X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;ds=sidebyside;f=index%2Fzrpn.c;h=8d5da192e68cf52847e9d80a766f3c71c6c0458c;hb=0ac0f444414651f31406b227b64a447fd324ac4a;hp=7ef2f446c5f5debb8795db47c4b0f4eaafd81239;hpb=80eb0a73f541530f383715296203036b3adbacc2;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 7ef2f44..8d5da19 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -4,7 +4,19 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.51 1996-06-11 10:54:15 quinn + * 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 @@ -185,7 +197,6 @@ #include "attribute.h" #include -#include #include #include #include @@ -252,248 +263,6 @@ 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; -}; - -static void heap_swap (struct trunc_info *ti, int i1, int i2) -{ - int swap; - - swap = ti->ptr[i1]; - ti->ptr[i1] = ti->ptr[i2]; - ti->ptr[i2] = swap; -} - -static void heap_delete (struct trunc_info *ti) -{ - int cur = 1, child = 2; - - heap_swap (ti, 1, 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) - { - heap_swap (ti, cur, child); - cur = child; - child = 2*cur; - } - else - break; - } -} - -static void heap_insert (struct trunc_info *ti, const char *buf, int indx) -{ - 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; - } -} - -static -struct trunc_info *heap_init (int size, int key_size, - int (*cmp)(const void *p1, const void *p2)) -{ - struct trunc_info *ti = xmalloc (sizeof(*ti)); - int i; - - ++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; ) - { - ti->ptr[i] = i; - ti->heap[i] = ti->buf + ti->keysize * i; - } - return ti; -} - -static void heap_close (struct trunc_info *ti) -{ - xfree (ti->ptr); - xfree (ti->indx); - xfree (ti->heap); - xfree (ti->swapbuf); - xfree (ti->tmpbuf); - xfree (ti); -} - -static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to, - int merge_chunk) -{ - RSET result; - RSFD result_rsfd; - rset_temp_parms parms; - - parms.key_size = sizeof(struct it_key); - result = rset_create (rset_kind_temp, &parms); - result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); - - if (to - from > merge_chunk) - { - RSFD *rsfd; - RSET *rset; - int i, i_add = (to-from)/merge_chunk + 1; - struct trunc_info *ti; - int rscur = 0; - int rsmax = (to-from)/i_add + 1; - - rset = xmalloc (sizeof(*rset) * rsmax); - rsfd = xmalloc (sizeof(*rsfd) * rsmax); - - for (i = from; i < to; i += i_add) - { - if (i_add <= to - i) - rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add, - merge_chunk); - else - rset[rscur] = rset_trunc_r (isam, isam_p, i, to, - merge_chunk); - rscur++; - } - ti = heap_init (rscur, sizeof(struct it_key), key_compare); - for (i = rscur; --i >= 0; ) - { - 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); - else - { - rset_close (rset[i], rsfd[i]); - rset_delete (rset[i]); - } - } - while (ti->heapnum) - { - int n = ti->indx[ti->ptr[1]]; - - rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); - - while (1) - { - if (!rset_read (rset[n], rsfd[n], ti->tmpbuf)) - { - heap_delete (ti); - rset_close (rset[n], rsfd[n]); - rset_delete (rset[n]); - break; - } - if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1) - { - heap_delete (ti); - heap_insert (ti, ti->tmpbuf, n); - break; - } - } - } - xfree (rset); - xfree (rsfd); - heap_close (ti); - } - else - { - ISPT *ispt; - int i; - struct trunc_info *ti; - - ispt = xmalloc (sizeof(*ispt) * (to-from)); - - ti = heap_init (to-from, sizeof(struct it_key), - key_compare); - for (i = to-from; --i >= 0; ) - { - ispt[i] = is_position (isam, isam_p[from+i]); - if (is_readkey (ispt[i], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, i); - else - is_pt_free (ispt[i]); - } - while (ti->heapnum) - { - int n = ti->indx[ti->ptr[1]]; - - 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)) - { - heap_delete (ti); - is_pt_free (ispt[n]); - break; - } - if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1) - { - heap_delete (ti); - heap_insert (ti, ti->tmpbuf, n); - break; - } - } -#endif - } - heap_close (ti); - xfree (ispt); - } - rset_close (result, result_rsfd); - return result; -} - -static int isam_trunc_cmp (const void *p1, const void *p2) -{ - ISAM_P i1 = *(ISAM_P*) p1; - ISAM_P i2 = *(ISAM_P*) p2; - int d; - - d = is_type (i1) - is_type (i2); - if (d) - return d; - return is_block (i1) - is_block (i2); -} - -static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no) -{ - - qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp); - return rset_trunc_r (isam, isam_p, 0, no, 100); -} - #define TERM_COUNT struct grep_info { @@ -544,19 +313,45 @@ static void add_isam_p (const char *info, struct grep_info *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; } +/* 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 = 1; + int dst_p; int w, d, i; int pos = 0; char numstr[20]; - *dst = '('; + 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) + { + strcpy (dst, "([0-9]+|-"); + dst_p = strlen (dst); + islt = 1; + } + else + { + strcpy (dst, "(-"); + islt = 0; + } + val = -val; + } + dst_p = strlen (dst); sprintf (numstr, "%d", val); for (w = strlen(numstr); --w >= 0; pos++) { @@ -677,7 +472,7 @@ static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, return 0; } logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict); - r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos, + 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); @@ -774,7 +569,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, 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, + 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); @@ -784,7 +579,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, 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, + dict_lookup_grep (zi->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); break; case 2: /* left truncation */ @@ -802,7 +597,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, else verbatim_char (term_sub[i], &j, term_dict); strcpy (term_dict+j, ")"); - r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, + 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", @@ -810,7 +605,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, break; case 102: /* regular expression */ sprintf (term_dict + j, "(%s)", term_sub); - r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, + 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", @@ -825,7 +620,7 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, cp += 2; } sprintf (term_dict + j, "(%s)", cp); - r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info, + r = dict_lookup_grep (zi->dict, term_dict, r, grep_info, &max_pos, j, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d", @@ -856,7 +651,7 @@ static void trans_scan_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, { Z_Term *term = zapt->term; char **map; - const char *cp = (const char *) term->u.general->buf; + char *cp = (char*) term->u.general->buf; const char *cp_end = cp + term->u.general->len; const char *src; int i = 0; @@ -896,7 +691,7 @@ static RSET rpn_search_APT_relevance (ZServerInfo *zi, parms.key_size = sizeof(struct it_key); parms.max_rec = 100; parms.cmp = key_compare; - parms.is = zi->wordIsam; + parms.is = zi->isam; parms.no_terms = 0; if (zapt->term->which != Z_Term_general) @@ -967,7 +762,6 @@ static RSET rpn_search_APT_cphrase (ZServerInfo *zi, 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; @@ -989,17 +783,7 @@ static RSET rpn_search_APT_cphrase (ZServerInfo *zi, if (field_term (zi, zapt, termz, 'p', attributeSet, &grep_info, num_bases, basenames)) return NULL; - if (grep_info.isam_p_indx < 1) - result = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx == 1) - { - parms.is = zi->wordIsam; - parms.pos = *grep_info.isam_p_buf; - result = rset_create (rset_kind_isam, &parms); - } - else - result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf, - grep_info.isam_p_indx); + result = rset_trunc (zi, grep_info.isam_p_buf, grep_info.isam_p_indx); #ifdef TERM_COUNT xfree(grep_info.term_no); #endif @@ -1151,20 +935,8 @@ static RSET rpn_search_APT_phrase (ZServerInfo *zi, if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info, num_bases, basenames)) 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, - grep_info.isam_p_indx); - 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); - } + 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; @@ -1391,6 +1163,7 @@ void count_set_save (RSET *r, int *count) rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO); while (rset_read (*r, rfd, &key)) { + logf (LOG_DEBUG, "sysno=%-7d seqno=%d", key.sysno, key.seqno); if (key.sysno != psysno) { rset_write (w, wfd, &key); @@ -1437,7 +1210,7 @@ int rpn_search (ZServerInfo *zi, oident *attrset; oid_value attributeSet; - dict_grep_cmap (zi->wordDict, map_chrs_input); + dict_grep_cmap (zi->dict, map_chrs_input); zlog_rpn (rpn); zi->errCode = 0; @@ -1531,17 +1304,17 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num); + if (attributeset == VAL_NONE) + attributeset = VAL_BIB1; + attr_init (&use, zapt, 1); - use_value = attr_find (&use, NULL); + use_value = attr_find (&use, &attributeset); logf (LOG_DEBUG, "use value %d", use_value); attr_init (&completeness, zapt, 6); completeness_value = attr_find (&completeness, NULL); logf (LOG_DEBUG, "completeness value %d", completeness_value); - if (attributeset == VAL_NONE) - attributeset = VAL_BIB1; - if (use_value == -1) use_value = 1016; for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++) @@ -1599,7 +1372,7 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, trans_scan_term (zi, zapt, termz+prefix_len); - dict_scan (zi->wordDict, termz, &before_tmp, &after_tmp, scan_info, + dict_scan (zi->dict, termz, &before_tmp, &after_tmp, scan_info, scan_handle); } glist = odr_malloc (zi->odr, (before+after)*sizeof(*glist)); @@ -1613,7 +1386,6 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, const char *mterm = NULL; const char *tst; RSET rset; - rset_isam_parms parms; for (j = 0; j < ord_no; j++) { @@ -1628,9 +1400,7 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, if (j0 == -1) break; scan_term_untrans (zi->odr, &glist[i+before].term, mterm); - parms.is = zi->wordIsam; - parms.pos = scan_info_array[j0].list[ptr[j0]].isam_p; - rset = rset_create (rset_kind_isam, &parms); + rset = rset_trunc (zi, &scan_info_array[j0].list[ptr[j0]].isam_p, 1); ptr[j0]++; for (j = j0+1; jwordIsam; - parms.pos = scan_info_array[j].list[ptr[j]].isam_p; - rset2 = rset_create (rset_kind_isam, &parms); + rset2 = + rset_trunc (zi, &scan_info_array[j].list[ptr[j]].isam_p, 1); bool_parms.key_size = sizeof(struct it_key); bool_parms.cmp = key_compare; @@ -1675,7 +1443,6 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, const char *mterm = NULL; const char *tst; RSET rset; - rset_isam_parms parms; for (j = 0; j odr, &glist[before-1-i].term, mterm); - parms.is = zi->wordIsam; - parms.pos = scan_info_array[j0].list[before-1-ptr[j0]].isam_p; - rset = rset_create (rset_kind_isam, &parms); + rset = rset_trunc + (zi, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1); ptr[j0]++; @@ -1704,13 +1470,11 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, (tst=scan_info_array[j].list[before-1-ptr[j]].term) && !strcmp (tst, mterm)) { - rset_isam_parms parms; rset_bool_parms bool_parms; RSET rset2; - parms.is = zi->wordIsam; - parms.pos = scan_info_array[j].list[before-1-ptr[j]].isam_p; - rset2 = rset_create (rset_kind_isam, &parms); + rset2 = rset_trunc (zi, + &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1); bool_parms.key_size = sizeof(struct it_key); bool_parms.cmp = key_compare;