X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzsets.c;h=4a8c74c39c8c26f6638ee1fd0e89069a0b05138a;hp=cf87884e7600adb8a1096df932362878492c8996;hb=d87f70faee1aeead5d32c388bfde8c41e859fb2f;hpb=e590cf3b85bc9dcdd908033b96c64bb5457f19e9 diff --git a/index/zsets.c b/index/zsets.c index cf87884..4a8c74c 100644 --- a/index/zsets.c +++ b/index/zsets.c @@ -1,8 +1,5 @@ -/* $Id: zsets.c,v 1.109 2006-08-07 10:14:59 adam Exp $ - Copyright (C) 1995-2006 - Index Data ApS - -This file is part of the Zebra server. +/* This file is part of the Zebra server. + Copyright (C) 1994-2011 Index Data Zebra is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free @@ -15,9 +12,9 @@ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License -along with Zebra; see the file LICENSE.zebra. If not, write to the -Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA -02111-1307, USA. +along with this program; if not, write to the Free Software +Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA + */ @@ -50,7 +47,7 @@ struct zebra_set { NMEM rset_nmem; /* for creating the rsets in */ zint hits; int num_bases; - char **basenames; + const char **basenames; Z_RPNQuery *rpn; Z_SortKeySpecList *sortSpec; struct zset_sort_info *sort_info; @@ -58,6 +55,7 @@ struct zebra_set { int term_entries_max; struct zebra_set *next; int locked; + int estimated_hit_count; zint cache_position; /* last position */ RSFD cache_rfd; /* rfd (NULL if not existing) */ @@ -83,7 +81,7 @@ static int log_level_searchhits=0; static int log_level_searchterms=0; static int log_level_resultsets=0; -static void loglevels() +static void loglevels(void) { if (log_level_set) return; @@ -94,28 +92,28 @@ static void loglevels() log_level_set = 1; } -ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem, - Z_RPNQuery *rpn, ZebraSet sset) + +static ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem, + Z_RPNQuery *rpn, ZebraSet sset) { RSET rset = 0; - oident *attrset; Z_SortKeySpecList *sort_sequence; int sort_status, i; ZEBRA_RES res = ZEBRA_OK; - zh->hits = 0; - sort_sequence = (Z_SortKeySpecList *) nmem_malloc(nmem, sizeof(*sort_sequence)); sort_sequence->num_specs = 10; /* FIXME - Hard-coded number */ sort_sequence->specs = (Z_SortKeySpec **) nmem_malloc(nmem, sort_sequence->num_specs * - sizeof(*sort_sequence->specs)); + sizeof(*sort_sequence->specs)); for (i = 0; inum_specs; i++) sort_sequence->specs[i] = 0; - attrset = oid_getentbyoid (rpn->attributeSetId); - res = rpn_search_top(zh, rpn->RPNStructure, attrset->value, + rpn_get_top_approx_limit(zh, rpn->RPNStructure, &sset->approx_limit); + + res = rpn_search_top(zh, rpn->RPNStructure, rpn->attributeSetId, + sset->approx_limit, nmem, rset_nmem, sort_sequence, sset->num_bases, sset->basenames, @@ -128,30 +126,33 @@ 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; + rset_set_hits_limit(rset, sset->approx_limit); + if (!i) { - res = resultSetRank (zh, sset, rset, rset_nmem); + res = resultSetRank(zh, sset, rset, rset_nmem); } else { - res = resultSetSortSingle (zh, nmem, sset, rset, - sort_sequence, &sort_status); + res = resultSetSortSingle(zh, nmem, sset, rset, + sort_sequence, &sort_status); } sset->rset = rset; return res; } -ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn, - int num_bases, char **basenames, - const char *setname) +ZEBRA_RES resultSetAddRPN(ZebraHandle zh, NMEM m, Z_RPNQuery *rpn, + int num_bases, char **basenames, + const char *setname, + zint *hits, int *estimated_hit_count) { ZebraSet zebraSet; int i; ZEBRA_RES res; - zh->hits = 0; + *hits = 0; + *estimated_hit_count = 0; zebraSet = resultSetAdd(zh, setname, 1); if (!zebraSet) @@ -163,13 +164,16 @@ ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn, zebraSet->num_bases = num_bases; zebraSet->basenames = - nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames)); + nmem_malloc(zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames)); for (i = 0; ibasenames[i] = nmem_strdup(zebraSet->nmem, basenames[i]); res = resultSetSearch(zh, zebraSet->nmem, zebraSet->rset_nmem, rpn, zebraSet); - zh->hits = zebraSet->hits; + *hits = zebraSet->hits; + if (zebraSet->estimated_hit_count) + *estimated_hit_count = 1; + if (zebraSet->rset) zebraSet->rpn = rpn; zebraSet->locked = 0; @@ -184,21 +188,21 @@ void resultSetAddTerm(ZebraHandle zh, ZebraSet s, int reg_type, { assert(zh); /* compiler shut up */ if (!s->nmem) - s->nmem = nmem_create (); + s->nmem = nmem_create(); if (!s->term_entries) { int i; s->term_entries_max = 1000; s->term_entries = - nmem_malloc (s->nmem, s->term_entries_max * - sizeof(*s->term_entries)); + nmem_malloc(s->nmem, s->term_entries_max * + sizeof(*s->term_entries)); for (i = 0; i < s->term_entries_max; i++) s->term_entries[i].term = 0; } if (s->hits < s->term_entries_max) { s->term_entries[s->hits].reg_type = reg_type; - s->term_entries[s->hits].db = nmem_strdup (s->nmem, db); + s->term_entries[s->hits].db = nmem_strdup(s->nmem, db); s->term_entries[s->hits].index_name = nmem_strdup(s->nmem, index_name); s->term_entries[s->hits].term = nmem_strdup(s->nmem, term); } @@ -211,7 +215,7 @@ ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov) int i; for (s = zh->sets; s; s = s->next) - if (!strcmp (s->name, name)) + if (!strcmp(s->name, name)) break; if (!log_level_set) @@ -225,36 +229,35 @@ ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov) { if (s->cache_rfd) rset_close(s->cache_rfd); - rset_delete (s->rset); + rset_delete(s->rset); } if (s->rset_nmem) - nmem_destroy (s->rset_nmem); + nmem_destroy(s->rset_nmem); if (s->nmem) - nmem_destroy (s->nmem); + nmem_destroy(s->nmem); } else { const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000"); yaz_log(log_level_resultsets, "adding result set %s", name); - s = (ZebraSet) xmalloc (sizeof(*s)); + s = (ZebraSet) xmalloc(sizeof(*s)); s->next = zh->sets; zh->sets = s; - s->name = (char *) xmalloc (strlen(name)+1); - strcpy (s->name, name); + s->name = xstrdup(name); s->sort_info = (struct zset_sort_info *) - xmalloc (sizeof(*s->sort_info)); + xmalloc(sizeof(*s->sort_info)); s->sort_info->max_entries = atoi(sort_max_str); if (s->sort_info->max_entries < 2) s->sort_info->max_entries = 2; s->sort_info->entries = (struct zset_sort_entry **) - xmalloc (sizeof(*s->sort_info->entries) * - s->sort_info->max_entries); + xmalloc(sizeof(*s->sort_info->entries) * + s->sort_info->max_entries); s->sort_info->all_entries = (struct zset_sort_entry *) - xmalloc (sizeof(*s->sort_info->all_entries) * - s->sort_info->max_entries); + xmalloc(sizeof(*s->sort_info->all_entries) * + s->sort_info->max_entries); for (i = 0; i < s->sort_info->max_entries; i++) s->sort_info->entries[i] = s->sort_info->all_entries + i; } @@ -269,6 +272,7 @@ ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov) s->cache_position = 0; s->cache_rfd = 0; s->approx_limit = zh->approx_limit; + s->estimated_hit_count = 0; return s; } @@ -277,30 +281,42 @@ ZebraSet resultSetGet(ZebraHandle zh, const char *name) ZebraSet s; for (s = zh->sets; s; s = s->next) - if (!strcmp (s->name, name)) + if (!strcmp(s->name, name)) { if (!s->term_entries && !s->rset && s->rpn) { - NMEM nmem = nmem_create (); + NMEM nmem = nmem_create(); yaz_log(log_level_resultsets, "research %s", name); if (!s->rset_nmem) - s->rset_nmem=nmem_create(); + 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); + resultSetSortSingle(zh, nmem, s, s->rset, s->sortSpec, + &sort_status); } - nmem_destroy (nmem); + nmem_destroy(nmem); } return s; } return NULL; } -void resultSetInvalidate (ZebraHandle zh) +ZEBRA_RES resultSetGetBaseNames(ZebraHandle zh, const char *setname, + const char ***basenames, int *num_bases) +{ + ZebraSet sset = resultSetGet(zh, setname); + if (!sset) + return ZEBRA_FAIL; + *basenames = sset->basenames; + *num_bases = sset->num_bases; + return ZEBRA_OK; + +} + +void resultSetInvalidate(ZebraHandle zh) { ZebraSet s = zh->sets; @@ -311,7 +327,7 @@ void resultSetInvalidate (ZebraHandle zh) { if (s->cache_rfd) rset_close(s->cache_rfd); - rset_delete (s->rset); + rset_delete(s->rset); } s->rset = 0; s->cache_rfd = 0; @@ -337,7 +353,7 @@ void resultSetDestroy(ZebraHandle zh, int num, char **names,int *statuses) if (num >= 0) { for (i = 0; iname, names[i])) + if (!strcmp(s->name, names[i])) { if (statuses) statuses[i] = Z_DeleteStatus_success; @@ -349,22 +365,22 @@ void resultSetDestroy(ZebraHandle zh, int num, char **names,int *statuses) { *ss = s->next; - xfree (s->sort_info->all_entries); - xfree (s->sort_info->entries); - xfree (s->sort_info); + xfree(s->sort_info->all_entries); + xfree(s->sort_info->entries); + xfree(s->sort_info); if (s->nmem) - nmem_destroy (s->nmem); + nmem_destroy(s->nmem); if (s->rset) { if (s->cache_rfd) rset_close(s->cache_rfd); - rset_delete (s->rset); + rset_delete(s->rset); } if (s->rset_nmem) nmem_destroy(s->rset_nmem); - xfree (s->name); - xfree (s); + xfree(s->name); + xfree(s); } else ss = &s->next; @@ -411,13 +427,13 @@ ZebraMetaRecord *zebra_meta_records_create(ZebraHandle zh, const char *name, if (!log_level_set) loglevels(); - if (!(sset = resultSetGet (zh, name))) + if (!(sset = resultSetGet(zh, name))) return NULL; if (!(rset = sset->rset)) { if (!sset->term_entries) return 0; - sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num); + sr = (ZebraMetaRecord *) xmalloc(sizeof(*sr) * num); for (i = 0; icache_rfd) rset_close(sset->cache_rfd); - rfd = rset_open (rset, RSETF_READ); + rfd = rset_open(rset, RSETF_READ); } - while (num_i < num && rset_read (rfd, &key, 0)) + while (num_i < num && rset_read(rfd, &key, 0)) { zint this_sys = key.mem[sysno_mem_index]; if (this_sys != psysno) @@ -507,7 +523,7 @@ ZebraMetaRecord *zebra_meta_records_create(ZebraHandle zh, const char *name, continue; } position++; - assert (num_i < num); + assert(num_i < num); if (position == positions[num_i]) { sr[num_i].sysno = psysno; @@ -525,24 +541,22 @@ ZebraMetaRecord *zebra_meta_records_create(ZebraHandle zh, const char *name, return sr; } -void zebra_meta_records_destroy (ZebraHandle zh, ZebraMetaRecord *records, - int num) +void zebra_meta_records_destroy(ZebraHandle zh, ZebraMetaRecord *records, + int num) { assert(zh); /* compiler shut up about unused arg */ - xfree (records); + xfree(records); } struct sortKeyInfo { int relation; -#if 0 - int attrUse; -#else - int ord; -#endif - int numerical; + int *ord; /* array of ord for each database searched */ + int *numerical; /* array of ord for each database searched */ + const char *index_type; }; void resultSetInsertSort(ZebraHandle zh, ZebraSet sset, + int database_no, struct sortKeyInfo *criteria, int num_criteria, zint sysno, char *cmp_buf[], char *tmp_cmp_buf[]) @@ -550,18 +564,53 @@ void resultSetInsertSort(ZebraHandle zh, ZebraSet sset, struct zset_sort_entry *new_entry = NULL; struct zset_sort_info *sort_info = sset->sort_info; int i, j; + WRBUF w = wrbuf_alloc(); - sortIdx_sysno (zh->reg->sortIdx, sysno); + zebra_sort_sysno(zh->reg->sort_index, sysno); for (i = 0; ireg->sort_index, criteria[i].ord[database_no]); + wrbuf_rewind(w); + if (zebra_sort_read(zh->reg->sort_index, 0, w)) + { + /* consider each sort entry and take lowest/highest one + of the one as sorting key depending on whether sort is + ascending/descending */ + int off = 0; + while (off != wrbuf_len(w)) + { + size_t l = strlen(wrbuf_buf(w)+off); + assert(off < wrbuf_len(w)); + + if (l >= SORT_IDX_ENTRYSIZE) + l = SORT_IDX_ENTRYSIZE-1; + if ( (off == 0) + || (criteria[i].relation == 'A' + && strcmp(wrbuf_buf(w)+off, this_entry_buf) < 0) + || (criteria[i].relation == 'D' + && strcmp(wrbuf_buf(w)+off, this_entry_buf) > 0) + ) + { + memcpy(this_entry_buf, wrbuf_buf(w)+off, l); + this_entry_buf[l] = '\0'; + } + off += 1 + strlen(wrbuf_buf(w)+off); + } + } + } + else { - sortIdx_type(zh->reg->sortIdx, criteria[i].ord); - sortIdx_read(zh->reg->sortIdx, this_entry_buf); + yaz_log(log_level_sort, "criteria[i].ord is -1 so not reading from sort index"); } } + wrbuf_destroy(w); i = sort_info->num_entries; while (--i >= 0) { @@ -571,20 +620,32 @@ void resultSetInsertSort(ZebraHandle zh, ZebraSet sset, char *this_entry_buf = tmp_cmp_buf[j]; char *other_entry_buf = cmp_buf[j] + i * SORT_IDX_ENTRYSIZE; - if (criteria[j].numerical) + if (criteria[j].numerical[database_no]) { - double diff = atof(this_entry_buf) - atof(other_entry_buf); - rel = 0; + char this_entry_org[1024]; + char other_entry_org[1024]; + double diff; + const char *index_type = criteria[j].index_type; + zebra_term_untrans(zh, index_type, this_entry_org, + this_entry_buf); + zebra_term_untrans(zh, index_type, other_entry_org, + other_entry_buf); + diff = atof(this_entry_org) - atof(other_entry_org); + if (diff > 0.0) rel = 1; else if (diff < 0.0) rel = -1; + else + rel = 0; } else { rel = memcmp(this_entry_buf, other_entry_buf, SORT_IDX_ENTRYSIZE); } + /* when the compare is equal, continue to next criteria, + else break out */ if (rel) break; } @@ -602,15 +663,19 @@ void resultSetInsertSort(ZebraHandle zh, ZebraSet sset, } } ++i; + yaz_log(log_level_sort, "ok, we want to insert record at position %d",i); j = sort_info->max_entries; - if (i == j) + if (i == j){ + yaz_log(log_level_sort, "sort_info->max_entries reached (%d) abort sort",j); return; + } if (sort_info->num_entries == j) --j; else j = (sort_info->num_entries)++; new_entry = sort_info->entries[j]; + /* move up all higher entries (to make room) */ while (j != i) { int k; @@ -623,8 +688,10 @@ void resultSetInsertSort(ZebraHandle zh, ZebraSet sset, sort_info->entries[j] = sort_info->entries[j-1]; --j; } + /* and insert the new entry at the correct place */ sort_info->entries[i] = new_entry; - assert (new_entry); + assert(new_entry); + /* and add this to the compare buffer */ for (i = 0; ientries[i] = new_entry; - assert (new_entry); + assert(new_entry); new_entry->sysno = sysno; new_entry->score = score; } @@ -742,7 +809,7 @@ ZebraSet resultSetClone(ZebraHandle zh, const char *setname, nset->num_bases = rset->num_bases; nset->basenames = - nmem_malloc (nset->nmem, nset->num_bases * sizeof(*rset->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]); @@ -774,8 +841,8 @@ ZEBRA_RES resultSetSort(ZebraHandle zh, NMEM nmem, if (!log_level_set) loglevels(); yaz_log(log_level_sort, "result set sort input=%s output=%s", - *input_setnames, output_setname); - sset = resultSetGet (zh, input_setnames[0]); + *input_setnames, output_setname); + sset = resultSetGet(zh, input_setnames[0]); if (!sset) { zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST, @@ -788,11 +855,11 @@ ZEBRA_RES resultSetSort(ZebraHandle zh, NMEM nmem, input_setnames[0]); return ZEBRA_FAIL; } - if (strcmp (output_setname, input_setnames[0])) + if (strcmp(output_setname, input_setnames[0])) sset = resultSetClone(zh, output_setname, sset); sset->sortSpec = copy_SortKeySpecList(sort_sequence, sset->nmem); - return resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, - sort_status); + return resultSetSortSingle(zh, nmem, sset, rset, sort_sequence, + sort_status); } ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, @@ -801,6 +868,7 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, int *sort_status) { int i; + int ib; int n = 0; zint kno = 0; zint psysno = 0; @@ -814,6 +882,9 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, TERMID *terms; int numTerms = 0; size_t sysno_mem_index = 0; + + int numbases = zh->num_basenames; + yaz_log(log_level_sort, "searching %d databases",numbases); if (zh->m_staticrank) sysno_mem_index = 1; @@ -829,14 +900,23 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, num_criteria = sort_sequence->num_specs; if (num_criteria > ZSET_SORT_MAX_LEVEL) num_criteria = ZSET_SORT_MAX_LEVEL; + /* set up the search criteria */ for (i = 0; i < num_criteria; i++) { Z_SortKeySpec *sks = sort_sequence->specs[i]; Z_SortKey *sk; - ZEBRA_RES res; - - sort_criteria[i].ord = -1; - sort_criteria[i].numerical = 0; + + sort_criteria[i].ord = (int *) + nmem_malloc(nmem, sizeof(int)*numbases); + sort_criteria[i].numerical = (int *) + nmem_malloc(nmem, sizeof(int)*numbases); + + /* initialize ord and numerical for each database */ + for (ib = 0; ib < numbases; ib++) + { + sort_criteria[i].ord[ib] = -1; + sort_criteria[i].numerical[ib] = 0; + } if (sks->which == Z_SortKeySpec_missingValueData) { @@ -866,46 +946,69 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, switch (sk->which) { case Z_SortKey_sortField: - yaz_log(log_level_sort, "key %d is of type sortField", - i+1); - sort_criteria[i].numerical = 0; - sort_criteria[i].ord = - zebraExplain_lookup_attr_str(zh->reg->zei, - zinfo_index_category_sort, - 's', - sk->u.sortField); - if (sks->which != Z_SortKeySpec_null - && sort_criteria[i].ord == -1) + yaz_log(log_level_sort, "key %d is of type sortField", i+1); + for (ib = 0; ib < numbases; ib++) { - zebra_setError(zh, - YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0); - return ZEBRA_FAIL; + zebraExplain_curDatabase(zh->reg->zei, zh->basenames[ib]); + sort_criteria[i].numerical[ib] = 0; + sort_criteria[i].ord[ib] = + zebraExplain_lookup_attr_str(zh->reg->zei, + zinfo_index_category_sort, + 0, sk->u.sortField); + if (sks->which != Z_SortKeySpec_null + && sort_criteria[i].ord[ib] == -1) + { + zebra_setError(zh, + YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0); + return ZEBRA_FAIL; + } } break; case Z_SortKey_elementSpec: - yaz_log(log_level_sort, "key %d is of type elementSpec", - i+1); + yaz_log(log_level_sort, "key %d is of type elementSpec", i+1); zebra_setError(zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0); return ZEBRA_FAIL; case Z_SortKey_sortAttributes: yaz_log(log_level_sort, "key %d is of type sortAttributes", i+1); - res = zebra_sort_get_ord(zh, sk->u.sortAttributes, - &sort_criteria[i].ord, - &sort_criteria[i].numerical); - if (sks->which != Z_SortKeySpec_null && res != ZEBRA_OK) - return ZEBRA_FAIL; + /* for every database we searched, get the sort index file + id (ord) and its numerical indication and store them in + the sort_criteria */ + for (ib = 0; ib < numbases; ib++) + { + zebraExplain_curDatabase(zh->reg->zei, zh->basenames[ib]); + if (zebra_sort_get_ord(zh, sk->u.sortAttributes, + &sort_criteria[i].ord[ib], + &sort_criteria[i].numerical[ib]) != + ZEBRA_OK && sks->which != Z_SortKeySpec_null) + return ZEBRA_FAIL; + } break; } + /* right now we look up the index type based on the first database + if the index_type's can differ between the indexes of different + databases (which i guess they can?) then we have to store the + index types for each database, just like the ord and numerical */ + if (zebraExplain_lookup_ord(zh->reg->zei, sort_criteria[i].ord[0], + &sort_criteria[i].index_type, + 0, 0)) + { + zebra_setError(zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0); + return ZEBRA_FAIL; + } } /* allocate space for each cmpare buf + one extra for tmp comparison */ + /* cmp_buf is an array of array, the first dimension is the criteria and the second dimension are + all other result entries to compare against. This is slowly filled when records are processed. + tmp_cmp_buf is an array with a value of the current record for each criteria + */ for (i = 0; isort_info->max_entries * SORT_IDX_ENTRYSIZE); tmp_cmp_buf[i] = xmalloc(SORT_IDX_ENTRYSIZE); } - rfd = rset_open (rset, RSETF_READ); - while (rset_read (rfd, &key, &termid)) + rfd = rset_open(rset, RSETF_READ); + while (rset_read(rfd, &key, &termid)) { zint this_sys = key.mem[sysno_mem_index]; if (log_level_searchhits) @@ -913,15 +1016,44 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem, kno++; if (this_sys != psysno) { + int database_no = 0; + if ((sset->hits & 255) == 0 && zh->break_handler_func) + { + if (zh->break_handler_func(zh->break_handler_data)) + { + rset_set_hits_limit(rset, 0); + break; + } + } (sset->hits)++; psysno = this_sys; - resultSetInsertSort(zh, sset, + + /* determine database from the term, but only bother if more than + one database is in use*/ + if (numbases > 1 && termid->ol) + { + const char *this_db = 0; + if (zebraExplain_lookup_ord(zh->reg->zei, termid->ol->ord, 0, &this_db, 0) + == 0 && this_db) + { + for (ib = 0; ib < numbases; ib++) + if (!strcmp(this_db, zh->basenames[ib])) + database_no = ib; + } + } +#if 0 + yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " database_no=%d", this_sys, + database_no); + ord_list_print(termid->ol); +#endif + resultSetInsertSort(zh, sset, database_no, sort_criteria, num_criteria, psysno, cmp_buf, tmp_cmp_buf); } } - rset_close (rfd); + rset_close(rfd); + /* free the compare buffers */ for (i = 0; ihits); for (i = 0; i < numTerms; i++) yaz_log(log_level_sort, "term=\"%s\" type=%s count=" ZINT_FORMAT, - terms[i]->name, terms[i]->flags, terms[i]->rset->hits_count); + terms[i]->name, terms[i]->flags, terms[i]->rset->hits_count); *sort_status = Z_SortResponse_success; return ZEBRA_OK; } @@ -941,7 +1073,7 @@ RSET resultSetRef(ZebraHandle zh, const char *resultSetId) { ZebraSet s; - if ((s = resultSetGet (zh, resultSetId))) + if ((s = resultSetGet(zh, resultSetId))) return s->rset; return NULL; } @@ -969,11 +1101,11 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, sort_info = zebraSet->sort_info; sort_info->num_entries = 0; zebraSet->hits = 0; + zebraSet->estimated_hit_count = 0; rset_getterms(rset, 0, 0, &n); terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n); rset_getterms(rset, terms, n, &numTerms); - rank_class = zebraRankLookup(zh, rank_handler_name); if (!rank_class) { @@ -987,10 +1119,8 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, struct rank_control *rc = rank_class->control; int score; zint count = 0; - - void *handle = - (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem, - terms, numTerms); + void *handle = (*rc->begin) (zh->reg, rank_class->class_handle, rset, + nmem, terms, numTerms); zint psysno = 0; /* previous doc id / sys no */ zint pstaticrank = 0; /* previous static rank */ int stop_flag = 0; @@ -1004,23 +1134,34 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, key_logdump_txt(log_level_searchhits, &key, termid->name); if (this_sys != psysno) { /* new record .. */ + if (!(rfd->counted_items & 255) && zh->break_handler_func) + { + if (zh->break_handler_func(zh->break_handler_data)) + { + yaz_log(YLOG_LOG, "Aborted search"); + stop_flag = 1; + } + } if (rfd->counted_items > rset->hits_limit) - break; + stop_flag = 1; + if (stop_flag) + { + zebraSet->estimated_hit_count = 1; + break; + } if (psysno) { /* only if we did have a previous record */ - score = (*rc->calc) (handle, psysno, pstaticrank, - &stop_flag); + score = (*rc->calc)(handle, psysno, pstaticrank, + &stop_flag); /* insert the hit. A=Ascending */ - resultSetInsertRank (zh, sort_info, psysno, score, 'A'); + 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); + (*rc->add)(handle, CAST_ZINT_TO_INT(seqno), termid); } /* no more items */ if (psysno) @@ -1030,8 +1171,8 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, resultSetInsertRank(zh, sort_info, psysno, score, 'A'); count++; } - (*rc->end) (zh->reg, handle); - rset_close (rfd); + (*rc->end)(zh->reg, handle); + rset_close(rfd); } zebraSet->hits = rset->hits_count; @@ -1049,7 +1190,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet, ZebraRankClass zebraRankLookup(ZebraHandle zh, const char *name) { ZebraRankClass p = zh->reg->rank_classes; - while (p && strcmp (p->control->name, name)) + while (p && strcmp(p->control->name, name)) p = p->next; if (p && !p->init_flag) { @@ -1062,10 +1203,10 @@ ZebraRankClass zebraRankLookup(ZebraHandle zh, const char *name) void zebraRankInstall(struct zebra_register *reg, struct rank_control *ctrl) { - ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p)); - p->control = (struct rank_control *) xmalloc (sizeof(*p->control)); - memcpy (p->control, ctrl, sizeof(*p->control)); - p->control->name = xstrdup (ctrl->name); + ZebraRankClass p = (ZebraRankClass) xmalloc(sizeof(*p)); + p->control = (struct rank_control *) xmalloc(sizeof(*p->control)); + memcpy(p->control, ctrl, sizeof(*p->control)); + p->control->name = xstrdup(ctrl->name); p->init_flag = 0; p->next = reg->rank_classes; reg->rank_classes = p; @@ -1166,7 +1307,11 @@ ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname, if (ret == (size_t)(-1)) *termlen = 0; else + { + yaz_iconv(zh->iconv_from_utf8, 0, 0, + &outbuf, &outleft); *termlen = outbuf - termbuf; + } } else { @@ -1204,7 +1349,7 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, struct it_key key; RSET rsets[2], rset_comb; RSET rset_temp = rset_create_temp(nmem, kc, kc->scope, - res_get (zh->res, "setTmpDir"),0 ); + res_get(zh->res, "setTmpDir"),0 ); TERMID termid; RSFD rsfd = rset_open(rset_temp, RSETF_WRITE); @@ -1214,8 +1359,8 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, key.mem[2] = 0; key.mem[3] = 0; key.len = 2; - rset_write (rsfd, &key); - rset_close (rsfd); + rset_write(rsfd, &key); + rset_close(rsfd); rsets[0] = rset_temp; rsets[1] = rset_dup(sset->rset); @@ -1231,7 +1376,7 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, struct ord_list *ol; for (ol = termid->ol; ol; ol = ol->next) { - zebra_snippets_append(snippets, key.mem[key.len-1], + zebra_snippets_append(snippets, key.mem[key.len-1], 0, ol->ord, termid->name); } } @@ -1245,9 +1390,139 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname, return ZEBRA_OK; } +static ZEBRA_RES zebra_recid_to_sysno(ZebraHandle zh, + const char **basenames, int num_bases, + zint recid, + zint *sysnos, int *no_sysnos) +{ + ZEBRA_RES res = ZEBRA_OK; + int sysnos_offset = 0; + int i; + + if (!zh->reg->isamb || !zh->m_segment_indexing) + { + if (sysnos_offset < *no_sysnos) + *sysnos = recid; + sysnos_offset++; + } + else + { + for (i = 0; res == ZEBRA_OK && i < num_bases; i++) + { + const char *database = basenames[i]; + if (zebraExplain_curDatabase(zh->reg->zei, database) == 0) + { + const char *index_type = "w"; + const char *use_string = "_ALLRECORDS"; + int ord; + zinfo_index_category_t cat = zinfo_index_category_alwaysmatches; + ord = zebraExplain_lookup_attr_str(zh->reg->zei, cat, + index_type, use_string); + if (ord != -1) + { + char ord_buf[32]; + int ord_len = key_SU_encode(ord, ord_buf); + char *info; + + ord_buf[ord_len] = '\0'; + + info = dict_lookup(zh->reg->dict, ord_buf); + if (info) + { + if (*info != sizeof(ISAM_P)) + { + res = ZEBRA_FAIL; + } + else + { + ISAM_P isam_p; + ISAMB_PP pt; + struct it_key key_until, key_found; + int i = 0; + int r; + + memcpy(&isam_p, info+1, sizeof(ISAM_P)); + + pt = isamb_pp_open(zh->reg->isamb, isam_p, 2); + if (!pt) + res = ZEBRA_FAIL; + else + { + key_until.mem[i++] = recid; + key_until.mem[i++] = 0; /* section_id */ + if (zh->m_segment_indexing) + key_until.mem[i++] = 0; /* segment */ + key_until.mem[i++] = 0; + key_until.len = i; + + r = isamb_pp_forward(pt, &key_found, &key_until); + while (r && key_found.mem[0] == recid) + { + if (sysnos_offset < *no_sysnos) + sysnos[sysnos_offset++] = + key_found.mem[key_found.len-1]; + r = isamb_pp_read(pt, &key_found); + } + isamb_pp_close(pt); + } + } + } + } + } + } + } + *no_sysnos = sysnos_offset; + return res; +} + +ZEBRA_RES zebra_result_recid_to_sysno(ZebraHandle zh, + const char *setname, + zint recid, + zint *sysnos, int *no_sysnos) +{ + const char **basenames; + int num_bases; + ZEBRA_RES res; + + res = resultSetGetBaseNames(zh, setname, &basenames, &num_bases); + if (res != ZEBRA_OK) + return ZEBRA_FAIL; + + return zebra_recid_to_sysno(zh, basenames, num_bases, + recid, sysnos, no_sysnos); +} + +void zebra_count_set(ZebraHandle zh, RSET rset, zint *count, + zint approx_limit) +{ + zint psysno = 0; + struct it_key key; + RSFD rfd; + + yaz_log(YLOG_DEBUG, "count_set"); + + rset->hits_limit = approx_limit; + + *count = 0; + rfd = rset_open(rset, RSETF_READ); + while (rset_read(rfd, &key,0 /* never mind terms */)) + { + if (key.mem[0] != psysno) + { + psysno = key.mem[0]; + if (rfd->counted_items >= rset->hits_limit) + break; + } + } + rset_close(rfd); + *count = rset->hits_count; +} + + /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab