X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=2a45b7d90839feba3c49fef8c653f4bd2c16491c;hb=9cf5dedda26c0ebaf38daa6fa3f15a763d6b2fcb;hp=c17ddfc450792aa13625d5d82df1235e37206b9a;hpb=c12de4a26665bcab5365d45a8ffc62470e71e9aa;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index c17ddfc..2a45b7d 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,10 +1,122 @@ /* - * Copyright (C) 1994-1995, Index Data I/S + * Copyright (C) 1994-1996, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.27 1995-10-12 17:07:22 adam + * 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. + * + * Revision 1.57 1996/11/11 13:38:02 adam + * Added proximity support in search. + * + * Revision 1.56 1996/11/08 11:10:32 adam + * Buffers used during file match got bigger. + * Compressed ISAM support everywhere. + * Bug fixes regarding masking characters in queries. + * Redesigned Regexp-2 queries. + * + * Revision 1.55 1996/11/04 14:07:44 adam + * Moved truncation code to trunc.c. + * + * Revision 1.54 1996/10/29 14:09:52 adam + * Use of cisam system - enabled if setting isamc is 1. + * + * Revision 1.53 1996/06/26 09:21:43 adam + * Bug fix: local attribute set wasn't obeyed in scan. + * + * Revision 1.52 1996/06/17 14:26:20 adam + * Function gen_regular_rel changed to handle negative numbers. + * + * Revision 1.51 1996/06/11 10:54:15 quinn + * Relevance work + * + * Revision 1.50 1996/06/07 08:51:53 adam + * Bug fix: Character mapping was broken (introducued by last revision). + * + * Revision 1.49 1996/06/04 10:18:11 adam + * Search/scan uses character mapping module. + * + * Revision 1.48 1996/05/28 15:15:01 adam + * Bug fix: Didn't handle unknown database correctly. + * + * Revision 1.47 1996/05/15 18:36:28 adam + * Function trans_term transforms unsearchable characters to blanks. + * + * Revision 1.46 1996/05/15 11:57:56 adam + * Fixed bug introduced by set/field mapping in search operations. + * + * Revision 1.45 1996/05/14 11:34:00 adam + * Scan support in multiple registers/databases. + * + * Revision 1.44 1996/05/14 06:16:44 adam + * Compact use/set bytes used in search service. + * + * Revision 1.43 1996/05/09 09:54:43 adam + * Server supports maps from one logical attributes to a list of physical + * attributes. + * The extraction process doesn't make space consuming 'any' keys. + * + * Revision 1.42 1996/05/09 07:28:56 quinn + * Work towards phrases and multiple registers + * + * Revision 1.41 1996/03/20 09:36:43 adam + * Function dict_lookup_grep got extra parameter, init_pos, which marks + * from which position in pattern approximate pattern matching should occur. + * Approximate pattern matching is used in relevance=re-2. + * + * 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 * Truncation works. * * Revision 1.26 1995/10/12 12:40:54 adam @@ -95,68 +207,17 @@ #include #include #include +#include #include "zserver.h" -#include +#include "attribute.h" -#include +#include #include #include #include #include -int index_word_prefix_map (char *string, oid_value attrSet, int attrUse) -{ - attent *attp; - - logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse); - attp = att_getentbyatt (attrSet, attrUse); - if (!attp) - 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); - } - } -} - typedef struct { int type; int major; @@ -218,329 +279,570 @@ static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, src->minor = 0; } -struct trunc_info { - int *ptr; - int *indx; - char **heap; - int heapnum; - int (*cmp)(const void *p1, const void *p2); - int keysize; - char *swapbuf; - char *tmpbuf; - char *buf; -}; +#define TERM_COUNT + +struct grep_info { +#ifdef TERM_COUNT + int *term_no; +#endif + ISAM_P *isam_p_buf; + int isam_p_size; + int isam_p_indx; +}; -static void heap_swap (struct trunc_info *ti, int i1, int i2) +static void add_isam_p (const char *info, struct grep_info *p) { - int swap; - - swap = ti->ptr[i1]; - ti->ptr[i1] = ti->ptr[i2]; - ti->ptr[i2] = swap; -} + if (p->isam_p_indx == p->isam_p_size) + { + ISAM_P *new_isam_p_buf; +#ifdef TERM_COUNT + int *new_term_no; +#endif + + p->isam_p_size = 2*p->isam_p_size + 100; + new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) * + p->isam_p_size); + if (p->isam_p_buf) + { + memcpy (new_isam_p_buf, p->isam_p_buf, + p->isam_p_indx * sizeof(*p->isam_p_buf)); + xfree (p->isam_p_buf); + } + p->isam_p_buf = new_isam_p_buf; -static void heap_delete (struct trunc_info *ti) -{ - int cur = 1, child = 2; - - assert (ti->heapnum > 0); - - heap_swap (ti, 1, ti->heapnum); - ti->heapnum--; - while (child <= ti->heapnum) { - if (child < ti->heapnum && - (*ti->cmp)(ti->heap[ti->ptr[child]], - ti->heap[ti->ptr[1+child]]) > 0) - child++; - if ((*ti->cmp)(ti->heap[ti->ptr[cur]], - ti->heap[ti->ptr[child]]) > 0) +#ifdef TERM_COUNT + new_term_no = xmalloc (sizeof(*new_term_no) * + p->isam_p_size); + if (p->term_no) { - heap_swap (ti, cur, child); - cur = child; - child = 2*cur; + memcpy (new_term_no, p->isam_p_buf, + p->isam_p_indx * sizeof(*p->term_no)); + xfree (p->term_no); } - else - break; + p->term_no = new_term_no; +#endif } + assert (*info == sizeof(*p->isam_p_buf)); + memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); + (p->isam_p_indx)++; } -static void heap_insert (struct trunc_info *ti, const char *buf, int indx) +static int grep_handle (char *name, const char *info, void *p) { - int cur, parent; - - cur = ++(ti->heapnum); - memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize); - ti->indx[ti->ptr[cur]] = indx; - parent = cur/2; - while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]], - ti->heap[ti->ptr[cur]]) > 0) - { - heap_swap (ti, cur, parent); - cur = parent; - parent = cur/2; - } + add_isam_p (info, p); + return 0; } -static -struct trunc_info *heap_init (int size, int key_size, - int (*cmp)(const void *p1, const void *p2)) +static int term_pre (char **src, const char *ct1, const char *ct2) { - struct trunc_info *ti = xmalloc (sizeof(*ti)); - int i; + char *s1, *s0 = *src; + char **map; - ++size; - ti->heapnum = 0; - ti->keysize = key_size; - ti->cmp = cmp; - ti->indx = xmalloc (size * sizeof(*ti->indx)); - ti->heap = xmalloc (size * sizeof(*ti->heap)); - ti->ptr = xmalloc (size * sizeof(*ti->ptr)); - ti->swapbuf = xmalloc (ti->keysize); - ti->tmpbuf = xmalloc (ti->keysize); - ti->buf = xmalloc (size * ti->keysize); - for (i = size; --i >= 0; ) + /* skip white space */ + while (*s0) { - ti->ptr[i] = i; - ti->heap[i] = ti->buf + ti->keysize * i; + if (ct1 && strchr (ct1, *s0)) + break; + if (ct2 && strchr (ct2, *s0)) + break; + s1 = s0; + map = map_chrs_input (&s1, strlen(s1)); + if (**map != *CHR_SPACE) + break; + s0 = s1; } - return ti; + *src = s0; + return *s0; } -static void heap_close (struct trunc_info *ti) +static int term_100 (char **src, char *dst, int space_split) { - xfree (ti->ptr); - xfree (ti->indx); - xfree (ti->heap); - xfree (ti->swapbuf); - xfree (ti->tmpbuf); - xfree (ti); + char *s0, *s1, **map; + int i = 0; + + if (!term_pre (src, NULL, NULL)) + return 0; + s0 = *src; + while (*s0) + { + s1 = s0; + map = map_chrs_input (&s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (!isalnum (*s1)) + dst[i++] = '\\'; + dst[i++] = *s1++; + } + } + dst[i] = '\0'; + *src = s0; + return i; } -static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int from, int to, - int merge_chunk) +static int term_101 (char **src, char *dst, int space_split) { - RSET result; - RSFD result_rsfd; - rset_temp_parms parms; - - logf (LOG_DEBUG, "rset_trunc, range=%d-%d", from, to-1); - - parms.key_size = sizeof(struct it_key); - result = rset_create (rset_kind_temp, &parms); - result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); + char *s0, *s1, **map; + int i = 0; - if (to - from > merge_chunk) + if (!term_pre (src, "#", "#")) + return 0; + s0 = *src; + while (*s0) { - RSFD *rsfd; - RSET *rset; - int i, i_add = (to-from)/merge_chunk + 1; - struct trunc_info *ti; - int rscur = 0; - - rset = xmalloc (sizeof(*rset) * merge_chunk); - rsfd = xmalloc (sizeof(*rsfd) * merge_chunk); - - for (i = from; i < to; i += i_add) + if (*s0 == '#') { - if (i_add <= to - i) - rset[rscur] = rset_trunc (isam, isam_p, i, i+i_add, i_add); - else - rset[rscur] = rset_trunc (isam, isam_p, i, to, to-i); - rscur++; + dst[i++] = '.'; + dst[i++] = '*'; + s0++; } - ti = heap_init (rscur, sizeof(struct it_key), key_compare); - for (i = rscur; --i >= 0; ) + else { - rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO); - if (rset_read (rset[i], rsfd[i], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, i); + s1 = s0; + map = map_chrs_input (&s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (!isalnum (*s1)) + dst[i++] = '\\'; + dst[i++] = *s1++; + } } - while (ti->heapnum) - { - int n = ti->indx[ti->ptr[1]]; + } + dst[i] = '\0'; + *src = s0; + return i; +} - rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); - heap_delete (ti); - if (rset_read (rset[n], rsfd[n], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, n); - } - for (i = rscur; --i >= 0; ) + +static int term_103 (char **src, char *dst, int *errors, int space_split) +{ + int i = 0; + char *s0, *s1, **map; + + if (!term_pre (src, "\\()[].*+?|", "(")) + return 0; + s0 = *src; + if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] && + isdigit (s0[1])) + { + *errors = s0[1] - '0'; + s0 += 3; + if (*errors > 3) + *errors = 3; + } + while (*s0) + { + if (strchr ("\\()[].*+?|-", *s0)) + dst[i++] = *s0++; + else { - rset_close (rset[i], rsfd[i]); - rset_delete (rset[i]); + s1 = s0; + map = map_chrs_input (&s0, strlen(s0)); + if (**map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (!isalnum (*s1)) + dst[i++] = '\\'; + dst[i++] = *s1++; + } } - xfree (rset); - xfree (rsfd); - heap_close (ti); } - else - { - ISPT *ispt; - int i; - struct trunc_info *ti; + dst[i] = '\0'; + *src = s0; + return i; +} - ispt = xmalloc (sizeof(*ispt) * (to-from)); +static int term_102 (char **src, char *dst, int space_split) +{ + return term_103 (src, dst, NULL, space_split); +} - ti = heap_init (to-from, sizeof(struct it_key), - key_compare); - for (i = to-from; --i >= 0; ) +/* gen_regular_rel - generate regular expression from relation + * val: border value (inclusive) + * islt: 1 if <=; 0 if >=. + */ +static void gen_regular_rel (char *dst, int val, int islt) +{ + int dst_p; + int w, d, i; + int pos = 0; + char numstr[20]; + + logf (LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt); + if (val >= 0) + { + if (islt) + strcpy (dst, "(-[0-9]+|"); + else + strcpy (dst, "("); + } + else + { + if (!islt) { - ispt[i] = is_position (isam, isam_p[from+i]); - if (is_readkey (ispt[i], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, i); + strcpy (dst, "([0-9]+|-"); + dst_p = strlen (dst); + islt = 1; } - while (ti->heapnum) + else { - int n = ti->indx[ti->ptr[1]]; - - rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); - heap_delete (ti); - if (is_readkey (ispt[n], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, n); + strcpy (dst, "(-"); + islt = 0; } - for (i = to-from; --i >= 0; ) - is_pt_free (ispt[i]); - heap_close (ti); - xfree (ispt); + val = -val; } - rset_close (result, result_rsfd); - return result; -} - -struct grep_info { - ISAM_P *isam_p_buf; - int isam_p_size; - int isam_p_indx; -}; - -static void add_isam_p (const char *info, struct grep_info *p) -{ - if (p->isam_p_indx == p->isam_p_size) + dst_p = strlen (dst); + sprintf (numstr, "%d", val); + for (w = strlen(numstr); --w >= 0; pos++) { - ISAM_P *new_isam_p_buf; + d = numstr[w]; + if (pos > 0) + { + if (islt) + { + if (d == '0') + continue; + d--; + } + else + { + if (d == '9') + continue; + d++; + } + } - p->isam_p_size = 2*p->isam_p_size + 100; - new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) * - p->isam_p_size); - if (p->isam_p_buf) + strcpy (dst + dst_p, numstr); + dst_p = strlen(dst) - pos - 1; + + if (islt) { - memcpy (new_isam_p_buf, p->isam_p_buf, - p->isam_p_indx * sizeof(*p->isam_p_buf)); - xfree (p->isam_p_buf); + 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; } - p->isam_p_buf = new_isam_p_buf; + 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; iisam_p_buf)); - memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); - (p->isam_p_indx)++; + dst[dst_p] = '\0'; + if (islt) + { + for (i=1; i="); + gen_regular_rel (term_dict + strlen(term_dict), term_value, 0); + break; + case 5: + if (!term_100 (term_sub, term_dict, 1)) + return 0; + term_value = atoi (term_dict); + 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->dict, term_dict, 0, grep_info, max_pos, + 0, 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 int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - const char *term_sub, - oid_value attributeSet, struct grep_info *grep_info) +static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, + char **term_sub, int regType, + oid_value attributeSet, struct grep_info *grep_info, + int num_bases, char **basenames, int space_split) { char term_dict[2*IT_MAX_WORD+2]; - int i, j; - const char *info; + int j, r, base_no; AttrType truncation; int truncation_value; AttrType use; int use_value; oid_value curAttributeSet = attributeSet; + 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); 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] == '#') + attent *attp; + data1_local_attribute *local_attr; + int max_pos, prefix_len = 0; + + termp = *term_sub; + attp = att_getentbyatt (curAttributeSet, use_value); + if (!attp) + { + logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", + curAttributeSet, use_value); + zi->errCode = 114; + return -1; + } + if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no])) + { + zi->errCode = 109; /* Database unavailable */ + zi->errString = basenames[base_no]; + return -1; + } + for (local_attr = attp->local_attributes; local_attr; + local_attr = local_attr->next) + { + int ord; + + ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, + local_attr->local); + if (ord < 0) + continue; + if (prefix_len) + term_dict[prefix_len++] = '|'; + else + term_dict[prefix_len++] = '('; + term_dict[prefix_len++] = 1; + term_dict[prefix_len++] = ord; + } + if (!prefix_len) + { + zi->errCode = 114; + return -1; + } + term_dict[prefix_len++] = ')'; + term_dict[prefix_len++] = 1; + term_dict[prefix_len++] = regType; + term_dict[prefix_len] = '\0'; + if (!relational_term (zi, zapt, &termp, term_dict, + attributeSet, grep_info, &max_pos)) + { + j = prefix_len; + logf (LOG_LOG, "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"); + switch (truncation_value) { - term_dict[j++] = '.'; - term_dict[j++] = '*'; + case -1: /* not specified */ + case 100: /* do not truncate */ + term_dict[j++] = '('; + logf (LOG_LOG, "termp=%s", termp); + if (!term_100 (&termp, term_dict + j, space_split)) + { + logf (LOG_LOG, "aaaaaaaaaaaaargh"); + return 0; + } + strcat (term_dict, ")"); + r = dict_lookup_grep (zi->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r); + break; + case 1: /* right truncation */ + 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, + &max_pos, 0, grep_handle); + break; + case 2: /* left truncation */ + case 3: /* left&right truncation */ + zi->errCode = 120; + return -1; + case 101: /* process # in term */ + 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, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", r); + break; + case 102: /* Regexp-1 */ + 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); + r = dict_lookup_grep (zi->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", + r); + break; + case 103: /* Regexp-1 */ + r = 1; + term_dict[j++] = '('; + if (!term_103 (&termp, term_dict + j, &r, space_split)) + return 0; + strcat (term_dict, ")"); + logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r); + r = dict_lookup_grep (zi->dict, term_dict, r, grep_info, + &max_pos, 2, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %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; + } } + *term_sub = termp; logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); - return 0; + return 1; } static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, char *termz) { - size_t i, sizez; + size_t sizez; Z_Term *term = zapt->term; sizez = term->u.general->len; - if (sizez > IT_MAX_WORD) - sizez = IT_MAX_WORD; - for (i = 0; i < sizez; i++) - termz[i] = index_char_cvt (term->u.general->buf[i]); + if (sizez > IT_MAX_WORD-1) + sizez = IT_MAX_WORD-1; + memcpy (termz, term->u.general->buf, sizez); + termz[sizez] = '\0'; +} + +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 *cp_end = cp + term->u.general->len; + const char *src; + int i = 0; + int prev_space = 0; + int len; + + while ((len = (cp_end - cp)) > 0) + { + map = map_chrs_input (&cp, len); + if (**map == *CHR_SPACE) + { + if (prev_space) + continue; + prev_space = 1; + } + else + prev_space = 0; + for (src = *map; *src; src++) + termz[i++] = *src; + } termz[i] = '\0'; } 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]; - char term_sub[IT_MAX_WORD+1]; + char *termp = termz; struct grep_info grep_info; - char *p0 = termz, *p1 = NULL; RSET result; + int term_index = 0; + int r; parms.key_size = sizeof(struct it_key); - parms.max_rec = 100; - parms.cmp = key_compare; - parms.is = zi->wordIsam; + parms.max_rec = 1000; + parms.cmp = key_compare_it; + parms.is = zi->isam; + parms.isc = zi->isamc; + parms.no_terms = 0; if (zapt->term->which != Z_Term_general) { @@ -548,44 +850,49 @@ static RSET rpn_search_APT_relevance (ZServerInfo *zi, return NULL; } trans_term (zi, zapt, termz); + +#ifdef TERM_COUNT + grep_info.term_no = 0; +#endif grep_info.isam_p_indx = 0; grep_info.isam_p_size = 0; grep_info.isam_p_buf = NULL; while (1) { - if ((p1 = strchr (p0, ' '))) - { - memcpy (term_sub, p0, p1-p0); - term_sub[p1-p0] = '\0'; - } - else - strcpy (term_sub, p0); - if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info)) - return NULL; - if (!p1) + r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info, + num_bases, basenames, 1); + if (r <= 0) break; - p0 = p1; - while (*++p0 == ' ') - ; +#ifdef TERM_COUNT + for (; term_index < grep_info.isam_p_indx; term_index++) + grep_info.term_no[term_index] = parms.no_terms; + parms.no_terms++; +#endif } + parms.term_no = grep_info.term_no; parms.isam_positions = grep_info.isam_p_buf; parms.no_isam_positions = grep_info.isam_p_indx; if (grep_info.isam_p_indx > 0) result = rset_create (rset_kind_relevance, &parms); else result = rset_create (rset_kind_null, NULL); +#ifdef TERM_COUNT + xfree(grep_info.term_no); +#endif xfree (grep_info.isam_p_buf); return result; } -static RSET rpn_search_APT_word (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet) +static RSET rpn_search_APT_cphrase (ZServerInfo *zi, + Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + int num_bases, char **basenames) { - rset_isam_parms parms; char termz[IT_MAX_WORD+1]; struct grep_info grep_info; RSET result; + char *termp = termz; + int r; if (zapt->term->which != Z_Term_general) { @@ -594,24 +901,109 @@ static RSET rpn_search_APT_word (ZServerInfo *zi, } trans_term (zi, zapt, termz); +#ifdef TERM_COUNT + grep_info.term_no = 0; +#endif grep_info.isam_p_indx = 0; grep_info.isam_p_size = 0; grep_info.isam_p_buf = NULL; - if (trunc_term (zi, zapt, termz, attributeSet, &grep_info)) - return NULL; - if (grep_info.isam_p_indx < 1) - result = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx == 1) + r = field_term (zi, zapt, &termp, 'p', attributeSet, &grep_info, + 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); +#endif + xfree (grep_info.isam_p_buf); + return result; +} + +static RSET rpn_proximity (RSET rset1, RSET rset2, int ordered, + int exclusion, int relation, int distance) +{ + int i; + RSFD rsfd1, rsfd2; + int more1, more2; + struct it_key buf1, buf2; + RSFD rsfd_result; + RSET result; + rset_temp_parms parms; + + rsfd1 = rset_open (rset1, RSETF_READ|RSETF_SORT_SYSNO); + more1 = rset_read (rset1, rsfd1, &buf1); + + rsfd2 = rset_open (rset2, RSETF_READ|RSETF_SORT_SYSNO); + more2 = rset_read (rset2, rsfd2, &buf2); + + parms.key_size = sizeof (struct it_key); + result = rset_create (rset_kind_temp, &parms); + rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); + + logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d", + exclusion, ordered, relation, distance); + while (more1 && more2) { - parms.is = zi->wordIsam; - parms.pos = *grep_info.isam_p_buf; - result = rset_create (rset_kind_isam, &parms); + int cmp = key_compare_it (&buf1, &buf2); + if (cmp < -1) + more1 = rset_read (rset1, rsfd1, &buf1); + else if (cmp > 1) + more2 = rset_read (rset2, rsfd2, &buf2); + else + { + int sysno = buf1.sysno; + int seqno[500]; + int n = 0; + + seqno[n++] = buf1.seqno; + while ((more1 = rset_read (rset1, rsfd1, &buf1)) && + sysno == buf1.sysno) + if (n < 500) + seqno[n++] = buf1.seqno; + do + { + for (i = 0; i= */ + if (diff >= distance) + excl = !excl; + break; + case 5: /* > */ + if (diff > distance) + excl = !excl; + break; + case 6: /* != */ + if (diff != distance) + excl = !excl; + break; + } + if (excl) + rset_write (result, rsfd_result, &buf2); + } + } while ((more2 = rset_read (rset2, rsfd2, &buf2)) && + sysno == buf2.sysno); + } } - else - result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf, 0, - grep_info.isam_p_indx, 400); - xfree (grep_info.isam_p_buf); + rset_close (result, rsfd_result); + rset_close (rset1, rsfd1); + rset_close (rset2, rsfd2); return result; } @@ -662,7 +1054,7 @@ static RSET rpn_prox (RSET *rset, int rset_no) *more = 0; break; } - cmp = key_compare (buf[i], buf[i-1]); + cmp = key_compare_it (buf[i], buf[i-1]); if (cmp > 1) { more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]); @@ -703,13 +1095,13 @@ 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]; - char *p0 = termz, *p1 = NULL; + char *termp = termz; RSET rset[60], result; - int i, rset_no = 0; + int i, r, rset_no = 0; struct grep_info grep_info; if (zapt->term->which != Z_Term_general) @@ -719,51 +1111,33 @@ static RSET rpn_search_APT_phrase (ZServerInfo *zi, } trans_term (zi, zapt, termz); +#ifdef TERM_COUNT + grep_info.term_no = 0; +#endif grep_info.isam_p_size = 0; grep_info.isam_p_buf = NULL; while (1) { - if ((p1 = strchr (p0, ' '))) - { - memcpy (term_sub, p0, p1-p0); - term_sub[p1-p0] = '\0'; - } - else - strcpy (term_sub, p0); - grep_info.isam_p_indx = 0; - if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info)) - return NULL; - if (grep_info.isam_p_indx == 0) - rset[rset_no] = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx > 1) - rset[rset_no] = rset_trunc (zi->wordIsam, - grep_info.isam_p_buf, 0, - grep_info.isam_p_indx, 400); - else - { - rset_isam_parms parms; - - parms.is = zi->wordIsam; - parms.pos = *grep_info.isam_p_buf; - rset[rset_no] = rset_create (rset_kind_isam, &parms); - } + r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info, + num_bases, basenames, 1); + if (r < 1) + break; + rset[rset_no] = rset_trunc (zi, grep_info.isam_p_buf, + grep_info.isam_p_indx); assert (rset[rset_no]); if (++rset_no >= sizeof(rset)/sizeof(*rset)) break; - if (!p1) - break; - p0 = p1; - while (*++p0 == ' ') - ; } +#ifdef TERM_COUNT + xfree(grep_info.term_no); +#endif xfree (grep_info.isam_p_buf); if (rset_no == 0) return rset_create (rset_kind_null, NULL); else if (rset_no == 1) return (rset[0]); - result = rpn_prox (rset, rset_no); for (i = 0; iwhich == Z_RPNStructure_complex) { + Z_Operator *zop = zs->u.complex->roperator; 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; + bool_parms.cmp = key_compare_it; - switch (zs->u.complex->operator->which) + switch (zop->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; + case Z_Operator_prox: + if (zop->u.prox->which != Z_ProxCode_known) + { + zi->errCode = 132; + return NULL; + } + if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word) + { + static char val[16]; + zi->errCode = 132; + zi->errString = val; + sprintf (val, "%d", *zop->u.prox->proximityUnitCode); + return NULL; + } + r = rpn_proximity (bool_parms.rset_l, bool_parms.rset_r, + *zop->u.prox->ordered, + (!zop->u.prox->exclusion ? 0 : + *zop->u.prox->exclusion), + *zop->u.prox->relationType, + *zop->u.prox->distance); break; default: - assert (0); + zi->errCode = 110; + return NULL; } } else if (zs->which == Z_RPNStructure_simple) @@ -910,7 +1339,7 @@ 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) { @@ -919,23 +1348,59 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, } else { - assert (0); + zi->errCode = 3; + return NULL; } } else { - assert (0); + zi->errCode = 3; + return NULL; } 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) + { + if (*count < 400) + rset_write (w, wfd, &key); + (*count)++; + psysno = key.sysno; + } + 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)) @@ -945,9 +1410,10 @@ 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, @@ -958,40 +1424,45 @@ int rpn_search (ZServerInfo *zi, oident *attrset; oid_value attributeSet; + dict_grep_cmap (zi->dict, map_chrs_input); + 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); return zi->errCode; } +struct scan_info_entry { + char *term; + ISAM_P isam_p; +}; + struct scan_info { - struct scan_entry *list; + struct scan_info_entry *list; ODR odr; int before, after; - ISAM isam; 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; - RSET rset; struct scan_info *scan_info = client; - rset_isam_parms parms; - len_prefix = strlen(scan_info->prefix); if (memcmp (name, scan_info->prefix, len_prefix)) return 1; @@ -1002,91 +1473,248 @@ static int scan_handle (Dict_char *name, const char *info, int pos, scan_info->list[idx].term = odr_malloc (scan_info->odr, strlen(name + len_prefix)+1); strcpy (scan_info->list[idx].term, name + len_prefix); - assert (*info == sizeof(isam_p)); - memcpy (&isam_p, info+1, sizeof(isam_p)); - parms.is = scan_info->isam; - parms.pos = isam_p; -#if 1 - rset = rset_create (rset_kind_isam, &parms); - count_set (rset, &scan_info->list[idx].occurrences); - rset_delete (rset); -#else - scan_info->list[idx].occurrences = 1; -#endif - logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name); + assert (*info == sizeof(ISAM_P)); + memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P)); return 0; } -int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt, + +static void scan_term_untrans (ODR odr, char **dstp, const char *src) +{ + char *dst = odr_malloc (odr, strlen(src)*2+1); + *dstp = dst; + + while (*src) + { + const char *cp = map_chrs_output (&src); + while (*cp) + *dst++ = *cp++; + } + *dst = '\0'; +} + +int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, + oid_value attributeset, + int num_bases, char **basenames, int *position, int *num_entries, struct scan_entry **list, int *status) { - int i, j, sizez; + int i; int pos = *position; int num = *num_entries; int before; int after; + int base_no; char termz[IT_MAX_WORD+20]; AttrType use; int use_value; - Z_Term *term = zapt->term; - struct scan_info scan_info; + AttrType completeness; + int completeness_value; + struct scan_info *scan_info_array; + struct scan_entry *glist; + int ords[32], ord_no = 0; + int ptr[32]; logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num); - scan_info.before = before = pos-1; - scan_info.after = after = 1+num-pos; - scan_info.odr = 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)); - for (j = 0; ju.general->len; - if (sizez > IT_MAX_WORD) - sizez = IT_MAX_WORD; - for (j = 0; ju.general->buf[j]); - termz[j+i] = '\0'; - - dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle); + for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++) + { + attent *attp; + data1_local_attribute *local_attr; - *status = BEND_SCAN_SUCCESS; + attp = att_getentbyatt (attributeset, use_value); + if (!attp) + { + logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", + attributeset, use_value); + return zi->errCode = 114; + } + if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no])) + { + zi->errString = basenames[base_no]; + return zi->errCode = 109; /* Database unavailable */ + } + for (local_attr = attp->local_attributes; local_attr && ord_no < 32; + local_attr = local_attr->next) + { + int ord; - for (i = 0; izti, attp->attset_ordinal, + local_attr->local); + if (ord > 0) + ords[ord_no++] = ord; + } + } + if (ord_no == 0) + return zi->errCode = 113; + before = pos-1; + after = 1+num-pos; + scan_info_array = odr_malloc (zi->odr, ord_no * sizeof(*scan_info_array)); + for (i = 0; i < ord_no; i++) + { + int j, prefix_len = 0; + int before_tmp = before, after_tmp = after; + struct scan_info *scan_info = scan_info_array + i; + + scan_info->before = before; + scan_info->after = after; + scan_info->odr = zi->odr; + + scan_info->list = odr_malloc (zi->odr, (before+after)* + sizeof(*scan_info->list)); + for (j = 0; jlist[j].term = NULL; + termz[prefix_len++] = ords[i]; + termz[prefix_len++] = + (completeness_value==2 || completeness_value==3) ? 'p': 'w'; + termz[prefix_len] = 0; + strcpy (scan_info->prefix, termz); + + trans_scan_term (zi, zapt, termz+prefix_len); + + dict_scan (zi->dict, termz, &before_tmp, &after_tmp, scan_info, + scan_handle); + } + glist = odr_malloc (zi->odr, (before+after)*sizeof(*glist)); + for (i = 0; i < ord_no; i++) + ptr[i] = before; + + *status = BEND_SCAN_SUCCESS; + for (i = 0; iodr, &glist[i+before].term, mterm); + rset = rset_trunc (zi, &scan_info_array[j0].list[ptr[j0]].isam_p, 1); + + ptr[j0]++; + for (j = j0+1; j 0)) + { + j0 = j; + mterm = tst; + } + } + if (j0 == -1) break; - if (i) - *status = BEND_SCAN_PARTIAL; - *position -= i; - *num_entries -= i; - *list = scan_info.list+i; /* list is set to first 'real' entry */ + scan_term_untrans (zi->odr, &glist[before-1-i].term, mterm); + + rset = rset_trunc + (zi, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1); + + ptr[j0]++; + + for (j = j0+1; jerrCode = 114; /* if no entries was found */ + ptr[j]++; + } + } + count_set (rset, &glist[before-1-i].occurrences); + rset_delete (rset); + } + i = before-i; + if (i) + { + *status = BEND_SCAN_PARTIAL; + *position -= i; + *num_entries -= i; + } + *list = glist + i; /* list is set to first 'real' entry */ + logf (LOG_DEBUG, "position = %d, num_entries = %d", *position, *num_entries); if (zi->errCode) logf (LOG_DEBUG, "scan error: %d", zi->errCode); - return 0; + return zi->errCode; } - -