X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzsets.c;h=98fd1dc4bd9e195075705c0e5ad7fd7edcbc702e;hb=746f36f2b12f2a342dc0213cb03458359a493c8f;hp=d729f93d429f391e013d42471af1bae96b31b067;hpb=989564485b832bc896b2c0bf73c8e4a7f28f1ad2;p=idzebra-moved-to-github.git diff --git a/index/zsets.c b/index/zsets.c index d729f93..98fd1dc 100644 --- a/index/zsets.c +++ b/index/zsets.c @@ -1,4 +1,4 @@ -/* $Id: zsets.c,v 1.84 2005-05-31 07:29:10 adam Exp $ +/* $Id: zsets.c,v 1.100 2006-05-03 09:31:26 marc Exp $ Copyright (C) 1995-2005 Index Data ApS @@ -30,6 +30,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #endif #include "index.h" +#include "rank.h" #include #include @@ -53,6 +54,7 @@ struct zebra_set { int num_bases; char **basenames; Z_RPNQuery *rpn; + Z_SortKeySpecList *sortSpec; struct zset_sort_info *sort_info; struct zebra_set_term_entry *term_entries; int term_entries_max; @@ -62,6 +64,7 @@ struct zebra_set { zint cache_position; /* last position */ RSFD cache_rfd; /* rfd (NULL if not existing) */ zint cache_psysno; /* sysno for last position */ + zint approx_limit; /* limit before we do approx */ }; struct zset_sort_entry { @@ -128,6 +131,7 @@ ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem, for (i = 0; sort_sequence->specs[i]; i++) ; sort_sequence->num_specs = i; + rset->hits_limit = sset->approx_limit; if (!i) { res = resultSetRank (zh, sset, rset, rset_nmem); @@ -152,7 +156,7 @@ ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn, zh->hits = 0; - zebraSet = resultSetAdd (zh, setname, 1); + zebraSet = resultSetAdd(zh, setname, 1); if (!zebraSet) return ZEBRA_FAIL; zebraSet->locked = 1; @@ -164,7 +168,7 @@ ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn, zebraSet->basenames = nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames)); for (i = 0; ibasenames[i] = nmem_strdup (zebraSet->nmem, basenames[i]); + zebraSet->basenames[i] = nmem_strdup(zebraSet->nmem, basenames[i]); res = resultSetSearch(zh, zebraSet->nmem, zebraSet->rset_nmem, rpn, zebraSet); @@ -205,7 +209,7 @@ void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type, (s->hits)++; } -ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) +ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov) { ZebraSet s; int i; @@ -265,12 +269,14 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) s->rset_nmem = 0; s->nmem = 0; s->rpn = 0; + s->sortSpec = 0; s->cache_position = 0; s->cache_rfd = 0; + s->approx_limit = zh->approx_limit; return s; } -ZebraSet resultSetGet (ZebraHandle zh, const char *name) +ZebraSet resultSetGet(ZebraHandle zh, const char *name) { ZebraSet s; @@ -284,6 +290,13 @@ ZebraSet resultSetGet (ZebraHandle zh, const char *name) if (!s->rset_nmem) s->rset_nmem=nmem_create(); resultSetSearch(zh, nmem, s->rset_nmem, s->rpn, s); + if (s->rset && s->sortSpec) + { + int sort_status; + yaz_log(log_level_resultsets, "resort %s", name); + resultSetSortSingle (zh, nmem, s, s->rset, s->sortSpec, + &sort_status); + } nmem_destroy (nmem); } return s; @@ -313,7 +326,7 @@ void resultSetInvalidate (ZebraHandle zh) } } -void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses) +void resultSetDestroy(ZebraHandle zh, int num, char **names,int *statuses) { ZebraSet * ss = &zh->sets; int i; @@ -395,6 +408,10 @@ ZebraMetaRecord *zebra_meta_records_create(ZebraHandle zh, const char *name, RSET rset; int i; struct zset_sort_info *sort_info; + size_t sysno_mem_index = 0; + + if (zh->m_staticrank) + sysno_mem_index = 1; if (!log_level_set) loglevels(); @@ -480,7 +497,7 @@ ZebraMetaRecord *zebra_meta_records_create(ZebraHandle zh, const char *name, } while (num_i < num && rset_read (rfd, &key, 0)) { - zint this_sys = key.mem[0]; + zint this_sys = key.mem[sysno_mem_index]; if (this_sys != psysno) { psysno = this_sys; @@ -647,6 +664,77 @@ void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info, new_entry->score = score; } +static Z_RPNQuery *copy_RPNQuery(Z_RPNQuery *src, NMEM nmem) +{ + Z_RPNQuery *dst = 0; + ODR encode = odr_createmem(ODR_ENCODE); + ODR decode = odr_createmem(ODR_DECODE); + + if (z_RPNQuery(encode, &src, 0, 0)) + { + int len; + char *buf = odr_getbuf(encode, &len, 0); + + if (buf) + { + odr_setbuf(decode, buf, len, 0); + z_RPNQuery(decode, &dst, 0, 0); + } + } + nmem_transfer(nmem, decode->mem); + odr_destroy(encode); + odr_destroy(decode); + return dst; +} + +static Z_SortKeySpecList *copy_SortKeySpecList(Z_SortKeySpecList *src, NMEM nmem) +{ + Z_SortKeySpecList *dst = 0; + ODR encode = odr_createmem(ODR_ENCODE); + ODR decode = odr_createmem(ODR_DECODE); + + if (z_SortKeySpecList(encode, &src, 0, 0)) + { + int len; + char *buf = odr_getbuf(encode, &len, 0); + + if (buf) + { + odr_setbuf(decode, buf, len, 0); + z_SortKeySpecList(decode, &dst, 0, 0); + } + } + nmem_transfer(nmem, decode->mem); + odr_destroy(encode); + odr_destroy(decode); + return dst; +} + +ZebraSet resultSetClone(ZebraHandle zh, const char *setname, + ZebraSet rset) +{ + ZebraSet nset; + int i; + + nset = resultSetAdd(zh, setname, 1); + if (!nset) + return 0; + + nset->nmem = nmem_create(); + + nset->num_bases = rset->num_bases; + nset->basenames = + nmem_malloc (nset->nmem, nset->num_bases * sizeof(*rset->basenames)); + for (i = 0; inum_bases; i++) + nset->basenames[i] = nmem_strdup(nset->nmem, rset->basenames[i]); + + if (rset->rset) + nset->rset = rset_dup(rset->rset); + if (rset->rpn) + nset->rpn = copy_RPNQuery(rset->rpn, nset->nmem); + return nset; +} + ZEBRA_RES resultSetSort(ZebraHandle zh, NMEM nmem, int num_input_setnames, const char **input_setnames, const char *output_setname, @@ -683,11 +771,8 @@ ZEBRA_RES resultSetSort(ZebraHandle zh, NMEM nmem, return ZEBRA_FAIL; } if (strcmp (output_setname, input_setnames[0])) - { - rset = rset_dup (rset); - sset = resultSetAdd (zh, output_setname, 1); - sset->rset = rset; - } + sset = resultSetClone(zh, output_setname, sset); + sset->sortSpec = copy_SortKeySpecList(sort_sequence, sset->nmem); return resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status); } @@ -708,6 +793,11 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, TERMID termid; TERMID *terms; int numTerms = 0; + size_t sysno_mem_index = 0; + + if (zh->m_staticrank) + sysno_mem_index = 1; + assert(nmem); /* compiler shut up about unused param */ sset->sort_info->num_entries = 0; @@ -782,7 +872,7 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, rfd = rset_open (rset, RSETF_READ); while (rset_read (rfd, &key, &termid)) { - zint this_sys = key.mem[0]; + zint this_sys = key.mem[sysno_mem_index]; if (log_level_searchhits) key_logdump_txt(log_level_searchhits, &key, termid->name); kno++; @@ -826,6 +916,10 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, ZebraRankClass rank_class; struct zset_sort_info *sort_info; const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1"); + size_t sysno_mem_index = 0; + + if (zh->m_staticrank) + sysno_mem_index = 1; if (!log_level_set) loglevels(); @@ -836,6 +930,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n); rset_getterms(rset, terms, n, &numTerms); + rank_class = zebraRankLookup(zh, rank_handler_name); if (!rank_class) { @@ -848,35 +943,49 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, RSFD rfd = rset_open(rset, RSETF_READ); struct rank_control *rc = rank_class->control; double score; + zint count = 0; void *handle = (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem, terms, numTerms); - zint psysno = 0; + zint psysno = 0; /* previous doc id / sys no */ + zint pstaticrank = 0; /* previous static rank */ + int stop_flag = 0; while (rset_read(rfd, &key, &termid)) { - zint this_sys = key.mem[0]; + zint this_sys = key.mem[sysno_mem_index]; + zint seqno = key.mem[key.len-1]; kno++; if (log_level_searchhits) key_logdump_txt(log_level_searchhits, &key, termid->name); - if (this_sys != psysno) - { - if (rfd->counted_items >= rset->hits_limit) + if (this_sys != psysno) + { /* new record .. */ + if (rfd->counted_items > rset->hits_limit) break; if (psysno) - { - score = (*rc->calc) (handle, psysno); + { /* only if we did have a previous record */ + score = (*rc->calc) (handle, psysno, pstaticrank, + &stop_flag); + /* insert the hit. A=Ascending */ resultSetInsertRank (zh, sort_info, psysno, score, 'A'); + count++; + if (stop_flag) + break; } psysno = this_sys; + if (zh->m_staticrank) + pstaticrank = key.mem[0]; } (*rc->add) (handle, CAST_ZINT_TO_INT(seqno), termid); } + /* no more items */ if (psysno) - { - score = (*rc->calc)(handle, psysno); + { /* we had - at least - one record */ + score = (*rc->calc)(handle, psysno, pstaticrank, &stop_flag); + /* insert the hit. A=Ascending */ resultSetInsertRank(zh, sort_info, psysno, score, 'A'); + count++; } (*rc->end) (zh->reg, handle); rset_close (rfd); @@ -954,8 +1063,9 @@ static int trav_rset_for_termids(RSET rset, TERMID *termid_array, if (approx_array) approx_array[no] = rset->hits_approx; #if 0 - yaz_log(YLOG_LOG, "rset=%p term=%s count=" ZINT_FORMAT, - rset, rset->term->name, rset->hits_count); + yaz_log(YLOG_LOG, "rset=%p term=%s limit=" ZINT_FORMAT + " count=" ZINT_FORMAT, + rset, rset->term->name, rset->hits_limit, rset->hits_count); #endif no++; } @@ -977,7 +1087,8 @@ ZEBRA_RES zebra_result_set_term_no(ZebraHandle zh, const char *setname, ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname, int no, zint *count, int *approx, - char *termbuf, size_t *termlen) + char *termbuf, size_t *termlen, + const char **term_ref_id) { ZebraSet sset = resultSetGet(zh, setname); if (sset) @@ -1023,6 +1134,8 @@ ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname, } termbuf[*termlen] = '\0'; } + if (term_ref_id) + *term_ref_id = term_array[no]->ref_id; xfree(term_array); xfree(hits_array); @@ -1032,3 +1145,60 @@ ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname, } return ZEBRA_FAIL; } + +ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, + zint sysno, zebra_snippets *snippets) +{ + ZebraSet sset = resultSetGet(zh, setname); + yaz_log(YLOG_DEBUG, "zebra_get_hit_vector setname=%s zysno=" ZINT_FORMAT, + setname, sysno); + if (!sset) + return ZEBRA_FAIL; + else + { + struct rset_key_control *kc = zebra_key_control_create(zh); + NMEM nmem = nmem_create(); + struct it_key key; + RSET rsets[2], rset_comb; + RSET rset_temp = rstemp_create(nmem, kc, kc->scope, + res_get (zh->res, "setTmpDir"),0 ); + + TERMID termid; + RSFD rsfd = rset_open(rset_temp, RSETF_WRITE); + + key.mem[0] = sysno; + key.mem[1] = 0; + key.mem[2] = 0; + key.mem[3] = 0; + key.len = 2; + rset_write (rsfd, &key); + rset_close (rsfd); + + rsets[0] = rset_temp; + rsets[1] = rset_dup(sset->rset); + + rset_comb = rsmulti_and_create(nmem, kc, kc->scope, 2, rsets); + + rsfd = rset_open(rset_comb, RSETF_READ); + + while (rset_read(rsfd, &key, &termid)) + { + if (termid) + { + struct ord_list *ol; + for (ol = termid->ol; ol; ol = ol->next) + { + zebra_snippets_append(snippets, key.mem[key.len-1], + ol->ord, termid->name); + } + } + } + rset_close(rsfd); + + rset_delete(rset_comb); + nmem_destroy(nmem); + kc->dec(kc); + } + return ZEBRA_OK; +} +