X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzsets.c;h=98fd1dc4bd9e195075705c0e5ad7fd7edcbc702e;hb=746f36f2b12f2a342dc0213cb03458359a493c8f;hp=ed60386944b4fe31a85b342f06f3b3da3bc0b4e1;hpb=342c28e71fe5122d17d760a930107cab0266e9c6;p=idzebra-moved-to-github.git diff --git a/index/zsets.c b/index/zsets.c index ed60386..98fd1dc 100644 --- a/index/zsets.c +++ b/index/zsets.c @@ -1,4 +1,4 @@ -/* $Id: zsets.c,v 1.93 2005-08-19 09:21:34 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; @@ -267,6 +269,7 @@ 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; @@ -287,6 +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; @@ -654,6 +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, @@ -690,11 +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); } @@ -872,6 +950,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, terms, numTerms); 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[sysno_mem_index]; @@ -886,10 +965,13 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, break; if (psysno) { /* only if we did have a previous record */ - score = (*rc->calc) (handle, psysno, pstaticrank); + 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) @@ -900,7 +982,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, /* no more items */ if (psysno) { /* we had - at least - one record */ - score = (*rc->calc)(handle, psysno, pstaticrank); + score = (*rc->calc)(handle, psysno, pstaticrank, &stop_flag); /* insert the hit. A=Ascending */ resultSetInsertRank(zh, sort_info, psysno, score, 'A'); count++; @@ -1068,7 +1150,7 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, zint sysno, zebra_snippets *snippets) { ZebraSet sset = resultSetGet(zh, setname); - yaz_log(YLOG_LOG, "zebra_get_hit_vector setname=%s zysno=" ZINT_FORMAT, + yaz_log(YLOG_DEBUG, "zebra_get_hit_vector setname=%s zysno=" ZINT_FORMAT, setname, sysno); if (!sset) return ZEBRA_FAIL; @@ -1107,7 +1189,6 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, for (ol = termid->ol; ol; ol = ol->next) { zebra_snippets_append(snippets, key.mem[key.len-1], - termid->reg_type, ol->ord, termid->name); } } @@ -1116,6 +1197,7 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, rset_delete(rset_comb); nmem_destroy(nmem); + kc->dec(kc); } return ZEBRA_OK; }