X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzsets.c;h=1109f81de3acd252454c242734df63c3c3152860;hp=aa2d16dc4c9156283199e59166f6e6f3c2aad3a5;hb=6c9fcd3b5d3108702fa1ffc92dab4ab6060f9a19;hpb=af5941499d0d0fe3f2d41f5aa1f82d6a04190349 diff --git a/index/zsets.c b/index/zsets.c index aa2d16d..1109f81 100644 --- a/index/zsets.c +++ b/index/zsets.c @@ -1,6 +1,6 @@ -/* $Id: zsets.c,v 1.58 2004-08-30 12:32:24 heikki Exp $ - Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 - Index Data Aps +/* $Id: zsets.c,v 1.77 2005-01-15 19:38:30 adam Exp $ + Copyright (C) 1995-2005 + Index Data ApS This file is part of the Zebra server. @@ -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 @@ -56,6 +56,10 @@ struct zebra_set { int term_entries_max; struct zebra_set *next; int locked; + + zint cache_position; /* last position */ + RSFD cache_rfd; /* rfd (NULL if not existing) */ + zint cache_psysno; /* sysno for last position */ }; struct zset_sort_entry { @@ -71,6 +75,23 @@ struct zset_sort_info { struct zset_sort_entry **entries; }; +static int log_level_set=0; +static int log_level_sorting=0; +static int log_level_searchhits=0; +static int log_level_searchterms=0; +static int log_level_resultsets=0; + +static void loglevels() +{ + if (log_level_set) + return; + log_level_sorting = yaz_log_module_level("sorting"); + log_level_searchhits = yaz_log_module_level("searchhits"); + log_level_searchterms = yaz_log_module_level("searchterms"); + log_level_resultsets = yaz_log_module_level("resultsets"); + log_level_set=1; +} + ZebraSet resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn, int num_bases, char **basenames, @@ -89,7 +110,7 @@ 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 = @@ -137,60 +158,6 @@ void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type, (s->hits)++; } -#if 0 /* FIXME - Delete this, we don't count terms no more */ -int zebra_resultSetTerms (ZebraHandle zh, const char *setname, - int no, zint *count, - int *type, char *out, size_t *len) -{ - ZebraSet s = resultSetGet (zh, setname); - int no_max = 0; - - if (count) - *count = 0; - if (!s || !s->rset) - return 0; - no_max = s->rset->no_rset_terms; - if (no < 0 || no >= no_max) - return 0; - if (count) - *count = s->rset->rset_terms[no]->count; - if (type) - *type = s->rset->rset_terms[no]->type; - - if (out) - { - char *inbuf = s->rset->rset_terms[no]->name; - size_t inleft = strlen(inbuf); - size_t outleft = *len - 1; - int converted = 0; - - if (zh->iconv_from_utf8 != 0) - { - char *outbuf = out; - size_t ret; - - ret = yaz_iconv(zh->iconv_from_utf8, &inbuf, &inleft, - &outbuf, &outleft); - if (ret == (size_t)(-1)) - *len = 0; - else - *len = outbuf - out; - converted = 1; - } - if (!converted) - { - if (inleft > outleft) - inleft = outleft; - *len = inleft; - memcpy (out, inbuf, *len); - } - out[*len] = 0; - } - return no_max; -} - -#endif - ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) { ZebraSet s; @@ -199,13 +166,22 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) for (s = zh->sets; s; s = s->next) if (!strcmp (s->name, name)) break; + + if (!log_level_set) + loglevels(); if (s) { - yaz_log (LOG_DEBUG, "updating result set %s", name); + yaz_log(log_level_resultsets, "updating result set %s", name); if (!ov || s->locked) return NULL; if (s->rset) + { + if (s->cache_rfd) + rset_close(s->cache_rfd); rset_delete (s->rset); + } + if (s->rset_nmem) + nmem_destroy (s->rset_nmem); if (s->nmem) nmem_destroy (s->nmem); } @@ -213,7 +189,7 @@ 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(log_level_resultsets, "adding result set %s", name); s = (ZebraSet) xmalloc (sizeof(*s)); s->next = zh->sets; zh->sets = s; @@ -242,6 +218,8 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) s->rset_nmem=0; s->nmem = 0; s->rpn = 0; + s->cache_position = 0; + s->cache_rfd = 0; return s; } @@ -255,7 +233,9 @@ 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(log_level_resultsets, "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); @@ -270,11 +250,18 @@ void resultSetInvalidate (ZebraHandle zh) { ZebraSet s = zh->sets; + yaz_log(log_level_resultsets, "invalidating result sets"); for (; s; s = s->next) { if (s->rset) + { + if (s->cache_rfd) + rset_close(s->cache_rfd); rset_delete (s->rset); + } s->rset = 0; + s->cache_rfd = 0; + s->cache_position = 0; if (s->rset_nmem) nmem_destroy(s->rset_nmem); s->rset_nmem=0; @@ -317,7 +304,11 @@ void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses) if (s->nmem) nmem_destroy (s->nmem); if (s->rset) + { + if (s->cache_rfd) + rset_close(s->cache_rfd); rset_delete (s->rset); + } if (s->rset_nmem) nmem_destroy(s->rset_nmem); xfree (s->name); @@ -328,22 +319,49 @@ void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses) } } -ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, - int num, int *positions) +ZebraMetaRecord *zebra_meta_records_create_range (ZebraHandle zh, + const char *name, + zint start, int num) +{ + zint pos_small[10]; + zint *pos = pos_small; + ZebraMetaRecord *mr; + int i; + + if (num > 10000 || num <= 0) + return 0; + + if (num > 10) + pos = xmalloc(sizeof(*pos) * num); + + for (i = 0; i 10) + xfree(pos); + return mr; +} + +ZebraMetaRecord *zebra_meta_records_create (ZebraHandle zh, const char *name, + int num, zint *positions) { ZebraSet sset; - ZebraPosSet sr = 0; + ZebraMetaRecord *sr = 0; RSET rset; int i; struct zset_sort_info *sort_info; + if (!log_level_set) + loglevels(); if (!(sset = resultSetGet (zh, name))) return NULL; if (!(rset = sset->rset)) { if (!sset->term_entries) return 0; - sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num); + sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num); for (i = 0; isort_info; if (sort_info) { - int position; + zint position; for (i = 0; i 0 && position <= sort_info->num_entries) { - yaz_log (LOG_DEBUG, "got pos=%d (sorted)", position); + yaz_log(log_level_sorting, "got pos=" ZINT_FORMAT + " (sorted)", position); sr[i].sysno = sort_info->entries[position-1]->sysno; sr[i].score = sort_info->entries[position-1]->score; } @@ -392,7 +411,7 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, } if (i < num) /* nope, get the rest, unsorted - sorry */ { - int position = 0; + zint position = 0; int num_i = 0; zint psysno = 0; RSFD rfd; @@ -402,14 +421,23 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, position = sort_info->num_entries; 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)) + + if (sset->cache_rfd && + num_i < num && positions[num_i] > sset->cache_position) + { + position = sset->cache_position; + rfd = sset->cache_rfd; + psysno = sset->cache_psysno; + } + else + { + if (sset->cache_rfd) + rset_close(sset->cache_rfd); + rfd = rset_open (rset, RSETF_READ); + } + 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; @@ -427,19 +455,22 @@ 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(log_level_sorting, "got pos=" ZINT_FORMAT " (unsorted)", position); sr[num_i].score = -1; num_i++; } } } - rset_close (rset, rfd); + sset->cache_position = position; + sset->cache_psysno = psysno; + sset->cache_rfd = rfd; } } return sr; } -void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num) +void zebra_meta_records_destroy (ZebraHandle zh, ZebraMetaRecord *records, + int num) { assert(zh); /* compiler shut up about unused arg */ xfree (records); @@ -591,7 +622,9 @@ void resultSetSort (ZebraHandle zh, NMEM nmem, zh->errCode = 230; return; } - yaz_log (LOG_DEBUG, "result set sort input=%s output=%s", + if (!log_level_set) + loglevels(); + yaz_log(log_level_sorting, "result set sort input=%s output=%s", *input_setnames, output_setname); sset = resultSetGet (zh, input_setnames[0]); if (!sset) @@ -620,16 +653,24 @@ 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 +703,20 @@ 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(log_level_sorting, "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(log_level_sorting, "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(log_level_sorting, "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(log_level_sorting, "use value = %d", sort_criteria[i].attrUse); if (sort_criteria[i].attrUse == -1) { zh->errCode = 116; @@ -690,13 +731,11 @@ 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 +744,13 @@ 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(log_level_sorting, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, sort", + kno, sset->hits); + for (i = 0; i < numTerms; i++) + yaz_log(log_level_sorting, "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 +762,15 @@ 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; @@ -745,43 +781,42 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset) zint esthits; double ratio; + if (!log_level_set) + loglevels(); 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[key.len-1]; /* FIXME - assumes scope==2 */ kno++; + key_logdump_txt(log_level_searchhits,&key," Got hit"); if (this_sys != psysno) { score = (*rc->calc) (handle, psysno); @@ -790,55 +825,45 @@ 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(log_level_searchhits, "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(log_level_searchterms, 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(log_level_searchterms, "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) @@ -874,9 +899,9 @@ void zebraRankDestroy (struct zebra_register *reg) ZebraRankClass p_next = p->next; if (p->init_flag && p->control->destroy) (*p->control->destroy)(reg, p->class_handle); - xfree (p->control->name); - xfree (p->control); - xfree (p); + xfree(p->control->name); + xfree(p->control); + xfree(p); p = p_next; } reg->rank_classes = NULL;