
382 /* term_103: handle term, where trunc=re-2 (regular expressions) */
383 static int term_103 (ZebraMaps zebra_maps, int reg_type, const char **src,
384 char *dst, int *errors, int space_split,
392 if (!term_pre (zebra_maps, reg_type, src, "^\\()[].*+?|", "("))
395 if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
398 *errors = s0[1] - '0';
405 if (strchr ("^\\()[].*+?|-", *s0))
413 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
414 if (**map == *CHR_SPACE)
418 if (strchr(REGEX_CHARS, *s1))
431 /* term_103: handle term, where trunc=re-1 (regular expressions) */
432 static int term_102 (ZebraMaps zebra_maps, int reg_type, const char **src,
433 char *dst, int space_split, char *dst_term)
435 return term_103 (zebra_maps, reg_type, src, dst, NULL, space_split,
440 /* term_104: handle term, where trunc=Process # and ! */
441 static int term_104 (ZebraMaps zebra_maps, int reg_type,
442 const char **src, char *dst, int space_split,
450 if (!term_pre (zebra_maps, reg_type, src, "?*#", "?*#"))
457 dst_term[j++] = *s0++;
458 if (*s0 >= '0' && *s0 <= '9')
461 while (*s0 >= '0' && *s0 <= '9')
463 limit = limit * 10 + (*s0 - '0');
464 dst_term[j++] = *s0++;
484 dst_term[j++] = *s0++;
489 dst_term[j++] = *s0++;
493 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
494 if (space_split && **map == *CHR_SPACE)
498 if (strchr(REGEX_CHARS, *s1))
506 dst_term[j++] = '\0';
511 /* term_105/106: handle term, where trunc=Process * and ! and right trunc */
512 static int term_105 (ZebraMaps zebra_maps, int reg_type,
513 const char **src, char *dst, int space_split,
514 char *dst_term, int right_truncate)
521 if (!term_pre (zebra_maps, reg_type, src, "*!", "*!"))
530 dst_term[j++] = *s0++;
535 dst_term[j++] = *s0++;
539 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
540 if (space_split && **map == *CHR_SPACE)
544 if (strchr(REGEX_CHARS, *s1))
558 dst_term[j++] = '\0';
564 /* gen_regular_rel - generate regular expression from relation
565 * val: border value (inclusive)
566 * islt: 1 if <=; 0 if >=.
568 static void gen_regular_rel (char *dst, int val, int islt)
575 logf (LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt);
579 strcpy (dst, "(-[0-9]+|(");
587 strcpy (dst, "([0-9]+|-(");
588 dst_p = strlen (dst);
598 dst_p = strlen (dst);
599 sprintf (numstr, "%d", val);
600 for (w = strlen(numstr); --w >= 0; pos++)
619 strcpy (dst + dst_p, numstr);
620 dst_p = strlen(dst) - pos - 1;
648 for (i = 0; i<pos; i++)
661 /* match everything less than 10^(pos-1) */
663 for (i=1; i<pos; i++)
664 strcat (dst, "[0-9]?");
668 /* match everything greater than 10^pos */
669 for (i = 0; i <= pos; i++)
670 strcat (dst, "[0-9]");
671 strcat (dst, "[0-9]*");
676 void string_rel_add_char (char **term_p, const char *src, int *indx)
678 if (src[*indx] == '\\')
679 *(*term_p)++ = src[(*indx)++];
680 *(*term_p)++ = src[(*indx)++];
684 * > abc ([b-].*|a[c-].*|ab[d-].*|abc.+)
685 * ([^-a].*|a[^-b].*ab[^-c].*|abc.+)
686 * >= abc ([b-].*|a[c-].*|ab[c-].*)
687 * ([^-a].*|a[^-b].*|ab[c-].*)
688 * < abc ([-0].*|a[-a].*|ab[-b].*)
689 * ([^a-].*|a[^b-].*|ab[^c-].*)
690 * <= abc ([-0].*|a[-a].*|ab[-b].*|abc)
691 * ([^a-].*|a[^b-].*|ab[^c-].*|abc)
693 static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
694 const char **term_sub, char *term_dict,
695 oid_value attributeSet,
696 int reg_type, int space_split, char *term_dst)
701 char *term_tmp = term_dict + strlen(term_dict);
702 char term_component[2*IT_MAX_WORD+20];
704 attr_init (&relation, zapt, 2);
705 relation_value = attr_find (&relation, NULL);
707 logf (LOG_DEBUG, "string relation value=%d", relation_value);
708 switch (relation_value)
711 if (!term_100 (zh->reg->zebra_maps, reg_type,
712 term_sub, term_component,
713 space_split, term_dst))
715 logf (LOG_DEBUG, "Relation <");
718 for (i = 0; term_component[i]; )
725 string_rel_add_char (&term_tmp, term_component, &j);
730 string_rel_add_char (&term_tmp, term_component, &i);
737 if ((term_tmp - term_dict) > IT_MAX_WORD)
744 if (!term_100 (zh->reg->zebra_maps, reg_type,
745 term_sub, term_component,
746 space_split, term_dst))
748 logf (LOG_DEBUG, "Relation <=");
751 for (i = 0; term_component[i]; )
756 string_rel_add_char (&term_tmp, term_component, &j);
760 string_rel_add_char (&term_tmp, term_component, &i);
769 if ((term_tmp - term_dict) > IT_MAX_WORD)
772 for (i = 0; term_component[i]; )
773 string_rel_add_char (&term_tmp, term_component, &i);
778 if (!term_100 (zh->reg->zebra_maps, reg_type,
779 term_sub, term_component, space_split, term_dst))
781 logf (LOG_DEBUG, "Relation >");
784 for (i = 0; term_component[i];)
789 string_rel_add_char (&term_tmp, term_component, &j);
794 string_rel_add_char (&term_tmp, term_component, &i);
802 if ((term_tmp - term_dict) > IT_MAX_WORD)
805 for (i = 0; term_component[i];)
806 string_rel_add_char (&term_tmp, term_component, &i);
813 if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub,
814 term_component, space_split, term_dst))
816 logf (LOG_DEBUG, "Relation >=");
819 for (i = 0; term_component[i];)
826 string_rel_add_char (&term_tmp, term_component, &j);
829 if (term_component[i+1])
833 string_rel_add_char (&term_tmp, term_component, &i);
837 string_rel_add_char (&term_tmp, term_component, &i);
844 if ((term_tmp - term_dict) > IT_MAX_WORD)
852 logf (LOG_DEBUG, "Relation =");
853 if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub,
854 term_component, space_split, term_dst))
856 strcat (term_tmp, "(");
857 strcat (term_tmp, term_component);
858 strcat (term_tmp, ")");
863 static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
864 const char **term_sub,
865 oid_value attributeSet, NMEM stream,
866 struct grep_info *grep_info,
867 int reg_type, int complete_flag,
868 int num_bases, char **basenames,
869 char *term_dst, int xpath_use);
871 static RSET term_trunc (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
872 const char **term_sub,
873 oid_value attributeSet, NMEM stream,
874 struct grep_info *grep_info,
875 int reg_type, int complete_flag,
876 int num_bases, char **basenames,
878 const char *rank_type, int xpath_use)
881 grep_info->isam_p_indx = 0;
882 r = string_term (zh, zapt, term_sub, attributeSet, stream, grep_info,
883 reg_type, complete_flag, num_bases, basenames,
884 term_dst, xpath_use);
887 logf (LOG_DEBUG, "term: %s", term_dst);
888 return rset_trunc (zh, grep_info->isam_p_buf,
889 grep_info->isam_p_indx, term_dst,
890 strlen(term_dst), rank_type, 1 /* preserve pos */,
895 static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
896 const char **term_sub,
897 oid_value attributeSet, NMEM stream,
898 struct grep_info *grep_info,
899 int reg_type, int complete_flag,
900 int num_bases, char **basenames,
901 char *term_dst, int xpath_use)
903 char term_dict[2*IT_MAX_WORD+4000];
906 int truncation_value;
909 const char *use_string = 0;
910 oid_value curAttributeSet = attributeSet;
912 struct rpn_char_map_info rcmi;
913 int space_split = complete_flag ? 0 : 1;
915 int bases_ok = 0; /* no of databases with OK attribute */
916 int errCode = 0; /* err code (if any is not OK) */
917 char *errString = 0; /* addinfo */
919 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
920 attr_init (&use, zapt, 1);
921 use_value = attr_find_ex (&use, &curAttributeSet, &use_string);
922 logf (LOG_DEBUG, "string_term, use value %d", use_value);
923 attr_init (&truncation, zapt, 5);
924 truncation_value = attr_find (&truncation, NULL);
925 logf (LOG_DEBUG, "truncation value %d", truncation_value);
927 if (use_value == -1) /* no attribute - assumy "any" */
929 for (base_no = 0; base_no < num_bases; base_no++)
932 data1_local_attribute id_xpath_attr;
933 data1_local_attribute *local_attr;
934 int max_pos, prefix_len = 0;
938 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
940 zh->errCode = 109; /* Database unavailable */
941 zh->errString = basenames[base_no];
944 if (xpath_use > 0 && use_value == -2)
946 use_value = xpath_use;
947 attp.local_attributes = &id_xpath_attr;
948 attp.attset_ordinal = VAL_IDXPATH;
949 id_xpath_attr.next = 0;
950 id_xpath_attr.local = use_value;
952 else if (curAttributeSet == VAL_IDXPATH)
954 attp.local_attributes = &id_xpath_attr;
955 attp.attset_ordinal = VAL_IDXPATH;
956 id_xpath_attr.next = 0;
957 id_xpath_attr.local = use_value;
961 if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value,
964 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
965 curAttributeSet, use_value, r);
968 /* set was found, but value wasn't defined */
971 errString = nmem_strdup(stream, use_string);
975 sprintf (val_str, "%d", use_value);
976 errString = nmem_strdup (stream, val_str);
982 struct oident oident;
984 oident.proto = PROTO_Z3950;
985 oident.oclass = CLASS_ATTSET;
986 oident.value = curAttributeSet;
987 oid_ent_to_oid (&oident, oid);
990 errString = nmem_strdup (stream, oident.desc);
995 for (local_attr = attp.local_attributes; local_attr;
996 local_attr = local_attr->next)
1002 ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal,
1007 term_dict[prefix_len++] = '|';
1009 term_dict[prefix_len++] = '(';
1011 ord_len = key_SU_encode (ord, ord_buf);
1012 for (i = 0; i<ord_len; i++)
1014 term_dict[prefix_len++] = 1;
1015 term_dict[prefix_len++] = ord_buf[i];
1024 sprintf (val_str, "%d", use_value);
1026 errString = nmem_strdup (stream, val_str);
1030 bases_ok++; /* this has OK attributes */
1032 term_dict[prefix_len++] = ')';
1033 term_dict[prefix_len++] = 1;
1034 term_dict[prefix_len++] = reg_type;
1035 logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
1036 term_dict[prefix_len] = '\0';
1038 switch (truncation_value)
1040 case -1: /* not specified */
1041 case 100: /* do not truncate */
1042 if (!string_relation (zh, zapt, &termp, term_dict,
1044 reg_type, space_split, term_dst))
1046 logf (LOG_LOG, "dict_lookup_grep: %s", term_dict+prefix_len);
1047 r = dict_lookup_grep (zh->reg->dict, term_dict, 0,
1048 grep_info, &max_pos, 0, grep_handle);
1050 logf (LOG_WARN, "dict_lookup_grep fail %d", r);
1052 case 1: /* right truncation */
1053 term_dict[j++] = '(';
1054 if (!term_100 (zh->reg->zebra_maps, reg_type,
1055 &termp, term_dict + j, space_split, term_dst))
1057 strcat (term_dict, ".*)");
1058 dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1059 &max_pos, 0, grep_handle);
1061 case 2: /* keft truncation */
1062 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1063 if (!term_100 (zh->reg->zebra_maps, reg_type,
1064 &termp, term_dict + j, space_split, term_dst))
1066 strcat (term_dict, ")");
1067 dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1068 &max_pos, 0, grep_handle);
1070 case 3: /* left&right truncation */
1071 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1072 if (!term_100 (zh->reg->zebra_maps, reg_type,
1073 &termp, term_dict + j, space_split, term_dst))
1075 strcat (term_dict, ".*)");
1076 dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1077 &max_pos, 0, grep_handle);
1081 case 101: /* process # in term */
1082 term_dict[j++] = '(';
1083 if (!term_101 (zh->reg->zebra_maps, reg_type,
1084 &termp, term_dict + j, space_split, term_dst))
1086 strcat (term_dict, ")");
1087 r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1088 &max_pos, 0, grep_handle);
1090 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", r);
1092 case 102: /* Regexp-1 */
1093 term_dict[j++] = '(';
1094 if (!term_102 (zh->reg->zebra_maps, reg_type,
1095 &termp, term_dict + j, space_split, term_dst))
1097 strcat (term_dict, ")");
1098 logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r);
1099 r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1100 &max_pos, 0, grep_handle);
1102 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
1105 case 103: /* Regexp-2 */
1107 term_dict[j++] = '(';
1108 if (!term_103 (zh->reg->zebra_maps, reg_type,
1109 &termp, term_dict + j, &r, space_split, term_dst))
1111 strcat (term_dict, ")");
1112 logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r);
1113 r = dict_lookup_grep (zh->reg->dict, term_dict, r, grep_info,
1114 &max_pos, 2, grep_handle);
1116 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
1119 case 104: /* process # and ! in term */
1120 term_dict[j++] = '(';
1121 if (!term_104 (zh->reg->zebra_maps, reg_type,
1122 &termp, term_dict + j, space_split, term_dst))
1124 strcat (term_dict, ")");
1125 r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1126 &max_pos, 0, grep_handle);
1128 logf (LOG_WARN, "dict_lookup_grep err, trunc=#/!: %d", r);
1130 case 105: /* process * and ! in term */
1131 term_dict[j++] = '(';
1132 if (!term_105 (zh->reg->zebra_maps, reg_type,
1133 &termp, term_dict + j, space_split, term_dst, 1))
1135 strcat (term_dict, ")");
1136 r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1137 &max_pos, 0, grep_handle);
1139 logf (LOG_WARN, "dict_lookup_grep err, trunc=*/!: %d", r);
1141 case 106: /* process * and ! in term */
1142 term_dict[j++] = '(';
1143 if (!term_105 (zh->reg->zebra_maps, reg_type,
1144 &termp, term_dict + j, space_split, term_dst, 0))
1146 strcat (term_dict, ")");
1147 r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info,
1148 &max_pos, 0, grep_handle);
1150 logf (LOG_WARN, "dict_lookup_grep err, trunc=*/!: %d", r);
1156 zh->errCode = errCode;
1157 zh->errString = errString;
1161 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1166 /* convert APT search term to UTF8 */
1167 static int zapt_term_to_utf8 (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1171 Z_Term *term = zapt->term;
1173 switch (term->which)
1175 case Z_Term_general:
1176 if (zh->iconv_to_utf8 != 0)
1178 char *inbuf = term->u.general->buf;
1179 size_t inleft = term->u.general->len;
1180 char *outbuf = termz;
1181 size_t outleft = IT_MAX_WORD-1;
1184 ret = yaz_iconv(zh->iconv_to_utf8, &inbuf, &inleft,
1186 if (ret == (size_t)(-1))
1188 ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0);
1196 sizez = term->u.general->len;
1197 if (sizez > IT_MAX_WORD-1)
1198 sizez = IT_MAX_WORD-1;
1199 memcpy (termz, term->u.general->buf, sizez);
1200 termz[sizez] = '\0';
1203 case Z_Term_characterString:
1204 sizez = strlen(term->u.characterString);
1205 if (sizez > IT_MAX_WORD-1)
1206 sizez = IT_MAX_WORD-1;
1207 memcpy (termz, term->u.characterString, sizez);
1208 termz[sizez] = '\0';
1217 /* convert APT SCAN term to internal cmap */
1218 static int trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1219 char *termz, int reg_type)
1221 char termz0[IT_MAX_WORD];
1223 if (zapt_term_to_utf8(zh, zapt, termz0))
1224 return -1; /* error */
1228 const char *cp = (const char *) termz0;
1229 const char *cp_end = cp + strlen(cp);
1232 const char *space_map = NULL;
1235 while ((len = (cp_end - cp)) > 0)
1237 map = zebra_maps_input (zh->reg->zebra_maps, reg_type, &cp, len);
1238 if (**map == *CHR_SPACE)
1243 for (src = space_map; *src; src++)
1246 for (src = *map; *src; src++)
1255 static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no,
1256 int ordered, int exclusion, int relation, int distance)
1261 struct it_key **buf;
1263 char prox_term[1024];
1264 int length_prox_term = 0;
1265 int min_nn = 10000000;
1267 int term_type = Z_Term_characterString;
1268 const char *flags = NULL;
1270 rsfd = (RSFD *) xmalloc (sizeof(*rsfd)*rset_no);
1271 more = (int *) xmalloc (sizeof(*more)*rset_no);
1272 buf = (struct it_key **) xmalloc (sizeof(*buf)*rset_no);
1275 for (i = 0; i<rset_no; i++)
1278 for (j = 0; j<rset[i]->no_rset_terms; j++)
1280 const char *nflags = rset[i]->rset_terms[j]->flags;
1281 char *term = rset[i]->rset_terms[j]->name;
1282 int lterm = strlen(term);
1283 if (lterm + length_prox_term < sizeof(prox_term)-1)
1285 if (length_prox_term)
1286 prox_term[length_prox_term++] = ' ';
1287 strcpy (prox_term + length_prox_term, term);
1288 length_prox_term += lterm;
1290 if (min_nn > rset[i]->rset_terms[j]->nn)
1291 min_nn = rset[i]->rset_terms[j]->nn;
1293 term_type = rset[i]->rset_terms[j]->type;
1295 /* only if all term types are of type characterString .. */
1296 /* the resulting term is of that type */
1297 if (term_type != Z_Term_characterString)
1298 term_type = Z_Term_general;
1301 for (i = 0; i<rset_no; i++)
1306 for (i = 0; i<rset_no; i++)
1308 buf[i] = (struct it_key *) xmalloc (sizeof(**buf));
1309 rsfd[i] = rset_open (rset[i], RSETF_READ);
1310 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i], &term_index)))
1315 /* at least one is empty ... return null set */
1316 rset_null_parms parms;
1318 parms.rset_term = rset_term_create (prox_term, length_prox_term,
1320 parms.rset_term->nn = 0;
1321 result = rset_create (rset_kind_null, &parms);
1323 else if (ordered && relation == 3 && exclusion == 0 && distance == 1)
1325 /* special proximity case = phrase search ... */
1326 rset_temp_parms parms;
1329 parms.rset_term = rset_term_create (prox_term, length_prox_term,
1331 parms.rset_term->nn = min_nn;
1332 parms.cmp = key_compare_it;
1333 parms.key_size = sizeof (struct it_key);
1334 parms.temp_path = res_get (zh->res, "setTmpDir");
1335 result = rset_create (rset_kind_temp, &parms);
1336 rsfd_result = rset_open (result, RSETF_WRITE);
1340 for (i = 1; i<rset_no; i++)
1349 cmp = key_compare_it (buf[i], buf[i-1]);
1352 more[i-1] = rset_read (rset[i-1], rsfd[i-1],
1353 buf[i-1], &term_index);
1358 if (buf[i-1]->seqno+1 != buf[i]->seqno)
1360 more[i-1] = rset_read (rset[i-1], rsfd[i-1],
1361 buf[i-1], &term_index);
1367 more[i] = rset_read (rset[i], rsfd[i], buf[i],
1374 rset_write (result, rsfd_result, buf[0]);
1375 more[0] = rset_read (*rset, *rsfd, *buf, &term_index);
1378 rset_close (result, rsfd_result);
1380 else if (rset_no == 2)
1382 /* generic proximity case (two input sets only) ... */
1383 rset_temp_parms parms;
1386 yaz_log (LOG_LOG, "generic prox, dist=%d, relation=%d, ordered=%d"
1388 distance, relation, ordered, exclusion);
1389 parms.rset_term = rset_term_create (prox_term, length_prox_term,
1391 parms.rset_term->nn = min_nn;
1392 parms.cmp = key_compare_it;
1393 parms.key_size = sizeof (struct it_key);
1394 parms.temp_path = res_get (zh->res, "setTmpDir");
1395 result = rset_create (rset_kind_temp, &parms);
1396 rsfd_result = rset_open (result, RSETF_WRITE);
1398 while (more[0] && more[1])
1400 int cmp = key_compare_it (buf[0], buf[1]);
1402 more[0] = rset_read (rset[0], rsfd[0], buf[0], &term_index);
1404 more[1] = rset_read (rset[1], rsfd[1], buf[1], &term_index);
1407 int sysno = buf[0]->sysno;
1411 seqno[n++] = buf[0]->seqno;
1412 while ((more[0] = rset_read (rset[0], rsfd[0], buf[0],
1414 sysno == buf[0]->sysno)
1416 seqno[n++] = buf[0]->seqno;
1419 for (i = 0; i<n; i++)
1421 int diff = buf[1]->seqno - seqno[i];
1422 int excl = exclusion;
1423 if (!ordered && diff < 0)
1428 if (diff < distance && diff >= 0)
1432 if (diff <= distance && diff >= 0)
1436 if (diff == distance && diff >= 0)
1440 if (diff >= distance && diff >= 0)
1444 if (diff > distance && diff >= 0)
1448 if (diff != distance && diff >= 0)
1454 rset_write (result, rsfd_result, buf[1]);
1458 } while ((more[1] = rset_read (rset[1], rsfd[1], buf[1],
1460 sysno == buf[1]->sysno);
1463 rset_close (result, rsfd_result);
1467 rset_null_parms parms;
1469 parms.rset_term = rset_term_create (prox_term, length_prox_term,
1471 parms.rset_term->nn = 0;
1472 result = rset_create (rset_kind_null, &parms);
1474 for (i = 0; i<rset_no; i++)
1477 rset_close (rset[i], rsfd[i]);
1487 char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1488 const char *termz, NMEM stream, unsigned reg_id)
1491 AttrType truncation;
1492 int truncation_value;
1495 attr_init (&truncation, zapt, 5);
1496 truncation_value = attr_find (&truncation, NULL);
1498 switch (truncation_value)
1518 wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list,
1519 termz, strlen(termz));
1521 return nmem_strdup(stream, termz);
1524 char *buf = (char*) nmem_malloc (stream, wrbuf_len(wrbuf)+1);
1525 memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
1526 buf[wrbuf_len(wrbuf)] = '\0';
1531 static void grep_info_delete (struct grep_info *grep_info)
1534 xfree(grep_info->term_no);
1536 xfree (grep_info->isam_p_buf);
1539 static int grep_info_prepare (ZebraHandle zh,
1540 Z_AttributesPlusTerm *zapt,
1541 struct grep_info *grep_info,
1546 int termset_value_numeric;
1547 const char *termset_value_string;
1550 grep_info->term_no = 0;
1552 grep_info->isam_p_size = 0;
1553 grep_info->isam_p_buf = NULL;
1555 grep_info->reg_type = reg_type;
1556 grep_info->termset = 0;
1560 attr_init (&termset, zapt, 8);
1561 termset_value_numeric =
1562 attr_find_ex (&termset, NULL, &termset_value_string);
1563 if (termset_value_numeric != -1)
1566 const char *termset_name = 0;
1567 if (termset_value_numeric != -2)
1570 sprintf (resname, "%d", termset_value_numeric);
1571 termset_name = resname;
1574 termset_name = termset_value_string;
1575 logf (LOG_LOG, "creating termset set %s", termset_name);
1576 grep_info->termset = resultSetAdd (zh, termset_name, 1);
1577 if (!grep_info->termset)
1580 zh->errString = nmem_strdup (stream, termset_name);
1588 static RSET rpn_search_APT_phrase (ZebraHandle zh,
1589 Z_AttributesPlusTerm *zapt,
1590 const char *termz_org,
1591 oid_value attributeSet,
1593 int reg_type, int complete_flag,
1594 const char *rank_type, int xpath_use,
1595 int num_bases, char **basenames)
1597 char term_dst[IT_MAX_WORD+1];
1598 RSET rset[60], result;
1600 struct grep_info grep_info;
1601 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1602 const char *termp = termz;
1605 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1609 logf (LOG_DEBUG, "APT_phrase termp=%s", termp);
1610 rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet,
1612 reg_type, complete_flag,
1613 num_bases, basenames,
1614 term_dst, rank_type,
1618 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1621 grep_info_delete (&grep_info);
1624 rset_null_parms parms;
1626 parms.rset_term = rset_term_create (termz, -1, rank_type,
1628 return rset_create (rset_kind_null, &parms);
1630 else if (rset_no == 1)
1632 result = rpn_prox (zh, rset, rset_no, 1, 0, 3, 1);
1633 for (i = 0; i<rset_no; i++)
1634 rset_delete (rset[i]);
1638 static RSET rpn_search_APT_or_list (ZebraHandle zh,
1639 Z_AttributesPlusTerm *zapt,
1640 const char *termz_org,
1641 oid_value attributeSet,
1643 int reg_type, int complete_flag,
1644 const char *rank_type,
1646 int num_bases, char **basenames)
1648 char term_dst[IT_MAX_WORD+1];
1649 RSET rset[60], result;
1651 struct grep_info grep_info;
1652 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1653 const char *termp = termz;
1655 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1659 logf (LOG_DEBUG, "APT_or_list termp=%s", termp);
1660 rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet,
1662 reg_type, complete_flag,
1663 num_bases, basenames,
1664 term_dst, rank_type,
1668 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1671 grep_info_delete (&grep_info);
1674 rset_null_parms parms;
1676 parms.rset_term = rset_term_create (termz, -1, rank_type,
1678 return rset_create (rset_kind_null, &parms);
1681 for (i = 1; i<rset_no; i++)
1683 rset_bool_parms bool_parms;
1685 bool_parms.rset_l = result;
1686 bool_parms.rset_r = rset[i];
1687 bool_parms.key_size = sizeof(struct it_key);
1688 bool_parms.cmp = key_compare_it;
1689 result = rset_create (rset_kind_or, &bool_parms);
1694 static RSET rpn_search_APT_and_list (ZebraHandle zh,
1695 Z_AttributesPlusTerm *zapt,
1696 const char *termz_org,
1697 oid_value attributeSet,
1699 int reg_type, int complete_flag,
1700 const char *rank_type,
1702 int num_bases, char **basenames)
1704 char term_dst[IT_MAX_WORD+1];
1705 RSET rset[60], result;
1707 struct grep_info grep_info;
1708 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1709 const char *termp = termz;
1711 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1715 logf (LOG_DEBUG, "APT_and_list termp=%s", termp);
1716 rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet,
1718 reg_type, complete_flag,
1719 num_bases, basenames,
1720 term_dst, rank_type,
1724 assert (rset[rset_no]);
1725 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1728 grep_info_delete (&grep_info);
1731 rset_null_parms parms;
1733 parms.rset_term = rset_term_create (termz, -1, rank_type,
1735 return rset_create (rset_kind_null, &parms);
1738 for (i = 1; i<rset_no; i++)
1740 rset_bool_parms bool_parms;
1742 bool_parms.rset_l = result;
1743 bool_parms.rset_r = rset[i];
1744 bool_parms.key_size = sizeof(struct it_key);
1745 bool_parms.cmp = key_compare_it;
1746 result = rset_create (rset_kind_and, &bool_parms);
1751 static int numeric_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1752 const char **term_sub,
1754 oid_value attributeSet,
1755 struct grep_info *grep_info,
1764 char *term_tmp = term_dict + strlen(term_dict);
1766 attr_init (&relation, zapt, 2);
1767 relation_value = attr_find (&relation, NULL);
1769 logf (LOG_DEBUG, "numeric relation value=%d", relation_value);
1771 if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1,
1774 term_value = atoi (term_tmp);
1775 switch (relation_value)
1778 logf (LOG_DEBUG, "Relation <");
1779 gen_regular_rel (term_tmp, term_value-1, 1);
1782 logf (LOG_DEBUG, "Relation <=");
1783 gen_regular_rel (term_tmp, term_value, 1);
1786 logf (LOG_DEBUG, "Relation >=");
1787 gen_regular_rel (term_tmp, term_value, 0);
1790 logf (LOG_DEBUG, "Relation >");
1791 gen_regular_rel (term_tmp, term_value+1, 0);
1795 logf (LOG_DEBUG, "Relation =");
1796 sprintf (term_tmp, "(0*%d)", term_value);
1798 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp);
1799 r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, max_pos,
1802 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
1803 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1807 static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1808 const char **term_sub,
1809 oid_value attributeSet, struct grep_info *grep_info,
1810 int reg_type, int complete_flag,
1811 int num_bases, char **basenames,
1812 char *term_dst, int xpath_use, NMEM stream)
1814 char term_dict[2*IT_MAX_WORD+2];
1818 const char *use_string = 0;
1819 oid_value curAttributeSet = attributeSet;
1821 struct rpn_char_map_info rcmi;
1823 int bases_ok = 0; /* no of databases with OK attribute */
1824 int errCode = 0; /* err code (if any is not OK) */
1825 char *errString = 0; /* addinfo */
1827 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
1828 attr_init (&use, zapt, 1);
1829 use_value = attr_find_ex (&use, &curAttributeSet, &use_string);
1831 if (use_value == -1)
1834 for (base_no = 0; base_no < num_bases; base_no++)
1837 data1_local_attribute id_xpath_attr;
1838 data1_local_attribute *local_attr;
1839 int max_pos, prefix_len = 0;
1842 if (use_value == -2) /* string attribute (assume IDXPATH/any) */
1844 use_value = xpath_use;
1845 attp.local_attributes = &id_xpath_attr;
1846 attp.attset_ordinal = VAL_IDXPATH;
1847 id_xpath_attr.next = 0;
1848 id_xpath_attr.local = use_value;
1850 else if (curAttributeSet == VAL_IDXPATH)
1852 attp.local_attributes = &id_xpath_attr;
1853 attp.attset_ordinal = VAL_IDXPATH;
1854 id_xpath_attr.next = 0;
1855 id_xpath_attr.local = use_value;
1859 if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value,
1862 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
1863 curAttributeSet, use_value, r);
1867 sprintf (val_str, "%d", use_value);
1868 errString = nmem_strdup (stream, val_str);
1876 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
1878 zh->errCode = 109; /* Database unavailable */
1879 zh->errString = basenames[base_no];
1882 for (local_attr = attp.local_attributes; local_attr;
1883 local_attr = local_attr->next)
1889 ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal,
1894 term_dict[prefix_len++] = '|';
1896 term_dict[prefix_len++] = '(';
1898 ord_len = key_SU_encode (ord, ord_buf);
1899 for (i = 0; i<ord_len; i++)
1901 term_dict[prefix_len++] = 1;
1902 term_dict[prefix_len++] = ord_buf[i];
1908 sprintf (val_str, "%d", use_value);
1910 errString = nmem_strdup (stream, val_str);
1914 term_dict[prefix_len++] = ')';
1915 term_dict[prefix_len++] = 1;
1916 term_dict[prefix_len++] = reg_type;
1917 logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
1918 term_dict[prefix_len] = '\0';
1919 if (!numeric_relation (zh, zapt, &termp, term_dict,
1920 attributeSet, grep_info, &max_pos, reg_type,
1926 zh->errCode = errCode;
1927 zh->errString = errString;
1931 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1935 static RSET rpn_search_APT_numeric (ZebraHandle zh,
1936 Z_AttributesPlusTerm *zapt,
1938 oid_value attributeSet,
1940 int reg_type, int complete_flag,
1941 const char *rank_type, int xpath_use,
1942 int num_bases, char **basenames)
1944 char term_dst[IT_MAX_WORD+1];
1945 const char *termp = termz;
1946 RSET rset[60], result;
1947 int i, r, rset_no = 0;
1948 struct grep_info grep_info;
1950 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1954 logf (LOG_DEBUG, "APT_numeric termp=%s", termp);
1955 grep_info.isam_p_indx = 0;
1956 r = numeric_term (zh, zapt, &termp, attributeSet, &grep_info,
1957 reg_type, complete_flag, num_bases, basenames,
1958 term_dst, xpath_use,
1962 logf (LOG_DEBUG, "term: %s", term_dst);
1963 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1964 grep_info.isam_p_indx, term_dst,
1965 strlen(term_dst), rank_type,
1966 0 /* preserve position */,
1968 assert (rset[rset_no]);
1969 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1972 grep_info_delete (&grep_info);
1975 rset_null_parms parms;
1977 parms.rset_term = rset_term_create (term_dst, -1, rank_type,
1979 return rset_create (rset_kind_null, &parms);
1982 for (i = 1; i<rset_no; i++)
1984 rset_bool_parms bool_parms;
1986 bool_parms.rset_l = result;
1987 bool_parms.rset_r = rset[i];
1988 bool_parms.key_size = sizeof(struct it_key);
1989 bool_parms.cmp = key_compare_it;
1990 result = rset_create (rset_kind_and, &bool_parms);
1995 static RSET rpn_search_APT_local (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1997 oid_value attributeSet,
1999 const char *rank_type)
2004 rset_temp_parms parms;
2006 parms.rset_term = rset_term_create (termz, -1, rank_type,
2008 parms.cmp = key_compare_it;
2009 parms.key_size = sizeof (struct it_key);
2010 parms.temp_path = res_get (zh->res, "setTmpDir");
2011 result = rset_create (rset_kind_temp, &parms);
2012 rsfd = rset_open (result, RSETF_WRITE);
2014 key.sysno = atoi (termz);
2018 rset_write (result, rsfd, &key);
2019 rset_close (result, rsfd);
2023 static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2024 oid_value attributeSet, NMEM stream,
2025 Z_SortKeySpecList *sort_sequence,
2026 const char *rank_type)
2028 rset_null_parms parms;
2030 int sort_relation_value;
2031 AttrType sort_relation_type;
2036 Z_AttributeElement *ae;
2041 attr_init (&sort_relation_type, zapt, 7);
2042 sort_relation_value = attr_find (&sort_relation_type, &attributeSet);
2044 attr_init (&use_type, zapt, 1);
2045 use_value = attr_find (&use_type, &attributeSet);
2047 if (!sort_sequence->specs)
2049 sort_sequence->num_specs = 10;
2050 sort_sequence->specs = (Z_SortKeySpec **)
2051 nmem_malloc (stream, sort_sequence->num_specs *
2052 sizeof(*sort_sequence->specs));
2053 for (i = 0; i<sort_sequence->num_specs; i++)
2054 sort_sequence->specs[i] = 0;
2056 if (zapt->term->which != Z_Term_general)
2059 i = atoi_n ((char *) zapt->term->u.general->buf,
2060 zapt->term->u.general->len);
2061 if (i >= sort_sequence->num_specs)
2063 sprintf (termz, "%d", i);
2065 oe.proto = PROTO_Z3950;
2066 oe.oclass = CLASS_ATTSET;
2067 oe.value = attributeSet;
2068 if (!oid_ent_to_oid (&oe, oid))
2071 sks = (Z_SortKeySpec *) nmem_malloc (stream, sizeof(*sks));
2072 sks->sortElement = (Z_SortElement *)
2073 nmem_malloc (stream, sizeof(*sks->sortElement));
2074 sks->sortElement->which = Z_SortElement_generic;
2075 sk = sks->sortElement->u.generic = (Z_SortKey *)
2076 nmem_malloc (stream, sizeof(*sk));
2077 sk->which = Z_SortKey_sortAttributes;
2078 sk->u.sortAttributes = (Z_SortAttributes *)
2079 nmem_malloc (stream, sizeof(*sk->u.sortAttributes));
2081 sk->u.sortAttributes->id = oid;
2082 sk->u.sortAttributes->list = (Z_AttributeList *)
2083 nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list));
2084 sk->u.sortAttributes->list->num_attributes = 1;
2085 sk->u.sortAttributes->list->attributes = (Z_AttributeElement **)
2086 nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list->attributes));
2087 ae = *sk->u.sortAttributes->list->attributes = (Z_AttributeElement *)
2088 nmem_malloc (stream, sizeof(**sk->u.sortAttributes->list->attributes));
2089 ae->attributeSet = 0;
2090 ae->attributeType = (int *)
2091 nmem_malloc (stream, sizeof(*ae->attributeType));
2092 *ae->attributeType = 1;
2093 ae->which = Z_AttributeValue_numeric;
2094 ae->value.numeric = (int *)
2095 nmem_malloc (stream, sizeof(*ae->value.numeric));
2096 *ae->value.numeric = use_value;
2098 sks->sortRelation = (int *)
2099 nmem_malloc (stream, sizeof(*sks->sortRelation));
2100 if (sort_relation_value == 1)
2101 *sks->sortRelation = Z_SortKeySpec_ascending;
2102 else if (sort_relation_value == 2)
2103 *sks->sortRelation = Z_SortKeySpec_descending;
2105 *sks->sortRelation = Z_SortKeySpec_ascending;
2107 sks->caseSensitivity = (int *)
2108 nmem_malloc (stream, sizeof(*sks->caseSensitivity));
2109 *sks->caseSensitivity = 0;
2111 sks->which = Z_SortKeySpec_null;
2112 sks->u.null = odr_nullval ();
2113 sort_sequence->specs[i] = sks;
2115 parms.rset_term = rset_term_create (termz, -1, rank_type,
2117 return rset_create (rset_kind_null, &parms);
2121 static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2122 oid_value attributeSet,
2123 struct xpath_location_step *xpath, int max, NMEM mem)
2125 oid_value curAttributeSet = attributeSet;
2127 const char *use_string = 0;
2129 attr_init (&use, zapt, 1);
2130 attr_find_ex (&use, &curAttributeSet, &use_string);
2132 if (!use_string || *use_string != '/')
2135 return zebra_parse_xpath_str(use_string, xpath, max, mem);
2140 static RSET xpath_trunc(ZebraHandle zh, NMEM stream,
2141 int reg_type, const char *term, int use,
2142 oid_value curAttributeSet)
2145 struct grep_info grep_info;
2146 char term_dict[2048];
2149 int ord = zebraExplain_lookupSU (zh->reg->zei, curAttributeSet, use);
2150 int ord_len, i, r, max_pos;
2151 int term_type = Z_Term_characterString;
2152 const char *flags = "void";
2154 if (grep_info_prepare (zh, 0 /* zapt */, &grep_info, '0', stream))
2156 rset_null_parms parms;
2158 parms.rset_term = rset_term_create (term, strlen(term),
2160 parms.rset_term->nn = 0;
2161 return rset_create (rset_kind_null, &parms);
2166 rset_null_parms parms;
2168 parms.rset_term = rset_term_create (term, strlen(term),
2170 parms.rset_term->nn = 0;
2171 return rset_create (rset_kind_null, &parms);
2174 term_dict[prefix_len++] = '|';
2176 term_dict[prefix_len++] = '(';
2178 ord_len = key_SU_encode (ord, ord_buf);
2179 for (i = 0; i<ord_len; i++)
2181 term_dict[prefix_len++] = 1;
2182 term_dict[prefix_len++] = ord_buf[i];
2184 term_dict[prefix_len++] = ')';
2185 term_dict[prefix_len++] = 1;
2186 term_dict[prefix_len++] = reg_type;
2188 strcpy (term_dict+prefix_len, term);
2190 grep_info.isam_p_indx = 0;
2191 r = dict_lookup_grep (zh->reg->dict, term_dict, 0,
2192 &grep_info, &max_pos, 0, grep_handle);
2193 yaz_log (LOG_LOG, "%s %d positions", term,
2194 grep_info.isam_p_indx);
2195 rset = rset_trunc (zh, grep_info.isam_p_buf,
2196 grep_info.isam_p_indx, term, strlen(term),
2197 flags, 1, term_type);
2198 grep_info_delete (&grep_info);
2202 static RSET rpn_search_xpath (ZebraHandle zh,
2203 oid_value attributeSet,
2204 int num_bases, char **basenames,
2205 NMEM stream, const char *rank_type, RSET rset,
2206 int xpath_len, struct xpath_location_step *xpath)
2208 oid_value curAttributeSet = attributeSet;
2215 yaz_log (LOG_LOG, "len=%d", xpath_len);
2216 for (i = 0; i<xpath_len; i++)
2218 yaz_log (LOG_LOG, "XPATH %d %s", i, xpath[i].part);
2222 curAttributeSet = VAL_IDXPATH;
2232 a[@attr=value]/b[@other=othervalue]
2234 /e/@a val range(e/,range(@a,freetext(w,1015,val),@a),e/)
2235 /a/b val range(b/a/,freetext(w,1016,val),b/a/)
2236 /a/b/@c val range(b/a/,range(@c,freetext(w,1016,val),@c),b/a/)
2237 /a/b[@c=y] val range(b/a/,freetext(w,1016,val),b/a/,@c=y)
2238 /a[@c=y]/b val range(a/,range(b/a/,freetext(w,1016,val),b/a/),a/,@c=y)
2239 /a[@c=x]/b[@c=y] range(a/,range(b/a/,freetext(w,1016,val),b/a/,@c=y),a/,@c=x)
2243 dict_grep_cmap (zh->reg->dict, 0, 0);
2245 for (base_no = 0; base_no < num_bases; base_no++)
2247 int level = xpath_len;
2250 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2252 zh->errCode = 109; /* Database unavailable */
2253 zh->errString = basenames[base_no];
2256 while (--level >= 0)
2258 char xpath_rev[128];
2260 rset_between_parms parms;
2261 RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0;
2265 for (i = level; i >= 1; --i)
2267 const char *cp = xpath[i].part;
2273 memcpy (xpath_rev + len, "[^/]*", 5);
2276 else if (*cp == ' ')
2279 xpath_rev[len++] = 1;
2280 xpath_rev[len++] = ' ';
2284 xpath_rev[len++] = *cp;
2285 xpath_rev[len++] = '/';
2287 else if (i == 1) /* // case */
2289 xpath_rev[len++] = '.';
2290 xpath_rev[len++] = '*';
2295 if (xpath[level].predicate &&
2296 xpath[level].predicate->which == XPATH_PREDICATE_RELATION &&
2297 xpath[level].predicate->u.relation.name[0])
2299 WRBUF wbuf = wrbuf_alloc();
2300 wrbuf_puts(wbuf, xpath[level].predicate->u.relation.name+1);
2301 if (xpath[level].predicate->u.relation.value)
2303 const char *cp = xpath[level].predicate->u.relation.value;
2304 wrbuf_putc(wbuf, '=');
2308 if (strchr(REGEX_CHARS, *cp))
2309 wrbuf_putc(wbuf, '\\');
2310 wrbuf_putc(wbuf, *cp);
2314 wrbuf_puts(wbuf, "");
2315 rset_attr = xpath_trunc (
2316 zh, stream, '0', wrbuf_buf(wbuf), 3, curAttributeSet);
2317 wrbuf_free(wbuf, 1);
2324 yaz_log (LOG_LOG, "xpath_rev (%d) = %s", level, xpath_rev);
2325 if (strlen(xpath_rev))
2327 rset_start_tag = xpath_trunc(zh, stream,
2328 '0', xpath_rev, 1, curAttributeSet);
2330 rset_end_tag = xpath_trunc(zh, stream,
2331 '0', xpath_rev, 2, curAttributeSet);
2333 parms.key_size = sizeof(struct it_key);
2334 parms.cmp = key_compare_it;
2335 parms.rset_l = rset_start_tag;
2336 parms.rset_m = rset;
2337 parms.rset_r = rset_end_tag;
2338 parms.rset_attr = rset_attr;
2339 parms.printer = key_print_it;
2340 rset = rset_create (rset_kind_between, &parms);
2351 static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2352 oid_value attributeSet, NMEM stream,
2353 Z_SortKeySpecList *sort_sequence,
2354 int num_bases, char **basenames)
2357 char *search_type = NULL;
2358 char rank_type[128];
2361 char termz[IT_MAX_WORD+1];
2365 struct xpath_location_step xpath[10];
2367 zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type,
2368 rank_type, &complete_flag, &sort_flag);
2370 logf (LOG_DEBUG, "reg_id=%c", reg_id);
2371 logf (LOG_DEBUG, "complete_flag=%d", complete_flag);
2372 logf (LOG_DEBUG, "search_type=%s", search_type);
2373 logf (LOG_DEBUG, "rank_type=%s", rank_type);
2375 if (zapt_term_to_utf8(zh, zapt, termz))
2379 return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence,
2381 xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream);
2385 if (xpath[xpath_len-1].part[0] == '@')
2389 if (!strcmp (search_type, "phrase"))
2391 rset = rpn_search_APT_phrase (zh, zapt, termz, attributeSet, stream,
2392 reg_id, complete_flag, rank_type,
2394 num_bases, basenames);
2396 else if (!strcmp (search_type, "and-list"))
2398 rset = rpn_search_APT_and_list (zh, zapt, termz, attributeSet, stream,
2399 reg_id, complete_flag, rank_type,
2401 num_bases, basenames);
2403 else if (!strcmp (search_type, "or-list"))
2405 rset = rpn_search_APT_or_list (zh, zapt, termz, attributeSet, stream,
2406 reg_id, complete_flag, rank_type,
2408 num_bases, basenames);
2410 else if (!strcmp (search_type, "local"))
2412 rset = rpn_search_APT_local (zh, zapt, termz, attributeSet, stream,
2415 else if (!strcmp (search_type, "numeric"))
2417 rset = rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream,
2418 reg_id, complete_flag, rank_type,
2420 num_bases, basenames);
2422 else if (!strcmp (search_type, "always"))
2428 return rpn_search_xpath (zh, attributeSet, num_bases, basenames,
2429 stream, rank_type, rset, xpath_len, xpath);
2432 static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs,
2433 oid_value attributeSet, NMEM stream,
2434 Z_SortKeySpecList *sort_sequence,
2435 int num_bases, char **basenames)
2438 if (zs->which == Z_RPNStructure_complex)
2440 Z_Operator *zop = zs->u.complex->roperator;
2441 rset_bool_parms bool_parms;
2443 bool_parms.rset_l = rpn_search_structure (zh, zs->u.complex->s1,
2444 attributeSet, stream,
2446 num_bases, basenames);
2447 if (bool_parms.rset_l == NULL)
2449 bool_parms.rset_r = rpn_search_structure (zh, zs->u.complex->s2,
2450 attributeSet, stream,
2452 num_bases, basenames);
2453 if (bool_parms.rset_r == NULL)
2455 rset_delete (bool_parms.rset_l);
2458 bool_parms.key_size = sizeof(struct it_key);
2459 bool_parms.cmp = key_compare_it;
2463 case Z_Operator_and:
2464 r = rset_create (rset_kind_and, &bool_parms);
2467 r = rset_create (rset_kind_or, &bool_parms);
2469 case Z_Operator_and_not:
2470 r = rset_create (rset_kind_not, &bool_parms);
2472 case Z_Operator_prox:
2473 if (zop->u.prox->which != Z_ProximityOperator_known)
2478 if (*zop->u.prox->u.known != Z_ProxUnit_word)
2480 char *val = (char *) nmem_malloc (stream, 16);
2482 zh->errString = val;
2483 sprintf (val, "%d", *zop->u.prox->u.known);
2490 rsets[0] = bool_parms.rset_l;
2491 rsets[1] = bool_parms.rset_r;
2493 r = rpn_prox (zh, rsets, 2,
2494 *zop->u.prox->ordered,
2495 (!zop->u.prox->exclusion ? 0 :
2496 *zop->u.prox->exclusion),
2497 *zop->u.prox->relationType,
2498 *zop->u.prox->distance);
2499 rset_delete (rsets[0]);
2500 rset_delete (rsets[1]);
2508 else if (zs->which == Z_RPNStructure_simple)
2510 if (zs->u.simple->which == Z_Operand_APT)
2512 logf (LOG_DEBUG, "rpn_search_APT");
2513 r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm,
2514 attributeSet, stream, sort_sequence,
2515 num_bases, basenames);
2517 else if (zs->u.simple->which == Z_Operand_resultSetId)
2519 logf (LOG_DEBUG, "rpn_search_ref");
2520 r = resultSetRef (zh, zs->u.simple->u.resultSetId);
2523 r = rset_create (rset_kind_null, NULL);
2526 nmem_strdup (stream, zs->u.simple->u.resultSetId);
2547 RSET rpn_search (ZebraHandle zh, NMEM nmem,
2548 Z_RPNQuery *rpn, int num_bases, char **basenames,
2549 const char *setname,
2554 oid_value attributeSet;
2555 Z_SortKeySpecList *sort_sequence;
2559 zh->errString = NULL;
2562 sort_sequence = (Z_SortKeySpecList *)
2563 nmem_malloc (nmem, sizeof(*sort_sequence));
2564 sort_sequence->num_specs = 10;
2565 sort_sequence->specs = (Z_SortKeySpec **)
2566 nmem_malloc (nmem, sort_sequence->num_specs *
2567 sizeof(*sort_sequence->specs));
2568 for (i = 0; i<sort_sequence->num_specs; i++)
2569 sort_sequence->specs[i] = 0;
2571 attrset = oid_getentbyoid (rpn->attributeSetId);
2572 attributeSet = attrset->value;
2573 rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet,
2574 nmem, sort_sequence, num_bases, basenames);
2579 logf (LOG_DEBUG, "search error: %d", zh->errCode);
2581 for (i = 0; sort_sequence->specs[i]; i++)
2583 sort_sequence->num_specs = i;
2585 resultSetRank (zh, sset, rset);
2588 logf (LOG_DEBUG, "resultSetSortSingle in rpn_search");
2589 resultSetSortSingle (zh, nmem, sset, rset,
2590 sort_sequence, &sort_status);
2593 logf (LOG_DEBUG, "resultSetSortSingle status = %d", zh->errCode);
2599 struct scan_info_entry {
2605 struct scan_info_entry *list;
2611 static int scan_handle (char *name, const char *info, int pos, void *client)
2613 int len_prefix, idx;
2614 struct scan_info *scan_info = (struct scan_info *) client;
2616 len_prefix = strlen(scan_info->prefix);
2617 if (memcmp (name, scan_info->prefix, len_prefix))
2619 if (pos > 0) idx = scan_info->after - pos + scan_info->before;
2622 scan_info->list[idx].term = (char *)
2623 odr_malloc (scan_info->odr, strlen(name + len_prefix)+1);
2624 strcpy (scan_info->list[idx].term, name + len_prefix);
2625 assert (*info == sizeof(ISAMS_P));
2626 memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAMS_P));
2630 static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type,
2631 char **dst, const char *src)
2633 char term_src[IT_MAX_WORD];
2634 char term_dst[IT_MAX_WORD];
2636 term_untrans (zh, reg_type, term_src, src);
2638 if (zh->iconv_from_utf8 != 0)
2641 char *inbuf = term_src;
2642 size_t inleft = strlen(term_src);
2643 char *outbuf = term_dst;
2644 size_t outleft = sizeof(term_dst)-1;
2647 ret = yaz_iconv (zh->iconv_from_utf8, &inbuf, &inleft,
2649 if (ret == (size_t)(-1))
2652 len = outbuf - term_dst;
2653 *dst = nmem_malloc (stream, len + 1);
2655 memcpy (*dst, term_dst, len);
2659 *dst = nmem_strdup (stream, term_src);
2662 static void count_set (RSET r, int *count)
2670 logf (LOG_DEBUG, "count_set");
2673 rfd = rset_open (r, RSETF_READ);
2674 while (rset_read (r, rfd, &key, &term_index))
2676 if (key.sysno != psysno)
2683 rset_close (r, rfd);
2684 logf (LOG_DEBUG, "%d keys, %d records", kno, *count);
2687 void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
2688 oid_value attributeset,
2689 int num_bases, char **basenames,
2690 int *position, int *num_entries, ZebraScanEntry **list,
2691 int *is_partial, RSET limit_set, int return_zero)
2694 int pos = *position;
2695 int num = *num_entries;
2699 char termz[IT_MAX_WORD+20];
2702 const char *use_string = 0;
2703 struct scan_info *scan_info_array;
2704 ZebraScanEntry *glist;
2705 int ords[32], ord_no = 0;
2708 int bases_ok = 0; /* no of databases with OK attribute */
2709 int errCode = 0; /* err code (if any is not OK) */
2710 char *errString = 0; /* addinfo */
2713 char *search_type = NULL;
2714 char rank_type[128];
2720 if (attributeset == VAL_NONE)
2721 attributeset = VAL_BIB1;
2726 int termset_value_numeric;
2727 const char *termset_value_string;
2728 attr_init (&termset, zapt, 8);
2729 termset_value_numeric =
2730 attr_find_ex (&termset, NULL, &termset_value_string);
2731 if (termset_value_numeric != -1)
2734 const char *termset_name = 0;
2736 if (termset_value_numeric != -2)
2739 sprintf (resname, "%d", termset_value_numeric);
2740 termset_name = resname;
2743 termset_name = termset_value_string;
2745 limit_set = resultSetRef (zh, termset_name);
2749 yaz_log (LOG_DEBUG, "position = %d, num = %d set=%d",
2750 pos, num, attributeset);
2752 attr_init (&use, zapt, 1);
2753 use_value = attr_find_ex (&use, &attributeset, &use_string);
2755 if (zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type,
2756 rank_type, &complete_flag, &sort_flag))
2762 yaz_log (LOG_DEBUG, "use_value = %d", use_value);
2764 if (use_value == -1)
2766 for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
2770 data1_local_attribute *local_attr;
2772 if ((r=att_getentbyatt (zh, &attp, attributeset, use_value,
2775 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
2776 attributeset, use_value);
2780 sprintf (val_str, "%d", use_value);
2782 errString = odr_strdup (stream, val_str);
2788 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2790 zh->errString = basenames[base_no];
2791 zh->errCode = 109; /* Database unavailable */
2796 for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
2797 local_attr = local_attr->next)
2801 ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal,
2804 ords[ord_no++] = ord;
2807 if (!bases_ok && errCode)
2809 zh->errCode = errCode;
2810 zh->errString = errString;
2816 sprintf (val_str, "%d", use_value);
2818 zh->errString = odr_strdup (stream, val_str);
2823 /* prepare dictionary scanning */
2826 scan_info_array = (struct scan_info *)
2827 odr_malloc (stream, ord_no * sizeof(*scan_info_array));
2828 for (i = 0; i < ord_no; i++)
2830 int j, prefix_len = 0;
2831 int before_tmp = before, after_tmp = after;
2832 struct scan_info *scan_info = scan_info_array + i;
2833 struct rpn_char_map_info rcmi;
2835 rpn_char_map_prepare (zh->reg, reg_id, &rcmi);
2837 scan_info->before = before;
2838 scan_info->after = after;
2839 scan_info->odr = stream;
2841 scan_info->list = (struct scan_info_entry *)
2842 odr_malloc (stream, (before+after) * sizeof(*scan_info->list));
2843 for (j = 0; j<before+after; j++)
2844 scan_info->list[j].term = NULL;
2846 prefix_len += key_SU_encode (ords[i], termz + prefix_len);
2847 termz[prefix_len++] = reg_id;
2848 termz[prefix_len] = 0;
2849 strcpy (scan_info->prefix, termz);
2851 if (trans_scan_term (zh, zapt, termz+prefix_len, reg_id))
2854 dict_scan (zh->reg->dict, termz, &before_tmp, &after_tmp,
2855 scan_info, scan_handle);
2857 glist = (ZebraScanEntry *)
2858 odr_malloc (stream, (before+after)*sizeof(*glist));
2860 /* consider terms after main term */
2861 for (i = 0; i < ord_no; i++)
2865 for (i = 0; i<after; i++)
2868 const char *mterm = NULL;
2872 for (j = 0; j < ord_no; j++)
2874 if (ptr[j] < before+after &&
2875 (tst=scan_info_array[j].list[ptr[j]].term) &&
2876 (!mterm || strcmp (tst, mterm) < 0))
2884 scan_term_untrans (zh, stream->mem, reg_id,
2885 &glist[i+before].term, mterm);
2886 rset = rset_trunc (zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
2887 glist[i+before].term, strlen(glist[i+before].term),
2888 NULL, 0, zapt->term->which);
2891 for (j = j0+1; j<ord_no; j++)
2893 if (ptr[j] < before+after &&
2894 (tst=scan_info_array[j].list[ptr[j]].term) &&
2895 !strcmp (tst, mterm))
2897 rset_bool_parms bool_parms;
2901 rset_trunc (zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
2902 glist[i+before].term,
2903 strlen(glist[i+before].term), NULL, 0,
2906 bool_parms.key_size = sizeof(struct it_key);
2907 bool_parms.cmp = key_compare_it;
2908 bool_parms.rset_l = rset;
2909 bool_parms.rset_r = rset2;
2911 rset = rset_create (rset_kind_or, &bool_parms);
2918 rset_bool_parms bool_parms;
2920 bool_parms.key_size = sizeof(struct it_key);
2921 bool_parms.cmp = key_compare_it;
2922 bool_parms.rset_l = rset;
2923 bool_parms.rset_r = rset_dup(limit_set);
2925 rset = rset_create (rset_kind_and, &bool_parms);
2927 count_set (rset, &glist[i+before].occurrences);
2932 *num_entries -= (after-i);
2936 /* consider terms before main term */
2937 for (i = 0; i<ord_no; i++)
2940 for (i = 0; i<before; i++)
2943 const char *mterm = NULL;
2947 for (j = 0; j <ord_no; j++)
2949 if (ptr[j] < before &&
2950 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
2951 (!mterm || strcmp (tst, mterm) > 0))
2960 scan_term_untrans (zh, stream->mem, reg_id,
2961 &glist[before-1-i].term, mterm);
2964 (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
2965 glist[before-1-i].term, strlen(glist[before-1-i].term),
2966 NULL, 0, zapt->term->which);
2970 for (j = j0+1; j<ord_no; j++)
2972 if (ptr[j] < before &&
2973 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
2974 !strcmp (tst, mterm))
2976 rset_bool_parms bool_parms;
2979 rset2 = rset_trunc (zh,
2980 &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
2981 glist[before-1-i].term,
2982 strlen(glist[before-1-i].term), NULL, 0,
2985 bool_parms.key_size = sizeof(struct it_key);
2986 bool_parms.cmp = key_compare_it;
2987 bool_parms.rset_l = rset;
2988 bool_parms.rset_r = rset2;
2990 rset = rset_create (rset_kind_or, &bool_parms);
2997 rset_bool_parms bool_parms;
2999 bool_parms.key_size = sizeof(struct it_key);
3000 bool_parms.cmp = key_compare_it;
3001 bool_parms.rset_l = rset;
3002 bool_parms.rset_r = rset_dup(limit_set);
3004 rset = rset_create (rset_kind_and, &bool_parms);
3006 count_set (rset, &glist[before-1-i].occurrences);
3016 *list = glist + i; /* list is set to first 'real' entry */
3018 logf (LOG_DEBUG, "position = %d, num_entries = %d",
3019 *position, *num_entries);
3021 logf (LOG_DEBUG, "scan error: %d", zh->errCode);