X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=f7a233084e789b0ebfb08a6dd4723e4c24b074ce;hb=c8ecacb7e0284165e2249e117a0b859d75e297f0;hp=5b50a4f217aac5e5b9f3bd3b36afbc89b0a7fe08;hpb=2038fcec0ea77e3fdb854ff968423d3ef24d53a8;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 5b50a4f..f7a2330 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -4,7 +4,51 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.28 1995-10-13 12:26:43 adam + * Revision 1.40 1996-02-02 13:44:44 adam + * The public dictionary functions simply use char instead of Dict_char + * to represent search strings. Dict_char is used internally only. + * + * Revision 1.39 1996/01/03 16:22:13 quinn + * operator->roperator + * + * Revision 1.38 1995/12/11 09:12:55 adam + * The rec_get function returns NULL if record doesn't exist - will + * happen in the server if the result set records have been deleted since + * the creation of the set (i.e. the search). + * The server saves a result temporarily if it is 'volatile', i.e. the + * set is register dependent. + * + * Revision 1.37 1995/12/06 15:05:28 adam + * More verbose in count_set. + * + * Revision 1.36 1995/12/06 12:41:27 adam + * New command 'stat' for the index program. + * Filenames can be read from stdin by specifying '-'. + * Bug fix/enhancement of the transformation from terms to regular + * expressons in the search engine. + * + * Revision 1.35 1995/11/27 09:29:00 adam + * Bug fixes regarding conversion to regular expressions. + * + * Revision 1.34 1995/11/16 17:00:56 adam + * Better logging of rpn query. + * + * Revision 1.33 1995/11/01 13:58:28 quinn + * Moving data1 to yaz/retrieval + * + * Revision 1.32 1995/10/27 14:00:11 adam + * Implemented detection of database availability. + * + * Revision 1.31 1995/10/17 18:02:10 adam + * New feature: databases. Implemented as prefix to words in dictionary. + * + * Revision 1.30 1995/10/16 09:32:38 adam + * More work on relational op. + * + * Revision 1.29 1995/10/13 16:01:49 adam + * Work on relations. + * + * Revision 1.28 1995/10/13 12:26:43 adam * Optimization of truncation. * * Revision 1.27 1995/10/12 17:07:22 adam @@ -98,9 +142,10 @@ #include #include #include +#include #include "zserver.h" -#include +#include "attribute.h" #include #include @@ -108,7 +153,8 @@ #include #include -int index_word_prefix_map (char *string, oid_value attrSet, int attrUse) +int index_word_prefix_map (char *string, oid_value attrSet, int attrUse, + char *basename) { attent *attp; @@ -118,46 +164,7 @@ int index_word_prefix_map (char *string, oid_value attrSet, int attrUse) return -1; logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal); return index_word_prefix (string, attp->attset_ordinal, - attp->local_attribute); -} - -/* - * attr_print: log attributes - */ -static void attr_print (Z_AttributesPlusTerm *t) -{ - int of, i; - for (of = 0; of < t->num_attributes; of++) - { - Z_AttributeElement *element; - element = t->attributeList[of]; - - switch (element->which) - { - case Z_AttributeValue_numeric: - logf (LOG_DEBUG, "attributeType=%d value=%d", - *element->attributeType, - *element->value.numeric); - break; - case Z_AttributeValue_complex: - logf (LOG_DEBUG, "attributeType=%d complex", - *element->attributeType); - for (i = 0; ivalue.complex->num_list; i++) - { - if (element->value.complex->list[i]->which == - Z_StringOrNumeric_string) - logf (LOG_DEBUG, " string: '%s'", - element->value.complex->list[i]->u.string); - else if (element->value.complex->list[i]->which == - Z_StringOrNumeric_numeric) - logf (LOG_DEBUG, " numeric: '%d'", - *element->value.complex->list[i]->u.numeric); - } - break; - default: - assert (0); - } - } + attp->local_attribute, basename); } typedef struct { @@ -412,12 +419,14 @@ static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to, rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); #if 0 +/* section that preserve all keys */ heap_delete (ti); if (is_readkey (ispt[n], ti->tmpbuf)) heap_insert (ti, ti->tmpbuf, n); else is_pt_free (ispt[n]); #else +/* section that preserve all keys with unique sysnos */ while (1) { if (!is_readkey (ispt[n], ti->tmpbuf)) @@ -489,20 +498,163 @@ static void add_isam_p (const char *info, struct grep_info *p) (p->isam_p_indx)++; } -static int grep_handle (Dict_char *name, const char *info, void *p) +static int grep_handle (char *name, const char *info, void *p) { logf (LOG_DEBUG, "dict name: %s", name); add_isam_p (info, p); return 0; } +static void gen_regular_rel (char *dst, int val, int islt) +{ + int dst_p = 1; + int w, d, i; + int pos = 0; + char numstr[20]; + + *dst = '('; + sprintf (numstr, "%d", val); + for (w = strlen(numstr); --w >= 0; pos++) + { + d = numstr[w]; + if (pos > 0) + { + if (islt) + { + if (d == '0') + continue; + d--; + } + else + { + if (d == '9') + continue; + d++; + } + } + + strcpy (dst + dst_p, numstr); + dst_p = strlen(dst) - pos - 1; + + if (islt) + { + if (d != '0') + { + dst[dst_p++] = '['; + dst[dst_p++] = '0'; + dst[dst_p++] = '-'; + dst[dst_p++] = d; + dst[dst_p++] = ']'; + } + else + dst[dst_p++] = d; + } + else + { + if (d != '9') + { + dst[dst_p++] = '['; + dst[dst_p++] = d; + dst[dst_p++] = '-'; + dst[dst_p++] = '9'; + dst[dst_p++] = ']'; + } + else + dst[dst_p++] = d; + } + for (i = 0; i="); + gen_regular_rel (term_dict + strlen(term_dict), term_value, 0); + break; + case 5: + if (term_value < 0) + term_value = 0; + logf (LOG_DEBUG, "Relation >"); + gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0); + break; + default: + return 0; + } + logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict); + r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos, + grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); + logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return 1; +} + +static void verbatim_char (int ch, int *indx, char *dst) +{ + if (!isalnum (ch)) + dst[(*indx)++] = '\\'; + dst[(*indx)++] = ch; +} + static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, const char *term_sub, - oid_value attributeSet, struct grep_info *grep_info) + oid_value attributeSet, struct grep_info *grep_info, + int num_bases, char **basenames) { char term_dict[2*IT_MAX_WORD+2]; - int i, j; - const char *info; + int i, j, r, base_no; AttrType truncation; int truncation_value; AttrType use; @@ -518,49 +670,82 @@ static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, if (use_value == -1) use_value = 1016; - i = index_word_prefix_map (term_dict, curAttributeSet, use_value); - if (i < 0) - { - zi->errCode = 114; - return -1; - } - - switch (truncation_value) + + for (base_no = 0; base_no < num_bases; base_no++) { - case -1: /* not specified */ - case 100: /* do not truncate */ - strcat (term_dict, term_sub); - logf (LOG_DEBUG, "dict_lookup: %s", term_dict); - if ((info = dict_lookup (zi->wordDict, term_dict))) - add_isam_p (info, grep_info); - break; - case 1: /* right truncation */ - strcat (term_dict, term_sub); - strcat (term_dict, ".*"); - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle); - break; - case 2: /* left truncation */ - case 3: /* left&right truncation */ - zi->errCode = 120; - return -1; - case 101: /* process # in term */ - for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++) - term_dict[j++] = term_sub[i]; - for (; term_sub[i]; i++) - if (term_sub[i] == '#') + int max_pos; + int prefix_len = index_word_prefix_map (term_dict, curAttributeSet, + use_value, + basenames[base_no]); + if (prefix_len < 0) + { + zi->errCode = 114; + return -1; + } + if (!relational_term (zi, zapt, term_sub, term_dict, + attributeSet, grep_info, &max_pos)) + { + switch (truncation_value) { - term_dict[j++] = '.'; - term_dict[j++] = '*'; + case -1: /* not specified */ + case 100: /* do not truncate */ + j = strlen(term_dict); + term_dict[j++] = '('; + for (i = 0; term_sub[i]; i++) + verbatim_char (term_sub[i], &j, term_dict); + strcpy (term_dict+j, ")"); + r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, + &max_pos, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r); + break; + case 1: /* right truncation */ + j = strlen(term_dict); + term_dict[j++] = '('; + for (i = 0; term_sub[i]; i++) + verbatim_char (term_sub[i], &j, term_dict); + strcpy (term_dict+j, ".*)"); + dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, + &max_pos, grep_handle); + break; + case 2: /* left truncation */ + case 3: /* left&right truncation */ + zi->errCode = 120; + return -1; + case 101: /* process # in term */ + j = strlen(term_dict); + term_dict[j++] = '('; + for (i=0; term_sub[i]; i++) + if (term_sub[i] == '#' && i > 2) + { + term_dict[j++] = '.'; + term_dict[j++] = '*'; + } + else + verbatim_char (term_sub[i], &j, term_dict); + strcpy (term_dict+j, ")"); + r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, + &max_pos, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", + r); + break; + case 102: /* regular expression */ + sprintf (term_dict + strlen(term_dict), "(%s)", term_sub); + r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, + &max_pos, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", + r); + break; } - else - term_dict[j++] = term_sub[i]; - term_dict[j] = '\0'; - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle); - break; - case 102: /* regular expression */ - strcat (term_dict, term_sub); - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle); - break; + } + if (max_pos <= strlen(basenames[base_no])) + { + zi->errCode = 109; /* Database unavailable */ + zi->errString = basenames[base_no]; + return -1; + } } logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); return 0; @@ -582,7 +767,8 @@ static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, static RSET rpn_search_APT_relevance (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet) + oid_value attributeSet, + int num_bases, char **basenames) { rset_relevance_parms parms; char termz[IT_MAX_WORD+1]; @@ -614,7 +800,8 @@ static RSET rpn_search_APT_relevance (ZServerInfo *zi, } else strcpy (term_sub, p0); - if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info)) + if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info, + num_bases, basenames)) return NULL; if (!p1) break; @@ -634,7 +821,8 @@ static RSET rpn_search_APT_relevance (ZServerInfo *zi, static RSET rpn_search_APT_word (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet) + oid_value attributeSet, + int num_bases, char **basenames) { rset_isam_parms parms; char termz[IT_MAX_WORD+1]; @@ -652,7 +840,8 @@ static RSET rpn_search_APT_word (ZServerInfo *zi, grep_info.isam_p_size = 0; grep_info.isam_p_buf = NULL; - if (trunc_term (zi, zapt, termz, attributeSet, &grep_info)) + if (trunc_term (zi, zapt, termz, attributeSet, &grep_info, + num_bases, basenames)) return NULL; if (grep_info.isam_p_indx < 1) result = rset_create (rset_kind_null, NULL); @@ -757,7 +946,8 @@ static RSET rpn_prox (RSET *rset, int rset_no) static RSET rpn_search_APT_phrase (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet) + oid_value attributeSet, + int num_bases, char **basenames) { char termz[IT_MAX_WORD+1]; char term_sub[IT_MAX_WORD+1]; @@ -787,7 +977,8 @@ static RSET rpn_search_APT_phrase (ZServerInfo *zi, strcpy (term_sub, p0); grep_info.isam_p_indx = 0; - if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info)) + if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info, + num_bases, basenames)) return NULL; if (grep_info.isam_p_indx == 0) rset[rset_no] = rset_create (rset_kind_null, NULL); @@ -851,9 +1042,9 @@ static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, return result; } - static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet) + oid_value attributeSet, + int num_bases, char **basenames) { AttrType relation; AttrType structure; @@ -868,17 +1059,23 @@ static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, { case -1: if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); - return rpn_search_APT_phrase (zi, zapt, attributeSet); + return rpn_search_APT_relevance (zi, zapt, attributeSet, + num_bases, basenames); + return rpn_search_APT_phrase (zi, zapt, attributeSet, + num_bases, basenames); case 1: /* phrase */ if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); - return rpn_search_APT_phrase (zi, zapt, attributeSet); + return rpn_search_APT_relevance (zi, zapt, attributeSet, + num_bases, basenames); + return rpn_search_APT_phrase (zi, zapt, attributeSet, + num_bases, basenames); break; case 2: /* word */ if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); - return rpn_search_APT_word (zi, zapt, attributeSet); + return rpn_search_APT_relevance (zi, zapt, attributeSet, + num_bases, basenames); + return rpn_search_APT_word (zi, zapt, attributeSet, + num_bases, basenames); case 3: /* key */ break; case 4: /* year */ @@ -886,7 +1083,8 @@ static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, case 5: /* date - normalized */ break; case 6: /* word list */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); + return rpn_search_APT_relevance (zi, zapt, attributeSet, + num_bases, basenames); case 100: /* date - un-normalized */ break; case 101: /* name - normalized */ @@ -898,13 +1096,16 @@ static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, case 104: /* urx */ break; case 105: /* free-form-text */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); + return rpn_search_APT_relevance (zi, zapt, attributeSet, + num_bases, basenames); case 106: /* document-text */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); + return rpn_search_APT_relevance (zi, zapt, attributeSet, + num_bases, basenames); case 107: /* local-number */ return rpn_search_APT_local (zi, zapt, attributeSet); case 108: /* string */ - return rpn_search_APT_word (zi, zapt, attributeSet); + return rpn_search_APT_word (zi, zapt, attributeSet, + num_bases, basenames); case 109: /* numeric string */ break; } @@ -922,37 +1123,45 @@ static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId) } static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, - oid_value attributeSet) + oid_value attributeSet, + int num_bases, char **basenames) { RSET r = NULL; if (zs->which == Z_RPNStructure_complex) { rset_bool_parms bool_parms; + int soft = 0; bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1, - attributeSet); + attributeSet, + num_bases, basenames); if (bool_parms.rset_l == NULL) return NULL; + if (rset_is_ranked(bool_parms.rset_l)) + soft = 1; bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2, - attributeSet); + attributeSet, + num_bases, basenames); if (bool_parms.rset_r == NULL) { rset_delete (bool_parms.rset_l); return NULL; } + if (rset_is_ranked(bool_parms.rset_r)) + soft = 1; bool_parms.key_size = sizeof(struct it_key); bool_parms.cmp = key_compare; - switch (zs->u.complex->operator->which) + switch (zs->u.complex->roperator->which) { case Z_Operator_and: - r = rset_create (rset_kind_and, &bool_parms); + r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms); break; case Z_Operator_or: - r = rset_create (rset_kind_or, &bool_parms); + r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms); break; case Z_Operator_and_not: - r = rset_create (rset_kind_not, &bool_parms); + r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms); break; default: assert (0); @@ -964,7 +1173,7 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, { logf (LOG_DEBUG, "rpn_search_APT"); r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm, - attributeSet); + attributeSet, num_bases, basenames); } else if (zs->u.simple->which == Z_Operand_resultSetId) { @@ -983,13 +1192,46 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, return r; } +void count_set_save (RSET *r, int *count) +{ + int psysno = 0; + int kno = 0; + struct it_key key; + RSFD rfd, wfd; + RSET w; + rset_temp_parms parms; + + logf (LOG_DEBUG, "count_set_save"); + *count = 0; + parms.key_size = sizeof(struct it_key); + 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); + while (rset_read (*r, rfd, &key)) + { + if (key.sysno != psysno) + { + rset_write (w, wfd, &key); + psysno = key.sysno; + (*count)++; + } + kno++; + } + rset_close (*r, rfd); + rset_delete (*r); + rset_close (w, wfd); + *r = w; + logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count); +} + static void count_set (RSET r, int *count) { int psysno = 0; + int kno = 0; struct it_key key; RSFD rfd; - logf (LOG_DEBUG, "rpn_save_set"); + logf (LOG_DEBUG, "count_set"); *count = 0; rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO); while (rset_read (r, rfd, &key)) @@ -999,9 +1241,10 @@ static void count_set (RSET r, int *count) psysno = key.sysno; (*count)++; } + kno++; } rset_close (r, rfd); - logf (LOG_DEBUG, "%d distinct sysnos", *count); + logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count); } int rpn_search (ZServerInfo *zi, @@ -1012,16 +1255,21 @@ int rpn_search (ZServerInfo *zi, oident *attrset; oid_value attributeSet; + zlog_rpn (rpn); + zi->errCode = 0; zi->errString = NULL; - + attrset = oid_getentbyoid (rpn->attributeSetId); attributeSet = attrset->value; - - rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet); + rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet, + num_bases, basenames); if (!rset) return zi->errCode; - count_set (rset, hits); + if (rset_is_volatile(rset)) + count_set_save(&rset,hits); + else + count_set (rset, hits); resultSetAdd (zi, setname, 1, rset); if (zi->errCode) logf (LOG_DEBUG, "search error: %d", zi->errCode); @@ -1036,8 +1284,7 @@ struct scan_info { char prefix[20]; }; -static int scan_handle (Dict_char *name, const char *info, int pos, - void *client) +static int scan_handle (char *name, const char *info, int pos, void *client) { int len_prefix, idx; ISAM_P isam_p; @@ -1071,11 +1318,18 @@ static int scan_handle (Dict_char *name, const char *info, int pos, return 0; } -int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt, + +static int dummy_handle (char *name, const char *info, void *p) +{ + return 0; +} + +int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, + int num_bases, char **basenames, int *position, int *num_entries, struct scan_entry **list, int *status) { - int i, j, sizez; + int i, j, sizez, max_pos; int pos = *position; int num = *num_entries; int before; @@ -1087,14 +1341,18 @@ int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt, struct scan_info scan_info; logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num); + + if (num_bases != 1) + return 111; scan_info.before = before = pos-1; scan_info.after = after = 1+num-pos; - scan_info.odr = odr; + scan_info.odr = zi->odr; logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after); scan_info.isam = zi->wordIsam; - scan_info.list = odr_malloc (odr, (before+after)*sizeof(*scan_info.list)); + scan_info.list = odr_malloc (zi->odr, (before+after)* + sizeof(*scan_info.list)); for (j = 0; jwordDict, termz, 0, NULL, &max_pos, + dummy_handle); + if (max_pos <= strlen(*basenames)) + { + zi->errString = *basenames; + return zi->errCode = 109; /* Database unavailable */ + } strcpy (scan_info.prefix, termz); sizez = term->u.general->len; if (sizez > IT_MAX_WORD)