1 /* $Id: zrpn.c,v 1.141.2.3 2004-11-15 22:52:48 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>
43 struct rpn_char_map_info {
52 Z_AttributesPlusTerm *zapt;
56 static const char **rpn_char_map_handler(void *vp, const char **from, int len)
58 struct rpn_char_map_info *p = (struct rpn_char_map_info *) vp;
59 const char **out = zebra_maps_input(p->zm, p->reg_type, from, len, 0);
63 const char *outp = *out;
64 yaz_log(LOG_LOG, "---");
67 yaz_log(LOG_LOG, "%02X", *outp);
75 static void rpn_char_map_prepare(struct zebra_register *reg, int reg_type,
76 struct rpn_char_map_info *map_info)
78 map_info->zm = reg->zebra_maps;
79 map_info->reg_type = reg_type;
80 dict_grep_cmap(reg->dict, map_info, rpn_char_map_handler);
83 static int attr_find_ex(AttrType *src, oid_value *attributeSetP,
84 const char **string_value)
88 num_attributes = src->zapt->attributes->num_attributes;
89 while (src->major < num_attributes)
91 Z_AttributeElement *element;
93 element = src->zapt->attributes->attributes[src->major];
94 if (src->type == *element->attributeType)
96 switch (element->which)
98 case Z_AttributeValue_numeric:
100 if (element->attributeSet && attributeSetP)
104 attrset = oid_getentbyoid(element->attributeSet);
105 *attributeSetP = attrset->value;
107 return *element->value.numeric;
109 case Z_AttributeValue_complex:
110 if (src->minor >= element->value.complex->num_list)
112 if (element->attributeSet && attributeSetP)
116 attrset = oid_getentbyoid(element->attributeSet);
117 *attributeSetP = attrset->value;
119 if (element->value.complex->list[src->minor]->which ==
120 Z_StringOrNumeric_numeric)
124 *element->value.complex->list[src->minor-1]->u.numeric;
126 else if (element->value.complex->list[src->minor]->which ==
127 Z_StringOrNumeric_string)
133 element->value.complex->list[src->minor-1]->u.string;
147 static int attr_find(AttrType *src, oid_value *attributeSetP)
149 return attr_find_ex(src, attributeSetP, 0);
152 static void attr_init(AttrType *src, Z_AttributesPlusTerm *zapt,
175 static void term_untrans(ZebraHandle zh, int reg_type,
176 char *dst, const char *src)
181 const char *cp = zebra_maps_output(zh->reg->zebra_maps,
183 if (!cp && len < IT_MAX_WORD-1)
186 while (*cp && len < IT_MAX_WORD-1)
192 static void add_isam_p(const char *name, const char *info,
195 if (p->isam_p_indx == p->isam_p_size)
197 ISAMS_P *new_isam_p_buf;
201 p->isam_p_size = 2*p->isam_p_size + 100;
202 new_isam_p_buf = (ISAMS_P *) xmalloc(sizeof(*new_isam_p_buf) *
206 memcpy(new_isam_p_buf, p->isam_p_buf,
207 p->isam_p_indx * sizeof(*p->isam_p_buf));
208 xfree(p->isam_p_buf);
210 p->isam_p_buf = new_isam_p_buf;
213 new_term_no = (int *) xmalloc(sizeof(*new_term_no) *
217 memcpy(new_term_no, p->isam_p_buf,
218 p->isam_p_indx * sizeof(*p->term_no));
221 p->term_no = new_term_no;
224 assert(*info == sizeof(*p->isam_p_buf));
225 memcpy(p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
232 char term_tmp[IT_MAX_WORD];
234 int len = key_SU_decode(&su_code, name);
236 term_untrans(p->zh, p->reg_type, term_tmp, name+len+1);
237 yaz_log(LOG_LOG, "grep: %d %c %s", su_code, name[len], term_tmp);
238 zebraExplain_lookup_ord(p->zh->reg->zei,
239 su_code, &db, &set, &use);
240 yaz_log(LOG_LOG, "grep: set=%d use=%d db=%s", set, use, db);
242 resultSetAddTerm(p->zh, p->termset, name[len], db,
249 static int grep_handle(char *name, const char *info, void *p)
251 add_isam_p(name, info, (struct grep_info *) p);
255 static int term_pre(ZebraMaps zebra_maps, int reg_type, const char **src,
256 const char *ct1, const char *ct2, int first)
258 const char *s1, *s0 = *src;
261 /* skip white space */
264 if (ct1 && strchr(ct1, *s0))
266 if (ct2 && strchr(ct2, *s0))
269 map = zebra_maps_input(zebra_maps, reg_type, &s1, strlen(s1), first);
270 if (**map != *CHR_SPACE)
278 #define REGEX_CHARS " []()|.*+?!"
280 /* term_100: handle term, where trunc=none(no operators at all) */
281 static int term_100(ZebraMaps zebra_maps, int reg_type,
282 const char **src, char *dst, int space_split,
290 const char *space_start = 0;
291 const char *space_end = 0;
293 if (!term_pre(zebra_maps, reg_type, src, NULL, NULL, !space_split))
299 map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0);
302 if (**map == *CHR_SPACE)
305 else /* complete subfield only. */
307 if (**map == *CHR_SPACE)
308 { /* save space mapping for later .. */
313 else if (space_start)
314 { /* reload last space */
315 while (space_start < space_end)
317 if (strchr(REGEX_CHARS, *space_start))
319 dst_term[j++] = *space_start;
320 dst[i++] = *space_start++;
323 space_start = space_end = 0;
326 /* add non-space char */
329 if (strchr(REGEX_CHARS, *s1))
341 /* term_101: handle term, where trunc=Process # */
342 static int term_101(ZebraMaps zebra_maps, int reg_type,
343 const char **src, char *dst, int space_split,
351 if (!term_pre(zebra_maps, reg_type, src, "#", "#", !space_split))
360 dst_term[j++] = *s0++;
365 map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0);
366 if (space_split && **map == *CHR_SPACE)
370 if (strchr(REGEX_CHARS, *s1))
378 dst_term[j++] = '\0';
383 /* term_103: handle term, where trunc=re-2 (regular expressions) */
384 static int term_103(ZebraMaps zebra_maps, int reg_type, const char **src,
385 char *dst, int *errors, int space_split,
393 if (!term_pre(zebra_maps, reg_type, src, "^\\()[].*+?|", "(", !space_split))
396 if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
399 *errors = s0[1] - '0';
406 if (strchr("^\\()[].*+?|-", *s0))
414 map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0);
415 if (**map == *CHR_SPACE)
419 if (strchr(REGEX_CHARS, *s1))
432 /* term_103: handle term, where trunc=re-1 (regular expressions) */
433 static int term_102 (ZebraMaps zebra_maps, int reg_type, const char **src,
434 char *dst, int space_split, char *dst_term)
436 return term_103(zebra_maps, reg_type, src, dst, NULL, space_split,
441 /* term_104: handle term, where trunc=Process # and ! */
442 static int term_104(ZebraMaps zebra_maps, int reg_type,
443 const char **src, char *dst, int space_split,
451 if (!term_pre(zebra_maps, reg_type, src, "?*#", "?*#", !space_split))
458 dst_term[j++] = *s0++;
459 if (*s0 >= '0' && *s0 <= '9')
462 while (*s0 >= '0' && *s0 <= '9')
464 limit = limit * 10 + (*s0 - '0');
465 dst_term[j++] = *s0++;
485 dst_term[j++] = *s0++;
490 dst_term[j++] = *s0++;
494 map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0);
495 if (space_split && **map == *CHR_SPACE)
499 if (strchr(REGEX_CHARS, *s1))
507 dst_term[j++] = '\0';
512 /* term_105/106: handle term, where trunc=Process * and ! and right trunc */
513 static int term_105 (ZebraMaps zebra_maps, int reg_type,
514 const char **src, char *dst, int space_split,
515 char *dst_term, int right_truncate)
522 if (!term_pre(zebra_maps, reg_type, src, "*!", "*!", !space_split))
531 dst_term[j++] = *s0++;
536 dst_term[j++] = *s0++;
540 map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0);
541 if (space_split && **map == *CHR_SPACE)
545 if (strchr(REGEX_CHARS, *s1))
559 dst_term[j++] = '\0';
565 /* gen_regular_rel - generate regular expression from relation
566 * val: border value (inclusive)
567 * islt: 1 if <=; 0 if >=.
569 static void gen_regular_rel(char *dst, int val, int islt)
576 yaz_log(LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt);
580 strcpy(dst, "(-[0-9]+|(");
588 strcpy(dst, "([0-9]+|-(");
600 sprintf(numstr, "%d", val);
601 for (w = strlen(numstr); --w >= 0; pos++)
620 strcpy(dst + dst_p, numstr);
621 dst_p = strlen(dst) - pos - 1;
649 for (i = 0; i<pos; i++)
662 /* match everything less than 10^(pos-1) */
664 for (i = 1; i<pos; i++)
665 strcat(dst, "[0-9]?");
669 /* match everything greater than 10^pos */
670 for (i = 0; i <= pos; i++)
671 strcat(dst, "[0-9]");
672 strcat(dst, "[0-9]*");
677 void string_rel_add_char(char **term_p, const char *src, int *indx)
679 if (src[*indx] == '\\')
680 *(*term_p)++ = src[(*indx)++];
681 *(*term_p)++ = src[(*indx)++];
685 * > abc ([b-].*|a[c-].*|ab[d-].*|abc.+)
686 * ([^-a].*|a[^-b].*ab[^-c].*|abc.+)
687 * >= abc ([b-].*|a[c-].*|ab[c-].*)
688 * ([^-a].*|a[^-b].*|ab[c-].*)
689 * < abc ([-0].*|a[-a].*|ab[-b].*)
690 * ([^a-].*|a[^b-].*|ab[^c-].*)
691 * <= abc ([-0].*|a[-a].*|ab[-b].*|abc)
692 * ([^a-].*|a[^b-].*|ab[^c-].*|abc)
694 static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
695 const char **term_sub, char *term_dict,
696 oid_value attributeSet,
697 int reg_type, int space_split, char *term_dst)
702 char *term_tmp = term_dict + strlen(term_dict);
703 char term_component[2*IT_MAX_WORD+20];
705 attr_init(&relation, zapt, 2);
706 relation_value = attr_find(&relation, NULL);
708 yaz_log(LOG_DEBUG, "string relation value=%d", relation_value);
709 switch (relation_value)
712 if (!term_100 (zh->reg->zebra_maps, reg_type,
713 term_sub, term_component,
714 space_split, term_dst))
716 yaz_log(LOG_DEBUG, "Relation <");
719 for (i = 0; term_component[i]; )
726 string_rel_add_char (&term_tmp, term_component, &j);
731 string_rel_add_char (&term_tmp, term_component, &i);
738 if ((term_tmp - term_dict) > IT_MAX_WORD)
745 if (!term_100 (zh->reg->zebra_maps, reg_type,
746 term_sub, term_component,
747 space_split, term_dst))
749 yaz_log(LOG_DEBUG, "Relation <=");
752 for (i = 0; term_component[i]; )
757 string_rel_add_char (&term_tmp, term_component, &j);
761 string_rel_add_char (&term_tmp, term_component, &i);
770 if ((term_tmp - term_dict) > IT_MAX_WORD)
773 for (i = 0; term_component[i]; )
774 string_rel_add_char (&term_tmp, term_component, &i);
779 if (!term_100 (zh->reg->zebra_maps, reg_type,
780 term_sub, term_component, space_split, term_dst))
782 yaz_log(LOG_DEBUG, "Relation >");
785 for (i = 0; term_component[i];)
790 string_rel_add_char (&term_tmp, term_component, &j);
795 string_rel_add_char (&term_tmp, term_component, &i);
803 if ((term_tmp - term_dict) > IT_MAX_WORD)
806 for (i = 0; term_component[i];)
807 string_rel_add_char (&term_tmp, term_component, &i);
814 if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub,
815 term_component, space_split, term_dst))
817 yaz_log(LOG_DEBUG, "Relation >=");
820 for (i = 0; term_component[i];)
827 string_rel_add_char (&term_tmp, term_component, &j);
830 if (term_component[i+1])
834 string_rel_add_char (&term_tmp, term_component, &i);
838 string_rel_add_char (&term_tmp, term_component, &i);
845 if ((term_tmp - term_dict) > IT_MAX_WORD)
853 yaz_log(LOG_DEBUG, "Relation =");
854 if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub,
855 term_component, space_split, term_dst))
857 strcat (term_tmp, "(");
858 strcat (term_tmp, term_component);
859 strcat (term_tmp, ")");
864 static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
865 const char **term_sub,
866 oid_value attributeSet, NMEM stream,
867 struct grep_info *grep_info,
868 int reg_type, int complete_flag,
869 int num_bases, char **basenames,
870 char *term_dst, int xpath_use);
872 static RSET term_trunc (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
873 const char **term_sub,
874 oid_value attributeSet, NMEM stream,
875 struct grep_info *grep_info,
876 int reg_type, int complete_flag,
877 int num_bases, char **basenames,
879 const char *rank_type, int xpath_use)
882 grep_info->isam_p_indx = 0;
883 r = string_term (zh, zapt, term_sub, attributeSet, stream, grep_info,
884 reg_type, complete_flag, num_bases, basenames,
885 term_dst, xpath_use);
888 yaz_log(LOG_DEBUG, "term: %s", term_dst);
889 return rset_trunc (zh, grep_info->isam_p_buf,
890 grep_info->isam_p_indx, term_dst,
891 strlen(term_dst), rank_type, 1 /* preserve pos */,
895 static char *nmem_strdup_i(NMEM nmem, int v)
898 sprintf (val_str, "%d", v);
899 return nmem_strdup (nmem, val_str);
902 static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
903 const char **term_sub,
904 oid_value attributeSet, NMEM stream,
905 struct grep_info *grep_info,
906 int reg_type, int complete_flag,
907 int num_bases, char **basenames,
908 char *term_dst, int xpath_use)
910 char term_dict[2*IT_MAX_WORD+4000];
913 int truncation_value;
916 const char *use_string = 0;
917 oid_value curAttributeSet = attributeSet;
919 struct rpn_char_map_info rcmi;
920 int space_split = complete_flag ? 0 : 1;
922 int bases_ok = 0; /* no of databases with OK attribute */
923 int errCode = 0; /* err code (if any is not OK) */
924 char *errString = 0; /* addinfo */
926 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
927 attr_init (&use, zapt, 1);
928 use_value = attr_find_ex (&use, &curAttributeSet, &use_string);
929 yaz_log(LOG_DEBUG, "string_term, use value %d", use_value);
930 attr_init (&truncation, zapt, 5);
931 truncation_value = attr_find (&truncation, NULL);
932 yaz_log(LOG_DEBUG, "truncation value %d", truncation_value);
934 if (use_value == -1) /* no attribute - assumy "any" */
936 for (base_no = 0; base_no < num_bases; base_no++)
942 data1_local_attribute id_xpath_attr;
943 data1_local_attribute *local_attr;
944 int max_pos, prefix_len = 0;
948 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
950 zh->errCode = 109; /* Database unavailable */
951 zh->errString = basenames[base_no];
954 if (xpath_use > 0 && use_value == -2)
956 use_value = xpath_use;
957 attp.local_attributes = &id_xpath_attr;
958 attp.attset_ordinal = VAL_IDXPATH;
959 id_xpath_attr.next = 0;
960 id_xpath_attr.local = use_value;
962 else if (curAttributeSet == VAL_IDXPATH)
964 attp.local_attributes = &id_xpath_attr;
965 attp.attset_ordinal = VAL_IDXPATH;
966 id_xpath_attr.next = 0;
967 id_xpath_attr.local = use_value;
971 if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value,
974 yaz_log(LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
975 curAttributeSet, use_value, r);
978 /* set was found, but value wasn't defined */
981 errString = nmem_strdup(stream, use_string);
983 errString = nmem_strdup_i(stream, use_value);
988 struct oident oident;
990 oident.proto = PROTO_Z3950;
991 oident.oclass = CLASS_ATTSET;
992 oident.value = curAttributeSet;
993 oid_ent_to_oid (&oident, oid);
996 errString = nmem_strdup (stream, oident.desc);
1001 for (local_attr = attp.local_attributes; local_attr;
1002 local_attr = local_attr->next)
1008 ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal,
1013 term_dict[prefix_len++] = '|';
1015 term_dict[prefix_len++] = '(';
1017 ord_len = key_SU_encode (ord, ord_buf);
1018 for (i = 0; i<ord_len; i++)
1020 term_dict[prefix_len++] = 1;
1021 term_dict[prefix_len++] = ord_buf[i];
1030 errString = nmem_strdup_i(stream, use_value);
1037 bases_ok++; /* this has OK attributes */
1040 term_dict[prefix_len++] = ')';
1041 term_dict[prefix_len++] = 1;
1042 term_dict[prefix_len++] = reg_type;
1043 yaz_log(LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
1044 term_dict[prefix_len] = '\0';
1046 switch (truncation_value)
1048 case -1: /* not specified */
1049 case 100: /* do not truncate */
1050 if (!string_relation (zh, zapt, &termp, term_dict,
1052 reg_type, space_split, term_dst))
1055 case 1: /* right truncation */
1056 term_dict[j++] = '(';
1057 if (!term_100 (zh->reg->zebra_maps, reg_type,
1058 &termp, term_dict + j, space_split, term_dst))
1060 strcat (term_dict, ".*)");
1062 case 2: /* keft truncation */
1063 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1064 if (!term_100 (zh->reg->zebra_maps, reg_type,
1065 &termp, term_dict + j, space_split, term_dst))
1067 strcat (term_dict, ")");
1069 case 3: /* left&right truncation */
1070 term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*';
1071 if (!term_100 (zh->reg->zebra_maps, reg_type,
1072 &termp, term_dict + j, space_split, term_dst))
1074 strcat (term_dict, ".*)");
1076 case 101: /* process # in term */
1077 term_dict[j++] = '(';
1078 if (!term_101 (zh->reg->zebra_maps, reg_type,
1079 &termp, term_dict + j, space_split, term_dst))
1081 strcat (term_dict, ")");
1083 case 102: /* Regexp-1 */
1084 term_dict[j++] = '(';
1085 if (!term_102 (zh->reg->zebra_maps, reg_type,
1086 &termp, term_dict + j, space_split, term_dst))
1088 strcat (term_dict, ")");
1090 case 103: /* Regexp-2 */
1092 term_dict[j++] = '(';
1094 if (!term_103 (zh->reg->zebra_maps, reg_type,
1095 &termp, term_dict + j, ®ex_range,
1096 space_split, term_dst))
1098 strcat (term_dict, ")");
1100 case 104: /* process # and ! in term */
1101 term_dict[j++] = '(';
1102 if (!term_104 (zh->reg->zebra_maps, reg_type,
1103 &termp, term_dict + j, space_split, term_dst))
1105 strcat (term_dict, ")");
1107 case 105: /* process * and ! in term */
1108 term_dict[j++] = '(';
1109 if (!term_105 (zh->reg->zebra_maps, reg_type,
1110 &termp, term_dict + j, space_split, term_dst, 1))
1112 strcat (term_dict, ")");
1114 case 106: /* process * and ! in term */
1115 term_dict[j++] = '(';
1116 if (!term_105 (zh->reg->zebra_maps, reg_type,
1117 &termp, term_dict + j, space_split, term_dst, 0))
1119 strcat (term_dict, ")");
1123 zh->errString = nmem_strdup_i(stream, truncation_value);
1128 yaz_log(LOG_DEBUG, "dict_lookup_grep: %s", term_dict+prefix_len);
1129 r = dict_lookup_grep (zh->reg->dict, term_dict, regex_range,
1130 grep_info, &max_pos, init_pos,
1133 yaz_log(LOG_WARN, "dict_lookup_grep fail %d", r);
1138 zh->errCode = errCode;
1139 zh->errString = errString;
1143 yaz_log(LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1148 /* convert APT search term to UTF8 */
1149 static int zapt_term_to_utf8 (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1153 Z_Term *term = zapt->term;
1155 switch (term->which)
1157 case Z_Term_general:
1158 if (zh->iconv_to_utf8 != 0)
1160 char *inbuf = term->u.general->buf;
1161 size_t inleft = term->u.general->len;
1162 char *outbuf = termz;
1163 size_t outleft = IT_MAX_WORD-1;
1166 ret = yaz_iconv(zh->iconv_to_utf8, &inbuf, &inleft,
1168 if (ret == (size_t)(-1))
1170 ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0);
1178 sizez = term->u.general->len;
1179 if (sizez > IT_MAX_WORD-1)
1180 sizez = IT_MAX_WORD-1;
1181 memcpy (termz, term->u.general->buf, sizez);
1182 termz[sizez] = '\0';
1185 case Z_Term_characterString:
1186 sizez = strlen(term->u.characterString);
1187 if (sizez > IT_MAX_WORD-1)
1188 sizez = IT_MAX_WORD-1;
1189 memcpy (termz, term->u.characterString, sizez);
1190 termz[sizez] = '\0';
1199 /* convert APT SCAN term to internal cmap */
1200 static int trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1201 char *termz, int reg_type)
1203 char termz0[IT_MAX_WORD];
1205 if (zapt_term_to_utf8(zh, zapt, termz0))
1206 return -1; /* error */
1210 const char *cp = (const char *) termz0;
1211 const char *cp_end = cp + strlen(cp);
1214 const char *space_map = NULL;
1217 while ((len = (cp_end - cp)) > 0)
1219 map = zebra_maps_input (zh->reg->zebra_maps, reg_type, &cp, len, 0);
1220 if (**map == *CHR_SPACE)
1225 for (src = space_map; *src; src++)
1228 for (src = *map; *src; src++)
1237 char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1238 const char *termz, NMEM stream, unsigned reg_id)
1241 AttrType truncation;
1242 int truncation_value;
1245 attr_init (&truncation, zapt, 5);
1246 truncation_value = attr_find (&truncation, NULL);
1248 switch (truncation_value)
1268 wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list,
1269 termz, strlen(termz));
1271 return nmem_strdup(stream, termz);
1274 char *buf = (char*) nmem_malloc (stream, wrbuf_len(wrbuf)+1);
1275 memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
1276 buf[wrbuf_len(wrbuf)] = '\0';
1281 static void grep_info_delete (struct grep_info *grep_info)
1284 xfree(grep_info->term_no);
1286 xfree (grep_info->isam_p_buf);
1289 static int grep_info_prepare (ZebraHandle zh,
1290 Z_AttributesPlusTerm *zapt,
1291 struct grep_info *grep_info,
1296 int termset_value_numeric;
1297 const char *termset_value_string;
1300 grep_info->term_no = 0;
1302 grep_info->isam_p_size = 0;
1303 grep_info->isam_p_buf = NULL;
1305 grep_info->reg_type = reg_type;
1306 grep_info->termset = 0;
1310 attr_init (&termset, zapt, 8);
1311 termset_value_numeric =
1312 attr_find_ex (&termset, NULL, &termset_value_string);
1313 if (termset_value_numeric != -1)
1316 const char *termset_name = 0;
1317 if (termset_value_numeric != -2)
1319 sprintf (resname, "%d", termset_value_numeric);
1320 termset_name = resname;
1323 termset_name = termset_value_string;
1324 yaz_log(LOG_LOG, "creating termset set %s", termset_name);
1325 grep_info->termset = resultSetAdd (zh, termset_name, 1);
1326 if (!grep_info->termset)
1329 zh->errString = nmem_strdup (stream, termset_name);
1337 static RSET rpn_search_APT_phrase (ZebraHandle zh,
1338 Z_AttributesPlusTerm *zapt,
1339 const char *termz_org,
1340 oid_value attributeSet,
1342 int reg_type, int complete_flag,
1343 const char *rank_type, int xpath_use,
1344 int num_bases, char **basenames)
1346 char term_dst[IT_MAX_WORD+1];
1347 RSET rset[60], result;
1349 struct grep_info grep_info;
1350 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1351 const char *termp = termz;
1354 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1358 yaz_log(LOG_DEBUG, "APT_phrase termp=%s", termp);
1359 rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet,
1361 reg_type, complete_flag,
1362 num_bases, basenames,
1363 term_dst, rank_type,
1367 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1370 grep_info_delete (&grep_info);
1373 rset_null_parms parms;
1375 parms.rset_term = rset_term_create (termz, -1, rank_type,
1377 return rset_create (rset_kind_null, &parms);
1379 else if (rset_no == 1)
1383 /* new / old prox */
1384 rset_prox_parms parms;
1387 parms.rset_no = rset_no;
1389 parms.exclusion = 0;
1392 parms.key_size = sizeof(struct it_key);
1393 parms.cmp = key_compare_it;
1394 parms.getseq = key_get_seq;
1395 parms.log_item = key_logdump_txt;
1396 result = rset_create(rset_kind_prox, &parms);
1401 static RSET rpn_search_APT_or_list (ZebraHandle zh,
1402 Z_AttributesPlusTerm *zapt,
1403 const char *termz_org,
1404 oid_value attributeSet,
1406 int reg_type, int complete_flag,
1407 const char *rank_type,
1409 int num_bases, char **basenames)
1411 char term_dst[IT_MAX_WORD+1];
1412 RSET rset[60], result;
1414 struct grep_info grep_info;
1415 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1416 const char *termp = termz;
1418 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1422 yaz_log(LOG_DEBUG, "APT_or_list termp=%s", termp);
1423 rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet,
1425 reg_type, complete_flag,
1426 num_bases, basenames,
1427 term_dst, rank_type,
1431 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1434 grep_info_delete (&grep_info);
1437 rset_null_parms parms;
1439 parms.rset_term = rset_term_create (termz, -1, rank_type,
1441 return rset_create (rset_kind_null, &parms);
1444 for (i = 1; i<rset_no; i++)
1446 rset_bool_parms bool_parms;
1448 bool_parms.rset_l = result;
1449 bool_parms.rset_r = rset[i];
1450 bool_parms.key_size = sizeof(struct it_key);
1451 bool_parms.cmp = key_compare_it;
1452 bool_parms.log_item = key_logdump_txt;
1453 result = rset_create (rset_kind_or, &bool_parms);
1458 static RSET rpn_search_APT_and_list (ZebraHandle zh,
1459 Z_AttributesPlusTerm *zapt,
1460 const char *termz_org,
1461 oid_value attributeSet,
1463 int reg_type, int complete_flag,
1464 const char *rank_type,
1466 int num_bases, char **basenames)
1468 char term_dst[IT_MAX_WORD+1];
1469 RSET rset[60], result;
1471 struct grep_info grep_info;
1472 char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
1473 const char *termp = termz;
1475 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1479 yaz_log(LOG_DEBUG, "APT_and_list termp=%s", termp);
1480 rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet,
1482 reg_type, complete_flag,
1483 num_bases, basenames,
1484 term_dst, rank_type,
1488 assert (rset[rset_no]);
1489 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1492 grep_info_delete (&grep_info);
1495 rset_null_parms parms;
1497 parms.rset_term = rset_term_create (termz, -1, rank_type,
1499 return rset_create (rset_kind_null, &parms);
1502 for (i = 1; i<rset_no; i++)
1504 rset_bool_parms bool_parms;
1506 bool_parms.rset_l = result;
1507 bool_parms.rset_r = rset[i];
1508 bool_parms.key_size = sizeof(struct it_key);
1509 bool_parms.cmp = key_compare_it;
1510 bool_parms.log_item = key_logdump_txt;
1511 result = rset_create (rset_kind_and, &bool_parms);
1516 static int numeric_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1517 const char **term_sub,
1519 oid_value attributeSet,
1520 struct grep_info *grep_info,
1529 char *term_tmp = term_dict + strlen(term_dict);
1531 attr_init (&relation, zapt, 2);
1532 relation_value = attr_find (&relation, NULL);
1534 yaz_log(LOG_DEBUG, "numeric relation value=%d", relation_value);
1536 if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1,
1539 term_value = atoi (term_tmp);
1540 switch (relation_value)
1543 yaz_log(LOG_DEBUG, "Relation <");
1544 gen_regular_rel (term_tmp, term_value-1, 1);
1547 yaz_log(LOG_DEBUG, "Relation <=");
1548 gen_regular_rel (term_tmp, term_value, 1);
1551 yaz_log(LOG_DEBUG, "Relation >=");
1552 gen_regular_rel (term_tmp, term_value, 0);
1555 yaz_log(LOG_DEBUG, "Relation >");
1556 gen_regular_rel (term_tmp, term_value+1, 0);
1560 yaz_log(LOG_DEBUG, "Relation =");
1561 sprintf (term_tmp, "(0*%d)", term_value);
1563 yaz_log(LOG_DEBUG, "dict_lookup_grep: %s", term_tmp);
1564 r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, max_pos,
1567 yaz_log(LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
1568 yaz_log(LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1572 static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1573 const char **term_sub,
1574 oid_value attributeSet, struct grep_info *grep_info,
1575 int reg_type, int complete_flag,
1576 int num_bases, char **basenames,
1577 char *term_dst, int xpath_use, NMEM stream)
1579 char term_dict[2*IT_MAX_WORD+2];
1583 const char *use_string = 0;
1584 oid_value curAttributeSet = attributeSet;
1586 struct rpn_char_map_info rcmi;
1588 int bases_ok = 0; /* no of databases with OK attribute */
1589 int errCode = 0; /* err code (if any is not OK) */
1590 char *errString = 0; /* addinfo */
1592 rpn_char_map_prepare (zh->reg, reg_type, &rcmi);
1593 attr_init (&use, zapt, 1);
1594 use_value = attr_find_ex (&use, &curAttributeSet, &use_string);
1596 if (use_value == -1)
1599 for (base_no = 0; base_no < num_bases; base_no++)
1602 data1_local_attribute id_xpath_attr;
1603 data1_local_attribute *local_attr;
1604 int max_pos, prefix_len = 0;
1607 if (use_value == -2) /* string attribute (assume IDXPATH/any) */
1609 use_value = xpath_use;
1610 attp.local_attributes = &id_xpath_attr;
1611 attp.attset_ordinal = VAL_IDXPATH;
1612 id_xpath_attr.next = 0;
1613 id_xpath_attr.local = use_value;
1615 else if (curAttributeSet == VAL_IDXPATH)
1617 attp.local_attributes = &id_xpath_attr;
1618 attp.attset_ordinal = VAL_IDXPATH;
1619 id_xpath_attr.next = 0;
1620 id_xpath_attr.local = use_value;
1624 if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value,
1627 yaz_log(LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
1628 curAttributeSet, use_value, r);
1631 errString = nmem_strdup_i (stream, use_value);
1639 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
1641 zh->errCode = 109; /* Database unavailable */
1642 zh->errString = basenames[base_no];
1645 for (local_attr = attp.local_attributes; local_attr;
1646 local_attr = local_attr->next)
1652 ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal,
1657 term_dict[prefix_len++] = '|';
1659 term_dict[prefix_len++] = '(';
1661 ord_len = key_SU_encode (ord, ord_buf);
1662 for (i = 0; i<ord_len; i++)
1664 term_dict[prefix_len++] = 1;
1665 term_dict[prefix_len++] = ord_buf[i];
1671 sprintf (val_str, "%d", use_value);
1673 errString = nmem_strdup (stream, val_str);
1677 term_dict[prefix_len++] = ')';
1678 term_dict[prefix_len++] = 1;
1679 term_dict[prefix_len++] = reg_type;
1680 yaz_log(LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
1681 term_dict[prefix_len] = '\0';
1682 if (!numeric_relation (zh, zapt, &termp, term_dict,
1683 attributeSet, grep_info, &max_pos, reg_type,
1689 zh->errCode = errCode;
1690 zh->errString = errString;
1694 yaz_log(LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
1698 static RSET rpn_search_APT_numeric (ZebraHandle zh,
1699 Z_AttributesPlusTerm *zapt,
1701 oid_value attributeSet,
1703 int reg_type, int complete_flag,
1704 const char *rank_type, int xpath_use,
1705 int num_bases, char **basenames)
1707 char term_dst[IT_MAX_WORD+1];
1708 const char *termp = termz;
1709 RSET rset[60], result;
1710 int i, r, rset_no = 0;
1711 struct grep_info grep_info;
1713 if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream))
1717 yaz_log(LOG_DEBUG, "APT_numeric termp=%s", termp);
1718 grep_info.isam_p_indx = 0;
1719 r = numeric_term (zh, zapt, &termp, attributeSet, &grep_info,
1720 reg_type, complete_flag, num_bases, basenames,
1721 term_dst, xpath_use,
1725 yaz_log(LOG_DEBUG, "term: %s", term_dst);
1726 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1727 grep_info.isam_p_indx, term_dst,
1728 strlen(term_dst), rank_type,
1729 0 /* preserve position */,
1731 assert (rset[rset_no]);
1732 if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset)))
1735 grep_info_delete (&grep_info);
1738 rset_null_parms parms;
1740 parms.rset_term = rset_term_create (term_dst, -1, rank_type,
1742 return rset_create (rset_kind_null, &parms);
1745 for (i = 1; i<rset_no; i++)
1747 rset_bool_parms bool_parms;
1749 bool_parms.rset_l = result;
1750 bool_parms.rset_r = rset[i];
1751 bool_parms.key_size = sizeof(struct it_key);
1752 bool_parms.cmp = key_compare_it;
1753 bool_parms.log_item = key_logdump_txt;
1754 result = rset_create (rset_kind_and, &bool_parms);
1759 static RSET rpn_search_APT_local (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1761 oid_value attributeSet,
1763 const char *rank_type)
1768 rset_temp_parms parms;
1770 parms.rset_term = rset_term_create (termz, -1, rank_type,
1772 parms.cmp = key_compare_it;
1773 parms.key_size = sizeof (struct it_key);
1774 parms.temp_path = res_get (zh->res, "setTmpDir");
1775 result = rset_create (rset_kind_temp, &parms);
1776 rsfd = rset_open (result, RSETF_WRITE);
1778 key.sysno = atoi (termz);
1782 rset_write (result, rsfd, &key);
1783 rset_close (result, rsfd);
1787 static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1788 oid_value attributeSet, NMEM stream,
1789 Z_SortKeySpecList *sort_sequence,
1790 const char *rank_type)
1792 rset_null_parms parms;
1794 int sort_relation_value;
1795 AttrType sort_relation_type;
1800 Z_AttributeElement *ae;
1805 attr_init (&sort_relation_type, zapt, 7);
1806 sort_relation_value = attr_find (&sort_relation_type, &attributeSet);
1808 attr_init (&use_type, zapt, 1);
1809 use_value = attr_find (&use_type, &attributeSet);
1811 if (!sort_sequence->specs)
1813 sort_sequence->num_specs = 10;
1814 sort_sequence->specs = (Z_SortKeySpec **)
1815 nmem_malloc (stream, sort_sequence->num_specs *
1816 sizeof(*sort_sequence->specs));
1817 for (i = 0; i<sort_sequence->num_specs; i++)
1818 sort_sequence->specs[i] = 0;
1820 if (zapt->term->which != Z_Term_general)
1823 i = atoi_n ((char *) zapt->term->u.general->buf,
1824 zapt->term->u.general->len);
1825 if (i >= sort_sequence->num_specs)
1827 sprintf (termz, "%d", i);
1829 oe.proto = PROTO_Z3950;
1830 oe.oclass = CLASS_ATTSET;
1831 oe.value = attributeSet;
1832 if (!oid_ent_to_oid (&oe, oid))
1835 sks = (Z_SortKeySpec *) nmem_malloc (stream, sizeof(*sks));
1836 sks->sortElement = (Z_SortElement *)
1837 nmem_malloc (stream, sizeof(*sks->sortElement));
1838 sks->sortElement->which = Z_SortElement_generic;
1839 sk = sks->sortElement->u.generic = (Z_SortKey *)
1840 nmem_malloc (stream, sizeof(*sk));
1841 sk->which = Z_SortKey_sortAttributes;
1842 sk->u.sortAttributes = (Z_SortAttributes *)
1843 nmem_malloc (stream, sizeof(*sk->u.sortAttributes));
1845 sk->u.sortAttributes->id = oid;
1846 sk->u.sortAttributes->list = (Z_AttributeList *)
1847 nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list));
1848 sk->u.sortAttributes->list->num_attributes = 1;
1849 sk->u.sortAttributes->list->attributes = (Z_AttributeElement **)
1850 nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list->attributes));
1851 ae = *sk->u.sortAttributes->list->attributes = (Z_AttributeElement *)
1852 nmem_malloc (stream, sizeof(**sk->u.sortAttributes->list->attributes));
1853 ae->attributeSet = 0;
1854 ae->attributeType = (int *)
1855 nmem_malloc (stream, sizeof(*ae->attributeType));
1856 *ae->attributeType = 1;
1857 ae->which = Z_AttributeValue_numeric;
1858 ae->value.numeric = (int *)
1859 nmem_malloc (stream, sizeof(*ae->value.numeric));
1860 *ae->value.numeric = use_value;
1862 sks->sortRelation = (int *)
1863 nmem_malloc (stream, sizeof(*sks->sortRelation));
1864 if (sort_relation_value == 1)
1865 *sks->sortRelation = Z_SortKeySpec_ascending;
1866 else if (sort_relation_value == 2)
1867 *sks->sortRelation = Z_SortKeySpec_descending;
1869 *sks->sortRelation = Z_SortKeySpec_ascending;
1871 sks->caseSensitivity = (int *)
1872 nmem_malloc (stream, sizeof(*sks->caseSensitivity));
1873 *sks->caseSensitivity = 0;
1875 sks->which = Z_SortKeySpec_null;
1876 sks->u.null = odr_nullval ();
1877 sort_sequence->specs[i] = sks;
1879 parms.rset_term = rset_term_create (termz, -1, rank_type,
1881 return rset_create (rset_kind_null, &parms);
1885 static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1886 oid_value attributeSet,
1887 struct xpath_location_step *xpath, int max, NMEM mem)
1889 oid_value curAttributeSet = attributeSet;
1891 const char *use_string = 0;
1893 attr_init (&use, zapt, 1);
1894 attr_find_ex (&use, &curAttributeSet, &use_string);
1896 if (!use_string || *use_string != '/')
1899 return zebra_parse_xpath_str(use_string, xpath, max, mem);
1904 static RSET xpath_trunc(ZebraHandle zh, NMEM stream,
1905 int reg_type, const char *term, int use,
1906 oid_value curAttributeSet)
1909 struct grep_info grep_info;
1910 char term_dict[2048];
1913 int ord = zebraExplain_lookupSU (zh->reg->zei, curAttributeSet, use);
1914 int ord_len, i, r, max_pos;
1915 int term_type = Z_Term_characterString;
1916 const char *flags = "void";
1918 if (grep_info_prepare (zh, 0 /* zapt */, &grep_info, '0', stream))
1920 rset_null_parms parms;
1922 parms.rset_term = rset_term_create (term, strlen(term),
1924 parms.rset_term->nn = 0;
1925 return rset_create (rset_kind_null, &parms);
1930 rset_null_parms parms;
1932 parms.rset_term = rset_term_create (term, strlen(term),
1934 parms.rset_term->nn = 0;
1935 return rset_create (rset_kind_null, &parms);
1938 term_dict[prefix_len++] = '|';
1940 term_dict[prefix_len++] = '(';
1942 ord_len = key_SU_encode (ord, ord_buf);
1943 for (i = 0; i<ord_len; i++)
1945 term_dict[prefix_len++] = 1;
1946 term_dict[prefix_len++] = ord_buf[i];
1948 term_dict[prefix_len++] = ')';
1949 term_dict[prefix_len++] = 1;
1950 term_dict[prefix_len++] = reg_type;
1952 strcpy (term_dict+prefix_len, term);
1954 grep_info.isam_p_indx = 0;
1955 r = dict_lookup_grep (zh->reg->dict, term_dict, 0,
1956 &grep_info, &max_pos, 0, grep_handle);
1957 yaz_log (LOG_LOG, "%s %d positions", term, grep_info.isam_p_indx);
1958 rset = rset_trunc (zh, grep_info.isam_p_buf,
1959 grep_info.isam_p_indx, term, strlen(term),
1960 flags, 1, term_type);
1961 grep_info_delete (&grep_info);
1965 static RSET rpn_search_xpath (ZebraHandle zh,
1966 oid_value attributeSet,
1967 int num_bases, char **basenames,
1968 NMEM stream, const char *rank_type, RSET rset,
1969 int xpath_len, struct xpath_location_step *xpath)
1971 oid_value curAttributeSet = attributeSet;
1978 yaz_log (LOG_DEBUG, "len=%d", xpath_len);
1979 for (i = 0; i<xpath_len; i++)
1981 yaz_log (LOG_DEBUG, "XPATH %d %s", i, xpath[i].part);
1984 curAttributeSet = VAL_IDXPATH;
1994 a[@attr=value]/b[@other=othervalue]
1996 /e/@a val range(e/,range(@a,freetext(w,1015,val),@a),e/)
1997 /a/b val range(b/a/,freetext(w,1016,val),b/a/)
1998 /a/b/@c val range(b/a/,range(@c,freetext(w,1016,val),@c),b/a/)
1999 /a/b[@c=y] val range(b/a/,freetext(w,1016,val),b/a/,@c=y)
2000 /a[@c=y]/b val range(a/,range(b/a/,freetext(w,1016,val),b/a/),a/,@c=y)
2001 /a[@c=x]/b[@c=y] range(a/,range(b/a/,freetext(w,1016,val),b/a/,@c=y),a/,@c=x)
2005 dict_grep_cmap (zh->reg->dict, 0, 0);
2007 for (base_no = 0; base_no < num_bases; base_no++)
2009 int level = xpath_len;
2012 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2014 zh->errCode = 109; /* Database unavailable */
2015 zh->errString = basenames[base_no];
2018 while (--level >= 0)
2020 char xpath_rev[128];
2022 rset_between_parms parms;
2023 RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0;
2027 for (i = level; i >= 1; --i)
2029 const char *cp = xpath[i].part;
2035 memcpy (xpath_rev + len, "[^/]*", 5);
2038 else if (*cp == ' ')
2041 xpath_rev[len++] = 1;
2042 xpath_rev[len++] = ' ';
2046 xpath_rev[len++] = *cp;
2047 xpath_rev[len++] = '/';
2049 else if (i == 1) /* // case */
2051 xpath_rev[len++] = '.';
2052 xpath_rev[len++] = '*';
2057 if (xpath[level].predicate &&
2058 xpath[level].predicate->which == XPATH_PREDICATE_RELATION &&
2059 xpath[level].predicate->u.relation.name[0])
2061 WRBUF wbuf = wrbuf_alloc();
2062 wrbuf_puts(wbuf, xpath[level].predicate->u.relation.name+1);
2063 if (xpath[level].predicate->u.relation.value)
2065 const char *cp = xpath[level].predicate->u.relation.value;
2066 wrbuf_putc(wbuf, '=');
2070 if (strchr(REGEX_CHARS, *cp))
2071 wrbuf_putc(wbuf, '\\');
2072 wrbuf_putc(wbuf, *cp);
2076 wrbuf_puts(wbuf, "");
2077 rset_attr = xpath_trunc (
2078 zh, stream, '0', wrbuf_buf(wbuf), 3, curAttributeSet);
2079 wrbuf_free(wbuf, 1);
2086 yaz_log (LOG_DEBUG, "xpath_rev (%d) = %s", level, xpath_rev);
2087 if (strlen(xpath_rev))
2089 rset_start_tag = xpath_trunc(zh, stream,
2090 '0', xpath_rev, 1, curAttributeSet);
2092 rset_end_tag = xpath_trunc(zh, stream,
2093 '0', xpath_rev, 2, curAttributeSet);
2095 parms.key_size = sizeof(struct it_key);
2096 parms.cmp = key_compare_it;
2097 parms.rset_l = rset_start_tag;
2098 parms.rset_m = rset;
2099 parms.rset_r = rset_end_tag;
2100 parms.rset_attr = rset_attr;
2101 parms.printer = key_print_it;
2102 rset = rset_create (rset_kind_between, &parms);
2113 static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
2114 oid_value attributeSet, NMEM stream,
2115 Z_SortKeySpecList *sort_sequence,
2116 int num_bases, char **basenames)
2119 char *search_type = NULL;
2120 char rank_type[128];
2123 char termz[IT_MAX_WORD+1];
2127 struct xpath_location_step xpath[10];
2129 zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type,
2130 rank_type, &complete_flag, &sort_flag);
2132 yaz_log(LOG_DEBUG, "reg_id=%c", reg_id);
2133 yaz_log(LOG_DEBUG, "complete_flag=%d", complete_flag);
2134 yaz_log(LOG_DEBUG, "search_type=%s", search_type);
2135 yaz_log(LOG_DEBUG, "rank_type=%s", rank_type);
2137 if (zapt_term_to_utf8(zh, zapt, termz))
2141 return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence,
2143 xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream);
2147 if (xpath[xpath_len-1].part[0] == '@')
2151 if (!strcmp (search_type, "phrase"))
2153 rset = rpn_search_APT_phrase (zh, zapt, termz, attributeSet, stream,
2154 reg_id, complete_flag, rank_type,
2156 num_bases, basenames);
2158 else if (!strcmp (search_type, "and-list"))
2160 rset = rpn_search_APT_and_list (zh, zapt, termz, attributeSet, stream,
2161 reg_id, complete_flag, rank_type,
2163 num_bases, basenames);
2165 else if (!strcmp (search_type, "or-list"))
2167 rset = rpn_search_APT_or_list (zh, zapt, termz, attributeSet, stream,
2168 reg_id, complete_flag, rank_type,
2170 num_bases, basenames);
2172 else if (!strcmp (search_type, "local"))
2174 rset = rpn_search_APT_local (zh, zapt, termz, attributeSet, stream,
2177 else if (!strcmp (search_type, "numeric"))
2179 rset = rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream,
2180 reg_id, complete_flag, rank_type,
2182 num_bases, basenames);
2184 else if (!strcmp (search_type, "always"))
2190 return rpn_search_xpath (zh, attributeSet, num_bases, basenames,
2191 stream, rank_type, rset, xpath_len, xpath);
2194 static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs,
2195 oid_value attributeSet, NMEM stream,
2196 Z_SortKeySpecList *sort_sequence,
2197 int num_bases, char **basenames)
2200 if (zs->which == Z_RPNStructure_complex)
2202 Z_Operator *zop = zs->u.complex->roperator;
2203 rset_bool_parms bool_parms;
2205 bool_parms.rset_l = rpn_search_structure (zh, zs->u.complex->s1,
2206 attributeSet, stream,
2208 num_bases, basenames);
2209 if (bool_parms.rset_l == NULL)
2211 bool_parms.rset_r = rpn_search_structure (zh, zs->u.complex->s2,
2212 attributeSet, stream,
2214 num_bases, basenames);
2215 if (bool_parms.rset_r == NULL)
2217 rset_delete (bool_parms.rset_l);
2220 bool_parms.key_size = sizeof(struct it_key);
2221 bool_parms.cmp = key_compare_it;
2222 bool_parms.log_item = key_logdump_txt;
2226 case Z_Operator_and:
2227 r = rset_create (rset_kind_and, &bool_parms);
2230 r = rset_create (rset_kind_or, &bool_parms);
2232 case Z_Operator_and_not:
2233 r = rset_create (rset_kind_not, &bool_parms);
2235 case Z_Operator_prox:
2236 if (zop->u.prox->which != Z_ProximityOperator_known)
2241 if (*zop->u.prox->u.known != Z_ProxUnit_word)
2243 char *val = (char *) nmem_malloc (stream, 16);
2245 zh->errString = val;
2246 sprintf (val, "%d", *zop->u.prox->u.known);
2251 /* new / old prox */
2252 rset_prox_parms parms;
2255 twosets[0] = bool_parms.rset_l;
2256 twosets[1] = bool_parms.rset_r;
2257 parms.rset = twosets;
2259 parms.ordered = *zop->u.prox->ordered;
2260 parms.exclusion = (!zop->u.prox->exclusion ? 0 :
2261 *zop->u.prox->exclusion);
2262 parms.relation = *zop->u.prox->relationType;
2263 parms.distance = *zop->u.prox->distance;
2264 parms.key_size = sizeof(struct it_key);
2265 parms.cmp = key_compare_it;
2266 parms.getseq = key_get_seq;
2267 parms.log_item = key_logdump_txt;
2268 r = rset_create(rset_kind_prox, &parms);
2276 else if (zs->which == Z_RPNStructure_simple)
2278 if (zs->u.simple->which == Z_Operand_APT)
2280 yaz_log(LOG_DEBUG, "rpn_search_APT");
2281 r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm,
2282 attributeSet, stream, sort_sequence,
2283 num_bases, basenames);
2285 else if (zs->u.simple->which == Z_Operand_resultSetId)
2287 yaz_log(LOG_DEBUG, "rpn_search_ref");
2288 r = resultSetRef (zh, zs->u.simple->u.resultSetId);
2291 r = rset_create (rset_kind_null, NULL);
2294 nmem_strdup (stream, zs->u.simple->u.resultSetId);
2315 RSET rpn_search (ZebraHandle zh, NMEM nmem,
2316 Z_RPNQuery *rpn, int num_bases, char **basenames,
2317 const char *setname,
2322 oid_value attributeSet;
2323 Z_SortKeySpecList *sort_sequence;
2327 zh->errString = NULL;
2330 sort_sequence = (Z_SortKeySpecList *)
2331 nmem_malloc (nmem, sizeof(*sort_sequence));
2332 sort_sequence->num_specs = 10;
2333 sort_sequence->specs = (Z_SortKeySpec **)
2334 nmem_malloc (nmem, sort_sequence->num_specs *
2335 sizeof(*sort_sequence->specs));
2336 for (i = 0; i<sort_sequence->num_specs; i++)
2337 sort_sequence->specs[i] = 0;
2339 attrset = oid_getentbyoid (rpn->attributeSetId);
2340 attributeSet = attrset->value;
2341 rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet,
2342 nmem, sort_sequence, num_bases, basenames);
2347 yaz_log(LOG_DEBUG, "search error: %d", zh->errCode);
2349 for (i = 0; sort_sequence->specs[i]; i++)
2351 sort_sequence->num_specs = i;
2353 resultSetRank (zh, sset, rset);
2356 yaz_log(LOG_DEBUG, "resultSetSortSingle in rpn_search");
2357 resultSetSortSingle (zh, nmem, sset, rset,
2358 sort_sequence, &sort_status);
2361 yaz_log(LOG_DEBUG, "resultSetSortSingle status = %d", zh->errCode);
2367 struct scan_info_entry {
2373 struct scan_info_entry *list;
2379 static int scan_handle (char *name, const char *info, int pos, void *client)
2381 int len_prefix, idx;
2382 struct scan_info *scan_info = (struct scan_info *) client;
2384 len_prefix = strlen(scan_info->prefix);
2385 if (memcmp (name, scan_info->prefix, len_prefix))
2387 if (pos > 0) idx = scan_info->after - pos + scan_info->before;
2390 scan_info->list[idx].term = (char *)
2391 odr_malloc (scan_info->odr, strlen(name + len_prefix)+1);
2392 strcpy (scan_info->list[idx].term, name + len_prefix);
2393 assert (*info == sizeof(ISAMS_P));
2394 memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAMS_P));
2398 static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type,
2399 char **dst, const char *src)
2401 char term_src[IT_MAX_WORD];
2402 char term_dst[IT_MAX_WORD];
2404 term_untrans (zh, reg_type, term_src, src);
2406 if (zh->iconv_from_utf8 != 0)
2409 char *inbuf = term_src;
2410 size_t inleft = strlen(term_src);
2411 char *outbuf = term_dst;
2412 size_t outleft = sizeof(term_dst)-1;
2415 ret = yaz_iconv (zh->iconv_from_utf8, &inbuf, &inleft,
2417 if (ret == (size_t)(-1))
2420 len = outbuf - term_dst;
2421 *dst = nmem_malloc (stream, len + 1);
2423 memcpy (*dst, term_dst, len);
2427 *dst = nmem_strdup (stream, term_src);
2430 static void count_set (RSET r, int *count)
2438 yaz_log(LOG_DEBUG, "count_set");
2441 rfd = rset_open (r, RSETF_READ);
2442 while (rset_read (r, rfd, &key, &term_index))
2444 if (key.sysno != psysno)
2451 rset_close (r, rfd);
2452 yaz_log(LOG_DEBUG, "%d keys, %d records", kno, *count);
2455 void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
2456 oid_value attributeset,
2457 int num_bases, char **basenames,
2458 int *position, int *num_entries, ZebraScanEntry **list,
2459 int *is_partial, RSET limit_set, int return_zero)
2462 int pos = *position;
2463 int num = *num_entries;
2467 char termz[IT_MAX_WORD+20];
2470 const char *use_string = 0;
2471 struct scan_info *scan_info_array;
2472 ZebraScanEntry *glist;
2473 int ords[32], ord_no = 0;
2476 int bases_ok = 0; /* no of databases with OK attribute */
2477 int errCode = 0; /* err code (if any is not OK) */
2478 char *errString = 0; /* addinfo */
2481 char *search_type = NULL;
2482 char rank_type[128];
2488 if (attributeset == VAL_NONE)
2489 attributeset = VAL_BIB1;
2494 int termset_value_numeric;
2495 const char *termset_value_string;
2496 attr_init (&termset, zapt, 8);
2497 termset_value_numeric =
2498 attr_find_ex (&termset, NULL, &termset_value_string);
2499 if (termset_value_numeric != -1)
2502 const char *termset_name = 0;
2504 if (termset_value_numeric != -2)
2507 sprintf (resname, "%d", termset_value_numeric);
2508 termset_name = resname;
2511 termset_name = termset_value_string;
2513 limit_set = resultSetRef (zh, termset_name);
2517 yaz_log (LOG_DEBUG, "position = %d, num = %d set=%d",
2518 pos, num, attributeset);
2520 attr_init (&use, zapt, 1);
2521 use_value = attr_find_ex (&use, &attributeset, &use_string);
2523 if (zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type,
2524 rank_type, &complete_flag, &sort_flag))
2530 yaz_log (LOG_DEBUG, "use_value = %d", use_value);
2532 if (use_value == -1)
2534 for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
2538 data1_local_attribute *local_attr;
2540 if ((r=att_getentbyatt (zh, &attp, attributeset, use_value,
2543 yaz_log(LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
2544 attributeset, use_value);
2548 sprintf (val_str, "%d", use_value);
2550 errString = odr_strdup (stream, val_str);
2556 if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no]))
2558 zh->errString = basenames[base_no];
2559 zh->errCode = 109; /* Database unavailable */
2564 for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
2565 local_attr = local_attr->next)
2569 ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal,
2572 ords[ord_no++] = ord;
2575 if (!bases_ok && errCode)
2577 zh->errCode = errCode;
2578 zh->errString = errString;
2584 sprintf (val_str, "%d", use_value);
2586 zh->errString = odr_strdup (stream, val_str);
2591 /* prepare dictionary scanning */
2594 scan_info_array = (struct scan_info *)
2595 odr_malloc (stream, ord_no * sizeof(*scan_info_array));
2596 for (i = 0; i < ord_no; i++)
2598 int j, prefix_len = 0;
2599 int before_tmp = before, after_tmp = after;
2600 struct scan_info *scan_info = scan_info_array + i;
2601 struct rpn_char_map_info rcmi;
2603 rpn_char_map_prepare (zh->reg, reg_id, &rcmi);
2605 scan_info->before = before;
2606 scan_info->after = after;
2607 scan_info->odr = stream;
2609 scan_info->list = (struct scan_info_entry *)
2610 odr_malloc (stream, (before+after) * sizeof(*scan_info->list));
2611 for (j = 0; j<before+after; j++)
2612 scan_info->list[j].term = NULL;
2614 prefix_len += key_SU_encode (ords[i], termz + prefix_len);
2615 termz[prefix_len++] = reg_id;
2616 termz[prefix_len] = 0;
2617 strcpy (scan_info->prefix, termz);
2619 if (trans_scan_term (zh, zapt, termz+prefix_len, reg_id))
2622 dict_scan (zh->reg->dict, termz, &before_tmp, &after_tmp,
2623 scan_info, scan_handle);
2625 glist = (ZebraScanEntry *)
2626 odr_malloc (stream, (before+after)*sizeof(*glist));
2628 /* consider terms after main term */
2629 for (i = 0; i < ord_no; i++)
2633 for (i = 0; i<after; i++)
2636 const char *mterm = NULL;
2640 for (j = 0; j < ord_no; j++)
2642 if (ptr[j] < before+after &&
2643 (tst=scan_info_array[j].list[ptr[j]].term) &&
2644 (!mterm || strcmp (tst, mterm) < 0))
2652 scan_term_untrans (zh, stream->mem, reg_id,
2653 &glist[i+before].term, mterm);
2654 rset = rset_trunc (zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
2655 glist[i+before].term, strlen(glist[i+before].term),
2656 NULL, 0, zapt->term->which);
2659 for (j = j0+1; j<ord_no; j++)
2661 if (ptr[j] < before+after &&
2662 (tst=scan_info_array[j].list[ptr[j]].term) &&
2663 !strcmp (tst, mterm))
2665 rset_bool_parms bool_parms;
2669 rset_trunc (zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
2670 glist[i+before].term,
2671 strlen(glist[i+before].term), NULL, 0,
2674 bool_parms.key_size = sizeof(struct it_key);
2675 bool_parms.cmp = key_compare_it;
2676 bool_parms.log_item = key_logdump_txt;
2677 bool_parms.rset_l = rset;
2678 bool_parms.rset_r = rset2;
2680 rset = rset_create (rset_kind_or, &bool_parms);
2687 rset_bool_parms bool_parms;
2689 bool_parms.key_size = sizeof(struct it_key);
2690 bool_parms.cmp = key_compare_it;
2691 bool_parms.log_item = key_logdump_txt;
2692 bool_parms.rset_l = rset;
2693 bool_parms.rset_r = rset_dup(limit_set);
2695 rset = rset_create (rset_kind_and, &bool_parms);
2697 count_set (rset, &glist[i+before].occurrences);
2702 *num_entries -= (after-i);
2706 /* consider terms before main term */
2707 for (i = 0; i<ord_no; i++)
2710 for (i = 0; i<before; i++)
2713 const char *mterm = NULL;
2717 for (j = 0; j <ord_no; j++)
2719 if (ptr[j] < before &&
2720 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
2721 (!mterm || strcmp (tst, mterm) > 0))
2730 scan_term_untrans (zh, stream->mem, reg_id,
2731 &glist[before-1-i].term, mterm);
2734 (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
2735 glist[before-1-i].term, strlen(glist[before-1-i].term),
2736 NULL, 0, zapt->term->which);
2740 for (j = j0+1; j<ord_no; j++)
2742 if (ptr[j] < before &&
2743 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
2744 !strcmp (tst, mterm))
2746 rset_bool_parms bool_parms;
2749 rset2 = rset_trunc (zh,
2750 &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
2751 glist[before-1-i].term,
2752 strlen(glist[before-1-i].term), NULL, 0,
2755 bool_parms.key_size = sizeof(struct it_key);
2756 bool_parms.cmp = key_compare_it;
2757 bool_parms.log_item = key_logdump_txt;
2758 bool_parms.rset_l = rset;
2759 bool_parms.rset_r = rset2;
2761 rset = rset_create (rset_kind_or, &bool_parms);
2768 rset_bool_parms bool_parms;
2770 bool_parms.key_size = sizeof(struct it_key);
2771 bool_parms.cmp = key_compare_it;
2772 bool_parms.log_item = key_logdump_txt;
2773 bool_parms.rset_l = rset;
2774 bool_parms.rset_r = rset_dup(limit_set);
2776 rset = rset_create (rset_kind_and, &bool_parms);
2778 count_set (rset, &glist[before-1-i].occurrences);
2788 *list = glist + i; /* list is set to first 'real' entry */
2790 yaz_log(LOG_DEBUG, "position = %d, num_entries = %d",
2791 *position, *num_entries);
2793 yaz_log(LOG_DEBUG, "scan error: %d", zh->errCode);