X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=fd48152666fe68608dfcdda2d08da2bba9860303;hb=a6908b6d93de84a850a3e558a2e4c17f02c3d651;hp=03bb798a3d9a6182a353fc0c5fec699e17547d23;hpb=75a4bfb9aee40405a751320769d74fd890f2c0b4;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 03bb798..fd48152 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,10 +1,36 @@ /* - * Copyright (C) 1994-1996, Index Data I/S + * Copyright (C) 1994-1997, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.58 1996-12-23 15:30:45 adam + * Revision 1.66 1997-09-25 14:58:03 adam + * Windows NT port. + * + * Revision 1.65 1997/09/22 12:39:06 adam + * Added get_pos method for the ranked result sets. + * + * Revision 1.64 1997/09/18 08:59:20 adam + * Extra generic handle for the character mapping routines. + * + * Revision 1.63 1997/09/17 12:19:18 adam + * Zebra version corresponds to YAZ version 1.4. + * Changed Zebra server so that it doesn't depend on global common_resource. + * + * Revision 1.62 1997/09/05 15:30:09 adam + * Changed prototype for chr_map_input - added const. + * Added support for C++, headers uses extern "C" for public definitions. + * + * Revision 1.61 1997/02/10 10:21:14 adam + * Bug fix: in search terms character (^) wasn't observed. + * + * Revision 1.60 1997/01/31 11:10:34 adam + * Bug fix: Leading and trailing white space weren't removed in scan tokens. + * + * Revision 1.59 1997/01/17 11:31:46 adam + * Bug fix: complete phrase search didn't work. + * + * Revision 1.58 1996/12/23 15:30:45 adam * Work on truncation. * Bug fix: result sets weren't deleted after server shut down. * @@ -203,11 +229,14 @@ */ #include #include +#ifdef WINDOWS +#include +#else #include +#endif #include #include "zserver.h" -#include "attribute.h" #include #include @@ -330,10 +359,10 @@ static int grep_handle (char *name, const char *info, void *p) return 0; } -static int term_pre (char **src, const char *ct1, const char *ct2) +static int term_pre (const char **src, const char *ct1, const char *ct2) { - char *s1, *s0 = *src; - char **map; + const char *s1, *s0 = *src; + const char **map; /* skip white space */ while (*s0) @@ -343,7 +372,7 @@ static int term_pre (char **src, const char *ct1, const char *ct2) if (ct2 && strchr (ct2, *s0)) break; s1 = s0; - map = map_chrs_input (&s1, strlen(s1)); + map = map_chrs_input (0, &s1, strlen(s1)); if (**map != *CHR_SPACE) break; s0 = s1; @@ -352,9 +381,10 @@ static int term_pre (char **src, const char *ct1, const char *ct2) return *s0; } -static int term_100 (char **src, char *dst) +static int term_100 (const char **src, char *dst, int space_split) { - char *s0, *s1, **map; + const char *s0, *s1; + const char **map; int i = 0; if (!term_pre (src, NULL, NULL)) @@ -363,8 +393,8 @@ static int term_100 (char **src, char *dst) while (*s0) { s1 = s0; - map = map_chrs_input (&s0, strlen(s0)); - if (**map == *CHR_SPACE) + map = map_chrs_input (0, &s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) break; while (s1 < s0) { @@ -378,9 +408,10 @@ static int term_100 (char **src, char *dst) return i; } -static int term_101 (char **src, char *dst) +static int term_101 (const char **src, char *dst, int space_split) { - char *s0, *s1, **map; + const char *s0, *s1; + const char **map; int i = 0; if (!term_pre (src, "#", "#")) @@ -397,8 +428,8 @@ static int term_101 (char **src, char *dst) else { s1 = s0; - map = map_chrs_input (&s0, strlen(s0)); - if (**map == *CHR_SPACE) + map = map_chrs_input (0, &s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) break; while (s1 < s0) { @@ -414,12 +445,13 @@ static int term_101 (char **src, char *dst) } -static int term_103 (char **src, char *dst, int *errors) +static int term_103 (const char **src, char *dst, int *errors, int space_split) { int i = 0; - char *s0, *s1, **map; + const char *s0, *s1; + const char **map; - if (!term_pre (src, "\\()[].*+?|", "(")) + if (!term_pre (src, "^\\()[].*+?|", "(")) return 0; s0 = *src; if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] && @@ -432,12 +464,12 @@ static int term_103 (char **src, char *dst, int *errors) } while (*s0) { - if (strchr ("\\()[].*+?|-", *s0)) + if (strchr ("^\\()[].*+?|-", *s0)) dst[i++] = *s0++; else { s1 = s0; - map = map_chrs_input (&s0, strlen(s0)); + map = map_chrs_input (0, &s0, strlen(s0)); if (**map == *CHR_SPACE) break; while (s1 < s0) @@ -453,9 +485,9 @@ static int term_103 (char **src, char *dst, int *errors) return i; } -static int term_102 (char **src, char *dst) +static int term_102 (const char **src, char *dst, int space_split) { - return term_103 (src, dst, NULL); + return term_103 (src, dst, NULL, space_split); } /* gen_regular_rel - generate regular expression from relation @@ -568,7 +600,7 @@ static void gen_regular_rel (char *dst, int val, int islt) } static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - char **term_sub, + const char **term_sub, char *term_dict, oid_value attributeSet, struct grep_info *grep_info, @@ -585,7 +617,7 @@ static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, switch (relation_value) { case 1: - if (!term_100 (term_sub, term_dict)) + if (!term_100 (term_sub, term_dict, 1)) return 0; term_value = atoi (term_dict); if (term_value <= 0) @@ -594,7 +626,7 @@ static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1); break; case 2: - if (!term_100 (term_sub, term_dict)) + if (!term_100 (term_sub, term_dict, 1)) return 0; term_value = atoi (term_dict); if (term_value < 0) @@ -603,7 +635,7 @@ static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, gen_regular_rel (term_dict + strlen(term_dict), term_value, 1); break; case 4: - if (!term_100 (term_sub, term_dict)) + if (!term_100 (term_sub, term_dict, 1)) return 0; term_value = atoi (term_dict); if (term_value < 0) @@ -612,7 +644,7 @@ static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, gen_regular_rel (term_dict + strlen(term_dict), term_value, 0); break; case 5: - if (!term_100 (term_sub, term_dict)) + if (!term_100 (term_sub, term_dict, 1)) return 0; term_value = atoi (term_dict); if (term_value < 0) @@ -633,9 +665,9 @@ static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, } static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - char **term_sub, int regType, + const char **term_sub, int regType, oid_value attributeSet, struct grep_info *grep_info, - int num_bases, char **basenames) + int num_bases, char **basenames, int space_split) { char term_dict[2*IT_MAX_WORD+2]; int j, r, base_no; @@ -644,11 +676,11 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, AttrType use; int use_value; oid_value curAttributeSet = attributeSet; - char *termp; + const char *termp; attr_init (&use, zapt, 1); use_value = attr_find (&use, &curAttributeSet); - logf (LOG_DEBUG, "use value %d", use_value); + logf (LOG_DEBUG, "field_term, use value %d", use_value); attr_init (&truncation, zapt, 5); truncation_value = attr_find (&truncation, NULL); logf (LOG_DEBUG, "truncation value %d", truncation_value); @@ -658,13 +690,12 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, for (base_no = 0; base_no < num_bases; base_no++) { - attent *attp; + attent attp; data1_local_attribute *local_attr; int max_pos, prefix_len = 0; termp = *term_sub; - attp = att_getentbyatt (curAttributeSet, use_value); - if (!attp) + if (!att_getentbyatt (zi, &attp, curAttributeSet, use_value)) { logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", curAttributeSet, use_value); @@ -677,12 +708,12 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, zi->errString = basenames[base_no]; return -1; } - for (local_attr = attp->local_attributes; local_attr; + for (local_attr = attp.local_attributes; local_attr; local_attr = local_attr->next) { int ord; - ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, + ord = zebTargetInfo_lookupSU (zi->zti, attp.attset_ordinal, local_attr->local); if (ord < 0) continue; @@ -710,8 +741,8 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, { case -1: /* not specified */ case 100: /* do not truncate */ - term_dict[j++] = '('; - if (!term_100 (&termp, term_dict + j)) + term_dict[j++] = '('; + if (!term_100 (&termp, term_dict + j, space_split)) return 0; strcat (term_dict, ")"); r = dict_lookup_grep (zi->dict, term_dict, 0, grep_info, @@ -721,7 +752,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, break; case 1: /* right truncation */ term_dict[j++] = '('; - if (!term_100 (&termp, term_dict + j)) + if (!term_100 (&termp, term_dict + j, space_split)) return 0; strcat (term_dict, ".*)"); dict_lookup_grep (zi->dict, term_dict, 0, grep_info, @@ -733,7 +764,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, return -1; case 101: /* process # in term */ term_dict[j++] = '('; - if (!term_101 (&termp, term_dict + j)) + if (!term_101 (&termp, term_dict + j, space_split)) return 0; strcat (term_dict, ")"); r = dict_lookup_grep (zi->dict, term_dict, 0, grep_info, @@ -743,7 +774,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, break; case 102: /* Regexp-1 */ term_dict[j++] = '('; - if (!term_102 (&termp, term_dict + j)) + if (!term_102 (&termp, term_dict + j, space_split)) return 0; strcat (term_dict, ")"); logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r); @@ -756,7 +787,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, case 103: /* Regexp-1 */ r = 1; term_dict[j++] = '('; - if (!term_103 (&termp, term_dict + j, &r)) + if (!term_103 (&termp, term_dict + j, &r, space_split)) return 0; strcat (term_dict, ")"); logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r); @@ -791,27 +822,28 @@ static void trans_scan_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, char *termz) { Z_Term *term = zapt->term; - char **map; - char *cp = (char*) term->u.general->buf; + const char **map; + const char *cp = (const char *) term->u.general->buf; const char *cp_end = cp + term->u.general->len; const char *src; int i = 0; - int prev_space = 0; + const char *space_map = NULL; int len; while ((len = (cp_end - cp)) > 0) { - map = map_chrs_input (&cp, len); + map = map_chrs_input (0, &cp, len); if (**map == *CHR_SPACE) - { - if (prev_space) - continue; - prev_space = 1; - } + space_map = *map; else - prev_space = 0; - for (src = *map; *src; src++) - termz[i++] = *src; + { + if (i && space_map) + for (src = space_map; *src; src++) + termz[i++] = *src; + space_map = NULL; + for (src = *map; *src; src++) + termz[i++] = *src; + } } termz[i] = '\0'; } @@ -823,15 +855,16 @@ static RSET rpn_search_APT_relevance (ZServerInfo *zi, { rset_relevance_parms parms; char termz[IT_MAX_WORD+1]; - char *termp = termz; + const char *termp = termz; struct grep_info grep_info; RSET result; int term_index = 0; int r; parms.key_size = sizeof(struct it_key); - parms.max_rec = 100; + parms.max_rec = 1000; parms.cmp = key_compare_it; + parms.get_pos = key_get_pos; parms.is = zi->isam; parms.isc = zi->isamc; parms.no_terms = 0; @@ -852,7 +885,7 @@ static RSET rpn_search_APT_relevance (ZServerInfo *zi, while (1) { r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info, - num_bases, basenames); + num_bases, basenames, 1); if (r <= 0) break; #ifdef TERM_COUNT @@ -883,7 +916,7 @@ static RSET rpn_search_APT_cphrase (ZServerInfo *zi, char termz[IT_MAX_WORD+1]; struct grep_info grep_info; RSET result; - char *termp = termz; + const char *termp = termz; int r; if (zapt->term->which != Z_Term_general) @@ -901,7 +934,7 @@ static RSET rpn_search_APT_cphrase (ZServerInfo *zi, grep_info.isam_p_buf = NULL; r = field_term (zi, zapt, &termp, 'p', attributeSet, &grep_info, - num_bases, basenames); + num_bases, basenames, 0); result = rset_trunc (zi, grep_info.isam_p_buf, grep_info.isam_p_indx); #ifdef TERM_COUNT xfree(grep_info.term_no); @@ -910,7 +943,8 @@ static RSET rpn_search_APT_cphrase (ZServerInfo *zi, return result; } -static RSET rpn_proximity (RSET rset1, RSET rset2, int ordered, +static RSET rpn_proximity (ZServerInfo *zi, RSET rset1, RSET rset2, + int ordered, int exclusion, int relation, int distance) { int i; @@ -928,6 +962,7 @@ static RSET rpn_proximity (RSET rset1, RSET rset2, int ordered, more2 = rset_read (rset2, rsfd2, &buf2); parms.key_size = sizeof (struct it_key); + parms.temp_path = res_get (zi->res, "setTmpDir"); result = rset_create (rset_kind_temp, &parms); rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); @@ -959,7 +994,6 @@ static RSET rpn_proximity (RSET rset1, RSET rset2, int ordered, int excl = exclusion; if (!ordered && diff < 0) diff = -diff; - logf (LOG_DEBUG, "l = %d r = %d", seqno[i], buf2.seqno); switch (relation) { case 1: /* < */ @@ -988,10 +1022,7 @@ static RSET rpn_proximity (RSET rset1, RSET rset2, int ordered, break; } if (excl) - { - logf (LOG_DEBUG, " match"); rset_write (result, rsfd_result, &buf2); - } } } while ((more2 = rset_read (rset2, rsfd2, &buf2)) && sysno == buf2.sysno); @@ -1003,7 +1034,7 @@ static RSET rpn_proximity (RSET rset1, RSET rset2, int ordered, return result; } -static RSET rpn_prox (RSET *rset, int rset_no) +static RSET rpn_prox (ZServerInfo *zi, RSET *rset, int rset_no) { int i; RSFD *rsfd; @@ -1036,6 +1067,7 @@ static RSET rpn_prox (RSET *rset, int rset_no) } } parms.key_size = sizeof (struct it_key); + parms.temp_path = res_get (zi->res, "setTmpDir"); result = rset_create (rset_kind_temp, &parms); rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); @@ -1095,7 +1127,7 @@ static RSET rpn_search_APT_phrase (ZServerInfo *zi, int num_bases, char **basenames) { char termz[IT_MAX_WORD+1]; - char *termp = termz; + const char *termp = termz; RSET rset[60], result; int i, r, rset_no = 0; struct grep_info grep_info; @@ -1117,7 +1149,7 @@ static RSET rpn_search_APT_phrase (ZServerInfo *zi, { grep_info.isam_p_indx = 0; r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info, - num_bases, basenames); + num_bases, basenames, 1); if (r < 1) break; rset[rset_no] = rset_trunc (zi, grep_info.isam_p_buf, @@ -1134,7 +1166,7 @@ static RSET rpn_search_APT_phrase (ZServerInfo *zi, return rset_create (rset_kind_null, NULL); else if (rset_no == 1) return (rset[0]); - result = rpn_prox (rset, rset_no); + result = rpn_prox (zi, rset, rset_no); for (i = 0; ires, "setTmpDir"); result = rset_create (rset_kind_temp, &parms); rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); @@ -1317,7 +1350,7 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, sprintf (val, "%d", *zop->u.prox->proximityUnitCode); return NULL; } - r = rpn_proximity (bool_parms.rset_l, bool_parms.rset_r, + r = rpn_proximity (zi, bool_parms.rset_l, bool_parms.rset_r, *zop->u.prox->ordered, (!zop->u.prox->exclusion ? 0 : *zop->u.prox->exclusion), @@ -1356,7 +1389,7 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, return r; } -void count_set_save (RSET *r, int *count) +void count_set_save (ZServerInfo *zi, RSET *r, int *count) { int psysno = 0; int kno = 0; @@ -1364,10 +1397,12 @@ void count_set_save (RSET *r, int *count) RSFD rfd, wfd; RSET w; rset_temp_parms parms; - + int maxResultSetSize = atoi (res_get_def (zi->res, + "maxResultSetSize", "400")); logf (LOG_DEBUG, "count_set_save"); *count = 0; parms.key_size = sizeof(struct it_key); + parms.temp_path = res_get (zi->res, "setTmpDir"); w = rset_create (rset_kind_temp, &parms); wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO); rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO); @@ -1375,7 +1410,7 @@ void count_set_save (RSET *r, int *count) { if (key.sysno != psysno) { - if (*count < 400) + if (*count < maxResultSetSize) rset_write (w, wfd, &key); (*count)++; psysno = key.sysno; @@ -1420,7 +1455,7 @@ int rpn_search (ZServerInfo *zi, oident *attrset; oid_value attributeSet; - dict_grep_cmap (zi->dict, map_chrs_input); + dict_grep_cmap (zi->dict, 0, map_chrs_input); zlog_rpn (rpn); zi->errCode = 0; @@ -1433,7 +1468,7 @@ int rpn_search (ZServerInfo *zi, if (!rset) return zi->errCode; if (rset_is_volatile(rset)) - count_set_save(&rset,hits); + count_set_save(zi, &rset,hits); else count_set (rset, hits); resultSetAdd (zi, setname, 1, rset); @@ -1466,7 +1501,6 @@ static int scan_handle (char *name, const char *info, int pos, void *client) idx = scan_info->after - pos + scan_info->before; else idx = - pos - 1; - logf (LOG_DEBUG, "%-3d %s", idx, name+len_prefix); scan_info->list[idx].term = odr_malloc (scan_info->odr, strlen(name + len_prefix)+1); strcpy (scan_info->list[idx].term, name + len_prefix); @@ -1529,11 +1563,10 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, use_value = 1016; for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++) { - attent *attp; + attent attp; data1_local_attribute *local_attr; - attp = att_getentbyatt (attributeset, use_value); - if (!attp) + if (!att_getentbyatt (zi, &attp, attributeset, use_value)) { logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", attributeset, use_value); @@ -1544,12 +1577,12 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, zi->errString = basenames[base_no]; return zi->errCode = 109; /* Database unavailable */ } - for (local_attr = attp->local_attributes; local_attr && ord_no < 32; + for (local_attr = attp.local_attributes; local_attr && ord_no < 32; local_attr = local_attr->next) { int ord; - ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, + ord = zebTargetInfo_lookupSU (zi->zti, attp.attset_ordinal, local_attr->local); if (ord > 0) ords[ord_no++] = ord;