X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzsets.c;h=efe14c78a54aa5c61f2ee6510b7a04daea275780;hb=73bedd5ba9152c9c107b502fae65723b551aff09;hp=6394f3da5dac9cc6defcaddff6568f30fd1d575e;hpb=ef696645cc3b7e0f4027008d1dc589c0f0f90c1f;p=idzebra-moved-to-github.git diff --git a/index/zsets.c b/index/zsets.c index 6394f3d..efe14c7 100644 --- a/index/zsets.c +++ b/index/zsets.c @@ -1,91 +1,26 @@ -/* - * Copyright (C) 1994-1999, Index Data - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: zsets.c,v $ - * Revision 1.23 1999-05-26 07:49:13 adam - * C++ compilation. - * - * Revision 1.22 1999/02/02 14:51:15 adam - * Updated WIN32 code specific sections. Changed header. - * - * Revision 1.21 1998/11/16 16:03:46 adam - * Moved loggin utilities to Yaz. Was implemented in file zlogs.c. - * - * Revision 1.20 1998/11/16 10:10:53 adam - * Fixed problem with zebraPosSetCreate that occurred when positions were - * less than 1. - * - * Revision 1.19 1998/09/22 10:48:22 adam - * Minor changes in search API. - * - * Revision 1.18 1998/09/22 10:03:45 adam - * Changed result sets to be persistent in the sense that they can - * be re-searched if needed. - * Fixed memory leak in rsm_or. - * - * Revision 1.17 1998/06/23 15:33:36 adam - * Added feature to specify sort criteria in query (type 7 specifies - * sort flags). - * - * Revision 1.16 1998/05/20 10:12:24 adam - * Implemented automatic EXPLAIN database maintenance. - * Modified Zebra to work with ASN.1 compiled version of YAZ. - * - * Revision 1.15 1998/03/05 08:45:14 adam - * New result set model and modular ranking system. Moved towards - * descent server API. System information stored as "SGML" records. - * - * Revision 1.14 1998/02/10 16:39:15 adam - * Minor change. - * - * Revision 1.13 1998/02/10 12:03:06 adam - * Implemented Sort. - * - * Revision 1.12 1997/09/25 14:57:36 adam - * Windows NT port. - * - * Revision 1.11 1996/12/23 15:30:46 adam - * Work on truncation. - * Bug fix: result sets weren't deleted after server shut down. - * - * Revision 1.10 1995/10/30 15:08:08 adam - * Bug fixes. - * - * Revision 1.9 1995/10/17 18:02:14 adam - * New feature: databases. Implemented as prefix to words in dictionary. - * - * Revision 1.8 1995/10/10 13:59:25 adam - * Function rset_open changed its wflag parameter to general flags. - * - * Revision 1.7 1995/10/06 14:38:01 adam - * New result set method: r_score. - * Local no (sysno) and score is transferred to retrieveCtrl. - * - * Revision 1.6 1995/09/28 09:19:49 adam - * xfree/xmalloc used everywhere. - * Extract/retrieve method seems to work for text records. - * - * Revision 1.5 1995/09/27 16:17:32 adam - * More work on retrieve. - * - * Revision 1.4 1995/09/07 13:58:36 adam - * New parameter: result-set file descriptor (RSFD) to support multiple - * positions within the same result-set. - * Boolean operators: and, or, not implemented. - * Result-set references. - * - * Revision 1.3 1995/09/06 16:11:19 adam - * Option: only one word key per file. - * - * Revision 1.2 1995/09/06 10:33:04 adam - * More work on present. Some log messages removed. - * - * Revision 1.1 1995/09/05 15:28:40 adam - * More work on search engine. - * - */ +/* $Id: zsets.c,v 1.69 2004-10-26 15:32:11 heikki Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 + Index Data Aps + +This file is part of the Zebra server. + +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 +Software Foundation; either version 2, or (at your option) any later +version. + +Zebra is distributed in the hope that it will be useful, but WITHOUT ANY +WARRANTY; without even the implied warranty of MERCHANTABILITY or +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. +*/ + + #include #include #ifdef WIN32 @@ -94,26 +29,37 @@ #include #endif -#include "zserver.h" -#include +#include "index.h" +#include #define SORT_IDX_ENTRYSIZE 64 #define ZSET_SORT_MAX_LEVEL 3 +struct zebra_set_term_entry { + int reg_type; + char *db; + int set; + int use; + char *term; +}; struct zebra_set { char *name; RSET rset; NMEM nmem; - int hits; + NMEM rset_nmem; /* for creating the rsets in */ + zint hits; int num_bases; char **basenames; Z_RPNQuery *rpn; struct zset_sort_info *sort_info; + struct zebra_set_term_entry *term_entries; + int term_entries_max; struct zebra_set *next; + int locked; }; struct zset_sort_entry { - int sysno; + zint sysno; int score; char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE]; }; @@ -125,11 +71,13 @@ struct zset_sort_info { struct zset_sort_entry **entries; }; -ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output, - Z_RPNQuery *rpn, int num_bases, char **basenames, - const char *setname) +ZebraSet resultSetAddRPN (ZebraHandle zh, NMEM m, + Z_RPNQuery *rpn, int num_bases, + char **basenames, + const char *setname) { ZebraSet zebraSet; + int i; zh->errCode = 0; zh->errString = NULL; @@ -137,22 +85,112 @@ ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output, zebraSet = resultSetAdd (zh, setname, 1); if (!zebraSet) - return 0; + return 0; + zebraSet->locked = 1; zebraSet->rpn = 0; + zebraSet->nmem = m; + zebraSet->rset_nmem=nmem_create(); + zebraSet->num_bases = num_bases; - zebraSet->basenames = basenames; - zebraSet->nmem = odr_extract_mem (input); + zebraSet->basenames = + nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames)); + for (i = 0; ibasenames[i] = nmem_strdup (zebraSet->nmem, basenames[i]); - zebraSet->rset = rpn_search (zh, output->mem, rpn, - zebraSet->num_bases, - zebraSet->basenames, zebraSet->name, - zebraSet); + + zebraSet->rset = rpn_search (zh, zebraSet->nmem, zebraSet->rset_nmem, + rpn, zebraSet->num_bases, + zebraSet->basenames, zebraSet->name, + zebraSet); zh->hits = zebraSet->hits; if (zebraSet->rset) zebraSet->rpn = rpn; + zebraSet->locked = 0; return zebraSet; } +void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type, + const char *db, int set, + int use, const char *term) +{ + assert(zh); /* compiler shut up */ + if (!s->nmem) + 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)); + 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].set = set; + s->term_entries[s->hits].use = use; + s->term_entries[s->hits].term = nmem_strdup (s->nmem, term); + } + (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; @@ -160,40 +198,52 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov) for (s = zh->sets; s; s = s->next) if (!strcmp (s->name, name)) - break; + break; if (s) { - logf (LOG_DEBUG, "updating result set %s", name); - if (!ov) - return NULL; - if (s->rset) - rset_delete (s->rset); - if (s->nmem) - nmem_destroy (s->nmem); + yaz_log (LOG_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); } else { - logf (LOG_DEBUG, "adding result set %s", name); - s = (ZebraSet) xmalloc (sizeof(*s)); - s->next = zh->sets; - zh->sets = s; - s->name = (char *) xmalloc (strlen(name)+1); - strcpy (s->name, name); - - s->sort_info = (struct zset_sort_info *) - xmalloc (sizeof(*s->sort_info)); - s->sort_info->max_entries = 1000; - s->sort_info->entries = (struct zset_sort_entry **) - 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); - for (i = 0; i < s->sort_info->max_entries; i++) - s->sort_info->entries[i] = s->sort_info->all_entries + i; + const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000"); + + yaz_log (LOG_DEBUG, "adding result set %s", name); + s = (ZebraSet) xmalloc (sizeof(*s)); + s->next = zh->sets; + zh->sets = s; + s->name = (char *) xmalloc (strlen(name)+1); + strcpy (s->name, name); + + s->sort_info = (struct zset_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); + s->sort_info->all_entries = (struct zset_sort_entry *) + 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; } + s->locked = 0; + s->term_entries = 0; + s->hits = 0; s->rset = 0; - s->nmem = 0; + s->rset_nmem=0; + s->nmem = 0; + s->rpn = 0; return s; } @@ -204,47 +254,89 @@ ZebraSet resultSetGet (ZebraHandle zh, const char *name) for (s = zh->sets; s; s = s->next) if (!strcmp (s->name, name)) { - if (!s->rset && s->rpn) + if (!s->term_entries && !s->rset && s->rpn) { NMEM nmem = nmem_create (); + yaz_log (LOG_LOG, "research %s", name); + if (!s->rset_nmem) + s->rset_nmem=nmem_create(); s->rset = - rpn_search (zh, nmem, s->rpn, s->num_bases, - s->basenames, s->name, s); + rpn_search (zh, nmem, s->rset_nmem, s->rpn, s->num_bases, + s->basenames, s->name, s); nmem_destroy (nmem); - } return s; } return NULL; } - -void resultSetDestroy (ZebraHandle zh) +void resultSetInvalidate (ZebraHandle zh) { - ZebraSet s, s1; - - for (s = zh->sets; s; s = s1) + ZebraSet s = zh->sets; + + for (; s; s = s->next) { - s1 = s->next; - - xfree (s->sort_info->all_entries); - xfree (s->sort_info->entries); - xfree (s->sort_info); + if (s->rset) + rset_delete (s->rset); + s->rset = 0; + if (s->rset_nmem) + nmem_destroy(s->rset_nmem); + s->rset_nmem=0; + } +} - if (s->nmem) - nmem_destroy (s->nmem); - rset_delete (s->rset); - xfree (s->name); - xfree (s); +void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses) +{ + ZebraSet * ss = &zh->sets; + int i; + + if (statuses) + for (i = 0; ierrCode = 0; + zh->errString = NULL; + while (*ss) + { + int i = -1; + ZebraSet s = *ss; + if (num >= 0) + { + for (i = 0; iname, names[i])) + { + if (statuses) + statuses[i] = Z_DeleteStatus_success; + i = -1; + break; + } + } + if (i < 0) + { + *ss = s->next; + + xfree (s->sort_info->all_entries); + xfree (s->sort_info->entries); + xfree (s->sort_info); + + if (s->nmem) + nmem_destroy (s->nmem); + if (s->rset) + rset_delete (s->rset); + if (s->rset_nmem) + nmem_destroy(s->rset_nmem); + xfree (s->name); + xfree (s); + } + else + ss = &s->next; } - zh->sets = NULL; } ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, - int num, int *positions) + int num, int *positions) { ZebraSet sset; - ZebraPosSet sr; + ZebraPosSet sr = 0; RSET rset; int i; struct zset_sort_info *sort_info; @@ -252,421 +344,531 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name, if (!(sset = resultSetGet (zh, name))) return NULL; if (!(rset = sset->rset)) - return NULL; - sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num); - for (i = 0; isort_info; - if (sort_info) - { - int position; - - for (i = 0; i 0 && position <= sort_info->num_entries) - { - logf (LOG_DEBUG, "got pos=%d (sorted)", position); - sr[i].sysno = sort_info->entries[position-1]->sysno; - sr[i].score = sort_info->entries[position-1]->score; - } - } - } - /* did we really get all entries using sort ? */ - for (i = 0; iterm_entries) + return 0; + sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num); + for (i = 0; iterm_entries_max) + { + sr[i].term = sset->term_entries[positions[i]-1].term; + sr[i].db = sset->term_entries[positions[i]-1].db; + } + } } - if (i < num) /* nope, get the rest, unsorted - sorry */ + else { - int position = 0; - int num_i = 0; - int psysno = 0; - int term_index; - RSFD rfd; - struct it_key key; - - if (sort_info) - 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, &term_index)) - { - if (key.sysno != psysno) - { - psysno = key.sysno; - if (sort_info) - { - /* determine we alreay have this in our set */ - for (i = sort_info->num_entries; --i >= 0; ) - if (psysno == sort_info->entries[i]->sysno) - break; - if (i >= 0) - continue; - } - position++; - assert (num_i < num); - if (position == positions[num_i]) - { - sr[num_i].sysno = psysno; - logf (LOG_DEBUG, "got pos=%d (unsorted)", position); - sr[num_i].score = -1; - num_i++; - } - } - } - rset_close (rset, rfd); + sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num); + for (i = 0; isort_info; + if (sort_info) + { + int position; + + for (i = 0; i 0 && position <= sort_info->num_entries) + { + yaz_log (LOG_DEBUG, "got pos=%d (sorted)", position); + sr[i].sysno = sort_info->entries[position-1]->sysno; + sr[i].score = sort_info->entries[position-1]->score; + } + } + } + /* did we really get all entries using sort ? */ + for (i = 0; inum_entries; + while (num_i < num && positions[num_i] < position) + num_i++; + rfd = rset_open (rset, RSETF_READ); + while (num_i < num && rset_read (rfd, &key, 0)) + { + zint this_sys = key.mem[0]; + if (this_sys != psysno) + { + psysno = this_sys; + if (sort_info) + { + /* determine we alreay have this in our set */ + for (i = sort_info->num_entries; --i >= 0; ) + if (psysno == sort_info->entries[i]->sysno) + break; + if (i >= 0) + continue; + } + position++; + assert (num_i < num); + if (position == positions[num_i]) + { + sr[num_i].sysno = psysno; + yaz_log (LOG_DEBUG, "got pos=%d (unsorted)", position); + sr[num_i].score = -1; + num_i++; + } + } + } + rset_close (rfd); + } } return sr; } void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num) { + assert(zh); /* compiler shut up about unused arg */ xfree (records); } -struct sortKey { +struct sortKeyInfo { int relation; int attrUse; + int numerical; }; void resultSetInsertSort (ZebraHandle zh, ZebraSet sset, - struct sortKey *criteria, int num_criteria, - int sysno) + struct sortKeyInfo *criteria, int num_criteria, + zint sysno) { struct zset_sort_entry this_entry; struct zset_sort_entry *new_entry = NULL; struct zset_sort_info *sort_info = sset->sort_info; int i, j; - sortIdx_sysno (zh->sortIdx, sysno); + sortIdx_sysno (zh->reg->sortIdx, sysno); for (i = 0; isortIdx, criteria[i].attrUse); - sortIdx_read (zh->sortIdx, this_entry.buf[i]); + sortIdx_type (zh->reg->sortIdx, criteria[i].attrUse); + sortIdx_read (zh->reg->sortIdx, this_entry.buf[i]); } i = sort_info->num_entries; while (--i >= 0) { - int rel = 0; - for (j = 0; jentries[i]->buf[j], - SORT_IDX_ENTRYSIZE); - if (rel) - break; - } - if (!rel) - break; - if (criteria[j].relation == 'A') + int rel = 0; + for (j = 0; j 0) - break; - } + if (criteria[j].numerical) + { + double diff = atof(this_entry.buf[j]) - + atof(sort_info->entries[i]->buf[j]); + rel = 0; + if (diff > 0.0) + rel = 1; + else if (diff < 0.0) + rel = -1; + } + else + { + rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j], + SORT_IDX_ENTRYSIZE); + } + if (rel) + break; + } + if (!rel) + break; + if (criteria[j].relation == 'A') + { + if (rel > 0) + break; + } else if (criteria[j].relation == 'D') - { - if (rel < 0) - break; - } + { + if (rel < 0) + break; + } } - j = sort_info->max_entries-1; - if (i == j) - return; ++i; + j = sort_info->max_entries; + if (i == j) + return; + + if (sort_info->num_entries == j) + --j; + else + j = (sort_info->num_entries)++; new_entry = sort_info->entries[j]; while (j != i) { - sort_info->entries[j] = sort_info->entries[j-1]; - --j; + sort_info->entries[j] = sort_info->entries[j-1]; + --j; } - sort_info->entries[j] = new_entry; + sort_info->entries[i] = new_entry; assert (new_entry); - if (sort_info->num_entries != sort_info->max_entries) - (sort_info->num_entries)++; for (i = 0; ibuf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE); + memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE); new_entry->sysno = sysno; new_entry->score = -1; } void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info, - int sysno, int score, int relation) + zint sysno, int score, int relation) { struct zset_sort_entry *new_entry = NULL; int i, j; + assert(zh); /* compiler shut up about unused arg */ i = sort_info->num_entries; while (--i >= 0) { - int rel = 0; - - rel = score - sort_info->entries[i]->score; - - if (relation == 'D') - { - if (rel >= 0) - break; - } - else if (relation == 'A') - { - if (rel <= 0) - break; - } + int rel = 0; + + rel = score - sort_info->entries[i]->score; + + if (relation == 'D') + { + if (rel >= 0) + break; + } + else if (relation == 'A') + { + if (rel <= 0) + break; + } } - j = sort_info->max_entries-1; - if (i == j) - return; ++i; + j = sort_info->max_entries; + if (i == j) + return; + + if (sort_info->num_entries == j) + --j; + else + j = (sort_info->num_entries)++; + new_entry = sort_info->entries[j]; while (j != i) { - sort_info->entries[j] = sort_info->entries[j-1]; - --j; + sort_info->entries[j] = sort_info->entries[j-1]; + --j; } - sort_info->entries[j] = new_entry; + sort_info->entries[i] = new_entry; assert (new_entry); - if (sort_info->num_entries != sort_info->max_entries) - (sort_info->num_entries)++; new_entry->sysno = sysno; new_entry->score = score; } void resultSetSort (ZebraHandle zh, NMEM nmem, - int num_input_setnames, const char **input_setnames, - const char *output_setname, - Z_SortKeySpecList *sort_sequence, int *sort_status) + int num_input_setnames, const char **input_setnames, + const char *output_setname, + Z_SortKeySpecList *sort_sequence, int *sort_status) { ZebraSet sset; RSET rset; if (num_input_setnames == 0) { - zh->errCode = 208; - return ; + zh->errCode = 208; + return ; } if (num_input_setnames > 1) { - zh->errCode = 230; - return; + zh->errCode = 230; + return; } - logf (LOG_DEBUG, "result set sort input=%s output=%s", - *input_setnames, output_setname); + yaz_log (LOG_DEBUG, "result set sort input=%s output=%s", + *input_setnames, output_setname); sset = resultSetGet (zh, input_setnames[0]); if (!sset) { - zh->errCode = 30; - zh->errString = nmem_strdup (nmem, input_setnames[0]); - return; + zh->errCode = 30; + zh->errString = nmem_strdup (nmem, input_setnames[0]); + return; } if (!(rset = sset->rset)) { - zh->errCode = 30; - zh->errString = nmem_strdup (nmem, input_setnames[0]); + zh->errCode = 30; + zh->errString = nmem_strdup (nmem, input_setnames[0]); return; } if (strcmp (output_setname, input_setnames[0])) { - rset = rset_dup (rset); - sset = resultSetAdd (zh, output_setname, 1); + rset = rset_dup (rset); + sset = resultSetAdd (zh, output_setname, 1); sset->rset = rset; } resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status); } void resultSetSortSingle (ZebraHandle zh, NMEM nmem, - ZebraSet sset, RSET rset, - Z_SortKeySpecList *sort_sequence, int *sort_status) + ZebraSet sset, RSET rset, + Z_SortKeySpecList *sort_sequence, int *sort_status) { - int i, psysno = 0; + int i; + zint psysno = 0; struct it_key key; - struct sortKey sort_criteria[3]; + struct sortKeyInfo sort_criteria[3]; int num_criteria; - int term_index; RSFD rfd; + TERMID termid; - logf (LOG_DEBUG, "resultSetSortSingle start"); + yaz_log (LOG_LOG, "resultSetSortSingle start"); + assert(nmem); /* compiler shut up about unused param */ sset->sort_info->num_entries = 0; sset->hits = 0; num_criteria = sort_sequence->num_specs; if (num_criteria > 3) - num_criteria = 3; + num_criteria = 3; for (i = 0; i < num_criteria; i++) { - Z_SortKeySpec *sks = sort_sequence->specs[i]; - Z_SortKey *sk; - - if (*sks->sortRelation == Z_SortRelation_ascending) - sort_criteria[i].relation = 'A'; - else if (*sks->sortRelation == Z_SortRelation_descending) - sort_criteria[i].relation = 'D'; - else - { - zh->errCode = 214; - return; - } - if (sks->sortElement->which == Z_SortElement_databaseSpecific) - { - zh->errCode = 210; - return; - } - else if (sks->sortElement->which != Z_SortElement_generic) - { - zh->errCode = 237; - return; - } - sk = sks->sortElement->u.generic; - switch (sk->which) - { - case Z_SortKey_sortField: - logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1); - zh->errCode = 207; - return; - case Z_SortKey_elementSpec: - logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1); - zh->errCode = 207; - return; - case Z_SortKey_sortAttributes: - logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1); - sort_criteria[i].attrUse = - zebra_maps_sort (zh->zebra_maps, sk->u.sortAttributes); - logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse); - if (sort_criteria[i].attrUse == -1) - { - zh->errCode = 116; - return; - } - if (sortIdx_type (zh->sortIdx, sort_criteria[i].attrUse)) - { - zh->errCode = 207; - return; - } - break; - } + Z_SortKeySpec *sks = sort_sequence->specs[i]; + Z_SortKey *sk; + + if (*sks->sortRelation == Z_SortKeySpec_ascending) + sort_criteria[i].relation = 'A'; + else if (*sks->sortRelation == Z_SortKeySpec_descending) + sort_criteria[i].relation = 'D'; + else + { + zh->errCode = 214; + return; + } + if (sks->sortElement->which == Z_SortElement_databaseSpecific) + { + zh->errCode = 210; + return; + } + else if (sks->sortElement->which != Z_SortElement_generic) + { + zh->errCode = 237; + return; + } + sk = sks->sortElement->u.generic; + switch (sk->which) + { + case Z_SortKey_sortField: + yaz_log (LOG_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); + zh->errCode = 207; + return; + case Z_SortKey_sortAttributes: + yaz_log (LOG_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); + if (sort_criteria[i].attrUse == -1) + { + zh->errCode = 116; + return; + } + if (sortIdx_type (zh->reg->sortIdx, sort_criteria[i].attrUse)) + { + zh->errCode = 207; + return; + } + break; + } } rfd = rset_open (rset, RSETF_READ); - while (rset_read (rset, rfd, &key, &term_index)) + while (rset_read (rfd, &key,&termid)) + /* FIXME - pass a TERMID *, and use it for something below !! */ { - if (key.sysno != psysno) + zint this_sys = key.mem[0]; + if (this_sys != psysno) { - (sset->hits)++; - psysno = key.sysno; - resultSetInsertSort (zh, sset, - sort_criteria, num_criteria, psysno); + (sset->hits)++; + psysno = this_sys; + resultSetInsertSort (zh, sset, + sort_criteria, num_criteria, psysno); } } - rset_close (rset, rfd); - - zh->errCode = 0; - *sort_status = Z_SortStatus_success; - logf (LOG_DEBUG, "resultSetSortSingle end"); + rset_close (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 + *sort_status = Z_SortResponse_success; + yaz_log (LOG_LOG, "resultSetSortSingle end"); } -RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId) +RSET resultSetRef (ZebraHandle zh, const char *resultSetId) { ZebraSet s; if ((s = resultSetGet (zh, resultSetId))) - return s->rset; + return s->rset; return NULL; } -void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset) +void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset, NMEM nmem) { - int kno = 0; + zint kno = 0; struct it_key key; RSFD rfd; - int term_index, i; + TERMID termid; + TERMID *terms; + int numterms; + int i,n; ZebraRankClass rank_class; struct rank_control *rc; struct zset_sort_info *sort_info; + const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1"); + double cur,tot; + zint est=-2; /* -2 not done, -1 can't do, >0 actual estimate*/ + zint esthits; + double ratio; sort_info = zebraSet->sort_info; sort_info->num_entries = 0; zebraSet->hits = 0; - rfd = rset_open (rset, RSETF_READ); + n=0; + rset_getterms(rset,0,0,&n); + terms=malloc( sizeof(*terms)*n); + numterms=0; + rset_getterms(rset,terms,n,&numterms); - logf (LOG_DEBUG, "resultSetRank"); - for (i = 0; i < rset->no_rset_terms; i++) - logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s", - rset->rset_terms[i]->name, - rset->rset_terms[i]->nn, - rset->rset_terms[i]->flags); + rfd = rset_open (rset, RSETF_READ); - rank_class = zebraRankLookup (zh, "rank-1"); + rank_class = zebraRankLookup (zh, rank_handler_name); + if (!rank_class) + { + yaz_log (LOG_WARN, "No such rank handler: %s", rank_handler_name); + return; + } rc = rank_class->control; - if (rset_read (rset, rfd, &key, &term_index)) + if (rset_read (rfd, &key, &termid)) + { + zint psysno = key.mem[0]; + int score; + void *handle = + (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem, + terms, numterms); + (zebraSet->hits)++; + esthits=atoi(res_get_def(zh->res,"estimatehits","0")); + if (!esthits) + est=-1; /* can not do */ + do + { + zint this_sys = key.mem[0]; /* FIXME - assumes scope==2 */ + zint seqno = key.mem[1]; /* FIXME - assumes scope==2 */ + kno++; + if (this_sys != psysno) + { + score = (*rc->calc) (handle, psysno); + + resultSetInsertRank (zh, sort_info, psysno, score, 'A'); + (zebraSet->hits)++; + psysno = this_sys; + } + (*rc->add) (handle, seqno, termid); + + 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); + 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++; + } + while (i--) est*=10; + zebraSet->hits=est; + } + } + } + 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 (rfd); + + + for (i = 0; i < numterms; i++) { - int psysno = key.sysno; - int score; - void *handle = - (*rc->begin) (zh, rank_class->class_handle, rset); - (zebraSet->hits)++; - do - { - kno++; - if (key.sysno != psysno) - { - score = (*rc->calc) (handle, psysno); - - resultSetInsertRank (zh, sort_info, psysno, score, 'A'); - (zebraSet->hits)++; - psysno = key.sysno; - } - (*rc->add) (handle, key.seqno, term_index); - } - while (rset_read (rset, rfd, &key, &term_index)); - score = (*rc->calc) (handle, psysno); - resultSetInsertRank (zh, sort_info, psysno, score, 'A'); - (*rc->end) (zh, handle); + yaz_log (LOG_LOG, "term=\"%s\" " + " type=%s count=" ZINT_FORMAT, + terms[i]->name, + terms[i]->flags, + rset_count(terms[i]->rset)); } - rset_close (rset, rfd); - logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits); + yaz_log (LOG_DEBUG, ZINT_FORMAT " keys, "ZINT_FORMAT" distinct sysnos", + kno, zebraSet->hits); } ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name) { - ZebraRankClass p = zh->rank_classes; + ZebraRankClass p = zh->reg->rank_classes; while (p && strcmp (p->control->name, name)) - p = p->next; + p = p->next; if (p && !p->init_flag) { - if (p->control->create) - p->class_handle = (*p->control->create)(zh); - p->init_flag = 1; + if (p->control->create) + p->class_handle = (*p->control->create)(zh); + p->init_flag = 1; } return p; } -void zebraRankInstall (ZebraHandle zh, struct rank_control *ctrl) +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); p->init_flag = 0; - p->next = zh->rank_classes; - zh->rank_classes = p; + p->next = reg->rank_classes; + reg->rank_classes = p; } -void zebraRankDestroy (ZebraHandle zh) +void zebraRankDestroy (struct zebra_register *reg) { - ZebraRankClass p = zh->rank_classes; + ZebraRankClass p = reg->rank_classes; while (p) { - ZebraRankClass p_next = p->next; - if (p->init_flag && p->control->destroy) - (*p->control->destroy)(zh, p->class_handle); - xfree (p->control->name); - xfree (p->control); - xfree (p); - p = p_next; + 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); + p = p_next; } - zh->rank_classes = NULL; + reg->rank_classes = NULL; }