X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzsets.c;h=cbdc8b177a020af4b7853f70b55fc7ccd0b7f7f1;hp=aa2d16dc4c9156283199e59166f6e6f3c2aad3a5;hb=5437b50633032595afe6f87dc0f989bc92a5aea8;hpb=af5941499d0d0fe3f2d41f5aa1f82d6a04190349 diff --git a/index/zsets.c b/index/zsets.c index aa2d16d..cbdc8b1 100644 --- a/index/zsets.c +++ b/index/zsets.c @@ -1,4 +1,4 @@ -/* $Id: zsets.c,v 1.58 2004-08-30 12:32:24 heikki Exp $ +/* $Id: zsets.c,v 1.71 2004-11-19 10:27:09 heikki Exp $ Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 Index Data Aps @@ -30,7 +30,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #endif #include "index.h" -#include +#include #define SORT_IDX_ENTRYSIZE 64 #define ZSET_SORT_MAX_LEVEL 3 @@ -89,7 +89,7 @@ ZebraSet resultSetAddRPN (ZebraHandle zh, NMEM m, zebraSet->locked = 1; zebraSet->rpn = 0; zebraSet->nmem = m; - zebraSet->rset_nmem=nmem_create(); /* FIXME - where to free this ?? */ + zebraSet->rset_nmem=nmem_create(); zebraSet->num_bases = num_bases; zebraSet->basenames = @@ -201,11 +201,13 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) break; if (s) { - yaz_log (LOG_DEBUG, "updating result set %s", name); + yaz_log(YLOG_DEBUG, "updating result set %s", name); if (!ov || s->locked) return NULL; if (s->rset) rset_delete (s->rset); + if (s->rset_nmem) + nmem_destroy (s->rset_nmem); if (s->nmem) nmem_destroy (s->nmem); } @@ -213,7 +215,7 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) { const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000"); - yaz_log (LOG_DEBUG, "adding result set %s", name); + yaz_log(YLOG_DEBUG, "adding result set %s", name); s = (ZebraSet) xmalloc (sizeof(*s)); s->next = zh->sets; zh->sets = s; @@ -255,7 +257,9 @@ ZebraSet resultSetGet (ZebraHandle zh, const char *name) if (!s->term_entries && !s->rset && s->rpn) { NMEM nmem = nmem_create (); - yaz_log (LOG_LOG, "research %s", name); + yaz_log(YLOG_LOG, "research %s", name); + if (!s->rset_nmem) + s->rset_nmem=nmem_create(); s->rset = rpn_search (zh, nmem, s->rset_nmem, s->rpn, s->num_bases, s->basenames, s->name, s); @@ -378,7 +382,7 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, position = positions[i]; if (position > 0 && position <= sort_info->num_entries) { - yaz_log (LOG_DEBUG, "got pos=%d (sorted)", position); + yaz_log(YLOG_DEBUG, "got pos=%d (sorted)", position); sr[i].sysno = sort_info->entries[position-1]->sysno; sr[i].score = sort_info->entries[position-1]->score; } @@ -403,13 +407,9 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, while (num_i < num && positions[num_i] < position) num_i++; rfd = rset_open (rset, RSETF_READ); - while (num_i < num && rset_read (rset, rfd, &key)) + while (num_i < num && rset_read (rfd, &key, 0)) { -#if IT_KEY_NEW zint this_sys = key.mem[0]; -#else - zint this_sys = key.sysno; -#endif if (this_sys != psysno) { psysno = this_sys; @@ -427,13 +427,13 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, if (position == positions[num_i]) { sr[num_i].sysno = psysno; - yaz_log (LOG_DEBUG, "got pos=%d (unsorted)", position); + yaz_log(YLOG_DEBUG, "got pos=%d (unsorted)", position); sr[num_i].score = -1; num_i++; } } } - rset_close (rset, rfd); + rset_close (rfd); } } return sr; @@ -591,7 +591,7 @@ void resultSetSort (ZebraHandle zh, NMEM nmem, zh->errCode = 230; return; } - yaz_log (LOG_DEBUG, "result set sort input=%s output=%s", + yaz_log(YLOG_DEBUG, "result set sort input=%s output=%s", *input_setnames, output_setname); sset = resultSetGet (zh, input_setnames[0]); if (!sset) @@ -620,16 +620,24 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem, Z_SortKeySpecList *sort_sequence, int *sort_status) { int i; + int n = 0; + zint kno = 0; zint psysno = 0; struct it_key key; struct sortKeyInfo sort_criteria[3]; int num_criteria; RSFD rfd; + TERMID termid; + TERMID *terms; + int numTerms = 0; - yaz_log (LOG_LOG, "resultSetSortSingle start"); assert(nmem); /* compiler shut up about unused param */ sset->sort_info->num_entries = 0; + rset_getterms(rset, 0, 0, &n); + terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n); + rset_getterms(rset, terms, n, &numTerms); + sset->hits = 0; num_criteria = sort_sequence->num_specs; if (num_criteria > 3) @@ -662,20 +670,20 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem, switch (sk->which) { case Z_SortKey_sortField: - yaz_log (LOG_DEBUG, "Sort: key %d is of type sortField", i+1); + yaz_log(YLOG_DEBUG, "Sort: key %d is of type sortField", i+1); zh->errCode = 207; return; case Z_SortKey_elementSpec: - yaz_log (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1); + yaz_log(YLOG_DEBUG, "Sort: key %d is of type elementSpec", i+1); zh->errCode = 207; return; case Z_SortKey_sortAttributes: - yaz_log (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1); + yaz_log(YLOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1); sort_criteria[i].attrUse = zebra_maps_sort (zh->reg->zebra_maps, sk->u.sortAttributes, &sort_criteria[i].numerical); - yaz_log (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse); + yaz_log(YLOG_DEBUG, "use value = %d", sort_criteria[i].attrUse); if (sort_criteria[i].attrUse == -1) { zh->errCode = 116; @@ -690,13 +698,11 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem, } } rfd = rset_open (rset, RSETF_READ); - while (rset_read (rset, rfd, &key)) + while (rset_read (rfd, &key, &termid)) + /* FIXME - pass a TERMID *, and use it for something below !! */ { -#if IT_KEY_NEW zint this_sys = key.mem[0]; -#else - zint this_sys = key.sysno; -#endif + kno++; if (this_sys != psysno) { (sset->hits)++; @@ -705,19 +711,13 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem, sort_criteria, num_criteria, psysno); } } - rset_close (rset, rfd); - -#if 0 - for (i = 0; i < rset->no_rset_terms; i++) - yaz_log (LOG_LOG, "term=\"%s\" nn=" ZINT_FORMAT - " type=%s count=" ZINT_FORMAT, - rset->rset_terms[i]->name, - rset->rset_terms[i]->nn, - rset->rset_terms[i]->flags, - rset->rset_terms[i]->count); -#endif + rset_close (rfd); + yaz_log(YLOG_DEBUG, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, sort", + kno, sset->hits); + for (i = 0; i < numTerms; i++) + yaz_log(YLOG_LOG, "term=\"%s\" type=%s count=" ZINT_FORMAT, + terms[i]->name, terms[i]->flags, rset_count(terms[i]->rset)); *sort_status = Z_SortResponse_success; - yaz_log (LOG_LOG, "resultSetSortSingle end"); } RSET resultSetRef (ZebraHandle zh, const char *resultSetId) @@ -729,12 +729,15 @@ RSET resultSetRef (ZebraHandle zh, const char *resultSetId) return NULL; } -void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset) +void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset, NMEM nmem) { zint kno = 0; struct it_key key; RSFD rfd; - /* int term_index; */ + TERMID termid; + TERMID *terms; + int numTerms = 0; + int n = 0; int i; ZebraRankClass rank_class; struct rank_control *rc; @@ -748,39 +751,35 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset) sort_info = zebraSet->sort_info; sort_info->num_entries = 0; zebraSet->hits = 0; - rfd = rset_open (rset, RSETF_READ); + rset_getterms(rset, 0, 0, &n); + terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n); + rset_getterms(rset, terms, n, &numTerms); - yaz_log (LOG_LOG, "resultSetRank"); + rfd = rset_open (rset, RSETF_READ); rank_class = zebraRankLookup (zh, rank_handler_name); if (!rank_class) { - yaz_log (LOG_WARN, "No such rank handler: %s", rank_handler_name); + yaz_log(YLOG_WARN, "No such rank handler: %s", rank_handler_name); return; } rc = rank_class->control; - if (rset_read (rset, rfd, &key)) + if (rset_read (rfd, &key, &termid)) { -#if IT_KEY_NEW zint psysno = key.mem[0]; -#else - zint psysno = key.sysno; -#endif int score; void *handle = - (*rc->begin) (zh->reg, rank_class->class_handle, rset); + (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem, + terms, numTerms); (zebraSet->hits)++; - esthits=atoi(res_get_def(zh->res,"estimatehits","0")); + esthits = atoi(res_get_def(zh->res, "estimatehits","0")); if (!esthits) - est=-1; /* can not do */ + est = -1; /* can not do */ do { -#if IT_KEY_NEW - zint this_sys = key.mem[0]; -#else - zint this_sys = key.sysno; -#endif + zint this_sys = key.mem[0]; /* FIXME - assumes scope==2 */ + zint seqno = key.mem[1]; /* FIXME - assumes scope==2 */ kno++; if (this_sys != psysno) { @@ -790,55 +789,45 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset) (zebraSet->hits)++; psysno = this_sys; } - /* FIXME - Ranking is broken, since rsets no longer have */ - /* term lists! */ - /* (*rc->add) (handle, this_sys, term_index); */ + (*rc->add) (handle, seqno, termid); - if ( (est==-2) && (zebraSet->hits==esthits)) - { /* time to estimate the hits */ - rset_pos(rset,rfd,&cur,&tot); - if (tot>0) { - ratio=cur/tot; - est=(zint)(0.5+zebraSet->hits/ratio); - logf(LOG_LOG, "Estimating hits (%s) " - "%0.1f->"ZINT_FORMAT - "; %0.1f->"ZINT_FORMAT, - rset->control->desc, - cur, zebraSet->hits, - tot,est); - i=0; /* round to 3 significant digits */ - while (est>1000) { - est/=10; - i++; + if ((est==-2) && (zebraSet->hits==esthits)) + { /* time to estimate the hits */ + rset_pos(rfd,&cur,&tot); + if (tot>0) { + ratio = cur/tot; + est = (zint)(0.5+zebraSet->hits/ratio); + yaz_log(YLOG_LOG, "Estimating hits (%s) " + "%0.1f->" ZINT_FORMAT + "; %0.1f->" ZINT_FORMAT, + rset->control->desc, + cur, zebraSet->hits, + tot, est); + i = 0; /* round to 3 significant digits */ + while (est>1000) { + est /= 10; + i++; + } + while (i--) + est *= 10; + zebraSet->hits = est; } - while (i--) est*=10; - zebraSet->hits=est; } } - } - while (rset_read (rset, rfd, &key) && (est<0) ); - - score = (*rc->calc) (handle, psysno); - resultSetInsertRank (zh, sort_info, psysno, score, 'A'); + while (rset_read (rfd, &key,&termid) && (est<0) ); + score = (*rc->calc)(handle, psysno); + resultSetInsertRank(zh, sort_info, psysno, score, 'A'); (*rc->end) (zh->reg, handle); } - rset_close (rset, rfd); -/* - for (i = 0; i < rset->no_rset_terms; i++) + rset_close (rfd); + + yaz_log(YLOG_DEBUG, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, rank", + kno, zebraSet->hits); + for (i = 0; i < numTerms; i++) { - if (est>0) - rset->rset_terms[i]->count = - est=(zint)(rset->rset_terms[i]->count/ratio); - yaz_log (LOG_LOG, "term=\"%s\" nn=" ZINT_FORMAT - " type=%s count=" ZINT_FORMAT, - rset->rset_terms[i]->name, - rset->rset_terms[i]->nn, - rset->rset_terms[i]->flags, - rset->rset_terms[i]->count); + yaz_log(YLOG_LOG, "term=\"%s\" type=%s count=" ZINT_FORMAT, + terms[i]->name, terms[i]->flags, rset_count(terms[i]->rset)); } -*/ - yaz_log (LOG_LOG, ZINT_FORMAT " keys, "ZINT_FORMAT" distinct sysnos", - kno, zebraSet->hits); } ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)